Fix matching repeated character classes against an 8-bit string containting multi-code-unit characters

pcre2-10.31-Fix-C-bug-with-repeated-character-classes-in-UTF-8-m.patch

@@ -0,0 +1,120 @@
+From ea6f7a508aaa2fd61eb60d7759fe00713f46cd5c Mon Sep 17 00:00:00 2001
+From: ph10 <ph10@6239d852-aaf2-0410-a92c-79f79f948069>
+Date: Mon, 19 Feb 2018 17:26:33 +0000
+Subject: [PATCH] Fix \C bug with repeated character classes in UTF-8 mode.
+MIME-Version: 1.0
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: 8bit
+
+git-svn-id: svn://vcs.exim.org/pcre2/code/trunk@918 6239d852-aaf2-0410-a92c-79f79f948069
+Petr Písař: Ported to 10.31.
+---
+ src/pcre2_match.c        | 16 ++++++++++++----
+ testdata/testinput22     |  3 +++
+ testdata/testoutput22-16 |  4 ++++
+ testdata/testoutput22-32 |  4 ++++
+ testdata/testoutput22-8  |  4 ++++
+
+diff --git a/src/pcre2_match.c b/src/pcre2_match.c
+index 79cc93f..ce96016 100644
+--- a/src/pcre2_match.c
++++ b/src/pcre2_match.c
+@@ -1962,11 +1962,15 @@ fprintf(stderr, "++ op=%d\n", *Fecode);
+ 
+           if (reptype == REPTYPE_POS) continue;    /* No backtracking */
+ 
++          /* After \C in UTF mode, Lstart_eptr might be in the middle of a
++          Unicode character. Use <= Lstart_eptr to ensure backtracking doesn't
++          go too far. */
++
+           for (;;)
+             {
+             RMATCH(Fecode, RM201);
+             if (rrc != MATCH_NOMATCH) RRETURN(rrc);
+-            if (Feptr-- == Lstart_eptr) break;  /* Tried at original position */
++            if (Feptr-- <= Lstart_eptr) break;  /* Tried at original position */
+             BACKCHAR(Feptr);
+             }
+           }
+@@ -2126,11 +2130,15 @@ fprintf(stderr, "++ op=%d\n", *Fecode);
+ 
+         if (reptype == REPTYPE_POS) continue;    /* No backtracking */
+ 
++        /* After \C in UTF mode, Lstart_eptr might be in the middle of a
++        Unicode character. Use <= Lstart_eptr to ensure backtracking doesn't
++        go too far. */
++
+         for(;;)
+           {
+           RMATCH(Fecode, RM101);
+           if (rrc != MATCH_NOMATCH) RRETURN(rrc);
+-          if (Feptr-- == Lstart_eptr) break;  /* Tried at original position */
++          if (Feptr-- <= Lstart_eptr) break;  /* Tried at original position */
+ #ifdef SUPPORT_UNICODE
+           if (utf) BACKCHAR(Feptr);
+ #endif
+@@ -4002,8 +4010,8 @@ fprintf(stderr, "++ op=%d\n", *Fecode);
+         if (reptype == REPTYPE_POS) continue;    /* No backtracking */
+ 
+         /* After \C in UTF mode, Lstart_eptr might be in the middle of a
+-        Unicode character. Use <= pp to ensure backtracking doesn't go too far.
+-        */
++        Unicode character. Use <= Lstart_eptr to ensure backtracking doesn't
++        go too far. */
+ 
+         for(;;)
+           {
+diff --git a/testdata/testinput22 b/testdata/testinput22
+index e6d4053..c218ea6 100644
+--- a/testdata/testinput22
++++ b/testdata/testinput22
+@@ -98,4 +98,7 @@
+ \= Expect no match - tests \C at end of subject
+     ab
+ 
++/\C[^\v]+\x80/utf
++    [AΏBŀC]
++
+ # End of testinput22
+diff --git a/testdata/testoutput22-16 b/testdata/testoutput22-16
+index 88f827c..5e23611 100644
+--- a/testdata/testoutput22-16
++++ b/testdata/testoutput22-16
+@@ -171,4 +171,8 @@ No match
+     ab
+ No match
+ 
++/\C[^\v]+\x80/utf
++    [AΏBŀC]
++No match
++
+ # End of testinput22
+diff --git a/testdata/testoutput22-32 b/testdata/testoutput22-32
+index ac485fc..8576f31 100644
+--- a/testdata/testoutput22-32
++++ b/testdata/testoutput22-32
+@@ -169,4 +169,8 @@ No match
+     ab
+ No match
+ 
++/\C[^\v]+\x80/utf
++    [AΏBŀC]
++No match
++
+ # End of testinput22
+diff --git a/testdata/testoutput22-8 b/testdata/testoutput22-8
+index 3d31fbc..8543652 100644
+--- a/testdata/testoutput22-8
++++ b/testdata/testoutput22-8
+@@ -173,4 +173,8 @@ No match
+     ab
+ No match
+ 
++/\C[^\v]+\x80/utf
++    [AΏBŀC]
++No match
++
+ # End of testinput22
+-- 
+2.13.6
+

pcre2.spec

@@ -57,6 +57,9 @@ Patch2:     pcre2-10.31-Fix-pcre2test-C-to-correctly-show-what-R-matches.patch
 # 2/2 Fix pcre2test -C to correctly show what \R matches,
 # in upstream after 10.31
 Patch3:     pcre2-10.31-Oops-forgot-about-C-bsr-in-previous-patch.patch
+# Fix matching repeated character classes against an 8-bit string containting
+# multi-code-unit characters, in upstream after 10.31
+Patch4:     pcre2-10.31-Fix-C-bug-with-repeated-character-classes-in-UTF-8-m.patch
 BuildRequires:  autoconf
 BuildRequires:  automake
 BuildRequires:  coreutils
@@ -136,6 +139,7 @@ Utilities demonstrating PCRE2 capabilities like pcre2grep or pcre2test.
 %patch1 -p1
 %patch2 -p1
 %patch3 -p1
+%patch4 -p1
 # Because of multilib patch
 libtoolize --copy --force
 autoreconf -vif
@@ -242,6 +246,8 @@ make %{?_smp_mflags} check VERBOSE=yes
 - Fix returning unset groups in POSIX interface if REG_STARTEND a non-zero
   starting offset (upstream bug #2244)
 - Fix pcre2test -C to correctly show what \R matches
+- Fix matching repeated character classes against an 8-bit string containting
+  multi-code-unit characters
 
 * Mon Feb 12 2018 Petr Pisar <ppisar@redhat.com> - 10.31-1
 - 10.31 bump