Find file History
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


README file for PCRE (Perl-compatible regular expression library)

The latest release of PCRE is always available from

There is a mailing list for discussion about the development of PCRE at

Please read the NEWS file if you are upgrading from a previous release.
The contents of this README file are:

  Documentation for PCRE
  Contributions by users of PCRE
  Building PCRE on non-Unix systems
  Building PCRE on Unix-like systems
  Retrieving configuration information on Unix-like systems
  Shared libraries on Unix-like systems
  Cross-compiling on Unix-like systems
  Using HP's ANSI C++ compiler (aCC)
  Making new tarballs
  Testing PCRE
  Character tables
  File manifest


PCRE is written in C, and it has its own API. The distribution also includes a
set of C++ wrapper functions (see the pcrecpp man page for details), courtesy
of Google Inc.

In addition, there is a set of C wrapper functions that are based on the POSIX
regular expression API (see the pcreposix man page). These end up in the
library called libpcreposix. Note that this just provides a POSIX calling
interface to PCRE; the regular expressions themselves still follow Perl syntax
and semantics. The POSIX API is restricted, and does not give full access to
all of PCRE's facilities.

The header file for the POSIX-style functions is called pcreposix.h. The
official POSIX name is regex.h, but I did not want to risk possible problems
with existing files of that name by distributing it that way. To use PCRE with
an existing program that uses the POSIX API, pcreposix.h will have to be
renamed or pointed at by a link.

If you are using the POSIX interface to PCRE and there is already a POSIX regex
library installed on your system, as well as worrying about the regex.h header
file (as mentioned above), you must also take care when linking programs to
ensure that they link with PCRE's libpcreposix library. Otherwise they may pick
up the POSIX functions of the same name from the other library.

One way of avoiding this confusion is to compile PCRE with the addition of
-Dregcomp=PCREregcomp (and similarly for the other POSIX functions) to the
compiler flags (CFLAGS if you are using "configure" -- see below). This has the
effect of renaming the functions so that the names no longer clash. Of course,
you have to do the same thing for your applications, or write them using the
new names.

Documentation for PCRE

If you install PCRE in the normal way on a Unix-like system, you will end up
with a set of man pages whose names all start with "pcre". The one that is just
called "pcre" lists all the others. In addition to these man pages, the PCRE
documentation is supplied in two other forms:

  1. There are files called doc/pcre.txt, doc/pcregrep.txt, and
     doc/pcretest.txt in the source distribution. The first of these is a
     concatenation of the text forms of all the section 3 man pages except
     those that summarize individual functions. The other two are the text
     forms of the section 1 man pages for the pcregrep and pcretest commands.
     These text forms are provided for ease of scanning with text editors or
     similar tools. They are installed in <prefix>/share/doc/pcre, where
     <prefix> is the installation prefix (defaulting to /usr/local).

  2. A set of files containing all the documentation in HTML form, hyperlinked
     in various ways, and rooted in a file called index.html, is distributed in
     doc/html and installed in <prefix>/share/doc/pcre/html.

Contributions by users of PCRE

You can find contributions from PCRE users in the directory

There is a README file giving brief descriptions of what they are. Some are
complete in themselves; others are pointers to URLs containing relevant files.
Some of this material is likely to be well out-of-date. Several of the earlier
contributions provided support for compiling PCRE on various flavours of
Windows (I myself do not use Windows). Nowadays there is more Windows support
in the standard distribution, so these contibutions have been archived.

Building PCRE on non-Unix systems

For a non-Unix system, please read the comments in the file NON-UNIX-USE,
though if your system supports the use of "configure" and "make" you may be
able to build PCRE in the same way as for Unix-like systems.

PCRE has been compiled on many different operating systems. It should be
straightforward to build PCRE on any system that has a Standard C compiler and
library, because it uses only Standard C functions.

Building PCRE on Unix-like systems

If you are using HP's ANSI C++ compiler (aCC), please see the special note
in the section entitled "Using HP's ANSI C++ compiler (aCC)" below.

The following instructions assume the use of the widely used "configure, make,
make install" process. There is also some experimental support for "cmake" in
the PCRE distribution, but it is incomplete and not documented. However, if you
are a "cmake" user, you might want to try it.

To build PCRE on a Unix-like system, first run the "configure" command from the
PCRE distribution directory, with your current directory set to the directory
where you want the files to be created. This command is a standard GNU
"autoconf" configuration script, for which generic instructions are supplied in
the file INSTALL.

Most commonly, people build PCRE within its own distribution directory, and in
this case, on many systems, just running "./configure" is sufficient. However,
the usual methods of changing standard defaults are available. For example:

CFLAGS='-O2 -Wall' ./configure --prefix=/opt/local

specifies that the C compiler should be run with the flags '-O2 -Wall' instead
of the default, and that "make install" should install PCRE under /opt/local
instead of the default /usr/local.

If you want to build in a different directory, just run "configure" with that
directory as current. For example, suppose you have unpacked the PCRE source
into /source/pcre/pcre-xxx, but you want to build it in /build/pcre/pcre-xxx:

cd /build/pcre/pcre-xxx

PCRE is written in C and is normally compiled as a C library. However, it is
possible to build it as a C++ library, though the provided building apparatus
does not have any features to support this.

There are some optional features that can be included or omitted from the PCRE
library. You can read more about them in the pcrebuild man page.

. If you want to suppress the building of the C++ wrapper library, you can add
  --disable-cpp to the "configure" command. Otherwise, when "configure" is run,
  it will try to find a C++ compiler and C++ header files, and if it succeeds,
  it will try to build the C++ wrapper.

. If you want to make use of the support for UTF-8 character strings in PCRE,
  you must add --enable-utf8 to the "configure" command. Without it, the code
  for handling UTF-8 is not included in the library. (Even when included, it
  still has to be enabled by an option at run time.)

. If, in addition to support for UTF-8 character strings, you want to include
  support for the \P, \p, and \X sequences that recognize Unicode character
  properties, you must add --enable-unicode-properties to the "configure"
  command. This adds about 30K to the size of the library (in the form of a
  property table); only the basic two-letter properties such as Lu are

. You can build PCRE to recognize either CR or LF or the sequence CRLF or any
  of the preceding, or any of the Unicode newline sequences as indicating the
  end of a line. Whatever you specify at build time is the default; the caller
  of PCRE can change the selection at run time. The default newline indicator
  is a single LF character (the Unix standard). You can specify the default
  newline indicator by adding --enable-newline-is-cr or --enable-newline-is-lf
  or --enable-newline-is-crlf or --enable-newline-is-anycrlf or
  --enable-newline-is-any to the "configure" command, respectively.

  If you specify --enable-newline-is-cr or --enable-newline-is-crlf, some of
  the standard tests will fail, because the lines in the test files end with
  LF. Even if the files are edited to change the line endings, there are likely
  to be some failures. With --enable-newline-is-anycrlf or
  --enable-newline-is-any, many tests should succeed, but there may be some

. When called via the POSIX interface, PCRE uses malloc() to get additional
  storage for processing capturing parentheses if there are more than 10 of
  them in a pattern. You can increase this threshold by setting, for example,


  on the "configure" command.

. PCRE has a counter that can be set to limit the amount of resources it uses.
  If the limit is exceeded during a match, the match fails. The default is ten
  million. You can change the default by setting, for example,


  on the "configure" command. This is just the default; individual calls to
  pcre_exec() can supply their own value. There is more discussion on the
  pcreapi man page.

. There is a separate counter that limits the depth of recursive function calls
  during a matching process. This also has a default of ten million, which is
  essentially "unlimited". You can change the default by setting, for example,


  Recursive function calls use up the runtime stack; running out of stack can
  cause programs to crash in strange ways. There is a discussion about stack
  sizes in the pcrestack man page.

. The default maximum compiled pattern size is around 64K. You can increase
  this by adding --with-link-size=3 to the "configure" command. You can
  increase it even more by setting --with-link-size=4, but this is unlikely
  ever to be necessary. Increasing the internal link size will reduce

. You can build PCRE so that its internal match() function that is called from
  pcre_exec() does not call itself recursively. Instead, it uses memory blocks
  obtained from the heap via the special functions pcre_stack_malloc() and
  pcre_stack_free() to save data that would otherwise be saved on the stack. To
  build PCRE like this, use


  on the "configure" command. PCRE runs more slowly in this mode, but it may be
  necessary in environments with limited stack sizes. This applies only to the
  pcre_exec() function; it does not apply to pcre_dfa_exec(), which does not
  use deeply nested recursion. There is a discussion about stack sizes in the
  pcrestack man page.

. For speed, PCRE uses four tables for manipulating and identifying characters
  whose code point values are less than 256. By default, it uses a set of
  tables for ASCII encoding that is part of the distribution. If you specify


  a program called dftables is compiled and run in the default C locale when
  you obey "make". It builds a source file called pcre_chartables.c. If you do
  not specify this option, pcre_chartables.c is created as a copy of
  pcre_chartables.c.dist. See "Character tables" below for further information.

. It is possible to compile PCRE for use on systems that use EBCDIC as their
  default character code (as opposed to ASCII) by specifying


  This automatically implies --enable-rebuild-chartables (see above).

The "configure" script builds the following files for the basic C library:

. Makefile is the makefile that builds the library
. config.h contains build-time configuration options for the library
. pcre.h is the public PCRE header file
. pcre-config is a script that shows the settings of "configure" options
. libpcre.pc is data for the pkg-config command
. libtool is a script that builds shared and/or static libraries
. RunTest is a script for running tests on the basic C library
. RunGrepTest is a script for running tests on the pcregrep command

Versions of config.h and pcre.h are distributed in the PCRE tarballs under
the names config.h.generic and pcre.h.generic. These are provided for the
benefit of those who have to built PCRE without the benefit of "configure". If
you use "configure", the .generic versions are not used.

If a C++ compiler is found, the following files are also built:

. libpcrecpp.pc is data for the pkg-config command
. pcrecpparg.h is a header file for programs that call PCRE via the C++ wrapper
. pcre_stringpiece.h is the header for the C++ "stringpiece" functions

The "configure" script also creates config.status, which is an executable
script that can be run to recreate the configuration, and config.log, which
contains compiler output from tests that "configure" runs.

Once "configure" has run, you can run "make". It builds two libraries, called
libpcre and libpcreposix, a test program called pcretest, a demonstration
program called pcredemo, and the pcregrep command. If a C++ compiler was found
on your system, "make" also builds the C++ wrapper library, which is called
libpcrecpp, and some test programs called pcrecpp_unittest,
pcre_scanner_unittest, and pcre_stringpiece_unittest. Building the C++ wrapper
can be disabled by adding --disable-cpp to the "configure" command.

The command "make check" runs all the appropriate tests. Details of the PCRE
tests are given below in a separate section of this document.

You can use "make install" to install PCRE into live directories on your
system. The following are installed (file names are all relative to the
<prefix> that is set when "configure" is run):

  Commands (bin):

  Libraries (lib):
    libpcrecpp (if C++ support is enabled)

  Configuration information (lib/pkgconfig):
    libpcrecpp.pc (if C++ support is enabled)

  Header files (include):
    pcre_scanner.h      )
    pcre_stringpiece.h  ) if C++ support is enabled
    pcrecpp.h           )
    pcrecpparg.h        )

  Man pages (share/man/man{1,3}):
    pcre*.3 (lots more pages, all starting "pcre")

  HTML documentation (share/doc/pcre/html):
    *.html (lots more pages, hyperlinked from index.html)

  Text file documentation (share/doc/pcre):
    pcre.txt       (a concatenation of the man(3) pages)
    pcretest.txt   the pcretest man page
    pcregrep.txt   the pcregrep man page

Note that the pcredemo program that is built by "configure" is *not* installed
anywhere. It is a demonstration for programmers wanting to use PCRE.

If you want to remove PCRE from your system, you can run "make uninstall".
This removes all the files that "make install" installed. However, it does not
remove any directories, because these are often shared with other programs.

Retrieving configuration information on Unix-like systems

Running "make install" installs the command pcre-config, which can be used to
recall information about the PCRE configuration and installation. For example:

  pcre-config --version

prints the version number, and

  pcre-config --libs

outputs information about where the library is installed. This command can be
included in makefiles for programs that use PCRE, saving the programmer from
having to remember too many details.

The pkg-config command is another system for saving and retrieving information
about installed libraries. Instead of separate commands for each library, a
single command is used. For example:

  pkg-config --cflags pcre

The data is held in *.pc files that are installed in a directory called

Shared libraries on Unix-like systems

The default distribution builds PCRE as shared libraries and static libraries,
as long as the operating system supports shared libraries. Shared library
support relies on the "libtool" script which is built as part of the
"configure" process.

The libtool script is used to compile and link both shared and static
libraries. They are placed in a subdirectory called .libs when they are newly
built. The programs pcretest and pcregrep are built to use these uninstalled
libraries (by means of wrapper scripts in the case of shared libraries). When
you use "make install" to install shared libraries, pcregrep and pcretest are
automatically re-built to use the newly installed shared libraries before being
installed themselves. However, the versions left in the build directory still
use the uninstalled libraries.

To build PCRE using static libraries only you must use --disable-shared when
configuring it. For example:

./configure --prefix=/usr/gnu --disable-shared

Then run "make" in the usual way. Similarly, you can use --disable-static to
build only shared libraries.

Cross-compiling on Unix-like systems

You can specify CC and CFLAGS in the normal way to the "configure" command, in
order to cross-compile PCRE for some other host. However, you should NOT
specify --enable-rebuild-chartables, because if you do, the dftables.c source
file is compiled and run on the local host, in order to generate the inbuilt
character tables (the pcre_chartables.c file). This will probably not work,
because dftables.c needs to be compiled with the local compiler, not the cross

When --enable-rebuild-chartables is not specified, pcre_chartables.c is created
by making a copy of pcre_chartables.c.dist, which is a default set of tables
that assumes ASCII code. Cross-compiling with the default tables should not be
a problem.

If you need to modify the character tables when cross-compiling, you should
move pcre_chartables.c.dist out of the way, then compile dftables.c by hand and
run it on the local host to make a new version of pcre_chartables.c.dist.
Then when you cross-compile PCRE this new version of the tables will be used.

Using HP's ANSI C++ compiler (aCC)

Unless C++ support is disabled by specifying the "--disable-cpp" option of the
"configure" script, you must include the "-AA" option in the CXXFLAGS
environment variable in order for the C++ components to compile correctly.

Also, note that the aCC compiler on PA-RISC platforms may have a defect whereby
needed libraries fail to get included when specifying the "-AA" compiler
option. If you experience unresolved symbols when linking the C++ programs,
use the workaround of specifying the following environment variable prior to
running the "configure" script:

  CXXLDFLAGS="-lstd_v2 -lCsup_v2"

Making new tarballs

The command "make dist" creates three PCRE tarballs, in tar.gz, tar.bz2, and
zip formats. The command "make distcheck" does the same, but then does a trial
build of the new distribution to ensure that it works.

If you have modified any of the man page sources in the doc directory, you
should first run the PrepareRelease script before making a distribution. This
script creates the .txt and HTML forms of the documentation from the man pages.

Testing PCRE

To test the basic PCRE library on a Unix system, run the RunTest script that is
created by the configuring process. There is also a script called RunGrepTest
that tests the options of the pcregrep command. If the C++ wrapper library is
built, three test programs called pcrecpp_unittest, pcre_scanner_unittest, and
pcre_stringpiece_unittest are also built.

Both the scripts and all the program tests are run if you obey "make check" or
"make test". For other systems, see the instructions in NON-UNIX-USE.

The RunTest script runs the pcretest test program (which is documented in its
own man page) on each of the testinput files in the testdata directory in
turn, and compares the output with the contents of the corresponding testoutput
files. A file called testtry is used to hold the main output from pcretest
(testsavedregex is also used as a working file). To run pcretest on just one of
the test files, give its number as an argument to RunTest, for example:

  RunTest 2

The first test file can also be fed directly into the script to
check that Perl gives the same results. The only difference you should see is
in the first few lines, where the Perl version is given instead of the PCRE

The second set of tests check pcre_fullinfo(), pcre_info(), pcre_study(),
pcre_copy_substring(), pcre_get_substring(), pcre_get_substring_list(), error
detection, and run-time flags that are specific to PCRE, as well as the POSIX
wrapper API. It also uses the debugging flags to check some of the internals of

If you build PCRE with a locale setting that is not the standard C locale, the
character tables may be different (see next paragraph). In some cases, this may
cause failures in the second set of tests. For example, in a locale where the
isprint() function yields TRUE for characters in the range 128-255, the use of
[:isascii:] inside a character class defines a different set of characters, and
this shows up in this test as a difference in the compiled code, which is being
listed for checking. Where the comparison test output contains [\x00-\x7f] the
test will contain [\x00-\xff], and similarly in some other cases. This is not a
bug in PCRE.

The third set of tests checks pcre_maketables(), the facility for building a
set of character tables for a specific locale and using them instead of the
default tables. The tests make use of the "fr_FR" (French) locale. Before
running the test, the script checks for the presence of this locale by running
the "locale" command. If that command fails, or if it doesn't include "fr_FR"
in the list of available locales, the third test cannot be run, and a comment
is output to say why. If running this test produces instances of the error

  ** Failed to set locale "fr_FR"

in the comparison output, it means that locale is not available on your system,
despite being listed by "locale". This does not mean that PCRE is broken.

[If you are trying to run this test on Windows, you may be able to get it to
work by changing "fr_FR" to "french" everywhere it occurs.]

The fourth test checks the UTF-8 support. It is not run automatically unless
PCRE is built with UTF-8 support. To do this you must set --enable-utf8 when
running "configure". This file can be also fed directly to the perltest script,
provided you are running Perl 5.8 or higher. (For Perl 5.6, a small patch,
commented in the script, can be be used.)

The fifth test checks error handling with UTF-8 encoding, and internal UTF-8
features of PCRE that are not relevant to Perl.

The sixth test checks the support for Unicode character properties. It it not
run automatically unless PCRE is built with Unicode property support. To to
this you must set --enable-unicode-properties when running "configure".

The seventh, eighth, and ninth tests check the pcre_dfa_exec() alternative
matching function, in non-UTF-8 mode, UTF-8 mode, and UTF-8 mode with Unicode
property support, respectively. The eighth and ninth tests are not run
automatically unless PCRE is build with the relevant support.

Character tables

For speed, PCRE uses four tables for manipulating and identifying characters
whose code point values are less than 256. The final argument of the
pcre_compile() function is a pointer to a block of memory containing the
concatenated tables. A call to pcre_maketables() can be used to generate a set
of tables in the current locale. If the final argument for pcre_compile() is
passed as NULL, a set of default tables that is built into the binary is used.

The source file called pcre_chartables.c contains the default set of tables. By
default, this is created as a copy of pcre_chartables.c.dist, which contains
tables for ASCII coding. However, if --enable-rebuild-chartables is specified
for ./configure, a different version of pcre_chartables.c is built by the
program dftables (compiled from dftables.c), which uses the ANSI C character
handling functions such as isalnum(), isalpha(), isupper(), islower(), etc. to
build the table sources. This means that the default C locale which is set for
your system will control the contents of these default tables. You can change
the default tables by editing pcre_chartables.c and then re-building PCRE. If
you do this, you should take care to ensure that the file does not get
automatically re-generated. The best way to do this is to move
pcre_chartables.c.dist out of the way and replace it with your customized

When the dftables program is run as a result of --enable-rebuild-chartables,
it uses the default C locale that is set on your system. It does not pay
attention to the LC_xxx environment variables. In other words, it uses the
system's default locale rather than whatever the compiling user happens to have
set. If you really do want to build a source set of character tables in a
locale that is specified by the LC_xxx variables, you can run the dftables
program by hand with the -L option. For example:

  ./dftables -L pcre_chartables.c.special

The first two 256-byte tables provide lower casing and case flipping functions,
respectively. The next table consists of three 32-byte bit maps which identify
digits, "word" characters, and white space, respectively. These are used when
building 32-byte bit maps that represent character classes for code points less
than 256.

The final 256-byte table has bits indicating various character types, as

    1   white space character
    2   letter
    4   decimal digit
    8   hexadecimal digit
   16   alphanumeric or '_'
  128   regular expression metacharacter or binary zero

You should not alter the set of characters that contain the 128 bit, as that
will cause PCRE to malfunction.

File manifest

The distribution should contain the following files:

(A) Source files of the PCRE library functions and their headers:

  dftables.c              auxiliary program for building pcre_chartables.c
                            when --enable-rebuild-chartables is specified

  pcre_chartables.c.dist  a default set of character tables that assume ASCII
                            coding; used, unless --enable-rebuild-chartables is
                            specified, by copying to pcre_chartables.c

  pcreposix.c             )
  pcre_compile.c          )
  pcre_config.c           )
  pcre_dfa_exec.c         )
  pcre_exec.c             )
  pcre_fullinfo.c         )
  pcre_get.c              ) sources for the functions in the library,
  pcre_globals.c          )   and some internal functions that they use
  pcre_info.c             )
  pcre_maketables.c       )
  pcre_newline.c          )
  pcre_ord2utf8.c         )
  pcre_refcount.c         )
  pcre_study.c            )
  pcre_tables.c           )
  pcre_try_flipped.c      )
  pcre_ucp_searchfuncs.c  )
  pcre_valid_utf8.c       )
  pcre_version.c          )
  pcre_xclass.c           )
  pcre_printint.src       ) debugging function that is #included in pcretest,
                          )   and can also be #included in pcre_compile()               template for pcre.h when built by "configure"
  pcreposix.h             header for the external POSIX wrapper API
  pcre_internal.h         header for internal use
  ucp.h                   ) headers concerned with
  ucpinternal.h           )   Unicode property handling
  ucptable.h              ) (this one is the data table)             template for config.h, which is built by "configure"

  pcrecpp.h               public header file for the C++ wrapper         template for another C++ header file
  pcre_scanner.h          public header file for C++ scanner functions              )         ) source for the C++ wrapper library   template for pcre_stringpiece.h, the header for the
                            C++ stringpiece functions     source for the C++ stringpiece functions

(B) Source files for programs that use PCRE:

  pcredemo.c              simple demonstration of coding calls to PCRE
  pcregrep.c              source of a grep utility that uses PCRE
  pcretest.c              comprehensive test program

(C) Auxiliary files:

  132html                 script to turn "man" pages into HTML
  AUTHORS                 information about the author of PCRE
  ChangeLog               log of changes to the code
  CleanTxt                script to clean nroff output for txt man pages
  Detrail                 script to remove trailing spaces
  HACKING                 some notes about the internals of PCRE
  INSTALL                 generic installation instructions
  LICENCE                 conditions for the use of PCRE
  COPYING                 the same, using GNU's standard name             ) template for Unix Makefile, which is built by
                          )   "configure"             ) the automake input that was used to create
  NEWS                    important changes in this release
  NON-UNIX-USE            notes on building PCRE on non-Unix systems
  PrepareRelease          script to make preparations for "make dist"
  README                  this file
  RunTest                 a Unix shell script for running tests
  RunGrepTest             a Unix shell script for pcregrep tests
  aclocal.m4              m4 macros (generated by "aclocal")
  config.guess            ) files used by libtool,
  config.sub              )   used only when building a shared library
  configure               a configuring shell script (built by autoconf)            ) the autoconf input that was used to build
                          )   "configure" and config.h
  depcomp                 ) script to find program dependencies, generated by
                          )   automake
  doc/*.3                 man page sources for the PCRE functions
  doc/*.1                 man page sources for pcregrep and pcretest
  doc/index.html.src      the base HTML page
  doc/html/*              HTML documentation
  doc/pcre.txt            plain text version of the man pages
  doc/pcretest.txt        plain text documentation of test program
  doc/perltest.txt        plain text documentation of Perl test program
  install-sh              a shell script for installing files           template for libpcre.pc for pkg-config        template for libpcrecpp.pc for pkg-config               file used to build a libtool script
  missing                 ) common stub for a few missing GNU programs while
                          )   installing, generated by automake
  mkinstalldirs           script for making install directories             Perl test program          source of script which retains PCRE information          )     ) test programs for the C++ wrapper )
  testdata/testinput*     test data for main library tests
  testdata/testoutput*    expected test results
  testdata/grep*          input and output for pcregrep tests

(D) Auxiliary files for cmake support


(E) Auxiliary files for VPASCAL


(F) Auxiliary files for building PCRE "by hand"

  pcre.h.generic          ) a version of the public PCRE header file
                          )   for use in non-"configure" environments
  config.h.generic        ) a version of config.h for use in non-"configure"
                          )   environments

(F) Miscellaneous

  RunTest.bat            a script for running tests under Windows

Philip Hazel
Email local part: ph10
Email domain:
Last updated: 24 April 2007