forked from osmocom/wireshark
8746eea297
Instead of using a heuristic to decide whether the form ${...} is
a macro or not, try to resolve the name to a registered protocol
field and use that instead.
This increases somewhat the surface for clobbering existing macro
names with new field registrations but we'll cross that bridge when
we get to it.
Rejecting protocol field types reduces this probability again but it
may not be intuitive to the user trying to mistakenly use a reference
to a protocol why it is parsed as a macro. The reasons for rejecting
FT_PROTOCOL types as not interesting field references are not
very strong but it seems reasonable.
$ dftest 'frame.number != ${frame.number}'
Filter: frame.number != ${frame.number}
Instructions:
00000 READ_TREE frame.number -> reg#0
00001 IF_FALSE_GOTO 5
00002 READ_REFERENCE ${frame.number} -> reg#1
00003 IF_FALSE_GOTO 5
00004 ALL_NE reg#0 != reg#1
00005 RETURN
$ dftest 'frame != ${frame}'
dftest: macro 'frame' does not exist
|
||
|---|---|---|
| .. | ||
| .editorconfig | ||
| CMakeLists.txt | ||
| dfilter-int.h | ||
| dfilter-macro.c | ||
| dfilter-macro.h | ||
| dfilter.c | ||
| dfilter.h | ||
| dfunctions.c | ||
| dfunctions.h | ||
| dfvm.c | ||
| dfvm.h | ||
| drange.c | ||
| drange.h | ||
| gencode.c | ||
| gencode.h | ||
| grammar.lemon | ||
| scanner.l | ||
| semcheck.c | ||
| semcheck.h | ||
| sttype-function.c | ||
| sttype-function.h | ||
| sttype-pointer.c | ||
| sttype-range.c | ||
| sttype-range.h | ||
| sttype-set.c | ||
| sttype-set.h | ||
| sttype-string.c | ||
| sttype-test.c | ||
| sttype-test.h | ||
| syntax-tree.c | ||
| syntax-tree.h | ||