Convert capture file regex search to PCRE2.

Replace the use of the obsolete GRegex with PCRE2.

Fixes a crash reported in issue #17500.

cfile.h

@@ -104,7 +104,7 @@ typedef struct _capture_file {
     guint32                     search_pos;           /* Byte position of last byte found in a hex search */
     guint32                     search_len;           /* Length of bytes matching the search */
     gboolean                    case_type;            /* TRUE if case-insensitive text search */
-    GRegex                     *regex;                /* Set if regular expression search */
+    ws_regex_t                 *regex;                /* Set if regular expression search */
     search_charset_t            scs_type;             /* Character set for text search */
     search_direction            dir;                  /* Direction in which to do searches */
     gboolean                    search_in_progress;   /* TRUE if user just clicked OK in the Find dialog or hit <control>N/B */

file.c

@@ -3146,7 +3146,7 @@ match_subtree_text(proto_node *node, gpointer data)
     }
 
     if (cf->regex) {
-        if (g_regex_match(cf->regex, label_ptr, (GRegexMatchFlags) 0, NULL)) {
+        if (ws_regex_matches(cf->regex, label_ptr)) {
             mdata->frame_matched = TRUE;
             mdata->finfo = fi;
             return;
@@ -3223,7 +3223,7 @@ match_summary_line(capture_file *cf, frame_data *fdata,
             info_column = get_column_text(edt.pi.cinfo, colx);
             info_column_len = strlen(info_column);
             if (cf->regex) {
-                if (g_regex_match(cf->regex, info_column, (GRegexMatchFlags) 0, NULL)) {
+                if (ws_regex_matches(cf->regex, info_column)) {
                     result = MR_MATCHED;
                     break;
                 }
@@ -3747,7 +3747,7 @@ match_regex(capture_file *cf, frame_data *fdata,
         wtap_rec *rec, Buffer *buf, void *criterion _U_)
 {
     match_result  result = MR_NOTMATCHED;
-    GMatchInfo   *match_info = NULL;
+    size_t result_pos[2] = {0, 0};
 
     /* Load the frame's data. */
     if (!cf_read_record(cf, fdata, rec, buf)) {
@@ -3755,13 +3755,13 @@ match_regex(capture_file *cf, frame_data *fdata,
         return MR_ERROR;
     }
 
-    if (g_regex_match_full(cf->regex, (const gchar *)ws_buffer_start_ptr(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;
+    if (ws_regex_matches_pos(cf->regex,
+                                (const gchar *)ws_buffer_start_ptr(buf),
+                                fdata->cap_len,
+                                result_pos)) {
+        //TODO: Fix cast.
+        cf->search_pos = (guint32)(result_pos[1] - 1); /* last byte = end position - 1 */
+        cf->search_len = (guint32)(result_pos[1] - result_pos[0]);
         result = MR_MATCHED;
     }
     return result;

packaging/debian/libwsutil0.symbols

@@ -439,6 +439,7 @@ libwsutil.so.0 libwsutil0 #MINVER#
  ws_regex_free@Base 3.7.0
  ws_regex_matches@Base 3.7.0
  ws_regex_matches_length@Base 3.7.0
+ ws_regex_matches_pos@Base 3.7.2
  ws_regex_pattern@Base 3.7.0
  ws_socket_ptoa@Base 3.1.1
  ws_strcasestr@Base 3.7.0

ui/qt/search_frame.cpp

@@ -17,6 +17,7 @@
 #include <epan/strutil.h>
 
 #include <wsutil/utf8_entities.h>
+#include <wsutil/regex.h>
 
 #include "main_application.h"
 #include <QKeyEvent>
@@ -63,7 +64,7 @@ SearchFrame::SearchFrame(QWidget *parent) :
 SearchFrame::~SearchFrame()
 {
     if (regex_) {
-        g_regex_unref(regex_);
+        ws_regex_free(regex_);
     }
     delete sf_ui_;
 }
@@ -140,13 +141,13 @@ void SearchFrame::keyPressEvent(QKeyEvent *event)
 
 bool SearchFrame::regexCompile()
 {
-    int flags = (G_REGEX_OPTIMIZE);
+    unsigned flags = 0;
     if (!sf_ui_->caseCheckBox->isChecked()) {
-        flags |= G_REGEX_CASELESS;
+        flags |= WS_REGEX_CASELESS;
     }
 
     if (regex_) {
-        g_regex_unref(regex_);
+        ws_regex_free(regex_);
     }
 
     if (sf_ui_->searchLineEdit->text().isEmpty()) {
@@ -154,12 +155,12 @@ bool SearchFrame::regexCompile()
         return false;
     }
 
-    GError *error = nullptr;
-    regex_ = g_regex_new(sf_ui_->searchLineEdit->text().toUtf8().constData(),
-                         (GRegexCompileFlags)flags, (GRegexMatchFlags) 0, &error);
-    if (error) {
-        regex_error_ = error->message;
-        g_error_free(error);
+    char *errmsg = nullptr;
+    regex_ = ws_regex_compile_ex(sf_ui_->searchLineEdit->text().toUtf8().constData(), -1,
+                         &errmsg, flags);
+
+    if (errmsg != nullptr) {
+        regex_error_ = errmsg;
     }
 
     return regex_ ? true : false;

ui/qt/search_frame.h

@@ -47,7 +47,7 @@ private:
 
     Ui::SearchFrame *sf_ui_;
     capture_file *cap_file_;
-    GRegex *regex_;
+    ws_regex_t *regex_;
     QString regex_error_;
 
 private slots:

wsutil/regex.c

@@ -103,14 +103,16 @@ ws_regex_compile(const char *patt, char **errmsg)
 
 
 static bool
-match_pcre2(pcre2_code *code, PCRE2_SPTR subject, PCRE2_SIZE length)
+match_pcre2(pcre2_code *code, const char *subject, ssize_t subj_length,
+                pcre2_match_data *match_data)
 {
-    pcre2_match_data *match_data;
+    PCRE2_SIZE length;
     int rc;
 
-    /* We don't use the matched substring but pcre2_match requires
-     * at least one pair of offsets. */
-    match_data = pcre2_match_data_create(1, NULL);
+    if (subj_length < 0)
+        length = PCRE2_ZERO_TERMINATED;
+    else
+        length = (PCRE2_SIZE)subj_length;
 
     rc = pcre2_match(code,
                     subject,
@@ -120,8 +122,6 @@ match_pcre2(pcre2_code *code, PCRE2_SPTR subject, PCRE2_SIZE length)
                     match_data,
                     NULL);
 
-    pcre2_match_data_free(match_data);
-
     if (rc < 0) {
         /* No match */
         if (rc != PCRE2_ERROR_NOMATCH) {
@@ -142,21 +142,49 @@ match_pcre2(pcre2_code *code, PCRE2_SPTR subject, PCRE2_SIZE length)
 bool
 ws_regex_matches(const ws_regex_t *re, const char *subj)
 {
+    return ws_regex_matches_length(re, subj, -1);
+}
+
+
+bool
+ws_regex_matches_length(const ws_regex_t *re,
+                        const char *subj, ssize_t subj_length)
+{
+    bool matched;
+    pcre2_match_data *match_data;
+
     ws_return_val_if_null(re, FALSE);
     ws_return_val_if_null(subj, FALSE);
 
-    return match_pcre2(re->code, (PCRE2_SPTR)subj, PCRE2_ZERO_TERMINATED);
+    /* We don't use the matched substring but pcre2_match requires
+     * at least one pair of offsets. */
+    match_data = pcre2_match_data_create(1, NULL);
+    matched = match_pcre2(re->code, subj, subj_length, match_data);
+    pcre2_match_data_free(match_data);
+    return matched;
 }
 
 
 bool
-ws_regex_matches_length(const ws_regex_t *re,
-                        const char *subj, size_t subj_length)
+ws_regex_matches_pos(const ws_regex_t *re,
+                        const char *subj, ssize_t subj_length,
+                        size_t pos_vect[2])
 {
+    bool matched;
+    pcre2_match_data *match_data;
+
     ws_return_val_if_null(re, FALSE);
     ws_return_val_if_null(subj, FALSE);
 
-    return match_pcre2(re->code, (PCRE2_SPTR)subj, (PCRE2_SIZE)subj_length);
+    match_data = pcre2_match_data_create(1, NULL);
+    matched = match_pcre2(re->code, subj, subj_length, match_data);
+    if (matched && pos_vect) {
+        PCRE2_SIZE *ovect = pcre2_get_ovector_pointer(match_data);
+        pos_vect[0] = ovect[0];
+        pos_vect[1] = ovect[1];
+    }
+    pcre2_match_data_free(match_data);
+    return matched;
 }
 
 

wsutil/regex.h

@@ -37,7 +37,18 @@ ws_regex_matches(const ws_regex_t *re, const char *subj);
 /** Matches a subject string length in 8 bit code units. */
 WS_DLL_PUBLIC bool
 ws_regex_matches_length(const ws_regex_t *re,
-                        const char *subj, size_t subj_length);
+                        const char *subj, ssize_t subj_length);
+
+/** Returns start and end position of the matched substring.
+ *
+ *  pos_vect[0] is first codepoint in the matched substring.
+ *  pos_vect[1] is the next to last codepoint in the matched substring.
+ *  pos_vect[1] - pos_vect[0] is the matched substring length.
+ */
+WS_DLL_PUBLIC bool
+ws_regex_matches_pos(const ws_regex_t *re,
+                        const char *subj, ssize_t subj_length,
+                        size_t pos_vect[2]);
 
 WS_DLL_PUBLIC void
 ws_regex_free(ws_regex_t *re);