walkero
/
pcre2
1
0
Fork 0
Code Issues Pull Requests Projects Releases 1 Wiki Activity

Fix match_unset_backref bug.

This commit is contained in:
Philip.Hazel 2014-10-10 16:42:03 +00:00
parent 623b497851
commit d961132ae1
3 changed files with 18 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
src/pcre2_match.c
View File

@ -2763,19 +2763,20 @@ for (;;)
continue; /* With the main loop */ continue; /* With the main loop */
} }
/* Handle repeated back references. If a set group has length zero, just /* Handle repeated back references. If a set group has length zero, just
continue with the main loop, because it matches however many times. For an continue with the main loop, because it matches however many times. For an
unset reference, in non-match-unset-backref mode, if the minimum is unset reference, if the minimum is zero, we can also just continue. We an
zero, we can continue at the same level without recursion. For any other also continue if PCRE2_MATCH_UNSET_BACKREF is set, because this makes unset
minimum, carrying on will result in NOMATCH. */ group be have as a zero-length group. For any other unset cases, carrying
on will result in NOMATCH. */
if (offset < offset_top && mb->ovector[offset] != PCRE2_UNSET) if (offset < offset_top && mb->ovector[offset] != PCRE2_UNSET)
{ {
if (mb->ovector[offset] == mb->ovector[offset + 1]) continue; if (mb->ovector[offset] == mb->ovector[offset + 1]) continue;
} }
else else /* Group is not set */
{ {
if (min == 0 && (mb->poptions & PCRE2_MATCH_UNSET_BACKREF) == 0) if (min == 0 || (mb->poptions & PCRE2_MATCH_UNSET_BACKREF) != 0)
continue; continue;
} }
@ -2856,7 +2857,7 @@ for (;;)
eptr += slength; eptr += slength;
} }
/* If the length matched for each repetiaion is the same as the length of /* If the length matched for each repetition is the same as the length of
the captured group, we can easily work backwards. This is the normal the captured group, we can easily work backwards. This is the normal
case. However, in caseless UTF-8 mode there are pairs of case-equivalent case. However, in caseless UTF-8 mode there are pairs of case-equivalent
characters whose lengths (in terms of code units) differ. However, this characters whose lengths (in terms of code units) differ. However, this

3
testdata/testinput2 vendored
View File

@ -4003,4 +4003,7 @@ a random value. /Ix
abd abd
xyd xyd
/\k<A>*(?<A>aa)(?<A>bb)/match_unset_backref,dupnames
aabb
# End of testinput2 # End of testinput2

6
testdata/testoutput2 vendored
View File

@ -13576,4 +13576,10 @@ Subject length lower bound = 0
xyd xyd
0: d 0: d
/\k<A>*(?<A>aa)(?<A>bb)/match_unset_backref,dupnames
aabb
0: aabb
1: aa
2: bb
# End of testinput2 # End of testinput2