github_mirrors/vectorscan
3
0
Fork 1
mirror of https://github.com/VectorCamp/vectorscan.git synced 2025-06-28 16:41:01 +03:00
Code Issues Packages Projects Releases Wiki Activity

Restore \Q..\E support in character classes

Browse Source
This commit is contained in:
Justin Viiret 2015-11-12 13:27:55 +11:00 committed by Matthew Barr
parent 2a2576e907
commit fd19168025
2 changed files with 3 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
src/parser/Parser.rl
View File

@ -893,11 +893,7 @@ unichar readUtf8CodePoint4c(const u8 *ts) {
throw LocatedParseError("Invalid POSIX named class"); throw LocatedParseError("Invalid POSIX named class");
}; };
'\\Q' => { '\\Q' => {
// fcall readQuotedClass; fcall readQuotedClass;
ostringstream str;
str << "\\Q..\\E sequences in character classes not supported at index "
<< ts - ptr << ".";
throw ParseError(str.str());
}; };
'\\E' => { /*noop*/}; '\\E' => { /*noop*/};
# Backspace (this is only valid for \b in char classes) # Backspace (this is only valid for \b in char classes)
@ -1131,10 +1127,7 @@ unichar readUtf8CodePoint4c(const u8 *ts) {
inCharClassEarly = false; inCharClassEarly = false;
}; };
# if we hit a quote before anything "real", handle it # if we hit a quote before anything "real", handle it
#'\\Q' => { fcall readQuotedClass; }; '\\Q' => { fcall readQuotedClass; };
'\\Q' => {
throw LocatedParseError("\\Q..\\E sequences in character classes not supported");
};
'\\E' => { /*noop*/}; '\\E' => { /*noop*/};
# time for the real work to happen # time for the real work to happen
@ -1170,6 +1163,7 @@ unichar readUtf8CodePoint4c(const u8 *ts) {
# Literal character # Literal character
any => { any => {
currentCls->add(*ts); currentCls->add(*ts);
inCharClassEarly = false;
}; };
*|; *|;

1
unit/hyperscan/bad_patterns.txt
View File

@ -85,7 +85,6 @@
84:/[=\]=]/ #Unsupported POSIX collating element at index 0. 84:/[=\]=]/ #Unsupported POSIX collating element at index 0.
85:/A(?!)+Z/ #Invalid repeat at index 5. 85:/A(?!)+Z/ #Invalid repeat at index 5.
86:/\X/ #\X unsupported at index 0. 86:/\X/ #\X unsupported at index 0.
87:/[a\Qz\E]/ #\Q..\E sequences in character classes not supported at index 2.
88:/[A-\d]/ #Invalid range in character class at index 3. 88:/[A-\d]/ #Invalid range in character class at index 3.
89:/[A-[:digit:]]/ #Invalid range in character class at index 3. 89:/[A-[:digit:]]/ #Invalid range in character class at index 3.
90:/B[--[:digit:]--]+/ #Invalid range in character class at index 4. 90:/B[--[:digit:]--]+/ #Invalid range in character class at index 4.