Change Log for PCRE2
--------------------

Before the move to GitHub, this was the only record of changes to PCRE2. Now
there is also the log of commit messages.

Internal changes which are not visible to clients of the library are mostly not
listed here.

Version 10.47 21-October-2025
-----------------------------

1. (#660, #655, #663) Expanded platforms tested by CI: FreeBSD, OpenBSD,
Solaris, MSYS (Cygwin), S390x, PPC64le, ARMv7, AARCH64, RiscV.

2. (#655) Made build clean of `/W3` warnings on MSVC. Further suppression of
minor build warnings on other platforms (various commits).

3. (#651) Added `--enable-Werror` flag to `./configure` to treat warnings as
errors (or `--enable-errwarn` for Solaris cc). Previously, you had to hackily
override the CPPFLAGS when calling make since you can't pass `-Werror` as a
CFLAG into `./configure` (it breaks compiler feature detection).

4. (#682) Added AM_MAINTAINER_MODE for Git tag releases. Users building with
the `./configure` script can check out the release tag using Git, which assigns
the current time as the modification time to each checked-out file. This caused
Autoconf to attempt to regenerate the configure script.

5. (#692) Add LICENSE file for sljit to the tarball release, to clarify that the
sljit code is governed separately by the 2-clause BSD license.

6. (#656, #695) Add full support for z/OS and native EBCDIC support. The z/OS
support is tested nightly using the XLC and IBM-Clang compilers, with Autoconf
and CMake. In addition, for test purposes, the EBCDIC support can now be enabled
on any platform using the new Autoconf `--enable-ebcdic-ignoring-compiler` and
CMake `-DPCRE2_EBCDIC_IGNORING_COMPILER` options.

7. (#700) Faster lookup of named capture groups during pattern compilation using
a hash table.

8. (#697, #756, #778) Improvements to pcre2test to increase code coverage:
-malloc argument; more detailed test assertions.

9. (#705, #710, #737, #738) Powerful new feature: pattern recursion of
the form "(?1(GROUP_NAME_OR_NUM,...))" acts as a subroutine call which
additionally returns the listed capturing groups to the calling context.

10. (#721) Add linker scripts to enable symbol versioning for the PCRE2 dynamic
libraries. Downstream Linux distributions may make use of this, or disable it
with the new Autoconf `--disable-symvers` and CMake `-DPCRE2_SYMVERS` options.
Currently, Linux, Solaris, and FreeBSD (GNU ld, LLVM lld, and Solaris ld) are
tested and supported.

11. (#733) New API function: pcre2_next_match(). This function makes it both
simpler and safer for clients to iterate over all matches in a subject. The
documentation in `pcre2api` also provides improved guidance in the section
"Iterating over all matches".

12. (#739, #744, #753) Modernize the CMake build files, to use standard
commands to export the PCRE2 targets. This makes use of the
"$<BUILD_INTERFACE:...>" and "$<INSTALL_INTERFACE:...>" expressions alongside
the built-in "install(EXPORT...)" command. This brings the CMake files in line
with the patches used by vcpkg to distribute PCRE2. The minimum CMake version
remains 3.15.

13. (#756) Improved error offsets and diagnostics for syntax errors during
pattern compilation.

14. (#708, #729, #724, #731, #777) Various updates to Bazel and Zig
build support.

15. (#775) Added PCRE2_CONFIG_EFFECTIVE_LINKSIZE option to pcre2_config(), to
report the actual number of code units used in compiled patterns for recording
string lengths and offsets.

16. (#801) Significant bugfix: Fix a crash in pcre2_callout_enumerate() which is
easily reachable on any pattern that contains a Unicode character class. If your
application uses this function, please read the details for this change and
evaluate its severity for your application.

17. (#806, #807) Improved input validation for pcre2_substitute() used with
PCRE2_SUBSTITUTE_MATCHED.

18. (#817) Add support for $+ replacement to pcre2_substitute().

19. (#818) New SIMD code generation in the JIT for AArch64.


Version 10.46 27-August-2025
----------------------------

1. (#771) (CVE-2025-58050) Security fix to prevent a read-past-the-end memory
error, of arbitrary length. An attacker-controlled regex pattern is required,
and it cannot be triggered by providing crafted subject (match) text. The
(*ACCEPT) and (*scs:) pattern features must be used together.

Release 10.44 and earlier are not affected.

This could have implications of denial-of-service or information disclosure,
and could potentially be used to escalate other vulnerabilities in a system
(such as information disclosure being used to escalate the severity of an
unrelated bug in another system).


Version 10.45 05-February-2025
------------------------------

1. (#418) Change 6 of 10.44 broke 32-bit tests because pcre2test's reporting of
memory size was changed to the entire compiled data block, instead of just the
pattern and tables data, so as to align with the new length restriction.
Because the block's header contains pointers, this meant the pcre2test output
was different in 32-bit mode. A patch by Carlo reverts to the previous state
and makes sure that any limit set by pcre2_set_max_pattern_compiled_length()
also avoids the internal struct overhead.

2. (#416, #622) Updates to build.zig.

3. (#427, et al.) Various fixes to pacify static analyzers.

4. (#428) Add --posix-pattern-file to pcre2grep to allow processing of empty
patterns through the -f option, as well as patterns that end in space
characters, for compatibility with other grep tools.

5. (4fa5b8bd) Fix a bug in the fuzz support quantifier-limiting code. It ignores
strings of more than 5 digits because they are necessarily numbers greater than
65535, the largest legal quantifier. However, it wasn't ignoring non-significant
leading zeros.

6. (6d82f0cd) The case-independent processing of the letter-matching Unicode
properties Ll, Lt, and Lu have been changed to match Perl (which changed a while
ago). When caseless matching is in force, all three of these properties are now
treated as Lc (cased letter).

7. (#433) The pcre2_jit_compile() function was updated by the addition of a new
option PCRE2_JIT_TEST_ALLOC which, if called with a NULL first argument, tests
not only the availability of JIT, but also its ability to allocate executable
memory. Update pcre2test to use this support to extend the -C option.

8. (75b1025a) The code for parsing Unicode property descriptions for \p and \P
been changed as follows:

  . White space etc. before ^ in a negated value such as \p{ ^L } was not being
    ignored.

  . The code wouldn't have worked if PCRE2 was compiled for UTF-8 support
    within an EBCDIC environment. Possibly nobody does this any more, but it
    should now work.

  . The documentation of the syntax of what can follow \p and \P has been
    updated.

9. (1c24ba01) There was an error in the table of lengths for parsed items for
the OPTIONS item, but fortuitously it could never have actually bitten. While
fixing this, some other code that could never be obeyed was discovered and
removed.

10. (674b6640) Removed some incorect optimization code from DFA matching that
has been there since PCRE1, but has just been found to cause a no match return
instead of a partial match in some cases. It involves partial matching when (*F)
is present so is unlikely to have actually affected anyone.

11. (b0f4ac17) Tidy the wording and formatting of some pcre2test error messages
concerned with bad modifiers. Also restrict single-letter modifier sequences to
the first item in a modifier list, as documented and always intended.

12. (1415565c) An iterator at the end of many assertions can always be
auto-possessified, but not at the end of variable-length lookbehinds. There was
a bug in the code that checks for such a lookbehind; it was looking only at the
first branch, which is wrong because some branches can be fixed length when
others are not, for example (?<=AB|CD?). Now all branches are checked for
variability.

13. (ead08288) Matching with pcre2_match() could give an incorrect result if a
variable-length lookbehind was used as the condition in a conditional group.
The condition could erroneously be treated as true if a branch matched but
overran the current position. This bug was in the interpreter only; matching
with JIT was correct.

14. (#443) Split out the sljit sub-project into a "Git submodule". Git users
must now run `git submodule init; git submodule update` after a Git checkout, or
the build will fail due to missing files in deps/sljit.

15. (#441) Add a new error code (PCRE2_ERROR_JIT_UNSUPPORTED) which is yielded
for unsupported jit features.

16. (#444) Fix bug in 'first code unit' and 'last code unit' optimization
combined with lookahead assertions.

17. (#445, #447, #449, #451, #452, #459, #563) Add a new feature called scan
substring. This feature is a new type of assertion which matches the content of
a capturing block to a sub-pattern.

18. (#450) Improvements to 'first code unit' / 'starting code units'
optimisation.

19. (#455) Many, many improvements to the JIT compiler.

20. Item 43 of 10.43 was incomplete because it addressed only \z and not \Z,
which was still misbehaving when matching fragments inside invalid UTF strings.

21. (d29e7290) Octal escapes of the form \045 or \111 were not being recognized
in substitution strings, and if encountered gave an error, though the \o{...}
form was recognized. This bug is now fixed.

22. (#463, #487) Fix 1 byte out-of-bounds read when parsing malformed limits
(e.g. LIMIT_HEAP)

23. Many improvements to test infrastructure. Many more platforms and
configurations are now run in Continuous Integration, and all the platforms now
run the full test suite, rather than a partial subset.

24. (#475) Implement title casing in substitution strings using Perl syntax.

25. (#478, #504) Disallow \x if not followed by { or a hex digit.

26. (#473) Implements Python-style backrefs in substitutions.

27. (#472) Fix error reporting for certain over-large octal escapes.

28. (#482) Fix parsing of named captures in replacement strings, allowing
non-ASCII capture names to be used.

29. (#477, #474, #488, #494, #496, #506, #508, #511, #518, #524, #540) Many
improvements to parsing and optimising of character classes.

30. (#483, #498) Add support for \g<n> and $<name> to replacement strings.

31. (#470) Add option flags PCRE2_EXTRA_NO_BS0 and PCRE2_EXTRA_PYTHON_OCTAL.

32. (#471) Add new API function pcre2_set_optimize() for controlling which
optimizations are enabled.

33. (#491) Adds $& $` $' and $_ to substitution replacements, as well as
interpreting \b and \v as characters.

34. (#499) Add option PCRE2_EXTRA_NEVER_CALLOUT to disable callouts.

35. (#503, #513) Update Unicode support to UCD 16.

36. (#512, #618, #638) Add new function pcre2_set_substitute_case_callout() to
allow clients to provide a custom callback with locale-aware case
transformation.

37. (#516) Fix case-insensitive matching of backreferences when using the
PCRE2_EXTRA_CASELESS_RESTRICT option.

38. (#519) In pcre2grep, add $& as an alias for $0

39. (c9bf8339, #534) Updated perltest.sh to enable locale setting.

40. (#521) Add support for Turkish I casefolding, using new options
PCRE2_EXTRA_TURKISH_CASING, and added pre-pattern flags (*TURKISH_CASING) and
(*CASELESS_RESTRICT).

41. (#523, #546, #547) Add support for UTS#18 compatible character classes,
using the new option PCRE2_ALT_EXTENDED_CLASS. This adds '[' as a metacharacter
within character classes and the operators '&&', '--' and '~~', allowing
subtractions and intersections of character classes to be easily expressed.

42. (#553, #586, #596, #597) Add support for Perl-style extended character
classes, using the syntax (?[...]). This also allows expressing subtractions and
intersections of character classes, but using a different syntax to UTS#18.

43. (#554) Fixed a bug in JIT affecting greedy bounded repeats. The upper limit
of repeats inside a repeated bracket might be incorrectly checked.

44. (#556) Fixed a bug in JIT affecting caseful matching of backreferences. When
utf is disabled, and dupnames is enabled, caseless matching was used even
if caseful matching was needed.

45. (f34fc0a3) Fixed a bug in pcre2grep reported by Alejandro Colomar
<alx@kernel.org> (GitHub issue #577). In certain cases, when lines of above and
below context were contiguous, a separator line was incorrectly being inserted.

46. (#594) Fix a small (one/two byte) out-of-bounds read on invalid UTF-8 input
in pcre2grep.

47. (#370) Fix the INSTALL_MSVC_PDB CMake flag.

48. (#366) Install cmake files in prefix/lib/cmake/pcre2 rather than
prefix/cmake. The new CMake flag PCRE2_INSTALL_CMAKEDIR allows customising this
location.

49. (#624, #626, #628, #632, #639, #641) Reduce code size of generated JIT code
for repeated character classes.

50. (#623) Update the Bazel build files.


Version 10.44 07-June-2024
--------------------------

1. If a pattern contained a variable-length lookbehind in which the first
branch was not the one with the shortest minimum length, and the lookbehind
contained a capturing group, and elsewhere in the pattern there was another
lookbehind that referenced that group, the pattern was incorrectly compiled,
leading to unpredictable results, including crashes in JIT compiling. An
example pattern is: /(((?<=123?456456|ABC)))(?<=\2)/

2. Further updates to the oss-fuzz support:

   (a) Limit quantifiers for groups and classes to be no more than 10. This
       avoids very long JIT compile times that happen in some cases when groups
       are replicated for quantification, and very long match times when
       classes contain a lot of non-ascii characters.

   (b) Added PCRE2_EXTENDED_MORE to the list of allowed options.

   (c) Arranged for text error messages to be shown in 16-bit and 32-bit modes.

   (d) Made the output in standalone mode more readable.

   (e) General code tidies.

   (f) Limit the size of compiled patterns to 10MB (see 6 below).

   (g) Do not run JIT on patterns whose compiled length is greater than 200K
       bytes because this takes a long time, causing oss-fuzz to time out.

   (h) Avoid compiling or matching twice with the same options (this could
       happen if the input didn't set any options).

3. Increase the maximum length of a name for a group from 32 to 128 because
there is a user for whom 32 is too small.

4. Cause pcre2test to output a message when pcre2_jit_compile() gives an error
return if either jitverify or info is specified.

5. Some auxiliary files for building under OpenVMS that were contributed by
Alexey Chupahin have been installed.

6. Added pcre2_set_max_pattern_compiled_length() to limit the size of compiled
patterns.

7. There was a bug in the implementation of \X caused by my (PH) misreading or
misunderstanding one of the grapheme sequence breaking rules in Unicode Annex
#29. A break should occur between two characters with the Extended Pictographic
break property unless a zero-width joiner intervenes. PCRE2 was not insisting
on the ZWJ, causing \X to match more than it should. See GitHub issue #410.

8. Avoid compilation issues with proprietary compilers in UNIX since 10.43.


Version 10.43 16-February-2024
------------------------------

1. The test program added by change 2 of 10.42 didn't work when the default
newline setting didn't include \n as a newline. One test needed (*LF) to ensure
that it worked.

2. Added the new freestanding POSIX test program to the ManyConfigTests script
in the maint directory (overlooked in 2 below). Also improved the selection
facilities in that script, and added a test with JIT in a non-source directory,
fixing an oversight that would have made such a test fail before.

3. Added pcre2_get_match_data_heapframes_size() and related pcre2test flags
to allow for finer control of the heap used when pcre2_match() without JIT is
used and the match_data might be reused. This began as PR #191, but has had
further refinement and documentation edits.

4. Applied PR #181, which tidies some casts in pcre2_valid_utf.c.

5. Applied PR #184, which avoids overflow issues with the heap limit
(introduced in 10.41/9).

6. Applied PR #192, which changes the timing units for pcre2test from
milliseconds to microseconds. This is more useful for modern CPUs.

7. Applied PR #193, which makes the requirement for C99 explicit in
configure.ac and CMakeLists.txt.

8. Fixed a bug in pcre2test when a ridiculously large string repeat required a
stupid amount of memory. It now gives a clean realloc() failure error.

9. Updates to restrict the interaction between ASCII and non-ASCII characters
for caseless matching and items like \d:

   (a) Added PCRE2_EXTRA_CASELESS_RESTRICT to lock out mixing of ASCII and
       non-ASCII when matching caselessly. This is also /r in pcre2test and
       (?r) within patterns.

   (b) Added PCRE2_EXTRA_ASCII_{BSD,BSS,BSW,POSIX} and corresponding (?aD) etc
       in patterns and /a in pcre2test.

   (c) Corresponding updates to pcre2test.

10. Unicode has been updated to 15.0.0.

11. The Python scripts and ucptest.c in maint have been updated (a) a minor
change needed for 9(a) above; (b) fix bugs in ucptest,

12. Integer overflow testing is now centralized in a new function.

13. Made PCRE2_UCP the default in UTF mode in pcre2grep, and added new options
--case-restrict and --no-ucp.

14. In the debugging printint module (which is normally only linked into
pcre2test), avoid the use of a variable called "not" because that's deprecated
in C and forbidden in C++. Also rewrite some code to avoid a goto into a block
that bypassed its initialization (though it didn't actually matter).

15. More minor code adjustments to avoid using reserved C++ words as variable
names ("new" and "typename") and another jump that bypassed an (irrelevant)
initialization.

16. Merged a pull request that removed pcre2_ucptables.c from the list of files
to compile in NON-AUTOTOOLS-BUILD because it is #included in pcre2_tables.c.
Also adjusted the BUILD.bazel and build.zig files, which had the same issue. At
the same time, fixed a typo in the Bazel file.

17. Add PCRE2_EXTRA_ASCII_DIGIT to allow [:digit:] to be kept on sync with \d
even in UCP mode.

18. Fix an invalid match of ascii word classes when invalid utf is enabled.

19. Add a --posix-digit to pcre2grep for compatibility with GNU grep, and
other tools that prefer the POSIX compatible unicode definition for \d.

20. Report the bit width of the library in use by pcre2test for usability.

21. A pathological pattern conversion test could result in a string longer than
the available input buffer. Cause such a test to fail.

22. Add a check that forces a compiler error if PCRE2_CODE_UNIT_WIDTH is not 8,
16, or 32 when compiling any of the library modules.

23. Update pcre2_compile() to treat a NULL pattern with zero length as an empty
string.

24. Add support for limited-length variable-length lookbehind assertions, with
default maximum length 255 characters (same as Perl) but with a function to
adjust the limit.

25. Applied pull request #262, which updates the zig configuration, and #278
which fixes a bug with out-of-source-tree CMake build testing.

26. Add support for LoongArch to JIT.

27. Fixed a bug in pcre2_match() in the code for handling the vector of
backtracking frames on the heap, which caused a heap overflow if *LIMIT_HEAP
restricted an attempt to extend to less than the frame size. Generally tidy up
the code for extending the heap frames vector. This fixes GitHub issue #275.

28. Update pcre2_fuzzsupport.c to avoid clang sanitize complaint about shifting
left by 16 when there are non-zeros in the top 16 bits.

29. Perl 5.34.0 changed the meaning of (for example) {,3} which did not used to
be treated as a quantifier. Now it is interpreted as {0,3} and PCRE2 has
changed to match. Note that {,} is still not a quantifier.

30. Perl allows spaces and/or horizontal tabs after { or before } in all items
that use braces, and also before or after the comma in quantifiers. PCRE2 now
does the same, except for \u{...}, which is recognized only when
PCRE2_EXTRA_ALT_BSUX is set. This an ECMAScript, non-Perl compatible,
extension, so PCRE2 follows ECMAScript rather than Perl.

31. Applied pull request #300 by Carlo, which fixes #261. The bug was that
pcre2_match() was not fully resetting all captures that had been set within a
(possibly recursive) subroutine call such as (?3).

32. Changed the meaning of \w (and its synonyms) in UCP mode to match Perl. It
now matches characters whose general categories are L or N or whose particular
categories are Mn (non-spacing mark) or Pc (combining punctuation). The latter
includes underscore.

33. Changed the meaning of [:xdigit:] in UCP mode to match Perl. It now also
matches the "fullwidth" versions of the hex digits. Just like it is done for
[:digit:], PCRE2_EXTRA_ASCII_DIGIT can be used to keep this class ASCII only
without affecting other POSIX classes.

34. GitHub PR305 fixes a potential integer overflow in pcre2_dfa_match().

35. Updated handling of \b and \B in UCP mode to match the changes to \w in 32
above because \b and \B are defined in terms of \w.

36. Within a pattern (?aT) and (?-aT) set and reset the PCRE2_EXTRA_ASCII_DIGIT
option, and (?aP) also sets (?aT) so that (?-aP) disables all ASCII
restrictions on POSIX classes.

37. If PCRE2_FIRSTLINE was set on an anchored pattern, pcre2_match() and
pcre2_dfa_match() misbehaved. PCRE2_FIRSTLINE is now ignored for anchored
patterns.

38. Add a test for ridiculous ovector offset values to the substring extraction
functions.

39. Make OP_REVERSE use IMM2_SIZE for its data instead of LINK_SIZE, for
consistency with OP_VREVERSE.

40. In some legacy environments with a pre C99 snprintf, pcre2_regerror could
return an incorrect value when the provided buffer was too small.

41. Applied pull request #342 which adds sanity checks for ctype functions and
locks out any accidental sign-extension.

42. In the 32-bit library, in non-UTF mode, a quantifier that followed a
literal character with a value greater than or equal to 0x80000000u caused
undefined behaviour.

43. \z was misbehaving when matching fragments inside invalid UTF strings.

44. Implement --group-separator and --no-group-separator for pcre2grep.

45. Fix \X matching in 32 bit mode without UTF in JIT.

46. Fix backref iterators when PCRE2_MATCH_UNSET_BACKREF is set in JIT.

47. Refactor the handling of whole-pattern recursion (?0) in pcre2_match() so
that its end is handled similarly to other recursions. This has altered the
behaviour of   /|(?0)./endanchored   which was previously not right.

48. Improved the test for looping recursion by checking the last referenced
character as well as the current character. This allows some patterns that
previously triggered the check to run to completion instead of giving the loop
error.

49. In 32-bit mode, the compiler looped for the pattern /[\x{ffffffff}]/ when
PCRE2_CASELESS and PCRE2_UCP (but not PCRE2_UTF) were set. Fixed by not trying
to look for other cases for characters above the Unicode range.

50. In caseless 32-bit mode with UCP (but not UTF) set, the character
0xffffffff incorrectly matched any character that has more than one other case,
in particular k and s.

51. Fix accept and endanchored interaction in JIT.

52. Fix backreferences with unset backref and non-greedy iterators in JIT.

53. Improve the logic that checks for a list of starting code units -- positive
lookahead assertions are now ignored if the immediately following item is one
that sets a mandatory starting character. For example, /a?(?=bc|)d/ used to set
all of a, b, and d as possible starting code units; now it sets only a and d.

54. Fix incorrect class character matches in JIT.

55. In pcre2test, ensure pcre2_jit_match() is used when jitfast is used with
substitution testing.

56. Insert omitted setting of subject length in match data at the end of
pcre2_jit_match().

57. Implemented PCRE2_DISABLE_RECURSELOOP_CHECK for pcre2_match() to enable
some apparently looping recursions to run to completion and therefore match the
JIT behaviour. With this set, real loops will eventually get caught by match or
heap limits or run out of resource.

58. AC did a lot of work on pcre2_fuzzsupport.c to extend it to 16-bit and
32-bit libraries and to compare JIT and non-JIT matching.


Version 10.42 11-December-2022
------------------------------

1. Change 19 of 10.41 wasn't quite right; it put the definition of a default,
empty value for PCRE2_CALL_CONVENTION in src/pcre2posix.c instead of
src/pcre2posix.h, which meant that programs that included pcre2posix.h but not
pcre2.h failed to compile.

2. To catch similar issues to the above in future, a new small test program
that includes pcre2posix.h but not pcre2.h has been added to the test suite.

3. When the -S option of pcre2test was used to set a stack size greater than
the allowed maximum, the error message displayed the hard limit incorrectly.
This was pointed out on GitHub pull request #171, but the suggested patch
didn't cope with all cases. Some further modification was required.

4. Supplying an ovector count of more than 65535 to pcre2_match_data_create()
caused a crash because the field in the match data block is only 16 bits. A
maximum of 65535 is now silently applied.

5. Merged @carenas patch #175 which fixes #86 - segfault on aarch64 (ARM),

6. The prototype for pcre2_substring_list_free() specified its argument as
PCRE2_SPTR * which is a const data type, whereas the yield from
pcre2_substring_list() is not const. This caused compiler warnings. I have
changed the argument of pcre2_substring_list_free() to be PCRE2_UCHAR ** to
remove this anomaly. This might cause new warnings in existing code where a
cast has been used to avoid previous ones.


Version 10.41 06-December-2022
------------------------------

1. Add fflush() before and after a fork callout in pcre2grep to get its output
to be the same on all systems. (There were previously ordering differences in
Alpine Linux).

2. Merged patch from @carenas (GitHub #110) for pthreads support in CMake.

3. SSF scorecards grumbled about possible overflow in an expression in
pcre2test. It never would have overflowed in practice, but some casts have been
added and at the some time there's been some tidying of fprints that output
size_t values.

4. PR #94 showed up an unused enum in pcre2_convert.c, which is now removed.

5. Minor code re-arrangement to remove gcc warning about realloc() in
pcre2test.

6. Change a number of int variables that hold buffer and line lengths in
pcre2grep to PCRE2_SIZE (aka size_t).

7. Added an #ifdef to cut out a call to PRIV(jit_free) when JIT is not
supported (even though that function would do nothing in that case) at the
request of a user who doesn't even want to link with pcre_jit_compile.o. Also
tidied up an untidy #ifdef arrangement in pcre2test.

8. Fixed an issue in the backtracking optimization of character repeats in
JIT. Furthermore optimize star repetitions, not just plus repetitions.

9. Removed the use of an initial backtracking frames vector on the system stack
in pcre2_match() so that it now always uses the heap. (In a multi-thread
environment with very small stacks there had been an issue.) This also is
tidier for JIT matching, which didn't need that vector. The heap vector is now
remembered in the match data block and re-used if that block itself is re-used.
It is freed with the match data block.

10. Adjusted the find_limits code in pcre2test to work with change 9 above.

11. Added find_limits_noheap to pcre2test, because the heap limits are now
different in different environments and so cannot be included in the standard
tests.

12. Created a test for pcre2_match() heap processing that is not part of the
tests run by 'make check', but can be run manually. The current output is from
a 64-bit system.

13. Implemented -Z aka --null in pcre2grep.

14. A minor change to pcre2test and the addition of several new pcre2grep tests
have improved LCOV coverage statistics. At the same time, code in pcre2grep and
elsewhere that can never be obeyed in normal testing has been excluded from
coverage.

15. Fixed a bug in pcre2grep that could cause an extra newline to be written
after output generated by --output.

16. If a file has a .bz2 extension but is not in fact compressed, pcre2grep
should process it as a plain text file. A bug stopped this happening; now fixed
and added to the tests.

17. When pcre2grep was running not in UTF mode, if a string specified by
--output or obtained from a callout in a pattern contained a character (byte)
greater than 127, it was incorrectly output in UTF-8 format.

18. Added some casts after warnings from Clang sanitize.

19. Merged patch from cbouc (GitHub #139): 4 function prototypes were missing
PCRE2_CALL_CONVENTION in src/pcre2posix.h. All function prototypes returning
pointers had out of place PCRE2_CALL_CONVENTION in src/pcre2.h.*. These
produced errors when building for Windows with #define PCRE2_CALL_CONVENTION
__stdcall.

20. A negative repeat value in a pcre2test subject line was not being
diagnosed, leading to infinite looping.

21. Updated RunGrepTest to discard the warning that Bash now gives when setting
LC_CTYPE to a bad value (because older versions didn't).

22. Updated pcre2grep so that it behaves like GNU grep when matching more than
one pattern and a later pattern matches at an earlier point in the subject when
the matched substrings are being identified by colour or by offsets.

23. Updated the PrepareRelease script so that the man page that it makes for
the pcre2demo demonstration program is more standard and does not cause errors
when processed by lexgrog or mandb -c (GitHub issue #160).

24. The JIT compiler was updated.


Version 10.40 15-April-2022
---------------------------

1. Merged patch from @carenas (GitHub #35, 7db87842) to fix pcre2grep incorrect
handling of multiple passes.

2. Merged patch from @carenas (GitHub #36, dae47509) to fix portability issue
in pcre2grep with buffered fseek(stdin).

3. Merged patch from @carenas (GitHub #37, acc520924) to fix tests when -S is
not supported.

4. Revert an unintended change in JIT repeat detection.

5. Merged patch from @carenas (GitHub #52, b037bfa1) to fix build on GNU Hurd.

6. Merged documentation and comments patches from @carenas (GitHub #47).

7. Merged patch from @carenas (GitHub #49) to remove obsolete JFriedl test code
from pcre2grep.

8. Merged patch from @carenas (GitHub #48) to fix CMake install issue #46.

9. Merged patch from @carenas (GitHub #53) fixing NULL checks in matching and
substituting.

10. Add null_subject and null_replacement modifiers to pcre2test.

11. Add check for NULL subject to POSIX regexec() function.

12. Add check for NULL replacement to pcre2_substitute().

13. For the subject arguments of pcre2_match(), pcre2_dfa_match(), and
pcre2_substitute(), and the replacement argument of the latter, if the pointer
is NULL and the length is zero, treat as an empty string. Apparently a number
of applications treat NULL/0 in this way.

14. Added support for Bidi_Class and a number of binary Unicode properties,
including Bidi_Control.

15. Fix some minor issues raised by clang sanitize.

16. Very minor code speed up for maximizing character property matches.

17. A number of changes to script matching for \p and \P:

    (a) Script extensions for a character are now coded as a bitmap instead of
        a list of script numbers, which should be faster and does not need a
        loop.

    (b) Added the syntax \p{script:xxx} and \p{script_extensions:xxx} (synonyms
        sc and scx).

    (c) Changed \p{scriptname} from being the same as \p{sc:scriptname} to being
        the same as \p{scx:scriptname} because this change happened in Perl at
        release 5.26.

    (d) The standard Unicode 4-letter abbreviations for script names are now
        recognized.

    (e) In accordance with Unicode and Perl's "loose matching" rules, spaces,
        hyphens, and underscores are ignored in property names, which are then
        matched independent of case.

18. The Python scripts in the maint directory have been refactored. There are
now three scripts that generate pcre2_ucd.c, pcre2_ucp.h, and pcre2_ucptables.c
(which is #included by pcre2_tables.c). The data lists that used to be
duplicated are now held in a single common Python module.

19. On CHERI, and thus Arm's Morello prototype, pointers are represented as
hardware capabilities, which consist of both an integer address and additional
metadata, meaning they are twice the size of the platform's size_t type, i.e.
16 bytes on a 64-bit system. The ovector member of heapframe happens to only be
8 byte aligned, and so computing frame_size ended up with a multiple of 8 but
not 16. Whilst the first frame was always suitably aligned, this then
misaligned the frame that follows, resulting in an alignment fault when storing
a pointer to Fecode at the start of match. Patch to fix this issue by Jessica
Clarke PR#72.

20. Added -LP and -LS listing options to pcre2test.

21. A user discovered that the library names in CMakeLists.txt for MSVC
debugger (PDB) files were incorrect - perhaps never tried for PCRE2?

22. An item such as [Aa] is optimized into a caseless single character match.
When this was quantified (e.g. [Aa]{2}) and was also the last literal item in a
pattern, the optimizing "must be present for a match" character check was not
being flagged as caseless, causing some matches that should have succeeded to
fail.

23. Fixed a unicode property matching issue in JIT. The character was not
fully read in caseless matching.

24. Fixed an issue affecting recursions in JIT caused by duplicated data
transfers.

25. Merged patch from @carenas (GitHub #96) which fixes some problems with
pcre2test and readline/readedit:

  * Use the right header for libedit in FreeBSD with autoconf
  * Really allow libedit with cmake
  * Avoid using readline headers with libedit


Version 10.39 29-October-2021
-----------------------------

1. Fix incorrect detection of alternatives in first character search in JIT.

2. Merged patch from @carenas (GitHub #28):

  Visual Studio 2013 includes support for %zu and %td, so let newer
  versions of it avoid the fallback, and while at it, make sure that
  the first check is for DISABLE_PERCENT_ZT so it will be always
  honoured if chosen.

  prtdiff_t is signed, so use a signed type instead, and make sure
  that an appropriate width is chosen if pointers are 64bit wide and
  long is not (ex: Windows 64bit).

  IMHO removing the cast (and therefore the possibility of truncation)
  make the code cleaner and the fallback is likely portable enough
  with all 64-bit POSIX systems doing LP64 except for Windows.

3. Merged patch from @carenas (GitHub #29) to update to Unicode 14.0.0.

4. Merged patch from @carenas (GitHub #30):

  * Cleanup: remove references to no longer used stdint.h

  Since 19c50b9d (Unconditionally use inttypes.h instead of trying for stdint.h
  (simplification) and remove the now unnecessary inclusion in
  pcre2_internal.h., 2018-11-14), stdint.h is no longer used.

  Remove checks for it in autotools and CMake and document better the expected
  build failures for systems that might have stdint.h (C99) and not inttypes.h
  (from POSIX), like old Windows.

  * Cleanup: remove detection for inttypes.h which is a hard dependency

  CMake checks for standard headers are not meant to be used for hard
  dependencies, so will prevent a possible fallback to work.

  Alternatively, the header could be checked to make the configuration fail
  instead of breaking the build, but that was punted, as it was missing anyway
  from autotools.

5. Merged patch from @carenas (GitHub #32):

  * jit: allow building with ancient MSVC versions

  Visual Studio older than 2013 fails to build with JIT enabled, because it is
  unable to parse non C89 compatible syntax, with mixed declarations and code.
  While most recent compilers wouldn't even report this as a warning since it
  is valid C99, it could be also made visible by adding to gcc/clang the
  -Wdeclaration-after-statement flag at build time.

  Move the code below the affected definitions.

  * pcre2grep: avoid mixing declarations with code

  Since d5a61ee8 (Patch to detect (and ignore) symlink loops in pcre2grep,
  2021-08-28), code will fail to build in a strict C89 compiler.

  Reformat slightly to make it C89 compatible again.


Version 10.38 01-October-2021
-----------------------------

1. Fix invalid single character repetition issues in JIT when the repetition
is inside a capturing bracket and the bracket is preceded by character
literals.

2. Installed revised CMake configuration files provided by Jan-Willem Blokland.
This extends the CMake build system to build both static and shared libraries
in one go, builds the static library with PIC, and exposes PCRE2 libraries
using the CMake config files. JWB provided these notes:

- Introduced CMake variable BUILD_STATIC_LIBS to build the static library.

- Make a small modification to config-cmake.h.in by removing the PCRE2_STATIC
  variable. Added PCRE2_STATIC variable to the static build using the
  target_compile_definitions() function.

- Extended the CMake config files.

  - Introduced CMake variable PCRE2_USE_STATIC_LIBS to easily switch between
    the static and shared libraries.

  - Added the PCRE_STATIC variable to the target compile definitions for the
    import of the static library.

Building static and shared libraries using MSVC results in a name clash of
the libraries. Both static and shared library builds create, for example, the
file pcre2-8.lib. Therefore, I decided to change the static library names by
adding "-static". For example, pcre2-8.lib has become pcre2-8-static.lib.
[Comment by PH: this is MSVC-specific. It doesn't happen on Linux.]

3. Increased the minimum release number for CMake to 3.0.0 because older than
2.8.12 is deprecated (it was set to 2.8.5) and causes warnings. Even 3.0.0 is
quite old; it was released in 2014.

4. Implemented a modified version of Thomas Tempelmann's pcre2grep patch for
detecting symlink loops. This is dependent on the availability of realpath(),
which is now tested for in ./configure and CMakeLists.txt.

5. Implemented a modified version of Thomas Tempelmann's patch for faster
case-independent "first code unit" searches for unanchored patterns in 8-bit
mode in the interpreters. Instead of just remembering whether one case matched
or not, it remembers the position of a previous match so as to avoid
unnecessary repeated searching.

6. Perl now locks out \K in lookarounds, so PCRE2 now does the same by default.
However, just in case anybody was relying on the old behaviour, there is an
option called PCRE2_EXTRA_ALLOW_LOOKAROUND_BSK that enables the old behaviour.
An option has also been added to pcre2grep to enable this.

7. Re-enable a JIT optimization which was unintentionally disabled in 10.35.

8. There is a loop counter to catch excessively crazy patterns when checking
the lengths of lookbehinds at compile time. This was incorrectly getting reset
whenever a lookahead was processed, leading to some fuzzer-generated patterns
taking a very long time to compile when (?|) was present in the pattern,
because (?|) disables caching of group lengths.


Version 10.37 26-May-2021
-------------------------

1. Change RunGrepTest to use tr instead of sed when testing with binary
zero bytes, because sed varies a lot from system to system and has problems
with binary zeros. This is from Bugzilla #2681. Patch from Jeremie
Courreges-Anglas via Nam Nguyen. This fixes RunGrepTest for OpenBSD. Later:
it broke it for at least one version of Solaris, where tr can't handle binary
zeros. However, that system had /usr/xpg4/bin/tr installed, which works OK, so
RunGrepTest now checks for that command and uses it if found.

2. Compiling with gcc 10.2's -fanalyzer option showed up a hypothetical problem
with a NULL dereference. I don't think this case could ever occur in practice,
but I have put in a check in order to get rid of the compiler error.

3. An alternative patch for CMakeLists.txt because 10.36 #4 breaks CMake on
Windows. Patch from email@cs-ware.de fixes bugzilla #2688.

4. Two bugs related to over-large numbers have been fixed so the behaviour is
now the same as Perl.

  (a) A pattern such as /\214748364/ gave an overflow error instead of being
  treated as the octal number \214 followed by literal digits.

  (b) A sequence such as {65536 that has no terminating } so is not a
  quantifier was nevertheless complaining that a quantifier number was too big.

5. A run of autoconf suggested that configure.ac was out-of-date with respect
to the latest autoconf. Running autoupdate made some valid changes, some valid
suggestions, and also some invalid changes, which were fixed by hand. Autoconf
now runs clean and the resulting "configure" seems to work, so I hope nothing
is broken. Later: the requirement for autoconf 2.70 broke some automatic test
robots. It doesn't seem to be necessary: trying a reduction to 2.60.

6. The pattern /a\K.(?0)*/ when matched against "abac" by the interpreter gave
the answer "bac", whereas Perl and JIT both yield "c". This was because the
effect of \K was not propagating back from the full pattern recursion. Other
recursions such as /(a\K.(?1)*)/ did not have this problem.

7. Restore single character repetition optimization in JIT. Currently fewer
character repetitions are optimized than in 10.34.

8. When the names of the functions in the POSIX wrapper were changed to
pcre2_regcomp() etc. (see change 10.33 #4 below), functions with the original
names were left in the library so that pre-compiled programs would still work.
However, this has proved troublesome when programs link with several libraries,
some of which use PCRE2 via the POSIX interface while others use a native POSIX
library. For this reason, the POSIX function names are removed in this release.
The macros in pcre2posix.h should ensure that re-compiling fixes any programs
that haven't been compiled since before 10.33.


Version 10.36 04-December-2020
------------------------------

1. Add CET_CFLAGS so that when Intel CET is enabled, pass -mshstk to
compiler. This fixes https://bugs.exim.org/show_bug.cgi?id=2578. Patch for
Makefile.am and configure.ac by H.J. Lu. Equivalent patch for CMakeLists.txt
invented by PH.

2. Fix infinite loop when a single byte newline is searched in JIT when
invalid utf8 mode is enabled.

3. Updated CMakeLists.txt with patch from Wolfgang Stöggl (Bugzilla #2584):

  - Include GNUInstallDirs and use ${CMAKE_INSTALL_LIBDIR} instead of hardcoded
    lib. This allows differentiation between lib and lib64.
    CMAKE_INSTALL_LIBDIR is used for installation of libraries and also for
    pkgconfig file generation.

  - Add the version of PCRE2 to the configuration summary like ./configure
    does.

  - Fix typo: MACTHED_STRING->MATCHED_STRING

4. Updated CMakeLists.txt with another patch from Wolfgang Stöggl (Bugzilla
#2588):

  - Add escaped double quotes around include directory in CMakeLists.txt to
    allow spaces in directory names.

  - This fixes a cmake error, if the path of the pcre2 source contains a space.

5. Updated CMakeLists.txt with a patch from B. Scott Michel: CMake's
documentation suggests using CHECK_SYMBOL_EXISTS over CHECK_FUNCTION_EXIST.
Moreover, these functions come from specific header files, which need to be
specified (and, thankfully, are the same on both the Linux and WinXX
platforms.)

6. Added a (uint32_t) cast to prevent a compiler warning in pcre2_compile.c.

7. Applied a patch from Wolfgang Stöggl (Bugzilla #2600) to fix postfix for
debug Windows builds using CMake. This also updated configure so that it
generates *.pc files and pcre2-config with the same content, as in the past.

8. If a pattern ended with (?(VERSION=n.d where n is any number but d is just a
single digit, the code unit beyond d was being read (i.e. there was a read
buffer overflow). Fixes ClusterFuzz 23779.

9. After the rework in r1235, certain character ranges were incorrectly
handled by an optimization in JIT. Furthermore a wrong offset was used to
read a value from a buffer which could lead to memory overread.
[--snip--]
