wireshark/file.c

4923 lines
150 KiB
C
Raw Normal View History

/* file.c
* File I/O routines
*
* Wireshark - Network traffic analyzer
* By Gerald Combs <gerald@wireshark.org>
* Copyright 1998 Gerald Combs
*
* SPDX-License-Identifier: GPL-2.0-or-later
*/
#include <config.h>
#include <time.h>
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
#include <ctype.h>
#include <errno.h>
#include <wsutil/tempfile.h>
#include <wsutil/file_util.h>
#include <wsutil/filesystem.h>
#include <version_info.h>
#include <wiretap/merge.h>
#include <epan/exceptions.h>
#include <epan/epan.h>
#include <epan/column.h>
#include <epan/packet.h>
#include <epan/column-utils.h>
#include <epan/expert.h>
#include <epan/prefs.h>
#include <epan/dfilter/dfilter.h>
#include <epan/epan_dissect.h>
#include <epan/tap.h>
#include <epan/dissectors/packet-ber.h>
#include <epan/timestamp.h>
#include <epan/dfilter/dfilter-macro.h>
#include <epan/strutil.h>
#include <epan/addr_resolv.h>
#include <epan/color_filters.h>
#include "cfile.h"
#include "file.h"
#include "fileset.h"
#include "frame_tvbuff.h"
#include "ui/alert_box.h"
#include "ui/simple_dialog.h"
#include "ui/main_statusbar.h"
#include "ui/progress_dlg.h"
#include "ui/ws_ui_util.h"
/* Needed for addrinfo */
#include <sys/types.h>
#ifdef HAVE_SYS_SOCKET_H
#include <sys/socket.h>
#endif
#ifdef HAVE_NETINET_IN_H
# include <netinet/in.h>
#endif
#ifdef _WIN32
# include <winsock2.h>
# include <ws2tcpip.h>
#endif
static gboolean read_record(capture_file *cf, dfilter_t *dfcode,
epan_dissect_t *edt, column_info *cinfo, gint64 offset);
static void rescan_packets(capture_file *cf, const char *action, const char *action_item, gboolean redissect);
Turn the code of "colorize_packet()" into a static routine that is given a word to use in the progress dialog, and a flag indicating whether the display filter is to be reevaluated or not, and: have "colorize_packet()" call that routine with "Colorizing" and FALSE as those arguments; have the filtering code call that routine with "Filtering" and TRUE as those arguments; add an exported routine to call that routine with "Reprocessing" and TRUE as those arguments, to use to re-generate the packet list and to re-filter the packets if a protocol preference has been changed. Keep track of whether preferences are changed from their initial value by a preferences file or a command-line option, or from their previous value by the "Preferences" dialog box; have "prefs_apply_all()" only call the "apply" callback for a module if they have. Call "prefs_apply_all()" after the command-line arguments have been parsed and after "OK" has been clicked in the "Preferences" dialog box, to notify modules of preference changes if they've registered a callback for that. After "OK" has been clicked in the "Preferences" dialog box, if any preferences have changed, call the reprocessing routine, as the summary line for some frames and/or the current display filter's value when applied to some frames may have changed as a result of a preference change. Do the same after "OK" or "Apply" has been clicked in the "Display Options" dialog box (as it controls a protocol preferences item. svn path=/trunk/; revision=2126
2000-07-09 03:29:42 +00:00
typedef enum {
MR_NOTMATCHED,
MR_MATCHED,
MR_ERROR
} match_result;
static match_result match_protocol_tree(capture_file *cf, frame_data *fdata,
void *criterion);
static void match_subtree_text(proto_node *node, gpointer data);
static match_result match_summary_line(capture_file *cf, frame_data *fdata,
void *criterion);
static match_result match_narrow_and_wide(capture_file *cf, frame_data *fdata,
void *criterion);
static match_result match_narrow(capture_file *cf, frame_data *fdata,
void *criterion);
static match_result match_wide(capture_file *cf, frame_data *fdata,
void *criterion);
static match_result match_binary(capture_file *cf, frame_data *fdata,
void *criterion);
static match_result match_regex(capture_file *cf, frame_data *fdata,
void *criterion);
static match_result match_dfilter(capture_file *cf, frame_data *fdata,
void *criterion);
static match_result match_marked(capture_file *cf, frame_data *fdata,
void *criterion);
static match_result match_time_reference(capture_file *cf, frame_data *fdata,
void *criterion);
static gboolean find_packet(capture_file *cf,
match_result (*match_function)(capture_file *, frame_data *, void *),
void *criterion, search_direction dir);
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
static void cf_rename_failure_alert_box(const char *filename, int err);
static void ref_time_packets(capture_file *cf);
/* Seconds spent processing packets between pushing UI updates. */
#define PROGBAR_UPDATE_INTERVAL 0.150
/* Show the progress bar after this many seconds. */
#define PROGBAR_SHOW_DELAY 0.5
/*
* We could probably use g_signal_...() instead of the callbacks below but that
* would require linking our CLI programs to libgobject and creating an object
* instance for the signals.
*/
typedef struct {
cf_callback_t cb_fct;
gpointer user_data;
} cf_callback_data_t;
static GList *cf_callbacks = NULL;
static void
cf_callback_invoke(int event, gpointer data)
{
cf_callback_data_t *cb;
GList *cb_item = cf_callbacks;
/* there should be at least one interested */
g_assert(cb_item != NULL);
while (cb_item != NULL) {
cb = (cf_callback_data_t *)cb_item->data;
cb->cb_fct(event, data, cb->user_data);
cb_item = g_list_next(cb_item);
}
}
void
cf_callback_add(cf_callback_t func, gpointer user_data)
{
cf_callback_data_t *cb;
cb = g_new(cf_callback_data_t,1);
cb->cb_fct = func;
cb->user_data = user_data;
cf_callbacks = g_list_prepend(cf_callbacks, cb);
}
void
cf_callback_remove(cf_callback_t func, gpointer user_data)
{
cf_callback_data_t *cb;
GList *cb_item = cf_callbacks;
while (cb_item != NULL) {
cb = (cf_callback_data_t *)cb_item->data;
if (cb->cb_fct == func && cb->user_data == user_data) {
cf_callbacks = g_list_remove(cf_callbacks, cb);
g_free(cb);
return;
}
cb_item = g_list_next(cb_item);
}
g_assert_not_reached();
}
void
cf_timestamp_auto_precision(capture_file *cf)
{
int i;
/* don't try to get the file's precision if none is opened */
if (cf->state == FILE_CLOSED) {
return;
}
/* Set the column widths of those columns that show the time in
"command-line-specified" format. */
for (i = 0; i < cf->cinfo.num_cols; i++) {
if (col_has_time_fmt(&cf->cinfo, i)) {
packet_list_resize_column(i);
}
}
}
gulong
cf_get_computed_elapsed(capture_file *cf)
{
return cf->computed_elapsed;
}
/*
* GLIB_CHECK_VERSION(2,28,0) adds g_get_real_time which could minimize or
* replace this
*/
static void compute_elapsed(capture_file *cf, GTimeVal *start_time)
{
gdouble delta_time;
GTimeVal time_now;
g_get_current_time(&time_now);
delta_time = (time_now.tv_sec - start_time->tv_sec) * 1e6 +
time_now.tv_usec - start_time->tv_usec;
cf->computed_elapsed = (gulong) (delta_time / 1000); /* ms */
}
static const nstime_t *
ws_get_frame_ts(struct packet_provider_data *prov, guint32 frame_num)
{
if (prov->prev_dis && prov->prev_dis->num == frame_num)
return &prov->prev_dis->abs_ts;
if (prov->prev_cap && prov->prev_cap->num == frame_num)
return &prov->prev_cap->abs_ts;
if (prov->frames) {
frame_data *fd = frame_data_sequence_find(prov->frames, frame_num);
return (fd) ? &fd->abs_ts : NULL;
}
return NULL;
}
static epan_t *
ws_epan_new(capture_file *cf)
{
static const struct packet_provider_funcs funcs = {
ws_get_frame_ts,
cap_file_provider_get_interface_name,
cap_file_provider_get_interface_description,
cap_file_provider_get_user_comment
};
return epan_new(&cf->provider, &funcs);
}
cf_status_t
cf_open(capture_file *cf, const char *fname, unsigned int type, gboolean is_tempfile, int *err)
{
wtap *wth;
gchar *err_info;
wth = wtap_open_offline(fname, type, err, &err_info, TRUE);
if (wth == NULL)
goto fail;
/* The open succeeded. Close whatever capture file we had open,
and fill in the information for this file. */
cf_close(cf);
/* Initialize the record metadata. */
wtap_rec_init(&cf->rec);
/* XXX - we really want to initialize this after we've read all
the packets, so we know how much we'll ultimately need. */
ws_buffer_init(&cf->buf, 1500);
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
/* We're about to start reading the file. */
cf->state = FILE_READ_IN_PROGRESS;
cf->provider.wth = wth;
cf->f_datalen = 0;
/* Set the file name because we need it to set the follow stream filter.
XXX - is that still true? We need it for other reasons, though,
in any case. */
cf->filename = g_strdup(fname);
/* Indicate whether it's a permanent or temporary file. */
cf->is_tempfile = is_tempfile;
/* No user changes yet. */
cf->unsaved_changes = FALSE;
cf->computed_elapsed = 0;
cf->cd_t = wtap_file_type_subtype(cf->provider.wth);
cf->open_type = type;
cf->linktypes = g_array_sized_new(FALSE, FALSE, (guint) sizeof(int), 1);
cf->count = 0;
cf->packet_comment_count = 0;
cf->displayed_count = 0;
cf->marked_count = 0;
cf->ignored_count = 0;
cf->ref_time_count = 0;
cf->drops_known = FALSE;
cf->drops = 0;
cf->snap = wtap_snapshot_length(cf->provider.wth);
/* Allocate a frame_data_sequence for the frames in this file */
cf->provider.frames = new_frame_data_sequence();
nstime_set_zero(&cf->elapsed_time);
cf->provider.ref = NULL;
cf->provider.prev_dis = NULL;
cf->provider.prev_cap = NULL;
cf->cum_bytes = 0;
/* Create new epan session for dissection.
* (The old one was freed in cf_close().)
*/
cf->epan = ws_epan_new(cf);
packet_list_queue_draw();
cf_callback_invoke(cf_cb_file_opened, cf);
if (cf->cd_t == WTAP_FILE_TYPE_SUBTYPE_BER) {
/* tell the BER dissector the file name */
ber_set_filename(cf->filename);
}
wtap_set_cb_new_ipv4(cf->provider.wth, add_ipv4_name);
wtap_set_cb_new_ipv6(cf->provider.wth, (wtap_new_ipv6_callback_t) add_ipv6_name);
return CF_OK;
fail:
cfile_open_failure_alert_box(fname, *err, err_info);
return CF_ERROR;
}
/*
* Add an encapsulation type to cf->linktypes.
*/
static void
cf_add_encapsulation_type(capture_file *cf, int encap)
{
guint i;
for (i = 0; i < cf->linktypes->len; i++) {
if (g_array_index(cf->linktypes, gint, i) == encap)
return; /* it's already there */
}
/* It's not already there - add it. */
g_array_append_val(cf->linktypes, encap);
}
/* Reset everything to a pristine state */
void
cf_close(capture_file *cf)
{
cf->stop_flag = FALSE;
if (cf->state == FILE_CLOSED)
return; /* Nothing to do */
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
/* Die if we're in the middle of reading a file. */
g_assert(cf->state != FILE_READ_IN_PROGRESS);
g_assert(!cf->read_lock);
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
cf_callback_invoke(cf_cb_file_closing, cf);
/* close things, if not already closed before */
color_filters_cleanup();
if (cf->provider.wth) {
wtap_close(cf->provider.wth);
cf->provider.wth = NULL;
}
/* We have no file open... */
if (cf->filename != NULL) {
/* If it's a temporary file, remove it. */
if (cf->is_tempfile)
ws_unlink(cf->filename);
g_free(cf->filename);
cf->filename = NULL;
}
/* ...which means we have no changes to that file to save. */
cf->unsaved_changes = FALSE;
/* no open_routine type */
cf->open_type = WTAP_TYPE_AUTO;
/* Clean up the record metadata. */
wtap_rec_cleanup(&cf->rec);
/* Free up the packet buffer. */
ws_buffer_free(&cf->buf);
dfilter_free(cf->rfcode);
cf->rfcode = NULL;
if (cf->provider.frames != NULL) {
free_frame_data_sequence(cf->provider.frames);
cf->provider.frames = NULL;
}
if (cf->provider.frames_user_comments) {
g_tree_destroy(cf->provider.frames_user_comments);
cf->provider.frames_user_comments = NULL;
}
cf_unselect_packet(cf); /* nothing to select */
Store the frame_data structures in a tree, rather than a linked list. This lets us get rid of the per-frame_data-structure prev and next pointers, saving memory (at least according to Activity Monitor's report of the virtual address space size on my Snow Leopard machine, it's a noticeable saving), and lets us look up frame_data structures by frame number in O(log2(number of frames)) time rather than O(number of frames) time. It seems to take more CPU time when reading in the file, but seems to go from "finished reading in all the packets" to "displaying the packets" faster and seems to free up the frame_data structures faster when closing the file. It *is* doing more copying, currently, as we now don't allocate the frame_data structure until after the packet has passed the read filter, so that might account for the additional CPU time. (Oh, and, for what it's worth, on an LP64 platform, a frame_data structure is exactly 128 bytes long. However, there's more stuff to remove, so the power-of-2 size is not guaranteed to remain, and it's not a power-of-2 size on an ILP32 platform.) It also means we don't need GLib 2.10 or later for the two-pass mode in TShark. It also means some code in the TCP dissector that was checking pinfo->fd->next to see if it's NULL, in order to see if this is the last packet in the file, no longer works, but that wasn't guaranteed to work anyway: we might be doing a one-pass read through the capture in TShark; we might be dissecting the frame while we're reading in the packets for the first time in Wireshark; we might be doing a live capture in Wireshark; in which case packets might be prematurely considered "the last packet". #if 0 the no-longer-working tests, pending figuring out a better way of doing it. svn path=/trunk/; revision=36849
2011-04-25 19:01:05 +00:00
cf->first_displayed = 0;
cf->last_displayed = 0;
/* No frames, no frame selected, no field in that frame selected. */
cf->count = 0;
Store the frame_data structures in a tree, rather than a linked list. This lets us get rid of the per-frame_data-structure prev and next pointers, saving memory (at least according to Activity Monitor's report of the virtual address space size on my Snow Leopard machine, it's a noticeable saving), and lets us look up frame_data structures by frame number in O(log2(number of frames)) time rather than O(number of frames) time. It seems to take more CPU time when reading in the file, but seems to go from "finished reading in all the packets" to "displaying the packets" faster and seems to free up the frame_data structures faster when closing the file. It *is* doing more copying, currently, as we now don't allocate the frame_data structure until after the packet has passed the read filter, so that might account for the additional CPU time. (Oh, and, for what it's worth, on an LP64 platform, a frame_data structure is exactly 128 bytes long. However, there's more stuff to remove, so the power-of-2 size is not guaranteed to remain, and it's not a power-of-2 size on an ILP32 platform.) It also means we don't need GLib 2.10 or later for the two-pass mode in TShark. It also means some code in the TCP dissector that was checking pinfo->fd->next to see if it's NULL, in order to see if this is the last packet in the file, no longer works, but that wasn't guaranteed to work anyway: we might be doing a one-pass read through the capture in TShark; we might be dissecting the frame while we're reading in the packets for the first time in Wireshark; we might be doing a live capture in Wireshark; in which case packets might be prematurely considered "the last packet". #if 0 the no-longer-working tests, pending figuring out a better way of doing it. svn path=/trunk/; revision=36849
2011-04-25 19:01:05 +00:00
cf->current_frame = 0;
cf->current_row = 0;
cf->finfo_selected = NULL;
/* No frame link-layer types, either. */
if (cf->linktypes != NULL) {
g_array_free(cf->linktypes, TRUE);
cf->linktypes = NULL;
}
/* Clear the packet list. */
packet_list_freeze();
packet_list_clear();
packet_list_thaw();
cf->f_datalen = 0;
nstime_set_zero(&cf->elapsed_time);
reset_tap_listeners();
epan_free(cf->epan);
cf->epan = NULL;
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
/* We have no file open. */
cf->state = FILE_CLOSED;
cf_callback_invoke(cf_cb_file_closed, cf);
}
/*
* TRUE if the progress dialog doesn't exist and it looks like we'll
* take > 2s to load, FALSE otherwise.
*/
static inline gboolean
progress_is_slow(progdlg_t *progdlg, GTimer *prog_timer, gint64 size, gint64 pos)
{
double elapsed;
if (progdlg) return FALSE;
elapsed = g_timer_elapsed(prog_timer, NULL);
if ((elapsed / 2 > PROGBAR_SHOW_DELAY && (size / pos) > 2) /* It looks like we're going to be slow. */
|| elapsed > PROGBAR_SHOW_DELAY) { /* We are indeed slow. */
return TRUE;
}
return FALSE;
}
static float
calc_progbar_val(capture_file *cf, gint64 size, gint64 file_pos, gchar *status_str, gulong status_size)
{
float progbar_val;
progbar_val = (gfloat) file_pos / (gfloat) size;
if (progbar_val > 1.0) {
/* The file probably grew while we were reading it.
* Update file size, and try again.
*/
size = wtap_file_size(cf->provider.wth, NULL);
if (size >= 0)
progbar_val = (gfloat) file_pos / (gfloat) size;
/* If it's still > 1, either "wtap_file_size()" failed (in which
* case there's not much we can do about it), or the file
* *shrank* (in which case there's not much we can do about
* it); just clip the progress value at 1.0.
*/
if (progbar_val > 1.0f)
progbar_val = 1.0f;
}
g_snprintf(status_str, status_size,
"%" G_GINT64_MODIFIER "dKB of %" G_GINT64_MODIFIER "dKB",
file_pos / 1024, size / 1024);
return progbar_val;
}
cf_read_status_t
cf_read(capture_file *cf, gboolean reloading)
{
int err = 0;
gchar *err_info = NULL;
gchar *name_ptr;
progdlg_t *volatile progbar = NULL;
GTimer *prog_timer = g_timer_new();
GTimeVal start_time;
epan_dissect_t edt;
dfilter_t *dfcode;
volatile gboolean create_proto_tree;
guint tap_flags;
gboolean compiled;
volatile gboolean is_read_aborted = FALSE;
/* The update_progress_dlg call below might end up accepting a user request to
* trigger redissection/rescans which can modify/destroy the dissection
* context ("cf->epan"). That condition should be prevented by callers, but in
* case it occurs let's fail gracefully.
*/
if (cf->read_lock) {
g_warning("Failing due to recursive cf_read(\"%s\", %d) call!",
cf->filename, reloading);
return CF_READ_ERROR;
}
cf->read_lock = TRUE;
/* Compile the current display filter.
* We assume this will not fail since cf->dfilter is only set in
* cf_filter IFF the filter was valid.
*/
compiled = dfilter_compile(cf->dfilter, &dfcode, NULL);
g_assert(!cf->dfilter || (compiled && dfcode));
/* Get the union of the flags for all tap listeners. */
tap_flags = union_of_tap_listener_flags();
/*
* Determine whether we need to create a protocol tree.
* We do if:
*
* we're going to apply a display filter;
*
* one of the tap listeners is going to apply a filter;
*
* one of the tap listeners requires a protocol tree;
*
* a postdissector wants field values or protocols on
* the first pass.
*/
create_proto_tree =
(dfcode != NULL || have_filtering_tap_listeners() ||
(tap_flags & TL_REQUIRES_PROTO_TREE) || postdissectors_want_hfids());
reset_tap_listeners();
name_ptr = g_filename_display_basename(cf->filename);
if (reloading)
cf_callback_invoke(cf_cb_file_reload_started, cf);
else
cf_callback_invoke(cf_cb_file_read_started, cf);
/* Record whether the file is compressed.
XXX - do we know this at open time? */
cf->iscompressed = wtap_iscompressed(cf->provider.wth);
/* The packet list window will be empty until the file is completly loaded */
packet_list_freeze();
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
cf->stop_flag = FALSE;
g_get_current_time(&start_time);
epan_dissect_init(&edt, cf->epan, create_proto_tree, FALSE);
TRY {
int count = 0;
gint64 size;
gint64 file_pos;
gint64 data_offset;
float progbar_val;
gchar status_str[100];
column_info *cinfo;
/* If any tap listeners require the columns, construct them. */
cinfo = (tap_flags & TL_REQUIRES_COLUMNS) ? &cf->cinfo : NULL;
/* Find the size of the file. */
size = wtap_file_size(cf->provider.wth, NULL);
g_timer_start(prog_timer);
while ((wtap_read(cf->provider.wth, &err, &err_info, &data_offset))) {
if (size >= 0) {
count++;
file_pos = wtap_read_so_far(cf->provider.wth);
/* Create the progress bar if necessary. */
if (progress_is_slow(progbar, prog_timer, size, file_pos)) {
progbar_val = calc_progbar_val(cf, size, file_pos, status_str, sizeof(status_str));
if (reloading)
progbar = delayed_create_progress_dlg(cf->window, "Reloading", name_ptr,
TRUE, &cf->stop_flag, &start_time, progbar_val);
else
progbar = delayed_create_progress_dlg(cf->window, "Loading", name_ptr,
TRUE, &cf->stop_flag, &start_time, progbar_val);
}
/*
* Update the progress bar, but do it only after
* PROGBAR_UPDATE_INTERVAL has elapsed. Calling update_progress_dlg
* and packets_bar_update will likely trigger UI paint events, which
* might take a while depending on the platform and display. Reset
* our timer *after* painting.
*/
if (progbar && g_timer_elapsed(prog_timer, NULL) > PROGBAR_UPDATE_INTERVAL) {
progbar_val = calc_progbar_val(cf, size, file_pos, status_str, sizeof(status_str));
/* update the packet bar content on the first run or frequently on very large files */
update_progress_dlg(progbar, progbar_val, status_str);
compute_elapsed(cf, &start_time);
packets_bar_update();
g_timer_start(prog_timer);
}
/*
* The previous GUI triggers should not have destroyed the running
* session. If that did happen, it could blow up when read_record tries
* to use the destroyed edt.session, so detect it right here.
*/
g_assert(edt.session == cf->epan);
}
if (cf->state == FILE_READ_ABORTED) {
/* Well, the user decided to exit Wireshark. Break out of the
loop, and let the code below (which is called even if there
aren't any packets left to read) exit. */
is_read_aborted = TRUE;
break;
}
if (cf->stop_flag) {
/* Well, the user decided to abort the read. He/She will be warned and
it might be enough for him/her to work with the already loaded
packets.
This is especially true for very large capture files, where you don't
want to wait loading the whole file (which may last minutes or even
hours even on fast machines) just to see that it was the wrong file. */
break;
}
read_record(cf, dfcode, &edt, cinfo, data_offset);
}
}
CATCH(OutOfMemoryError) {
simple_message_box(ESD_TYPE_ERROR, NULL,
"More information and workarounds can be found at\n"
"https://wiki.wireshark.org/KnownBugs/OutOfMemory",
"Sorry, but Wireshark has run out of memory and has to terminate now.");
#if 0
/* Could we close the current capture and free up memory from that? */
#else
/* we have to terminate, as we cannot recover from the memory error */
exit(1);
#endif
}
ENDTRY;
/* Free the display name */
g_free(name_ptr);
/* Cleanup and release all dfilter resources */
dfilter_free(dfcode);
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
epan_dissect_cleanup(&edt);
/* We're done reading the file; destroy the progress bar if it was created. */
if (progbar != NULL)
destroy_progress_dlg(progbar);
g_timer_destroy(prog_timer);
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
/* We're done reading sequentially through the file. */
cf->state = FILE_READ_DONE;
/* Close the sequential I/O side, to free up memory it requires. */
wtap_sequential_close(cf->provider.wth);
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
/* Allow the protocol dissectors to free up memory that they
* don't need after the sequential run-through of the packets. */
postseq_cleanup_all_protocols();
/* compute the time it took to load the file */
compute_elapsed(cf, &start_time);
/* Set the file encapsulation type now; we don't know what it is until
we've looked at all the packets, as we don't know until then whether
there's more than one type (and thus whether it's
WTAP_ENCAP_PER_PACKET). */
cf->lnk_t = wtap_file_encap(cf->provider.wth);
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
cf->current_frame = frame_data_sequence_find(cf->provider.frames, cf->first_displayed);
cf->current_row = 0;
packet_list_thaw();
if (reloading)
cf_callback_invoke(cf_cb_file_reload_finished, cf);
else
cf_callback_invoke(cf_cb_file_read_finished, cf);
/* If we have any displayed packets to select, select the first of those
packets by making the first row the selected row. */
if (cf->first_displayed != 0) {
packet_list_select_first_row();
}
/* It is safe again to execute redissections. */
g_assert(cf->read_lock);
cf->read_lock = FALSE;
if (is_read_aborted) {
/*
* Well, the user decided to exit Wireshark while reading this *offline*
* capture file (Live captures are handled by something like
* cf_continue_tail). Clean up accordingly.
*/
cf_close(cf);
cf->redissection_queued = RESCAN_NONE;
return CF_READ_ABORTED;
}
if (cf->redissection_queued != RESCAN_NONE) {
/* Redissection was queued up. Clear the request and perform it now. */
gboolean redissect = cf->redissection_queued == RESCAN_REDISSECT;
rescan_packets(cf, "Reprocessing", "all packets", redissect);
}
if (cf->stop_flag) {
simple_message_box(ESD_TYPE_WARN, NULL,
"The remaining packets in the file were discarded.\n"
"\n"
"As a lot of packets from the original file will be missing,\n"
"remember to be careful when saving the current content to a file.\n",
"File loading was cancelled.");
return CF_READ_ERROR;
}
if (err != 0) {
/* Put up a message box noting that the read failed somewhere along
the line. Don't throw out the stuff we managed to read, though,
if any. */
cfile_read_failure_alert_box(NULL, err, err_info);
return CF_READ_ERROR;
} else
return CF_READ_OK;
}
#ifdef HAVE_LIBPCAP
cf_read_status_t
cf_continue_tail(capture_file *cf, volatile int to_read, int *err)
{
gchar *err_info;
volatile int newly_displayed_packets = 0;
dfilter_t *dfcode;
epan_dissect_t edt;
gboolean create_proto_tree;
guint tap_flags;
gboolean compiled;
/* Compile the current display filter.
* We assume this will not fail since cf->dfilter is only set in
* cf_filter IFF the filter was valid.
*/
compiled = dfilter_compile(cf->dfilter, &dfcode, NULL);
g_assert(!cf->dfilter || (compiled && dfcode));
/* Get the union of the flags for all tap listeners. */
tap_flags = union_of_tap_listener_flags();
/*
* Determine whether we need to create a protocol tree.
* We do if:
*
* we're going to apply a display filter;
*
* one of the tap listeners is going to apply a filter;
*
* one of the tap listeners requires a protocol tree;
*
* a postdissector wants field values or protocols on
* the first pass.
*/
create_proto_tree =
(dfcode != NULL || have_filtering_tap_listeners() ||
(tap_flags & TL_REQUIRES_PROTO_TREE) || postdissectors_want_hfids());
*err = 0;
/* Don't freeze/thaw the list when doing live capture */
/*packet_list_freeze();*/
/*g_log(NULL, G_LOG_LEVEL_MESSAGE, "cf_continue_tail: %u new: %u", cf->count, to_read);*/
epan_dissect_init(&edt, cf->epan, create_proto_tree, FALSE);
TRY {
gint64 data_offset = 0;
column_info *cinfo;
/* If any tap listeners require the columns, construct them. */
cinfo = (tap_flags & TL_REQUIRES_COLUMNS) ? &cf->cinfo : NULL;
while (to_read != 0) {
wtap_cleareof(cf->provider.wth);
if (!wtap_read(cf->provider.wth, err, &err_info, &data_offset)) {
break;
}
if (cf->state == FILE_READ_ABORTED) {
/* Well, the user decided to exit Wireshark. Break out of the
loop, and let the code below (which is called even if there
aren't any packets left to read) exit. */
break;
}
if (read_record(cf, dfcode, &edt, (column_info *) cinfo, data_offset)) {
newly_displayed_packets++;
}
to_read--;
}
}
CATCH(OutOfMemoryError) {
simple_message_box(ESD_TYPE_ERROR, NULL,
"More information and workarounds can be found at\n"
"https://wiki.wireshark.org/KnownBugs/OutOfMemory",
"Sorry, but Wireshark has run out of memory and has to terminate now.");
#if 0
/* Could we close the current capture and free up memory from that? */
return CF_READ_ABORTED;
#else
/* we have to terminate, as we cannot recover from the memory error */
exit(1);
#endif
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
}
ENDTRY;
/* Update the file encapsulation; it might have changed based on the
packets we've read. */
cf->lnk_t = wtap_file_encap(cf->provider.wth);
/* Cleanup and release all dfilter resources */
dfilter_free(dfcode);
epan_dissect_cleanup(&edt);
/*g_log(NULL, G_LOG_LEVEL_MESSAGE, "cf_continue_tail: count %u state: %u err: %u",
cf->count, cf->state, *err);*/
/* Don't freeze/thaw the list when doing live capture */
/*packet_list_thaw();*/
/* With the new packet list the first packet
* isn't automatically selected.
*/
if (!cf->current_frame)
packet_list_select_first_row();
/* moving to the end of the packet list - if the user requested so and
we have some new packets. */
if (newly_displayed_packets && cf->count != 0)
packet_list_moveto_end();
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
if (cf->state == FILE_READ_ABORTED) {
/* Well, the user decided to exit Wireshark. Return CF_READ_ABORTED
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
so that our caller can kill off the capture child process;
this will cause an EOF on the pipe from the child, so
"cf_finish_tail()" will be called, and it will clean up
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
and exit. */
return CF_READ_ABORTED;
} else if (*err != 0) {
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
/* We got an error reading the capture file.
XXX - pop up a dialog box instead? */
if (err_info != NULL) {
g_warning("Error \"%s\" while reading \"%s\" (\"%s\")",
wtap_strerror(*err), cf->filename, err_info);
g_free(err_info);
} else {
g_warning("Error \"%s\" while reading \"%s\"",
wtap_strerror(*err), cf->filename);
}
return CF_READ_ERROR;
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
} else
return CF_READ_OK;
}
void
cf_fake_continue_tail(capture_file *cf) {
cf->state = FILE_READ_DONE;
}
cf_read_status_t
cf_finish_tail(capture_file *cf, int *err)
{
gchar *err_info;
gint64 data_offset;
dfilter_t *dfcode;
column_info *cinfo;
epan_dissect_t edt;
gboolean create_proto_tree;
guint tap_flags;
gboolean compiled;
/* Compile the current display filter.
* We assume this will not fail since cf->dfilter is only set in
* cf_filter IFF the filter was valid.
*/
compiled = dfilter_compile(cf->dfilter, &dfcode, NULL);
g_assert(!cf->dfilter || (compiled && dfcode));
/* Get the union of the flags for all tap listeners. */
tap_flags = union_of_tap_listener_flags();
/* If any tap listeners require the columns, construct them. */
cinfo = (tap_flags & TL_REQUIRES_COLUMNS) ? &cf->cinfo : NULL;
/*
* Determine whether we need to create a protocol tree.
* We do if:
*
* we're going to apply a display filter;
*
* one of the tap listeners is going to apply a filter;
*
* one of the tap listeners requires a protocol tree;
*
* a postdissector wants field values or protocols on
* the first pass.
*/
create_proto_tree =
(dfcode != NULL || have_filtering_tap_listeners() ||
(tap_flags & TL_REQUIRES_PROTO_TREE) || postdissectors_want_hfids());
if (cf->provider.wth == NULL) {
cf_close(cf);
return CF_READ_ERROR;
}
/* Don't freeze/thaw the list when doing live capture */
/*packet_list_freeze();*/
epan_dissect_init(&edt, cf->epan, create_proto_tree, FALSE);
while ((wtap_read(cf->provider.wth, err, &err_info, &data_offset))) {
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
if (cf->state == FILE_READ_ABORTED) {
/* Well, the user decided to abort the read. Break out of the
loop, and let the code below (which is called even if there
aren't any packets left to read) exit. */
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
break;
}
read_record(cf, dfcode, &edt, cinfo, data_offset);
}
/* Cleanup and release all dfilter resources */
dfilter_free(dfcode);
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
epan_dissect_cleanup(&edt);
/* Don't freeze/thaw the list when doing live capture */
/*packet_list_thaw();*/
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
if (cf->state == FILE_READ_ABORTED) {
/* Well, the user decided to abort the read. We're only called
when the child capture process closes the pipe to us (meaning
it's probably exited), so we can just close the capture
file; we return CF_READ_ABORTED so our caller can do whatever
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
is appropriate when that happens. */
cf_close(cf);
return CF_READ_ABORTED;
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
}
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
/* We're done reading sequentially through the file. */
cf->state = FILE_READ_DONE;
/* We're done reading sequentially through the file; close the
sequential I/O side, to free up memory it requires. */
wtap_sequential_close(cf->provider.wth);
/* Allow the protocol dissectors to free up memory that they
* don't need after the sequential run-through of the packets. */
postseq_cleanup_all_protocols();
/* Update the file encapsulation; it might have changed based on the
packets we've read. */
cf->lnk_t = wtap_file_encap(cf->provider.wth);
/* Update the details in the file-set dialog, as the capture file
* has likely grown since we first stat-ed it */
fileset_update_file(cf->filename);
if (*err != 0) {
Add routines to Wiretap to allow a client of Wiretap to get: a pointer to the "wtap_pkthdr" structure for an open capture file; a pointer to the "wtap_pseudo_header" union for an open capture file; a pointer to the packet buffer for an open capture file; so that a program using "wtap_read()" in a loop can get at those items. Keep, in a "capture_file" structure, an indicator of whether: no file is open; a file is open, and being read; a file is open, and is being read, but the user tried to quit out of reading the file (e.g., by doing "File/Quit"); a file is open, and has been completely read. Abort if we try to close a capture that's being read if the user hasn't tried to quit out of the read. Have "File/Quit" check if a file is being read; if so, just set the state indicator to "user tried to quit out of it", so that the code reading the file can do what's appropriate to clean up, rather than closing the file out from under that code and causing crashes. Have "read_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, close the capture and return an indication that the read was aborted by the user. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "continue_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have "finish_tail_cap_file()" read the capture file with a loop using "wtap_read()", rather than by using "wtap_loop()"; have it check after reading each packet whether the user tried to abort the read and, if so, quit the loop, and after the loop finishes (even if it read no packets), close the capture and return an indication that the read was aborted by the user if that happened. Otherwise, return an indication of whether the read completely succeeded or failed in the middle (and, if it failed, return the error code through a pointer). Have their callers check whether the read was aborted or not and, if it was, bail out in the appropriate fashion (exit if it's reading a file specified by "-r" on the command line; exit the main loop if it's reading a file specified with File->Open; kill the capture child if it's "continue_tail_cap_file()"; exit the main loop if it's "finish_tail_cap_file()". svn path=/trunk/; revision=2095
2000-06-27 07:13:42 +00:00
/* We got an error reading the capture file.
XXX - pop up a dialog box? */
if (err_info != NULL) {
g_warning("Error \"%s\" while reading \"%s\" (\"%s\")",
wtap_strerror(*err), cf->filename, err_info);
g_free(err_info);
} else {
g_warning("Error \"%s\" while reading \"%s\"",
wtap_strerror(*err), cf->filename);
}
return CF_READ_ERROR;
} else {
return CF_READ_OK;
}
}
#endif /* HAVE_LIBPCAP */
gchar *
cf_get_display_name(capture_file *cf)
{
gchar *displayname;
/* Return a name to use in displays */
if (!cf->is_tempfile) {
/* Get the last component of the file name, and use that. */
if (cf->filename) {
displayname = g_filename_display_basename(cf->filename);
} else {
displayname=g_strdup("(No file)");
}
} else {
/* The file we read is a temporary file from a live capture or
a merge operation; we don't mention its name, but, if it's
from a capture, give the source of the capture. */
if (cf->source) {
displayname = g_strdup(cf->source);
} else {
displayname = g_strdup("(Untitled)");
}
}
return displayname;
}
gchar *
cf_get_basename(capture_file *cf)
{
gchar *displayname;
/* Return a name to use in the GUI for the basename for files to
which we save statistics */
if (!cf->is_tempfile) {
/* Get the last component of the file name, and use that. */
if (cf->filename) {
displayname = g_filename_display_basename(cf->filename);
/* If the file name ends with any extension that corresponds
to a file type we support - including compressed versions
of those files - strip it off. */
size_t displayname_len = strlen(displayname);
GSList *extensions = wtap_get_all_file_extensions_list();
GSList *suffix;
for (suffix = extensions; suffix != NULL; suffix = g_slist_next(suffix)) {
/* Does the file name end with that extension? */
const char *extension = (char *)suffix->data;
size_t extension_len = strlen(extension);
if (displayname_len > extension_len &&
displayname[displayname_len - extension_len - 1] == '.' &&
strcmp(&displayname[displayname_len - extension_len], extension) == 0) {
/* Yes. Strip the extension off, and return the result. */
displayname[displayname_len - extension_len - 1] = '\0';
break;
}
}
wtap_free_extensions_list(extensions);
} else {
displayname=g_strdup("");
}
} else {
/* The file we read is a temporary file from a live capture or
a merge operation; we don't mention its name, but, if it's
from a capture, give the source of the capture. */
if (cf->source) {
displayname = g_strdup(cf->source);
} else {
displayname = g_strdup("");
}
}
return displayname;
}
void cf_set_tempfile_source(capture_file *cf, gchar *source) {
if (cf->source) {
g_free(cf->source);
}
if (source) {
cf->source = g_strdup(source);
} else {
cf->source = g_strdup("");
}
}
const gchar *cf_get_tempfile_source(capture_file *cf) {
if (!cf->source) {
return "";
}
return cf->source;
}
/* XXX - use a macro instead? */
int
cf_get_packet_count(capture_file *cf)
{
return cf->count;
}
/* XXX - use a macro instead? */
gboolean
cf_is_tempfile(capture_file *cf)
{
return cf->is_tempfile;
}
void cf_set_tempfile(capture_file *cf, gboolean is_tempfile)
{
cf->is_tempfile = is_tempfile;
}
/* XXX - use a macro instead? */
void cf_set_drops_known(capture_file *cf, gboolean drops_known)
{
cf->drops_known = drops_known;
}
/* XXX - use a macro instead? */
void cf_set_drops(capture_file *cf, guint32 drops)
{
cf->drops = drops;
}
/* XXX - use a macro instead? */
gboolean cf_get_drops_known(capture_file *cf)
{
return cf->drops_known;
}
/* XXX - use a macro instead? */
guint32 cf_get_drops(capture_file *cf)
{
return cf->drops;
}
void cf_set_rfcode(capture_file *cf, dfilter_t *rfcode)
{
cf->rfcode = rfcode;
}
static void
add_packet_to_packet_list(frame_data *fdata, capture_file *cf,
epan_dissect_t *edt, dfilter_t *dfcode, column_info *cinfo,
wtap_rec *rec, const guint8 *buf, gboolean add_to_packet_list)
{
frame_data_set_before_dissect(fdata, &cf->elapsed_time,
&cf->provider.ref, cf->provider.prev_dis);
cf->provider.prev_cap = fdata;
if (dfcode != NULL) {
epan_dissect_prime_with_dfilter(edt, dfcode);
}
#if 0
/* Prepare coloring rules, this ensures that display filter rules containing
* frame.color_rule references are still processed.
* TODO: actually detect that situation or maybe apply other optimizations? */
if (edt->tree && color_filters_used()) {
color_filters_prime_edt(edt);
fdata->flags.need_colorize = 1;
}
#endif
if (!fdata->flags.visited) {
/* This is the first pass, so prime the epan_dissect_t with the
hfids postdissectors want on the first pass. */
prime_epan_dissect_with_postdissector_wanted_hfids(edt);
}
/* Dissect the frame. */
epan_dissect_run_with_taps(edt, cf->cd_t, rec,
frame_tvbuff_new(&cf->provider, fdata, buf),
fdata, cinfo);
/* If we don't have a display filter, set "passed_dfilter" to 1. */
if (dfcode != NULL) {
fdata->flags.passed_dfilter = dfilter_apply_edt(dfcode, edt) ? 1 : 0;
Fix https://bugs.wireshark.org/bugzilla/show_bug.cgi?id=3315 - make Save-As/Displayed/All-Packets save not only the displayed packets but also any other packets needed (e.g., for reassembly) to fully dissect the displayed packets. This works only for the "All packets" case; choosing only the Selected packet, the Marked packets, or a range of packets would require actually storing which packets depend on which (too much memory) or going through the packet list many times (too slow). Also, this behavior is always the case: you can't save the displayed packets without their dependencies (I don't see why this would be desirable). So far this is done for SCTP and things using the reassembly routines (TCP has been tested). The Win32 dialog was modified but hasn't been tested yet. One confusing aspect of the UI is that the Displayed count in the Save-As dialog does not match the number of displayed packets. (I tried renaming the button "Displayed + Dependencies" but it looked too big.) The tooltip tries to explain this and the fact that this works only in the All-Packets case; suggestions for improvement are welcome. Implementation details: Dissectors (or the reassembly code) can list frames which were needed to build the current frame's tree. If the current frame passes the display filter then each listed frame is marked as "depended upon" (this takes up the last free frame_data flag). When performing a Save-As/Displayed/All-Packets then choose packets which passed the dfilter _or_ are depended upon. svn path=/trunk/; revision=41216
2012-02-28 03:19:49 +00:00
if (fdata->flags.passed_dfilter) {
/* This frame passed the display filter but it may depend on other
* (potentially not displayed) frames. Find those frames and mark them
* as depended upon.
*/
g_slist_foreach(edt->pi.dependent_frames, find_and_mark_frame_depended_upon, cf->provider.frames);
}
} else
fdata->flags.passed_dfilter = 1;
if (fdata->flags.passed_dfilter || fdata->flags.ref_time)
cf->displayed_count++;
if (add_to_packet_list) {
/* We fill the needed columns from new_packet_list */
packet_list_append(cinfo, fdata);
}
if (fdata->flags.passed_dfilter || fdata->flags.ref_time)
{
frame_data_set_after_dissect(fdata, &cf->cum_bytes);
cf->provider.prev_dis = fdata;
/* If we haven't yet seen the first frame, this is it. */
Store the frame_data structures in a tree, rather than a linked list. This lets us get rid of the per-frame_data-structure prev and next pointers, saving memory (at least according to Activity Monitor's report of the virtual address space size on my Snow Leopard machine, it's a noticeable saving), and lets us look up frame_data structures by frame number in O(log2(number of frames)) time rather than O(number of frames) time. It seems to take more CPU time when reading in the file, but seems to go from "finished reading in all the packets" to "displaying the packets" faster and seems to free up the frame_data structures faster when closing the file. It *is* doing more copying, currently, as we now don't allocate the frame_data structure until after the packet has passed the read filter, so that might account for the additional CPU time. (Oh, and, for what it's worth, on an LP64 platform, a frame_data structure is exactly 128 bytes long. However, there's more stuff to remove, so the power-of-2 size is not guaranteed to remain, and it's not a power-of-2 size on an ILP32 platform.) It also means we don't need GLib 2.10 or later for the two-pass mode in TShark. It also means some code in the TCP dissector that was checking pinfo->fd->next to see if it's NULL, in order to see if this is the last packet in the file, no longer works, but that wasn't guaranteed to work anyway: we might be doing a one-pass read through the capture in TShark; we might be dissecting the frame while we're reading in the packets for the first time in Wireshark; we might be doing a live capture in Wireshark; in which case packets might be prematurely considered "the last packet". #if 0 the no-longer-working tests, pending figuring out a better way of doing it. svn path=/trunk/; revision=36849
2011-04-25 19:01:05 +00:00
if (cf->first_displayed == 0)
cf->first_displayed = fdata->num;
/* This is the last frame we've seen so far. */
Store the frame_data structures in a tree, rather than a linked list. This lets us get rid of the per-frame_data-structure prev and next pointers, saving memory (at least according to Activity Monitor's report of the virtual address space size on my Snow Leopard machine, it's a noticeable saving), and lets us look up frame_data structures by frame number in O(log2(number of frames)) time rather than O(number of frames) time. It seems to take more CPU time when reading in the file, but seems to go from "finished reading in all the packets" to "displaying the packets" faster and seems to free up the frame_data structures faster when closing the file. It *is* doing more copying, currently, as we now don't allocate the frame_data structure until after the packet has passed the read filter, so that might account for the additional CPU time. (Oh, and, for what it's worth, on an LP64 platform, a frame_data structure is exactly 128 bytes long. However, there's more stuff to remove, so the power-of-2 size is not guaranteed to remain, and it's not a power-of-2 size on an ILP32 platform.) It also means we don't need GLib 2.10 or later for the two-pass mode in TShark. It also means some code in the TCP dissector that was checking pinfo->fd->next to see if it's NULL, in order to see if this is the last packet in the file, no longer works, but that wasn't guaranteed to work anyway: we might be doing a one-pass read through the capture in TShark; we might be dissecting the frame while we're reading in the packets for the first time in Wireshark; we might be doing a live capture in Wireshark; in which case packets might be prematurely considered "the last packet". #if 0 the no-longer-working tests, pending figuring out a better way of doing it. svn path=/trunk/; revision=36849
2011-04-25 19:01:05 +00:00
cf->last_displayed = fdata->num;
}
epan_dissect_reset(edt);
}
/*
* Read in a new record.
* Returns TRUE if the packet was added to the packet (record) list,
* FALSE otherwise.
*/
static gboolean
read_record(capture_file *cf, dfilter_t *dfcode, epan_dissect_t *edt,
column_info *cinfo, gint64 offset)
We can't trust "cf->current_frame" to refer to the frame that was selected before we started re-colorizing or re-filtering the display, as when the first row is added to the clist, that may be selected and thus made the current frame. This means that we can't find the row corresponding to the previously-selected frame, if any, by checking as each packet is colorized/filtered and see whether its "frame_data" structure is equal to "cf->current_frame", as that'll always say that the first frame in the display is the selected frame. Instead, we recored the value of "cf->current_frame" before we do anything to the clist, have "add_packet_to_packet_list()" return either the row number of the frame (if it passed the filter and thus was added to the clist) or -1 (if it didn't pass the filter and thus wasn't added to the clist), and, after "add_packet_to_packet_list()", if the current frame is the one that was the selected row, remember its row number (if any), and, when we're finished colorizing/filtering the display, make that row the current row if it's not -1 (-1 means that the selected row didn't pass the filter). Also, don't do that until after we've thawed the clist, as the vertical adjustment for the clist doesn't reflect reality until then, and attempting to go to a given row won't work right until the vertical adjustment for the clist reflects reality. Shove all the code to set the selected and focus rows, and to make said row visible, into a routine, so the "Find Frame" and "Go To Frame" code can use it as well. svn path=/trunk/; revision=1959
2000-05-15 01:50:16 +00:00
{
wtap_rec *rec = wtap_get_rec(cf->provider.wth);
const guint8 *buf = wtap_get_buf_ptr(cf->provider.wth);
Store the frame_data structures in a tree, rather than a linked list. This lets us get rid of the per-frame_data-structure prev and next pointers, saving memory (at least according to Activity Monitor's report of the virtual address space size on my Snow Leopard machine, it's a noticeable saving), and lets us look up frame_data structures by frame number in O(log2(number of frames)) time rather than O(number of frames) time. It seems to take more CPU time when reading in the file, but seems to go from "finished reading in all the packets" to "displaying the packets" faster and seems to free up the frame_data structures faster when closing the file. It *is* doing more copying, currently, as we now don't allocate the frame_data structure until after the packet has passed the read filter, so that might account for the additional CPU time. (Oh, and, for what it's worth, on an LP64 platform, a frame_data structure is exactly 128 bytes long. However, there's more stuff to remove, so the power-of-2 size is not guaranteed to remain, and it's not a power-of-2 size on an ILP32 platform.) It also means we don't need GLib 2.10 or later for the two-pass mode in TShark. It also means some code in the TCP dissector that was checking pinfo->fd->next to see if it's NULL, in order to see if this is the last packet in the file, no longer works, but that wasn't guaranteed to work anyway: we might be doing a one-pass read through the capture in TShark; we might be dissecting the frame while we're reading in the packets for the first time in Wireshark; we might be doing a live capture in Wireshark; in which case packets might be prematurely considered "the last packet". #if 0 the no-longer-working tests, pending figuring out a better way of doing it. svn path=/trunk/; revision=36849
2011-04-25 19:01:05 +00:00
frame_data fdlocal;
frame_data *fdata;
gboolean passed = TRUE;
gboolean added = FALSE;
/* Add this packet's link-layer encapsulation type to cf->linktypes, if
it's not already there.
XXX - yes, this is O(N), so if every packet had a different
link-layer encapsulation type, it'd be O(N^2) to read the file, but
there are probably going to be a small number of encapsulation types
in a file. */
if (rec->rec_type == REC_TYPE_PACKET) {
cf_add_encapsulation_type(cf, rec->rec_header.packet_header.pkt_encap);
}
/* The frame number of this packet, if we add it to the set of frames,
would be one more than the count of frames in the file so far. */
frame_data_init(&fdlocal, cf->count + 1, rec, offset, cf->cum_bytes);
if (cf->rfcode) {
epan_dissect_t rf_edt;
epan_dissect_init(&rf_edt, cf->epan, TRUE, FALSE);
epan_dissect_prime_with_dfilter(&rf_edt, cf->rfcode);
epan_dissect_run(&rf_edt, cf->cd_t, rec,
frame_tvbuff_new(&cf->provider, &fdlocal, buf),
&fdlocal, NULL);
passed = dfilter_apply_edt(cf->rfcode, &rf_edt);
epan_dissect_cleanup(&rf_edt);
}
if (passed) {
added = TRUE;
Store the frame_data structures in a tree, rather than a linked list. This lets us get rid of the per-frame_data-structure prev and next pointers, saving memory (at least according to Activity Monitor's report of the virtual address space size on my Snow Leopard machine, it's a noticeable saving), and lets us look up frame_data structures by frame number in O(log2(number of frames)) time rather than O(number of frames) time. It seems to take more CPU time when reading in the file, but seems to go from "finished reading in all the packets" to "displaying the packets" faster and seems to free up the frame_data structures faster when closing the file. It *is* doing more copying, currently, as we now don't allocate the frame_data structure until after the packet has passed the read filter, so that might account for the additional CPU time. (Oh, and, for what it's worth, on an LP64 platform, a frame_data structure is exactly 128 bytes long. However, there's more stuff to remove, so the power-of-2 size is not guaranteed to remain, and it's not a power-of-2 size on an ILP32 platform.) It also means we don't need GLib 2.10 or later for the two-pass mode in TShark. It also means some code in the TCP dissector that was checking pinfo->fd->next to see if it's NULL, in order to see if this is the last packet in the file, no longer works, but that wasn't guaranteed to work anyway: we might be doing a one-pass read through the capture in TShark; we might be dissecting the frame while we're reading in the packets for the first time in Wireshark; we might be doing a live capture in Wireshark; in which case packets might be prematurely considered "the last packet". #if 0 the no-longer-working tests, pending figuring out a better way of doing it. svn path=/trunk/; revision=36849
2011-04-25 19:01:05 +00:00
/* This does a shallow copy of fdlocal, which is good enough. */
fdata = frame_data_sequence_add(cf->provider.frames, &fdlocal);
cf->count++;
if (rec->opt_comment != NULL)
cf->packet_comment_count++;
Store the frame_data structures in a tree, rather than a linked list. This lets us get rid of the per-frame_data-structure prev and next pointers, saving memory (at least according to Activity Monitor's report of the virtual address space size on my Snow Leopard machine, it's a noticeable saving), and lets us look up frame_data structures by frame number in O(log2(number of frames)) time rather than O(number of frames) time. It seems to take more CPU time when reading in the file, but seems to go from "finished reading in all the packets" to "displaying the packets" faster and seems to free up the frame_data structures faster when closing the file. It *is* doing more copying, currently, as we now don't allocate the frame_data structure until after the packet has passed the read filter, so that might account for the additional CPU time. (Oh, and, for what it's worth, on an LP64 platform, a frame_data structure is exactly 128 bytes long. However, there's more stuff to remove, so the power-of-2 size is not guaranteed to remain, and it's not a power-of-2 size on an ILP32 platform.) It also means we don't need GLib 2.10 or later for the two-pass mode in TShark. It also means some code in the TCP dissector that was checking pinfo->fd->next to see if it's NULL, in order to see if this is the last packet in the file, no longer works, but that wasn't guaranteed to work anyway: we might be doing a one-pass read through the capture in TShark; we might be dissecting the frame while we're reading in the packets for the first time in Wireshark; we might be doing a live capture in Wireshark; in which case packets might be prematurely considered "the last packet". #if 0 the no-longer-working tests, pending figuring out a better way of doing it. svn path=/trunk/; revision=36849
2011-04-25 19:01:05 +00:00
cf->f_datalen = offset + fdlocal.cap_len;
/* When a redissection is in progress (or queued), do not process packets.
* This will be done once all (new) packets have been scanned. */
if (!cf->redissecting && cf->redissection_queued == RESCAN_NONE) {
add_packet_to_packet_list(fdata, cf, edt, dfcode,
cinfo, rec, buf, TRUE);
}
}
return added;
}
typedef struct _callback_data_t {
gpointer pd_window;
gint64 f_len;
GTimeVal start_time;
progdlg_t *progbar;
GTimer *prog_timer;
gboolean stop_flag;
} callback_data_t;
static gboolean
merge_callback(merge_event event, int num _U_,
const merge_in_file_t in_files[], const guint in_file_count,
void *data)
{
guint i;
callback_data_t *cb_data = (callback_data_t*) data;
g_assert(cb_data != NULL);
switch (event) {
case MERGE_EVENT_INPUT_FILES_OPENED:
/* do nothing */
break;
case MERGE_EVENT_FRAME_TYPE_SELECTED:
/* do nothing */
break;
case MERGE_EVENT_READY_TO_MERGE:
/* Get the sum of the sizes of all the files. */
for (i = 0; i < in_file_count; i++)
cb_data->f_len += in_files[i].size;
cb_data->prog_timer = g_timer_new();
g_timer_start(cb_data->prog_timer);
g_get_current_time(&cb_data->start_time);
break;
case MERGE_EVENT_RECORD_WAS_READ:
{
/* Create the progress bar if necessary.
We check on every iteration of the loop, so that it takes no
longer than the standard time to create it (otherwise, for a
large file, we might take considerably longer than that standard
time in order to get to the next progress bar step). */
if (cb_data->progbar == NULL) {
cb_data->progbar = delayed_create_progress_dlg(cb_data->pd_window, "Merging", "files",
FALSE, &cb_data->stop_flag, &cb_data->start_time, 0.0f);
}
/*
* Update the progress bar, but do it only after
* PROGBAR_UPDATE_INTERVAL has elapsed. Calling update_progress_dlg
* and packets_bar_update will likely trigger UI paint events, which
* might take a while depending on the platform and display. Reset
* our timer *after* painting.
*/
if (g_timer_elapsed(cb_data->prog_timer, NULL) > PROGBAR_UPDATE_INTERVAL) {
float progbar_val;
gint64 file_pos = 0;
/* Get the sum of the seek positions in all of the files. */
for (i = 0; i < in_file_count; i++)
file_pos += wtap_read_so_far(in_files[i].wth);
progbar_val = (gfloat) file_pos / (gfloat) cb_data->f_len;
if (progbar_val > 1.0f) {
/* Some file probably grew while we were reading it.
That "shouldn't happen", so we'll just clip the progress
value at 1.0. */
progbar_val = 1.0f;
}
if (cb_data->progbar != NULL) {
gchar status_str[100];
g_snprintf(status_str, sizeof(status_str),
"%" G_GINT64_MODIFIER "dKB of %" G_GINT64_MODIFIER "dKB",
file_pos / 1024, cb_data->f_len / 1024);
update_progress_dlg(cb_data->progbar, progbar_val, status_str);
}
g_timer_start(cb_data->prog_timer);
}
}
break;
case MERGE_EVENT_DONE:
/* We're done merging the files; destroy the progress bar if it was created. */
if (cb_data->progbar != NULL)
destroy_progress_dlg(cb_data->progbar);
g_timer_destroy(cb_data->prog_timer);
break;
}
return cb_data->stop_flag;
}
cf_status_t
cf_merge_files_to_tempfile(gpointer pd_window, char **out_filenamep,
int in_file_count, char *const *in_filenames,
int file_type, gboolean do_append)
{
int err = 0;
gchar *err_info = NULL;
guint err_fileno;
guint32 err_framenum;
merge_result status;
merge_progress_callback_t cb;
callback_data_t *cb_data = g_new0(callback_data_t, 1);
/* prepare our callback routine */
cb_data->pd_window = pd_window;
cb.callback_func = merge_callback;
cb.data = cb_data;
cf_callback_invoke(cf_cb_file_merge_started, NULL);
/* merge the files */
status = merge_files_to_tempfile(out_filenamep, "wireshark", file_type,
(const char *const *) in_filenames,
in_file_count, do_append,
IDB_MERGE_MODE_ALL_SAME, 0 /* snaplen */,
"Wireshark", &cb, &err, &err_info,
&err_fileno, &err_framenum);
g_free(cb.data);
switch (status) {
case MERGE_OK:
break;
case MERGE_USER_ABORTED:
/* this isn't really an error, though we will return CF_ERROR later */
break;
case MERGE_ERR_CANT_OPEN_INFILE:
cfile_open_failure_alert_box(in_filenames[err_fileno], err, err_info);
break;
case MERGE_ERR_CANT_OPEN_OUTFILE:
cfile_dump_open_failure_alert_box(*out_filenamep, err, file_type);
break;
case MERGE_ERR_CANT_READ_INFILE:
cfile_read_failure_alert_box(in_filenames[err_fileno], err, err_info);
break;
case MERGE_ERR_BAD_PHDR_INTERFACE_ID:
simple_error_message_box("Record %u of \"%s\" has an interface ID that does not match any IDB in its file.",
err_framenum, in_filenames[err_fileno]);
break;
case MERGE_ERR_CANT_WRITE_OUTFILE:
cfile_write_failure_alert_box(in_filenames[err_fileno],
*out_filenamep, err, err_info,
err_framenum, file_type);
break;
case MERGE_ERR_CANT_CLOSE_OUTFILE:
cfile_close_failure_alert_box(*out_filenamep, err);
break;
default:
simple_error_message_box("Unknown merge_files error %d", status);
break;
}
cf_callback_invoke(cf_cb_file_merge_finished, NULL);
if (status != MERGE_OK) {
/* Callers aren't expected to treat an error or an explicit abort
differently - we put up error dialogs ourselves, so they don't
have to. */
return CF_ERROR;
} else
return CF_OK;
}
cf_status_t
cf_filter_packets(capture_file *cf, gchar *dftext, gboolean force)
{
const char *filter_new = dftext ? dftext : "";
const char *filter_old = cf->dfilter ? cf->dfilter : "";
dfilter_t *dfcode;
gchar *err_msg;
GTimeVal start_time;
/* if new filter equals old one, do nothing unless told to do so */
if (!force && strcmp(filter_new, filter_old) == 0) {
return CF_OK;
}
dfcode=NULL;
if (dftext == NULL) {
/* The new filter is an empty filter (i.e., display all packets).
* so leave dfcode==NULL
*/
} else {
/*
Clean up the handling of filter strings: have "filter_packets()" make a copy of the filter string handed to it, as it may save the filter string in the "capture_file" structure, and the caller of "filter_packets()" shouldn't have to worry about the string it passed to "filter_packets()" being stashed away somewhere so that it can't just free that string or change it; have callers of "filter_packets()" free up the string they handed to it, if the string was allocated and they're done with it; plug some memory leaks in "match_selected_cb_do()". Check for an illegal "action" argument being passed to "match_selected_cb_do()". Move some keys out of "keys.h" into "gtk/main.c", as they're only used in "gtk/main.c". Make the pointer to the filter list a data item for the combo box, as it's a copy of the list of strings for the combo box, rather than attaching it to the widgets that activate the filter (a pointer to the combo box *itself* is a data item for those widgets). In "filter_activate_cb()", make a copy of the text from the text entry field as soon as we fetch it, and use that copy. Free that copy if we didn't add the filter to the filter list. Don't make a copy of the entire filter list and use that to set the combo box's list of items - just use the list itself. Also, when the list is changed, make the new value the data for the combo box (the list pointer will actually not be changed, because we happen to be using "g_list_append()", but let's not rely on that). svn path=/trunk/; revision=5368
2002-05-03 03:24:47 +00:00
* We have a filter; make a copy of it (as we'll be saving it),
* and try to compile it.
*/
Clean up the handling of filter strings: have "filter_packets()" make a copy of the filter string handed to it, as it may save the filter string in the "capture_file" structure, and the caller of "filter_packets()" shouldn't have to worry about the string it passed to "filter_packets()" being stashed away somewhere so that it can't just free that string or change it; have callers of "filter_packets()" free up the string they handed to it, if the string was allocated and they're done with it; plug some memory leaks in "match_selected_cb_do()". Check for an illegal "action" argument being passed to "match_selected_cb_do()". Move some keys out of "keys.h" into "gtk/main.c", as they're only used in "gtk/main.c". Make the pointer to the filter list a data item for the combo box, as it's a copy of the list of strings for the combo box, rather than attaching it to the widgets that activate the filter (a pointer to the combo box *itself* is a data item for those widgets). In "filter_activate_cb()", make a copy of the text from the text entry field as soon as we fetch it, and use that copy. Free that copy if we didn't add the filter to the filter list. Don't make a copy of the entire filter list and use that to set the combo box's list of items - just use the list itself. Also, when the list is changed, make the new value the data for the combo box (the list pointer will actually not be changed, because we happen to be using "g_list_append()", but let's not rely on that). svn path=/trunk/; revision=5368
2002-05-03 03:24:47 +00:00
dftext = g_strdup(dftext);
if (!dfilter_compile(dftext, &dfcode, &err_msg)) {
/* The attempt failed; report an error. */
simple_message_box(ESD_TYPE_ERROR, NULL,
"See the help for a description of the display filter syntax.",
"\"%s\" isn't a valid display filter: %s",
dftext, err_msg);
g_free(err_msg);
g_free(dftext);
return CF_ERROR;
}
/* Was it empty? */
if (dfcode == NULL) {
/* Yes - free the filter text, and set it to null. */
g_free(dftext);
dftext = NULL;
}
}
/* We have a valid filter. Replace the current filter. */
g_free(cf->dfilter);
cf->dfilter = dftext;
g_get_current_time(&start_time);
/* Now rescan the packet list, applying the new filter, but not
* throwing away information constructed on a previous pass.
* If a dissection is already in progress, queue it.
*/
if (cf->redissection_queued == RESCAN_NONE) {
if (cf->read_lock) {
cf->redissection_queued = RESCAN_SCAN;
} else if (cf->state != FILE_CLOSED) {
if (dftext == NULL) {
rescan_packets(cf, "Resetting", "Filter", FALSE);
} else {
rescan_packets(cf, "Filtering", dftext, FALSE);
}
}
}
/* Cleanup and release all dfilter resources */
dfilter_free(dfcode);
return CF_OK;
}
void
cf_reftime_packets(capture_file *cf)
{
ref_time_packets(cf);
}
Turn the code of "colorize_packet()" into a static routine that is given a word to use in the progress dialog, and a flag indicating whether the display filter is to be reevaluated or not, and: have "colorize_packet()" call that routine with "Colorizing" and FALSE as those arguments; have the filtering code call that routine with "Filtering" and TRUE as those arguments; add an exported routine to call that routine with "Reprocessing" and TRUE as those arguments, to use to re-generate the packet list and to re-filter the packets if a protocol preference has been changed. Keep track of whether preferences are changed from their initial value by a preferences file or a command-line option, or from their previous value by the "Preferences" dialog box; have "prefs_apply_all()" only call the "apply" callback for a module if they have. Call "prefs_apply_all()" after the command-line arguments have been parsed and after "OK" has been clicked in the "Preferences" dialog box, to notify modules of preference changes if they've registered a callback for that. After "OK" has been clicked in the "Preferences" dialog box, if any preferences have changed, call the reprocessing routine, as the summary line for some frames and/or the current display filter's value when applied to some frames may have changed as a result of a preference change. Do the same after "OK" or "Apply" has been clicked in the "Display Options" dialog box (as it controls a protocol preferences item. svn path=/trunk/; revision=2126
2000-07-09 03:29:42 +00:00
void
cf_redissect_packets(capture_file *cf)
Turn the code of "colorize_packet()" into a static routine that is given a word to use in the progress dialog, and a flag indicating whether the display filter is to be reevaluated or not, and: have "colorize_packet()" call that routine with "Colorizing" and FALSE as those arguments; have the filtering code call that routine with "Filtering" and TRUE as those arguments; add an exported routine to call that routine with "Reprocessing" and TRUE as those arguments, to use to re-generate the packet list and to re-filter the packets if a protocol preference has been changed. Keep track of whether preferences are changed from their initial value by a preferences file or a command-line option, or from their previous value by the "Preferences" dialog box; have "prefs_apply_all()" only call the "apply" callback for a module if they have. Call "prefs_apply_all()" after the command-line arguments have been parsed and after "OK" has been clicked in the "Preferences" dialog box, to notify modules of preference changes if they've registered a callback for that. After "OK" has been clicked in the "Preferences" dialog box, if any preferences have changed, call the reprocessing routine, as the summary line for some frames and/or the current display filter's value when applied to some frames may have changed as a result of a preference change. Do the same after "OK" or "Apply" has been clicked in the "Display Options" dialog box (as it controls a protocol preferences item. svn path=/trunk/; revision=2126
2000-07-09 03:29:42 +00:00
{
if (cf->read_lock || cf->redissection_queued == RESCAN_SCAN) {
/* Dissection in progress, signal redissection rather than rescanning. That
* would destroy the current (in-progress) dissection in "cf_read" which
* will cause issues when "cf_read" tries to add packets to the list.
* If a previous rescan was requested, "upgrade" it to a full redissection.
*/
cf->redissection_queued = RESCAN_REDISSECT;
}
if (cf->redissection_queued != RESCAN_NONE) {
/* Redissection is (already) queued, wait for "cf_read" to finish. */
return;
}
if (cf->state != FILE_CLOSED) {
/* Restart dissection in case no cf_read is pending. */
rescan_packets(cf, "Reprocessing", "all packets", TRUE);
}
Turn the code of "colorize_packet()" into a static routine that is given a word to use in the progress dialog, and a flag indicating whether the display filter is to be reevaluated or not, and: have "colorize_packet()" call that routine with "Colorizing" and FALSE as those arguments; have the filtering code call that routine with "Filtering" and TRUE as those arguments; add an exported routine to call that routine with "Reprocessing" and TRUE as those arguments, to use to re-generate the packet list and to re-filter the packets if a protocol preference has been changed. Keep track of whether preferences are changed from their initial value by a preferences file or a command-line option, or from their previous value by the "Preferences" dialog box; have "prefs_apply_all()" only call the "apply" callback for a module if they have. Call "prefs_apply_all()" after the command-line arguments have been parsed and after "OK" has been clicked in the "Preferences" dialog box, to notify modules of preference changes if they've registered a callback for that. After "OK" has been clicked in the "Preferences" dialog box, if any preferences have changed, call the reprocessing routine, as the summary line for some frames and/or the current display filter's value when applied to some frames may have changed as a result of a preference change. Do the same after "OK" or "Apply" has been clicked in the "Display Options" dialog box (as it controls a protocol preferences item. svn path=/trunk/; revision=2126
2000-07-09 03:29:42 +00:00
}
gboolean
cf_read_record_r(capture_file *cf, const frame_data *fdata,
wtap_rec *rec, Buffer *buf)
{
int err;
gchar *err_info;
if (!wtap_seek_read(cf->provider.wth, fdata->file_off, rec, buf, &err, &err_info)) {
cfile_read_failure_alert_box(cf->filename, err, err_info);
return FALSE;
}
return TRUE;
}
gboolean
cf_read_record(capture_file *cf, frame_data *fdata)
{
return cf_read_record_r(cf, fdata, &cf->rec, &cf->buf);
}
Turn the code of "colorize_packet()" into a static routine that is given a word to use in the progress dialog, and a flag indicating whether the display filter is to be reevaluated or not, and: have "colorize_packet()" call that routine with "Colorizing" and FALSE as those arguments; have the filtering code call that routine with "Filtering" and TRUE as those arguments; add an exported routine to call that routine with "Reprocessing" and TRUE as those arguments, to use to re-generate the packet list and to re-filter the packets if a protocol preference has been changed. Keep track of whether preferences are changed from their initial value by a preferences file or a command-line option, or from their previous value by the "Preferences" dialog box; have "prefs_apply_all()" only call the "apply" callback for a module if they have. Call "prefs_apply_all()" after the command-line arguments have been parsed and after "OK" has been clicked in the "Preferences" dialog box, to notify modules of preference changes if they've registered a callback for that. After "OK" has been clicked in the "Preferences" dialog box, if any preferences have changed, call the reprocessing routine, as the summary line for some frames and/or the current display filter's value when applied to some frames may have changed as a result of a preference change. Do the same after "OK" or "Apply" has been clicked in the "Display Options" dialog box (as it controls a protocol preferences item. svn path=/trunk/; revision=2126
2000-07-09 03:29:42 +00:00
/* Rescan the list of packets, reconstructing the CList.
"action" describes why we're doing this; it's used in the progress
dialog box.
"action_item" describes what we're doing; it's used in the progress
dialog box.
"redissect" is TRUE if we need to make the dissectors reconstruct
any state information they have (because a preference that affects
some dissector has changed, meaning some dissector might construct
its state differently from the way it was constructed the last time). */
Turn the code of "colorize_packet()" into a static routine that is given a word to use in the progress dialog, and a flag indicating whether the display filter is to be reevaluated or not, and: have "colorize_packet()" call that routine with "Colorizing" and FALSE as those arguments; have the filtering code call that routine with "Filtering" and TRUE as those arguments; add an exported routine to call that routine with "Reprocessing" and TRUE as those arguments, to use to re-generate the packet list and to re-filter the packets if a protocol preference has been changed. Keep track of whether preferences are changed from their initial value by a preferences file or a command-line option, or from their previous value by the "Preferences" dialog box; have "prefs_apply_all()" only call the "apply" callback for a module if they have. Call "prefs_apply_all()" after the command-line arguments have been parsed and after "OK" has been clicked in the "Preferences" dialog box, to notify modules of preference changes if they've registered a callback for that. After "OK" has been clicked in the "Preferences" dialog box, if any preferences have changed, call the reprocessing routine, as the summary line for some frames and/or the current display filter's value when applied to some frames may have changed as a result of a preference change. Do the same after "OK" or "Apply" has been clicked in the "Display Options" dialog box (as it controls a protocol preferences item. svn path=/trunk/; revision=2126
2000-07-09 03:29:42 +00:00
static void
rescan_packets(capture_file *cf, const char *action, const char *action_item, gboolean redissect)
{
/* Rescan packets new packet list */
guint32 framenum;
frame_data *fdata;
progdlg_t *progbar = NULL;
GTimer *prog_timer = g_timer_new();
int count;
frame_data *selected_frame, *preceding_frame, *following_frame, *prev_frame;
int selected_frame_num, preceding_frame_num, following_frame_num, prev_frame_num;
gboolean selected_frame_seen;
float progbar_val;
GTimeVal start_time;
gchar status_str[100];
epan_dissect_t edt;
dfilter_t *dfcode;
column_info *cinfo;
gboolean create_proto_tree;
guint tap_flags;
gboolean add_to_packet_list = FALSE;
gboolean compiled;
guint32 frames_count;
gboolean queued_rescan_type = RESCAN_NONE;
/* Rescan in progress, clear pending actions. */
cf->redissection_queued = RESCAN_NONE;
g_assert(!cf->read_lock);
cf->read_lock = TRUE;
/* Compile the current display filter.
* We assume this will not fail since cf->dfilter is only set in
* cf_filter IFF the filter was valid.
*/
compiled = dfilter_compile(cf->dfilter, &dfcode, NULL);
g_assert(!cf->dfilter || (compiled && dfcode));
We can't trust "cf->current_frame" to refer to the frame that was selected before we started re-colorizing or re-filtering the display, as when the first row is added to the clist, that may be selected and thus made the current frame. This means that we can't find the row corresponding to the previously-selected frame, if any, by checking as each packet is colorized/filtered and see whether its "frame_data" structure is equal to "cf->current_frame", as that'll always say that the first frame in the display is the selected frame. Instead, we recored the value of "cf->current_frame" before we do anything to the clist, have "add_packet_to_packet_list()" return either the row number of the frame (if it passed the filter and thus was added to the clist) or -1 (if it didn't pass the filter and thus wasn't added to the clist), and, after "add_packet_to_packet_list()", if the current frame is the one that was the selected row, remember its row number (if any), and, when we're finished colorizing/filtering the display, make that row the current row if it's not -1 (-1 means that the selected row didn't pass the filter). Also, don't do that until after we've thawed the clist, as the vertical adjustment for the clist doesn't reflect reality until then, and attempting to go to a given row won't work right until the vertical adjustment for the clist reflects reality. Shove all the code to set the selected and focus rows, and to make said row visible, into a routine, so the "Find Frame" and "Go To Frame" code can use it as well. svn path=/trunk/; revision=1959
2000-05-15 01:50:16 +00:00
/* Get the union of the flags for all tap listeners. */
tap_flags = union_of_tap_listener_flags();
/* If any tap listeners require the columns, construct them. */
cinfo = (tap_flags & TL_REQUIRES_COLUMNS) ? &cf->cinfo : NULL;
/*
* Determine whether we need to create a protocol tree.
* We do if:
*
* we're going to apply a display filter;
*
* one of the tap listeners is going to apply a filter;
*
* one of the tap listeners requires a protocol tree;
*
* we're redissecting and a postdissector wants field
* values or protocols on the first pass.
*/
create_proto_tree =
(dfcode != NULL || have_filtering_tap_listeners() ||
(tap_flags & TL_REQUIRES_PROTO_TREE) ||
(redissect && postdissectors_want_hfids()));
reset_tap_listeners();
We can't trust "cf->current_frame" to refer to the frame that was selected before we started re-colorizing or re-filtering the display, as when the first row is added to the clist, that may be selected and thus made the current frame. This means that we can't find the row corresponding to the previously-selected frame, if any, by checking as each packet is colorized/filtered and see whether its "frame_data" structure is equal to "cf->current_frame", as that'll always say that the first frame in the display is the selected frame. Instead, we recored the value of "cf->current_frame" before we do anything to the clist, have "add_packet_to_packet_list()" return either the row number of the frame (if it passed the filter and thus was added to the clist) or -1 (if it didn't pass the filter and thus wasn't added to the clist), and, after "add_packet_to_packet_list()", if the current frame is the one that was the selected row, remember its row number (if any), and, when we're finished colorizing/filtering the display, make that row the current row if it's not -1 (-1 means that the selected row didn't pass the filter). Also, don't do that until after we've thawed the clist, as the vertical adjustment for the clist doesn't reflect reality until then, and attempting to go to a given row won't work right until the vertical adjustment for the clist reflects reality. Shove all the code to set the selected and focus rows, and to make said row visible, into a routine, so the "Find Frame" and "Go To Frame" code can use it as well. svn path=/trunk/; revision=1959
2000-05-15 01:50:16 +00:00
/* Which frame, if any, is the currently selected frame?
XXX - should the selected frame or the focus frame be the "current"
frame, that frame being the one from which "Find Frame" searches
start? */
selected_frame = cf->current_frame;
/* Mark frame num as not found */
selected_frame_num = -1;
/* Freeze the packet list while we redo it, so we don't get any
screen updates while it happens. */
packet_list_freeze();
if (redissect) {
/* We need to re-initialize all the state information that protocols
keep, because some preference that controls a dissector has changed,
which might cause the state information to be constructed differently
by that dissector. */
/* We might receive new packets while redissecting, and we don't
want to dissect those before their time. */
cf->redissecting = TRUE;
/* 'reset' dissection session */
epan_free(cf->epan);
if (cf->edt && cf->edt->pi.fd) {
/* All pointers in "per frame proto data" for the currently selected
packet are allocated in wmem_file_scope() and deallocated in epan_free().
Free them here to avoid unintended usage in packet_list_clear(). */
frame_data_destroy(cf->edt->pi.fd);
}
cf->epan = ws_epan_new(cf);
cf->cinfo.epan = cf->epan;
/* A new Lua tap listener may be registered in lua_prime_all_fields()
called via epan_new() / init_dissection() when reloading Lua plugins. */
if (!create_proto_tree && have_filtering_tap_listeners()) {
create_proto_tree = TRUE;
}
/* We need to redissect the packets so we have to discard our old
* packet list store. */
packet_list_clear();
add_to_packet_list = TRUE;
}
/* We don't yet know which will be the first and last frames displayed. */
Store the frame_data structures in a tree, rather than a linked list. This lets us get rid of the per-frame_data-structure prev and next pointers, saving memory (at least according to Activity Monitor's report of the virtual address space size on my Snow Leopard machine, it's a noticeable saving), and lets us look up frame_data structures by frame number in O(log2(number of frames)) time rather than O(number of frames) time. It seems to take more CPU time when reading in the file, but seems to go from "finished reading in all the packets" to "displaying the packets" faster and seems to free up the frame_data structures faster when closing the file. It *is* doing more copying, currently, as we now don't allocate the frame_data structure until after the packet has passed the read filter, so that might account for the additional CPU time. (Oh, and, for what it's worth, on an LP64 platform, a frame_data structure is exactly 128 bytes long. However, there's more stuff to remove, so the power-of-2 size is not guaranteed to remain, and it's not a power-of-2 size on an ILP32 platform.) It also means we don't need GLib 2.10 or later for the two-pass mode in TShark. It also means some code in the TCP dissector that was checking pinfo->fd->next to see if it's NULL, in order to see if this is the last packet in the file, no longer works, but that wasn't guaranteed to work anyway: we might be doing a one-pass read through the capture in TShark; we might be dissecting the frame while we're reading in the packets for the first time in Wireshark; we might be doing a live capture in Wireshark; in which case packets might be prematurely considered "the last packet". #if 0 the no-longer-working tests, pending figuring out a better way of doing it. svn path=/trunk/; revision=36849
2011-04-25 19:01:05 +00:00
cf->first_displayed = 0;
cf->last_displayed = 0;
/* We currently don't display any packets */
cf->displayed_count = 0;
Turn the code of "colorize_packet()" into a static routine that is given a word to use in the progress dialog, and a flag indicating whether the display filter is to be reevaluated or not, and: have "colorize_packet()" call that routine with "Colorizing" and FALSE as those arguments; have the filtering code call that routine with "Filtering" and TRUE as those arguments; add an exported routine to call that routine with "Reprocessing" and TRUE as those arguments, to use to re-generate the packet list and to re-filter the packets if a protocol preference has been changed. Keep track of whether preferences are changed from their initial value by a preferences file or a command-line option, or from their previous value by the "Preferences" dialog box; have "prefs_apply_all()" only call the "apply" callback for a module if they have. Call "prefs_apply_all()" after the command-line arguments have been parsed and after "OK" has been clicked in the "Preferences" dialog box, to notify modules of preference changes if they've registered a callback for that. After "OK" has been clicked in the "Preferences" dialog box, if any preferences have changed, call the reprocessing routine, as the summary line for some frames and/or the current display filter's value when applied to some frames may have changed as a result of a preference change. Do the same after "OK" or "Apply" has been clicked in the "Display Options" dialog box (as it controls a protocol preferences item. svn path=/trunk/; revision=2126
2000-07-09 03:29:42 +00:00
/* Iterate through the list of frames. Call a routine for each frame
to check whether it should be displayed and, if so, add it to
the display list. */
cf->provider.ref = NULL;
cf->provider.prev_dis = NULL;
cf->provider.prev_cap = NULL;
cf->cum_bytes = 0;
cf_callback_invoke(cf_cb_file_rescan_started, cf);
g_timer_start(prog_timer);
/* Count of packets at which we've looked. */
count = 0;
/* Progress so far. */
progbar_val = 0.0f;
cf->stop_flag = FALSE;
g_get_current_time(&start_time);
/* no previous row yet */
prev_frame_num = -1;
prev_frame = NULL;
preceding_frame_num = -1;
preceding_frame = NULL;
following_frame_num = -1;
following_frame = NULL;
selected_frame_seen = FALSE;
frames_count = cf->count;
epan_dissect_init(&edt, cf->epan, create_proto_tree, FALSE);
for (framenum = 1; framenum <= frames_count; framenum++) {
fdata = frame_data_sequence_find(cf->provider.frames, framenum);
/* Create the progress bar if necessary.
We check on every iteration of the loop, so that it takes no
longer than the standard time to create it (otherwise, for a
large file, we might take considerably longer than that standard
time in order to get to the next progress bar step). */
if (progbar == NULL)
progbar = delayed_create_progress_dlg(cf->window, action, action_item, TRUE,
&cf->stop_flag,
&start_time,
progbar_val);
/*
* Update the progress bar, but do it only after PROGBAR_UPDATE_INTERVAL
* has elapsed. Calling update_progress_dlg and packets_bar_update will
* likely trigger UI paint events, which might take a while depending on
* the platform and display. Reset our timer *after* painting.
*/
if (g_timer_elapsed(prog_timer, NULL) > PROGBAR_UPDATE_INTERVAL) {
/* let's not divide by zero. I should never be started
* with count == 0, so let's assert that
*/
g_assert(cf->count > 0);
progbar_val = (gfloat) count / frames_count;
if (progbar != NULL) {
g_snprintf(status_str, sizeof(status_str),
"%4u of %u frames", count, frames_count);
update_progress_dlg(progbar, progbar_val, status_str);
}
g_timer_start(prog_timer);
}
queued_rescan_type = cf->redissection_queued;
if (queued_rescan_type != RESCAN_NONE) {
/* A redissection was requested while an existing redissection was
* pending. */
break;
}
if (cf->stop_flag) {
/* Well, the user decided to abort the filtering. Just stop.
XXX - go back to the previous filter? Users probably just
want not to wait for a filtering operation to finish;
unless we cancel by having no filter, reverting to the
previous filter will probably be even more expensive than
continuing the filtering, as it involves going back to the
beginning and filtering, and even with no filter we currently
have to re-generate the entire clist, which is also expensive.
I'm not sure what Network Monitor does, but it doesn't appear
to give you an unfiltered display if you cancel. */
break;
}
count++;
if (redissect) {
/* Since all state for the frame was destroyed, mark the frame
* as not visited, free the GSList referring to the state
* data (the per-frame data itself was freed by
* "init_dissection()"), and null out the GSList pointer. */
frame_data_reset(fdata);
frames_count = cf->count;
}
/* Frame dependencies from the previous dissection/filtering are no longer valid. */
fdata->flags.dependent_of_displayed = 0;
if (!cf_read_record(cf, fdata))
break; /* error reading the frame */
/* If the previous frame is displayed, and we haven't yet seen the
selected frame, remember that frame - it's the closest one we've
yet seen before the selected frame. */
if (prev_frame_num != -1 && !selected_frame_seen && prev_frame->flags.passed_dfilter) {
preceding_frame_num = prev_frame_num;
preceding_frame = prev_frame;
}
add_packet_to_packet_list(fdata, cf, &edt, dfcode,
cinfo, &cf->rec,
ws_buffer_start_ptr(&cf->buf),
add_to_packet_list);
/* If this frame is displayed, and this is the first frame we've
seen displayed after the selected frame, remember this frame -
it's the closest one we've yet seen at or after the selected
frame. */
if (fdata->flags.passed_dfilter && selected_frame_seen && following_frame_num == -1) {
following_frame_num = fdata->num;
following_frame = fdata;
}
if (fdata == selected_frame) {
selected_frame_seen = TRUE;
if (fdata->flags.passed_dfilter)
selected_frame_num = fdata->num;
}
/* Remember this frame - it'll be the previous frame
on the next pass through the loop. */
prev_frame_num = fdata->num;
prev_frame = fdata;
}
epan_dissect_cleanup(&edt);
/* We are done redissecting the packet list. */
cf->redissecting = FALSE;
if (redissect) {
frames_count = cf->count;
/* Clear out what remains of the visited flags and per-frame data
pointers.
XXX - that may cause various forms of bogosity when dissecting
these frames, as they won't have been seen by this sequential
pass, but the only alternative I see is to keep scanning them
even though the user requested that the scan stop, and that
would leave the user stuck with an Wireshark grinding on
until it finishes. Should we just stick them with that? */
for (; framenum <= frames_count; framenum++) {
fdata = frame_data_sequence_find(cf->provider.frames, framenum);
frame_data_reset(fdata);
}
}
/* We're done filtering the packets; destroy the progress bar if it
was created. */
if (progbar != NULL)
destroy_progress_dlg(progbar);
g_timer_destroy(prog_timer);
We can't trust "cf->current_frame" to refer to the frame that was selected before we started re-colorizing or re-filtering the display, as when the first row is added to the clist, that may be selected and thus made the current frame. This means that we can't find the row corresponding to the previously-selected frame, if any, by checking as each packet is colorized/filtered and see whether its "frame_data" structure is equal to "cf->current_frame", as that'll always say that the first frame in the display is the selected frame. Instead, we recored the value of "cf->current_frame" before we do anything to the clist, have "add_packet_to_packet_list()" return either the row number of the frame (if it passed the filter and thus was added to the clist) or -1 (if it didn't pass the filter and thus wasn't added to the clist), and, after "add_packet_to_packet_list()", if the current frame is the one that was the selected row, remember its row number (if any), and, when we're finished colorizing/filtering the display, make that row the current row if it's not -1 (-1 means that the selected row didn't pass the filter). Also, don't do that until after we've thawed the clist, as the vertical adjustment for the clist doesn't reflect reality until then, and attempting to go to a given row won't work right until the vertical adjustment for the clist reflects reality. Shove all the code to set the selected and focus rows, and to make said row visible, into a routine, so the "Find Frame" and "Go To Frame" code can use it as well. svn path=/trunk/; revision=1959
2000-05-15 01:50:16 +00:00
/* Unfreeze the packet list. */
if (!add_to_packet_list)
packet_list_recreate_visible_rows();
We can't trust "cf->current_frame" to refer to the frame that was selected before we started re-colorizing or re-filtering the display, as when the first row is added to the clist, that may be selected and thus made the current frame. This means that we can't find the row corresponding to the previously-selected frame, if any, by checking as each packet is colorized/filtered and see whether its "frame_data" structure is equal to "cf->current_frame", as that'll always say that the first frame in the display is the selected frame. Instead, we recored the value of "cf->current_frame" before we do anything to the clist, have "add_packet_to_packet_list()" return either the row number of the frame (if it passed the filter and thus was added to the clist) or -1 (if it didn't pass the filter and thus wasn't added to the clist), and, after "add_packet_to_packet_list()", if the current frame is the one that was the selected row, remember its row number (if any), and, when we're finished colorizing/filtering the display, make that row the current row if it's not -1 (-1 means that the selected row didn't pass the filter). Also, don't do that until after we've thawed the clist, as the vertical adjustment for the clist doesn't reflect reality until then, and attempting to go to a given row won't work right until the vertical adjustment for the clist reflects reality. Shove all the code to set the selected and focus rows, and to make said row visible, into a routine, so the "Find Frame" and "Go To Frame" code can use it as well. svn path=/trunk/; revision=1959
2000-05-15 01:50:16 +00:00
/* Compute the time it took to filter the file */
compute_elapsed(cf, &start_time);
packet_list_thaw();
cf_callback_invoke(cf_cb_file_rescan_finished, cf);
if (selected_frame_num == -1) {
/* The selected frame didn't pass the filter. */
if (selected_frame == NULL) {
/* That's because there *was* no selected frame. Make the first
displayed frame the current frame. */
selected_frame_num = 0;
} else {
/* Find the nearest displayed frame to the selected frame (whether
it's before or after that frame) and make that the current frame.
If the next and previous displayed frames are equidistant from the
selected frame, choose the next one. */
g_assert(following_frame == NULL ||
following_frame->num >= selected_frame->num);
g_assert(preceding_frame == NULL ||
preceding_frame->num <= selected_frame->num);
if (following_frame == NULL) {
/* No frame after the selected frame passed the filter, so we
have to select the last displayed frame before the selected
frame. */
selected_frame_num = preceding_frame_num;
selected_frame = preceding_frame;
} else if (preceding_frame == NULL) {
/* No frame before the selected frame passed the filter, so we
have to select the first displayed frame after the selected
frame. */
selected_frame_num = following_frame_num;
selected_frame = following_frame;
} else {
/* Frames before and after the selected frame passed the filter, so
we'll select the previous frame */
selected_frame_num = preceding_frame_num;
selected_frame = preceding_frame;
}
}
}
if (selected_frame_num == -1) {
/* There are no frames displayed at all. */
cf_unselect_packet(cf);
} else {
/* Either the frame that was selected passed the filter, or we've
found the nearest displayed frame to that frame. Select it, make
it the focus row, and make it visible. */
/* Set to invalid to force update of packet list and packet details */
cf->current_row = -1;
if (selected_frame_num == 0) {
packet_list_select_first_row();
}else{
if (!packet_list_select_row_from_data(selected_frame)) {
/* We didn't find a row corresponding to this frame.
This means that the frame isn't being displayed currently,
so we can't select it. */
simple_message_box(ESD_TYPE_INFO, NULL,
"The capture file is probably not fully dissected.",
"End of capture exceeded.");
}
}
}
/* Cleanup and release all dfilter resources */
dfilter_free(dfcode);
/* It is safe again to execute redissections. */
g_assert(cf->read_lock);
cf->read_lock = FALSE;
/* If another rescan (due to dfilter change) or redissection (due to profile
* change) was requested, the rescan above is aborted and restarted here. */
if (queued_rescan_type != RESCAN_NONE) {
redissect = redissect || queued_rescan_type == RESCAN_REDISSECT;
rescan_packets(cf, "Reprocessing", "all packets", redissect);
}
}
/*
* Scan through all frame data and recalculate the ref time
* without rereading the file.
* XXX - do we need a progres bar or is this fast enough?
*/
static void
ref_time_packets(capture_file *cf)
{
guint32 framenum;
frame_data *fdata;
nstime_t rel_ts;
cf->provider.ref = NULL;
cf->provider.prev_dis = NULL;
cf->cum_bytes = 0;
for (framenum = 1; framenum <= cf->count; framenum++) {
fdata = frame_data_sequence_find(cf->provider.frames, framenum);
/* just add some value here until we know if it is being displayed or not */
fdata->cum_bytes = cf->cum_bytes + fdata->pkt_len;
/*
*Timestamps
*/
/* If we don't have the time stamp of the first packet in the
capture, it's because this is the first packet. Save the time
stamp of this packet as the time stamp of the first packet. */
if (cf->provider.ref == NULL)
cf->provider.ref = fdata;
/* if this frames is marked as a reference time frame, reset
firstsec and firstusec to this frame */
if (fdata->flags.ref_time)
cf->provider.ref = fdata;
/* If we don't have the time stamp of the previous displayed packet,
it's because this is the first displayed packet. Save the time
stamp of this packet as the time stamp of the previous displayed
packet. */
if (cf->provider.prev_dis == NULL) {
cf->provider.prev_dis = fdata;
}
/* Get the time elapsed between the first packet and this packet. */
fdata->frame_ref_num = (fdata != cf->provider.ref) ? cf->provider.ref->num : 0;
nstime_delta(&rel_ts, &fdata->abs_ts, &cf->provider.ref->abs_ts);
/* If it's greater than the current elapsed time, set the elapsed time
to it (we check for "greater than" so as not to be confused by
time moving backwards). */
if ((gint32)cf->elapsed_time.secs < rel_ts.secs
|| ((gint32)cf->elapsed_time.secs == rel_ts.secs && (gint32)cf->elapsed_time.nsecs < rel_ts.nsecs)) {
cf->elapsed_time = rel_ts;
}
Fix https://bugs.wireshark.org/bugzilla/show_bug.cgi?id=3315 - make Save-As/Displayed/All-Packets save not only the displayed packets but also any other packets needed (e.g., for reassembly) to fully dissect the displayed packets. This works only for the "All packets" case; choosing only the Selected packet, the Marked packets, or a range of packets would require actually storing which packets depend on which (too much memory) or going through the packet list many times (too slow). Also, this behavior is always the case: you can't save the displayed packets without their dependencies (I don't see why this would be desirable). So far this is done for SCTP and things using the reassembly routines (TCP has been tested). The Win32 dialog was modified but hasn't been tested yet. One confusing aspect of the UI is that the Displayed count in the Save-As dialog does not match the number of displayed packets. (I tried renaming the button "Displayed + Dependencies" but it looked too big.) The tooltip tries to explain this and the fact that this works only in the All-Packets case; suggestions for improvement are welcome. Implementation details: Dissectors (or the reassembly code) can list frames which were needed to build the current frame's tree. If the current frame passes the display filter then each listed frame is marked as "depended upon" (this takes up the last free frame_data flag). When performing a Save-As/Displayed/All-Packets then choose packets which passed the dfilter _or_ are depended upon. svn path=/trunk/; revision=41216
2012-02-28 03:19:49 +00:00
/* If this frame is displayed, get the time elapsed between the
previous displayed packet and this packet. */
if ( fdata->flags.passed_dfilter ) {
fdata->prev_dis_num = cf->provider.prev_dis->num;
cf->provider.prev_dis = fdata;
}
/*
* Byte counts
*/
if ( (fdata->flags.passed_dfilter) || (fdata->flags.ref_time) ) {
/* This frame either passed the display filter list or is marked as
a time reference frame. All time reference frames are displayed
even if they don't pass the display filter */
if (fdata->flags.ref_time) {
/* if this was a TIME REF frame we should reset the cum_bytes field */
cf->cum_bytes = fdata->pkt_len;
fdata->cum_bytes = cf->cum_bytes;
} else {
/* increase cum_bytes with this packets length */
cf->cum_bytes += fdata->pkt_len;
}
}
}
}
typedef enum {
PSP_FINISHED,
PSP_STOPPED,
PSP_FAILED
} psp_return_t;
static psp_return_t
process_specified_records(capture_file *cf, packet_range_t *range,
const char *string1, const char *string2, gboolean terminate_is_stop,
gboolean (*callback)(capture_file *, frame_data *,
wtap_rec *, const guint8 *, void *),
void *callback_args,
gboolean show_progress_bar)
{
guint32 framenum;
frame_data *fdata;
wtap_rec rec;
Buffer buf;
psp_return_t ret = PSP_FINISHED;
progdlg_t *progbar = NULL;
GTimer *prog_timer = g_timer_new();
int progbar_count;
float progbar_val;
GTimeVal progbar_start_time;
gchar progbar_status_str[100];
range_process_e process_this;
wtap_rec_init(&rec);
ws_buffer_init(&buf, 1500);
g_timer_start(prog_timer);
/* Count of packets at which we've looked. */
progbar_count = 0;
/* Progress so far. */
progbar_val = 0.0f;
if (cf->read_lock) {
g_warning("Failing due to nested process_specified_records(\"%s\") call!", cf->filename);
return PSP_FAILED;
}
cf->read_lock = TRUE;
cf->stop_flag = FALSE;
g_get_current_time(&progbar_start_time);
if (range != NULL)
packet_range_process_init(range);
/* Iterate through all the packets, printing the packets that
were selected by the current display filter. */
for (framenum = 1; framenum <= cf->count; framenum++) {
fdata = frame_data_sequence_find(cf->provider.frames, framenum);
/* Create the progress bar if necessary.
We check on every iteration of the loop, so that it takes no
longer than the standard time to create it (otherwise, for a
large file, we might take considerably longer than that standard
time in order to get to the next progress bar step). */
if (show_progress_bar && progbar == NULL)
progbar = delayed_create_progress_dlg(cf->window, string1, string2,
terminate_is_stop,
&cf->stop_flag,
&progbar_start_time,
progbar_val);
/*
* Update the progress bar, but do it only after PROGBAR_UPDATE_INTERVAL
* has elapsed. Calling update_progress_dlg and packets_bar_update will
* likely trigger UI paint events, which might take a while depending on
* the platform and display. Reset our timer *after* painting.
*/
if (progbar && g_timer_elapsed(prog_timer, NULL) > PROGBAR_UPDATE_INTERVAL) {
/* let's not divide by zero. I should never be started
* with count == 0, so let's assert that
*/
g_assert(cf->count > 0);
progbar_val = (gfloat) progbar_count / cf->count;
g_snprintf(progbar_status_str, sizeof(progbar_status_str),
"%4u of %u packets", progbar_count, cf->count);
update_progress_dlg(progbar, progbar_val, progbar_status_str);
g_timer_start(prog_timer);
}
if (cf->stop_flag) {
/* Well, the user decided to abort the operation. Just stop,
and arrange to return PSP_STOPPED to our caller, so they know
it was stopped explicitly. */
ret = PSP_STOPPED;
break;
}
progbar_count++;
if (range != NULL) {
/* do we have to process this packet? */
process_this = packet_range_process_packet(range, fdata);
if (process_this == range_process_next) {
/* this packet uninteresting, continue with next one */
continue;
} else if (process_this == range_processing_finished) {
/* all interesting packets processed, stop the loop */
break;
}
}
/* Get the packet */
if (!cf_read_record_r(cf, fdata, &rec, &buf)) {
/* Attempt to get the packet failed. */
ret = PSP_FAILED;
break;
}
/* Process the packet */
if (!callback(cf, fdata, &rec, ws_buffer_start_ptr(&buf), callback_args)) {
/* Callback failed. We assume it reported the error appropriately. */
ret = PSP_FAILED;
break;
}
}
/* We're done printing the packets; destroy the progress bar if
it was created. */
if (progbar != NULL)
destroy_progress_dlg(progbar);
g_timer_destroy(prog_timer);
g_assert(cf->read_lock);
cf->read_lock = FALSE;
wtap_rec_cleanup(&rec);
ws_buffer_free(&buf);
return ret;
}
typedef struct {
epan_dissect_t edt;
column_info *cinfo;
} retap_callback_args_t;
static gboolean
retap_packet(capture_file *cf, frame_data *fdata, wtap_rec *rec,
const guint8 *pd, void *argsp)
{
retap_callback_args_t *args = (retap_callback_args_t *)argsp;
epan_dissect_run_with_taps(&args->edt, cf->cd_t, rec,
frame_tvbuff_new(&cf->provider, fdata, pd),
fdata, args->cinfo);
epan_dissect_reset(&args->edt);
return TRUE;
}
cf_read_status_t
cf_retap_packets(capture_file *cf)
{
packet_range_t range;
retap_callback_args_t callback_args;
gboolean create_proto_tree;
guint tap_flags;
psp_return_t ret;
/* Presumably the user closed the capture file. */
if (cf == NULL) {
return CF_READ_ABORTED;
}
cf_callback_invoke(cf_cb_file_retap_started, cf);
/* Get the union of the flags for all tap listeners. */
tap_flags = union_of_tap_listener_flags();
/* If any tap listeners require the columns, construct them. */
callback_args.cinfo = (tap_flags & TL_REQUIRES_COLUMNS) ? &cf->cinfo : NULL;
/*
* Determine whether we need to create a protocol tree.
* We do if:
*
* one of the tap listeners is going to apply a filter;
*
* one of the tap listeners requires a protocol tree.
*/
create_proto_tree =
(have_filtering_tap_listeners() || (tap_flags & TL_REQUIRES_PROTO_TREE));
/* Reset the tap listeners. */
reset_tap_listeners();
epan_dissect_init(&callback_args.edt, cf->epan, create_proto_tree, FALSE);
/* Iterate through the list of packets, dissecting all packets and
re-running the taps. */
packet_range_init(&range, cf);
packet_range_process_init(&range);
ret = process_specified_records(cf, &range, "Recalculating statistics on",
"all packets", TRUE, retap_packet,
&callback_args, TRUE);
packet_range_cleanup(&range);
epan_dissect_cleanup(&callback_args.edt);
cf_callback_invoke(cf_cb_file_retap_finished, cf);
switch (ret) {
case PSP_FINISHED:
/* Completed successfully. */
return CF_READ_OK;
case PSP_STOPPED:
/* Well, the user decided to abort the refiltering.
Return CF_READ_ABORTED so our caller knows they did that. */
return CF_READ_ABORTED;
case PSP_FAILED:
/* Error while retapping. */
return CF_READ_ERROR;
}
g_assert_not_reached();
return CF_READ_OK;
}
typedef struct {
print_args_t *print_args;
gboolean print_header_line;
char *header_line_buf;
int header_line_buf_len;
gboolean print_formfeed;
gboolean print_separator;
char *line_buf;
int line_buf_len;
gint *col_widths;
int num_visible_cols;
gint *visible_cols;
epan_dissect_t edt;
} print_callback_args_t;
static gboolean
print_packet(capture_file *cf, frame_data *fdata, wtap_rec *rec,
const guint8 *pd, void *argsp)
{
print_callback_args_t *args = (print_callback_args_t *)argsp;
int i;
char *cp;
int line_len;
int column_len;
int cp_off;
char bookmark_name[9+10+1]; /* "__frameNNNNNNNNNN__\0" */
char bookmark_title[6+10+1]; /* "Frame NNNNNNNNNN__\0" */
col_item_t* col_item;
/* Fill in the column information if we're printing the summary
information. */
if (args->print_args->print_summary) {
col_custom_prime_edt(&args->edt, &cf->cinfo);
epan_dissect_run(&args->edt, cf->cd_t, rec,
frame_tvbuff_new(&cf->provider, fdata, pd),
fdata, &cf->cinfo);
epan_dissect_fill_in_columns(&args->edt, FALSE, TRUE);
} else
epan_dissect_run(&args->edt, cf->cd_t, rec,
frame_tvbuff_new(&cf->provider, fdata, pd), fdata, NULL);
if (args->print_formfeed) {
if (!new_page(args->print_args->stream))
goto fail;
/*
* Print another header line if we print a packet summary on the
* new page.
*/
if (args->print_args->print_col_headings)
args->print_header_line = TRUE;
} else {
if (args->print_separator) {
if (!print_line(args->print_args->stream, 0, ""))
goto fail;
}
}
/*
* We generate bookmarks, if the output format supports them.
* The name is "__frameN__".
*/
g_snprintf(bookmark_name, sizeof bookmark_name, "__frame%u__", fdata->num);
if (args->print_args->print_summary) {
if (!args->print_args->print_col_headings)
args->print_header_line = FALSE;
if (args->print_header_line) {
if (!print_line(args->print_args->stream, 0, args->header_line_buf))
goto fail;
args->print_header_line = FALSE; /* we might not need to print any more */
}
cp = &args->line_buf[0];
line_len = 0;
for (i = 0; i < args->num_visible_cols; i++) {
col_item = &cf->cinfo.columns[args->visible_cols[i]];
/* Find the length of the string for this column. */
column_len = (int) strlen(col_item->col_data);
if (args->col_widths[i] > column_len)
column_len = args->col_widths[i];
/* Make sure there's room in the line buffer for the column; if not,
double its length. */
line_len += column_len + 1; /* "+1" for space */
if (line_len > args->line_buf_len) {
cp_off = (int) (cp - args->line_buf);
args->line_buf_len = 2 * line_len;
args->line_buf = (char *)g_realloc(args->line_buf, args->line_buf_len + 1);
cp = args->line_buf + cp_off;
}
/* Right-justify the packet number column. */
if (col_item->col_fmt == COL_NUMBER)
g_snprintf(cp, column_len+1, "%*s", args->col_widths[i], col_item->col_data);
else
g_snprintf(cp, column_len+1, "%-*s", args->col_widths[i], col_item->col_data);
cp += column_len;
if (i != args->num_visible_cols - 1)
*cp++ = ' ';
}
*cp = '\0';
/*
* Generate a bookmark, using the summary line as the title.
*/
if (!print_bookmark(args->print_args->stream, bookmark_name,
args->line_buf))
goto fail;
if (!print_line(args->print_args->stream, 0, args->line_buf))
goto fail;
} else {
/*
* Generate a bookmark, using "Frame N" as the title, as we're not
* printing the summary line.
*/
g_snprintf(bookmark_title, sizeof bookmark_title, "Frame %u", fdata->num);
if (!print_bookmark(args->print_args->stream, bookmark_name,
bookmark_title))
goto fail;
} /* if (print_summary) */
if (args->print_args->print_dissections != print_dissections_none) {
if (args->print_args->print_summary) {
/* Separate the summary line from the tree with a blank line. */
if (!print_line(args->print_args->stream, 0, ""))
goto fail;
}
/* Print the information in that tree. */
if (!proto_tree_print(args->print_args->print_dissections,
args->print_args->print_hex, &args->edt, NULL,
args->print_args->stream))
goto fail;
/* Print a blank line if we print anything after this (aka more than one packet). */
args->print_separator = TRUE;
/* Print a header line if we print any more packet summaries */
if (args->print_args->print_col_headings)
args->print_header_line = TRUE;
}
if (args->print_args->print_hex) {
if (args->print_args->print_summary || (args->print_args->print_dissections != print_dissections_none)) {
if (!print_line(args->print_args->stream, 0, ""))
goto fail;
}
/* Print the full packet data as hex. */
if (!print_hex_data(args->print_args->stream, &args->edt))
goto fail;
/* Print a blank line if we print anything after this (aka more than one packet). */
args->print_separator = TRUE;
/* Print a header line if we print any more packet summaries */
if (args->print_args->print_col_headings)
args->print_header_line = TRUE;
} /* if (args->print_args->print_dissections != print_dissections_none) */
epan_dissect_reset(&args->edt);
/* do we want to have a formfeed between each packet from now on? */
if (args->print_args->print_formfeed) {
args->print_formfeed = TRUE;
}
return TRUE;
fail:
epan_dissect_reset(&args->edt);
return FALSE;
}
cf_print_status_t
cf_print_packets(capture_file *cf, print_args_t *print_args,
gboolean show_progress_bar)
{
print_callback_args_t callback_args;
gint data_width;
char *cp;
int i, cp_off, column_len, line_len;
int num_visible_col = 0, last_visible_col = 0, visible_col_count;
psp_return_t ret;
GList *clp;
fmt_data *cfmt;
gboolean proto_tree_needed;
callback_args.print_args = print_args;
callback_args.print_header_line = print_args->print_col_headings;
callback_args.header_line_buf = NULL;
callback_args.header_line_buf_len = 256;
callback_args.print_formfeed = FALSE;
callback_args.print_separator = FALSE;
callback_args.line_buf = NULL;
callback_args.line_buf_len = 256;
callback_args.col_widths = NULL;
callback_args.num_visible_cols = 0;
callback_args.visible_cols = NULL;
if (!print_preamble(print_args->stream, cf->filename, get_ws_vcs_version_info())) {
destroy_print_stream(print_args->stream);
return CF_PRINT_WRITE_ERROR;
}
if (print_args->print_summary) {
/* We're printing packet summaries. Allocate the header line buffer
and get the column widths. */
callback_args.header_line_buf = (char *)g_malloc(callback_args.header_line_buf_len + 1);
/* Find the number of visible columns and the last visible column */
for (i = 0; i < prefs.num_cols; i++) {
clp = g_list_nth(prefs.col_list, i);
if (clp == NULL) /* Sanity check, Invalid column requested */
continue;
cfmt = (fmt_data *) clp->data;
if (cfmt->visible) {
num_visible_col++;
last_visible_col = i;
}
}
/* if num_visible_col is 0, we are done */
if (num_visible_col == 0) {
g_free(callback_args.header_line_buf);
return CF_PRINT_OK;
}
/* Find the widths for each of the columns - maximum of the
width of the title and the width of the data - and construct
a buffer with a line containing the column titles. */
callback_args.num_visible_cols = num_visible_col;
callback_args.col_widths = (gint *) g_malloc(sizeof(gint) * num_visible_col);
callback_args.visible_cols = (gint *) g_malloc(sizeof(gint) * num_visible_col);
cp = &callback_args.header_line_buf[0];
line_len = 0;
visible_col_count = 0;
for (i = 0; i < cf->cinfo.num_cols; i++) {
clp = g_list_nth(prefs.col_list, i);
if (clp == NULL) /* Sanity check, Invalid column requested */
continue;
cfmt = (fmt_data *) clp->data;
if (cfmt->visible == FALSE)
continue;
/* Save the order of visible columns */
callback_args.visible_cols[visible_col_count] = i;
/* Don't pad the last column. */
if (i == last_visible_col)
callback_args.col_widths[visible_col_count] = 0;
else {
callback_args.col_widths[visible_col_count] = (gint) strlen(cf->cinfo.columns[i].col_title);
data_width = get_column_char_width(get_column_format(i));
if (data_width > callback_args.col_widths[visible_col_count])
callback_args.col_widths[visible_col_count] = data_width;
}
/* Find the length of the string for this column. */
column_len = (int) strlen(cf->cinfo.columns[i].col_title);
if (callback_args.col_widths[visible_col_count] > column_len)
column_len = callback_args.col_widths[visible_col_count];
/* Make sure there's room in the line buffer for the column; if not,
double its length. */
line_len += column_len + 1; /* "+1" for space */
if (line_len > callback_args.header_line_buf_len) {
cp_off = (int) (cp - callback_args.header_line_buf);
callback_args.header_line_buf_len = 2 * line_len;
callback_args.header_line_buf = (char *)g_realloc(callback_args.header_line_buf,
callback_args.header_line_buf_len + 1);
cp = callback_args.header_line_buf + cp_off;
}
/* Right-justify the packet number column. */
/* if (cf->cinfo.col_fmt[i] == COL_NUMBER)
g_snprintf(cp, column_len+1, "%*s", callback_args.col_widths[visible_col_count], cf->cinfo.columns[i].col_title);
else*/
g_snprintf(cp, column_len+1, "%-*s", callback_args.col_widths[visible_col_count], cf->cinfo.columns[i].col_title);
cp += column_len;
if (i != cf->cinfo.num_cols - 1)
*cp++ = ' ';
visible_col_count++;
}
*cp = '\0';
/* Now start out the main line buffer with the same length as the
header line buffer. */
callback_args.line_buf_len = callback_args.header_line_buf_len;
callback_args.line_buf = (char *)g_malloc(callback_args.line_buf_len + 1);
} /* if (print_summary) */
/* Create the protocol tree, and make it visible, if we're printing
the dissection or the hex data.
XXX - do we need it if we're just printing the hex data? */
proto_tree_needed =
callback_args.print_args->print_dissections != print_dissections_none ||
callback_args.print_args->print_hex ||
have_custom_cols(&cf->cinfo) || have_field_extractors();
epan_dissect_init(&callback_args.edt, cf->epan, proto_tree_needed, proto_tree_needed);
/* Iterate through the list of packets, printing the packets we were
told to print. */
ret = process_specified_records(cf, &print_args->range, "Printing",
"selected packets", TRUE, print_packet,
&callback_args, show_progress_bar);
epan_dissect_cleanup(&callback_args.edt);
g_free(callback_args.header_line_buf);
g_free(callback_args.line_buf);
g_free(callback_args.col_widths);
g_free(callback_args.visible_cols);
switch (ret) {
case PSP_FINISHED:
/* Completed successfully. */
break;
case PSP_STOPPED:
/* Well, the user decided to abort the printing.
XXX - note that what got generated before they did that
will get printed if we're piping to a print program; we'd
have to write to a file and then hand that to the print
program to make it actually not print anything. */
break;
case PSP_FAILED:
/* Error while printing.
XXX - note that what got generated before they did that
will get printed if we're piping to a print program; we'd
have to write to a file and then hand that to the print
program to make it actually not print anything. */
destroy_print_stream(print_args->stream);
return CF_PRINT_WRITE_ERROR;
}
if (!print_finale(print_args->stream)) {
destroy_print_stream(print_args->stream);
return CF_PRINT_WRITE_ERROR;
}
if (!destroy_print_stream(print_args->stream))
return CF_PRINT_WRITE_ERROR;
return CF_PRINT_OK;
}
typedef struct {
FILE *fh;
epan_dissect_t edt;
print_args_t *print_args;
} write_packet_callback_args_t;
static gboolean
write_pdml_packet(capture_file *cf, frame_data *fdata, wtap_rec *rec,
const guint8 *pd, void *argsp)
{
write_packet_callback_args_t *args = (write_packet_callback_args_t *)argsp;
/* Create the protocol tree, but don't fill in the column information. */
epan_dissect_run(&args->edt, cf->cd_t, rec,
frame_tvbuff_new(&cf->provider, fdata, pd), fdata, NULL);
/* Write out the information in that tree. */
write_pdml_proto_tree(NULL, NULL, PF_NONE, &args->edt, &cf->cinfo, args->fh, FALSE);
epan_dissect_reset(&args->edt);
return !ferror(args->fh);
}
cf_print_status_t
cf_write_pdml_packets(capture_file *cf, print_args_t *print_args)
{
write_packet_callback_args_t callback_args;
FILE *fh;
psp_return_t ret;
fh = ws_fopen(print_args->file, "w");
if (fh == NULL)
return CF_PRINT_OPEN_ERROR; /* attempt to open destination failed */
write_pdml_preamble(fh, cf->filename);
if (ferror(fh)) {
fclose(fh);
return CF_PRINT_WRITE_ERROR;
}
callback_args.fh = fh;
callback_args.print_args = print_args;
epan_dissect_init(&callback_args.edt, cf->epan, TRUE, TRUE);
/* Iterate through the list of packets, printing the packets we were
told to print. */
ret = process_specified_records(cf, &print_args->range, "Writing PDML",
"selected packets", TRUE,
write_pdml_packet, &callback_args, TRUE);
epan_dissect_cleanup(&callback_args.edt);
switch (ret) {
case PSP_FINISHED:
/* Completed successfully. */
break;
case PSP_STOPPED:
/* Well, the user decided to abort the printing. */
break;
case PSP_FAILED:
/* Error while printing. */
fclose(fh);
return CF_PRINT_WRITE_ERROR;
}
write_pdml_finale(fh);
if (ferror(fh)) {
fclose(fh);
return CF_PRINT_WRITE_ERROR;
}
/* XXX - check for an error */
fclose(fh);
return CF_PRINT_OK;
}
static gboolean
write_psml_packet(capture_file *cf, frame_data *fdata, wtap_rec *rec,
const guint8 *pd, void *argsp)
{
write_packet_callback_args_t *args = (write_packet_callback_args_t *)argsp;
/* Fill in the column information */
col_custom_prime_edt(&args->edt, &cf->cinfo);
epan_dissect_run(&args->edt, cf->cd_t, rec,
frame_tvbuff_new(&cf->provider, fdata, pd),
fdata, &cf->cinfo);
epan_dissect_fill_in_columns(&args->edt, FALSE, TRUE);
/* Write out the column information. */
write_psml_columns(&args->edt, args->fh, FALSE);
epan_dissect_reset(&args->edt);
return !ferror(args->fh);
}
cf_print_status_t
cf_write_psml_packets(capture_file *cf, print_args_t *print_args)
{
write_packet_callback_args_t callback_args;
FILE *fh;
psp_return_t ret;
gboolean proto_tree_needed;
fh = ws_fopen(print_args->file, "w");
if (fh == NULL)
return CF_PRINT_OPEN_ERROR; /* attempt to open destination failed */
write_psml_preamble(&cf->cinfo, fh);
if (ferror(fh)) {
fclose(fh);
return CF_PRINT_WRITE_ERROR;
}
callback_args.fh = fh;
callback_args.print_args = print_args;
/* Fill in the column information, only create the protocol tree
if having custom columns or field extractors. */
proto_tree_needed = have_custom_cols(&cf->cinfo) || have_field_extractors();
epan_dissect_init(&callback_args.edt, cf->epan, proto_tree_needed, proto_tree_needed);
/* Iterate through the list of packets, printing the packets we were
told to print. */
ret = process_specified_records(cf, &print_args->range, "Writing PSML",
"selected packets", TRUE,
write_psml_packet, &callback_args, TRUE);
epan_dissect_cleanup(&callback_args.edt);
switch (ret) {
case PSP_FINISHED:
/* Completed successfully. */
break;
case PSP_STOPPED:
/* Well, the user decided to abort the printing. */
break;
case PSP_FAILED:
/* Error while printing. */
fclose(fh);
return CF_PRINT_WRITE_ERROR;
}
write_psml_finale(fh);
if (ferror(fh)) {
fclose(fh);
return CF_PRINT_WRITE_ERROR;
}
/* XXX - check for an error */
fclose(fh);
return CF_PRINT_OK;
}
static gboolean
write_csv_packet(capture_file *cf, frame_data *fdata, wtap_rec *rec,
const guint8 *pd, void *argsp)
{
write_packet_callback_args_t *args = (write_packet_callback_args_t *)argsp;
/* Fill in the column information */
col_custom_prime_edt(&args->edt, &cf->cinfo);
epan_dissect_run(&args->edt, cf->cd_t, rec,
frame_tvbuff_new(&cf->provider, fdata, pd),
fdata, &cf->cinfo);
epan_dissect_fill_in_columns(&args->edt, FALSE, TRUE);
/* Write out the column information. */
write_csv_columns(&args->edt, args->fh);
epan_dissect_reset(&args->edt);
return !ferror(args->fh);
}
cf_print_status_t
cf_write_csv_packets(capture_file *cf, print_args_t *print_args)
{
write_packet_callback_args_t callback_args;
gboolean proto_tree_needed;
FILE *fh;
psp_return_t ret;
fh = ws_fopen(print_args->file, "w");
if (fh == NULL)
return CF_PRINT_OPEN_ERROR; /* attempt to open destination failed */
write_csv_column_titles(&cf->cinfo, fh);
if (ferror(fh)) {
fclose(fh);
return CF_PRINT_WRITE_ERROR;
}
callback_args.fh = fh;
callback_args.print_args = print_args;
/* only create the protocol tree if having custom columns or field extractors. */
proto_tree_needed = have_custom_cols(&cf->cinfo) || have_field_extractors();
epan_dissect_init(&callback_args.edt, cf->epan, proto_tree_needed, proto_tree_needed);
/* Iterate through the list of packets, printing the packets we were
told to print. */
ret = process_specified_records(cf, &print_args->range, "Writing CSV",
"selected packets", TRUE,
write_csv_packet, &callback_args, TRUE);
epan_dissect_cleanup(&callback_args.edt);
switch (ret) {
case PSP_FINISHED:
/* Completed successfully. */
break;
case PSP_STOPPED:
/* Well, the user decided to abort the printing. */
break;
case PSP_FAILED:
/* Error while printing. */
fclose(fh);
return CF_PRINT_WRITE_ERROR;
}
/* XXX - check for an error */
fclose(fh);
return CF_PRINT_OK;
}
static gboolean
carrays_write_packet(capture_file *cf, frame_data *fdata, wtap_rec *rec,
const guint8 *pd, void *argsp)
{
write_packet_callback_args_t *args = (write_packet_callback_args_t *)argsp;
epan_dissect_run(&args->edt, cf->cd_t, rec,
frame_tvbuff_new(&cf->provider, fdata, pd), fdata, NULL);
write_carrays_hex_data(fdata->num, args->fh, &args->edt);
epan_dissect_reset(&args->edt);
return !ferror(args->fh);
}
cf_print_status_t
cf_write_carrays_packets(capture_file *cf, print_args_t *print_args)
{
write_packet_callback_args_t callback_args;
FILE *fh;
psp_return_t ret;
fh = ws_fopen(print_args->file, "w");
if (fh == NULL)
return CF_PRINT_OPEN_ERROR; /* attempt to open destination failed */
if (ferror(fh)) {
fclose(fh);
return CF_PRINT_WRITE_ERROR;
}
callback_args.fh = fh;
callback_args.print_args = print_args;
epan_dissect_init(&callback_args.edt, cf->epan, TRUE, TRUE);
/* Iterate through the list of packets, printing the packets we were
told to print. */
ret = process_specified_records(cf, &print_args->range,
"Writing C Arrays",
"selected packets", TRUE,
carrays_write_packet, &callback_args, TRUE);
epan_dissect_cleanup(&callback_args.edt);
switch (ret) {
case PSP_FINISHED:
/* Completed successfully. */
break;
case PSP_STOPPED:
/* Well, the user decided to abort the printing. */
break;
case PSP_FAILED:
/* Error while printing. */
fclose(fh);
return CF_PRINT_WRITE_ERROR;
}
fclose(fh);
return CF_PRINT_OK;
}
static gboolean
write_json_packet(capture_file *cf, frame_data *fdata, wtap_rec *rec,
const guint8 *pd, void *argsp)
{
write_packet_callback_args_t *args = (write_packet_callback_args_t *)argsp;
/* Create the protocol tree, but don't fill in the column information. */
epan_dissect_run(&args->edt, cf->cd_t, rec,
frame_tvbuff_new(&cf->provider, fdata, pd), fdata, NULL);
/* Write out the information in that tree. */
write_json_proto_tree(NULL, args->print_args->print_dissections,
args->print_args->print_hex, NULL, PF_NONE,
&args->edt, &cf->cinfo, proto_node_group_children_by_unique, args->fh);
epan_dissect_reset(&args->edt);
return !ferror(args->fh);
}
cf_print_status_t
cf_write_json_packets(capture_file *cf, print_args_t *print_args)
{
write_packet_callback_args_t callback_args;
FILE *fh;
psp_return_t ret;
fh = ws_fopen(print_args->file, "w");
if (fh == NULL)
return CF_PRINT_OPEN_ERROR; /* attempt to open destination failed */
write_json_preamble(fh);
if (ferror(fh)) {
fclose(fh);
return CF_PRINT_WRITE_ERROR;
}
callback_args.fh = fh;
callback_args.print_args = print_args;
epan_dissect_init(&callback_args.edt, cf->epan, TRUE, TRUE);
/* Iterate through the list of packets, printing the packets we were
told to print. */
ret = process_specified_records(cf, &print_args->range, "Writing PDML",
"selected packets", TRUE,
write_json_packet, &callback_args, TRUE);
epan_dissect_cleanup(&callback_args.edt);
switch (ret) {
case PSP_FINISHED:
/* Completed successfully. */
break;
case PSP_STOPPED:
/* Well, the user decided to abort the printing. */
break;
case PSP_FAILED:
/* Error while printing. */
fclose(fh);
return CF_PRINT_WRITE_ERROR;
}
write_json_finale(fh);
if (ferror(fh)) {
fclose(fh);
return CF_PRINT_WRITE_ERROR;
}
/* XXX - check for an error */
fclose(fh);
return CF_PRINT_OK;
}
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
gboolean
cf_find_packet_protocol_tree(capture_file *cf, const char *string,
search_direction dir)
{
match_data mdata;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
mdata.string = string;
mdata.string_len = strlen(string);
return find_packet(cf, match_protocol_tree, &mdata, dir);
}
gboolean
cf_find_string_protocol_tree(capture_file *cf, proto_tree *tree, match_data *mdata)
{
mdata->frame_matched = FALSE;
mdata->string = convert_string_case(cf->sfilter, cf->case_type);
mdata->string_len = strlen(mdata->string);
mdata->cf = cf;
/* Iterate through all the nodes looking for matching text */
proto_tree_children_foreach(tree, match_subtree_text, mdata);
return mdata->frame_matched ? MR_MATCHED : MR_NOTMATCHED;
}
static match_result
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
match_protocol_tree(capture_file *cf, frame_data *fdata, void *criterion)
{
match_data *mdata = (match_data *)criterion;
epan_dissect_t edt;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
/* Load the frame's data. */
if (!cf_read_record(cf, fdata)) {
/* Attempt to get the packet failed. */
return MR_ERROR;
}
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
/* Construct the protocol tree, including the displayed text */
epan_dissect_init(&edt, cf->epan, TRUE, TRUE);
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
/* We don't need the column information */
epan_dissect_run(&edt, cf->cd_t, &cf->rec,
frame_tvbuff_new_buffer(&cf->provider, fdata, &cf->buf),
fdata, NULL);
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
/* Iterate through all the nodes, seeing if they have text that matches. */
mdata->cf = cf;
mdata->frame_matched = FALSE;
proto_tree_children_foreach(edt.tree, match_subtree_text, mdata);
epan_dissect_cleanup(&edt);
return mdata->frame_matched ? MR_MATCHED : MR_NOTMATCHED;
}
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
static void
match_subtree_text(proto_node *node, gpointer data)
{
match_data *mdata = (match_data *) data;
const gchar *string = mdata->string;
size_t string_len = mdata->string_len;
capture_file *cf = mdata->cf;
field_info *fi = PNODE_FINFO(node);
gchar label_str[ITEM_LABEL_LENGTH];
gchar *label_ptr;
size_t label_len;
guint32 i;
guint8 c_char;
size_t c_match = 0;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
/* dissection with an invisible proto tree? */
g_assert(fi);
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
if (mdata->frame_matched) {
/* We already had a match; don't bother doing any more work. */
return;
}
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
/* Don't match invisible entries. */
if (PROTO_ITEM_IS_HIDDEN(node))
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
return;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
/* was a free format label produced? */
if (fi->rep) {
label_ptr = fi->rep->representation;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
} else {
/* no, make a generic label */
label_ptr = label_str;
proto_item_fill_label(fi, label_str);
}
if (cf->regex) {
if (g_regex_match(cf->regex, label_ptr, (GRegexMatchFlags) 0, NULL)) {
mdata->frame_matched = TRUE;
mdata->finfo = fi;
return;
}
} else {
/* Does that label match? */
label_len = strlen(label_ptr);
for (i = 0; i < label_len; i++) {
c_char = label_ptr[i];
if (cf->case_type)
c_char = g_ascii_toupper(c_char);
if (c_char == string[c_match]) {
c_match++;
if (c_match == string_len) {
/* No need to look further; we have a match */
mdata->frame_matched = TRUE;
mdata->finfo = fi;
return;
}
} else
c_match = 0;
}
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
}
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
/* Recurse into the subtree, if it exists */
if (node->first_child != NULL)
proto_tree_children_foreach(node, match_subtree_text, mdata);
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
}
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
gboolean
cf_find_packet_summary_line(capture_file *cf, const char *string,
search_direction dir)
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
{
match_data mdata;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
mdata.string = string;
mdata.string_len = strlen(string);
return find_packet(cf, match_summary_line, &mdata, dir);
}
static match_result
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
match_summary_line(capture_file *cf, frame_data *fdata, void *criterion)
{
match_data *mdata = (match_data *)criterion;
const gchar *string = mdata->string;
size_t string_len = mdata->string_len;
epan_dissect_t edt;
const char *info_column;
size_t info_column_len;
match_result result = MR_NOTMATCHED;
gint colx;
guint32 i;
guint8 c_char;
size_t c_match = 0;
/* Load the frame's data. */
if (!cf_read_record(cf, fdata)) {
/* Attempt to get the packet failed. */
return MR_ERROR;
}
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
/* Don't bother constructing the protocol tree */
epan_dissect_init(&edt, cf->epan, FALSE, FALSE);
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
/* Get the column information */
epan_dissect_run(&edt, cf->cd_t, &cf->rec,
frame_tvbuff_new_buffer(&cf->provider, fdata, &cf->buf),
fdata, &cf->cinfo);
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
/* Find the Info column */
for (colx = 0; colx < cf->cinfo.num_cols; colx++) {
if (cf->cinfo.columns[colx].fmt_matx[COL_INFO]) {
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
/* Found it. See if we match. */
info_column = edt.pi.cinfo->columns[colx].col_data;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
info_column_len = strlen(info_column);
if (cf->regex) {
if (g_regex_match(cf->regex, info_column, (GRegexMatchFlags) 0, NULL)) {
result = MR_MATCHED;
break;
}
} else {
for (i = 0; i < info_column_len; i++) {
c_char = info_column[i];
if (cf->case_type)
c_char = g_ascii_toupper(c_char);
if (c_char == string[c_match]) {
c_match++;
if (c_match == string_len) {
result = MR_MATCHED;
break;
}
} else
c_match = 0;
}
}
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
break;
}
}
epan_dissect_cleanup(&edt);
return result;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
}
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
typedef struct {
const guint8 *data;
size_t data_len;
} cbs_t; /* "Counted byte string" */
/*
* The current match_* routines only support ASCII case insensitivity and don't
* convert UTF-8 inputs to UTF-16 for matching.
*
* We could modify them to use the GLib Unicode routines or the International
* Components for Unicode library but it's not apparent that we could do so
* without consuming a lot more CPU and memory or that searching would be
* significantly better.
*/
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
gboolean
cf_find_packet_data(capture_file *cf, const guint8 *string, size_t string_size,
search_direction dir)
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
{
cbs_t info;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
info.data = string;
info.data_len = string_size;
/* Regex, String or hex search? */
if (cf->regex) {
/* Regular Expression search */
return find_packet(cf, match_regex, NULL, dir);
} else if (cf->string) {
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
/* String search - what type of string? */
switch (cf->scs_type) {
case SCS_NARROW_AND_WIDE:
return find_packet(cf, match_narrow_and_wide, &info, dir);
case SCS_NARROW:
return find_packet(cf, match_narrow, &info, dir);
case SCS_WIDE:
return find_packet(cf, match_wide, &info, dir);
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
default:
g_assert_not_reached();
return FALSE;
}
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
} else
return find_packet(cf, match_binary, &info, dir);
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
}
static match_result
match_narrow_and_wide(capture_file *cf, frame_data *fdata, void *criterion)
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
{
cbs_t *info = (cbs_t *)criterion;
const guint8 *ascii_text = info->data;
size_t textlen = info->data_len;
match_result result;
guint32 buf_len;
guint8 *pd;
guint32 i;
guint8 c_char;
size_t c_match = 0;
/* Load the frame's data. */
if (!cf_read_record(cf, fdata)) {
/* Attempt to get the packet failed. */
return MR_ERROR;
}
result = MR_NOTMATCHED;
buf_len = fdata->cap_len;
pd = ws_buffer_start_ptr(&cf->buf);
i = 0;
while (i < buf_len) {
c_char = pd[i];
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
if (cf->case_type)
c_char = g_ascii_toupper(c_char);
if (c_char != '\0') {
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
if (c_char == ascii_text[c_match]) {
c_match += 1;
if (c_match == textlen) {
result = MR_MATCHED;
cf->search_pos = i; /* Save the position of the last character
for highlighting the field. */
cf->search_len = (guint32)textlen;
break;
}
}
else {
g_assert(i>=c_match);
i -= (guint32)c_match;
c_match = 0;
}
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
}
i += 1;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
}
return result;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
}
static match_result
match_narrow(capture_file *cf, frame_data *fdata, void *criterion)
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
{
guint8 *pd;
cbs_t *info = (cbs_t *)criterion;
const guint8 *ascii_text = info->data;
size_t textlen = info->data_len;
match_result result;
guint32 buf_len;
guint32 i;
guint8 c_char;
size_t c_match = 0;
/* Load the frame's data. */
if (!cf_read_record(cf, fdata)) {
/* Attempt to get the packet failed. */
return MR_ERROR;
}
result = MR_NOTMATCHED;
buf_len = fdata->cap_len;
pd = ws_buffer_start_ptr(&cf->buf);
i = 0;
while (i < buf_len) {
c_char = pd[i];
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
if (cf->case_type)
c_char = g_ascii_toupper(c_char);
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
if (c_char == ascii_text[c_match]) {
c_match += 1;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
if (c_match == textlen) {
result = MR_MATCHED;
cf->search_pos = i; /* Save the position of the last character
for highlighting the field. */
cf->search_len = (guint32)textlen;
break;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
}
}
else {
g_assert(i>=c_match);
i -= (guint32)c_match;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
c_match = 0;
}
i += 1;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
}
return result;
}
static match_result
match_wide(capture_file *cf, frame_data *fdata, void *criterion)
{
cbs_t *info = (cbs_t *)criterion;
const guint8 *ascii_text = info->data;
size_t textlen = info->data_len;
match_result result;
guint32 buf_len;
guint8 *pd;
guint32 i;
guint8 c_char;
size_t c_match = 0;
/* Load the frame's data. */
if (!cf_read_record(cf, fdata)) {
/* Attempt to get the packet failed. */
return MR_ERROR;
}
result = MR_NOTMATCHED;
buf_len = fdata->cap_len;
pd = ws_buffer_start_ptr(&cf->buf);
i = 0;
while (i < buf_len) {
c_char = pd[i];
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
if (cf->case_type)
c_char = g_ascii_toupper(c_char);
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
if (c_char == ascii_text[c_match]) {
c_match += 1;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
if (c_match == textlen) {
result = MR_MATCHED;
cf->search_pos = i; /* Save the position of the last character
for highlighting the field. */
cf->search_len = (guint32)textlen;
break;
}
i += 1;
}
else {
g_assert(i>=(c_match*2));
i -= (guint32)c_match*2;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
c_match = 0;
}
i += 1;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
}
return result;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
}
static match_result
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
match_binary(capture_file *cf, frame_data *fdata, void *criterion)
{
cbs_t *info = (cbs_t *)criterion;
const guint8 *binary_data = info->data;
size_t datalen = info->data_len;
match_result result;
guint32 buf_len;
guint8 *pd;
guint32 i;
size_t c_match = 0;
/* Load the frame's data. */
if (!cf_read_record(cf, fdata)) {
/* Attempt to get the packet failed. */
return MR_ERROR;
}
result = MR_NOTMATCHED;
buf_len = fdata->cap_len;
pd = ws_buffer_start_ptr(&cf->buf);
i = 0;
while (i < buf_len) {
if (pd[i] == binary_data[c_match]) {
c_match += 1;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
if (c_match == datalen) {
result = MR_MATCHED;
cf->search_pos = i; /* Save the position of the last character
for highlighting the field. */
cf->search_len = (guint32)datalen;
break;
}
}
else {
g_assert(i>=c_match);
i -= (guint32)c_match;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
c_match = 0;
}
i += 1;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
}
return result;
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
}
static match_result
match_regex(capture_file *cf, frame_data *fdata, void *criterion _U_)
{
match_result result = MR_NOTMATCHED;
GMatchInfo *match_info = NULL;
/* Load the frame's data. */
if (!cf_read_record(cf, fdata)) {
/* Attempt to get the packet failed. */
return MR_ERROR;
}
if (g_regex_match_full(cf->regex, (const gchar *)ws_buffer_start_ptr(&cf->buf), fdata->cap_len,
0, (GRegexMatchFlags) 0, &match_info, NULL))
{
gint start_pos = 0, end_pos = 0;
g_match_info_fetch_pos (match_info, 0, &start_pos, &end_pos);
cf->search_pos = end_pos - 1;
cf->search_len = end_pos - start_pos;
result = MR_MATCHED;
}
return result;
}
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
gboolean
cf_find_packet_dfilter(capture_file *cf, dfilter_t *sfcode,
search_direction dir)
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
{
return find_packet(cf, match_dfilter, sfcode, dir);
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
}
gboolean
cf_find_packet_dfilter_string(capture_file *cf, const char *filter,
search_direction dir)
{
dfilter_t *sfcode;
gboolean result;
if (!dfilter_compile(filter, &sfcode, NULL)) {
/*
* XXX - this shouldn't happen, as the filter string is machine
* generated
*/
return FALSE;
}
if (sfcode == NULL) {
/*
* XXX - this shouldn't happen, as the filter string is machine
* generated.
*/
return FALSE;
}
result = find_packet(cf, match_dfilter, sfcode, dir);
dfilter_free(sfcode);
return result;
}
static match_result
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
match_dfilter(capture_file *cf, frame_data *fdata, void *criterion)
{
dfilter_t *sfcode = (dfilter_t *)criterion;
epan_dissect_t edt;
match_result result;
/* Load the frame's data. */
if (!cf_read_record(cf, fdata)) {
/* Attempt to get the packet failed. */
return MR_ERROR;
}
epan_dissect_init(&edt, cf->epan, TRUE, FALSE);
epan_dissect_prime_with_dfilter(&edt, sfcode);
epan_dissect_run(&edt, cf->cd_t, &cf->rec,
frame_tvbuff_new_buffer(&cf->provider, fdata, &cf->buf),
fdata, NULL);
result = dfilter_apply_edt(sfcode, &edt) ? MR_MATCHED : MR_NOTMATCHED;
epan_dissect_cleanup(&edt);
return result;
}
gboolean
cf_find_packet_marked(capture_file *cf, search_direction dir)
{
return find_packet(cf, match_marked, NULL, dir);
}
static match_result
match_marked(capture_file *cf _U_, frame_data *fdata, void *criterion _U_)
{
return fdata->flags.marked ? MR_MATCHED : MR_NOTMATCHED;
}
gboolean
cf_find_packet_time_reference(capture_file *cf, search_direction dir)
{
return find_packet(cf, match_time_reference, NULL, dir);
}
static match_result
match_time_reference(capture_file *cf _U_, frame_data *fdata, void *criterion _U_)
{
return fdata->flags.ref_time ? MR_MATCHED : MR_NOTMATCHED;
}
Get rid of the EBCDIC stuff in the find dialog - it's not supported yet, so we shouldn't torment the users by offering it. Check the string type and convert it to an internal representation in the GUI code; have the search code deal only with the internal representation. Save the case-sensitivity flag, and the indication of where string searches look, along with other search parameters. Upper-casify the string, for case-insensitive searches, in the GUI code; don't save the upper-casified string, so it doesn't SHOUT at you when you next pop up a "find" dialog. Convert the hex value string to raw binary data in the GUI code, rather than doing so in the search code. Check that it's a valid string. Connect the signals to the radio buttons after the pointers have been attached to various GUI items - the signal handlers expect some of those pointers to be attached, and aren't happy if they're not. Have "find_packet()" contain a framework for searching, but not contain the matching code; instead, pass it a pointer to a matching routine and an opaque pointer to be passed to the matching routine. Have all the routines that do different types of searching have their own matching routines, and use the common "find_packet()" code, rather than duplicating that code. Search for the Info column by column type, not by name (the user can change the name). When matching on the protocol tree, don't format the entire protocol tree into a big buffer - just have a routine that matches the text representation of a protocol tree item against a string, and, if it finds a match, sets a "we found a match flag" and returns; have that routine not bother doing any more work if that flag is set. (Unfortunately, you can't abort "g_node_children_foreach()" in the middle of a traversal.) Free the generated display filter code after a find-by-display-filter finishes. svn path=/trunk/; revision=8306
2003-08-29 04:03:46 +00:00
static gboolean
find_packet(capture_file *cf,
match_result (*match_function)(capture_file *, frame_data *, void *),
void *criterion, search_direction dir)
{
frame_data *start_fd;
guint32 framenum;
guint32 prev_framenum;
frame_data *fdata;
frame_data *new_fd = NULL;
progdlg_t *progbar = NULL;
GTimer *prog_timer = g_timer_new();
int count;
gboolean found;
float progbar_val;
GTimeVal start_time;
gchar status_str[100];
const char *title;
match_result result;
start_fd = cf->current_frame;
if (start_fd != NULL) {
prev_framenum = start_fd->num;
} else {
prev_framenum = 0; /* No start packet selected. */
}
/* Iterate through the list of packets, starting at the packet we've
picked, calling a routine to run the filter on the packet, see if
it matches, and stop if so. */
count = 0;
framenum = prev_framenum;
g_timer_start(prog_timer);
/* Progress so far. */
progbar_val = 0.0f;
cf->stop_flag = FALSE;
g_get_current_time(&start_time);
title = cf->sfilter?cf->sfilter:"";
for (;;) {
/* Create the progress bar if necessary.
We check on every iteration of the loop, so that it takes no
longer than the standard time to create it (otherwise, for a
large file, we might take considerably longer than that standard
time in order to get to the next progress bar step). */
if (progbar == NULL)
progbar = delayed_create_progress_dlg(cf->window, "Searching", title,
FALSE, &cf->stop_flag, &start_time, progbar_val);
/*
* Update the progress bar, but do it only after PROGBAR_UPDATE_INTERVAL
* has elapsed. Calling update_progress_dlg and packets_bar_update will
* likely trigger UI paint events, which might take a while depending on
* the platform and display. Reset our timer *after* painting.
*/
if (g_timer_elapsed(prog_timer, NULL) > PROGBAR_UPDATE_INTERVAL) {
/* let's not divide by zero. I should never be started
* with count == 0, so let's assert that
*/
g_assert(cf->count > 0);
progbar_val = (gfloat) count / cf->count;
g_snprintf(status_str, sizeof(status_str),
"%4u of %u packets", count, cf->count);
update_progress_dlg(progbar, progbar_val, status_str);
g_timer_start(prog_timer);
}
if (cf->stop_flag) {
/* Well, the user decided to abort the search. Go back to the
frame where we started. */
new_fd = start_fd;
break;
}
/* Go past the current frame. */
if (dir == SD_BACKWARD) {
/* Go on to the previous frame. */
if (framenum <= 1) {
/*
* XXX - other apps have a bit more of a detailed message
* for this, and instead of offering "OK" and "Cancel",
* they offer things such as "Continue" and "Cancel";
* we need an API for popping up alert boxes with
* {Verb} and "Cancel".
*/
if (prefs.gui_find_wrap) {
statusbar_push_temporary_msg("Search reached the beginning. Continuing at end.");
framenum = cf->count; /* wrap around */
} else {
statusbar_push_temporary_msg("Search reached the beginning.");
framenum = prev_framenum; /* stay on previous packet */
}
} else
framenum--;
} else {
/* Go on to the next frame. */
if (framenum == cf->count) {
if (prefs.gui_find_wrap) {
statusbar_push_temporary_msg("Search reached the end. Continuing at beginning.");
framenum = 1; /* wrap around */
} else {
statusbar_push_temporary_msg("Search reached the end.");
framenum = prev_framenum; /* stay on previous packet */
}
} else
framenum++;
}
fdata = frame_data_sequence_find(cf->provider.frames, framenum);
count++;
/* Is this packet in the display? */
if (fdata && fdata->flags.passed_dfilter) {
/* Yes. Does it match the search criterion? */
result = (*match_function)(cf, fdata, criterion);
if (result == MR_ERROR) {
/* Error; our caller has reported the error. Go back to the frame
where we started. */
new_fd = start_fd;
break;
} else if (result == MR_MATCHED) {
/* Yes. Go to the new frame. */
new_fd = fdata;
break;
}
}
if (fdata == start_fd) {
/* We're back to the frame we were on originally, and that frame
doesn't match the search filter. The search failed. */
break;
}
}
/* We're done scanning the packets; destroy the progress bar if it
was created. */
if (progbar != NULL)
destroy_progress_dlg(progbar);
g_timer_destroy(prog_timer);
if (new_fd != NULL) {
/* Find and select */
cf->search_in_progress = TRUE;
found = packet_list_select_row_from_data(new_fd);
cf->search_in_progress = FALSE;
cf->search_pos = 0; /* Reset the position */
cf->search_len = 0; /* Reset length */
if (!found) {
/* We didn't find a row corresponding to this frame.
This means that the frame isn't being displayed currently,
so we can't select it. */
simple_message_box(ESD_TYPE_INFO, NULL,
"The capture file is probably not fully dissected.",
"End of capture exceeded.");
return FALSE;
}
return TRUE; /* success */
} else
return FALSE; /* failure */
}
gboolean
cf_goto_frame(capture_file *cf, guint fnumber)
{
frame_data *fdata;
if (cf == NULL || cf->provider.frames == NULL) {
/* we don't have a loaded capture file - fix for bugs 11810 & 11989 */
statusbar_push_temporary_msg("There is no file loaded");
return FALSE; /* we failed to go to that packet */
}
fdata = frame_data_sequence_find(cf->provider.frames, fnumber);
if (fdata == NULL) {
/* we didn't find a packet with that packet number */
statusbar_push_temporary_msg("There is no packet number %u.", fnumber);
return FALSE; /* we failed to go to that packet */
}
if (!fdata->flags.passed_dfilter) {
/* that packet currently isn't displayed */
/* XXX - add it to the set of displayed packets? */
statusbar_push_temporary_msg("Packet number %u isn't displayed.", fnumber);
return FALSE; /* we failed to go to that packet */
}
if (!packet_list_select_row_from_data(fdata)) {
/* We didn't find a row corresponding to this frame.
This means that the frame isn't being displayed currently,
so we can't select it. */
simple_message_box(ESD_TYPE_INFO, NULL,
"The capture file is probably not fully dissected.",
"End of capture exceeded.");
return FALSE;
}
return TRUE; /* we got to that packet */
}
/*
* Go to frame specified by currently selected protocol tree item.
*/
gboolean
cf_goto_framenum(capture_file *cf)
{
header_field_info *hfinfo;
guint32 framenum;
if (cf->finfo_selected) {
hfinfo = cf->finfo_selected->hfinfo;
g_assert(hfinfo);
if (hfinfo->type == FT_FRAMENUM) {
framenum = fvalue_get_uinteger(&cf->finfo_selected->value);
if (framenum != 0)
return cf_goto_frame(cf, framenum);
}
}
return FALSE;
}
/* Select the packet on a given row. */
void
cf_select_packet(capture_file *cf, int row)
{
epan_dissect_t *old_edt;
frame_data *fdata;
We already set the foreground and background color for every frame, which means we're already doing a "do something to the last row in the packet list" operation on every frame we add to the list, so adding a call to "gtk_clist_set_row_data()" won't make matters worse. In addition, we already set one column in a row on a "change time format" operation, so finding the row for a frame by calling "gtk_clist_find_row_from_data()" doesn't turn a constant-time operation into a linear-time operation, it just cranks the proportionality constant up - it was quadratic before, alas, and it's still quadratic. Adding calls to "gtk_clist_find_row_from_data()" to the "Find Frame" and "Go To Frame" code does add an extra linear operation there, but those operations shouldn't be common - and "Go To Frame", going to the last frame on an ~100,000-frame big capture file, was quick, at least on my 450 MHz Pentium II machine, so maybe it won't be too bad. And "select_packet()" either has to search the frame table for the frame with the specified row number, or has to call "gtk_clist_get_row_data()" to do that - the first is linear in the position of the frame in the frame table, and the latter is linear in its position in the CList, and the latter is less than or equal to the former, so the only thing making it worse would be a change in the proportionality constant. So it probably won't hurt performance by much. Furthermore, if we add the ability to sort the display on an arbitrary column, or to delete frames from the display - both of which are in the wish list - storing the row number of the frame in the "frame_data" structure won't necessarily work, as the row number can change out from under us. Therefore, reinstate the old way of doing things, where we associate with each row a pointer to the "frame_data" structure for the row, using "gtk_clist_set_row_data()". svn path=/trunk/; revision=1703
2000-03-08 06:48:01 +00:00
/* Get the frame data struct pointer for this frame */
fdata = packet_list_get_row_data(row);
if (fdata == NULL) {
return;
}
/* Get the data in that frame. */
if (!cf_read_record (cf, fdata)) {
return;
}
We can't trust "cf->current_frame" to refer to the frame that was selected before we started re-colorizing or re-filtering the display, as when the first row is added to the clist, that may be selected and thus made the current frame. This means that we can't find the row corresponding to the previously-selected frame, if any, by checking as each packet is colorized/filtered and see whether its "frame_data" structure is equal to "cf->current_frame", as that'll always say that the first frame in the display is the selected frame. Instead, we recored the value of "cf->current_frame" before we do anything to the clist, have "add_packet_to_packet_list()" return either the row number of the frame (if it passed the filter and thus was added to the clist) or -1 (if it didn't pass the filter and thus wasn't added to the clist), and, after "add_packet_to_packet_list()", if the current frame is the one that was the selected row, remember its row number (if any), and, when we're finished colorizing/filtering the display, make that row the current row if it's not -1 (-1 means that the selected row didn't pass the filter). Also, don't do that until after we've thawed the clist, as the vertical adjustment for the clist doesn't reflect reality until then, and attempting to go to a given row won't work right until the vertical adjustment for the clist reflects reality. Shove all the code to set the selected and focus rows, and to make said row visible, into a routine, so the "Find Frame" and "Go To Frame" code can use it as well. svn path=/trunk/; revision=1959
2000-05-15 01:50:16 +00:00
/* Record that this frame is the current frame. */
cf->current_frame = fdata;
cf->current_row = row;
old_edt = cf->edt;
/* Create the logical protocol tree. */
/* We don't need the columns here. */
cf->edt = epan_dissect_new(cf->epan, TRUE, TRUE);
tap_build_interesting(cf->edt);
epan_dissect_run(cf->edt, cf->cd_t, &cf->rec,
frame_tvbuff_new_buffer(&cf->provider, cf->current_frame, &cf->buf),
cf->current_frame, NULL);
dfilter_macro_build_ftv_cache(cf->edt->tree);
if (old_edt != NULL)
epan_dissect_free(old_edt);
}
/* Unselect the selected packet, if any. */
void
cf_unselect_packet(capture_file *cf)
{
epan_dissect_t *old_edt = cf->edt;
cf->edt = NULL;
/* No packet is selected. */
cf->current_frame = NULL;
cf->current_row = 0;
/* Destroy the epan_dissect_t for the unselected packet. */
if (old_edt != NULL)
epan_dissect_free(old_edt);
}
/*
* Mark a particular frame.
*/
void
cf_mark_frame(capture_file *cf, frame_data *frame)
{
if (! frame->flags.marked) {
frame->flags.marked = TRUE;
if (cf->count > cf->marked_count)
cf->marked_count++;
}
}
/*
* Unmark a particular frame.
*/
void
cf_unmark_frame(capture_file *cf, frame_data *frame)
{
if (frame->flags.marked) {
frame->flags.marked = FALSE;
if (cf->marked_count > 0)
cf->marked_count--;
}
}
/*
* Ignore a particular frame.
*/
void
cf_ignore_frame(capture_file *cf, frame_data *frame)
{
if (! frame->flags.ignored) {
frame->flags.ignored = TRUE;
if (cf->count > cf->ignored_count)
cf->ignored_count++;
}
}
/*
* Un-ignore a particular frame.
*/
void
cf_unignore_frame(capture_file *cf, frame_data *frame)
{
if (frame->flags.ignored) {
frame->flags.ignored = FALSE;
if (cf->ignored_count > 0)
cf->ignored_count--;
}
}
/*
* Read the section comment.
*/
const gchar *
cf_read_section_comment(capture_file *cf)
{
wtap_block_t shb_inf;
char *shb_comment;
/* Get the SHB. */
/* XXX - support multiple SHBs */
shb_inf = wtap_file_get_shb(cf->provider.wth);
/* Get the first comment from the SHB. */
/* XXX - support multiple comments */
if (wtap_block_get_nth_string_option_value(shb_inf, OPT_COMMENT, 0, &shb_comment) != WTAP_OPTTYPE_SUCCESS)
return NULL;
return shb_comment;
}
/*
* Modify the section comment.
*/
void
cf_update_section_comment(capture_file *cf, gchar *comment)
{
wtap_block_t shb_inf;
gchar *shb_comment;
/* Get the SHB. */
/* XXX - support multiple SHBs */
shb_inf = wtap_file_get_shb(cf->provider.wth);
/* Get the first comment from the SHB. */
/* XXX - support multiple comments */
if (wtap_block_get_nth_string_option_value(shb_inf, OPT_COMMENT, 0, &shb_comment) != WTAP_OPTTYPE_SUCCESS) {
/* There's no comment - add one. */
wtap_block_add_string_option(shb_inf, OPT_COMMENT, comment, strlen(comment));
} else {
/* See if the comment has changed or not */
if (strcmp(shb_comment, comment) == 0) {
g_free(comment);
return;
}
/* The comment has changed, let's update it */
wtap_block_set_nth_string_option_value(shb_inf, OPT_COMMENT, 0, comment, strlen(comment));
}
/* Mark the file as having unsaved changes */
cf->unsaved_changes = TRUE;
}
/*
* Get the comment on a packet (record).
* If the comment has been edited, it returns the result of the edit,
* otherwise it returns the comment from the file.
*/
char *
cf_get_packet_comment(capture_file *cf, const frame_data *fd)
{
char *comment;
/* fetch user comment */
if (fd->flags.has_user_comment)
return g_strdup(cap_file_provider_get_user_comment(&cf->provider, fd));
/* fetch phdr comment */
if (fd->flags.has_phdr_comment) {
wtap_rec rec; /* Record metadata */
Buffer buf; /* Record data */
wtap_rec_init(&rec);
ws_buffer_init(&buf, 1500);
if (!cf_read_record_r(cf, fd, &rec, &buf))
{ /* XXX, what we can do here? */ }
comment = rec.opt_comment;
wtap_rec_cleanup(&rec);
ws_buffer_free(&buf);
return comment;
}
return NULL;
}
/*
* Update(replace) the comment on a capture from a frame
*/
gboolean
cf_set_user_packet_comment(capture_file *cf, frame_data *fd, const gchar *new_comment)
{
char *pkt_comment = cf_get_packet_comment(cf, fd);
/* Check if the comment has changed */
if (!g_strcmp0(pkt_comment, new_comment)) {
g_free(pkt_comment);
return FALSE;
}
g_free(pkt_comment);
if (pkt_comment)
cf->packet_comment_count--;
if (new_comment)
cf->packet_comment_count++;
cap_file_provider_set_user_comment(&cf->provider, fd, new_comment);
expert_update_comment_count(cf->packet_comment_count);
/* OK, we have unsaved changes. */
cf->unsaved_changes = TRUE;
return TRUE;
}
/*
* What types of comments does this capture file have?
*/
guint32
cf_comment_types(capture_file *cf)
{
guint32 comment_types = 0;
if (cf_read_section_comment(cf) != NULL)
comment_types |= WTAP_COMMENT_PER_SECTION;
if (cf->packet_comment_count != 0)
comment_types |= WTAP_COMMENT_PER_PACKET;
return comment_types;
}
/*
* Add a resolved address to this file's list of resolved addresses.
*/
gboolean
cf_add_ip_name_from_string(capture_file *cf, const char *addr, const char *name)
{
/*
* XXX - support multiple resolved address lists, and add to the one
* attached to this file?
*/
if (!add_ip_name_from_string(addr, name))
return FALSE;
/* OK, we have unsaved changes. */
cf->unsaved_changes = TRUE;
return TRUE;
}
typedef struct {
wtap_dumper *pdh;
const char *fname;
int file_type;
} save_callback_args_t;
/*
* Save a capture to a file, in a particular format, saving either
* all packets, all currently-displayed packets, or all marked packets.
*
* Returns TRUE if it succeeds, FALSE otherwise; if it fails, it pops
* up a message box for the failure.
*/
static gboolean
save_record(capture_file *cf, frame_data *fdata, wtap_rec *rec,
const guint8 *pd, void *argsp)
{
save_callback_args_t *args = (save_callback_args_t *)argsp;
wtap_rec new_rec;
int err;
gchar *err_info;
const char *pkt_comment;
/* Copy the record information from what was read in from the file. */
new_rec = *rec;
/* Make changes based on anything that the user has done but that
hasn't been saved yet. */
if (fdata->flags.has_user_comment)
pkt_comment = cap_file_provider_get_user_comment(&cf->provider, fdata);
else
pkt_comment = rec->opt_comment;
new_rec.opt_comment = g_strdup(pkt_comment);
new_rec.has_comment_changed = fdata->flags.has_user_comment ? TRUE : FALSE;
/* XXX - what if times have been shifted? */
/* and save the packet */
if (!wtap_dump(args->pdh, &new_rec, pd, &err, &err_info)) {
cfile_write_failure_alert_box(NULL, args->fname, err, err_info, fdata->num,
args->file_type);
return FALSE;
}
g_free(new_rec.opt_comment);
return TRUE;
}
/*
* Can this capture file be written out in any format using Wiretap
* rather than by copying the raw data?
*/
gboolean
cf_can_write_with_wiretap(capture_file *cf)
{
/* We don't care whether we support the comments in this file or not;
if we can't, we'll offer the user the option of discarding the
comments. */
return wtap_dump_can_write(cf->linktypes, 0);
}
/*
* Should we let the user do a save?
*
* We should if:
*
* the file has unsaved changes, and we can save it in some
* format through Wiretap
*
* or
*
* the file is a temporary file and has no unsaved changes (so
* that "saving" it just means copying it).
*
* XXX - we shouldn't allow files to be edited if they can't be saved,
* so cf->unsaved_changes should be true only if the file can be saved.
*
* We don't care whether we support the comments in this file or not;
* if we can't, we'll offer the user the option of discarding the
* comments.
*/
gboolean
cf_can_save(capture_file *cf)
{
if (cf->unsaved_changes && wtap_dump_can_write(cf->linktypes, 0)) {
/* Saved changes, and we can write it out with Wiretap. */
return TRUE;
}
if (cf->is_tempfile && !cf->unsaved_changes) {
/*
* Temporary file with no unsaved changes, so we can just do a
* raw binary copy.
*/
return TRUE;
}
/* Nothing to save. */
return FALSE;
}
/*
* Should we let the user do a "save as"?
*
* That's true if:
*
* we can save it in some format through Wiretap
*
* or
*
* the file is a temporary file and has no unsaved changes (so
* that "saving" it just means copying it).
*
* XXX - we shouldn't allow files to be edited if they can't be saved,
* so cf->unsaved_changes should be true only if the file can be saved.
*
* We don't care whether we support the comments in this file or not;
* if we can't, we'll offer the user the option of discarding the
* comments.
*/
gboolean
cf_can_save_as(capture_file *cf)
{
if (wtap_dump_can_write(cf->linktypes, 0)) {
/* We can write it out with Wiretap. */
return TRUE;
}
if (cf->is_tempfile && !cf->unsaved_changes) {
/*
* Temporary file with no unsaved changes, so we can just do a
* raw binary copy.
*/
return TRUE;
}
/* Nothing to save. */
return FALSE;
}
/*
* Does this file have unsaved data?
*/
gboolean
cf_has_unsaved_data(capture_file *cf)
{
/*
* If this is a temporary file, or a file with unsaved changes, it
* has unsaved data.
*/
return (cf->is_tempfile && cf->count>0) || cf->unsaved_changes;
}
/*
* Quick scan to find packet offsets.
*/
static cf_read_status_t
rescan_file(capture_file *cf, const char *fname, gboolean is_tempfile)
{
const wtap_rec *rec;
int err;
gchar *err_info;
gchar *name_ptr;
gint64 data_offset;
progdlg_t *progbar = NULL;
GTimer *prog_timer = g_timer_new();
gint64 size;
float progbar_val;
GTimeVal start_time;
gchar status_str[100];
guint32 framenum;
frame_data *fdata;
int count = 0;
/* Close the old handle. */
wtap_close(cf->provider.wth);
/* Open the new file. */
/* XXX: this will go through all open_routines for a matching one. But right
now rescan_file() is only used when a file is being saved to a different
format than the original, and the user is not given a choice of which
reader to use (only which format to save it in), so doing this makes
sense for now. */
cf->provider.wth = wtap_open_offline(fname, WTAP_TYPE_AUTO, &err, &err_info, TRUE);
if (cf->provider.wth == NULL) {
cfile_open_failure_alert_box(fname, err, err_info);
return CF_READ_ERROR;
}
/* We're scanning a file whose contents should be the same as what
we had before, so we don't discard dissection state etc.. */
cf->f_datalen = 0;
/* Set the file name because we need it to set the follow stream filter.
XXX - is that still true? We need it for other reasons, though,
in any case. */
cf->filename = g_strdup(fname);
/* Indicate whether it's a permanent or temporary file. */
cf->is_tempfile = is_tempfile;
/* No user changes yet. */
cf->unsaved_changes = FALSE;
cf->cd_t = wtap_file_type_subtype(cf->provider.wth);
cf->linktypes = g_array_sized_new(FALSE, FALSE, (guint) sizeof(int), 1);
cf->snap = wtap_snapshot_length(cf->provider.wth);
name_ptr = g_filename_display_basename(cf->filename);
cf_callback_invoke(cf_cb_file_rescan_started, cf);
/* Record whether the file is compressed.
XXX - do we know this at open time? */
cf->iscompressed = wtap_iscompressed(cf->provider.wth);
/* Find the size of the file. */
size = wtap_file_size(cf->provider.wth, NULL);
g_timer_start(prog_timer);
cf->stop_flag = FALSE;
g_get_current_time(&start_time);
framenum = 0;
rec = wtap_get_rec(cf->provider.wth);
while ((wtap_read(cf->provider.wth, &err, &err_info, &data_offset))) {
framenum++;
fdata = frame_data_sequence_find(cf->provider.frames, framenum);
fdata->file_off = data_offset;
if (size >= 0) {
count++;
cf->f_datalen = wtap_read_so_far(cf->provider.wth);
/* Create the progress bar if necessary. */
if (progress_is_slow(progbar, prog_timer, size, cf->f_datalen)) {
progbar_val = calc_progbar_val(cf, size, cf->f_datalen, status_str, sizeof(status_str));
progbar = delayed_create_progress_dlg(cf->window, "Rescanning", name_ptr,
TRUE, &cf->stop_flag, &start_time, progbar_val);
}
/*
* Update the progress bar, but do it only after PROGBAR_UPDATE_INTERVAL
* has elapsed. Calling update_progress_dlg and packets_bar_update will
* likely trigger UI paint events, which might take a while depending on
* the platform and display. Reset our timer *after* painting.
*/
if (progbar && g_timer_elapsed(prog_timer, NULL) > PROGBAR_UPDATE_INTERVAL) {
progbar_val = calc_progbar_val(cf, size, cf->f_datalen, status_str, sizeof(status_str));
/* update the packet bar content on the first run or frequently on very large files */
update_progress_dlg(progbar, progbar_val, status_str);
compute_elapsed(cf, &start_time);
packets_bar_update();
g_timer_start(prog_timer);
}
}
if (cf->stop_flag) {
/* Well, the user decided to abort the rescan. Sadly, as this
isn't a reread, recovering is difficult, so we'll just
close the current capture. */
break;
}
/* Add this packet's link-layer encapsulation type to cf->linktypes, if
it's not already there.
XXX - yes, this is O(N), so if every packet had a different
link-layer encapsulation type, it'd be O(N^2) to read the file, but
there are probably going to be a small number of encapsulation types
in a file. */
if (rec->rec_type == REC_TYPE_PACKET) {
cf_add_encapsulation_type(cf, rec->rec_header.packet_header.pkt_encap);
}
}
/* Free the display name */
g_free(name_ptr);
/* We're done reading the file; destroy the progress bar if it was created. */
if (progbar != NULL)
destroy_progress_dlg(progbar);
g_timer_destroy(prog_timer);
/* We're done reading sequentially through the file. */
cf->state = FILE_READ_DONE;
/* Close the sequential I/O side, to free up memory it requires. */
wtap_sequential_close(cf->provider.wth);
/* compute the time it took to load the file */
compute_elapsed(cf, &start_time);
/* Set the file encapsulation type now; we don't know what it is until
we've looked at all the packets, as we don't know until then whether
there's more than one type (and thus whether it's
WTAP_ENCAP_PER_PACKET). */
cf->lnk_t = wtap_file_encap(cf->provider.wth);
cf_callback_invoke(cf_cb_file_rescan_finished, cf);
if (cf->stop_flag) {
/* Our caller will give up at this point. */
return CF_READ_ABORTED;
}
if (err != 0) {
/* Put up a message box noting that the read failed somewhere along
the line. Don't throw out the stuff we managed to read, though,
if any. */
cfile_read_failure_alert_box(NULL, err, err_info);
return CF_READ_ERROR;
} else
return CF_READ_OK;
}
cf_write_status_t
cf_save_records(capture_file *cf, const char *fname, guint save_format,
gboolean compressed, gboolean discard_comments,
gboolean dont_reopen)
{
gchar *err_info;
gchar *fname_new = NULL;
wtap_dumper *pdh;
frame_data *fdata;
addrinfo_lists_t *addr_lists;
guint framenum;
int err;
#ifdef _WIN32
gchar *display_basename;
#endif
enum {
SAVE_WITH_MOVE,
SAVE_WITH_COPY,
SAVE_WITH_WTAP
} how_to_save;
save_callback_args_t callback_args;
ERF_TYPE_META write and comment support Support per-packet comments in ERF_TYPE_META through a new Anchor ID extension header with per-Host unique 48-bit Anchor ID which links an ERF_TYPE_META record with a packet record. There may be more than one Anchor ID associated with a packet, where they are grouped by Host ID extension header in the extension header list. Like other ERF_TYPE_META existing comments should not be overwritten and instead a new record generated. See erf_write_anchor_meta_update_phdr() for detailed comments on the extension header stack required. As Wireshark only supports one comment currently, use the one one with the latest metadata generation time (gen_time). Do this for capture comment too. Write various wtap metadata in periodic per-second ERF_TYPE_META records if non-WTAP_ENCAP_ERF or we have an updated capture comment. Refactor erf_dump to create fake ERF header first then follow common pseudoheadr and payload write code rather than two separate code paths. Support an ERF_HOST_ID environment variable to define Wireshark's Host ID when writing. Defaults to 0 for now. ERF dissector updates to support Anchor ID extension header with basic frame linking. Update ERF_TYPE_META naming and descriptions to official name (Provenance) Core changes: Add has_comment_changed to wtap_pkthdr, TRUE when a packet opt_comment has unsaved changes by the user. Add needs_reload to wtap_dumper which forces a full reload of the file on save, otherwise wireshark gets confused by additional packets being written. Change-Id: I0bb04411548c7bcd2d6ed82af689fbeed104546c Ping-Bug: 12303 Reviewed-on: https://code.wireshark.org/review/21873 Petri-Dish: Anders Broman <a.broman58@gmail.com> Tested-by: Petri Dish Buildbot <buildbot-no-reply@wireshark.org> Reviewed-by: Stephen Donnelly <stephen.donnelly@endace.com> Reviewed-by: Guy Harris <guy@alum.mit.edu>
2017-06-01 08:34:25 +00:00
gboolean needs_reload = FALSE;
/* XXX caller should avoid saving the file while a read is pending
* (e.g. by delaying the save action) */
if (cf->read_lock) {
g_warning("cf_save_records(\"%s\") while the file is being read, potential crash ahead", fname);
}
cf_callback_invoke(cf_cb_file_save_started, (gpointer)fname);
addr_lists = get_addrinfo_list();
if (save_format == cf->cd_t && compressed == cf->iscompressed
&& !discard_comments && !cf->unsaved_changes
&& (wtap_addrinfo_list_empty(addr_lists) || !wtap_dump_has_name_resolution(save_format))) {
/* We're saving in the format it's already in, and we're not discarding
comments, and there are no changes we have in memory that aren't saved
to the file, and we have no name resolution information to write or
the file format we're saving in doesn't support writing name
resolution information, so we can just move or copy the raw data. */
if (cf->is_tempfile) {
/* The file being saved is a temporary file from a live
capture, so it doesn't need to stay around under that name;
first, try renaming the capture buffer file to the new name.
This acts as a "safe save", in that, if the file already
exists, the existing file will be removed only if the rename
succeeds.
Sadly, on Windows, as we have the current capture file
open, even MoveFileEx() with MOVEFILE_REPLACE_EXISTING
(to cause the rename to remove an existing target), as
done by ws_stdio_rename() (ws_rename() is #defined to
be ws_stdio_rename() on Windows) will fail.
According to the MSDN documentation for CreateFile(), if,
when we open a capture file, we were to directly do a CreateFile(),
opening with FILE_SHARE_DELETE|FILE_SHARE_READ, and then
convert it to a file descriptor with _open_osfhandle(),
that would allow the file to be renamed out from under us.
However, that doesn't work in practice. Perhaps the problem
is that the process doing the rename is the process that
has the file open. */
#ifndef _WIN32
if (ws_rename(cf->filename, fname) == 0) {
/* That succeeded - there's no need to copy the source file. */
how_to_save = SAVE_WITH_MOVE;
} else {
if (errno == EXDEV) {
/* They're on different file systems, so we have to copy the
file. */
how_to_save = SAVE_WITH_COPY;
} else {
/* The rename failed, but not because they're on different
file systems - put up an error message. (Or should we
just punt and try to copy? The only reason why I'd
expect the rename to fail and the copy to succeed would
be if we didn't have permission to remove the file from
the temporary directory, and that might be fixable - but
is it worth requiring the user to go off and fix it?) */
cf_rename_failure_alert_box(fname, errno);
goto fail;
}
}
#else
/* Windows - copy the file to its new location. */
how_to_save = SAVE_WITH_COPY;
#endif
} else {
/* It's a permanent file, so we should copy it, and not remove the
original. */
how_to_save = SAVE_WITH_COPY;
}
if (how_to_save == SAVE_WITH_COPY) {
/* Copy the file, if we haven't moved it. If we're overwriting
an existing file, we do it with a "safe save", by writing
to a new file and, if the write succeeds, renaming the
new file on top of the old file. */
if (file_exists(fname)) {
fname_new = g_strdup_printf("%s~", fname);
if (!copy_file_binary_mode(cf->filename, fname_new))
goto fail;
} else {
if (!copy_file_binary_mode(cf->filename, fname))
goto fail;
}
}
} else {
/* Either we're saving in a different format or we're saving changes,
such as added, modified, or removed comments, that haven't yet
been written to the underlying file; we can't do that by copying
or moving the capture file, we have to do it by writing the packets
out in Wiretap. */
GArray *shb_hdrs = NULL;
wtapng_iface_descriptions_t *idb_inf = NULL;
GArray *nrb_hdrs = NULL;
int encap;
/* XXX: what free's this shb_hdr? */
shb_hdrs = wtap_file_get_shb_for_new_file(cf->provider.wth);
idb_inf = wtap_file_get_idb_info(cf->provider.wth);
nrb_hdrs = wtap_file_get_nrb_for_new_file(cf->provider.wth);
/* Determine what file encapsulation type we should use. */
encap = wtap_dump_file_encap_type(cf->linktypes);
if (file_exists(fname)) {
/* We're overwriting an existing file; write out to a new file,
and, if that succeeds, rename the new file on top of the
old file. That makes this a "safe save", so that we don't
lose the old file if we have a problem writing out the new
file. (If the existing file is the current capture file,
we *HAVE* to do that, otherwise we're overwriting the file
from which we're reading the packets that we're writing!) */
fname_new = g_strdup_printf("%s~", fname);
pdh = wtap_dump_open_ng(fname_new, save_format, encap, cf->snap,
compressed, shb_hdrs, idb_inf, nrb_hdrs, &err);
} else {
pdh = wtap_dump_open_ng(fname, save_format, encap, cf->snap,
compressed, shb_hdrs, idb_inf, nrb_hdrs, &err);
}
g_free(idb_inf);
idb_inf = NULL;
if (pdh == NULL) {
cfile_dump_open_failure_alert_box(fname, err, save_format);
goto fail;
}
/* Add address resolution */
wtap_dump_set_addrinfo_list(pdh, addr_lists);
/* Iterate through the list of packets, processing all the packets. */
callback_args.pdh = pdh;
callback_args.fname = fname;
callback_args.file_type = save_format;
switch (process_specified_records(cf, NULL, "Saving", "packets",
TRUE, save_record, &callback_args, TRUE)) {
case PSP_FINISHED:
/* Completed successfully. */
break;
case PSP_STOPPED:
/* The user decided to abort the saving.
If we're writing to a temporary file, remove it.
XXX - should we do so even if we're not writing to a
temporary file? */
wtap_dump_close(pdh, &err);
if (fname_new != NULL)
ws_unlink(fname_new);
cf_callback_invoke(cf_cb_file_save_stopped, NULL);
return CF_WRITE_ABORTED;
case PSP_FAILED:
/* Error while saving.
If we're writing to a temporary file, remove it. */
if (fname_new != NULL)
ws_unlink(fname_new);
wtap_dump_close(pdh, &err);
goto fail;
}
ERF_TYPE_META write and comment support Support per-packet comments in ERF_TYPE_META through a new Anchor ID extension header with per-Host unique 48-bit Anchor ID which links an ERF_TYPE_META record with a packet record. There may be more than one Anchor ID associated with a packet, where they are grouped by Host ID extension header in the extension header list. Like other ERF_TYPE_META existing comments should not be overwritten and instead a new record generated. See erf_write_anchor_meta_update_phdr() for detailed comments on the extension header stack required. As Wireshark only supports one comment currently, use the one one with the latest metadata generation time (gen_time). Do this for capture comment too. Write various wtap metadata in periodic per-second ERF_TYPE_META records if non-WTAP_ENCAP_ERF or we have an updated capture comment. Refactor erf_dump to create fake ERF header first then follow common pseudoheadr and payload write code rather than two separate code paths. Support an ERF_HOST_ID environment variable to define Wireshark's Host ID when writing. Defaults to 0 for now. ERF dissector updates to support Anchor ID extension header with basic frame linking. Update ERF_TYPE_META naming and descriptions to official name (Provenance) Core changes: Add has_comment_changed to wtap_pkthdr, TRUE when a packet opt_comment has unsaved changes by the user. Add needs_reload to wtap_dumper which forces a full reload of the file on save, otherwise wireshark gets confused by additional packets being written. Change-Id: I0bb04411548c7bcd2d6ed82af689fbeed104546c Ping-Bug: 12303 Reviewed-on: https://code.wireshark.org/review/21873 Petri-Dish: Anders Broman <a.broman58@gmail.com> Tested-by: Petri Dish Buildbot <buildbot-no-reply@wireshark.org> Reviewed-by: Stephen Donnelly <stephen.donnelly@endace.com> Reviewed-by: Guy Harris <guy@alum.mit.edu>
2017-06-01 08:34:25 +00:00
needs_reload = wtap_dump_get_needs_reload(pdh);
if (!wtap_dump_close(pdh, &err)) {
cfile_close_failure_alert_box(fname, err);
goto fail;
}
how_to_save = SAVE_WITH_WTAP;
}
if (fname_new != NULL) {
/* We wrote out to fname_new, and should rename it on top of
fname. fname_new is now closed, so that should be possible even
on Windows. However, on Windows, we first need to close whatever
file descriptors we have open for fname. */
#ifdef _WIN32
wtap_fdclose(cf->provider.wth);
#endif
/* Now do the rename. */
if (ws_rename(fname_new, fname) == -1) {
/* Well, the rename failed. */
cf_rename_failure_alert_box(fname, errno);
#ifdef _WIN32
/* Attempt to reopen the random file descriptor using the
current file's filename. (At this point, the sequential
file descriptor is closed.) */
if (!wtap_fdreopen(cf->provider.wth, cf->filename, &err)) {
/* Oh, well, we're screwed. */
display_basename = g_filename_display_basename(cf->filename);
simple_error_message_box(
file_open_error_message(err, FALSE), display_basename);
g_free(display_basename);
}
#endif
goto fail;
}
}
/* If this was a temporary file, and we didn't do the save by doing
a move, so the tempoary file is still around under its old name,
remove it. */
if (cf->is_tempfile && how_to_save != SAVE_WITH_MOVE) {
/* If this fails, there's not much we can do, so just ignore errors. */
ws_unlink(cf->filename);
}
cf_callback_invoke(cf_cb_file_save_finished, NULL);
cf->unsaved_changes = FALSE;
if (!dont_reopen) {
switch (how_to_save) {
case SAVE_WITH_MOVE:
/* We just moved the file, so the wtap structure refers to the
new file, and all the information other than the filename
and the "is temporary" status applies to the new file; just
update that. */
g_free(cf->filename);
cf->filename = g_strdup(fname);
cf->is_tempfile = FALSE;
cf_callback_invoke(cf_cb_file_fast_save_finished, cf);
break;
case SAVE_WITH_COPY:
/* We just copied the file, so all the information other than
the wtap structure, the filename, and the "is temporary"
status applies to the new file; just update that. */
wtap_close(cf->provider.wth);
/* Although we're just "copying" and then opening the copy, it will
try all open_routine readers to open the copy, so we need to
reset the cfile's open_type. */
cf->open_type = WTAP_TYPE_AUTO;
cf->provider.wth = wtap_open_offline(fname, WTAP_TYPE_AUTO, &err, &err_info, TRUE);
if (cf->provider.wth == NULL) {
cfile_open_failure_alert_box(fname, err, err_info);
cf_close(cf);
} else {
g_free(cf->filename);
cf->filename = g_strdup(fname);
cf->is_tempfile = FALSE;
}
cf_callback_invoke(cf_cb_file_fast_save_finished, cf);
break;
case SAVE_WITH_WTAP:
/* Open and read the file we saved to.
XXX - this is somewhat of a waste; we already have the
packets, all this gets us is updated file type information
(which we could just stuff into "cf"), and having the new
file be the one we have opened and from which we're reading
the data, and it means we have to spend time opening and
reading the file, which could be a significant amount of
time if the file is large.
If the capture-file-writing code were to return the
seek offset of each packet it writes, we could save that
in the frame_data structure for the frame, and just open
the file without reading it again...
...as long as, for gzipped files, the process of writing
out the file *also* generates the information needed to
support fast random access to the compressed file. */
/* rescan_file will cause us to try all open_routines, so
reset cfile's open_type */
cf->open_type = WTAP_TYPE_AUTO;
ERF_TYPE_META write and comment support Support per-packet comments in ERF_TYPE_META through a new Anchor ID extension header with per-Host unique 48-bit Anchor ID which links an ERF_TYPE_META record with a packet record. There may be more than one Anchor ID associated with a packet, where they are grouped by Host ID extension header in the extension header list. Like other ERF_TYPE_META existing comments should not be overwritten and instead a new record generated. See erf_write_anchor_meta_update_phdr() for detailed comments on the extension header stack required. As Wireshark only supports one comment currently, use the one one with the latest metadata generation time (gen_time). Do this for capture comment too. Write various wtap metadata in periodic per-second ERF_TYPE_META records if non-WTAP_ENCAP_ERF or we have an updated capture comment. Refactor erf_dump to create fake ERF header first then follow common pseudoheadr and payload write code rather than two separate code paths. Support an ERF_HOST_ID environment variable to define Wireshark's Host ID when writing. Defaults to 0 for now. ERF dissector updates to support Anchor ID extension header with basic frame linking. Update ERF_TYPE_META naming and descriptions to official name (Provenance) Core changes: Add has_comment_changed to wtap_pkthdr, TRUE when a packet opt_comment has unsaved changes by the user. Add needs_reload to wtap_dumper which forces a full reload of the file on save, otherwise wireshark gets confused by additional packets being written. Change-Id: I0bb04411548c7bcd2d6ed82af689fbeed104546c Ping-Bug: 12303 Reviewed-on: https://code.wireshark.org/review/21873 Petri-Dish: Anders Broman <a.broman58@gmail.com> Tested-by: Petri Dish Buildbot <buildbot-no-reply@wireshark.org> Reviewed-by: Stephen Donnelly <stephen.donnelly@endace.com> Reviewed-by: Guy Harris <guy@alum.mit.edu>
2017-06-01 08:34:25 +00:00
/* There are cases when SAVE_WITH_WTAP can result in new packets
being written to the file, e.g ERF records
In that case, we need to reload the whole file */
if(needs_reload) {
if (cf_open(cf, fname, WTAP_TYPE_AUTO, FALSE, &err) == CF_OK) {
if (cf_read(cf, TRUE) != CF_READ_OK) {
/* The rescan failed; just close the file. Either
a dialog was popped up for the failure, so the
user knows what happened, or they stopped the
rescan, in which case they know what happened. */
/* XXX: This is inconsistent with normal open/reload behaviour. */
cf_close(cf);
}
}
}
else {
if (rescan_file(cf, fname, FALSE) != CF_READ_OK) {
/* The rescan failed; just close the file. Either
a dialog was popped up for the failure, so the
user knows what happened, or they stopped the
rescan, in which case they know what happened. */
cf_close(cf);
}
}
break;
}
/* If we were told to discard the comments, do so. */
if (discard_comments) {
/* Remove SHB comment, if any. */
wtap_write_shb_comment(cf->provider.wth, NULL);
/* remove all user comments */
for (framenum = 1; framenum <= cf->count; framenum++) {
fdata = frame_data_sequence_find(cf->provider.frames, framenum);
fdata->flags.has_phdr_comment = FALSE;
fdata->flags.has_user_comment = FALSE;
}
if (cf->provider.frames_user_comments) {
g_tree_destroy(cf->provider.frames_user_comments);
cf->provider.frames_user_comments = NULL;
}
cf->packet_comment_count = 0;
}
}
return CF_WRITE_OK;
fail:
if (fname_new != NULL) {
/* We were trying to write to a temporary file; get rid of it if it
exists. (We don't care whether this fails, as, if it fails,
there's not much we can do about it. I guess if it failed for
a reason other than "it doesn't exist", we could report an
error, so the user knows there's a junk file that they might
want to clean up.) */
ws_unlink(fname_new);
g_free(fname_new);
}
cf_callback_invoke(cf_cb_file_save_failed, NULL);
return CF_WRITE_ERROR;
}
cf_write_status_t
cf_export_specified_packets(capture_file *cf, const char *fname,
packet_range_t *range, guint save_format,
gboolean compressed)
{
gchar *fname_new = NULL;
int err;
wtap_dumper *pdh;
save_callback_args_t callback_args;
GArray *shb_hdrs = NULL;
wtapng_iface_descriptions_t *idb_inf = NULL;
GArray *nrb_hdrs = NULL;
int encap;
packet_range_process_init(range);
/* We're writing out specified packets from the specified capture
file to another file. Even if all captured packets are to be
written, don't special-case the operation - read each packet
and then write it out if it's one of the specified ones. */
/* XXX: what free's this shb_hdr? */
shb_hdrs = wtap_file_get_shb_for_new_file(cf->provider.wth);
idb_inf = wtap_file_get_idb_info(cf->provider.wth);
nrb_hdrs = wtap_file_get_nrb_for_new_file(cf->provider.wth);
/* Determine what file encapsulation type we should use. */
encap = wtap_dump_file_encap_type(cf->linktypes);
if (file_exists(fname)) {
/* We're overwriting an existing file; write out to a new file,
and, if that succeeds, rename the new file on top of the
old file. That makes this a "safe save", so that we don't
lose the old file if we have a problem writing out the new
file. (If the existing file is the current capture file,
we *HAVE* to do that, otherwise we're overwriting the file
from which we're reading the packets that we're writing!) */
fname_new = g_strdup_printf("%s~", fname);
pdh = wtap_dump_open_ng(fname_new, save_format, encap, cf->snap,
compressed, shb_hdrs, idb_inf, nrb_hdrs, &err);
} else {
pdh = wtap_dump_open_ng(fname, save_format, encap, cf->snap,
compressed, shb_hdrs, idb_inf, nrb_hdrs, &err);
}
g_free(idb_inf);
idb_inf = NULL;
if (pdh == NULL) {
cfile_dump_open_failure_alert_box(fname, err, save_format);
goto fail;
}
/* Add address resolution */
wtap_dump_set_addrinfo_list(pdh, get_addrinfo_list());
/* Iterate through the list of packets, processing the packets we were
told to process.
XXX - we've already called "packet_range_process_init(range)", but
"process_specified_records()" will do it again. Fortunately,
that's harmless in this case, as we haven't done anything to
"range" since we initialized it. */
callback_args.pdh = pdh;
callback_args.fname = fname;
callback_args.file_type = save_format;
switch (process_specified_records(cf, range, "Writing", "specified records",
TRUE, save_record, &callback_args, TRUE)) {
case PSP_FINISHED:
/* Completed successfully. */
break;
case PSP_STOPPED:
/* The user decided to abort the saving.
If we're writing to a temporary file, remove it.
XXX - should we do so even if we're not writing to a
temporary file? */
wtap_dump_close(pdh, &err);
if (fname_new != NULL)
ws_unlink(fname_new);
return CF_WRITE_ABORTED;
break;
case PSP_FAILED:
/* Error while saving.
If we're writing to a temporary file, remove it. */
if (fname_new != NULL)
ws_unlink(fname_new);
wtap_dump_close(pdh, &err);
goto fail;
}
if (!wtap_dump_close(pdh, &err)) {
cfile_close_failure_alert_box(fname, err);
goto fail;
}
if (fname_new != NULL) {
/* We wrote out to fname_new, and should rename it on top of
fname; fname is now closed, so that should be possible even
on Windows. Do the rename. */
if (ws_rename(fname_new, fname) == -1) {
/* Well, the rename failed. */
cf_rename_failure_alert_box(fname, errno);
goto fail;
}
}
return CF_WRITE_OK;
fail:
if (fname_new != NULL) {
/* We were trying to write to a temporary file; get rid of it if it
exists. (We don't care whether this fails, as, if it fails,
there's not much we can do about it. I guess if it failed for
a reason other than "it doesn't exist", we could report an
error, so the user knows there's a junk file that they might
want to clean up.) */
ws_unlink(fname_new);
g_free(fname_new);
}
return CF_WRITE_ERROR;
}
/*
* XXX - whether we mention the source pathname, the target pathname,
* or both depends on the error and on what we find if we look for
* one or both of them.
*/
static void
cf_rename_failure_alert_box(const char *filename, int err)
{
gchar *display_basename;
display_basename = g_filename_display_basename(filename);
switch (err) {
case ENOENT:
/* XXX - should check whether the source exists and, if not,
report it as the problem and, if so, report the destination
as the problem. */
simple_error_message_box("The path to the file \"%s\" doesn't exist.",
display_basename);
break;
case EACCES:
/* XXX - if we're doing a rename after a safe save, we should
probably say something else. */
simple_error_message_box("You don't have permission to move the capture file to \"%s\".",
display_basename);
break;
default:
/* XXX - this should probably mention both the source and destination
pathnames. */
simple_error_message_box("The file \"%s\" could not be moved: %s.",
display_basename, wtap_strerror(err));
break;
}
g_free(display_basename);
}
/* Reload the current capture file. */
void
cf_reload(capture_file *cf) {
gchar *filename;
gboolean is_tempfile;
int err;
if (cf->read_lock) {
g_warning("Failing cf_reload(\"%s\") since a read is in progress", cf->filename);
return;
}
/* If the file could be opened, "cf_open()" calls "cf_close()"
to get rid of state for the old capture file before filling in state
for the new capture file. "cf_close()" will remove the file if
it's a temporary file; we don't want that to happen (for one thing,
it'd prevent subsequent reopens from working). Remember whether it's
a temporary file, mark it as not being a temporary file, and then
reopen it as the type of file it was.
Also, "cf_close()" will free "cf->filename", so we must make
a copy of it first. */
filename = g_strdup(cf->filename);
is_tempfile = cf->is_tempfile;
cf->is_tempfile = FALSE;
if (cf_open(cf, filename, cf->open_type, is_tempfile, &err) == CF_OK) {
switch (cf_read(cf, TRUE)) {
case CF_READ_OK:
case CF_READ_ERROR:
/* Just because we got an error, that doesn't mean we were unable
to read any of the file; we handle what we could get from the
file. */
break;
case CF_READ_ABORTED:
/* The user bailed out of re-reading the capture file; the
capture file has been closed - just free the capture file name
string and return (without changing the last containing
directory). */
g_free(filename);
return;
}
} else {
/* The open failed, so "cf->is_tempfile" wasn't set to "is_tempfile".
Instead, the file was left open, so we should restore "cf->is_tempfile"
ourselves.
XXX - change the menu? Presumably "cf_open()" will do that;
make sure it does! */
cf->is_tempfile = is_tempfile;
}
/* "cf_open()" made a copy of the file name we handed it, so
we should free up our copy. */
g_free(filename);
}
/*
* Editor modelines
*
* Local Variables:
* c-basic-offset: 2
* tab-width: 8
* indent-tabs-mode: nil
* End:
*
* ex: set shiftwidth=2 tabstop=8 expandtab:
* :indentSize=2:tabSize=8:noTabs=true:
*/