[pcre/f20] Fix empty-matching possessive zero-repeat groups in interpreted mode
Petr Pisar
ppisar at fedoraproject.org
Mon Jul 14 13:34:54 UTC 2014
commit 0d90f2c1f9ad2e78076e9f04f2ee71ad6be4d187
Author: Petr Písař <ppisar at redhat.com>
Date: Mon Jul 14 13:15:27 2014 +0200
Fix empty-matching possessive zero-repeat groups in interpreted mode
...atching-possessive-zero-repeat-groups-bug.patch | 187 ++++++++++++++++++++
pcre.spec | 10 +-
2 files changed, 196 insertions(+), 1 deletions(-)
---
diff --git a/pcre-8.33-Fix-empty-matching-possessive-zero-repeat-groups-bug.patch b/pcre-8.33-Fix-empty-matching-possessive-zero-repeat-groups-bug.patch
new file mode 100644
index 0000000..a2db8d0
--- /dev/null
+++ b/pcre-8.33-Fix-empty-matching-possessive-zero-repeat-groups-bug.patch
@@ -0,0 +1,187 @@
+From 93c413c5fac105d90f77ab5d03e31e0f64fc6142 Mon Sep 17 00:00:00 2001
+From: ph10 <ph10 at 2f5784b3-3f2a-0410-8824-cb99058d5e15>
+Date: Tue, 27 May 2014 13:18:31 +0000
+Subject: [PATCH] Fix empty-matching possessive zero-repeat groups bug.
+MIME-Version: 1.0
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: 8bit
+
+git-svn-id: svn://vcs.exim.org/pcre/code/trunk@1478 2f5784b3-3f2a-0410-8824-cb99058d5e15
+
+Petr Pisar: Ported to 8.33.
+
+Signed-off-by: Petr Písař <ppisar at redhat.com>
+---
+ pcre_exec.c | 41 ++++++++++++++++++++++++++---------------
+ testdata/testinput1 | 9 +++++++++
+ testdata/testinput8 | 6 ++++++
+ testdata/testoutput1 | 12 ++++++++++++
+ testdata/testoutput8 | 8 ++++++++
+ 5 files changed, 61 insertions(+), 15 deletions(-)
+
+diff --git a/pcre_exec.c b/pcre_exec.c
+index ab76d02..481e899 100644
+--- a/pcre_exec.c
++++ b/pcre_exec.c
+@@ -1169,10 +1169,15 @@ for (;;)
+ if (rrc == MATCH_KETRPOS)
+ {
+ offset_top = md->end_offset_top;
+- eptr = md->end_match_ptr;
+ ecode = md->start_code + code_offset;
+ save_capture_last = md->capture_last;
+ matched_once = TRUE;
++ if (eptr == md->end_match_ptr) /* Matched an empty string */
++ {
++ do ecode += GET(ecode, 1); while (*ecode == OP_ALT);
++ break;
++ }
++ eptr = md->end_match_ptr;
+ continue;
+ }
+
+@@ -1242,9 +1247,14 @@ for (;;)
+ if (rrc == MATCH_KETRPOS)
+ {
+ offset_top = md->end_offset_top;
+- eptr = md->end_match_ptr;
+ ecode = md->start_code + code_offset;
+ matched_once = TRUE;
++ if (eptr == md->end_match_ptr) /* Matched an empty string */
++ {
++ do ecode += GET(ecode, 1); while (*ecode == OP_ALT);
++ break;
++ }
++ eptr = md->end_match_ptr;
+ continue;
+ }
+
+@@ -1976,7 +1986,7 @@ for (;;)
+ case OP_KETRMAX:
+ case OP_KETRPOS:
+ prev = ecode - GET(ecode, 1);
+-
++
+ /* If this was a group that remembered the subject start, in order to break
+ infinite repeats of empty string matches, retrieve the subject start from
+ the chain. Otherwise, set it NULL. */
+@@ -2001,7 +2011,7 @@ for (;;)
+ md->start_match_ptr = mstart;
+ RRETURN(MATCH_MATCH); /* Sets md->mark */
+ }
+-
++
+ /* For capturing groups we have to check the group number back at the start
+ and if necessary complete handling an extraction by setting the offsets and
+ bumping the high water mark. Whole-pattern recursion is coded as a recurse
+@@ -2061,6 +2071,18 @@ for (;;)
+ }
+ }
+
++ /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
++ and return the MATCH_KETRPOS. This makes it possible to do the repeats one
++ at a time from the outer level, thus saving stack. This must precede the
++ empty string test - in this case that test is done at the outer level. */
++
++ if (*ecode == OP_KETRPOS)
++ {
++ md->end_match_ptr = eptr;
++ md->end_offset_top = offset_top;
++ RRETURN(MATCH_KETRPOS);
++ }
++
+ /* For an ordinary non-repeating ket, just continue at this level. This
+ also happens for a repeating ket if no characters were matched in the
+ group. This is the forcible breaking of infinite loops as implemented in
+@@ -2083,17 +2105,6 @@ for (;;)
+ break;
+ }
+
+- /* OP_KETRPOS is a possessive repeating ket. Remember the current position,
+- and return the MATCH_KETRPOS. This makes it possible to do the repeats one
+- at a time from the outer level, thus saving stack. */
+-
+- if (*ecode == OP_KETRPOS)
+- {
+- md->end_match_ptr = eptr;
+- md->end_offset_top = offset_top;
+- RRETURN(MATCH_KETRPOS);
+- }
+-
+ /* The normal repeating kets try the rest of the pattern or restart from
+ the preceding bracket, in the appropriate order. In the second case, we can
+ use tail recursion to avoid using another stack frame, unless we have an
+diff --git a/testdata/testinput1 b/testdata/testinput1
+index d77d8ac..6bde9ec 100644
+--- a/testdata/testinput1
++++ b/testdata/testinput1
+@@ -5614,4 +5614,13 @@ AbcdCBefgBhiBqz
+ /[\Q]a\E]+/
+ aa]]
+
++'\A(?:[^\"]++|\"(?:[^\"]*+|\"\")*+\")++'
++ NON QUOTED \"QUOT\"\"ED\" AFTER \"NOT MATCHED
++
++'\A(?:[^\"]++|\"(?:[^\"]++|\"\")*+\")++'
++ NON QUOTED \"QUOT\"\"ED\" AFTER \"NOT MATCHED
++
++'\A(?:[^\"]++|\"(?:[^\"]++|\"\")++\")++'
++ NON QUOTED \"QUOT\"\"ED\" AFTER \"NOT MATCHED
++
+ /-- End of testinput1 --/
+diff --git a/testdata/testinput8 b/testdata/testinput8
+index d91013b..98a0b38 100644
+--- a/testdata/testinput8
++++ b/testdata/testinput8
+@@ -4801,4 +4801,10 @@
+ /abcd/
+ abcd\O0
+
++'\A(?:[^\"]++|\"(?:[^\"]*+|\"\")*+\")++'
++ NON QUOTED \"QUOT\"\"ED\" AFTER \"NOT MATCHED
++
++'\A(?:[^\"]++|\"(?:[^\"]++|\"\")*+\")++'
++ NON QUOTED \"QUOT\"\"ED\" AFTER \"NOT MATCHED
++
+ /-- End of testinput8 --/
+diff --git a/testdata/testoutput1 b/testdata/testoutput1
+index 1b0b8dc..cb9592d 100644
+--- a/testdata/testoutput1
++++ b/testdata/testoutput1
+@@ -9208,4 +9208,16 @@ No match
+ aa]]
+ 0: aa]]
+
++'\A(?:[^\"]++|\"(?:[^\"]*+|\"\")*+\")++'
++ NON QUOTED \"QUOT\"\"ED\" AFTER \"NOT MATCHED
++ 0: NON QUOTED "QUOT""ED" AFTER
++
++'\A(?:[^\"]++|\"(?:[^\"]++|\"\")*+\")++'
++ NON QUOTED \"QUOT\"\"ED\" AFTER \"NOT MATCHED
++ 0: NON QUOTED "QUOT""ED" AFTER
++
++'\A(?:[^\"]++|\"(?:[^\"]++|\"\")++\")++'
++ NON QUOTED \"QUOT\"\"ED\" AFTER \"NOT MATCHED
++ 0: NON QUOTED "QUOT""ED" AFTER
++
+ /-- End of testinput1 --/
+diff --git a/testdata/testoutput8 b/testdata/testoutput8
+index 75affbe..666b67e 100644
+--- a/testdata/testoutput8
++++ b/testdata/testoutput8
+@@ -8020,4 +8020,12 @@ Error -30 (invalid data in workspace for DFA restart)
+ abcd\O0
+ Matched, but offsets vector is too small to show all matches
+
++'\A(?:[^\"]++|\"(?:[^\"]*+|\"\")*+\")++'
++ NON QUOTED \"QUOT\"\"ED\" AFTER \"NOT MATCHED
++ 0: NON QUOTED "QUOT""ED" AFTER
++
++'\A(?:[^\"]++|\"(?:[^\"]++|\"\")*+\")++'
++ NON QUOTED \"QUOT\"\"ED\" AFTER \"NOT MATCHED
++ 0: NON QUOTED "QUOT""ED" AFTER
++
+ /-- End of testinput8 --/
+--
+1.9.3
+
diff --git a/pcre.spec b/pcre.spec
index 12ea70f..bc06187 100644
--- a/pcre.spec
+++ b/pcre.spec
@@ -2,7 +2,7 @@
#%%global rcversion RC1
Name: pcre
Version: 8.33
-Release: %{?rcversion:0.}5%{?rcversion:.%rcversion}%{?dist}
+Release: %{?rcversion:0.}6%{?rcversion:.%rcversion}%{?dist}
%global myversion %{version}%{?rcversion:-%rcversion}
Summary: Perl-compatible regular expression library
Group: System Environment/Libraries
@@ -22,6 +22,9 @@ Patch3: pcre-8.33-Fix-bad-starting-data-when-char-with-more-than-one-o.patch
# Fix character class with a literal quotation, bug #1111054,
# upstream bug #1494, in upstream after 8.35
Patch4: pcre-8.33-Fix-bad-compile-of-Qx-.-where-x-is-any-character.patch
+# Fix empty-matching possessive zero-repeat groups in interpreted mode,
+# bug #1119241, upstream bug #1500, in upstream after 8.35
+Patch5: pcre-8.33-Fix-empty-matching-possessive-zero-repeat-groups-bug.patch
BuildRequires: readline-devel
# New libtool to get rid of rpath
BuildRequires: autoconf, automake, libtool
@@ -66,6 +69,7 @@ Utilities demonstrating PCRE capabilities like pcregrep or pcretest.
%patch2 -p1 -b .jitted_range_check
%patch3 -p1 -b .starting_data
%patch4 -p1 -b .class_with_literal
+%patch5 -p1 -b .empty_zero_repeat_group
# Because of rpath patch
libtoolize --copy --force && autoreconf -vif
# One contributor's name is non-UTF-8
@@ -129,6 +133,10 @@ make check
%{_mandir}/man1/pcretest.*
%changelog
+* Mon Jul 14 2014 Petr Pisar <ppisar at redhat.com> - 8.33-6
+- Fix empty-matching possessive zero-repeat groups in interpreted mode
+ (bug #1119241)
+
* Thu Jun 19 2014 Petr Pisar <ppisar at redhat.com> - 8.33-5
- Fix bad starting data when char with more than one other case follows
circumflex in multiline UTF mode (bug #1110620)
More information about the scm-commits
mailing list