zdoom/raze-gles
0
0
Fork 0
mirror of https://github.com/ZDoom/raze-gles.git synced 2025-01-11 18:50:46 +00:00
Code Issues Projects Releases Packages Wiki Activity

- added compile tools from GZDoom repo:

Browse source 
- updaterevision for automated revision string generation
- re2c as a prerequisite to use sc_man in the future
- zipdir to automatically generate an engine resource file.
This commit is contained in:
Christoph Oelckers 2019-09-22 08:52:30 +02:00
parent 0d98e7f256
commit 8c95516224
161 changed files with 29182 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

7
tools/CMakeLists.txt Normal file
View file

@ -0,0 +1,7 @@
cmake_minimum_required( VERSION 2.8.7 )
add_subdirectory( re2c )
add_subdirectory( updaterevision )
add_subdirectory( zipdir )
set( CROSS_EXPORTS ${CROSS_EXPORTS} PARENT_SCOPE )

403
tools/re2c/CHANGELOG Normal file
View file

@ -0,0 +1,403 @@
Version 0.16 (2016-01-21)
---------------------------
- Fixed bug #127 "code generation error with wide chars and bitmaps (omitted 'goto' statement)"
- Added DFA minimization and option '--dfa-minimization <table | moore>'
- Fixed bug #128 "very slow DFA construction (resulting in a very large DFA)"
- Fixed bug #132 "test failure on big endian archs with 0.15.3"
Version 0.15.3 (2015-12-02)
---------------------------
- Fixed bugs and applied patches:
#122 "clang does not compile re2c 0.15.x" (reported and fixed by Oleksii Taran).
#124 "Get rid of UINT32_MAX and friends" (patch by Sergei Trofimovich, fixes FreeBSD builds).
#125 "[OS X] git reports changes not staged for commit in newly cloned repository" (by Oleksii Taran, this fix also applies to Windows).
- Added option --no-version that allows to omit version information.
- Reduced memory and time consumed with -Wundefined-control-flow.
- Improved coverage of input data generated with -S --skeleton.
Version 0.15.2 (2015-11-23)
---------------------------
- Fixed build system: lexer depends on bison-generated parser
(Gentoo bug: https://bugs.gentoo.org/show_bug.cgi?id=566620)
Version 0.15.1 (2015-11-22)
---------------------------
- Fixed test failures caused by locale-sensitive 'sort'.
Version 0.15 (2015-11-22)
-------------------------
- Updated website http://re2c.org:
added examples
updated docs
added news
added web feed (Atom 1.0)
- Added options:
-S, --skeleton
--empty-class <match-empty | match-none | error>
- Added warnings:
-W
-Werror
-W<warning>
-Wno-<warning>
-Werror-<warning>
-Wno-error-<warning>
- Added individual warnings:
-Wundefined-control-flow
-Wunreachable-rules
-Wcondition-order
-Wuseless-escape
-Wempty-character-class
-Wswapped-range
-Wmatch-empty-string
- Fixed options:
-- (interpret remaining arguments as non-options)
- Deprecated options:
-1 --single-pass (single pass is by default now)
- Reduced size of the generated .dot files.
- Fixed bugs:
#27 re2c crashes reading files containing %{ %} (patch by Rui)
#51 default rule doesn't work in reuse mode
#52 eliminate multiple passes
#59 bogus yyaccept in -c mode
#60 redundant use of YYMARKER
#61 empty character class [] matches empty string
#115 flex-style named definitions cause ambiguity in re2c grammar
#119 -f with -b/-g generates incorrect dispatch on fill labels
#116 empty string with non-empty trailing context consumes code units
- Added test options:
-j, -j <N> (run tests in N threads, defaults to the number of CPUs)
--wine (test windows builds using wine)
--skeleton (generate skeleton programs, compile and execute them)
--keep-tmp-files (don't delete intermediate files for successful tests)
- Updated build system:
support out of source builds
support `make distcheck`
added `make bootstrap` (rebuild re2c after building with precomplied .re files)
added `make tests` (run tests with -j)
added `make vtests` (run tests with --valgrind -j)
added `make wtests` (run tests with --wine -j 1)
added Autoconf tests for CXXFLAGS. By default try the following options:
-W -Wall -Wextra -Weffc++ -pedantic -Wformat=2 -Wredundant-decls
-Wsuggest-attribute=format -Wconversion -Wsign-conversion -O2 -Weverything),
respect user-defined CXXFLAGS
support Mingw builds: `configure -host i686-w64-mingw32`
structured source files
removed old MSVC files
- Moved development to github (https://github.com/skvadrik/re2c), keep a mirror on sourceforge.
Version 0.14.3 (2015-05-20)
-----------------------------
- applied patch '#27 re2c crashes reading files containing %{ %}' by Rui
- dropped distfiles for MSVC (they are broken anyway)
Version 0.14.2 (2015-03-25)
-----------------------------
- fixed #57 Wrong result only if another rule is present
Version 0.14.1 (2015-02-27)
-----------------------------
- fixed #55 re2c-0.14: re2c -V outputs null byte
Version 0.14 (2015-02-23)
-----------------------------
- Added generic input API 21 (#21 Support to configure how re2c code interfaced with the symbol buffer?)
- fixed #46 re2c generates an infinite loop, depends on existence of previous parser
- fixed #47 Dot output label escaped characters
Version 0.13.7.5 (2014-08-22)
-----------------------------
- Fixed Gentoo bug: https://bugs.gentoo.org/show_bug.cgi?id=518904 (PHP lexer)
Version 0.13.7.4 (2014-07-29)
-----------------------------
- Enabled 'make docs' only if configured with '--enable-docs'
- Disallowed to use yacc/byacc instead of bison to build parser
- Removed non-portable sed feature in script that runs tests
Version 0.13.7.3 (2014-07-28)
-----------------------------
- Fixed CXX warning
- Got rid of asciidoc build-time dependency
Version 0.13.7.2 (2014-07-27)
-----------------------------
- Included man page into dist, respect users CXXFLAGS.
Version 0.13.7.1 (2014-07-26)
-----------------------------
- Added missing files to tarball
Version 0.13.7 (2014-07-25)
---------------------------
- Added UTF-8 support
- Added UTF-16 support
- Added default rule
- Added option to control ill-formed Unicode
Version 0.13.6 (2013-07-04)
---------------------------
- Fixed #2535084 uint problem with Sun C 5.8
- #3308400: allow Yacc-style %{code brackets}%
- #2506253: allow C++ // comments
- Fixed inplace configuration in -e mode.
- Applied #2482572 Typos in error messages.
- Applied #2482561 Error in manual section on -r mode.
- Fixed #2478216 Wrong start_label in -c mode.
- Fixed #2186718 Unescaped backslash in file name of #line directive.
- Fixed #2102138 Duplicate case labels on EBCDIC.
- Fixed #2088583 Compile problem on AIX.
- Fixed #2038610 Ebcdic problem.
- improve dot support: make char intervals (e.g. [A-Z]) instead of one edge per char
Version 0.13.5 (2008-05-25)
---------------------------
- Fixed #1952896 Segfault in re2c::Scanner::scan.
- Fixed #1952842 Regression.
Version 0.13.4 (2008-04-05)
---------------------------
- Added transparent handling of #line directives in input files.
- Added re2c:yyfill:check inplace configuration.
- Added re2c:define:YYSETSTATE:naked inplace configuration.
- Added re2c:flags:w and re2c:flags:u inplace configurations.
- Added the ability to add rules in 'use:re2c' blocks.
- Changed -r flag to accept only 'rules:re2c' and 'use:re2c' blocks.
Version 0.13.3 (2008-03-14)
---------------------------
- Added -r flag to allow reuse of scanner definitions.
- Added -F flag to support flex syntax in rules.
- Fixed SEGV in scanner that occurs with very large blocks.
- Fixed issue with unused yybm.
- Partial support for flex syntax.
- Changed to allow /* comments with -c switch.
- Added flag -D/--emit-dot.
Version 0.13.2 (2008-02-14)
---------------------------
- Added flag --case-inverted.
- Added flag --case-insensitive.
- Added support for '<!...>' to enable rule setup.
- Added support for '=>' style rules.
- Added support for ':=' style rules.
- Added support for ':=>' style rules.
- Added re2c:cond:divider and re2c:con:goto inplace configuration.
- Fixed code generation to emit space after 'if'.
Version 0.13.1 (2007-08-24)
---------------------------
- Added custom build rules for Visual Studio 2005 (re2c.rules). (William Swanson)
- Fixed issue with some compilers.
- Fixed #1776177 Build on AIX.
- Fixed #1743180 fwrite with 0 length crashes on OS X.
Version 0.13.0 (2007-06-24)
---------------------------
- Added -c and -t to generate scanners with (f)lex-like condition support.
- Fixed issue with short form of switches and parameter if not first switch.
- Fixed #1708378 segfault in actions.cc.
Version 0.12.3 (2007-08-24)
---------------------------
- Fixed issue with some compilers.
- Fixed #1776177 Build on AIX.
- Fixed #1743180 fwrite with 0 length crashes on OS X.
Version 0.12.2 (2007-06-26)
---------------------------
- Fixed #1743180 fwrite with 0 length crashes on OS X.
Version 0.12.1 (2007-05-23)
---------------------------
- Fixed #1711240 problem with '"' and 7F on EBCDIC plattforms.
Version 0.12.0 (2007-05-01)
---------------------------
- Re-release of 0.11.3 as new stable branch.
- Fixed issue with short form of switches and parameter if not first switch.
- Fixed #1708378 segfault in actions.cc.
Version 0.11.3 (2007-04-01)
---------------------------
- Added support for underscores in named definitions.
- Added new option --no-generation-date.
- Fixed issue with long form of switches.
Version 0.11.2 (2007-03-01)
---------------------------
- Added inplace configuration 're2c:yyfill:parameter'.
- Added inplace configuration 're2c:yych:conversion'.
- Fixed -u switch code generation.
- Added ability to avoid defines and overwrite variable and label names.
Version 0.11.1 (2007-02-20)
---------------------------
- Applied #1647875 add const to yybm vector.
Version 0.11.0 (2007-01-01)
---------------------------
- Added -u switch to support unicode.
Version 0.10.8 (2007-04-01)
---------------------------
- Fixed issue with long form of switches.
Version 0.10.7 (2007-02-20)
---------------------------
- Applied #1647875 add const to yybm vector.
Version 0.10.6 (2006-08-05)
---------------------------
- Fixed #1529351 Segv bug on unterminated code blocks.
- Fixed #1528269 Invalid code generation.
Version 0.10.5 (2006-06-11)
---------------------------
- Fixed long form of -1 switch to --single-pass as noted in man page and help.
- Added MSVC 2003 project files and renamed old 2002 ones.
Version 0.10.4 (2006-06-01)
---------------------------
- Fix whitespace in generated code.
Version 0.10.3 (2006-05-14)
---------------------------
- Fixed issue with -wb and -ws.
- Added -g switch to support gcc's computed goto's.
- Changed to use nested if's instead of "switch(yyaccept)" in -s mode.
Version 0.10.2 (2006-05-01)
---------------------------
- Changed to generate YYMARKER only when needed or in single pass mode.
- Added -1 switch to force single pass generation and make two pass the default.
- Fixed -i switch.
- Added configuration 'yyfill:enable' to allow suppression of YYFILL() blocks.
- Added tutorial like lessons to re2c.
- Added /*!ignore:re2c */ to support documenting of re2c source.
- Fixed issue with multiline re2c comments (/*!max:re2c ... */ and alike).
- Fixed generation of YYDEBUG() when using -d switch.
- Added /*!getstate:re2c */ which triggers generation of the YYGETSTATE() block.
- Added configuration 'state:abort'.
- Changed to not generate yyNext unless configuration 'state:nextlabel' is used.
- Changed to not generate yyaccept code unless needed.
- Changed to use if- instead of switch-expression when yyaccpt has only one case.
- Added docu, examples and tests to .src.zip package (0.10.1 zip was repackaged).
- Fixed #1479044 incorrect code generated when using -b.
- Fixed #1472770 re2c creates an infinite loop.
- Fixed #1454253 Piece of code saving a backtracking point not generated.
- Fixed #1463639 Missing forward declaration.
- Implemented #1187127 savable state support for multiple re2c blocks.
Version 0.10.1 (2006-02-28)
---------------------------
- Added support for Solaris and native SUN compiler.
- Applied #1438160 expose YYCTXMARKER.
Version 0.10.0 (2006-02-18)
---------------------------
- Added make target zip to create windows source packages as zip files.
- Added re2c:startlabel configuration.
- Fixed code generation to not generate unreachable code for initial state.
- Added support for c/c++ compatible \u and \U unicode notation.
- Added ability to control indendation.
- Made scanner error out in case an ambiguous /* is found.
- Fixed indendation of generated code.
- Added support for DOS line endings.
- Added experimental unicode support.
- Added config_w32.h to build out of the box on windows (using msvc 2002+).
- Added Microsoft Visual C .NET 2005 build files.
- Applied #1411087 variable length trailing context.
- Applied #1408326 do not generate goto next state.
- Applied #1408282 CharSet initialization fix.
- Applied #1408278 readsome with MSVC.
- Applied #1307467 Unicode patch for 0.9.7.
Version 0.9.12 (2005-12-28)
---------------------------
- Fixed bug #1390174 re2c cannot accept {0,}.
Version 0.9.11 (2005-12-18)
---------------------------
- Fixed #1313083 -e (EBCDIC cross compile) broken.
- Fixed #1297658 underestimation of n in YYFILL(n).
- Applied #1339483 Avoid rebuilds of re2c when running subtargets.
- Implemented #1335305 symbol table reimplementation, just slightly modifed.
Version 0.9.10 (2005-09-04)
---------------------------
- Added -i switch to avoid generating #line information.
- Fixed bug #1251653 re2c generate some invalid #line on WIN32.
Version 0.9.9 (2005-07-21)
--------------------------
- Implemented #1232777 negated char classes '[^...]' and the dot operator '.'.
- Added hexadecimal character definitions.
- Added consistency check for octal character definitions.
Version 0.9.8 (2005-06-26)
--------------------------
- Fixed code generation for -b switch.
- Added Microsoft Visual C .NET build files.
Version 0.9.7 (2005-04-30)
--------------------------
- Applied #1181535 storable state patch.
- Added -d flag which outputs a debugable parser.
- Fixed generation of '#line' directives (according to ISO-C99).
- Fixed bug #1187785 Re2c fails to generate valid code.
- Fixed bug #1187452 unused variable `yyaccept'.
Version 0.9.6 (2005-04-14)
--------------------------
- Fixed build with gcc >= 3.4.
Version 0.9.5 (2005-04-08)
--------------------------
- Added /*!max:re2c */ which emits a '#define YYMAXFILL <max>\n' line
This allows to define buffers of the minimum required length. Occurence
must follow '/*re2c */ and cannot preceed it.
- Changed re2c to two pass generation to output warning free code.
- Fixed bug #1163046 re2c hangs when processing valid re-file.
- Fixed bug #1022799 re2c scanner has buffering bug.
Version 0.9.4 (2005-03-12)
--------------------------
- Added --vernum support.
- Fixed bug #1054496 incorrect code generated with -b option.
- Fixed bug #1012748 re2c does not emit last line if '\n' missing.
- Fixed bug #999104 --output=output option does not work as documented.
- Fixed bug #999103 Invalid options prefixed with two dashes cause program crash.
Version 0.9.3 (2004-05-26)
--------------------------
- Fixes one small possible bug in the generated output. ych instead of yych is
output in certain circumstances
Version 0.9.2 (2004-05-26)
--------------------------
- Added -o option to specify the output file which also will set the #line
directives to something useful.
- Print version to cout instead cerr.
- Added -h and -- style options.
- Moved development to http://sourceforge.net/projects/re2c .
- Fixed bug #960144 minor cosmetic problem.
- Fixed bug #953181 cannot compile with.
- Fixed bug #939277 Windows support.
- Fixed bug #914462 automake build patch
- Fixed bug #891940 braced quantifiers: {\d+(,|,\d+)?} style.
- Fixed bug #869298 Add case insensitive string literals.
- Fixed bug #869297 Input buffer overrun.
Version 0.9.1 (2003-12-13)
--------------------------
- Removed rcs comments in source files.
Version 0.9
-----------
- Redistribution based on version 0.5.
- Added parentheses to assignment expressions in 'if' statements.
- Rearranged class members to match initialization order.
- Substr fix.
- Use array delete [] when necessary.
- Other minor fixes for subduing compiler warnings.

104
tools/re2c/CMakeLists.txt Normal file
View file

@ -0,0 +1,104 @@
cmake_minimum_required( VERSION 2.8.7 )
if( NOT CMAKE_CROSSCOMPILING )
include( CheckFunctionExists )
include( CheckTypeSize )
if( MSVC )
# Runtime type information is required and don't complain about uint32_t to bool conversions
set( CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /GR /wd4800" )
endif()
set( PACKAGE_NAME re2c )
set( PACKAGE_TARNAME re2c )
set( PACKAGE_VERSION 0.16 )
set( PACKAGE_STRING "re2c 0.16" )
set( PACKAGE_BUGREPORT "re2c-general@lists.sourceforge.net" )
CHECK_FUNCTION_EXISTS( strdup HAVE_STRDUP )
CHECK_FUNCTION_EXISTS( strndup HAVE_STRNDUP )
CHECK_TYPE_SIZE( "0i8" SIZEOF_0I8 )
CHECK_TYPE_SIZE( "0l" SIZEOF_0L )
CHECK_TYPE_SIZE( "0ll" SIZEOF_0LL )
CHECK_TYPE_SIZE( char SIZEOF_CHAR )
CHECK_TYPE_SIZE( short SIZEOF_SHORT )
CHECK_TYPE_SIZE( int SIZEOF_INT )
CHECK_TYPE_SIZE( long SIZEOF_LONG )
CHECK_TYPE_SIZE( "long long" SIZEOF_LONG_LONG )
CHECK_TYPE_SIZE( "void *" SIZEOF_VOID_P )
CHECK_TYPE_SIZE( __int64 SIZEOF___INT_64 )
configure_file( ${CMAKE_CURRENT_SOURCE_DIR}/config.h.in ${CMAKE_CURRENT_BINARY_DIR}/config.h )
include_directories( ${CMAKE_CURRENT_BINARY_DIR} ${CMAKE_CURRENT_SOURCE_DIR} )
add_definitions( -DHAVE_CONFIG_H )
file( GLOB SRC_HDR
src/codegen/*.h
src/conf/*.h
src/ir/*.h
src/*.h
src/parse/*.h
src/util/*.h )
add_executable( re2c
${SRC_HDR}
src/codegen/bitmap.cc
src/codegen/emit_action.cc
src/codegen/emit_dfa.cc
src/codegen/label.cc
src/codegen/go_construct.cc
src/codegen/go_destruct.cc
src/codegen/go_emit.cc
src/codegen/go_used_labels.cc
src/codegen/input_api.cc
src/codegen/output.cc
src/codegen/print.cc
src/conf/msg.cc
src/conf/opt.cc
src/conf/parse_opts.cc
src/conf/warn.cc
src/ir/nfa/calc_size.cc
src/ir/nfa/nfa.cc
src/ir/nfa/split.cc
src/ir/adfa/adfa.cc
src/ir/adfa/prepare.cc
src/ir/dfa/determinization.cc
src/ir/dfa/fillpoints.cc
src/ir/dfa/minimization.cc
src/ir/regexp/display.cc
src/ir/regexp/encoding/enc.cc
src/ir/regexp/encoding/range_suffix.cc
src/ir/regexp/encoding/utf8/utf8_regexp.cc
src/ir/regexp/encoding/utf8/utf8_range.cc
src/ir/regexp/encoding/utf8/utf8.cc
src/ir/regexp/encoding/utf16/utf16_regexp.cc
src/ir/regexp/encoding/utf16/utf16.cc
src/ir/regexp/encoding/utf16/utf16_range.cc
src/ir/regexp/fixed_length.cc
src/ir/regexp/regexp.cc
src/ir/compile.cc
src/ir/rule_rank.cc
src/ir/skeleton/control_flow.cc
src/ir/skeleton/generate_code.cc
src/ir/skeleton/generate_data.cc
src/ir/skeleton/match_empty.cc
src/ir/skeleton/maxlen.cc
src/ir/skeleton/skeleton.cc
src/ir/skeleton/unreachable.cc
src/ir/skeleton/way.cc
src/main.cc
src/parse/code.cc
src/parse/input.cc
src/parse/lex.cc
src/parse/lex_conf.cc
src/parse/parser.cc
src/parse/scanner.cc
src/parse/unescape.cc
src/util/s_to_n32_unsafe.cc
src/util/range.cc )
set( CROSS_EXPORTS ${CROSS_EXPORTS} re2c PARENT_SCOPE )
endif()

2
tools/re2c/NO_WARRANTY Normal file
View file

@ -0,0 +1,2 @@
re2c is distributed with no warranty whatever. The author and any other
contributors take no responsibility for the consequences of its use.

159
tools/re2c/README Normal file
View file

@ -0,0 +1,159 @@
re2c
--------------------------------------------------------------------------------
DESCRIPTION
--------------------------------------------------------------------------------
re2c is a tool for generating C-based recognizers from regular expressions.
re2c-based scanners are efficient: for programming languages, given similar
specifications, a re2c-based scanner is typically almost twice as fast as a
flex-based scanner with little or no increase in size (possibly a decrease
on cisc architectures). Indeed, re2c-based scanners are quite competitive with
hand-crafted ones.
Unlike flex, re2c does not generate complete scanners: the user must supply some
interface code. While this code is not bulky (about 50-100 lines for a
flex-like scanner; see the man page and examples in the distribution) careful
coding is required for efficiency (and correctness). One advantage of this
arrangement is that the generated code is not tied to any particular input
model.
--------------------------------------------------------------------------------
DOWNLOAD
--------------------------------------------------------------------------------
The re2c distribution can be found at:
https://sourceforge.net/projects/re2c/
Download the latest tarball:
https://sourceforge.net/projects/re2c/files/latest/download
Clone git repo:
git clone git://git.code.sf.net/p/re2c/code-git
--------------------------------------------------------------------------------
BUILD
--------------------------------------------------------------------------------
Contents:
1. simple build
2. bootstrap
3. out-of-source build
4. testing
5. rebuild documentation
6. build for windows with mingw
7. build from git
1. Simplest possible build:
$ ./configure [--prefix=<prefix>]
$ make
$ make install
This will build re2c and install it (binary and man page) to <prefix> (defaults
to /usr/local).
2. Bootstrap and rebuild:
$ ./configure [--prefix=<prefix>]
$ make bootstrap
$ make install
Usual bootstrap procedure: re2c uses re2c to compile its lexer.
1. build lexer (if make finds re2c binary in build directory, it will build lexer
from source, otherwize it will use prebuilt lexer)
2. build re2c
3. build lexer from source using re2c binary in build directory
4. rebuild re2c
3. Out-of-source build:
$ mkdir <build-directory>
$ cd <build-directory>
$ <path-to-configure>/configure [--prefix=<prefix>]
$ make
$ make install
4. Testing:
$ make check
This will redirect test script output to file. If you want to see progress:
$ make tests
Testing under valgrind (takes a long time):
$ make vtests
5. Rebuild documentation (requires rst2man.py):
$ ./configure --enable-docs [--prefix=<prefix>]
$ make docs
$ make install
6. Build for windows using mingw:
$ ../configure --host i686-w64-mingw32 [--prefix=<prefix>]
$ make
This will result into an executable re2c.exe, which can be tested with wine:
$ make wtests
7. If you want to build from git, you'll first need to generate autotools files:
$ ./autogen.sh
--------------------------------------------------------------------------------
INFO
--------------------------------------------------------------------------------
$ man re2c
re2c home page:
re2c.org
re2c manual:
re2c.org/manual.html
Ulya Trofimovich's blog on re2c:
skvadrik.github.io/aleph_null/re2c.html
Original paper on re2c: "RE2C: a More Versatile Parser Generator" (1994, Peter
Bumbulis and Donald D. Cowan).
Examples can be found in 'examples' directory.
--------------------------------------------------------------------------------
MAILING LISTS
--------------------------------------------------------------------------------
re2c-general:
re2c-general@lists.sourceforge.net
re2c-devel:
re2c-devel@lists.sourceforge.net
You are welcome to ask for help or share your thoughts and ideas about re2c :)
--------------------------------------------------------------------------------
BUGS
--------------------------------------------------------------------------------
Please report any bugs and send feature requests to:
https://sourceforge.net/p/re2c/_list/tickets
--------------------------------------------------------------------------------
AUTHORS
--------------------------------------------------------------------------------
Originally written by Peter Bumbulis (peter@csg.uwaterloo.ca)
Currently maintained by:
Ulya Trofimovich <skvadrik@gmail.com>
Dan Nuffer <nuffer@users.sourceforge.net>
Marcus Boerger <helly@users.sourceforge.net>
Hartmut Kaiser <hkaiser@users.sourceforge.net>
--------------------------------------------------------------------------------
LICENSE
--------------------------------------------------------------------------------
re2c is distributed with no warranty whatever. The code is certain to contain
errors. Neither the author nor any contributor takes responsibility for any
consequences of its use.
re2c is in the public domain. The data structures and algorithms used in re2c
are all either taken from documents available to the general public or are
inventions of the authors. Programs generated by re2c may be distributed freely.
re2c itself may be distributed freely, in source or binary, unchanged or
modified. Distributors may charge whatever fees they can obtain for re2c.
If you do make use of re2c, or incorporate it into a larger project an
acknowledgement somewhere (documentation, research report, etc.) would be
appreciated.
--------------------------------------------------------------------------------

58
tools/re2c/config.h.in Normal file
View file

@ -0,0 +1,58 @@
/* config.h.in. Generated from configure.ac by autoheader. */
/* Define to 1 if you have the <stdint.h> header file. */
#cmakedefine HAVE_STDINT_H @HAVE_STDINT_H@
/* Name of package */
#cmakedefine PACKAGE "@PACKAGE_NAME@"
/* Define to the address where bug reports for this package should be sent. */
#cmakedefine PACKAGE_BUGREPORT "@PACKAGE_BUGREPORT@"
/* Define to the full name of this package. */
#cmakedefine PACKAGE_NAME "@PACKAGE_NAME@"
/* Define to the full name and version of this package. */
#cmakedefine PACKAGE_STRING "@PACKAGE_STRING@"
/* Define to the one symbol short name of this package. */
#cmakedefine PACKAGE_TARNAME "@PACKAGE_TARNAME@"
/* Define to the home page for this package. */
#cmakedefine PACKAGE_URL "@PACKAGE_URL@"
/* Define to the version of this package. */
#cmakedefine PACKAGE_VERSION "@PACKAGE_VERSION@"
/* The size of `0i8', as computed by sizeof. */
#cmakedefine SIZEOF_0I8 @SIZEOF_0I8@
/* The size of `0l', as computed by sizeof. */
#cmakedefine SIZEOF_0L @SIZEOF_0L@
/* The size of `0ll', as computed by sizeof. */
#cmakedefine SIZEOF_0LL @SIZEOF_0LL@
/* The size of `char', as computed by sizeof. */
#cmakedefine SIZEOF_CHAR @SIZEOF_CHAR@
/* The size of `int', as computed by sizeof. */
#cmakedefine SIZEOF_INT @SIZEOF_INT@
/* The size of `long', as computed by sizeof. */
#cmakedefine SIZEOF_LONG @SIZEOF_LONG@
/* The size of `long long', as computed by sizeof. */
#cmakedefine SIZEOF_LONG_LONG @SIZEOF_LONG_LONG@
/* The size of `short', as computed by sizeof. */
#cmakedefine SIZEOF_SHORT @SIZEOF_SHORT@
/* The size of `void *', as computed by sizeof. */
#cmakedefine SIZEOF_VOID_P @SIZEOF_VOID_P@
/* The size of `__int64', as computed by sizeof. */
#cmakedefine SIZEOF___INT64 @SIZEOF___INT64@
/* Version number of package */
#cmakedefine VERSION @PACKAGE_VERSION@

62
tools/re2c/config.msc.h Normal file
View file

@ -0,0 +1,62 @@
/* config.h.in. Generated from configure.ac by autoheader. */
/* Define to 1 if you have the <stdint.h> header file. */
/* #undef HAVE_STDINT_H */
/* Name of package */
/* #undef PACKAGE */
/* Define to the address where bug reports for this package should be sent. */
#define PACKAGE_BUGREPORT "re2c-general@lists.sourceforge.net"
/* Define to the full name of this package. */
#define PACKAGE_NAME "re2c"
/* Define to the full name and version of this package. */
#define PACKAGE_STRING "re2c 0.16"
/* Define to the one symbol short name of this package. */
#define PACKAGE_TARNAME "re2c"
/* Define to the home page for this package. */
/* #undef PACKAGE_URL */
/* Define to the version of this package. */
#define PACKAGE_VERSION "0.16"
/* The size of `0i8', as computed by sizeof. */
#define SIZEOF_0I8 1
/* The size of `0l', as computed by sizeof. */
#define SIZEOF_0L 4
/* The size of `0ll', as computed by sizeof. */
#define SIZEOF_0LL 8
/* The size of `char', as computed by sizeof. */
#define SIZEOF_CHAR 1
/* The size of `int', as computed by sizeof. */
#define SIZEOF_INT 4
/* The size of `long', as computed by sizeof. */
#define SIZEOF_LONG 4
/* The size of `long long', as computed by sizeof. */
#define SIZEOF_LONG_LONG 8
/* The size of `short', as computed by sizeof. */
#define SIZEOF_SHORT 2
/* The size of `void *', as computed by sizeof. */
#ifdef _M_X64
#define SIZEOF_VOID_P 8
#else
#define SIZEOF_VOID_P 4
#endif
/* The size of `__int64', as computed by sizeof. */
#define SIZEOF___INT64 8
/* Version number of package */
/* #undef VERSION */

83
tools/re2c/examples/001_upn_calculator/README Normal file
View file

@ -0,0 +1,83 @@
re2c lesson 001_upn_calculator, (c) M. Boerger 2006
This lesson gets you started with re2c. In the end you will have an easy RPN
(reverse polish notation) calculator for use at command line.
You will learn about the basic interface of re2c when scanning input strings.
How to detect the end of the input and use that to stop scanning in order to
avoid problems.
Once you have successfully installed re2c you can use it to generate *.c files
from the *.re files presented in this lesson. Actually the expected *.c files
are already present. So you should name them *.cc or something alike or just
give them a different name like test.c. To do so you simply change into the
directory and execute the following command:
re2c calc_001.re > test.c
Then use your compiler to compile that code and run it. If you are using gcc
you simply do the following:
gcc -o test.o test.c
./test.o <input_file_name>
If you are using windows you might want to read till the end of this lesson.
When you want to debug the code it helps to make re2c generate working #line
information. To do so you simply specify the output file using the -o switch
followed by the output filename:
re2c -o test.c calc_001.re
The input files *.re each contain basic step by comments that explain what is
going on and what you can see in the examples.
In order to optimize the generated code we will use the -s command line switch
of re2c. This tells re2c to generate code that uses if statements rather
then endless switch/case expressions where appropriate. Note that the file name
extension is actually '.s.re' to tell the test system to use the -s switch. To
invoke re2 you do the following:
re2c -s -o test.c calc_006.s.re
Finally we use the -b switch to have the code use a decision table. The -b
switch also contains the -s behavior.
re2c -b -o test.c calc_007.b.re
-------------------------------------------------------------------------------
For windows users Lynn Allan provided some additional stuff to get you started
in the Microsoft world. This addon resides in the windows subdirectory and
gives you something to expereiment with. The code in that directory is based
on the first step and has the following changes:
* vc6 .dsp/.dsw and vc7/vc8 .sln/.vcproj project files that have "Custom Build
Steps" that can tell when main.re changes, and know how to generate main.c
from main.re. They assume that you unpacked the zip package and have re2c
itself build or installed in Release and Release-2005 directory respectively.
If re2c cannot be found you need to modify the custom build step and correct
the path to re2c.
* BuildAndRun.bat to do command line rec2 and then cl and then run the
executable (discontinues with message if errors).
* built-in cppunit-like test to confirm it worked as expected.
* array of test strings "fed" to scan rather than file contents to facilitate
testing and also reduce the newbie learning curve.
* HiResTimer output for 10,000 loops and 100,000 loops. While this might be
excessive for this lesson, it illustrates how to do it for subsequent lessons
and your own stuff using windows. Also it shows that Release build is as fast
as strncmp for this test and can probably be made significantly faster.
* If you want to build the other steps of this lesson using windows tools
simply copy the *.re files into the windows directory as main.re and rebuild.
-------------------------------------------------------------------------------
Sidenote: UPN is the german translation of RPN, somehow hardcoded into the
authors brain :-)

84
tools/re2c/examples/001_upn_calculator/calc_001.re Normal file
View file

@ -0,0 +1,84 @@
/* re2c lesson 001_upn_calculator, calc_001, (c) M. Boerger 2006 - 2007 */
/*!ignore:re2c
- basic interface for string reading
. We define the macros YYCTYPE, YYCURSOR, YYLIMIT, YYMARKER, YYFILL
. YYCTYPE is the type re2c operates on or in other words the type that
it generates code for. While it is not a big difference when we were
using 'unsigned char' here we would need to run re2c with option -w
to fully support types with sieof() > 1.
. YYCURSOR is used internally and holds the current scanner position. In
expression handlers, the code blocks after re2c expressions, this can be
used to identify the end of the token.
. YYMARKER is not always being used so we set an initial value to avoid
a compiler warning. Here we could also omit it compleley.
. YYLIMIT stores the end of the input. Unfortunatley we have to use strlen()
in this lesson. In the next example we see one way to get rid of it.
. We use a 'for(;;)'-loop around the scanner block. We could have used a
'while(1)'-loop instead but some compilers generate a warning for it.
. To make the output more readable we use 're2c:indent:top' scanner
configuration that configures re2c to prepend a single tab (the default)
to the beginning of each output line.
. The following lines are expressions and for each expression we output the
token name and continue the scanner loop.
. The second last token detects the end of our input, the terminating zero in
our input string. In other scanners detecting the end of input may vary.
For example binary code may contain \0 as valid input.
. The last expression accepts any input character. It tells re2c to accept
the opposit of the empty range. This includes numbers and our tokens but
as re2c goes from top to botton when evaluating the expressions this is no
problem.
. The first three rules show that re2c actually prioritizes the expressions
from top to bottom. Octal number require a starting "0" and the actual
number. Normal numbers start with a digit greater 0. And zero is finally a
special case. A single "0" is detected by the last rule of this set. And
valid ocal number is already being detected by the first rule. This even
includes multi "0" sequences that in octal notation also means zero.
Another way would be to only use two rules:
"0" [0-9]+
"0" | ( [1-9] [0-9]* )
A full description of re2c rule syntax can be found in the manual.
*/
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
int scan(char *s, int l)
{
char *p = s;
char *q = 0;
#define YYCTYPE char
#define YYCURSOR p
#define YYLIMIT (s+l)
#define YYMARKER q
#define YYFILL(n)
for(;;)
{
/*!re2c
re2c:indent:top = 2;
"0"[0-9]+ { printf("Oct\n"); continue; }
[1-9][0-9]* { printf("Num\n"); continue; }
"0" { printf("Num\n"); continue; }
"+" { printf("+\n"); continue; }
"-" { printf("-\n"); continue; }
"\000" { printf("EOF\n"); return 0; }
[^] { printf("ERR\n"); return 1; }
*/
}
}
int main(int argc, char **argv)
{
if (argc > 1)
{
return scan(argv[1], strlen(argv[1]));
}
else
{
fprintf(stderr, "%s <expr>\n", argv[0]);
return 1;
}
}

69
tools/re2c/examples/001_upn_calculator/calc_002.re Normal file
View file

@ -0,0 +1,69 @@
/* re2c lesson 001_upn_calculator, calc_002, (c) M. Boerger 2006 - 2007 */
/*!ignore:re2c
- making use of YYFILL
. Here we modified the scanner to not require strlen() on the call. Instead
we compute limit on the fly. That is whenever more input is needed we
search for the terminating \0 in the next n chars the scanner needs.
. If there is not enough input we quit the scanner.
. Note that in lesson_001 YYLIMIT was a character pointer computed only once.
Here is of course also of type YYCTYPE but a variable that gets reevaluated
by YYFILL().
. To make the code smaller we take advantage of the fact that our loop has no
break so far. This allows us to use break here and have the code that is
used for YYFILL() not contain the printf in every occurence. That way the
generated code gets smaller.
*/
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
int fill(char *p, int n, char **l)
{
while (*++p && n--) ;
* l = p;
return n <= 0;
}
int scan(char *s)
{
char *p = s;
char *l = s;
char *q = 0;
#define YYCTYPE char
#define YYCURSOR p
#define YYLIMIT l
#define YYMARKER q
#define YYFILL(n) { if (!fill(p, n, &l)) break; }
for(;;)
{
/*!re2c
re2c:indent:top = 2;
"0"[0-9]+ { printf("Oct\n"); continue; }
[1-9][0-9]* { printf("Num\n"); continue; }
"0" { printf("Num\n"); continue; }
"+" { printf("+\n"); continue; }
"-" { printf("+\n"); continue; }
"\000" { printf("EOF\n"); return 0; }
[^] { printf("ERR\n"); return 1; }
*/
}
printf("OOD\n"); return 2;
}
int main(int argc, char **argv)
{
if (argc > 1)
{
return scan(argv[1]);
}
else
{
fprintf(stderr, "%s <expr>\n", argv[0]);
return 0;
}
}

61
tools/re2c/examples/001_upn_calculator/calc_003.re Normal file
View file

@ -0,0 +1,61 @@
/* re2c lesson 001_upn_calculator, calc_003, (c) M. Boerger 2006 - 2007 */
/*!ignore:re2c
- making use of YYFILL
. Again provide the length of the input to generate the limit only once. Now
we can use YYFILL() to detect the end and simply return since YYFILL() is
only being used if the next scanner run might use more chars then YYLIMIT
allows.
. Note that we now use (s+l+2) instead of (s+l) as we did in lesson_001. In
the first lesson we did not quit from YYFILL() and used a special rule to
detect the end of input. Here we use the fact that we know the exact end
of input and that this length does not include the terminating zero. Since
YYLIMIT points to the first character behind the used buffer we use "+ 2".
If we would use "+1" we could drop the "\000" rule but could no longer
distinguish between end of input and out of data.
*/
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
int scan(char *s, int l)
{
char *p = s;
char *q = 0;
#define YYCTYPE char
#define YYCURSOR p
#define YYLIMIT (s+l+2)
#define YYMARKER q
#define YYFILL(n) { printf("OOD\n"); return 2; }
for(;;)
{
/*!re2c
re2c:indent:top = 2;
"0"[0-9]+ { printf("Oct\n"); continue; }
[1-9][0-9]* { printf("Num\n"); continue; }
"0" { printf("Num\n"); continue; }
"+" { printf("+\n"); continue; }
"-" { printf("+\n"); continue; }
"\000" { printf("EOF\n"); return 0; }
[^] { printf("ERR\n"); return 1; }
*/
}
return 0;
}
int main(int argc, char **argv)
{
if (argc > 1)
{
return scan(argv[1], strlen(argv[1]));
}
else
{
fprintf(stderr, "%s <expr>\n", argv[0]);
return 0;
}
}

78
tools/re2c/examples/001_upn_calculator/calc_004.re Normal file
View file

@ -0,0 +1,78 @@
/* re2c lesson 001_upn_calculator, calc_004, (c) M. Boerger 2006 - 2007 */
/*!ignore:re2c
- making use of definitions
. We provide complex rules as definitions. We can even have definitions made
up from other definitions. And we could also use definitions as part of
rules and not only as full rules as shown in this lesson.
- showing the tokens
. re2c does not store the beginning of a token on its own but we can easily
do this by providing variable, in our case t, that is set to YYCURSOR on
every loop. If we were not using a loop here the token, we could have used
s instead of a new variable instead.
. As we use the token for an output function that requires a terminating zero
we copy the token. Alternatively we could store the end of the token, then
replace it with a zero character and replace it after the token has been
used. However that approach is not always acceptable.
*/
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
char * tokendup(const char *t, const char *l)
{
size_t n = l -t + 1;
char *r = (char*)malloc(n);
memmove(r, t, n-1);
r[n] = '\0';
return r;
}
int scan(char *s, int l)
{
char *p = s;
char *q = 0;
char *t;
#define YYCTYPE char
#define YYCURSOR p
#define YYLIMIT (s+l+2)
#define YYMARKER q
#define YYFILL(n) { printf("OOD\n"); return 2; }
for(;;)
{
t = p;
/*!re2c
re2c:indent:top = 2;
DIGIT = [0-9] ;
OCT = "0" DIGIT+ ;
INT = "0" | ( [1-9] DIGIT* ) ;
OCT { t = tokendup(t, p); printf("Oct: %s\n", t); free(t); continue; }
INT { t = tokendup(t, p); printf("Num: %s\n", t); free(t); continue; }
"+" { printf("+\n"); continue; }
"-" { printf("+\n"); continue; }
"\000" { printf("EOF\n"); return 0; }
[^] { printf("ERR\n"); return 1; }
*/
}
return 0;
}
int main(int argc, char **argv)
{
if (argc > 1)
{
return scan(argv[1], strlen(argv[1]));
}
else
{
fprintf(stderr, "%s <expr>\n", argv[0]);
return 0;
}
}

144
tools/re2c/examples/001_upn_calculator/calc_005.re Normal file
View file

@ -0,0 +1,144 @@
/* re2c lesson 001_upn_calculator, calc_005, (c) M. Boerger 2006 - 2007 */
/*!ignore:re2c
- turning this lesson into an easy calculator
. We are going to write an UPN calculator so we need an additional rule to
ignore white space.
. Then we need to store the scanned input somewhere and do our math on it.
. Also we need to scan all arguments since the main c code gets the input
split up into chunks.
. In contrast to what we did before we now add a variable res that holds the
scanner state. We initialize that variable to 0 and quit the loop when it
is non zero. This will also be our return value so that we can use it in
function main to generate error information.
. To support operating systems where ' and " get passed in program arguments
we check for them being first and last input character. If so we correct
input pointer and input length. Since now our scanner might not see a
terminating zero we change YYLIMIT again and drop the special zero rule.
*/
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#define DEBUG(stmt) stmt
int stack[4];
int depth = 0;
int push_num(const char *t, const char *l, int radix)
{
int num = 0;
if (depth >= sizeof(stack))
{
return 3;
}
--t;
while(++t < l)
{
num = num * radix + (*t - '0');
}
DEBUG(printf("Num: %d\n", num));
stack[depth++] = num;
return 0;
}
int stack_add()
{
if (depth < 2) return 4;
--depth;
stack[depth-1] = stack[depth-1] + stack[depth];
return 0;
}
int stack_sub()
{
if (depth < 2) return 4;
--depth;
stack[depth-1] = stack[depth-1] - stack[depth];
return 0;
}
int scan(char *s, int l)
{
char *p = s;
char *q = 0;
char *t;
int res = 0;
#define YYCTYPE char
#define YYCURSOR p
#define YYLIMIT (s+l+1)
#define YYMARKER q
#define YYFILL(n) { return depth == 1 ? 0 : 2; }
while(!res)
{
t = p;
/*!re2c
re2c:indent:top = 2;
DIGIT = [0-9] ;
OCT = "0" DIGIT+ ;
INT = "0" | ( [1-9] DIGIT* ) ;
WS = [ \t]+ ;
WS { continue; }
OCT { res = push_num(t, p, 8); continue; }
INT { res = push_num(t, p, 10); continue; }
"+" { res = stack_add(); continue; }
"-" { res = stack_sub(); continue; }
[^] { res = 1; continue; }
*/
}
return res;
}
int main(int argc, char **argv)
{
if (argc > 1)
{
char *inp;
int res = 0, argp = 0, len;
while(!res && ++argp < argc)
{
inp = argv[argp];
len = strlen(inp);
if (inp[0] == '\"' && inp[len-1] == '\"')
{
++inp;
len -=2;
}
res = scan(inp, len);
}
switch(res)
{
case 0:
printf("Result: %d\n", stack[0]);
return 0;
case 1:
fprintf(stderr, "Illegal character in input.\n");
return 1;
case 2:
fprintf(stderr, "Premature end of input.\n");
return 2;
case 3:
fprintf(stderr, "Stack overflow.\n");
return 3;
case 4:
fprintf(stderr, "Stack underflow.\n");
return 4;
}
}
else
{
fprintf(stderr, "%s <expr>\n", argv[0]);
return 0;
}
}

162
tools/re2c/examples/001_upn_calculator/calc_006.s.re Normal file
View file

@ -0,0 +1,162 @@
/* re2c lesson 001_upn_calculator, calc_006, (c) M. Boerger 2006 - 2007 */
/*!ignore:re2c
- avoiding YYFILL()
. We use the inplace configuration re2c:yyfill to suppress generation of
YYFILL() blocks. This of course means we no longer have to provide the
macro.
. We also drop the YYMARKER stuff since we know that re2c does not generate
it for this example.
. Since re2c does no longer check for out of data situations we must do this.
For that reason we first reintroduce our zero rule and second we need to
ensure that the scanner does not take more than one bytes in one go.
In the example suppose "0" is passed. The scanner reads the first "0" and
then is in an undecided state. The scanner can earliest decide on the next
char what the token is. In case of a zero the input ends and it was a
number, 0 to be precise. In case of a digit it is an octal number and the
next character needs to be read. In case of any other character the scanner
will detect an error with the any rule [^].
Now the above shows that the scanner may read two characters directly. But
only if the first is a "0". So we could easily check that if the first char
is "0" and the next char is a digit then yet another charcter is present.
But we require our inut to be zero terminated. And that means we do not
have to check anything for this scanner.
However with other rule sets re2c might read more then one character in a
row. In those cases it is normally hard to impossible to avoid YYFILL.
- optimizing the generated code by using -s command line switch of re2c
. This tells re2c to generate code that uses if statements rather
then endless switch/case expressions where appropriate. Note that the
generated code now requires the input to be unsigned char rather than char
due to the way comparisons are generated.
*/
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#define DEBUG(stmt) stmt
int stack[4];
int depth = 0;
int push_num(const unsigned char *t, const unsigned char *l, int radix)
{
int num = 0;
if (depth >= sizeof(stack))
{
return 3;
}
--t;
while(++t < l)
{
num = num * radix + (*t - (unsigned char)'0');
}
DEBUG(printf("Num: %d\n", num));
stack[depth++] = num;
return 0;
}
int stack_add()
{
if (depth < 2) return 4;
--depth;
stack[depth-1] = stack[depth-1] + stack[depth];
DEBUG(printf("+\n"));
return 0;
}
int stack_sub()
{
if (depth < 2) return 4;
--depth;
stack[depth-1] = stack[depth-1] - stack[depth];
DEBUG(printf("-\n"));
return 0;
}
int scan(char *s)
{
unsigned char *p = (unsigned char*)s;
unsigned char *t;
int res = 0;
#define YYCTYPE unsigned char
#define YYCURSOR p
while(!res)
{
t = p;
/*!re2c
re2c:indent:top = 2;
re2c:yyfill:enable = 0;
DIGIT = [0-9] ;
OCT = "0" DIGIT+ ;
INT = "0" | ( [1-9] DIGIT* ) ;
WS = [ \t]+ ;
WS { continue; }
OCT { res = push_num(t, p, 8); continue; }
INT { res = push_num(t, p, 10); continue; }
"+" { res = stack_add(); continue; }
"-" { res = stack_sub(); continue; }
"\000" { res = depth == 1 ? 0 : 2; break; }
[^] { res = 1; continue; }
*/
}
return res;
}
int main(int argc, char **argv)
{
if (argc > 1)
{
char *inp;
int res = 0, argp = 0, len;
while(!res && ++argp < argc)
{
inp = strdup(argv[argp]);
len = strlen(inp);
if (inp[0] == '\"' && inp[len-1] == '\"')
{
inp[len - 1] = '\0';
++inp;
}
res = scan(inp);
free(inp);
}
switch(res)
{
case 0:
printf("Result: %d\n", stack[0]);
return 0;
case 1:
fprintf(stderr, "Illegal character in input.\n");
return 1;
case 2:
fprintf(stderr, "Premature end of input.\n");
return 2;
case 3:
fprintf(stderr, "Stack overflow.\n");
return 3;
case 4:
fprintf(stderr, "Stack underflow.\n");
return 4;
}
}
else
{
fprintf(stderr, "%s <expr>\n", argv[0]);
return 0;
}
}

135
tools/re2c/examples/001_upn_calculator/calc_007.b.re Normal file
View file

@ -0,0 +1,135 @@
/* re2c lesson 001_upn_calculator, calc_007, (c) M. Boerger 2006 - 2007 */
/*!ignore:re2c
- optimizing the generated code by using -b command line switch of re2c
. This tells re2c to generate code that uses a decision table. The -b switch
also contains the -s behavior. And -b also requires the input to be
unsigned chars.
*/
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#define DEBUG(stmt) stmt
int stack[4];
int depth = 0;
int push_num(const unsigned char *t, const unsigned char *l, int radix)
{
int num = 0;
if (depth >= sizeof(stack))
{
return 3;
}
--t;
while(++t < l)
{
num = num * radix + (*t - (unsigned char)'0');
}
DEBUG(printf("Num: %d\n", num));
stack[depth++] = num;
return 0;
}
int stack_add()
{
if (depth < 2) return 4;
--depth;
stack[depth-1] = stack[depth-1] + stack[depth];
DEBUG(printf("+\n"));
return 0;
}
int stack_sub()
{
if (depth < 2) return 4;
--depth;
stack[depth-1] = stack[depth-1] - stack[depth];
DEBUG(printf("+\n"));
return 0;
}
int scan(char *s)
{
unsigned char *p = (unsigned char*)s;
unsigned char *t;
int res = 0;
#define YYCTYPE unsigned char
#define YYCURSOR p
while(!res)
{
t = p;
/*!re2c
re2c:indent:top = 2;
re2c:yyfill:enable = 0;
DIGIT = [0-9] ;
OCT = "0" DIGIT+ ;
INT = "0" | ( [1-9] DIGIT* ) ;
WS = [ \t]+ ;
WS { continue; }
OCT { res = push_num(t, p, 8); continue; }
INT { res = push_num(t, p, 10); continue; }
"+" { res = stack_add(); continue; }
"-" { res = stack_sub(); continue; }
"\000" { res = depth == 1 ? 0 : 2; break; }
[^] { res = 1; continue; }
*/
}
return res;
}
int main(int argc, char **argv)
{
if (argc > 1)
{
char *inp;
int res = 0, argp = 0, len;
while(!res && ++argp < argc)
{
inp = strdup(argv[argp]);
len = strlen(inp);
if (inp[0] == '\"' && inp[len-1] == '\"')
{
inp[len - 1] = '\0';
++inp;
}
res = scan(inp);
free(inp);
}
switch(res)
{
case 0:
printf("Result: %d\n", stack[0]);
return 0;
case 1:
fprintf(stderr, "Illegal character in input.\n");
return 1;
case 2:
fprintf(stderr, "Premature end of input.\n");
return 2;
case 3:
fprintf(stderr, "Stack overflow.\n");
return 3;
case 4:
fprintf(stderr, "Stack underflow.\n");
return 4;
}
}
else
{
fprintf(stderr, "%s <expr>\n", argv[0]);
return 0;
}
}

158
tools/re2c/examples/001_upn_calculator/calc_008.b.re Normal file
View file

@ -0,0 +1,158 @@
/* re2c lesson 001_upn_calculator, calc_008, (c) M. Boerger 2006 - 2007 */
/*!ignore:re2c
- using -b with signed character input
. Since the code is being generated with -b switch re2c requires the internal
character variable yych to use an unsigned character type. For that reason
the previous lessons had a conversion at the beginning of their scan()
function. Other re2c generated code often have the scanners work completely
on unsigned input. Thus requesting a conversion.
To avoid the conversion on input, re2c allows to do the conversion when
reading the internal yych variable. To enable that conversion you need to
use the implace configuration 're2c:yych:conversion' and set it to 1. This
will change the generated code to insert conversions to YYCTYPE whenever
yych is being read.
- More inplace configurations for better/nicer code
. re2c allows to overwrite the generation of any define, label or variable
used in the generated code. For example we overwrite the 'yych' variable
name to 'curr' using inplace configuration 're2c:variable:yych = curr;'.
. We further more use inplace configurations instead of defines. This allows
to use correct conversions to 'unsigned char' instead of having to convert
to 'YYCTYPE' when placing 're2c:define:YYCTYPE = "unsigned char";' infront
of 're2c:yych:conversion'. Note that we have to use apostrophies for the
first setting as it contains a space.
. Last but not least we use 're2c:labelprefix = scan' to change the prefix
of generated labels.
*/
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#define DEBUG(stmt) stmt
int stack[4];
int depth = 0;
int push_num(const char *t, const char *l, int radix)
{
int num = 0;
if (depth >= sizeof(stack))
{
return 3;
}
--t;
while(++t < l)
{
num = num * radix + (*t - '0');
}
DEBUG(printf("Num: %d\n", num));
stack[depth++] = num;
return 0;
}
int stack_add()
{
if (depth < 2) return 4;
--depth;
stack[depth-1] = stack[depth-1] + stack[depth];
DEBUG(printf("+\n"));
return 0;
}
int stack_sub()
{
if (depth < 2) return 4;
--depth;
stack[depth-1] = stack[depth-1] - stack[depth];
DEBUG(printf("+\n"));
return 0;
}
int scan(char *p)
{
char *t;
int res = 0;
while(!res)
{
t = p;
/*!re2c
re2c:define:YYCTYPE = "unsigned char";
re2c:define:YYCURSOR = p;
re2c:variable:yych = curr;
re2c:indent:top = 2;
re2c:yyfill:enable = 0;
re2c:yych:conversion = 1;
re2c:labelprefix = scan;
DIGIT = [0-9] ;
OCT = "0" DIGIT+ ;
INT = "0" | ( [1-9] DIGIT* ) ;
WS = [ \t]+ ;
WS { continue; }
OCT { res = push_num(t, p, 8); continue; }
INT { res = push_num(t, p, 10); continue; }
"+" { res = stack_add(); continue; }
"-" { res = stack_sub(); continue; }
"\000" { res = depth == 1 ? 0 : 2; break; }
[^] { res = 1; continue; }
*/
}
return res;
}
int main(int argc, char **argv)
{
if (argc > 1)
{
char *inp;
int res = 0, argp = 0, len;
while(!res && ++argp < argc)
{
inp = strdup(argv[argp]);
len = strlen(inp);
if (inp[0] == '\"' && inp[len-1] == '\"')
{
inp[len - 1] = '\0';
++inp;
}
res = scan(inp);
free(inp);
}
switch(res)
{
case 0:
printf("Result: %d\n", stack[0]);
return 0;
case 1:
fprintf(stderr, "Illegal character in input.\n");
return 1;
case 2:
fprintf(stderr, "Premature end of input.\n");
return 2;
case 3:
fprintf(stderr, "Stack overflow.\n");
return 3;
case 4:
fprintf(stderr, "Stack underflow.\n");
return 4;
}
}
else
{
fprintf(stderr, "%s <expr>\n", argv[0]);
return 0;
}
}

54
tools/re2c/examples/001_upn_calculator/windows/HiResTimer.h Normal file
View file

@ -0,0 +1,54 @@
/**
* @file HiResTimer.h
* @brief
* @note
*/
#ifndef _HI_RES_TIMER_H_
#define _HI_RES_TIMER_H_
#ifdef WIN32
#include <windows.h> // probably already done in stdafx.h
static LARGE_INTEGER start;
static LARGE_INTEGER stop;
static LARGE_INTEGER freq;
static _int64 elapsedCounts;
static double elapsedMillis;
static double elapsedMicros;
static HANDLE processHandle;
static DWORD prevPriorityClass;
void HrtInit()
{
processHandle = GetCurrentProcess();
prevPriorityClass = GetPriorityClass(processHandle);
QueryPerformanceFrequency(&freq);
}
void HrtStart()
{
QueryPerformanceCounter(&start);
}
void HrtSetPriority(DWORD priority)
{
int flag;
prevPriorityClass = GetPriorityClass(processHandle);
flag = SetPriorityClass(processHandle, priority);
}
void HrtResetPriority(void)
{
int flag = SetPriorityClass(processHandle, prevPriorityClass);
}
double HrtElapsedMillis()
{
QueryPerformanceCounter(&stop);
elapsedCounts = (stop.QuadPart - start.QuadPart);
elapsedMillis = ((elapsedCounts * 1000.0) / freq.QuadPart);
return elapsedMillis;
}
#endif
#endif

291
tools/re2c/examples/001_upn_calculator/windows/main.b.re Normal file
View file

@ -0,0 +1,291 @@
/* re2c lesson 001_upn_calculator, main.b.re, (c) M. Boerger, L. Allan 2006 */
/*!ignore:re2c
- basic interface for string reading
. We define the macros YYCTYPE, YYCURSOR, YYLIMIT, YYMARKER, YYFILL
. YYCTYPE is the type re2c operates on or in other words the type that
it generates code for. While it is not a big difference when we were
using 'unsigned char' here we would need to run re2c with option -w
to fully support types with sieof() > 1.
. YYCURSOR is used internally and holds the current scanner position. In
expression handlers, the code blocks after re2c expressions, this can be
used to identify the end of the token.
. YYMARKER is not always being used so we set an initial value to avoid
a compiler warning.
. YYLIMIT stores the end of the input. Unfortunatley we have to use strlen()
in this lesson. In the next example we see one way to get rid of it.
. We use a 'for(;;)'-loop around the scanner block. We could have used a
'while(1)'-loop instead but some compilers generate a warning for it.
. To make the output more readable we use 're2c:indent:top' scanner
configuration that configures re2c to prepend a single tab (the default)
to the beginning of each output line.
. The following lines are expressions and for each expression we output the
token name and continue the scanner loop.
. The second last token detects the end of our input, the terminating zero in
our input string. In other scanners detecting the end of input may vary.
For example binary code may contain \0 as valid input.
. The last expression accepts any input character. It tells re2c to accept
the opposit of the empty range. This includes numbers and our tokens but
as re2c goes from top to botton when evaluating the expressions this is no
problem.
. The first three rules show that re2c actually prioritizes the expressions
from top to bottom. Octal number require a starting "0" and the actual
number. Normal numbers start with a digit greater 0. And zero is finally a
special case. A single "0" is detected by the last rule of this set. And
valid ocal number is already being detected by the first rule. This even
includes multi "0" sequences that in octal notation also means zero.
Another way would be to only use two rules:
"0" [0-9]+
"0" | ( [1-9] [0-9]* )
A full description of re2c rule syntax can be found in the manual.
*/
#define VC_EXTRALEAN // Exclude rarely-used stuff from Windows headers
#if _MSC_VER > 1200
#define WINVER 0x0400 // Change this to the appropriate value to target Windows 98 and Windows 2000 or later.
#endif // Prevents warning from vc7.1 complaining about redefinition
#include <stdio.h>
#include <stdlib.h>
#include <assert.h>
#include <string.h>
#include <windows.h>
#include "HiResTimer.h"
static char gTestBuf[1000] = "";
/**
* @brief Setup HiResolution timer and confirm it is working ok
*/
void InitHiResTimerAndVerifyWorking(void)
{
double elapsed;
HrtInit();
HrtSetPriority(ABOVE_NORMAL_PRIORITY_CLASS);
HrtStart();
Sleep(100);
elapsed = HrtElapsedMillis();
if ((elapsed < 90) || (elapsed > 110)) {
printf("HiResTimer misbehaving: %f\n", elapsed);
exit(2);
}
}
/**
* @brief Scan for numbers in different formats
*/
int ScanFullSpeed(char *pzStrToScan, size_t lenStrToScan)
{
unsigned char *pzCurScanPos = (unsigned char*)pzStrToScan;
unsigned char *pzBacktrackInfo = 0;
#define YYCTYPE unsigned char
#define YYCURSOR pzCurScanPos
#define YYLIMIT (pzStrToScan+lenStrToScan)
#define YYMARKER pzBacktrackInfo
#define YYFILL(n)
for(;;)
{
/*!re2c
re2c:indent:top = 2;
[1-9][0-9]* { continue; }
[0][0-9]+ { continue; }
"+" { continue; }
"-" { continue; }
"\000" { return 0; }
[^] { return 1; }
*/
}
}
/**
* @brief Scan for numbers in different formats
*/
int scan(char *pzStrToScan, size_t lenStrToScan)
{
unsigned char *pzCurScanPos = (unsigned char*)pzStrToScan;
unsigned char *pzBacktrackInfo = 0;
#define YYCTYPE unsigned char
#define YYCURSOR pzCurScanPos
#define YYLIMIT (pzStrToScan+lenStrToScan)
#define YYMARKER pzBacktrackInfo
#define YYFILL(n)
for(;;)
{
/*!re2c
re2c:indent:top = 2;
[1-9][0-9]* { printf("Num\n"); strcat(gTestBuf, "Num "); continue; }
[0][0-9]+ { printf("Oct\n"); strcat(gTestBuf, "Oct "); continue; }
"+" { printf("+\n"); strcat(gTestBuf, "+ "); continue; }
"-" { printf("-\n"); strcat(gTestBuf, "- "); continue; }
"\000" { printf("EOF\n"); return 0; }
[^] { printf("ERR\n"); strcat(gTestBuf, "ERR "); return 1; }
*/
}
}
/**
* @brief Show high resolution elapsed time for 10,000 and 100,000 loops
*/
void DoTimingsOfStrnCmp(void)
{
char testStr[] = "Hello, world";
int totLoops = 10000;
int totFoundCount = 0;
int foundCount = 0;
int loop;
int rc;
const int progressAnd = 0xFFFFF000;
double elapsed;
printf("\n\n%d loops with * every %d loops to confirm\n", totLoops, ((~progressAnd) + 1));
HrtStart();
for (loop = 0; loop < totLoops; ++loop) {
foundCount = 0;
rc = strncmp(testStr, "Hello", 5);
if (rc == 0) {
foundCount++;
totFoundCount++;
if ((totFoundCount & progressAnd) == totFoundCount) {
printf("*");
}
}
}
elapsed = HrtElapsedMillis();
printf("\nstrncmp Elapsed for %7d loops milliseconds: %7.3f\n", totLoops, elapsed);
printf("FoundCount each loop: %d\n", foundCount);
printf("TotalFoundCount for all loops: %d\n", totFoundCount);
totLoops = 100000;
HrtStart();
for (loop = 0; loop < totLoops; ++loop) {
foundCount = 0;
rc = strncmp(testStr, "Hello", 5);
if (rc == 0) {
foundCount++;
totFoundCount++;
if ((totFoundCount & progressAnd) == totFoundCount) {
printf("*");
}
}
}
elapsed = HrtElapsedMillis();
printf("\nstrncmp Elapsed for %7d loops milliseconds: %7.3f\n", totLoops, elapsed);
printf("FoundCount each loop: %d\n", foundCount);
printf("TotalFoundCount for all loops: %d\n", totFoundCount);
}
/**
* @brief Show high resolution elapsed time for 10,000 and 100,000 loops
*/
void DoTimingsOfRe2c(void)
{
char* testStrings[] = { "123", "1234", "+123", "01234", "-04321", "abc", "123abc" };
const int testCount = sizeof(testStrings) / sizeof(testStrings[0]);
int i;
int totLoops = 10000 / testCount; // Doing more than one per loop
int totFoundCount = 0;
int foundCount = 0;
int loop;
int rc;
const int progressAnd = 0xFFFFF000;
double elapsed;
printf("\n\n%d loops with * every %d loops to confirm\n", totLoops, ((~progressAnd) + 1));
HrtStart();
for (loop = 0; loop < totLoops; ++loop) {
foundCount = 0;
strcpy(gTestBuf, "");
for (i = 0; i < testCount; ++i) {
char* pzCurStr = testStrings[i];
size_t len = strlen(pzCurStr); // Calc of strlen slows things down ... std::string?
rc = ScanFullSpeed(pzCurStr, len);
if (rc == 0) {
foundCount++;
totFoundCount++;
if ((totFoundCount & progressAnd) == totFoundCount) {
printf("*");
}
}
}
}
elapsed = HrtElapsedMillis();
printf("\nRe2c Elapsed for %7d loops milliseconds: %7.3f\n", totLoops, elapsed);
printf("FoundCount each loop: %d\n", foundCount);
printf("TotalFoundCount for all loops: %d\n", totFoundCount);
totLoops = 100000 / testCount;
printf("\n\n%d loops with * every %d loops to confirm\n", totLoops, ((~progressAnd) + 1));
HrtStart();
for (loop = 0; loop < totLoops; ++loop) {
foundCount = 0;
strcpy(gTestBuf, "");
for (i = 0; i < testCount; ++i) {
char* pzCurStr = testStrings[i];
size_t len = strlen(pzCurStr); // Calc of strlen slows things down ... std::string?
rc = ScanFullSpeed(pzCurStr, len);
if (rc == 0) {
foundCount++;
totFoundCount++;
if ((totFoundCount & progressAnd) == totFoundCount) {
printf("*");
}
}
}
}
elapsed = HrtElapsedMillis();
printf("\nRe2c Elapsed for %7d loops milliseconds: %7.3f\n", totLoops, elapsed);
printf("FoundCount each loop: %d\n", foundCount);
printf("TotalFoundCount for all loops: %d\n", totFoundCount);
}
/**
* @brief Entry point for console app
*/
int main(int argc, char **argv)
{
char testStr_A[] = "123";
char* testStr_B = "456";
char* testStrings[] = { "123", "1234", "+123", "01234", "-04321", "abc", "123abc" };
const int testCount = sizeof(testStrings) / sizeof(testStrings[0]);
int i;
int rc = scan(testStr_A, 3);
printf("rc: %d\n", rc);
rc = scan(testStr_B, 3);
printf("rc: %d\n", rc);
rc = scan("789", 3);
printf("rc: %d\n", rc);
strcpy(gTestBuf, "");
for (i = 0; i < testCount; ++i) {
char* pzCurStr = testStrings[i];
size_t len = strlen(pzCurStr);
scan(pzCurStr, len);
}
printf("%s\n", gTestBuf);
rc = strcmp(gTestBuf, "Num Num + Num Oct - Oct ERR Num ERR ");
if (rc == 0) {
printf("Success\n");
}
else {
printf("Failure\n");
}
assert(0 == rc); // Doesn't work with Release build
InitHiResTimerAndVerifyWorking();
DoTimingsOfStrnCmp();
DoTimingsOfRe2c();
return 0;
}

21
tools/re2c/examples/002_strip_comments/README Normal file
View file

@ -0,0 +1,21 @@
re2c lesson 002_strip_comments, (c) M. Boerger 2006
In this lesson you will learn how to use multiple scanner blocks and how to
read the input from a file instead of a zero terminated string. In the end you
will have a scanner that filters comments out of c source files but keeps re2c
comments.
The first scanner can be generated with:
re2c -s -o t.c strip_001.s.re
In the second step we will learn about YYMARKER that stores backtracking
information.
re2c -s -0 t.c strip_002.b.re
The third step brings trailing contexts that are stored in YYCTXMARKER. We also
change to use -b instead of -s option since the scanner gets more and more
complex.
re2c -b -0 t.c strip_002.b.re

147
tools/re2c/examples/002_strip_comments/strip_001.s.re Normal file
View file

@ -0,0 +1,147 @@
/* re2c lesson 002_strip_comments, strip_001.s, (c) M. Boerger 2006 - 2007 */
/*!ignore:re2c
- basic interface for file reading
. This scanner will read chunks of input from a file. The easiest way would
be to read the whole file into a memory buffer and use that a zero
terminated string.
. Instead we want to read input chunks of a reasonable size as they are neede
by the scanner. Thus we basically need YYFILL(n) to call fread(n).
. Before we provide a buffer that we constantly reallocate we instead use
one buffer that we get from the stack or global memory just once. When we
reach the end of the buffer we simply move the beginning of our input
that is somewhere in our buffer to the beginning of our buffer and then
append the next chunk of input to the correct end inside our buffer.
. As re2c scanners might read more than one character we need to ensure our
buffer is long enough. We can use re2c to inform about the maximum size
by placing a "!max:re2c" comment somewhere. This gets translated to a
"#define YYMAXFILL <n>" line where <n> is the maximum length value. This
define can be used as precompiler condition.
- multiple scanner blocks
. We use a main scanner block that outputs every input character unless the
input is two /s or a / followed by a *. In the latter two cases we switch
to a special c++ comment and a comment block respectively.
. Both special blocks simply detect their end ignore any other character.
. The c++ block is a bit special. Since the terminating new line needs to
be output and that can either be a new line or a carridge return followed
by a new line.
. In order to ensure that we do not read behind our buffer we reset the token
pointer to the cursor on every scanner run.
*/
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
/*!max:re2c */
#define BSIZE 128
#if BSIZE < YYMAXFILL
# error BSIZE must be greater YYMAXFILL
#endif
#define YYCTYPE unsigned char
#define YYCURSOR s.cur
#define YYLIMIT s.lim
#define YYFILL(n) { if ((res = fill(&s, n)) >= 0) break; }
typedef struct Scanner
{
FILE *fp;
unsigned char *cur, *tok, *lim, *eof;
unsigned char buffer[BSIZE];
} Scanner;
int fill(Scanner *s, int len)
{
if (!len)
{
s->cur = s->tok = s->lim = s->buffer;
s->eof = 0;
}
if (!s->eof)
{
int got, cnt = s->tok - s->buffer;
if (cnt > 0)
{
memcpy(s->buffer, s->tok, s->lim - s->tok);
s->tok -= cnt;
s->cur -= cnt;
s->lim -= cnt;
}
cnt = BSIZE - cnt;
if ((got = fread(s->lim, 1, cnt, s->fp)) != cnt)
{
s->eof = &s->lim[got];
}
s->lim += got;
}
else if (s->cur + len > s->eof)
{
return 0; /* not enough input data */
}
return -1;
}
int scan(FILE *fp)
{
int res = 0;
Scanner s;
if (!fp)
{
return 1; /* no file was opened */
}
s.fp = fp;
fill(&s, 0);
for(;;)
{
s.tok = s.cur;
/*!re2c
re2c:indent:top = 2;
NL = "\r"? "\n" ;
ANY = [^] ;
"/" "/" { goto cppcomment; }
"/" "*" { goto comment; }
ANY { fputc(*s.tok, stdout); continue; }
*/
comment:
s.tok = s.cur;
/*!re2c
"*" "/" { continue; }
ANY { goto comment; }
*/
cppcomment:
s.tok = s.cur;
/*!re2c
NL { fwrite(s.tok, 1, s.cur - s.tok, stdout); continue; }
ANY { goto cppcomment; }
*/
}
if (fp != stdin)
{
fclose(fp); /* close only if not stdin */
}
return res; /* return result */
}
int main(int argc, char **argv)
{
if (argc > 1)
{
return scan(!strcmp(argv[1], "-") ? stdin : fopen(argv[1], "r"));
}
else
{
fprintf(stderr, "%s <expr>\n", argv[0]);
return 1;
}
}

162
tools/re2c/examples/002_strip_comments/strip_002.s.re Normal file
View file

@ -0,0 +1,162 @@
/* re2c lesson 002_strip_comments, strip_002.s, (c) M. Boerger 2006 - 2007 */
/*!ignore:re2c
- complexity
. When a comment is preceeded by a new line and followed by whitespace and a
new line then we can drop the trailing whitespace and new line.
. Additional to what we strip out already what about two consequtive comment
blocks? When two comments are only separated by whitespace we want to drop
both. In other words when detecting the end of a comment block we need to
check whether it is followed by only whitespace and the a new comment in
which case we continure ignoring the input. If it is followed only by white
space and a new line we strip out the new white space and new line. In any
other case we start outputting all that follows.
But we cannot simply use the following two rules:
"*" "/" WS* "/" "*" { continue; }
"*" "/" WS* NL { continue; }
The main problem is that WS* can get bigger then our buffer, so we need a
new scanner.
. Meanwhile our scanner gets a bit more complex and we have to add two more
things. First the scanner code now uses a YYMARKER to store backtracking
information.
- backtracking information
. When the scanner has two rules that can have the same beginning but a
different ending then it needs to store the position that identifies the
common part. This is called backtracking. As mentioned above re2c expects
you to provide compiler define YYMARKER and a pointer variable.
. When shifting buffer contents as done in our fill function the marker needs
to be corrected, too.
*/
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
/*!max:re2c */
#define BSIZE 128
#if BSIZE < YYMAXFILL
# error BSIZE must be greater YYMAXFILL
#endif
#define YYCTYPE unsigned char
#define YYCURSOR s.cur
#define YYLIMIT s.lim
#define YYMARKER s.mrk
#define YYFILL(n) { if ((res = fill(&s, n)) >= 0) break; }
typedef struct Scanner
{
FILE *fp;
unsigned char *cur, *tok, *lim, *eof, *mrk;
unsigned char buffer[BSIZE];
} Scanner;
int fill(Scanner *s, int len)
{
if (!len)
{
s->cur = s->tok = s->lim = s->mrk = s->buffer;
s->eof = 0;
}
if (!s->eof)
{
int got, cnt = s->tok - s->buffer;
if (cnt > 0)
{
memcpy(s->buffer, s->tok, s->lim - s->tok);
s->tok -= cnt;
s->cur -= cnt;
s->lim -= cnt;
s->mrk -= cnt;
}
cnt = BSIZE - cnt;
if ((got = fread(s->lim, 1, cnt, s->fp)) != cnt)
{
s->eof = &s->lim[got];
}
s->lim += got;
}
else if (s->cur + len > s->eof)
{
return 0; /* not enough input data */
}
return -1;
}
void echo(Scanner *s)
{
fwrite(s->tok, 1, s->cur - s->tok, stdout);
}
int scan(FILE *fp)
{
int res = 0;
Scanner s;
if (!fp)
{
return 1; /* no file was opened */
}
s.fp = fp;
fill(&s, 0);
for(;;)
{
s.tok = s.cur;
/*!re2c
re2c:indent:top = 2;
NL = "\r"? "\n" ;
WS = [\r\n\t ] ;
ANY = [^] ;
"/" "/" { goto cppcomment; }
"/" "*" { goto comment; }
ANY { fputc(*s.tok, stdout); continue; }
*/
comment:
s.tok = s.cur;
/*!re2c
"*" "/" { goto commentws; }
ANY { goto comment; }
*/
commentws:
s.tok = s.cur;
/*!re2c
NL { echo(&s); continue; }
WS { goto commentws; }
ANY { echo(&s); continue; }
*/
cppcomment:
s.tok = s.cur;
/*!re2c
NL { echo(&s); continue; }
ANY { goto cppcomment; }
*/
}
if (fp != stdin)
{
fclose(fp); /* close only if not stdin */
}
return res; /* return result */
}
int main(int argc, char **argv)
{
if (argc > 1)
{
return scan(!strcmp(argv[1], "-") ? stdin : fopen(argv[1], "r"));
}
else
{
fprintf(stderr, "%s <expr>\n", argv[0]);
return 1;
}
}

179
tools/re2c/examples/002_strip_comments/strip_003.b.re Normal file
View file

@ -0,0 +1,179 @@
/* re2c lesson 002_strip_comments, strip_003.b, (c) M. Boerger 2006 - 2007 */
/*!ignore:re2c
- more complexity
. Right now we strip out trailing white space and new lines after a comment
block. This can be a problem when the comment block was not preceeded by
a new line.
. The solution is to use trailing contexts.
- trailing contexts
. Re2c allows to check for a portion of input and only recognize it when it
is followed by another portion. This is called a trailing context.
. The trailing context is not part of the identified input. That means that
it follows exactly at the cursor. A consequence is that the scanner has
already read more input and on the next run you need to restore begining
of input, in our case s.tok, from the cursor, here s.cur, rather then
restoring to the beginning of the buffer. This way the scanner can reuse
the portion it has already read.
. The position of the trailing context is stored in YYCTXMARKER for which
a pointer variable needs to be provided.
. As with YYMARKER the corrsponding variable needs to be corrected if we
shift in some buffer.
. Still this is not all we need to solve the problem. What is left is that
the information whether we detected a trailing context was detected has to
be stored somewhere. This is done by the new variable nlcomment.
- formatting
. Until now we only used single line expression code and we always had the
opening { on the same line as the rule itself. If we have multiline rule
code and care for formatting we can no longer rely on re2c. Now we have
to indent the rule code ourself. Also we need to take care of the opening
{. If we keep it on the same line as the rule then re2c will indent it
correctly and the emitted #line informations will be correct. If we place
it on the next line then the #line directive will also point to that line
and not to the rule.
*/
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
/*!max:re2c */
#define BSIZE 128
#if BSIZE < YYMAXFILL
# error BSIZE must be greater YYMAXFILL
#endif
#define YYCTYPE unsigned char
#define YYCURSOR s.cur
#define YYLIMIT s.lim
#define YYMARKER s.mrk
#define YYCTXMARKER s.ctx
#define YYFILL(n) { if ((res = fill(&s, n)) >= 0) break; }
typedef struct Scanner
{
FILE *fp;
unsigned char *cur, *tok, *lim, *eof, *ctx, *mrk;
unsigned char buffer[BSIZE];
} Scanner;
int fill(Scanner *s, int len)
{
if (!len)
{
s->cur = s->tok = s->lim = s->mrk = s->buffer;
s->eof = 0;
}
if (!s->eof)
{
int got, cnt = s->tok - s->buffer;
if (cnt > 0)
{
memcpy(s->buffer, s->tok, s->lim - s->tok);
s->tok -= cnt;
s->cur -= cnt;
s->lim -= cnt;
s->mrk -= cnt;
s->ctx -= cnt;
}
cnt = BSIZE - cnt;
if ((got = fread(s->lim, 1, cnt, s->fp)) != cnt)
{
s->eof = &s->lim[got];
}
s->lim += got;
}
else if (s->cur + len > s->eof)
{
return 0; /* not enough input data */
}
return -1;
}
void echo(Scanner *s)
{
fwrite(s->tok, 1, s->cur - s->tok, stdout);
}
int scan(FILE *fp)
{
int res = 0;
int nlcomment = 0;
Scanner s;
if (!fp)
{
return 1; /* no file was opened */
}
s.fp = fp;
fill(&s, 0);
for(;;)
{
s.tok = s.cur;
/*!re2c
re2c:indent:top = 2;
NL = "\r"? "\n" ;
WS = [\r\n\t ] ;
ANY = [^] ;
"/" "/" { goto cppcomment; }
NL / "/""*" { echo(&s); nlcomment = 1; continue; }
"/" "*" { goto comment; }
ANY { fputc(*s.tok, stdout); continue; }
*/
comment:
s.tok = s.cur;
/*!re2c
"*" "/" { goto commentws; }
ANY { goto comment; }
*/
commentws:
s.tok = s.cur;
/*!re2c
NL? "/" "*" { goto comment; }
NL {
if (!nlcomment)
{
echo(&s);
}
nlcomment = 0;
continue;
}
WS { goto commentws; }
ANY { echo(&s); nlcomment = 0; continue; }
*/
cppcomment:
s.tok = s.cur;
/*!re2c
NL { echo(&s); continue; }
ANY { goto cppcomment; }
*/
}
if (fp != stdin)
{
fclose(fp); /* close only if not stdin */
}
return res; /* return result */
}
int main(int argc, char **argv)
{
if (argc > 1)
{
return scan(!strcmp(argv[1], "-") ? stdin : fopen(argv[1], "r"));
}
else
{
fprintf(stderr, "%s <expr>\n", argv[0]);
return 1;
}
}

35
tools/re2c/examples/input_custom/fixed.re Normal file
View file

@ -0,0 +1,35 @@
// Build with "--input custom" re2c switch.
//
// This is an example of handling fixed-length buffer with "--input custom":
// on each YYPEEK we check for the end of input, thus YYFILL generation
// can be safely suppressed.
//
// Note that YYLIMIT points not to terminating NULL, but to the previous
// character: we emulate the case when input has no terminating NULL.
//
// For a real-life example see https://github.com/sopyer/mjson
// or mjson.re from re2c test collection.
bool lex (const char * cursor, const char * const limit)
{
const char * marker;
const char * ctxmarker;
# define YYCTYPE char
# define YYPEEK() (cursor >= limit ? 0 : *cursor)
# define YYSKIP() ++cursor
# define YYBACKUP() marker = cursor
# define YYBACKUPCTX() ctxmarker = cursor
# define YYRESTORE() cursor = marker
# define YYRESTORECTX() cursor = ctxmarker
/*!re2c
re2c:yyfill:enable = 0;
"int buffer " / "[" [0-9]+ "]" { return true; }
* { return false; }
*/
}
int main ()
{
char buffer [] = "int buffer [1024]";
return !lex (buffer, buffer + sizeof (buffer) - 1);
}

20
tools/re2c/examples/input_custom/simple/README Normal file
View file

@ -0,0 +1,20 @@
Build with "--input custom" re2c switch.
These are three examples of "--input custom" usage:
- input_custom_default.re:
implements default re2c input model (pointers to plain buffer)
- input_custom_fgetc:
implements C-style file input (using <stdio.h>)
- input_custom_fgetc:
implements std::istringstream input
Note that these examples are very simple and don't need
to implement YYFILL; the only reason they don't use
"re2c:yyfill:enable = 0;" is to keep YYLESSTHAN and YYLIMIT
(for the sake of example).
In real-life programs one will need to care for correct
end-of-input handling.

24
tools/re2c/examples/input_custom/simple/default.re Normal file
View file

@ -0,0 +1,24 @@
bool lex (const char * cursor, const char * const limit)
{
const char * marker;
const char * ctxmarker;
# define YYCTYPE char
# define YYPEEK() *cursor
# define YYSKIP() ++cursor
# define YYBACKUP() marker = cursor
# define YYBACKUPCTX() ctxmarker = cursor
# define YYRESTORE() cursor = marker
# define YYRESTORECTX() cursor = ctxmarker
# define YYLESSTHAN(n) limit - cursor < n
# define YYFILL(n) {}
/*!re2c
"int buffer " / "[" [0-9]+ "]" { return true; }
* { return false; }
*/
}
int main ()
{
char buffer [] = "int buffer [1024]";
return !lex (buffer, buffer + sizeof (buffer));
}

43
tools/re2c/examples/input_custom/simple/fgetc.re Normal file
View file

@ -0,0 +1,43 @@
#include <stdio.h>
char peek (FILE * f)
{
char c = fgetc (f);
ungetc (c, f);
return c;
}
bool lex (FILE * f, const long limit)
{
long marker;
long ctxmarker;
# define YYCTYPE char
# define YYPEEK() peek (f)
# define YYSKIP() fgetc (f)
# define YYBACKUP() marker = ftell (f)
# define YYBACKUPCTX() ctxmarker = ftell (f)
# define YYRESTORE() fseek (f, marker, SEEK_SET)
# define YYRESTORECTX() fseek (f, ctxmarker, SEEK_SET)
# define YYLESSTHAN(n) limit - ftell (f) < n
# define YYFILL(n) {}
/*!re2c
"int buffer " / "[" [0-9]+ "]" { return true; }
* { return false; }
*/
}
int main ()
{
const char buffer [] = "int buffer [1024]";
const char fn [] = "input.txt";
FILE * f = fopen (fn, "w");
fwrite (buffer, 1, sizeof (buffer), f);
fclose (f);
f = fopen (fn, "rb");
int result = !lex (f, sizeof (buffer));
fclose (f);
return result;
}

27
tools/re2c/examples/input_custom/simple/istringstream.re Normal file
View file

@ -0,0 +1,27 @@
#include <sstream>
bool lex (std::istringstream & is, const std::streampos limit)
{
std::streampos marker;
std::streampos ctxmarker;
# define YYCTYPE char
# define YYPEEK() is.peek ()
# define YYSKIP() is.ignore ()
# define YYBACKUP() marker = is.tellg ()
# define YYBACKUPCTX() ctxmarker = is.tellg ()
# define YYRESTORE() is.seekg (marker)
# define YYRESTORECTX() is.seekg (ctxmarker)
# define YYLESSTHAN(n) limit - is.tellg () < n
# define YYFILL(n) {}
/*!re2c
"int buffer " / "[" [0-9]+ "]" { return true; }
* { return false; }
*/
}
int main ()
{
const char buffer [] = "int buffer [1024]";
std::istringstream is (buffer);
return !lex (is, sizeof (buffer));
}

272
tools/re2c/examples/langs/c.re Normal file
View file

@ -0,0 +1,272 @@
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#define ADDEQ 257
#define ANDAND 258
#define ANDEQ 259
#define ARRAY 260
#define ASM 261
#define AUTO 262
#define BREAK 263
#define CASE 264
#define CHAR 265
#define CONST 266
#define CONTINUE 267
#define DECR 268
#define DEFAULT 269
#define DEREF 270
#define DIVEQ 271
#define DO 272
#define DOUBLE 273
#define ELLIPSIS 274
#define ELSE 275
#define ENUM 276
#define EQL 277
#define EXTERN 278
#define FCON 279
#define FLOAT 280
#define FOR 281
#define FUNCTION 282
#define GEQ 283
#define GOTO 284
#define ICON 285
#define ID 286
#define IF 287
#define INCR 288
#define INT 289
#define LEQ 290
#define LONG 291
#define LSHIFT 292
#define LSHIFTEQ 293
#define MODEQ 294
#define MULEQ 295
#define NEQ 296
#define OREQ 297
#define OROR 298
#define POINTER 299
#define REGISTER 300
#define RETURN 301
#define RSHIFT 302
#define RSHIFTEQ 303
#define SCON 304
#define SHORT 305
#define SIGNED 306
#define SIZEOF 307
#define STATIC 308
#define STRUCT 309
#define SUBEQ 310
#define SWITCH 311
#define TYPEDEF 312
#define UNION 313
#define UNSIGNED 314
#define VOID 315
#define VOLATILE 316
#define WHILE 317
#define XOREQ 318
#define EOI 319
typedef unsigned int uint;
typedef unsigned char uchar;
#define BSIZE 8192
#define YYCTYPE uchar
#define YYCURSOR cursor
#define YYLIMIT s->lim
#define YYMARKER s->ptr
#define YYFILL(n) {cursor = fill(s, cursor);}
#define RET(i) {s->cur = cursor; return i;}
typedef struct Scanner {
int fd;
uchar *bot, *tok, *ptr, *cur, *pos, *lim, *top, *eof;
uint line;
} Scanner;
uchar *fill(Scanner *s, uchar *cursor){
if(!s->eof) {
uint cnt = s->tok - s->bot;
if(cnt){
memcpy(s->bot, s->tok, s->lim - s->tok);
s->tok = s->bot;
s->ptr -= cnt;
cursor -= cnt;
s->pos -= cnt;
s->lim -= cnt;
}
if((s->top - s->lim) < BSIZE){
uchar *buf = (uchar*) malloc(((s->lim - s->bot) + BSIZE)*sizeof(uchar));
memcpy(buf, s->tok, s->lim - s->tok);
s->tok = buf;
s->ptr = &buf[s->ptr - s->bot];
cursor = &buf[cursor - s->bot];
s->pos = &buf[s->pos - s->bot];
s->lim = &buf[s->lim - s->bot];
s->top = &s->lim[BSIZE];
free(s->bot);
s->bot = buf;
}
if((cnt = read(s->fd, (char*) s->lim, BSIZE)) != BSIZE){
s->eof = &s->lim[cnt]; *(s->eof)++ = '\n';
}
s->lim += cnt;
}
return cursor;
}
int scan(Scanner *s){
uchar *cursor = s->cur;
std:
s->tok = cursor;
/*!re2c
any = [\000-\377];
O = [0-7];
D = [0-9];
L = [a-zA-Z_];
H = [a-fA-F0-9];
E = [Ee] [+-]? D+;
FS = [fFlL];
IS = [uUlL]*;
ESC = [\\] ([abfnrtv?'"\\] | "x" H+ | O+);
*/
/*!re2c
"/*" { goto comment; }
"auto" { RET(AUTO); }
"break" { RET(BREAK); }
"case" { RET(CASE); }
"char" { RET(CHAR); }
"const" { RET(CONST); }
"continue" { RET(CONTINUE); }
"default" { RET(DEFAULT); }
"do" { RET(DO); }
"double" { RET(DOUBLE); }
"else" { RET(ELSE); }
"enum" { RET(ENUM); }
"extern" { RET(EXTERN); }
"float" { RET(FLOAT); }
"for" { RET(FOR); }
"goto" { RET(GOTO); }
"if" { RET(IF); }
"int" { RET(INT); }
"long" { RET(LONG); }
"register" { RET(REGISTER); }
"return" { RET(RETURN); }
"short" { RET(SHORT); }
"signed" { RET(SIGNED); }
"sizeof" { RET(SIZEOF); }
"static" { RET(STATIC); }
"struct" { RET(STRUCT); }
"switch" { RET(SWITCH); }
"typedef" { RET(TYPEDEF); }
"union" { RET(UNION); }
"unsigned" { RET(UNSIGNED); }
"void" { RET(VOID); }
"volatile" { RET(VOLATILE); }
"while" { RET(WHILE); }
L (L|D)* { RET(ID); }
("0" [xX] H+ IS?) | ("0" D+ IS?) | (D+ IS?) |
(['] (ESC|any\[\n\\'])* ['])
{ RET(ICON); }
(D+ E FS?) | (D* "." D+ E? FS?) | (D+ "." D* E? FS?)
{ RET(FCON); }
(["] (ESC|any\[\n\\"])* ["])
{ RET(SCON); }
"..." { RET(ELLIPSIS); }
">>=" { RET(RSHIFTEQ); }
"<<=" { RET(LSHIFTEQ); }
"+=" { RET(ADDEQ); }
"-=" { RET(SUBEQ); }
"*=" { RET(MULEQ); }
"/=" { RET(DIVEQ); }
"%=" { RET(MODEQ); }
"&=" { RET(ANDEQ); }
"^=" { RET(XOREQ); }
"|=" { RET(OREQ); }
">>" { RET(RSHIFT); }
"<<" { RET(LSHIFT); }
"++" { RET(INCR); }
"--" { RET(DECR); }
"->" { RET(DEREF); }
"&&" { RET(ANDAND); }
"||" { RET(OROR); }
"<=" { RET(LEQ); }
">=" { RET(GEQ); }
"==" { RET(EQL); }
"!=" { RET(NEQ); }
";" { RET(';'); }
"{" { RET('{'); }
"}" { RET('}'); }
"," { RET(','); }
":" { RET(':'); }
"=" { RET('='); }
"(" { RET('('); }
")" { RET(')'); }
"[" { RET('['); }
"]" { RET(']'); }
"." { RET('.'); }
"&" { RET('&'); }
"!" { RET('!'); }
"~" { RET('~'); }
"-" { RET('-'); }
"+" { RET('+'); }
"*" { RET('*'); }
"/" { RET('/'); }
"%" { RET('%'); }
"<" { RET('<'); }
">" { RET('>'); }
"^" { RET('^'); }
"|" { RET('|'); }
"?" { RET('?'); }
[ \t\v\f]+ { goto std; }
"\n"
{
if(cursor == s->eof) RET(EOI);
s->pos = cursor; s->line++;
goto std;
}
any
{
printf("unexpected character: %c\n", *s->tok);
goto std;
}
*/
comment:
/*!re2c
"*/" { goto std; }
"\n"
{
if(cursor == s->eof) RET(EOI);
s->tok = s->pos = cursor; s->line++;
goto comment;
}
any { goto comment; }
*/
}
main(){
Scanner in;
int t;
memset((char*) &in, 0, sizeof(in));
in.fd = 0;
while((t = scan(&in)) != EOI){
/*
printf("%d\t%.*s\n", t, in.cur - in.tok, in.tok);
printf("%d\n", t);
*/
}
close(in.fd);
}

203
tools/re2c/examples/langs/modula.re Normal file
View file

@ -0,0 +1,203 @@
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
typedef unsigned int uint;
typedef unsigned char uchar;
#define BSIZE 8192
#define YYCTYPE uchar
#define YYCURSOR cursor
#define YYLIMIT s->lim
#define YYMARKER s->ptr
#define YYCTXMARKER s->ctx
#define YYFILL {cursor = fill(s, cursor);}
#define RETURN(i) {s->cur = cursor; return i;}
typedef struct Scanner {
int fd;
uchar *bot, *tok, *ptr, *ctx, *cur, *pos, *lim, *top, *eof;
uint line;
} Scanner;
uchar *fill(Scanner *s, uchar *cursor){
if(!s->eof){
uint cnt = s->tok - s->bot;
if(cnt){
memcpy(s->bot, s->tok, s->lim - s->tok);
s->tok = s->bot;
s->ptr -= cnt;
cursor -= cnt;
s->pos -= cnt;
s->lim -= cnt;
}
if((s->top - s->lim) < BSIZE){
uchar *buf = (uchar*) malloc(((s->lim - s->bot) + BSIZE)*sizeof(uchar));
memcpy(buf, s->tok, s->lim - s->tok);
s->tok = buf;
s->ptr = &buf[s->ptr - s->bot];
cursor = &buf[cursor - s->bot];
s->pos = &buf[s->pos - s->bot];
s->lim = &buf[s->lim - s->bot];
s->top = &s->lim[BSIZE];
free(s->bot);
s->bot = buf;
}
if((cnt = read(s->fd, (char*) s->lim, BSIZE)) != BSIZE){
s->eof = &s->lim[cnt]; *(s->eof)++ = '\n';
}
s->lim += cnt;
}
return cursor;
}
int scan(Scanner *s){
uchar *cursor = s->cur;
uint depth;
std:
s->tok = cursor;
/*!re2c
any = [\000-\377];
digit = [0-9];
letter = [a-zA-Z];
*/
/*!re2c
"(*" { depth = 1; goto comment; }
digit + {RETURN(1);}
digit + / ".." {RETURN(1);}
[0-7] + "B" {RETURN(2);}
[0-7] + "C" {RETURN(3);}
digit [0-9A-F] * "H" {RETURN(4);}
digit + "." digit * ("E" ([+-]) ? digit +) ? {RETURN(5);}
['] (any\[\n']) * ['] | ["] (any\[\n"]) * ["] {RETURN(6);}
"#" {RETURN(7);}
"&" {RETURN(8);}
"(" {RETURN(9);}
")" {RETURN(10);}
"*" {RETURN(11);}
"+" {RETURN(12);}
"," {RETURN(13);}
"-" {RETURN(14);}
"." {RETURN(15);}
".." {RETURN(16);}
"/" {RETURN(17);}
":" {RETURN(18);}
":=" {RETURN(19);}
";" {RETURN(20);}
"<" {RETURN(21);}
"<=" {RETURN(22);}
"<>" {RETURN(23);}
"=" {RETURN(24);}
">" {RETURN(25);}
">=" {RETURN(26);}
"[" {RETURN(27);}
"]" {RETURN(28);}
"^" {RETURN(29);}
"{" {RETURN(30);}
"|" {RETURN(31);}
"}" {RETURN(32);}
"~" {RETURN(33);}
"AND" {RETURN(34);}
"ARRAY" {RETURN(35);}
"BEGIN" {RETURN(36);}
"BY" {RETURN(37);}
"CASE" {RETURN(38);}
"CONST" {RETURN(39);}
"DEFINITION" {RETURN(40);}
"DIV" {RETURN(41);}
"DO" {RETURN(42);}
"ELSE" {RETURN(43);}
"ELSIF" {RETURN(44);}
"END" {RETURN(45);}
"EXIT" {RETURN(46);}
"EXPORT" {RETURN(47);}
"FOR" {RETURN(48);}
"FROM" {RETURN(49);}
"IF" {RETURN(50);}
"IMPLEMENTATION" {RETURN(51);}
"IMPORT" {RETURN(52);}
"IN" {RETURN(53);}
"LOOP" {RETURN(54);}
"MOD" {RETURN(55);}
"MODULE" {RETURN(56);}
"NOT" {RETURN(57);}
"OF" {RETURN(58);}
"OR" {RETURN(59);}
"POINTER" {RETURN(60);}
"PROCEDURE" {RETURN(61);}
"QUALIFIED" {RETURN(62);}
"RECORD" {RETURN(63);}
"REPEAT" {RETURN(64);}
"RETURN" {RETURN(65);}
"SET" {RETURN(66);}
"THEN" {RETURN(67);}
"TO" {RETURN(68);}
"TYPE" {RETURN(69);}
"UNTIL" {RETURN(70);}
"VAR" {RETURN(71);}
"WHILE" {RETURN(72);}
"WITH" {RETURN(73);}
letter (letter | digit) * {RETURN(74);}
[ \t]+ { goto std; }
"\n"
{
if(cursor == s->eof) RETURN(0);
s->pos = cursor; s->line++;
goto std;
}
any
{
printf("unexpected character: %c\n", *s->tok);
goto std;
}
*/
comment:
/*!re2c
"*)"
{
if(--depth == 0)
goto std;
else
goto comment;
}
"(*" { ++depth; goto comment; }
"\n"
{
if(cursor == s->eof) RETURN(0);
s->tok = s->pos = cursor; s->line++;
goto comment;
}
any { goto comment; }
*/
}
/*
void putStr(FILE *o, char *s, uint l){
while(l-- > 0)
putc(*s++, o);
}
*/
main(){
Scanner in;
memset((char*) &in, 0, sizeof(in));
in.fd = 0;
while(scan(&in)){
/*
putc('<', stdout);
putStr(stdout, (char*) in.tok, in.cur - in.tok);
putc('>', stdout);
putc('\n', stdout);
*/
}
}

319
tools/re2c/examples/langs/rexx.re Normal file
View file

@ -0,0 +1,319 @@
#include "scanio.h"
#include "scanner.h"
#define CURSOR ch
#define LOADCURSOR ch = *cursor;
#define ADVANCE cursor++;
#define BACK(n) cursor -= (n);
#define CHECK(n) if((ScanCB.lim - cursor) < (n)){cursor = ScanFill(cursor);}
#define MARK(n) ScanCB.ptr = cursor; sel = (n);
#define REVERT cursor = ScanCB.ptr;
#define MARKER sel
#define RETURN(i) {ScanCB.cur = cursor; return i;}
int ScanToken(){
uchar *cursor = ScanCB.cur;
unsigned sel;
uchar ch;
ScanCB.tok = cursor;
ScanCB.eot = NULL;
/*!re2c
all = [\000-\377];
eof = [\000];
any = all\eof;
letter = [a-z]|[A-Z];
digit = [0-9];
symchr = letter|digit|[.!?_];
const = (digit|[.])symchr*([eE][+-]?digit+)?;
simple = (symchr\(digit|[.]))(symchr\[.])*;
stem = simple [.];
symbol = symchr*;
sqstr = ['] ((any\['\n])|(['][']))* ['];
dqstr = ["] ((any\["\n])|(["]["]))* ["];
str = sqstr|dqstr;
ob = [ \t]*;
not = [\\~];
A = [aA];
B = [bB];
C = [cC];
D = [dD];
E = [eE];
F = [fF];
G = [gG];
H = [hH];
I = [iI];
J = [jJ];
K = [kK];
L = [lL];
M = [mM];
N = [nN];
O = [oO];
P = [pP];
Q = [qQ];
R = [rR];
S = [sS];
T = [tT];
U = [uU];
V = [vV];
W = [wW];
X = [xX];
Y = [yY];
Z = [zZ];
*/
scan:
/*!re2c
"\n"
{
++(ScanCB.lineNum);
ScanCB.linePos = ScanCB.pos + (cursor - ScanCB.mrk);
RETURN(SU_EOL);
}
"|" ob "|"
{ RETURN(OP_CONCAT); }
"+"
{ RETURN(OP_PLUS); }
"-"
{ RETURN(OP_MINUS); }
"*"
{ RETURN(OP_MULT); }
"/"
{ RETURN(OP_DIV); }
"%"
{ RETURN(OP_IDIV); }
"/" ob "/"
{ RETURN(OP_REMAIN); }
"*" ob "*"
{ RETURN(OP_POWER); }
"="
{ RETURN(OP_EQUAL); }
not ob "=" | "<" ob ">" | ">" ob "<"
{ RETURN(OP_EQUAL_N); }
">"
{ RETURN(OP_GT); }
"<"
{ RETURN(OP_LT); }
">" ob "=" | not ob "<"
{ RETURN(OP_GE); }
"<" ob "=" | not ob ">"
{ RETURN(OP_LE); }
"=" ob "="
{ RETURN(OP_EQUAL_EQ); }
not ob "=" ob "="
{ RETURN(OP_EQUAL_EQ_N); }
">" ob ">"
{ RETURN(OP_GT_STRICT); }
"<" ob "<"
{ RETURN(OP_LT_STRICT); }
">" ob ">" ob "=" | not ob "<" ob "<"
{ RETURN(OP_GE_STRICT); }
"<" ob "<" ob "=" | not ob ">" ob ">"
{ RETURN(OP_LE_STRICT); }
"&"
{ RETURN(OP_AND); }
"|"
{ RETURN(OP_OR); }
"&" ob "&"
{ RETURN(OP_XOR); }
not
{ RETURN(OP_NOT); }
":"
{ RETURN(SU_COLON); }
","
{ RETURN(SU_COMMA); }
"("
{ RETURN(SU_POPEN); }
")"
{ RETURN(SU_PCLOSE); }
";"
{ RETURN(SU_EOC); }
A D D R E S S
{ RETURN(RX_ADDRESS); }
A R G
{ RETURN(RX_ARG); }
C A L L
{ RETURN(RX_CALL); }
D O
{ RETURN(RX_DO); }
D R O P
{ RETURN(RX_DROP); }
E L S E
{ RETURN(RX_ELSE); }
E N D
{ RETURN(RX_END); }
E X I T
{ RETURN(RX_EXIT); }
I F
{ RETURN(RX_IF); }
I N T E R P R E T
{ RETURN(RX_INTERPRET); }
I T E R A T E
{ RETURN(RX_ITERATE); }
L E A V E
{ RETURN(RX_LEAVE); }
N O P
{ RETURN(RX_NOP); }
N U M E R I C
{ RETURN(RX_NUMERIC); }
O P T I O N S
{ RETURN(RX_OPTIONS); }
O T H E R W I S E
{ RETURN(RX_OTHERWISE); }
P A R S E
{ RETURN(RX_PARSE); }
P R O C E D U R E
{ RETURN(RX_PROCEDURE); }
P U L L
{ RETURN(RX_PULL); }
P U S H
{ RETURN(RX_PUSH); }
Q U E U E
{ RETURN(RX_QUEUE); }
R E T U R N
{ RETURN(RX_RETURN); }
S A Y
{ RETURN(RX_SAY); }
S E L E C T
{ RETURN(RX_SELECT); }
S I G N A L
{ RETURN(RX_SIGNAL); }
T H E N
{ RETURN(RX_THEN); }
T R A C E
{ RETURN(RX_TRACE); }
W H E N
{ RETURN(RX_WHEN); }
O F F
{ RETURN(RXS_OFF); }
O N
{ RETURN(RXS_ON); }
B Y
{ RETURN(RXS_BY); }
D I G I T S
{ RETURN(RXS_DIGITS); }
E N G I N E E R I N G
{ RETURN(RXS_ENGINEERING); }
E R R O R
{ RETURN(RXS_ERROR); }
E X P O S E
{ RETURN(RXS_EXPOSE); }
F A I L U R E
{ RETURN(RXS_FAILURE); }
F O R
{ RETURN(RXS_FOR); }
F O R E V E R
{ RETURN(RXS_FOREVER); }
F O R M
{ RETURN(RXS_FORM); }
F U Z Z
{ RETURN(RXS_FUZZ); }
H A L T
{ RETURN(RXS_HALT); }
L I N E I N
{ RETURN(RXS_LINEIN); }
N A M E
{ RETURN(RXS_NAME); }
N O T R E A D Y
{ RETURN(RXS_NOTREADY); }
N O V A L U E
{ RETURN(RXS_NOVALUE); }
S C I E N T I F I C
{ RETURN(RXS_SCIENTIFIC); }
S O U R C E
{ RETURN(RXS_SOURCE); }
S Y N T A X
{ RETURN(RXS_SYNTAX); }
T O
{ RETURN(RXS_TO); }
U N T I L
{ RETURN(RXS_UNTIL); }
U P P E R
{ RETURN(RXS_UPPER); }
V A L U E
{ RETURN(RXS_VALUE); }
V A R
{ RETURN(RXS_VAR); }
V E R S I O N
{ RETURN(RXS_VERSION); }
W H I L E
{ RETURN(RXS_WHILE); }
W I T H
{ RETURN(RXS_WITH); }
const
{ RETURN(SU_CONST); }
simple
{ RETURN(SU_SYMBOL); }
stem
{ RETURN(SU_SYMBOL_STEM); }
symbol
{ RETURN(SU_SYMBOL_COMPOUND); }
str
{ RETURN(SU_LITERAL); }
str [bB] / (all\symchr)
{ RETURN(SU_LITERAL_BIN); }
str [xX] / (all\symchr)
{ RETURN(SU_LITERAL_HEX); }
eof
{ RETURN(SU_EOF); }
any
{ RETURN(SU_ERROR); }
*/
}
bool StripToken(){
uchar *cursor = ScanCB.cur;
unsigned depth;
uchar ch;
bool blanks = FALSE;
ScanCB.eot = cursor;
strip:
/*!re2c
"/*"
{
depth = 1;
goto comment;
}
"\r"
{ goto strip; }
[ \t]
{
blanks = TRUE;
goto strip;
}
[] / all
{ RETURN(blanks); }
*/
comment:
/*!re2c
"*/"
{
if(--depth == 0)
goto strip;
else
goto comment;
}
"\n"
{
++(ScanCB.lineNum);
ScanCB.linePos = ScanCB.pos + (cursor - ScanCB.mrk);
goto comment;
}
"/*"
{
++depth;
goto comment;
}
eof
{ RETURN(blanks); }
any
{
goto comment;
}
*/
}

340
tools/re2c/examples/push_model/push.re Normal file
View file

@ -0,0 +1,340 @@
/*
* A push-model scanner example for re2c -f
* Written Mon Apr 11 2005 by mgix@mgix.com
* This file is in the public domain.
*
*/
// ----------------------------------------------------------------------
#include <fcntl.h>
#include <stdio.h>
#include <stddef.h>
#include <stdlib.h>
#include <string.h>
#if defined(WIN32)
typedef signed char int8_t;
typedef signed short int16_t;
typedef signed int int32_t;
typedef unsigned char uint8_t;
typedef unsigned short uint16_t;
typedef unsigned int uint32_t;
#else
#include <stdint.h>
#include <unistd.h>
#ifndef O_BINARY
#define O_BINARY 0
#endif
#endif
// ----------------------------------------------------------------------
#define TOKENS \
\
TOK(kEOF) \
TOK(kEOL) \
TOK(kUnknown) \
TOK(kIdentifier) \
TOK(kDecimalConstant) \
\
TOK(kEqual) \
TOK(kLeftParen) \
TOK(kRightParen) \
TOK(kMinus) \
TOK(kPlus) \
TOK(kStar) \
TOK(kSlash) \
\
TOK(kIf) \
TOK(kFor) \
TOK(kElse) \
TOK(kGoto) \
TOK(kBreak) \
TOK(kWhile) \
TOK(kReturn) \
// ----------------------------------------------------------------------
static const char *tokenNames[] =
{
#define TOK(x) #x,
TOKENS
#undef TOK
};
// ----------------------------------------------------------------------
class PushScanner
{
public:
enum Token
{
#define TOK(x) x,
TOKENS
#undef TOK
};
private:
bool eof;
int32_t state;
uint8_t *limit;
uint8_t *start;
uint8_t *cursor;
uint8_t *marker;
uint8_t *buffer;
uint8_t *bufferEnd;
uint8_t yych;
uint32_t yyaccept;
public:
// ----------------------------------------------------------------------
PushScanner()
{
limit = 0;
start = 0;
state = -1;
cursor = 0;
marker = 0;
buffer = 0;
eof = false;
bufferEnd = 0;
}
// ----------------------------------------------------------------------
~PushScanner()
{
}
// ----------------------------------------------------------------------
void send(
Token token
)
{
size_t tokenSize = cursor-start;
const char *tokenName = tokenNames[token];
printf(
"scanner is pushing out a token of type %d (%s)",
token,
tokenName
);
if(token==kEOF) putchar('\n');
else
{
size_t tokenNameSize = strlen(tokenNames[token]);
size_t padSize = 20-(20<tokenNameSize ? 20 : tokenNameSize);
for(size_t i=0; i<padSize; ++i) putchar(' ');
printf(" : ---->");
fwrite(
start,
tokenSize,
1,
stdout
);
printf("<----\n");
}
}
// ----------------------------------------------------------------------
uint32_t push(
const void *input,
ssize_t inputSize
)
{
printf(
"scanner is receiving a new data batch of length %d\n"
"scanner continues with saved state = %d\n",
inputSize,
state
);
/*
* Data source is signaling end of file when batch size
* is less than maxFill. This is slightly annoying because
* maxFill is a value that can only be known after re2c does
* its thing. Practically though, maxFill is never bigger than
* the longest keyword, so given our grammar, 32 is a safe bet.
*/
uint8_t null[64];
const ssize_t maxFill = 32;
if(inputSize<maxFill)
{
eof = true;
input = null;
inputSize = sizeof(null);
memset(null, 0, sizeof(null));
}
/*
* When we get here, we have a partially
* consumed buffer which is in the following state:
* last valid char last valid buffer spot
* v v
* +-------------------+-------------+---------------+-------------+----------------------+
* ^ ^ ^ ^ ^ ^
* buffer start marker cursor limit bufferEnd
*
* We need to stretch the buffer and concatenate the new chunk of input to it
*
*/
size_t used = limit-buffer;
size_t needed = used+inputSize;
size_t allocated = bufferEnd-buffer;
if(allocated<needed)
{
size_t limitOffset = limit-buffer;
size_t startOffset = start-buffer;
size_t markerOffset = marker-buffer;
size_t cursorOffset = cursor-buffer;
buffer = (uint8_t*)realloc(buffer, needed);
bufferEnd = needed+buffer;
marker = markerOffset + buffer;
cursor = cursorOffset + buffer;
start = buffer + startOffset;
limit = limitOffset + buffer;
}
memcpy(limit, input, inputSize);
limit += inputSize;
// The scanner starts here
#define YYLIMIT limit
#define YYCURSOR cursor
#define YYMARKER marker
#define YYCTYPE uint8_t
#define SKIP(x) { start = cursor; goto yy0; }
#define SEND(x) { send(x); SKIP(); }
#define YYFILL(n) { goto fill; }
#define YYGETSTATE() state
#define YYSETSTATE(x) { state = (x); }
start:
/*!re2c
re2c:startlabel = 1;
eol = "\n";
eof = "\000";
digit = [0-9];
integer = digit+;
alpha = [A-Za-z_];
any = [\000-\377];
space = [ \h\t\v\f\r];
"if" { SEND(kIf); }
"for" { SEND(kFor); }
"else" { SEND(kElse); }
"goto" { SEND(kGoto); }
"break" { SEND(kBreak); }
"while" { SEND(kWhile); }
"return" { SEND(kReturn); }
alpha (alpha|digit)* { SEND(kIdentifier); }
integer { SEND(kDecimalConstant);}
"=" { SEND(kEqual); }
"(" { SEND(kLeftParen); }
")" { SEND(kRightParen); }
"-" { SEND(kMinus); }
"+" { SEND(kPlus); }
"*" { SEND(kStar); }
"/" { SEND(kSlash); }
eol { SKIP(); }
space { SKIP(); }
eof { send(kEOF); return 1; }
any { SEND(kUnknown); }
*/
fill:
ssize_t unfinishedSize = cursor-start;
printf(
"scanner needs a refill. Exiting for now with:\n"
" saved fill state = %d\n"
" unfinished token size = %d\n",
state,
unfinishedSize
);
if(0<unfinishedSize && start<limit)
{
printf(" unfinished token is :");
fwrite(start, 1, cursor-start, stdout);
putchar('\n');
}
putchar('\n');
/*
* Once we get here, we can get rid of
* everything before start and after limit.
*/
if(eof==true) goto start;
if(buffer<start)
{
size_t startOffset = start-buffer;
memmove(buffer, start, limit-start);
marker -= startOffset;
cursor -= startOffset;
limit -= startOffset;
start -= startOffset;
}
return 0;
}
};
// ----------------------------------------------------------------------
int main(
int argc,
char **argv
)
{
// Parse cmd line
int input = 0;
if(1<argc)
{
input = open(argv[1], O_RDONLY | O_BINARY);
if(input<0)
{
fprintf(
stderr,
"could not open file %s\n",
argv[1]
);
exit(1);
}
}
/*
* Tokenize input file by pushing batches
* of data one by one into the scanner.
*/
const size_t batchSize = 256;
uint8_t buffer[batchSize];
PushScanner scanner;
while(1)
{
ssize_t n = read(input, buffer, batchSize);
scanner.push(buffer, n);
if(n<batchSize) break;
}
scanner.push(0, -1);
close(input);
// Done
return 0;
}

168
tools/re2c/src/codegen/bitmap.cc Normal file
View file

@ -0,0 +1,168 @@
#include <algorithm> // min
#include <string.h> // memset
#include "src/codegen/bitmap.h"
#include "src/codegen/go.h"
#include "src/codegen/output.h"
#include "src/conf/opt.h"
#include "src/globals.h"
namespace re2c
{
BitMap *BitMap::first = NULL;
BitMap::BitMap(const Go *g, const State *x)
: go(g)
, on(x)
, next(first)
, i(0)
, m(0)
{
first = this;
}
BitMap::~BitMap()
{
delete next;
}
const BitMap *BitMap::find(const Go *g, const State *x)
{
for (const BitMap *b = first; b; b = b->next)
{
if (matches(b->go->span, b->go->nSpans, b->on, g->span, g->nSpans, x))
{
return b;
}
}
return new BitMap(g, x);
}
const BitMap *BitMap::find(const State *x)
{
for (const BitMap *b = first; b; b = b->next)
{
if (b->on == x)
{
return b;
}
}
return NULL;
}
static void doGen(const Go *g, const State *s, uint32_t *bm, uint32_t f, uint32_t m)
{
Span *b = g->span, *e = &b[g->nSpans];
uint32_t lb = 0;
for (; b < e; ++b)
{
if (b->to == s)
{
for (; lb < b->ub && lb < 256; ++lb)
{
bm[lb-f] |= m;
}
}
lb = b->ub;
}
}
void BitMap::gen(OutputFile & o, uint32_t ind, uint32_t lb, uint32_t ub)
{
if (first && bUsedYYBitmap)
{
o.wind(ind).ws("static const unsigned char ").wstring(opts->yybm).ws("[] = {");
uint32_t c = 1, n = ub - lb;
const BitMap *cb = first;
while((cb = cb->next) != NULL) {
++c;
}
BitMap *b = first;
uint32_t *bm = new uint32_t[n];
for (uint32_t i = 0, t = 1; b; i += n, t += 8)
{
memset(bm, 0, n * sizeof(uint32_t));
for (uint32_t m = 0x80; b && m; m >>= 1)
{
b->i = i;
b->m = m;
doGen(b->go, b->on, bm, lb, m);
b = const_cast<BitMap*>(b->next);
}
if (c > 8)
{
o.ws("\n").wind(ind+1).ws("/* table ").wu32(t).ws(" .. ").wu32(std::min(c, t+7)).ws(": ").wu32(i).ws(" */");
}
for (uint32_t j = 0; j < n; ++j)
{
if (j % 8 == 0)
{
o.ws("\n").wind(ind+1);
}
if (opts->yybmHexTable)
{
o.wu32_hex(bm[j]);
}
else
{
o.wu32_width(bm[j], 3);
}
o.ws(", ");
}
}
o.ws("\n").wind(ind).ws("};\n");
delete[] bm;
}
}
// All spans in b1 that lead to s1 are pairwise equal to that in b2 leading to s2
bool matches(const Span * b1, uint32_t n1, const State * s1, const Span * b2, uint32_t n2, const State * s2)
{
const Span * e1 = &b1[n1];
uint32_t lb1 = 0;
const Span * e2 = &b2[n2];
uint32_t lb2 = 0;
for (;;)
{
for (; b1 < e1 && b1->to != s1; ++b1)
{
lb1 = b1->ub;
}
for (; b2 < e2 && b2->to != s2; ++b2)
{
lb2 = b2->ub;
}
if (b1 == e1)
{
return b2 == e2;
}
if (b2 == e2)
{
return false;
}
if (lb1 != lb2 || b1->ub != b2->ub)
{
return false;
}
++b1;
++b2;
}
}
} // end namespace re2c

45
tools/re2c/src/codegen/bitmap.h Normal file
View file

@ -0,0 +1,45 @@
#ifndef _RE2C_CODEGEN_BITMAP_
#define _RE2C_CODEGEN_BITMAP_
#include "src/util/c99_stdint.h"
#include "src/util/forbid_copy.h"
namespace re2c
{
struct Go;
struct Span;
struct State;
struct OutputFile;
class BitMap
{
public:
static BitMap *first;
const Go *go;
const State *on;
const BitMap *next;
uint32_t i;
uint32_t m;
public:
static const BitMap *find(const Go*, const State*);
static const BitMap *find(const State*);
static void gen(OutputFile &, uint32_t ind, uint32_t, uint32_t);
BitMap(const Go*, const State*);
~BitMap();
FORBID_COPY (BitMap);
};
bool matches(const Span * b1, uint32_t n1, const State * s1, const Span * b2, uint32_t n2, const State * s2);
#ifdef _MSC_VER
# pragma warning(disable: 4355) /* 'this' : used in base member initializer list */
#endif
} // end namespace re2c
#endif // _RE2C_CODEGEN_BITMAP_

43
tools/re2c/src/codegen/emit.h Normal file
View file

@ -0,0 +1,43 @@
#ifndef _RE2C_CODEGEN_EMIT_
#define _RE2C_CODEGEN_EMIT_
#include "src/codegen/output.h"
#include "src/ir/adfa/adfa.h"
namespace re2c {
typedef std::vector<std::string> RegExpIndices;
void emit_action
( const Action & action
, OutputFile & o
, uint32_t ind
, bool & readCh
, const State * const s
, const std::string & condName
, const Skeleton * skeleton
, const std::set<label_t> & used_labels
, bool save_yyaccept
);
// helpers
void genGoTo (OutputFile & o, uint32_t ind, const State * from, const State * to, bool & readCh);
template<typename _Ty> std::string replaceParam (std::string str, const std::string & param, const _Ty & value)
{
if (!param.empty ())
{
std::ostringstream strValue;
strValue << value;
std::string::size_type pos;
while((pos = str.find(param)) != std::string::npos)
{
str.replace(pos, param.length(), strValue.str());
}
}
return str;
}
} // namespace re2c
#endif // _RE2C_CODEGEN_EMIT_

388
tools/re2c/src/codegen/emit_action.cc Normal file
View file

@ -0,0 +1,388 @@
#include "src/util/c99_stdint.h"
#include <stddef.h>
#include <set>
#include <string>
#include "src/codegen/emit.h"
#include "src/codegen/input_api.h"
#include "src/codegen/output.h"
#include "src/conf/opt.h"
#include "src/globals.h"
#include "src/ir/adfa/action.h"
#include "src/ir/adfa/adfa.h"
#include "src/ir/regexp/regexp.h"
#include "src/ir/regexp/regexp_rule.h"
#include "src/ir/skeleton/skeleton.h"
#include "src/parse/code.h"
#include "src/parse/loc.h"
namespace re2c
{
class label_t;
static void need (OutputFile & o, uint32_t ind, bool & readCh, size_t n, bool bSetMarker);
static void emit_match (OutputFile & o, uint32_t ind, bool & readCh, const State * const s);
static void emit_initial (OutputFile & o, uint32_t ind, bool & readCh, const State * const s, const Initial & init, const std::set<label_t> & used_labels);
static void emit_save (OutputFile & o, uint32_t ind, bool & readCh, const State * const s, uint32_t save, bool save_yyaccept);
static void emit_accept_binary (OutputFile & o, uint32_t ind, bool & readCh, const State * const s, const accept_t & accept, size_t l, size_t r);
static void emit_accept (OutputFile & o, uint32_t ind, bool & readCh, const State * const s, const accept_t & accept);
static void emit_rule (OutputFile & o, uint32_t ind, const State * const s, const RuleOp * const rule, const std::string & condName, const Skeleton * skeleton);
static void genYYFill (OutputFile & o, size_t need);
static void genSetCondition (OutputFile & o, uint32_t ind, const std::string & newcond);
static void genSetState (OutputFile & o, uint32_t ind, uint32_t fillIndex);
void emit_action
( const Action & action
, OutputFile & o
, uint32_t ind
, bool & readCh
, const State * const s
, const std::string & condName
, const Skeleton * skeleton
, const std::set<label_t> & used_labels
, bool save_yyaccept
)
{
switch (action.type)
{
case Action::MATCH:
emit_match (o, ind, readCh, s);
break;
case Action::INITIAL:
emit_initial (o, ind, readCh, s, * action.info.initial, used_labels);
break;
case Action::SAVE:
emit_save (o, ind, readCh, s, action.info.save, save_yyaccept);
break;
case Action::MOVE:
break;
case Action::ACCEPT:
emit_accept (o, ind, readCh, s, * action.info.accepts);
break;
case Action::RULE:
emit_rule (o, ind, s, action.info.rule, condName, skeleton);
break;
}
if (s->isPreCtxt && opts->target != opt_t::DOT)
{
o.wstring(opts->input_api.stmt_backupctx (ind));
}
}
void emit_match (OutputFile & o, uint32_t ind, bool & readCh, const State * const s)
{
if (opts->target == opt_t::DOT)
{
return;
}
const bool read_ahead = s
&& s->next
&& s->next->action.type != Action::RULE;
if (s->fill != 0)
{
o.wstring(opts->input_api.stmt_skip (ind));
}
else if (!read_ahead)
{
/* do not read next char if match */
o.wstring(opts->input_api.stmt_skip (ind));
readCh = true;
}
else
{
o.wstring(opts->input_api.stmt_skip_peek (ind));
readCh = false;
}
if (s->fill != 0)
{
need(o, ind, readCh, s->fill, false);
}
}
void emit_initial (OutputFile & o, uint32_t ind, bool & readCh, const State * const s, const Initial & initial, const std::set<label_t> & used_labels)
{
if (opts->target == opt_t::DOT)
{
return;
}
if (used_labels.count(s->label))
{
if (s->fill != 0)
{
o.wstring(opts->input_api.stmt_skip (ind));
}
else
{
o.wstring(opts->input_api.stmt_skip_peek (ind));
}
}
if (used_labels.count(initial.label))
{
o.wstring(opts->labelPrefix).wlabel(initial.label).ws(":\n");
}
if (opts->dFlag)
{
o.wind(ind).wstring(opts->yydebug).ws("(").wlabel(initial.label).ws(", *").wstring(opts->yycursor).ws(");\n");
}
if (s->fill != 0)
{
need(o, ind, readCh, s->fill, initial.setMarker);
}
else
{
if (initial.setMarker)
{
o.wstring(opts->input_api.stmt_backup (ind));
}
readCh = false;
}
}
void emit_save (OutputFile & o, uint32_t ind, bool & readCh, const State * const s, uint32_t save, bool save_yyaccept)
{
if (opts->target == opt_t::DOT)
{
return;
}
if (save_yyaccept)
{
o.wind(ind).wstring(opts->yyaccept).ws(" = ").wu32(save).ws(";\n");
}
if (s->fill != 0)
{
o.wstring(opts->input_api.stmt_skip_backup (ind));
need(o, ind, readCh, s->fill, false);
}
else
{
o.wstring(opts->input_api.stmt_skip_backup_peek (ind));
readCh = false;
}
}
void emit_accept_binary (OutputFile & o, uint32_t ind, bool & readCh, const State * const s, const accept_t & accepts, size_t l, size_t r)
{
if (l < r)
{
const size_t m = (l + r) >> 1;
o.wind(ind).ws("if (").wstring(opts->yyaccept).ws(r == l+1 ? " == " : " <= ").wu64(m).ws(") {\n");
emit_accept_binary (o, ++ind, readCh, s, accepts, l, m);
o.wind(--ind).ws("} else {\n");
emit_accept_binary (o, ++ind, readCh, s, accepts, m + 1, r);
o.wind(--ind).ws("}\n");
}
else
{
genGoTo(o, ind, s, accepts[l], readCh);
}
}
void emit_accept (OutputFile & o, uint32_t ind, bool & readCh, const State * const s, const accept_t & accepts)
{
const size_t accepts_size = accepts.size ();
if (accepts_size > 0)
{
if (opts->target != opt_t::DOT)
{
o.wstring(opts->input_api.stmt_restore (ind));
}
if (readCh) // shouldn't be necessary, but might become at some point
{
o.wstring(opts->input_api.stmt_peek (ind));
readCh = false;
}
if (accepts_size > 1)
{
if (opts->gFlag && accepts_size >= opts->cGotoThreshold)
{
o.wind(ind++).ws("{\n");
o.wind(ind++).ws("static void *").wstring(opts->yytarget).ws("[").wu64(accepts_size).ws("] = {\n");
for (uint32_t i = 0; i < accepts_size; ++i)
{
o.wind(ind).ws("&&").wstring(opts->labelPrefix).wlabel(accepts[i]->label).ws(",\n");
}
o.wind(--ind).ws("};\n");
o.wind(ind).ws("goto *").wstring(opts->yytarget).ws("[").wstring(opts->yyaccept).ws("];\n");
o.wind(--ind).ws("}\n");
}
else if (opts->sFlag || (accepts_size == 2 && opts->target != opt_t::DOT))
{
emit_accept_binary (o, ind, readCh, s, accepts, 0, accepts_size - 1);
}
else if (opts->target == opt_t::DOT)
{
for (uint32_t i = 0; i < accepts_size; ++i)
{
o.wlabel(s->label).ws(" -> ").wlabel(accepts[i]->label);
o.ws(" [label=\"yyaccept=").wu32(i).ws("\"]\n");
}
}
else
{
o.wind(ind).ws("switch (").wstring(opts->yyaccept).ws(") {\n");
for (uint32_t i = 0; i < accepts_size - 1; ++i)
{
o.wind(ind).ws("case ").wu32(i).ws(": \t");
genGoTo(o, 0, s, accepts[i], readCh);
}
o.wind(ind).ws("default:\t");
genGoTo(o, 0, s, accepts[accepts_size - 1], readCh);
o.wind(ind).ws("}\n");
}
}
else
{
// no need to write if statement here since there is only case 0.
genGoTo(o, ind, s, accepts[0], readCh);
}
}
}
void emit_rule (OutputFile & o, uint32_t ind, const State * const s, const RuleOp * const rule, const std::string & condName, const Skeleton * skeleton)
{
if (opts->target == opt_t::DOT)
{
o.wlabel(s->label);
if (rule->code)
{
o.ws(" [label=\"").wstring(rule->code->loc.filename).ws(":").wu32(rule->code->loc.line).ws("\"]");
}
o.ws("\n");
return;
}
uint32_t back = rule->ctx->fixedLength();
if (back != 0u && opts->target != opt_t::DOT)
{
o.wstring(opts->input_api.stmt_restorectx (ind));
}
if (opts->target == opt_t::SKELETON)
{
skeleton->emit_action (o, ind, rule->rank);
}
else
{
if (!rule->newcond.empty () && condName != rule->newcond)
{
genSetCondition(o, ind, rule->newcond);
}
if (rule->code)
{
if (!yySetupRule.empty ())
{
o.wind(ind).wstring(yySetupRule).ws("\n");
}
o.wline_info(rule->code->loc.line, rule->code->loc.filename.c_str ())
.wind(ind).wstring(rule->code->text).ws("\n")
.wdelay_line_info ();
}
else if (!rule->newcond.empty ())
{
o.wind(ind).wstring(replaceParam(opts->condGoto, opts->condGotoParam, opts->condPrefix + rule->newcond)).ws("\n");
}
}
}
void need (OutputFile & o, uint32_t ind, bool & readCh, size_t n, bool bSetMarker)
{
if (opts->target == opt_t::DOT)
{
return;
}
uint32_t fillIndex = last_fill_index;
if (opts->fFlag)
{
last_fill_index++;
genSetState (o, ind, fillIndex);
}
if (opts->fill_use && n > 0)
{
o.wind(ind);
if (n == 1)
{
if (opts->fill_check)
{
o.ws("if (").wstring(opts->input_api.expr_lessthan_one ()).ws(") ");
}
genYYFill(o, n);
}
else
{
if (opts->fill_check)
{
o.ws("if (").wstring(opts->input_api.expr_lessthan (n)).ws(") ");
}
genYYFill(o, n);
}
}
if (opts->fFlag)
{
o.wstring(opts->yyfilllabel).wu32(fillIndex).ws(":\n");
}
if (n > 0)
{
if (bSetMarker)
{
o.wstring(opts->input_api.stmt_backup_peek (ind));
}
else
{
o.wstring(opts->input_api.stmt_peek (ind));
}
readCh = false;
}
}
void genYYFill (OutputFile & o, size_t need)
{
o.wstring(replaceParam (opts->fill, opts->fill_arg, need));
if (!opts->fill_naked)
{
if (opts->fill_arg_use)
{
o.ws("(").wu64(need).ws(")");
}
o.ws(";");
}
o.ws("\n");
}
void genSetCondition(OutputFile & o, uint32_t ind, const std::string& newcond)
{
o.wind(ind).wstring(replaceParam (opts->cond_set, opts->cond_set_arg, opts->condEnumPrefix + newcond));
if (!opts->cond_set_naked)
{
o.ws("(").wstring(opts->condEnumPrefix).wstring(newcond).ws(");");
}
o.ws("\n");
}
void genSetState(OutputFile & o, uint32_t ind, uint32_t fillIndex)
{
o.wind(ind).wstring(replaceParam (opts->state_set, opts->state_set_arg, fillIndex));
if (!opts->state_set_naked)
{
o.ws("(").wu32(fillIndex).ws(");");
}
o.ws("\n");
}
} // namespace re2c

348
tools/re2c/src/codegen/emit_dfa.cc Normal file
View file

@ -0,0 +1,348 @@
#include "src/util/c99_stdint.h"
#include <stddef.h>
#include <set>
#include <string>
#include <utility>
#include <vector>
#include "src/codegen/bitmap.h"
#include "src/codegen/emit.h"
#include "src/codegen/go.h"
#include "src/codegen/input_api.h"
#include "src/codegen/label.h"
#include "src/codegen/output.h"
#include "src/conf/opt.h"
#include "src/globals.h"
#include "src/ir/adfa/action.h"
#include "src/ir/adfa/adfa.h"
#include "src/ir/skeleton/skeleton.h"
#include "src/util/counter.h"
namespace re2c
{
static std::string genGetCondition ();
static void genCondGotoSub (OutputFile & o, uint32_t ind, const std::vector<std::string> & condnames, uint32_t cMin, uint32_t cMax);
static void genCondTable (OutputFile & o, uint32_t ind, const std::vector<std::string> & condnames);
static void genCondGoto (OutputFile & o, uint32_t ind, const std::vector<std::string> & condnames);
static void emit_state (OutputFile & o, uint32_t ind, const State * s, bool used_label);
std::string genGetCondition()
{
return opts->cond_get_naked
? opts->cond_get
: opts->cond_get + "()";
}
void genGoTo(OutputFile & o, uint32_t ind, const State *from, const State *to, bool & readCh)
{
if (opts->target == opt_t::DOT)
{
o.wlabel(from->label).ws(" -> ").wlabel(to->label).ws("\n");
return;
}
if (readCh && from->next != to)
{
o.wstring(opts->input_api.stmt_peek (ind));
readCh = false;
}
o.wind(ind).ws("goto ").wstring(opts->labelPrefix).wlabel(to->label).ws(";\n");
}
void emit_state (OutputFile & o, uint32_t ind, const State * s, bool used_label)
{
if (opts->target != opt_t::DOT)
{
if (used_label)
{
o.wstring(opts->labelPrefix).wlabel(s->label).ws(":\n");
}
if (opts->dFlag && (s->action.type != Action::INITIAL))
{
o.wind(ind).wstring(opts->yydebug).ws("(").wlabel(s->label).ws(", ").wstring(opts->input_api.expr_peek ()).ws(");\n");
}
}
}
void DFA::count_used_labels (std::set<label_t> & used, label_t start, label_t initial, bool force_start) const
{
// In '-f' mode, default state is always state 0
if (opts->fFlag)
{
used.insert (label_t::first ());
}
if (force_start)
{
used.insert (start);
}
for (State * s = head; s; s = s->next)
{
s->go.used_labels (used);
}
for (uint32_t i = 0; i < accepts.size (); ++i)
{
used.insert (accepts[i]->label);
}
// must go last: it needs the set of used labels
if (used.count (head->label))
{
used.insert (initial);
}
}
void DFA::emit_body (OutputFile & o, uint32_t& ind, const std::set<label_t> & used_labels, label_t initial) const
{
// If DFA has transitions to initial state, then initial state
// has a piece of code that advances input position. Wee must
// skip it when entering DFA.
if (used_labels.count(head->label))
{
o.wind(ind).ws("goto ").wstring(opts->labelPrefix).wlabel(initial).ws(";\n");
}
const bool save_yyaccept = accepts.size () > 1;
for (State * s = head; s; s = s->next)
{
bool readCh = false;
emit_state (o, ind, s, used_labels.count (s->label) != 0);
emit_action (s->action, o, ind, readCh, s, cond, skeleton, used_labels, save_yyaccept);
s->go.emit(o, ind, readCh);
}
}
void DFA::emit(Output & output, uint32_t& ind, bool isLastCond, bool& bPrologBrace)
{
OutputFile & o = output.source;
bool bProlog = (!opts->cFlag || !bWroteCondCheck);
// start_label points to the beginning of current re2c block
// (prior to condition dispatch in '-c' mode)
// it can forced by configuration 're2c:startlabel = <integer>;'
label_t start_label = o.label_counter.next ();
// initial_label points to the beginning of DFA
// in '-c' mode this is NOT equal to start_label
label_t initial_label = bProlog && opts->cFlag
? o.label_counter.next ()
: start_label;
for (State * s = head; s; s = s->next)
{
s->label = o.label_counter.next ();
}
std::set<label_t> used_labels;
count_used_labels (used_labels, start_label, initial_label, o.get_force_start_label ());
head->action.set_initial (initial_label, head->action.type == Action::SAVE);
skeleton->warn_undefined_control_flow ();
skeleton->warn_unreachable_rules ();
skeleton->warn_match_empty ();
if (opts->target == opt_t::SKELETON)
{
if (output.skeletons.insert (name).second)
{
skeleton->emit_data (o.file_name);
skeleton->emit_start (o, max_fill, need_backup, need_backupctx, need_accept);
uint32_t i = 2;
emit_body (o, i, used_labels, initial_label);
skeleton->emit_end (o, need_backup, need_backupctx);
}
}
else
{
// Generate prolog
if (bProlog)
{
o.ws("\n").wdelay_line_info ();
if (opts->target == opt_t::DOT)
{
bPrologBrace = true;
o.ws("digraph re2c {\n");
}
else if ((!opts->fFlag && o.get_used_yyaccept ())
|| (!opts->fFlag && opts->bEmitYYCh)
|| (opts->bFlag && !opts->cFlag && BitMap::first)
|| (opts->cFlag && !bWroteCondCheck && opts->gFlag)
|| (opts->fFlag && !bWroteGetState && opts->gFlag)
)
{
bPrologBrace = true;
o.wind(ind++).ws("{\n");
}
else if (ind == 0)
{
ind = 1;
}
if (!opts->fFlag && opts->target != opt_t::DOT)
{
if (opts->bEmitYYCh)
{
o.wind(ind).wstring(opts->yyctype).ws(" ").wstring(opts->yych).ws(";\n");
}
o.wdelay_yyaccept_init (ind);
}
else
{
o.ws("\n");
}
}
if (opts->bFlag && !opts->cFlag && BitMap::first)
{
BitMap::gen(o, ind, lbChar, ubChar <= 256 ? ubChar : 256);
}
if (bProlog)
{
if (opts->cFlag && !bWroteCondCheck && opts->gFlag)
{
genCondTable(o, ind, output.types);
}
o.wdelay_state_goto (ind);
if (opts->cFlag && opts->target != opt_t::DOT)
{
if (used_labels.count(start_label))
{
o.wstring(opts->labelPrefix).wlabel(start_label).ws(":\n");
}
}
o.wuser_start_label ();
if (opts->cFlag && !bWroteCondCheck)
{
genCondGoto(o, ind, output.types);
}
}
if (opts->cFlag && !cond.empty())
{
if (opts->condDivider.length())
{
o.wstring(replaceParam(opts->condDivider, opts->condDividerParam, cond)).ws("\n");
}
if (opts->target == opt_t::DOT)
{
o.wstring(cond).ws(" -> ").wlabel(head->label).ws("\n");
}
else
{
o.wstring(opts->condPrefix).wstring(cond).ws(":\n");
}
}
if (opts->cFlag && opts->bFlag && BitMap::first)
{
o.wind(ind++).ws("{\n");
BitMap::gen(o, ind, lbChar, ubChar <= 256 ? ubChar : 256);
}
// Generate code
emit_body (o, ind, used_labels, initial_label);
if (opts->cFlag && opts->bFlag && BitMap::first)
{
o.wind(--ind).ws("}\n");
}
// Generate epilog
if ((!opts->cFlag || isLastCond) && bPrologBrace)
{
o.wind(--ind).ws("}\n");
}
}
// Cleanup
if (BitMap::first)
{
delete BitMap::first;
BitMap::first = NULL;
}
}
void genCondTable(OutputFile & o, uint32_t ind, const std::vector<std::string> & condnames)
{
const size_t conds = condnames.size ();
o.wind(ind++).ws("static void *").wstring(opts->yyctable).ws("[").wu64(conds).ws("] = {\n");
for (size_t i = 0; i < conds; ++i)
{
o.wind(ind).ws("&&").wstring(opts->condPrefix).wstring(condnames[i]).ws(",\n");
}
o.wind(--ind).ws("};\n");
}
void genCondGotoSub(OutputFile & o, uint32_t ind, const std::vector<std::string> & condnames, uint32_t cMin, uint32_t cMax)
{
if (cMin == cMax)
{
o.wind(ind).ws("goto ").wstring(opts->condPrefix).wstring(condnames[cMin]).ws(";\n");
}
else
{
uint32_t cMid = cMin + ((cMax - cMin + 1) / 2);
o.wind(ind).ws("if (").wstring(genGetCondition()).ws(" < ").wu32(cMid).ws(") {\n");
genCondGotoSub(o, ind + 1, condnames, cMin, cMid - 1);
o.wind(ind).ws("} else {\n");
genCondGotoSub(o, ind + 1, condnames, cMid, cMax);
o.wind(ind).ws("}\n");
}
}
/*
* note [condition order]
*
* In theory re2c makes no guarantee about the order of conditions in
* the generated lexer. Users should define condition type 'YYCONDTYPE'
* and use values of this type with 'YYGETCONDITION' and 'YYSETCONDITION'.
* This way code is independent of internal re2c condition numbering.
*
* However, it is possible to manually hardcode condition numbers and make
* re2c generate condition dispatch without explicit use of condition names
* (nested 'if' statements with '-b' or computed 'goto' table with '-g').
* This code is syntactically valid (compiles), but unsafe:
* - change of re2c options may break compilation
* - change of internal re2c condition numbering may break runtime
*
* re2c has to preserve the existing numbering scheme.
*
* re2c warns about implicit assumptions about condition order, unless:
* - condition type is defined with 'types:re2c' or '-t, --type-header'
* - dispatch is independent of condition order: either it uses
* explicit condition names or there's only one condition and
* dispatch shrinks to unconditional jump
*/
void genCondGoto(OutputFile & o, uint32_t ind, const std::vector<std::string> & condnames)
{
const size_t conds = condnames.size ();
if (opts->target == opt_t::DOT)
{
o.warn_condition_order = false; // see note [condition order]
for (size_t i = 0; i < conds; ++i)
{
const std::string cond = condnames[i];
o.ws("0 -> ").wstring(cond).ws(" [label=\"state=").wstring(cond).ws("\"]\n");
}
}
else if (opts->gFlag)
{
o.wind(ind).ws("goto *").wstring(opts->yyctable).ws("[").wstring(genGetCondition()).ws("];\n");
}
else if (opts->sFlag)
{
if (conds == 1)
{
o.warn_condition_order = false; // see note [condition order]
}
genCondGotoSub(o, ind, condnames, 0, static_cast<uint32_t> (conds) - 1);
}
else
{
o.warn_condition_order = false; // see note [condition order]
o.wind(ind).ws("switch (").wstring(genGetCondition()).ws(") {\n");
for (size_t i = 0; i < conds; ++i)
{
const std::string & cond = condnames[i];
o.wind(ind).ws("case ").wstring(opts->condEnumPrefix).wstring(cond).ws(": goto ").wstring(opts->condPrefix).wstring(cond).ws(";\n");
}
o.wind(ind).ws("}\n");
}
o.wdelay_warn_condition_order ();
bWroteCondCheck = true;
}
} // end namespace re2c

216
tools/re2c/src/codegen/go.h Normal file
View file

@ -0,0 +1,216 @@
#ifndef _RE2C_CODEGEN_GO_
#define _RE2C_CODEGEN_GO_
#include <iostream>
#include <set>
#include <vector>
#include "src/codegen/output.h"
#include "src/util/c99_stdint.h"
#include "src/util/forbid_copy.h"
namespace re2c
{
class BitMap;
struct State;
struct If;
struct Span
{
uint32_t ub;
State * to;
FORBID_COPY (Span);
};
struct Case
{
std::vector<std::pair<uint32_t, uint32_t> > ranges;
const State * to;
void emit (OutputFile & o, uint32_t ind);
inline Case ()
: ranges ()
, to (NULL)
{}
FORBID_COPY (Case);
};
struct Cases
{
const State * def;
Case * cases;
uint32_t cases_size;
void add (uint32_t lb, uint32_t ub, State * to);
Cases (const Span * s, uint32_t n);
~Cases ();
void emit (OutputFile & o, uint32_t ind, bool & readCh);
void used_labels (std::set<label_t> & used);
FORBID_COPY (Cases);
};
struct Cond
{
std::string compare;
uint32_t value;
Cond (const std::string & cmp, uint32_t val);
};
struct Binary
{
Cond * cond;
If * thn;
If * els;
Binary (const Span * s, uint32_t n, const State * next);
~Binary ();
void emit (OutputFile & o, uint32_t ind, bool & readCh);
void used_labels (std::set<label_t> & used);
FORBID_COPY (Binary);
};
struct Linear
{
std::vector<std::pair<const Cond *, const State *> > branches;
Linear (const Span * s, uint32_t n, const State * next);
~Linear ();
void emit (OutputFile & o, uint32_t ind, bool & readCh);
void used_labels (std::set<label_t> & used);
};
struct If
{
enum type_t
{
BINARY,
LINEAR
} type;
union
{
Binary * binary;
Linear * linear;
} info;
If (type_t t, const Span * sp, uint32_t nsp, const State * next);
~If ();
void emit (OutputFile & o, uint32_t ind, bool & readCh);
void used_labels (std::set<label_t> & used);
};
struct SwitchIf
{
enum
{
SWITCH,
IF
} type;
union
{
Cases * cases;
If * ifs;
} info;
SwitchIf (const Span * sp, uint32_t nsp, const State * next);
~SwitchIf ();
void emit (OutputFile & o, uint32_t ind, bool & readCh);
void used_labels (std::set<label_t> & used);
};
struct GoBitmap
{
const BitMap * bitmap;
const State * bitmap_state;
SwitchIf * hgo;
SwitchIf * lgo;
GoBitmap (const Span * span, uint32_t nSpans, const Span * hspan, uint32_t hSpans, const BitMap * bm, const State * bm_state, const State * next);
~GoBitmap ();
void emit (OutputFile & o, uint32_t ind, bool & readCh);
void used_labels (std::set<label_t> & used);
FORBID_COPY (GoBitmap);
};
struct CpgotoTable
{
static const uint32_t TABLE_SIZE;
const State ** table;
CpgotoTable (const Span * span, uint32_t nSpans);
~CpgotoTable ();
void emit (OutputFile & o, uint32_t ind);
void used_labels (std::set<label_t> & used);
private:
label_t max_label () const;
FORBID_COPY (CpgotoTable);
};
struct Cpgoto
{
SwitchIf * hgo;
CpgotoTable * table;
Cpgoto (const Span * span, uint32_t nSpans, const Span * hspan, uint32_t hSpans, const State * next);
~Cpgoto ();
void emit (OutputFile & o, uint32_t ind, bool & readCh);
void used_labels (std::set<label_t> & used);
FORBID_COPY (Cpgoto);
};
struct Dot
{
const State * from;
Cases * cases;
Dot (const Span * sp, uint32_t nsp, const State * from);
~Dot ();
void emit (OutputFile & o);
FORBID_COPY (Dot);
};
struct Go
{
uint32_t nSpans; // number of spans
Span * span;
enum
{
EMPTY,
SWITCH_IF,
BITMAP,
CPGOTO,
DOT
} type;
union
{
SwitchIf * switchif;
GoBitmap * bitmap;
Cpgoto * cpgoto;
Dot * dot;
} info;
Go ();
~Go ();
void init (const State * from);
void emit (OutputFile & o, uint32_t ind, bool & readCh);
void used_labels (std::set<label_t> & used);
Go (const Go & g)
: nSpans (g.nSpans)
, span (g.span)
, type (g.type)
, info (g.info)
{}
Go & operator = (const Go & g)
{
nSpans = g.nSpans;
span = g.span;
type = g.type;
info = g.info;
return * this;
}
};
} // namespace re2c
#endif // _RE2C_CODEGEN_GO_

284
tools/re2c/src/codegen/go_construct.cc Normal file
View file

@ -0,0 +1,284 @@
#include <stddef.h>
#include "src/util/c99_stdint.h"
#include <string>
#include <utility>
#include <vector>
#include "src/codegen/bitmap.h"
#include "src/codegen/go.h"
#include "src/conf/opt.h"
#include "src/globals.h"
#include "src/ir/adfa/adfa.h"
#include "src/util/allocate.h"
namespace re2c
{
static uint32_t unmap (Span * new_span, const Span * old_span, uint32_t old_nspans, const State * x);
Cases::Cases (const Span * span, uint32_t span_size)
: def (span_size == 0 ? NULL : span[span_size - 1].to)
, cases (new Case[span_size])
, cases_size (0)
{
for (uint32_t i = 0, lb = 0; i < span_size; ++ i)
{
add (lb, span[i].ub, span[i].to);
lb = span[i].ub;
}
}
void Cases::add (uint32_t lb, uint32_t ub, State * to)
{
for (uint32_t i = 0; i < cases_size; ++i)
{
if (cases[i].to == to)
{
cases[i].ranges.push_back (std::make_pair (lb, ub));
return;
}
}
cases[cases_size].ranges.push_back (std::make_pair (lb, ub));
cases[cases_size].to = to;
++cases_size;
}
Cond::Cond (const std::string & cmp, uint32_t val)
: compare (cmp)
, value (val)
{}
Binary::Binary (const Span * s, uint32_t n, const State * next)
: cond (NULL)
, thn (NULL)
, els (NULL)
{
const uint32_t l = n / 2;
const uint32_t h = n - l;
cond = new Cond ("<=", s[l - 1].ub - 1);
thn = new If (l > 4 ? If::BINARY : If::LINEAR, &s[0], l, next);
els = new If (h > 4 ? If::BINARY : If::LINEAR, &s[l], h, next);
}
Linear::Linear (const Span * s, uint32_t n, const State * next)
: branches ()
{
for (;;)
{
const State *bg = s[0].to;
while (n >= 3 && s[2].to == bg && (s[1].ub - s[0].ub) == 1)
{
if (s[1].to == next && n == 3)
{
branches.push_back (std::make_pair (new Cond ("!=", s[0].ub), bg));
return ;
}
else
{
branches.push_back (std::make_pair (new Cond ("==", s[0].ub), s[1].to));
}
n -= 2;
s += 2;
}
if (n == 1)
{
if (next == NULL || s[0].to != next)
{
branches.push_back (std::make_pair (static_cast<const Cond *> (NULL), s[0].to));
}
return;
}
else if (n == 2 && bg == next)
{
branches.push_back (std::make_pair (new Cond (">=", s[0].ub), s[1].to));
return;
}
else
{
branches.push_back (std::make_pair (new Cond ("<=", s[0].ub - 1), bg));
n -= 1;
s += 1;
}
}
}
If::If (type_t t, const Span * sp, uint32_t nsp, const State * next)
: type (t)
, info ()
{
switch (type)
{
case BINARY:
info.binary = new Binary (sp, nsp, next);
break;
case LINEAR:
info.linear = new Linear (sp, nsp, next);
break;
}
}
SwitchIf::SwitchIf (const Span * sp, uint32_t nsp, const State * next)
: type (IF)
, info ()
{
if ((!opts->sFlag && nsp > 2) || (nsp > 8 && (sp[nsp - 2].ub - sp[0].ub <= 3 * (nsp - 2))))
{
type = SWITCH;
info.cases = new Cases (sp, nsp);
}
else if (nsp > 5)
{
info.ifs = new If (If::BINARY, sp, nsp, next);
}
else
{
info.ifs = new If (If::LINEAR, sp, nsp, next);
}
}
GoBitmap::GoBitmap (const Span * span, uint32_t nSpans, const Span * hspan, uint32_t hSpans, const BitMap * bm, const State * bm_state, const State * next)
: bitmap (bm)
, bitmap_state (bm_state)
, hgo (NULL)
, lgo (NULL)
{
Span * bspan = allocate<Span> (nSpans);
uint32_t bSpans = unmap (bspan, span, nSpans, bm_state);
lgo = bSpans == 0
? NULL
: new SwitchIf (bspan, bSpans, next);
// if there are any low spans, then next state for high spans
// must be NULL to trigger explicit goto generation in linear 'if'
hgo = hSpans == 0
? NULL
: new SwitchIf (hspan, hSpans, lgo ? NULL : next);
operator delete (bspan);
}
const uint32_t CpgotoTable::TABLE_SIZE = 0x100;
CpgotoTable::CpgotoTable (const Span * span, uint32_t nSpans)
: table (new const State * [TABLE_SIZE])
{
uint32_t c = 0;
for (uint32_t i = 0; i < nSpans; ++i)
{
for(; c < span[i].ub && c < TABLE_SIZE; ++c)
{
table[c] = span[i].to;
}
}
}
Cpgoto::Cpgoto (const Span * span, uint32_t nSpans, const Span * hspan, uint32_t hSpans, const State * next)
: hgo (hSpans == 0 ? NULL : new SwitchIf (hspan, hSpans, next))
, table (new CpgotoTable (span, nSpans))
{}
Dot::Dot (const Span * sp, uint32_t nsp, const State * s)
: from (s)
, cases (new Cases (sp, nsp))
{}
Go::Go ()
: nSpans (0)
, span (NULL)
, type (EMPTY)
, info ()
{}
void Go::init (const State * from)
{
if (nSpans == 0)
{
return;
}
// initialize high (wide) spans
uint32_t hSpans = 0;
const Span * hspan = NULL;
for (uint32_t i = 0; i < nSpans; ++i)
{
if (span[i].ub > 0x100)
{
hspan = &span[i];
hSpans = nSpans - i;
break;
}
}
// initialize bitmaps
uint32_t nBitmaps = 0;
const BitMap * bitmap = NULL;
const State * bitmap_state = NULL;
for (uint32_t i = 0; i < nSpans; ++i)
{
if (span[i].to->isBase)
{
const BitMap *b = BitMap::find (span[i].to);
if (b && matches(b->go->span, b->go->nSpans, b->on, span, nSpans, span[i].to))
{
if (bitmap == NULL)
{
bitmap = b;
bitmap_state = span[i].to;
}
nBitmaps++;
}
}
}
const uint32_t dSpans = nSpans - hSpans - nBitmaps;
if (opts->target == opt_t::DOT)
{
type = DOT;
info.dot = new Dot (span, nSpans, from);
}
else if (opts->gFlag && (dSpans >= opts->cGotoThreshold))
{
type = CPGOTO;
info.cpgoto = new Cpgoto (span, nSpans, hspan, hSpans, from->next);
}
else if (opts->bFlag && (nBitmaps > 0))
{
type = BITMAP;
info.bitmap = new GoBitmap (span, nSpans, hspan, hSpans, bitmap, bitmap_state, from->next);
bUsedYYBitmap = true;
}
else
{
type = SWITCH_IF;
info.switchif = new SwitchIf (span, nSpans, from->next);
}
}
/*
* Find all spans, that map to the given state. For each of them,
* find upper adjacent span, that maps to another state (if such
* span exists, otherwize try lower one).
* If input contains single span that maps to the given state,
* then output contains 0 spans.
*/
uint32_t unmap (Span * new_span, const Span * old_span, uint32_t old_nspans, const State * x)
{
uint32_t new_nspans = 0;
for (uint32_t i = 0; i < old_nspans; ++i)
{
if (old_span[i].to != x)
{
if (new_nspans > 0 && new_span[new_nspans - 1].to == old_span[i].to)
new_span[new_nspans - 1].ub = old_span[i].ub;
else
{
new_span[new_nspans].to = old_span[i].to;
new_span[new_nspans].ub = old_span[i].ub;
++new_nspans;
}
}
}
if (new_nspans > 0)
new_span[new_nspans - 1].ub = old_span[old_nspans - 1].ub;
return new_nspans;
}
} // namespace re2c

99
tools/re2c/src/codegen/go_destruct.cc Normal file
View file

@ -0,0 +1,99 @@
#include "src/util/c99_stdint.h"
#include <utility>
#include <vector>
#include "src/codegen/go.h"
namespace re2c
{
Cases::~Cases ()
{
delete [] cases;
}
Binary::~Binary ()
{
delete cond;
delete thn;
delete els;
}
Linear::~Linear ()
{
for (uint32_t i = 0; i < branches.size (); ++i)
{
delete branches[i].first;
}
}
If::~If ()
{
switch (type)
{
case BINARY:
delete info.binary;
break;
case LINEAR:
delete info.linear;
break;
}
}
SwitchIf::~SwitchIf ()
{
switch (type)
{
case SWITCH:
delete info.cases;
break;
case IF:
delete info.ifs;
break;
}
}
GoBitmap::~GoBitmap ()
{
delete hgo;
delete lgo;
}
CpgotoTable::~CpgotoTable ()
{
delete [] table;
}
Cpgoto::~Cpgoto ()
{
delete hgo;
delete table;
}
Dot::~Dot ()
{
delete cases;
}
Go::~Go ()
{
switch (type)
{
case EMPTY:
break;
case SWITCH_IF:
delete info.switchif;
break;
case BITMAP:
delete info.bitmap;
break;
case CPGOTO:
delete info.cpgoto;
break;
case DOT:
delete info.dot;
break;
}
}
} // namespace re2c

271
tools/re2c/src/codegen/go_emit.cc Normal file
View file

@ -0,0 +1,271 @@
#include <stddef.h>
#include "src/util/c99_stdint.h"
#include <string>
#include <utility>
#include <vector>
#include "src/codegen/bitmap.h"
#include "src/codegen/go.h"
#include "src/codegen/input_api.h"
#include "src/codegen/label.h"
#include "src/codegen/output.h"
#include "src/codegen/print.h"
#include "src/conf/opt.h"
#include "src/globals.h"
#include "src/ir/adfa/adfa.h"
#include "src/ir/regexp/encoding/enc.h"
namespace re2c
{
static void output_if (OutputFile & o, uint32_t ind, bool & readCh, const std::string & compare, uint32_t value);
static void output_goto (OutputFile & o, uint32_t ind, bool & readCh, label_t to);
static std::string output_yych (bool & readCh);
static std::string output_hgo (OutputFile & o, uint32_t ind, bool & readCh, SwitchIf * hgo);
std::string output_yych (bool & readCh)
{
if (readCh)
{
readCh = false;
return "(" + opts->input_api.expr_peek_save () + ")";
}
else
{
return opts->yych;
}
}
void output_if (OutputFile & o, uint32_t ind, bool & readCh, const std::string & compare, uint32_t value)
{
o.wind(ind).ws("if (").wstring(output_yych (readCh)).ws(" ").wstring(compare).ws(" ").wc_hex (value).ws(") ");
}
void output_goto (OutputFile & o, uint32_t ind, bool & readCh, label_t to)
{
if (readCh)
{
o.wstring(opts->input_api.stmt_peek (ind));
readCh = false;
}
o.wind(ind).ws("goto ").wstring(opts->labelPrefix).wlabel(to).ws(";\n");
}
std::string output_hgo (OutputFile & o, uint32_t ind, bool & readCh, SwitchIf * hgo)
{
std::string yych = output_yych (readCh);
if (hgo != NULL)
{
o.wind(ind).ws("if (").wstring(yych).ws(" & ~0xFF) {\n");
hgo->emit (o, ind + 1, readCh);
o.wind(ind).ws("} else ");
yych = opts->yych;
}
else
{
o.wind(ind);
}
return yych;
}
void Case::emit (OutputFile & o, uint32_t ind)
{
for (uint32_t i = 0; i < ranges.size (); ++i)
{
for (uint32_t b = ranges[i].first; b < ranges[i].second; ++b)
{
o.wind(ind).ws("case ").wc_hex (b).ws(":");
if (opts->dFlag && opts->encoding.type () == Enc::EBCDIC)
{
const uint32_t c = opts->encoding.decodeUnsafe (b);
if (is_print (c))
o.ws(" /* ").wc(static_cast<char> (c)).ws(" */");
}
bool last_case = i == ranges.size () - 1 && b == ranges[i].second - 1;
if (!last_case)
{
o.ws("\n");
}
}
}
}
void Cases::emit (OutputFile & o, uint32_t ind, bool & readCh)
{
o.wind(ind).ws("switch (").wstring(output_yych (readCh)).ws(") {\n");
for (uint32_t i = 0; i < cases_size; ++i)
{
if (cases[i].to != def)
{
cases[i].emit (o, ind);
output_goto (o, 1, readCh, cases[i].to->label);
}
}
o.wind(ind).ws("default:");
output_goto (o, 1, readCh, def->label);
o.wind(ind).ws("}\n");
}
void Binary::emit (OutputFile & o, uint32_t ind, bool & readCh)
{
output_if (o, ind, readCh, cond->compare, cond->value);
o.ws("{\n");
thn->emit (o, ind + 1, readCh);
o.wind(ind).ws("} else {\n");
els->emit (o, ind + 1, readCh);
o.wind(ind).ws("}\n");
}
void Linear::emit (OutputFile & o, uint32_t ind, bool & readCh)
{
for (uint32_t i = 0; i < branches.size (); ++i)
{
if (branches[i].first != NULL)
{
output_if (o, ind, readCh, branches[i].first->compare, branches[i].first->value);
output_goto (o, 0, readCh, branches[i].second->label);
}
else
{
output_goto (o, ind, readCh, branches[i].second->label);
}
}
}
void If::emit (OutputFile & o, uint32_t ind, bool & readCh)
{
switch (type)
{
case BINARY:
info.binary->emit (o, ind, readCh);
break;
case LINEAR:
info.linear->emit (o, ind, readCh);
break;
}
}
void SwitchIf::emit (OutputFile & o, uint32_t ind, bool & readCh)
{
switch (type)
{
case SWITCH:
info.cases->emit (o, ind, readCh);
break;
case IF:
info.ifs->emit (o, ind, readCh);
break;
}
}
void GoBitmap::emit (OutputFile & o, uint32_t ind, bool & readCh)
{
std::string yych = output_hgo (o, ind, readCh, hgo);
o.ws("if (").wstring(opts->yybm).ws("[").wu32(bitmap->i).ws("+").wstring(yych).ws("] & ");
if (opts->yybmHexTable)
{
o.wu32_hex(bitmap->m);
}
else
{
o.wu32(bitmap->m);
}
o.ws(") {\n");
output_goto (o, ind + 1, readCh, bitmap_state->label);
o.wind(ind).ws("}\n");
if (lgo != NULL)
{
lgo->emit (o, ind, readCh);
}
}
label_t CpgotoTable::max_label () const
{
label_t max = label_t::first ();
for (uint32_t i = 0; i < TABLE_SIZE; ++i)
{
if (max < table[i]->label)
{
max = table[i]->label;
};
}
return max;
}
void CpgotoTable::emit (OutputFile & o, uint32_t ind)
{
o.wind(ind).ws("static void *").wstring(opts->yytarget).ws("[256] = {\n");
o.wind(++ind);
const uint32_t max_digits = max_label ().width ();
for (uint32_t i = 0; i < TABLE_SIZE; ++i)
{
o.ws("&&").wstring(opts->labelPrefix).wlabel(table[i]->label);
if (i == TABLE_SIZE - 1)
{
o.ws("\n");
}
else if (i % 8 == 7)
{
o.ws(",\n").wind(ind);
}
else
{
const uint32_t padding = max_digits - table[i]->label.width () + 1;
o.ws(",").wstring(std::string (padding, ' '));
}
}
o.wind(--ind).ws("};\n");
}
void Cpgoto::emit (OutputFile & o, uint32_t ind, bool & readCh)
{
std::string yych = output_hgo (o, ind, readCh, hgo);
o.ws("{\n");
table->emit (o, ++ind);
o.wind(ind).ws("goto *").wstring(opts->yytarget).ws("[").wstring(yych).ws("];\n");
o.wind(--ind).ws("}\n");
}
void Dot::emit (OutputFile & o)
{
const uint32_t n = cases->cases_size;
if (n == 1)
{
o.wlabel(from->label).ws(" -> ").wlabel(cases->cases[0].to->label).ws("\n");
}
else
{
for (uint32_t i = 0; i < n; ++i)
{
o.wlabel(from->label).ws(" -> ").wlabel(cases->cases[i].to->label).ws(" [label=\"");
for (uint32_t j = 0; j < cases->cases[i].ranges.size (); ++j)
{
o.wrange(cases->cases[i].ranges[j].first, cases->cases[i].ranges[j].second);
}
o.ws("\"]\n");
}
}
}
void Go::emit (OutputFile & o, uint32_t ind, bool & readCh)
{
switch (type)
{
case EMPTY:
break;
case SWITCH_IF:
info.switchif->emit (o, ind, readCh);
break;
case BITMAP:
info.bitmap->emit (o, ind, readCh);
break;
case CPGOTO:
info.cpgoto->emit (o, ind, readCh);
break;
case DOT:
info.dot->emit (o);
break;
}
}
} // namespace re2c

111
tools/re2c/src/codegen/go_used_labels.cc Normal file
View file

@ -0,0 +1,111 @@
#include <stddef.h>
#include "src/util/c99_stdint.h"
#include <set>
#include <utility>
#include <vector>
#include "src/codegen/go.h"
#include "src/codegen/label.h"
#include "src/ir/adfa/adfa.h"
namespace re2c
{
void Cases::used_labels (std::set<label_t> & used)
{
for (uint32_t i = 0; i < cases_size; ++i)
{
used.insert (cases[i].to->label);
}
}
void Binary::used_labels (std::set<label_t> & used)
{
thn->used_labels (used);
els->used_labels (used);
}
void Linear::used_labels (std::set<label_t> & used)
{
for (uint32_t i = 0; i < branches.size (); ++i)
{
used.insert (branches[i].second->label);
}
}
void If::used_labels (std::set<label_t> & used)
{
switch (type)
{
case BINARY:
info.binary->used_labels (used);
break;
case LINEAR:
info.linear->used_labels (used);
break;
}
}
void SwitchIf::used_labels (std::set<label_t> & used)
{
switch (type)
{
case SWITCH:
info.cases->used_labels (used);
break;
case IF:
info.ifs->used_labels (used);
break;
}
}
void GoBitmap::used_labels (std::set<label_t> & used)
{
if (hgo != NULL)
{
hgo->used_labels (used);
}
used.insert (bitmap_state->label);
if (lgo != NULL)
{
lgo->used_labels (used);
}
}
void CpgotoTable::used_labels (std::set<label_t> & used)
{
for (uint32_t i = 0; i < TABLE_SIZE; ++i)
{
used.insert (table[i]->label);
}
}
void Cpgoto::used_labels (std::set<label_t> & used)
{
if (hgo != NULL)
{
hgo->used_labels (used);
}
table->used_labels (used);
}
void Go::used_labels (std::set<label_t> & used)
{
switch (type)
{
case EMPTY:
case DOT:
break;
case SWITCH_IF:
info.switchif->used_labels (used);
break;
case BITMAP:
info.bitmap->used_labels (used);
break;
case CPGOTO:
info.cpgoto->used_labels (used);
break;
}
}
} // namespace re2c

24
tools/re2c/src/codegen/indent.h Normal file
View file

@ -0,0 +1,24 @@
#ifndef _RE2C_CODEGEN_INDENT_
#define _RE2C_CODEGEN_INDENT_
#include <string>
#include "src/globals.h"
namespace re2c
{
inline std::string indent (uint32_t ind)
{
std::string str;
while (opts->target != opt_t::DOT && ind-- > 0)
{
str += opts->indString;
}
return str;
}
} // end namespace re2c
#endif // _RE2C_CODEGEN_INDENT_

175
tools/re2c/src/codegen/input_api.cc Normal file
View file

@ -0,0 +1,175 @@
#include <sstream>
#include "src/codegen/input_api.h"
#include "src/codegen/indent.h"
#include "src/conf/opt.h"
#include "src/globals.h"
namespace re2c
{
InputAPI::InputAPI ()
: type_ (DEFAULT)
{}
InputAPI::type_t InputAPI::type () const
{
return type_;
}
void InputAPI::set (type_t t)
{
type_ = t;
}
std::string InputAPI::expr_peek () const
{
std::string s;
switch (type_)
{
case DEFAULT:
s = "*" + opts->yycursor;
break;
case CUSTOM:
s = opts->yypeek + " ()";
break;
}
return s;
}
std::string InputAPI::expr_peek_save () const
{
return opts->yych + " = " + opts.yychConversion () + expr_peek ();
}
std::string InputAPI::stmt_peek (uint32_t ind) const
{
return indent (ind) + expr_peek_save () + ";\n";
}
std::string InputAPI::stmt_skip (uint32_t ind) const
{
std::string s;
switch (type_)
{
case DEFAULT:
s = "++" + opts->yycursor;
break;
case CUSTOM:
s = opts->yyskip + " ()";
break;
}
return indent (ind) + s + ";\n";
}
std::string InputAPI::stmt_backup (uint32_t ind) const
{
std::string s;
switch (type_)
{
case DEFAULT:
s = opts->yymarker + " = " + opts->yycursor;
break;
case CUSTOM:
s = opts->yybackup + " ()";
break;
}
return indent (ind) + s + ";\n";
}
std::string InputAPI::stmt_backupctx (uint32_t ind) const
{
std::string s;
switch (type_)
{
case DEFAULT:
s = opts->yyctxmarker + " = " + opts->yycursor;
break;
case CUSTOM:
s = opts->yybackupctx + " ()";
break;
}
return indent (ind) + s + ";\n";
}
std::string InputAPI::stmt_restore (uint32_t ind) const
{
std::string s;
switch (type_)
{
case DEFAULT:
s = opts->yycursor + " = " + opts->yymarker;
break;
case CUSTOM:
s = opts->yyrestore + " ()";
break;
}
return indent (ind) + s + ";\n";
}
std::string InputAPI::stmt_restorectx (uint32_t ind) const
{
std::string s;
switch (type_)
{
case DEFAULT:
s = indent (ind) + opts->yycursor + " = " + opts->yyctxmarker + ";\n";
break;
case CUSTOM:
s = indent (ind) + opts->yyrestorectx + " ();\n";
break;
}
return s;
}
std::string InputAPI::stmt_skip_peek (uint32_t ind) const
{
return type_ == DEFAULT
? indent (ind) + opts->yych + " = " + opts.yychConversion () + "*++" + opts->yycursor + ";\n"
: stmt_skip (ind) + stmt_peek (ind);
}
std::string InputAPI::stmt_skip_backup (uint32_t ind) const
{
return type_ == DEFAULT
? indent (ind) + opts->yymarker + " = ++" + opts->yycursor + ";\n"
: stmt_skip (ind) + stmt_backup (ind);
}
std::string InputAPI::stmt_backup_peek (uint32_t ind) const
{
return type_ == DEFAULT
? indent (ind) + opts->yych + " = " + opts.yychConversion () + "*(" + opts->yymarker + " = " + opts->yycursor + ");\n"
: stmt_backup (ind) + stmt_peek (ind);
}
std::string InputAPI::stmt_skip_backup_peek (uint32_t ind) const
{
return type_ == DEFAULT
? indent (ind) + opts->yych + " = " + opts.yychConversion () + "*(" + opts->yymarker + " = ++" + opts->yycursor + ");\n"
: stmt_skip (ind) + stmt_backup (ind) + stmt_peek (ind);
}
std::string InputAPI::expr_lessthan_one () const
{
return type_ == DEFAULT
? opts->yylimit + " <= " + opts->yycursor
: expr_lessthan (1);
}
std::string InputAPI::expr_lessthan (size_t n) const
{
std::ostringstream s;
switch (type_)
{
case DEFAULT:
s << "(" << opts->yylimit << " - " << opts->yycursor << ") < " << n;
break;
case CUSTOM:
s << opts->yylessthan << " (" << n << ")";
break;
}
return s.str ();
}
} // end namespace re2c

43
tools/re2c/src/codegen/input_api.h Normal file
View file

@ -0,0 +1,43 @@
#ifndef _RE2C_CODEGEN_INPUT_API_
#define _RE2C_CODEGEN_INPUT_API_
#include "src/util/c99_stdint.h"
#include <string>
namespace re2c
{
class InputAPI
{
public:
enum type_t
{ DEFAULT
, CUSTOM
};
private:
type_t type_;
public:
InputAPI ();
type_t type () const;
void set (type_t t);
std::string expr_peek () const;
std::string expr_peek_save () const;
std::string stmt_peek (uint32_t ind) const;
std::string stmt_skip (uint32_t ind) const;
std::string stmt_backup (uint32_t ind) const;
std::string stmt_backupctx (uint32_t ind) const;
std::string stmt_restore (uint32_t ind) const;
std::string stmt_restorectx (uint32_t ind) const;
std::string stmt_skip_peek (uint32_t ind) const;
std::string stmt_skip_backup (uint32_t ind) const;
std::string stmt_backup_peek (uint32_t ind) const;
std::string stmt_skip_backup_peek (uint32_t ind) const;
std::string expr_lessthan_one () const;
std::string expr_lessthan (size_t n) const;
};
} // end namespace re2c
#endif // _RE2C_CODEGEN_INPUT_API_

42
tools/re2c/src/codegen/label.cc Normal file
View file

@ -0,0 +1,42 @@
#include <ostream>
#include "src/codegen/label.h"
namespace re2c {
const uint32_t label_t::FIRST = 0;
label_t::label_t ()
: value (FIRST)
{}
void label_t::inc ()
{
++value;
}
label_t label_t::first ()
{
return label_t ();
}
bool label_t::operator < (const label_t & l) const
{
return value < l.value;
}
uint32_t label_t::width () const
{
uint32_t v = value;
uint32_t n = 0;
while (v /= 10) ++n;
return n;
}
std::ostream & operator << (std::ostream & o, label_t l)
{
o << l.value;
return o;
}
} // namespace re2c

39
tools/re2c/src/codegen/label.h Normal file
View file

@ -0,0 +1,39 @@
#ifndef _RE2C_CODEGEN_LABEL_
#define _RE2C_CODEGEN_LABEL_
#include <iosfwd> // ostream
#include "src/util/c99_stdint.h"
namespace re2c {
template <typename num_t> class counter_t;
// label public API:
// - get first label
// - compare labels
// - get label width
// - output label to std::ostream
//
// label private API (for label counter):
// - get initial label
// - get next label
class label_t
{
static const uint32_t FIRST;
uint32_t value;
label_t ();
void inc ();
public:
static label_t first ();
bool operator < (const label_t & l) const;
uint32_t width () const;
friend std::ostream & operator << (std::ostream & o, label_t l);
friend class counter_t<label_t>;
};
} // namespace re2c
#endif // _RE2C_CODEGEN_LABEL_

465
tools/re2c/src/codegen/output.cc Normal file
View file

@ -0,0 +1,465 @@
#include <stdio.h>
#include <time.h>
#include <iomanip>
#include "src/codegen/indent.h"
#include "src/codegen/output.h"
#include "src/codegen/print.h"
#include "src/conf/opt.h"
#include "src/conf/warn.h"
#include "src/globals.h"
#include "src/ir/rule_rank.h"
namespace re2c
{
OutputFragment::OutputFragment (type_t t, uint32_t i)
: type (t)
, stream ()
, indent (i)
{}
uint32_t OutputFragment::count_lines ()
{
uint32_t lines = 0;
const std::string content = stream.str ();
const char * p = content.c_str ();
for (uint32_t i = 0; i < content.size (); ++i)
{
if (p[i] == '\n')
{
++lines;
}
}
return lines;
}
OutputBlock::OutputBlock ()
: fragments ()
, used_yyaccept (false)
, force_start_label (false)
, user_start_label ()
, line (0)
{
fragments.push_back (new OutputFragment (OutputFragment::CODE, 0));
}
OutputBlock::~OutputBlock ()
{
for (unsigned int i = 0; i < fragments.size (); ++i)
{
delete fragments[i];
}
}
OutputFile::OutputFile (const char * fn)
: file_name (fn)
, file (NULL)
, blocks ()
, label_counter ()
, warn_condition_order (!opts->tFlag) // see note [condition order]
{
new_block ();
}
bool OutputFile::open ()
{
if (file_name == NULL)
{
file_name = "<stdout>";
file = stdout;
}
else
{
file = fopen (file_name, "wb");
}
return file != NULL;
}
OutputFile::~OutputFile ()
{
if (file != NULL && file != stdout)
{
fclose (file);
}
for (unsigned int i = 0; i < blocks.size (); ++i)
{
delete blocks[i];
}
}
std::ostream & OutputFile::stream ()
{
return blocks.back ()->fragments.back ()->stream;
}
OutputFile & OutputFile::wraw (const char * s, size_t n)
{
stream ().write (s, static_cast<std::streamsize> (n));
return *this;
}
OutputFile & OutputFile::wu32_hex (uint32_t n)
{
prtHex (stream (), n);
return *this;
}
OutputFile & OutputFile::wc_hex (uint32_t n)
{
prtChOrHex (stream (), n);
return *this;
}
OutputFile & OutputFile::wrange (uint32_t l, uint32_t u)
{
printSpan (stream (), l, u);
return *this;
}
OutputFile & OutputFile::wu32_width (uint32_t n, int w)
{
stream () << std::setw (w);
stream () << n;
return *this;
}
OutputFile & OutputFile::wline_info (uint32_t l, const char * fn)
{
output_line_info (stream (), l, fn);
return *this;
}
OutputFile & OutputFile::wversion_time ()
{
output_version_time (stream ());
return *this;
}
OutputFile & OutputFile::wuser_start_label ()
{
const std::string label = blocks.back ()->user_start_label;
if (!label.empty ())
{
wstring(label).ws(":\n");
}
return *this;
}
OutputFile & OutputFile::wc (char c)
{
stream () << c;
return *this;
}
OutputFile & OutputFile::wu32 (uint32_t n)
{
stream () << n;
return *this;
}
OutputFile & OutputFile::wu64 (uint64_t n)
{
stream () << n;
return *this;
}
OutputFile & OutputFile::wstring (const std::string & s)
{
stream () << s;
return *this;
}
OutputFile & OutputFile::ws (const char * s)
{
stream () << s;
return *this;
}
OutputFile & OutputFile::wlabel (label_t l)
{
stream () << l;
return *this;
}
OutputFile & OutputFile::wrank (rule_rank_t r)
{
stream () << r;
return *this;
}
OutputFile & OutputFile::wind (uint32_t ind)
{
stream () << indent(ind);
return *this;
}
void OutputFile::insert_code ()
{
blocks.back ()->fragments.push_back (new OutputFragment (OutputFragment::CODE, 0));
}
OutputFile & OutputFile::wdelay_line_info ()
{
blocks.back ()->fragments.push_back (new OutputFragment (OutputFragment::LINE_INFO, 0));
insert_code ();
return *this;
}
OutputFile & OutputFile::wdelay_state_goto (uint32_t ind)
{
if (opts->fFlag && !bWroteGetState)
{
blocks.back ()->fragments.push_back (new OutputFragment (OutputFragment::STATE_GOTO, ind));
insert_code ();
bWroteGetState = true;
}
return *this;
}
OutputFile & OutputFile::wdelay_types ()
{
warn_condition_order = false; // see note [condition order]
blocks.back ()->fragments.push_back (new OutputFragment (OutputFragment::TYPES, 0));
insert_code ();
return *this;
}
OutputFile & OutputFile::wdelay_warn_condition_order ()
{
blocks.back ()->fragments.push_back (new OutputFragment (OutputFragment::WARN_CONDITION_ORDER, 0));
insert_code ();
return *this;
}
OutputFile & OutputFile::wdelay_yyaccept_init (uint32_t ind)
{
blocks.back ()->fragments.push_back (new OutputFragment (OutputFragment::YYACCEPT_INIT, ind));
insert_code ();
return *this;
}
OutputFile & OutputFile::wdelay_yymaxfill ()
{
blocks.back ()->fragments.push_back (new OutputFragment (OutputFragment::YYMAXFILL, 0));
insert_code ();
return *this;
}
void OutputFile::set_used_yyaccept ()
{
blocks.back ()->used_yyaccept = true;
}
bool OutputFile::get_used_yyaccept () const
{
return blocks.back ()->used_yyaccept;
}
void OutputFile::set_force_start_label (bool force)
{
blocks.back ()->force_start_label = force;
}
void OutputFile::set_user_start_label (const std::string & label)
{
blocks.back ()->user_start_label = label;
}
bool OutputFile::get_force_start_label () const
{
return blocks.back ()->force_start_label;
}
void OutputFile::set_block_line (uint32_t l)
{
blocks.back ()->line = l;
}
uint32_t OutputFile::get_block_line () const
{
return blocks.back ()->line;
}
void OutputFile::new_block ()
{
blocks.push_back (new OutputBlock ());
insert_code ();
}
void OutputFile::emit
( const std::vector<std::string> & types
, size_t max_fill
)
{
if (file != NULL)
{
unsigned int line_count = 1;
for (unsigned int j = 0; j < blocks.size (); ++j)
{
OutputBlock & b = * blocks[j];
for (unsigned int i = 0; i < b.fragments.size (); ++i)
{
OutputFragment & f = * b.fragments[i];
switch (f.type)
{
case OutputFragment::CODE:
break;
case OutputFragment::LINE_INFO:
output_line_info (f.stream, line_count + 1, file_name);
break;
case OutputFragment::STATE_GOTO:
output_state_goto (f.stream, f.indent, 0);
break;
case OutputFragment::TYPES:
output_types (f.stream, f.indent, types);
break;
case OutputFragment::WARN_CONDITION_ORDER:
if (warn_condition_order) // see note [condition order]
{
warn.condition_order (b.line);
}
break;
case OutputFragment::YYACCEPT_INIT:
output_yyaccept_init (f.stream, f.indent, b.used_yyaccept);
break;
case OutputFragment::YYMAXFILL:
output_yymaxfill (f.stream, max_fill);
break;
}
std::string content = f.stream.str ();
fwrite (content.c_str (), 1, content.size (), file);
line_count += f.count_lines ();
}
}
}
}
HeaderFile::HeaderFile (const char * fn)
: stream ()
// header is always generated, but not always dumped to file
// NULL filename crashes 'operator <<' on some platforms
// TODO: generate header only if necessary
, file_name (fn ? fn : "<stdout>.h")
, file (NULL)
{}
bool HeaderFile::open ()
{
file = fopen (file_name, "wb");
return file != NULL;
}
void HeaderFile::emit (const std::vector<std::string> & types)
{
output_version_time (stream);
output_line_info (stream, 3, file_name);
stream << "\n";
output_types (stream, 0, types);
}
HeaderFile::~HeaderFile ()
{
if (file != NULL)
{
std::string content = stream.str ();
fwrite (content.c_str (), 1, content.size (), file);
fclose (file);
}
}
Output::Output (const char * source_name, const char * header_name)
: source (source_name)
, header (header_name)
, types ()
, skeletons ()
, max_fill (1)
{}
Output::~Output ()
{
if (!warn.error ())
{
source.emit (types, max_fill);
header.emit (types);
}
}
void output_state_goto (std::ostream & o, uint32_t ind, uint32_t start_label)
{
o << indent(ind) << "switch (" << output_get_state() << ") {\n";
if (opts->bUseStateAbort)
{
o << indent(ind) << "default: abort();\n";
o << indent(ind) << "case -1: goto " << opts->labelPrefix << start_label << ";\n";
}
else
{
o << indent(ind) << "default: goto " << opts->labelPrefix << start_label << ";\n";
}
for (uint32_t i = 0; i < last_fill_index; ++i)
{
o << indent(ind) << "case " << i << ": goto " << opts->yyfilllabel << i << ";\n";
}
o << indent(ind) << "}\n";
if (opts->bUseStateNext)
{
o << opts->yynext << ":\n";
}
}
void output_yyaccept_init (std::ostream & o, uint32_t ind, bool used_yyaccept)
{
if (used_yyaccept)
{
o << indent (ind) << "unsigned int " << opts->yyaccept << " = 0;\n";
}
}
void output_yymaxfill (std::ostream & o, size_t max_fill)
{
o << "#define YYMAXFILL " << max_fill << "\n";
}
void output_line_info (std::ostream & o, uint32_t line_number, const char * file_name)
{
if (!opts->iFlag)
{
o << "#line " << line_number << " \"" << file_name << "\"\n";
}
}
void output_types (std::ostream & o, uint32_t ind, const std::vector<std::string> & types)
{
o << indent (ind++) << "enum " << opts->yycondtype << " {\n";
for (unsigned int i = 0; i < types.size (); ++i)
{
o << indent (ind) << opts->condEnumPrefix << types[i] << ",\n";
}
o << indent (--ind) << "};\n";
}
void output_version_time (std::ostream & o)
{
o << "/* Generated by re2c";
if (opts->version)
{
o << " " << PACKAGE_VERSION;
}
if (!opts->bNoGenerationDate)
{
o << " on ";
time_t now = time (NULL);
o.write (ctime (&now), 24);
}
o << " */" << "\n";
}
std::string output_get_state ()
{
return opts->state_get_naked
? opts->state_get
: opts->state_get + "()";
}
} // namespace re2c

158
tools/re2c/src/codegen/output.h Normal file
View file

@ -0,0 +1,158 @@
#ifndef _RE2C_CODEGEN_OUTPUT_
#define _RE2C_CODEGEN_OUTPUT_
#include "src/util/c99_stdint.h"
#include <stddef.h>
#include <stdio.h>
#include <fstream>
#include <set>
#include <sstream>
#include <string>
#include <vector>
#include "src/codegen/label.h"
#include "src/util/counter.h"
#include "src/util/forbid_copy.h"
namespace re2c
{
class rule_rank_t;
struct OutputFragment
{
enum type_t
{ CODE
// , CONFIG
, LINE_INFO
, STATE_GOTO
, TYPES
, WARN_CONDITION_ORDER
, YYACCEPT_INIT
, YYMAXFILL
};
type_t type;
std::ostringstream stream;
uint32_t indent;
OutputFragment (type_t t, uint32_t i);
uint32_t count_lines ();
};
struct OutputBlock
{
std::vector<OutputFragment *> fragments;
bool used_yyaccept;
bool force_start_label;
std::string user_start_label;
uint32_t line;
OutputBlock ();
~OutputBlock ();
};
struct OutputFile
{
public:
const char * file_name;
private:
FILE * file;
std::vector<OutputBlock *> blocks;
public:
counter_t<label_t> label_counter;
bool warn_condition_order;
private:
std::ostream & stream ();
void insert_code ();
public:
OutputFile (const char * fn);
~OutputFile ();
bool open ();
void new_block ();
// immediate output
OutputFile & wraw (const char * s, size_t n);
OutputFile & wc (char c);
OutputFile & wc_hex (uint32_t n);
OutputFile & wu32 (uint32_t n);
OutputFile & wu32_hex (uint32_t n);
OutputFile & wu32_width (uint32_t n, int w);
OutputFile & wu64 (uint64_t n);
OutputFile & wstring (const std::string & s);
OutputFile & ws (const char * s);
OutputFile & wlabel (label_t l);
OutputFile & wrank (rule_rank_t l);
OutputFile & wrange (uint32_t u, uint32_t l);
OutputFile & wline_info (uint32_t l, const char * fn);
OutputFile & wversion_time ();
OutputFile & wuser_start_label ();
OutputFile & wind (uint32_t ind);
// delayed output
OutputFile & wdelay_line_info ();
OutputFile & wdelay_state_goto (uint32_t ind);
OutputFile & wdelay_types ();
OutputFile & wdelay_warn_condition_order ();
OutputFile & wdelay_yyaccept_init (uint32_t ind);
OutputFile & wdelay_yymaxfill ();
void set_used_yyaccept ();
bool get_used_yyaccept () const;
void set_force_start_label (bool force);
void set_user_start_label (const std::string & label);
bool get_force_start_label () const;
void set_block_line (uint32_t l);
uint32_t get_block_line () const;
void emit (const std::vector<std::string> & types, size_t max_fill);
FORBID_COPY (OutputFile);
};
struct HeaderFile
{
HeaderFile (const char * fn);
~HeaderFile ();
bool open ();
void emit (const std::vector<std::string> & types);
private:
std::ostringstream stream;
const char * file_name;
FILE * file;
FORBID_COPY (HeaderFile);
};
struct Output
{
OutputFile source;
HeaderFile header;
std::vector<std::string> types;
std::set<std::string> skeletons;
size_t max_fill;
Output (const char * source_name, const char * header_name);
~Output ();
};
void output_line_info (std::ostream &, uint32_t, const char *);
void output_state_goto (std::ostream &, uint32_t, uint32_t);
void output_types (std::ostream &, uint32_t, const std::vector<std::string> &);
void output_version_time (std::ostream &);
void output_yyaccept_init (std::ostream &, uint32_t, bool);
void output_yymaxfill (std::ostream &, size_t);
// helpers
std::string output_get_state ();
} // namespace re2c
#endif // _RE2C_CODEGEN_OUTPUT_

156
tools/re2c/src/codegen/print.cc Normal file
View file

@ -0,0 +1,156 @@
#include <iostream>
#include "src/codegen/print.h"
#include "src/conf/opt.h"
#include "src/globals.h"
#include "src/ir/regexp/encoding/enc.h"
namespace re2c
{
bool is_print (uint32_t c)
{
return c >= 0x20 && c < 0x7F;
}
bool is_space (uint32_t c)
{
switch (c)
{
case '\t':
case '\f':
case '\v':
case '\n':
case '\r':
case ' ':
return true;
default:
return false;
}
}
char hexCh(uint32_t c)
{
static const char * sHex = "0123456789ABCDEF";
return sHex[c & 0x0F];
}
void prtChOrHex(std::ostream& o, uint32_t c)
{
if (opts->encoding.type () != Enc::EBCDIC
&& (is_print (c) || is_space (c)))
{
o << '\'';
prtCh(o, c);
o << '\'';
}
else
{
prtHex(o, c);
}
}
void prtHex(std::ostream& o, uint32_t c)
{
o << "0x";
const uint32_t cunit_size = opts->encoding.szCodeUnit ();
if (cunit_size >= 4)
{
o << hexCh (c >> 28u)
<< hexCh (c >> 24u)
<< hexCh (c >> 20u)
<< hexCh (c >> 16u);
}
if (cunit_size >= 2)
{
o << hexCh (c >> 12u)
<< hexCh (c >> 8u);
}
o << hexCh (c >> 4u)
<< hexCh (c);
}
void prtCh(std::ostream& o, uint32_t c)
{
const bool dot = opts->target == opt_t::DOT;
switch (c)
{
case '\'':
o << (dot ? "'" : "\\'");
break;
case '"':
o << (dot ? "\\\"" : "\"");
break;
case '\n':
o << (dot ? "\\\\n" : "\\n");
break;
case '\t':
o << (dot ? "\\\\t" : "\\t");
break;
case '\v':
o << (dot ? "\\\\v" : "\\v");
break;
case '\b':
o << (dot ? "\\\\b" : "\\b");
break;
case '\r':
o << (dot ? "\\\\r" : "\\r");
break;
case '\f':
o << (dot ? "\\\\f" : "\\f");
break;
case '\a':
o << (dot ? "\\\\a" :"\\a");
break;
case '\\':
o << "\\\\"; // both .dot and C/C++ code expect "\\"
break;
default:
o << static_cast<char> (c);
break;
}
}
void prtChOrHexForSpan(std::ostream& o, uint32_t c)
{
if (opts->encoding.type () != Enc::EBCDIC
&& is_print (c)
&& (c != ']'))
{
prtCh(o, c);
}
else
{
prtHex(o, c);
}
}
void printSpan(std::ostream& o, uint32_t lb, uint32_t ub)
{
o << "[";
if ((ub - lb) == 1)
{
prtChOrHexForSpan(o, lb);
}
else
{
prtChOrHexForSpan(o, lb);
o << "-";
prtChOrHexForSpan(o, ub - 1);
}
o << "]";
}
} // end namespace re2c

20
tools/re2c/src/codegen/print.h Normal file
View file

@ -0,0 +1,20 @@
#ifndef _RE2C_CODEGEN_PRINT_
#define _RE2C_CODEGEN_PRINT_
#include "src/util/c99_stdint.h"
#include <iosfwd>
namespace re2c
{
bool is_print (uint32_t c);
bool is_space (uint32_t c);
char hexCh(uint32_t c);
void prtCh(std::ostream&, uint32_t);
void prtHex(std::ostream&, uint32_t);
void prtChOrHex(std::ostream&, uint32_t);
void printSpan(std::ostream&, uint32_t, uint32_t);
} // end namespace re2c
#endif // _RE2C_CODEGEN_PRINT_

258
tools/re2c/src/conf/msg.cc Normal file
View file

@ -0,0 +1,258 @@
#include <stdarg.h>
#include <stdio.h>
#include <string>
#if defined(_MSC_VER) && _MSC_VER < 1500
#include "config.msc.h"
#else
#include "config.h"
#endif
#include "src/conf/msg.h"
namespace re2c {
void error (const char * fmt, ...)
{
fprintf (stderr, "re2c: error: ");
va_list args;
va_start (args, fmt);
vfprintf (stderr, fmt, args);
va_end (args);
fprintf (stderr, "\n");
}
void error_encoding ()
{
error ("only one of switches -e, -w, -x, -u and -8 must be set");
}
void error_arg (const char * option)
{
error ("expected argument to option %s", option);
}
void warning_start (uint32_t line, bool error)
{
static const char * msg = error ? "error" : "warning";
fprintf (stderr, "re2c: %s: line %u: ", msg, line);
}
void warning_end (const char * type, bool error)
{
if (type != NULL)
{
const char * prefix = error ? "error-" : "";
fprintf (stderr, " [-W%s%s]", prefix, type);
}
fprintf (stderr, "\n");
}
void warning (const char * type, uint32_t line, bool error, const char * fmt, ...)
{
warning_start (line, error);
va_list args;
va_start (args, fmt);
vfprintf (stderr, fmt, args);
va_end (args);
warning_end (type, error);
}
void usage ()
{
fprintf (stderr,
"usage: re2c [-bcdDefFghirsuvVwx18] [-o of] [-t th] file\n"
"\n"
"-? -h --help Display this info.\n"
"\n"
"-b --bit-vectors Implies -s. Use bit vectors as well in the attempt to\n"
" coax better code out of the compiler. Most useful for\n"
" specifications with more than a few keywords (e.g. for\n"
" most programming languages).\n"
"\n"
"-c --conditions Require start conditions.\n"
"\n"
"-d --debug-output Creates a parser that dumps information during\n"
" about the current position and in which state the\n"
" parser is.\n"
"\n"
"-D --emit-dot Emit a Graphviz dot view of the DFA graph\n"
"\n"
"-e --ecb Generate a parser that supports EBCDIC. The generated code\n"
" can deal with any character up to 0xFF. In this mode re2c\n"
" assumes that input character size is 1 byte. This switch is\n"
" incompatible with -w, -u, -x and -8\n"
"\n"
"-f --storable-state Generate a scanner that supports storable states.\n"
"\n"
"-F --flex-syntax Partial support for flex syntax.\n"
"\n"
"-g --computed-gotos Implies -b. Generate computed goto code (only useable\n"
" with gcc).\n"
"\n"
"-i --no-debug-info Do not generate '#line' info (useful for versioning).\n"
"\n"
"-o of --output=of Specify the output file (of) instead of stdout\n"
"\n"
"-r --reusable Allow reuse of scanner definitions.\n"
"\n"
"-s --nested-ifs Generate nested ifs for some switches. Many compilers\n"
" need this assist to generate better code.\n"
"\n"
"-t th --type-header=th Generate a type header file (th) with type definitions.\n"
"\n"
"-u --unicode Generate a parser that supports UTF-32. The generated code\n"
" can deal with any valid Unicode character up to 0x10FFFF.\n"
" In this mode re2c assumes that input character size is 4 bytes.\n"
" This switch is incompatible with -e, -w, -x and -8. It implies -s.\n"
"\n"
"-v --version Show version information.\n"
"\n"
"-V --vernum Show version as one number.\n"
"\n"
"-w --wide-chars Generate a parser that supports UCS-2. The generated code can\n"
" deal with any valid Unicode character up to 0xFFFF. In this mode\n"
" re2c assumes that input character size is 2 bytes. This switch is\n"
" incompatible with -e, -x, -u and -8. It implies -s."
"\n"
"-x --utf-16 Generate a parser that supports UTF-16. The generated code can\n"
" deal with any valid Unicode character up to 0x10FFFF. In this mode\n"
" re2c assumes that input character size is 2 bytes. This switch is\n"
" incompatible with -e, -w, -u and -8. It implies -s."
"\n"
"-8 --utf-8 Generate a parser that supports UTF-8. The generated code can\n"
" deal with any valid Unicode character up to 0x10FFFF. In this mode\n"
" re2c assumes that input character size is 1 byte. This switch is\n"
" incompatible with -e, -w, -x and -u."
"\n"
"--no-generation-date Suppress date output in the generated file.\n"
"\n"
"--no-version Suppress version output in the generated file.\n"
"\n"
"--case-insensitive All strings are case insensitive, so all \"-expressions\n"
" are treated in the same way '-expressions are.\n"
"\n"
"--case-inverted Invert the meaning of single and double quoted strings.\n"
" With this switch single quotes are case sensitive and\n"
" double quotes are case insensitive.\n"
"\n"
"--encoding-policy ep Specify what re2c should do when given bad code unit.\n"
" ep can be one of the following: fail, substitute, ignore.\n"
"\n"
"--input i Specify re2c input API.\n"
" i can be one of the following: default, custom.\n"
"\n"
"--skeleton Instead of embedding re2c-generated code into C/C++ source,\n"
" generate a self-contained program for the same DFA.\n"
" Most useful for correctness and performance testing.\n"
"\n"
"--empty-class policy What to do if user inputs empty character class. policy can be\n"
" one of the following: 'match-empty' (match empty input, default),\n"
" 'match-none' (fail to match on any input), 'error' (compilation\n"
" error). Note that there are various ways to construct empty class,\n"
" e.g: [], [^\\x00-\\xFF], [\\x00-\\xFF]\\[\\x00-\\xFF].\n"
"\n"
"--dfa-minimization <table | moore>\n"
" Internal algorithm used by re2c to minimize DFA (defaults to\n"
" 'moore'). Both table filling and Moore's algorithms should\n"
" produce identical DFA (up to states relabelling). Table filling\n"
" algorithm is much simpler and slower; it serves as a reference\n"
" implementation.\n"
"\n"
"-1 --single-pass Deprecated and does nothing (single pass is by default now).\n"
"\n"
"-W Turn on all warnings.\n"
"\n"
"-Werror Turn warnings into errors. Note that this option along doesn't\n"
" turn on any warnings, it only affects those warnings that have\n"
" been turned on so far or will be turned on later.\n"
"\n"
"-W<warning> Turn on individual warning.\n"
"\n"
"-Wno-<warning> Turn off individual warning.\n"
"\n"
"-Werror-<warning> Turn on individual warning and treat it as error (this implies\n"
" '-W<warning>').\n"
"\n"
"-Wno-error-<warning> Don't treat this particular warning as error. This doesn't turn\n"
" off the warning itself.\n"
"\n"
"Warnings:\n"
"\n"
"-Wcondition-order Warn if the generated program makes implicit assumptions about\n"
" condition numbering. One should use either '-t, --type-header'\n"
" option or '/*!types:re2c*/' directive to generate mapping of\n"
" condition names to numbers and use autogenerated condition names.\n"
"\n"
"-Wempty-character-class Warn if regular expression contains empty character class. From\n"
" the rational point of view trying to match empty character class\n"
" makes no sense: it should always fail. However, for backwards\n"
" compatibility reasons re2c allows empty character class and treats\n"
" it as empty string. Use '--empty-class' option to change default\n"
" behaviour.\n"
"\n"
"-Wmatch-empty-string Warn if regular expression in a rule is nullable (matches empty\n"
" string). If DFA runs in a loop and empty match is unintentional\n"
" (input position in not advanced manually), lexer may get stuck\n"
" in eternal loop.\n"
"\n"
"-Wswapped-range Warn if range lower bound is greater that upper bound. Default\n"
" re2c behaviour is to silently swap range bounds.\n"
"\n"
"-Wundefined-control-flow\n"
" Warn if some input strings cause undefined control flow in lexer\n"
" (the faulty patterns are reported). This is the most dangerous\n"
" and common mistake. It can be easily fixed by adding default rule\n"
" '*' (this rule has the lowest priority, matches any code unit\n"
" and consumes exactly one code unit).\n"
"\n"
"-Wuseless-escape Warn if a symbol is escaped when it shouldn't be. By default re2c\n"
" silently ignores escape, but this may as well indicate a typo\n"
" or an error in escape sequence.\n"
"\n"
);
}
void vernum ()
{
std::string vernum (PACKAGE_VERSION);
if (vernum[1] == '.')
{
vernum.insert(0, "0");
}
vernum.erase(2, 1);
if (vernum[3] == '.')
{
vernum.insert(2, "0");
}
vernum.erase(4, 1);
if (vernum.length() < 6 || vernum[5] < '0' || vernum[5] > '9')
{
vernum.insert(4, "0");
}
vernum.resize(6, '0');
printf ("%s\n", vernum.c_str ());
}
void version ()
{
printf ("re2c %s\n", PACKAGE_VERSION);
}
std::string incond (const std::string & cond)
{
std::string s;
if (!cond.empty ())
{
s += "in condition '";
s += cond;
s += "' ";
}
return s;
}
} // namespace re2c

24
tools/re2c/src/conf/msg.h Normal file
View file

@ -0,0 +1,24 @@
#ifndef _RE2C_CONF_MSG_
#define _RE2C_CONF_MSG_
#include <string>
#include "src/util/attribute.h"
#include "src/util/c99_stdint.h"
namespace re2c {
void error (const char * fmt, ...) RE2C_GXX_ATTRIBUTE ((format (printf, 1, 2)));
void error_encoding ();
void error_arg (const char * option);
void warning_start (uint32_t line, bool error);
void warning_end (const char * type, bool error);
void warning (const char * type, uint32_t line, bool error, const char * fmt, ...) RE2C_GXX_ATTRIBUTE ((format (printf, 4, 5)));
void usage ();
void vernum ();
void version ();
std::string incond (const std::string & cond);
} // namespace re2c
#endif // _RE2C_CONF_MSG_

331
tools/re2c/src/conf/opt.cc Normal file
View file

@ -0,0 +1,331 @@
#include "src/conf/msg.h"
#include "src/conf/opt.h"
namespace re2c
{
Opt opts;
opt_t::opt_t ()
#define OPT1(type, name, value) : name (value)
#define OPT(type, name, value) , name (value)
RE2C_OPTS
#undef OPT1
#undef OPT
{}
opt_t::opt_t (const opt_t & opt)
#define OPT1(type, name, value) : name (opt.name)
#define OPT(type, name, value) , name (opt.name)
RE2C_OPTS
#undef OPT1
#undef OPT
{}
opt_t & opt_t::operator = (const opt_t & opt)
{
#define OPT1 OPT
#define OPT(type, name, value) name = opt.name;
RE2C_OPTS
#undef OPT1
#undef OPT
return *this;
}
void opt_t::fix ()
{
// some options either make no sense or must have fixed value
// with current target: reset them to default
switch (target)
{
case DOT:
// default code generation options
sFlag = Opt::baseopt.sFlag;
bFlag = Opt::baseopt.bFlag;
gFlag = Opt::baseopt.gFlag;
cGotoThreshold = Opt::baseopt.cGotoThreshold;
// default environment-insensitive formatting
yybmHexTable = Opt::baseopt.yybmHexTable;
// fallthrough
case SKELETON:
// default line information
iFlag = Opt::baseopt.iFlag;
// default environment-sensitive formatting
topIndent = Opt::baseopt.topIndent;
indString = Opt::baseopt.indString;
condDivider = Opt::baseopt.condDivider;
condDividerParam = Opt::baseopt.condDividerParam;
// default environment bindings
tFlag = Opt::baseopt.tFlag;
header_file = Opt::baseopt.header_file;
yycondtype = Opt::baseopt.yycondtype;
cond_get = Opt::baseopt.cond_get;
cond_get_naked = Opt::baseopt.cond_get_naked;
cond_set = Opt::baseopt.cond_set;
cond_set_arg = Opt::baseopt.cond_set_arg;
cond_set_naked = Opt::baseopt.cond_set_naked;
yyctable = Opt::baseopt.yyctable;
condPrefix = Opt::baseopt.condPrefix;
condEnumPrefix = Opt::baseopt.condEnumPrefix;
condGoto = Opt::baseopt.condGoto;
condGotoParam = Opt::baseopt.condGotoParam;
fFlag = Opt::baseopt.fFlag;
state_get = Opt::baseopt.state_get;
state_get_naked = Opt::baseopt.state_get_naked;
state_set = Opt::baseopt.state_set;
state_set_arg = Opt::baseopt.state_set_arg;
state_set_naked = Opt::baseopt.state_set_naked;
yyfilllabel = Opt::baseopt.yyfilllabel;
yynext = Opt::baseopt.yynext;
yyaccept = Opt::baseopt.yyaccept;
bUseStateAbort = Opt::baseopt.bUseStateAbort;
bUseStateNext = Opt::baseopt.bUseStateNext;
yybm = Opt::baseopt.yybm;
yytarget = Opt::baseopt.yytarget;
input_api = Opt::baseopt.input_api;
yycursor = Opt::baseopt.yycursor;
yymarker = Opt::baseopt.yymarker;
yyctxmarker = Opt::baseopt.yyctxmarker;
yylimit = Opt::baseopt.yylimit;
yypeek = Opt::baseopt.yypeek;
yyskip = Opt::baseopt.yyskip;
yybackup = Opt::baseopt.yybackup;
yybackupctx = Opt::baseopt.yybackupctx;
yyrestore = Opt::baseopt.yyrestore;
yyrestorectx = Opt::baseopt.yyrestorectx;
yylessthan = Opt::baseopt.yylessthan;
dFlag = Opt::baseopt.dFlag;
yydebug = Opt::baseopt.yydebug;
yyctype = Opt::baseopt.yyctype;
yych = Opt::baseopt.yych;
bEmitYYCh = Opt::baseopt.bEmitYYCh;
yychConversion = Opt::baseopt.yychConversion;
fill = Opt::baseopt.fill;
fill_use = Opt::baseopt.fill_use;
fill_check = Opt::baseopt.fill_check;
fill_arg = Opt::baseopt.fill_arg;
fill_arg_use = Opt::baseopt.fill_arg_use;
fill_naked = Opt::baseopt.fill_naked;
labelPrefix = Opt::baseopt.labelPrefix;
break;
default:
break;
}
if (bCaseInsensitive)
{
bCaseInverted = Opt::baseopt.bCaseInverted;
}
// respect hierarchy
if (!cFlag)
{
tFlag = Opt::baseopt.tFlag;
header_file = Opt::baseopt.header_file;
yycondtype = Opt::baseopt.yycondtype;
cond_get = Opt::baseopt.cond_get;
cond_get_naked = Opt::baseopt.cond_get_naked;
cond_set = Opt::baseopt.cond_set;
cond_set_arg = Opt::baseopt.cond_set_arg;
cond_set_naked = Opt::baseopt.cond_set_naked;
yyctable = Opt::baseopt.yyctable;
condPrefix = Opt::baseopt.condPrefix;
condEnumPrefix = Opt::baseopt.condEnumPrefix;
condDivider = Opt::baseopt.condDivider;
condDividerParam = Opt::baseopt.condDividerParam;
condGoto = Opt::baseopt.condGoto;
condGotoParam = Opt::baseopt.condGotoParam;
}
if (!fFlag)
{
state_get = Opt::baseopt.state_get;
state_get_naked = Opt::baseopt.state_get_naked;
state_set = Opt::baseopt.state_set;
state_set_arg = Opt::baseopt.state_set_arg;
state_set_naked = Opt::baseopt.state_set_naked;
yyfilllabel = Opt::baseopt.yyfilllabel;
yynext = Opt::baseopt.yynext;
yyaccept = Opt::baseopt.yyaccept;
bUseStateAbort = Opt::baseopt.bUseStateAbort;
bUseStateNext = Opt::baseopt.bUseStateNext;
}
if (!bFlag)
{
yybmHexTable = Opt::baseopt.yybmHexTable;
yybm = Opt::baseopt.yybm;
}
if (!gFlag)
{
cGotoThreshold = Opt::baseopt.cGotoThreshold;
yytarget = Opt::baseopt.yytarget;
}
if (input_api.type () != InputAPI::DEFAULT)
{
yycursor = Opt::baseopt.yycursor;
yymarker = Opt::baseopt.yymarker;
yyctxmarker = Opt::baseopt.yyctxmarker;
yylimit = Opt::baseopt.yylimit;
}
if (input_api.type () != InputAPI::CUSTOM)
{
yypeek = Opt::baseopt.yypeek;
yyskip = Opt::baseopt.yyskip;
yybackup = Opt::baseopt.yybackup;
yybackupctx = Opt::baseopt.yybackupctx;
yyrestore = Opt::baseopt.yyrestore;
yyrestorectx = Opt::baseopt.yyrestorectx;
yylessthan = Opt::baseopt.yylessthan;
}
if (!dFlag)
{
yydebug = Opt::baseopt.yydebug;
}
if (!fill_use)
{
fill = Opt::baseopt.fill;
fill_check = Opt::baseopt.fill_check;
fill_arg = Opt::baseopt.fill_arg;
fill_arg_use = Opt::baseopt.fill_arg_use;
fill_naked = Opt::baseopt.fill_naked;
}
// force individual options
switch (target)
{
case DOT:
iFlag = true;
break;
case SKELETON:
iFlag = true;
input_api.set (InputAPI::CUSTOM);
indString = " ";
topIndent = 2;
break;
default:
break;
}
switch (encoding.type ())
{
case Enc::UCS2:
case Enc::UTF16:
case Enc::UTF32:
sFlag = true;
break;
default:
break;
}
if (bFlag)
{
sFlag = true;
}
if (gFlag)
{
bFlag = true;
sFlag = true;
}
if (header_file != NULL)
{
tFlag = true;
}
}
realopt_t::realopt_t (useropt_t & opt)
: real ()
, user (opt)
{}
const opt_t * realopt_t::operator -> ()
{
sync ();
return &real;
}
void realopt_t::sync ()
{
if (user.diverge)
{
real = user.opt;
real.fix ();
user.diverge = false;
}
}
useropt_t::useropt_t ()
: opt ()
, diverge (true)
{}
opt_t * useropt_t::operator -> ()
{
diverge = true;
return &opt;
}
const opt_t Opt::baseopt;
bool Opt::source (const char * s)
{
if (source_file)
{
error ("multiple source files: %s, %s", source_file, s);
return false;
}
else
{
source_file = s;
return true;
}
}
bool Opt::output (const char * s)
{
if (output_file)
{
error ("multiple output files: %s, %s", output_file, s);
return false;
}
else
{
output_file = s;
return true;
}
}
void Opt::reset_encoding (const Enc & enc)
{
useropt->encoding = enc;
}
void Opt::reset_mapCodeName ()
{
// historically arranged set of names
// no actual reason why these particular options should be reset
useropt->cond_get = Opt::baseopt.cond_get;
useropt->cond_set = Opt::baseopt.cond_set;
useropt->fill = Opt::baseopt.fill;
useropt->state_get = Opt::baseopt.state_get;
useropt->state_set = Opt::baseopt.state_set;
useropt->yybackup = Opt::baseopt.yybackup;
useropt->yybackupctx = Opt::baseopt.yybackupctx;
useropt->yycondtype = Opt::baseopt.yycondtype;
useropt->yyctxmarker = Opt::baseopt.yyctxmarker;
useropt->yyctype = Opt::baseopt.yyctype;
useropt->yycursor = Opt::baseopt.yycursor;
useropt->yydebug = Opt::baseopt.yydebug;
useropt->yylessthan = Opt::baseopt.yylessthan;
useropt->yylimit = Opt::baseopt.yylimit;
useropt->yymarker = Opt::baseopt.yymarker;
useropt->yypeek = Opt::baseopt.yypeek;
useropt->yyrestore = Opt::baseopt.yyrestore;
useropt->yyrestorectx = Opt::baseopt.yyrestorectx;
useropt->yyskip = Opt::baseopt.yyskip;
useropt->yyfilllabel = Opt::baseopt.yyfilllabel;
useropt->yynext = Opt::baseopt.yynext;
useropt->yyaccept = Opt::baseopt.yyaccept;
useropt->yybm = Opt::baseopt.yybm;
useropt->yych = Opt::baseopt.yych;
useropt->yyctable = Opt::baseopt.yyctable;
useropt->yytarget = Opt::baseopt.yytarget;
}
} // namespace re2c

218
tools/re2c/src/conf/opt.h Normal file
View file

@ -0,0 +1,218 @@
#ifndef _RE2C_CONF_OPT_
#define _RE2C_CONF_OPT_
#include "src/util/c99_stdint.h"
#include <stddef.h>
#include <string>
#include "src/codegen/input_api.h"
#include "src/ir/dfa/dfa.h"
#include "src/ir/regexp/empty_class_policy.h"
#include "src/ir/regexp/encoding/enc.h"
#include "src/util/forbid_copy.h"
namespace re2c
{
#define RE2C_OPTS \
/* target */ \
OPT1 (opt_t::target_t, target, CODE) \
/* fingerprint */ \
OPT (bool, bNoGenerationDate, false) \
OPT (bool, version, true) \
/* regular expressions */ \
OPT (Enc, encoding, Enc ()) \
OPT (bool, bCaseInsensitive, false) \
OPT (bool, bCaseInverted, false) \
OPT (empty_class_policy_t, empty_class_policy, EMPTY_CLASS_MATCH_EMPTY) \
/* conditions */ \
OPT (bool, cFlag, false) \
OPT (bool, tFlag, false) \
OPT (const char *, header_file, NULL) \
OPT (std::string, yycondtype, "YYCONDTYPE") \
OPT (std::string, cond_get, "YYGETCONDITION") \
OPT (bool, cond_get_naked, false) \
OPT (std::string, cond_set, "YYSETCONDITION" ) \
OPT (std::string, cond_set_arg, "@@" ) \
OPT (bool, cond_set_naked, false ) \
OPT (std::string, yyctable, "yyctable") \
OPT (std::string, condPrefix, "yyc_") \
OPT (std::string, condEnumPrefix, "yyc") \
OPT (std::string, condDivider, "/* *********************************** */") \
OPT (std::string, condDividerParam, "@@") \
OPT (std::string, condGoto, "goto @@;") \
OPT (std::string, condGotoParam, "@@") \
/* states */ \
OPT (bool, fFlag, false) \
OPT (std::string, state_get, "YYGETSTATE") \
OPT (bool, state_get_naked, false) \
OPT (std::string, state_set, "YYSETSTATE") \
OPT (std::string, state_set_arg, "@@") \
OPT (bool, state_set_naked, false) \
OPT (std::string, yyfilllabel, "yyFillLabel") \
OPT (std::string, yynext, "yyNext") \
OPT (std::string, yyaccept, "yyaccept") \
OPT (bool, bUseStateAbort, false) \
OPT (bool, bUseStateNext, false) \
/* reuse */ \
OPT (bool, rFlag, false) \
/* partial flex syntax support */ \
OPT (bool, FFlag, false) \
/* code generation */ \
OPT (bool, sFlag, false) \
OPT (bool, bFlag, false) \
OPT (std::string, yybm, "yybm") \
OPT (bool, yybmHexTable, false) \
OPT (bool, gFlag, false) \
OPT (std::string, yytarget, "yytarget") \
OPT (uint32_t, cGotoThreshold, 9) \
/* formatting */ \
OPT (uint32_t, topIndent, 0) \
OPT (std::string, indString, "\t") \
/* input API */ \
OPT (InputAPI, input_api, InputAPI ()) \
OPT (std::string, yycursor, "YYCURSOR") \
OPT (std::string, yymarker, "YYMARKER") \
OPT (std::string, yyctxmarker, "YYCTXMARKER") \
OPT (std::string, yylimit, "YYLIMIT") \
OPT (std::string, yypeek, "YYPEEK") \
OPT (std::string, yyskip, "YYSKIP") \
OPT (std::string, yybackup, "YYBACKUP") \
OPT (std::string, yybackupctx, "YYBACKUPCTX") \
OPT (std::string, yyrestore, "YYRESTORE") \
OPT (std::string, yyrestorectx, "YYRESTORECTX") \
OPT (std::string, yylessthan, "YYLESSTHAN") \
/* #line directives */ \
OPT (bool, iFlag, false) \
/* debug */ \
OPT (bool, dFlag, false) \
OPT (std::string, yydebug, "YYDEBUG") \
/* yych */ \
OPT (std::string, yyctype, "YYCTYPE") \
OPT (std::string, yych, "yych") \
OPT (bool, bEmitYYCh, true) \
OPT (bool, yychConversion, false) \
/* YYFILL */ \
OPT (std::string, fill, "YYFILL") \
OPT (bool, fill_use, true) \
OPT (bool, fill_check, true) \
OPT (std::string, fill_arg, "@@") \
OPT (bool, fill_arg_use, true) \
OPT (bool, fill_naked, false) \
/* labels */ \
OPT (std::string, labelPrefix, "yy") \
/* internals */ \
OPT (dfa_minimization_t, dfa_minimization, DFA_MINIMIZATION_MOORE)
struct opt_t
{
enum target_t
{
CODE,
DOT,
SKELETON
};
#define OPT1 OPT
#define OPT(type, name, value) type name;
RE2C_OPTS
#undef OPT1
#undef OPT
opt_t ();
opt_t (const opt_t & opt);
opt_t & operator = (const opt_t & opt);
void fix ();
};
class useropt_t;
class realopt_t
{
opt_t real;
useropt_t & user;
public:
realopt_t (useropt_t & opt);
const opt_t * operator -> ();
void sync ();
};
class useropt_t
{
opt_t opt;
bool diverge;
public:
useropt_t ();
opt_t * operator -> ();
friend void realopt_t::sync ();
};
struct Opt
{
static const opt_t baseopt;
const char * source_file;
const char * output_file;
private:
useropt_t useropt;
realopt_t realopt;
public:
Opt ()
: source_file (NULL)
, output_file (NULL)
, useropt ()
, realopt (useropt)
{}
// read-only access, forces options syncronization
const opt_t * operator -> ()
{
return realopt.operator -> ();
}
bool source (const char * s);
bool output (const char * s);
// Inplace configurations are applied immediately when parsed.
// This is very bad: first, re2c behaviour is changed in the middle
// of the block; second, config is resynced too often (every
// attempt to read config that has been updated results in
// automatic resync). It is much better to set all options at once.
bool set_encoding (Enc::type_t t) { return useropt->encoding.set (t); }
void unset_encoding (Enc::type_t t) { useropt->encoding.unset (t); }
void set_encoding_policy (Enc::policy_t p) { useropt->encoding.setPolicy (p); }
void set_input_api (InputAPI::type_t t) { useropt->input_api.set (t); }
#define OPT1 OPT
#define OPT(type, name, value) void set_##name (type arg) { useropt->name = arg; }
RE2C_OPTS
#undef OPT1
#undef OPT
// helpers
std::string yychConversion ()
{
return realopt->yychConversion
? "(" + realopt->yyctype + ")"
: "";
}
// bad temporary hacks, should be fixed by proper scoping of config (parts).
void reset_encoding (const Enc & enc);
void reset_mapCodeName ();
FORBID_COPY (Opt);
};
enum parse_opts_t
{
OK,
EXIT_OK,
EXIT_FAIL
};
parse_opts_t parse_opts (char ** argv, Opt & opts);
} // namespace re2c
#endif // _RE2C_CONF_OPT_

2846
tools/re2c/src/conf/parse_opts.cc Normal file
View file

File diff suppressed because it is too large Load diff

261
tools/re2c/src/conf/parse_opts.re Normal file
View file

@ -0,0 +1,261 @@
#include "src/codegen/input_api.h"
#include "src/conf/msg.h"
#include "src/conf/opt.h"
#include "src/conf/warn.h"
#include "src/globals.h"
#include "src/ir/regexp/empty_class_policy.h"
#include "src/ir/regexp/encoding/enc.h"
namespace re2c
{
static inline bool next (char * & arg, char ** & argv)
{
arg = *++argv;
return arg != NULL;
}
parse_opts_t parse_opts (char ** argv, Opt & opts)
{
#define YYCTYPE unsigned char
char * YYCURSOR;
char * YYMARKER;
Warn::option_t option;
/*!re2c
re2c:yyfill:enable = 0;
re2c:yych:conversion = 1;
end = "\x00";
filename = [^\x00-] [^\x00]*;
*/
opt:
if (!next (YYCURSOR, argv))
{
goto end;
}
/*!re2c
*
{
error ("bad option: %s", *argv);
return EXIT_FAIL;
}
"--" end
{
// all remaining arguments are non-options
// so they must be input files
// re2c expects exactly one input file
for (char * f; next (f, argv);)
{
if (!opts.source (f))
{
return EXIT_FAIL;
}
}
goto end;
}
"-" end { if (!opts.source ("<stdin>")) return EXIT_FAIL; goto opt; }
filename end { if (!opts.source (*argv)) return EXIT_FAIL; goto opt; }
"-" { goto opt_short; }
"--" { goto opt_long; }
"-W" end { warn.set_all (); goto opt; }
"-Werror" end { warn.set_all_error (); goto opt; }
"-W" { option = Warn::W; goto opt_warn; }
"-Wno-" { option = Warn::WNO; goto opt_warn; }
"-Werror-" { option = Warn::WERROR; goto opt_warn; }
"-Wno-error-" { option = Warn::WNOERROR; goto opt_warn; }
*/
opt_warn:
/*!re2c
*
{
error ("bad warning: %s", *argv);
return EXIT_FAIL;
}
"condition-order" end { warn.set (Warn::CONDITION_ORDER, option); goto opt; }
"empty-character-class" end { warn.set (Warn::EMPTY_CHARACTER_CLASS, option); goto opt; }
"match-empty-string" end { warn.set (Warn::MATCH_EMPTY_STRING, option); goto opt; }
"swapped-range" end { warn.set (Warn::SWAPPED_RANGE, option); goto opt; }
"undefined-control-flow" end { warn.set (Warn::UNDEFINED_CONTROL_FLOW, option); goto opt; }
"unreachable-rules" end { warn.set (Warn::UNREACHABLE_RULES, option); goto opt; }
"useless-escape" end { warn.set (Warn::USELESS_ESCAPE, option); goto opt; }
*/
opt_short:
/*!re2c
*
{
error ("bad short option: %s", *argv);
return EXIT_FAIL;
}
end { goto opt; }
[?h] { usage (); return EXIT_OK; }
"v" { version (); return EXIT_OK; }
"V" { vernum (); return EXIT_OK; }
"b" { opts.set_bFlag (true); goto opt_short; }
"c" { opts.set_cFlag (true); goto opt_short; }
"d" { opts.set_dFlag (true); goto opt_short; }
"D" { opts.set_target (opt_t::DOT); goto opt_short; }
"f" { opts.set_fFlag (true); goto opt_short; }
"F" { opts.set_FFlag (true); goto opt_short; }
"g" { opts.set_gFlag (true); goto opt_short; }
"i" { opts.set_iFlag (true); goto opt_short; }
"r" { opts.set_rFlag (true); goto opt_short; }
"s" { opts.set_sFlag (true); goto opt_short; }
"S" { opts.set_target (opt_t::SKELETON); goto opt_short; }
"e" { if (!opts.set_encoding (Enc::EBCDIC)) { error_encoding (); return EXIT_FAIL; } goto opt_short; }
"u" { if (!opts.set_encoding (Enc::UTF32)) { error_encoding (); return EXIT_FAIL; } goto opt_short; }
"w" { if (!opts.set_encoding (Enc::UCS2)) { error_encoding (); return EXIT_FAIL; } goto opt_short; }
"x" { if (!opts.set_encoding (Enc::UTF16)) { error_encoding (); return EXIT_FAIL; } goto opt_short; }
"8" { if (!opts.set_encoding (Enc::UTF8)) { error_encoding (); return EXIT_FAIL; } goto opt_short; }
"o" end { if (!next (YYCURSOR, argv)) { error_arg ("-o, --output"); return EXIT_FAIL; } goto opt_output; }
"o" { *argv = YYCURSOR; goto opt_output; }
"t" end { if (!next (YYCURSOR, argv)) { error_arg ("-t, --type-header"); return EXIT_FAIL; } goto opt_header; }
"t" { *argv = YYCURSOR; goto opt_header; }
"1" { goto opt_short; } // deprecated
*/
opt_long:
/*!re2c
*
{
error ("bad long option: %s", *argv);
return EXIT_FAIL;
}
"help" end { usage (); return EXIT_OK; }
"version" end { version (); return EXIT_OK; }
"vernum" end { vernum (); return EXIT_OK; }
"bit-vectors" end { opts.set_bFlag (true); goto opt; }
"start-conditions" end { opts.set_cFlag (true); goto opt; }
"debug-output" end { opts.set_dFlag (true); goto opt; }
"emit-dot" end { opts.set_target (opt_t::DOT); goto opt; }
"storable-state" end { opts.set_fFlag (true); goto opt; }
"flex-syntax" end { opts.set_FFlag (true); goto opt; }
"computed-gotos" end { opts.set_gFlag (true); goto opt; }
"no-debug-info" end { opts.set_iFlag (true); goto opt; }
"reusable" end { opts.set_rFlag (true); goto opt; }
"nested-ifs" end { opts.set_sFlag (true); goto opt; }
"no-generation-date" end { opts.set_bNoGenerationDate (true); goto opt; }
"no-version" end { opts.set_version (false); goto opt; }
"case-insensitive" end { opts.set_bCaseInsensitive (true); goto opt; }
"case-inverted" end { opts.set_bCaseInverted (true); goto opt; }
"skeleton" end { opts.set_target (opt_t::SKELETON); goto opt; }
"ecb" end { if (!opts.set_encoding (Enc::EBCDIC)) { error_encoding (); return EXIT_FAIL; } goto opt; }
"unicode" end { if (!opts.set_encoding (Enc::UTF32)) { error_encoding (); return EXIT_FAIL; } goto opt; }
"wide-chars" end { if (!opts.set_encoding (Enc::UCS2)) { error_encoding (); return EXIT_FAIL; } goto opt; }
"utf-16" end { if (!opts.set_encoding (Enc::UTF16)) { error_encoding (); return EXIT_FAIL; } goto opt; }
"utf-8" end { if (!opts.set_encoding (Enc::UTF8)) { error_encoding (); return EXIT_FAIL; } goto opt; }
"output" end { if (!next (YYCURSOR, argv)) { error_arg ("-o, --output"); return EXIT_FAIL; } goto opt_output; }
"type-header" end { if (!next (YYCURSOR, argv)) { error_arg ("-t, --type-header"); return EXIT_FAIL; } goto opt_header; }
"encoding-policy" end { goto opt_encoding_policy; }
"input" end { goto opt_input; }
"empty-class" end { goto opt_empty_class; }
"dfa-minimization" end { goto opt_dfa_minimization; }
"single-pass" end { goto opt; } // deprecated
*/
opt_output:
/*!re2c
*
{
error ("bad argument to option -o, --output: %s", *argv);
return EXIT_FAIL;
}
filename end { if (!opts.output (*argv)) return EXIT_FAIL; goto opt; }
*/
opt_header:
/*!re2c
*
{
error ("bad argument to option -t, --type-header: %s", *argv);
return EXIT_FAIL;
}
filename end { opts.set_header_file (*argv); goto opt; }
*/
opt_encoding_policy:
if (!next (YYCURSOR, argv))
{
error_arg ("--encoding-policy");
return EXIT_FAIL;
}
/*!re2c
*
{
error ("bad argument to option --encoding-policy (expected: ignore | substitute | fail): %s", *argv);
return EXIT_FAIL;
}
"ignore" end { opts.set_encoding_policy (Enc::POLICY_IGNORE); goto opt; }
"substitute" end { opts.set_encoding_policy (Enc::POLICY_SUBSTITUTE); goto opt; }
"fail" end { opts.set_encoding_policy (Enc::POLICY_FAIL); goto opt; }
*/
opt_input:
if (!next (YYCURSOR, argv))
{
error_arg ("--input");
return EXIT_FAIL;
}
/*!re2c
*
{
error ("bad argument to option --input (expected: default | custom): %s", *argv);
return EXIT_FAIL;
}
"default" end { opts.set_input_api (InputAPI::DEFAULT); goto opt; }
"custom" end { opts.set_input_api (InputAPI::CUSTOM); goto opt; }
*/
opt_empty_class:
if (!next (YYCURSOR, argv))
{
error_arg ("--empty-class");
return EXIT_FAIL;
}
/*!re2c
*
{
error ("bad argument to option --empty-class (expected: match-empty | match-none | error): %s", *argv);
return EXIT_FAIL;
}
"match-empty" end { opts.set_empty_class_policy (EMPTY_CLASS_MATCH_EMPTY); goto opt; }
"match-none" end { opts.set_empty_class_policy (EMPTY_CLASS_MATCH_NONE); goto opt; }
"error" end { opts.set_empty_class_policy (EMPTY_CLASS_ERROR); goto opt; }
*/
opt_dfa_minimization:
if (!next (YYCURSOR, argv))
{
error_arg ("--minimization");
return EXIT_FAIL;
}
/*!re2c
*
{
error ("bad argument to option --dfa-minimization (expected: table | moore): %s", *argv);
return EXIT_FAIL;
}
"table" end { opts.set_dfa_minimization (DFA_MINIMIZATION_TABLE); goto opt; }
"moore" end { opts.set_dfa_minimization (DFA_MINIMIZATION_MOORE); goto opt; }
*/
end:
if (!opts.source_file)
{
error ("no source file");
return EXIT_FAIL;
}
return OK;
#undef YYCTYPE
}
} // namespace re2c

200
tools/re2c/src/conf/warn.cc Normal file
View file

@ -0,0 +1,200 @@
#include <stddef.h>
#include <stdio.h>
#include <algorithm>
#include <utility>
#include "src/conf/msg.h"
#include "src/conf/warn.h"
namespace re2c {
Warn warn;
const uint32_t Warn::SILENT = 0;
const uint32_t Warn::WARNING = 1u << 0;
const uint32_t Warn::ERROR = 1u << 1;
const char * Warn::names [TYPES] =
{
#define W(x, y) y
RE2C_WARNING_TYPES
#undef W
};
Warn::Warn ()
: mask ()
, error_accuml (false)
{
for (uint32_t i = 0; i < TYPES; ++i)
{
mask[i] = SILENT;
}
}
bool Warn::error () const
{
return error_accuml;
}
void Warn::set (type_t t, option_t o)
{
switch (o)
{
case W:
mask[t] |= WARNING;
break;
case WNO:
mask[t] &= ~WARNING;
break;
case WERROR:
// unlike -Werror, -Werror-<warning> implies -W<warning>
mask[t] |= (WARNING | ERROR);
break;
case WNOERROR:
mask[t] &= ~ERROR;
break;
}
}
void Warn::set_all ()
{
for (uint32_t i = 0; i < TYPES; ++i)
{
mask[i] |= WARNING;
}
}
// -Werror doesn't set any warnings: it only guarantees that if a warning
// has been set by now or will be set later then it will result into error.
void Warn::set_all_error ()
{
for (uint32_t i = 0; i < TYPES; ++i)
{
mask[i] |= ERROR;
}
}
void Warn::fail (type_t t, uint32_t line, const char * s)
{
if (mask[t] & WARNING)
{
// -Werror has no effect
warning (names[t], line, false, "%s", s);
}
}
void Warn::condition_order (uint32_t line)
{
if (mask[CONDITION_ORDER] & WARNING)
{
const bool e = mask[CONDITION_ORDER] & ERROR;
error_accuml |= e;
warning (names[CONDITION_ORDER], line, e,
"looks like you use hardcoded numbers instead of autogenerated condition names: "
"better add '/*!types:re2c*/' directive or '-t, --type-header' option "
"and don't rely on fixed condition order.");
}
}
void Warn::empty_class (uint32_t line)
{
if (mask[EMPTY_CHARACTER_CLASS] & WARNING)
{
const bool e = mask[EMPTY_CHARACTER_CLASS] & ERROR;
error_accuml |= e;
warning (names[EMPTY_CHARACTER_CLASS], line, e, "empty character class");
}
}
void Warn::match_empty_string (uint32_t line)
{
if (mask[MATCH_EMPTY_STRING] & WARNING)
{
const bool e = mask[MATCH_EMPTY_STRING] & ERROR;
error_accuml |= e;
warning (names[MATCH_EMPTY_STRING], line, e, "rule matches empty string");
}
}
void Warn::swapped_range (uint32_t line, uint32_t l, uint32_t u)
{
if (mask[SWAPPED_RANGE] & WARNING)
{
const bool e = mask[SWAPPED_RANGE] & ERROR;
error_accuml |= e;
warning (names[SWAPPED_RANGE], line, e, "range lower bound (0x%X) is greater than upper bound (0x%X), swapping", l, u);
}
}
void Warn::undefined_control_flow (uint32_t line, const std::string & cond, std::vector<way_t> & ways, bool overflow)
{
if (mask[UNDEFINED_CONTROL_FLOW] & WARNING)
{
const bool e = mask[UNDEFINED_CONTROL_FLOW] & ERROR;
error_accuml |= e;
// report shorter patterns first
std::sort (ways.begin (), ways.end (), cmp_ways);
warning_start (line, e);
fprintf (stderr, "control flow %sis undefined for strings that match ", incond (cond).c_str ());
const size_t count = ways.size ();
if (count == 1)
{
fprint_way (stderr, ways[0]);
}
else
{
for (size_t i = 0; i < count; ++i)
{
fprintf (stderr, "\n\t");
fprint_way (stderr, ways[i]);
}
fprintf (stderr, "\n");
}
if (overflow)
{
fprintf (stderr, " ... and a few more");
}
fprintf (stderr, ", use default rule '*'");
warning_end (names[UNDEFINED_CONTROL_FLOW], e);
}
}
void Warn::unreachable_rule (const std::string & cond, const rule_info_t & rule, const rules_t & rules)
{
if (mask[UNREACHABLE_RULES] & WARNING)
{
const bool e = mask[UNREACHABLE_RULES] & ERROR;
error_accuml |= e;
warning_start (rule.line, e);
fprintf (stderr, "unreachable rule %s", incond (cond).c_str ());
const size_t shadows = rule.shadow.size ();
if (shadows > 0)
{
const char * pl = shadows > 1
? "s"
: "";
std::set<rule_rank_t>::const_iterator i = rule.shadow.begin ();
fprintf (stderr, "(shadowed by rule%s at line%s %u", pl, pl, rules.find (*i)->second.line);
for (++i; i != rule.shadow.end (); ++i)
{
fprintf (stderr, ", %u", rules.find (*i)->second.line);
}
fprintf (stderr, ")");
}
warning_end (names[UNREACHABLE_RULES], e);
}
}
void Warn::useless_escape (uint32_t line, uint32_t col, char c)
{
if (mask[USELESS_ESCAPE] & WARNING)
{
const bool e = mask[USELESS_ESCAPE] & ERROR;
error_accuml |= e;
warning (names[USELESS_ESCAPE], line, e, "column %u: escape has no effect: '\\%c'", col, c);
}
}
} // namespace re2c

67
tools/re2c/src/conf/warn.h Normal file
View file

@ -0,0 +1,67 @@
#ifndef _RE2C_CONF_WARN_
#define _RE2C_CONF_WARN_
#include "src/util/c99_stdint.h"
#include <string>
#include <vector>
#include "src/ir/skeleton/way.h"
#include "src/parse/rules.h"
namespace re2c {
#define RE2C_WARNING_TYPES \
W (CONDITION_ORDER, "condition-order"), \
W (EMPTY_CHARACTER_CLASS, "empty-character-class"), \
W (MATCH_EMPTY_STRING, "match-empty-string"), \
W (SWAPPED_RANGE, "swapped-range"), \
W (UNDEFINED_CONTROL_FLOW, "undefined-control-flow"), \
W (UNREACHABLE_RULES, "unreachable-rules"), \
W (USELESS_ESCAPE, "useless-escape"),
class Warn
{
public:
enum type_t
{
#define W(x, y) x
RE2C_WARNING_TYPES
#undef W
TYPES // count
};
enum option_t
{
W,
WNO,
WERROR,
WNOERROR
};
private:
static const uint32_t SILENT;
static const uint32_t WARNING;
static const uint32_t ERROR;
static const char * names [TYPES];
uint32_t mask[TYPES];
bool error_accuml;
public:
Warn ();
bool error () const;
void set (type_t t, option_t o);
void set_all ();
void set_all_error ();
void fail (type_t t, uint32_t line, const char * s);
void condition_order (uint32_t line);
void empty_class (uint32_t line);
void match_empty_string (uint32_t line);
void swapped_range (uint32_t line, uint32_t l, uint32_t u);
void undefined_control_flow (uint32_t line, const std::string & cond, std::vector<way_t> & ways, bool overflow);
void unreachable_rule (const std::string & cond, const rule_info_t & rule, const rules_t & rules);
void useless_escape (uint32_t line, uint32_t col, char c);
};
} // namespace re2c
#endif // _RE2C_CONF_WARN_

24
tools/re2c/src/globals.h Normal file
View file

@ -0,0 +1,24 @@
#ifndef _RE2C_GLOBALS_
#define _RE2C_GLOBALS_
#include <string>
#include "src/conf/opt.h"
#include "src/conf/warn.h"
#include "src/util/c99_stdint.h"
namespace re2c
{
extern bool bUsedYYBitmap;
extern bool bWroteGetState;
extern bool bWroteCondCheck;
extern uint32_t last_fill_index;
extern std::string yySetupRule;
extern Opt opts;
extern Warn warn;
} // end namespace re2c
#endif // _RE2C_GLOBALS_

109
tools/re2c/src/ir/adfa/action.h Normal file
View file

@ -0,0 +1,109 @@
#ifndef _RE2C_IR_ADFA_ACTION_
#define _RE2C_IR_ADFA_ACTION_
#include <vector>
#include "src/codegen/label.h"
#include "src/util/c99_stdint.h"
#include "src/util/uniq_vector.h"
namespace re2c
{
struct OutputFile;
class RuleOp;
struct State;
struct Initial
{
label_t label;
bool setMarker;
inline Initial (label_t l, bool b)
: label (l)
, setMarker (b)
{}
};
typedef uniq_vector_t<const State *> accept_t;
class Action
{
public:
enum type_t
{
MATCH,
INITIAL,
SAVE,
MOVE,
ACCEPT,
RULE
} type;
union
{
Initial * initial;
uint32_t save;
const accept_t * accepts;
const RuleOp * rule;
} info;
public:
inline Action ()
: type (MATCH)
, info ()
{}
~Action ()
{
clear ();
}
void set_initial (label_t label, bool used_marker)
{
clear ();
type = INITIAL;
info.initial = new Initial (label, used_marker);
}
void set_save (uint32_t save)
{
clear ();
type = SAVE;
info.save = save;
}
void set_move ()
{
clear ();
type = MOVE;
}
void set_accept (const accept_t * accepts)
{
clear ();
type = ACCEPT;
info.accepts = accepts;
}
void set_rule (const RuleOp * const rule)
{
clear ();
type = RULE;
info.rule = rule;
}
private:
void clear ()
{
switch (type)
{
case INITIAL:
delete info.initial;
break;
case MATCH:
case SAVE:
case MOVE:
case ACCEPT:
case RULE:
break;
}
}
};
} // namespace re2c
#endif // _RE2C_IR_ADFA_ACTION_

135
tools/re2c/src/ir/adfa/adfa.cc Normal file
View file

@ -0,0 +1,135 @@
#include <assert.h>
#include <queue>
#include <set>
#include <vector>
#include <utility>
#include "src/codegen/go.h"
#include "src/ir/adfa/adfa.h"
#include "src/ir/dfa/dfa.h"
#include "src/ir/skeleton/skeleton.h"
#include "src/util/allocate.h"
namespace re2c
{
DFA::DFA
( const dfa_t &dfa
, const std::vector<size_t> &fill
, Skeleton *skel
, const charset_t &charset
, const std::string &n
, const std::string &c
, uint32_t l
)
: accepts ()
, skeleton (skel)
, name (n)
, cond (c)
, line (l)
, lbChar(0)
, ubChar(charset.back())
, nStates(0)
, head(NULL)
// statistics
, max_fill (0)
, need_backup (false)
, need_backupctx (false)
, need_accept (false)
{
const size_t nstates = dfa.states.size();
const size_t nchars = dfa.nchars;
State **i2s = new State*[nstates];
for (size_t i = 0; i < nstates; ++i)
{
i2s[i] = new State;
}
State **p = &head;
for (size_t i = 0; i < nstates; ++i)
{
dfa_state_t *t = dfa.states[i];
State *s = i2s[i];
++nStates;
*p = s;
p = &s->next;
s->isPreCtxt = t->ctx;
s->rule = t->rule;
s->fill = fill[i];
s->go.span = allocate<Span>(nchars);
uint32_t j = 0;
for (uint32_t c = 0; c < nchars; ++j)
{
const size_t to = t->arcs[c];
for (;++c < nchars && t->arcs[c] == to;);
s->go.span[j].to = to == dfa_t::NIL ? NULL : i2s[to];
s->go.span[j].ub = charset[c];
}
s->go.nSpans = j;
}
*p = NULL;
delete[] i2s;
}
DFA::~DFA()
{
State *s;
while ((s = head))
{
head = s->next;
delete s;
}
delete skeleton;
}
void DFA::reorder()
{
std::vector<State*> ord;
ord.reserve(nStates);
std::queue<State*> todo;
todo.push(head);
std::set<State*> done;
done.insert(head);
for(;!todo.empty();)
{
State *s = todo.front();
todo.pop();
ord.push_back(s);
for(uint32_t i = 0; i < s->go.nSpans; ++i)
{
State *q = s->go.span[i].to;
if(q && done.insert(q).second)
{
todo.push(q);
}
}
}
assert(nStates == ord.size());
ord.push_back(NULL);
for(uint32_t i = 0; i < nStates; ++i)
{
ord[i]->next = ord[i + 1];
}
}
void DFA::addState(State *s, State *next)
{
++nStates;
s->next = next->next;
next->next = s;
}
} // namespace re2c

101
tools/re2c/src/ir/adfa/adfa.h Normal file
View file

@ -0,0 +1,101 @@
#ifndef _RE2C_IR_ADFA_ADFA_
#define _RE2C_IR_ADFA_ADFA_
#include <stddef.h>
#include "src/util/c99_stdint.h"
#include <set>
#include <string>
#include "src/codegen/go.h"
#include "src/codegen/label.h"
#include "src/ir/adfa/action.h"
#include "src/ir/regexp/regexp.h"
#include "src/util/forbid_copy.h"
namespace re2c
{
struct Skeleton;
struct Output;
struct OutputFile;
struct dfa_t;
struct State
{
label_t label;
RuleOp * rule;
State * next;
size_t fill;
bool isPreCtxt;
bool isBase;
Go go;
Action action;
State ()
: label (label_t::first ())
, rule (NULL)
, next (0)
, fill (0)
, isPreCtxt (false)
, isBase (false)
, go ()
, action ()
{}
~State ()
{
operator delete (go.span);
}
FORBID_COPY (State);
};
class DFA
{
accept_t accepts;
Skeleton * skeleton;
public:
const std::string name;
const std::string cond;
const uint32_t line;
uint32_t lbChar;
uint32_t ubChar;
uint32_t nStates;
State * head;
// statistics
size_t max_fill;
bool need_backup;
bool need_backupctx;
bool need_accept;
public:
DFA ( const dfa_t &dfa
, const std::vector<size_t> &fill
, Skeleton *skel
, const charset_t &charset
, const std::string &n
, const std::string &c
, uint32_t l
);
~DFA ();
void reorder();
void prepare();
void calc_stats();
void emit (Output &, uint32_t &, bool, bool &);
private:
void addState(State*, State *);
void split (State *);
void findBaseState ();
void count_used_labels (std::set<label_t> & used, label_t prolog, label_t start, bool force_start) const;
void emit_body (OutputFile &, uint32_t &, const std::set<label_t> & used_labels, label_t initial) const;
FORBID_COPY (DFA);
};
} // namespace re2c
#endif // _RE2C_IR_ADFA_ADFA_

268
tools/re2c/src/ir/adfa/prepare.cc Normal file
View file

@ -0,0 +1,268 @@
#include "src/util/c99_stdint.h"
#include <string.h>
#include <map>
#include "src/codegen/bitmap.h"
#include "src/codegen/go.h"
#include "src/globals.h"
#include "src/ir/adfa/action.h"
#include "src/ir/adfa/adfa.h"
#include "src/ir/regexp/regexp_rule.h"
#include "src/ir/rule_rank.h"
#include "src/util/allocate.h"
namespace re2c {
void DFA::split(State *s)
{
State *move = new State;
addState(move, s);
move->action.set_move ();
move->rule = s->rule;
move->fill = s->fill;
move->go = s->go;
s->rule = NULL;
s->go.nSpans = 1;
s->go.span = allocate<Span> (1);
s->go.span[0].ub = ubChar;
s->go.span[0].to = move;
}
static uint32_t merge(Span *x0, State *fg, State *bg)
{
Span *x = x0, *f = fg->go.span, *b = bg->go.span;
uint32_t nf = fg->go.nSpans, nb = bg->go.nSpans;
State *prev = NULL, *to;
// NB: we assume both spans are for same range
for (;;)
{
if (f->ub == b->ub)
{
to = f->to == b->to ? bg : f->to;
if (to == prev)
{
--x;
}
else
{
x->to = prev = to;
}
x->ub = f->ub;
++x;
++f;
--nf;
++b;
--nb;
if (nf == 0 && nb == 0)
{
return static_cast<uint32_t> (x - x0);
}
}
while (f->ub < b->ub)
{
to = f->to == b->to ? bg : f->to;
if (to == prev)
{
--x;
}
else
{
x->to = prev = to;
}
x->ub = f->ub;
++x;
++f;
--nf;
}
while (b->ub < f->ub)
{
to = b->to == f->to ? bg : f->to;
if (to == prev)
{
--x;
}
else
{
x->to = prev = to;
}
x->ub = b->ub;
++x;
++b;
--nb;
}
}
}
void DFA::findBaseState()
{
Span *span = allocate<Span> (ubChar - lbChar);
for (State *s = head; s; s = s->next)
{
if (s->fill == 0)
{
for (uint32_t i = 0; i < s->go.nSpans; ++i)
{
State *to = s->go.span[i].to;
if (to->isBase)
{
to = to->go.span[0].to;
uint32_t nSpans = merge(span, s, to);
if (nSpans < s->go.nSpans)
{
operator delete (s->go.span);
s->go.nSpans = nSpans;
s->go.span = allocate<Span> (nSpans);
memcpy(s->go.span, span, nSpans*sizeof(Span));
}
break;
}
}
}
}
operator delete (span);
}
void DFA::prepare ()
{
bUsedYYBitmap = false;
// create rule states
std::map<rule_rank_t, State *> rules;
for (State * s = head; s; s = s->next)
{
if (s->rule)
{
if (rules.find (s->rule->rank) == rules.end ())
{
State *n = new State;
n->action.set_rule (s->rule);
rules[s->rule->rank] = n;
addState(n, s);
}
for (uint32_t i = 0; i < s->go.nSpans; ++i)
{
if (!s->go.span[i].to)
{
s->go.span[i].to = rules[s->rule->rank];
}
}
}
}
// create default state (if needed)
State * default_state = NULL;
for (State * s = head; s; s = s->next)
{
for (uint32_t i = 0; i < s->go.nSpans; ++i)
{
if (!s->go.span[i].to)
{
if (!default_state)
{
default_state = new State;
addState(default_state, s);
}
s->go.span[i].to = default_state;
}
}
}
// find backup states and create accept state (if needed)
if (default_state)
{
for (State * s = head; s; s = s->next)
{
if (s->rule)
{
for (uint32_t i = 0; i < s->go.nSpans; ++i)
{
if (!s->go.span[i].to->rule && s->go.span[i].to->action.type != Action::RULE)
{
const uint32_t accept = static_cast<uint32_t> (accepts.find_or_add (rules[s->rule->rank]));
s->action.set_save (accept);
}
}
}
}
default_state->action.set_accept (&accepts);
}
// split ``base'' states into two parts
for (State * s = head; s; s = s->next)
{
s->isBase = false;
if (s->fill != 0)
{
for (uint32_t i = 0; i < s->go.nSpans; ++i)
{
if (s->go.span[i].to == s)
{
s->isBase = true;
split(s);
if (opts->bFlag)
{
BitMap::find(&s->next->go, s);
}
s = s->next;
break;
}
}
}
}
// find ``base'' state, if possible
findBaseState();
for (State * s = head; s; s = s->next)
{
s->go.init (s);
}
}
void DFA::calc_stats ()
{
// calculate 'YYMAXFILL'
max_fill = 0;
for (State * s = head; s; s = s->next)
{
if (max_fill < s->fill)
{
max_fill = s->fill;
}
}
// determine if 'YYMARKER' or 'YYBACKUP'/'YYRESTORE' pair is used
need_backup = accepts.size () > 0;
// determine if 'YYCTXMARKER' or 'YYBACKUPCTX'/'YYRESTORECTX' pair is used
for (State * s = head; s; s = s->next)
{
if (s->isPreCtxt)
{
need_backupctx = true;
}
}
// determine if 'yyaccept' variable is used
need_accept = accepts.size () > 1;
}
} // namespace re2c

104
tools/re2c/src/ir/compile.cc Normal file
View file

@ -0,0 +1,104 @@
#include <algorithm>
#include <ostream>
#include <set>
#include "src/codegen/output.h"
#include "src/ir/compile.h"
#include "src/ir/adfa/adfa.h"
#include "src/ir/dfa/dfa.h"
#include "src/ir/nfa/nfa.h"
#include "src/ir/regexp/regexp.h"
#include "src/ir/skeleton/skeleton.h"
#include "src/parse/spec.h"
namespace re2c {
static std::string make_name(const std::string &cond, uint32_t line)
{
std::ostringstream os;
os << "line" << line;
std::string name = os.str();
if (!cond.empty ())
{
name += "_";
name += cond;
}
return name;
}
smart_ptr<DFA> compile (Spec & spec, Output & output, const std::string & cond, uint32_t cunits)
{
const uint32_t line = output.source.get_block_line();
const std::string name = make_name(cond, line);
// The original set of code units (charset) might be very large.
// A common trick it is to split charset into disjoint character ranges
// and choose a representative of each range (we choose lower bound).
// The set of all representatives is the new (compacted) charset.
// Don't forget to include zero and upper bound, even if they
// do not explicitely apper in ranges.
std::set<uint32_t> bounds;
spec.re->split(bounds);
bounds.insert(0);
bounds.insert(cunits);
charset_t cs;
for (std::set<uint32_t>::const_iterator i = bounds.begin(); i != bounds.end(); ++i)
{
cs.push_back(*i);
}
nfa_t nfa(spec.re);
dfa_t dfa(nfa, cs, spec.rules);
// skeleton must be constructed after DFA construction
// but prior to any other DFA transformations
Skeleton *skeleton = new Skeleton(dfa, cs, spec.rules, name, cond, line);
minimization(dfa);
// find YYFILL states and calculate argument to YYFILL
std::vector<size_t> fill;
fillpoints(dfa, fill);
// ADFA stands for 'DFA with actions'
DFA *adfa = new DFA(dfa, fill, skeleton, cs, name, cond, line);
/*
* note [reordering DFA states]
*
* re2c-generated code depends on the order of states in DFA: simply
* flipping two states may change the output significantly.
* The order of states is affected by many factors, e.g.:
* - flipping left and right subtrees of alternative when constructing
* AST (also applies to iteration and counted repetition)
* - changing the order in which graph nodes are visited (applies to
* any intermediate representation: bytecode, NFA, DFA, etc.)
*
* To make the resulting code independent of such changes, we hereby
* reorder DFA states. The ordering scheme is very simple:
*
* Starting with DFA root, walk DFA nodes in breadth-first order.
* Child nodes are ordered accoding to the (alphabetically) first symbol
* leading to each node. Each node must be visited exactly once.
* Default state (NULL) is always the last state.
*/
adfa->reorder();
// skeleton is constructed, do further DFA transformations
adfa->prepare();
// finally gather overall DFA statistics
adfa->calc_stats();
// accumulate global statistics from this particular DFA
output.max_fill = std::max (output.max_fill, adfa->max_fill);
if (adfa->need_accept)
{
output.source.set_used_yyaccept ();
}
return make_smart_ptr(adfa);
}
} // namespace re2c

20
tools/re2c/src/ir/compile.h Normal file
View file

@ -0,0 +1,20 @@
#ifndef _RE2C_IR_COMPILE_
#define _RE2C_IR_COMPILE_
#include "src/util/c99_stdint.h"
#include <string>
#include "src/util/smart_ptr.h"
namespace re2c
{
class DFA;
struct Output;
struct Spec;
smart_ptr<DFA> compile (Spec & spec, Output & output, const std::string & cond, uint32_t cunits);
} // namespace re2c
#endif // _RE2C_IR_COMPILE_

197
tools/re2c/src/ir/dfa/determinization.cc Normal file
View file

@ -0,0 +1,197 @@
#include <algorithm>
#include <limits>
#include <map>
#include <set>
#include <vector>
#include "src/ir/dfa/dfa.h"
#include "src/ir/nfa/nfa.h"
#include "src/ir/regexp/regexp.h"
#include "src/ir/regexp/regexp_rule.h"
#include "src/ir/rule_rank.h"
#include "src/parse/rules.h"
#include "src/util/ord_hash_set.h"
#include "src/util/range.h"
namespace re2c
{
const size_t dfa_t::NIL = std::numeric_limits<size_t>::max();
/*
* note [marking DFA states]
*
* DFA state is a set of NFA states.
* However, DFA state includes not all NFA states that are in
* epsilon-closure (NFA states that have only epsilon-transitions
* and are not context of final states are omitted).
* The included states are called 'kernel' states.
*
* We mark visited NFA states during closure construction.
* These marks serve two purposes:
* - avoid loops in NFA
* - avoid duplication of NFA states in kernel
*
* Note that after closure construction:
* - all non-kernel states must be unmarked (these states are
* not stored in kernel and it is impossible to unmark them
* afterwards)
* - all kernel states must be marked (because we may later
* extend this kernel with epsilon-closure of another NFA
* state). Kernel states are unmarked later (before finding
* or adding DFA state).
*/
static nfa_state_t **closure(nfa_state_t **cP, nfa_state_t *n)
{
if (!n->mark)
{
n->mark = true;
switch (n->type)
{
case nfa_state_t::ALT:
cP = closure(cP, n->value.alt.out2);
cP = closure(cP, n->value.alt.out1);
n->mark = false;
break;
case nfa_state_t::CTX:
*(cP++) = n;
cP = closure(cP, n->value.ctx.out);
break;
default:
*(cP++) = n;
break;
}
}
return cP;
}
static size_t find_state
( nfa_state_t **kernel
, nfa_state_t **end
, ord_hash_set_t &kernels
)
{
// zero-sized kernel corresponds to default state
if (kernel == end)
{
return dfa_t::NIL;
}
// see note [marking DFA states]
for (nfa_state_t **p = kernel; p != end; ++p)
{
(*p)->mark = false;
}
// sort kernel states: we need this to get stable hash
// and to compare states with simple 'memcmp'
std::sort(kernel, end);
const size_t size = static_cast<size_t>(end - kernel) * sizeof(nfa_state_t*);
return kernels.insert(kernel, size);
}
dfa_t::dfa_t(const nfa_t &nfa, const charset_t &charset, rules_t &rules)
: states()
, nchars(charset.size() - 1) // (n + 1) bounds for n ranges
{
std::map<size_t, std::set<RuleOp*> > s2rules;
ord_hash_set_t kernels;
nfa_state_t **const buffer = new nfa_state_t*[nfa.size];
std::vector<std::vector<nfa_state_t*> > arcs(nchars);
find_state(buffer, closure(buffer, nfa.root), kernels);
for (size_t i = 0; i < kernels.size(); ++i)
{
dfa_state_t *s = new dfa_state_t;
states.push_back(s);
nfa_state_t **kernel;
const size_t kernel_size = kernels.deref<nfa_state_t*>(i, kernel);
for (size_t j = 0; j < kernel_size; ++j)
{
nfa_state_t *n = kernel[j];
switch (n->type)
{
case nfa_state_t::RAN:
{
nfa_state_t *m = n->value.ran.out;
size_t c = 0;
for (Range *r = n->value.ran.ran; r; r = r->next ())
{
for (; charset[c] != r->lower(); ++c);
for (; charset[c] != r->upper(); ++c)
{
arcs[c].push_back(m);
}
}
break;
}
case nfa_state_t::CTX:
s->ctx = true;
break;
case nfa_state_t::FIN:
s2rules[i].insert(n->value.fin.rule);
break;
default:
break;
}
}
s->arcs = new size_t[nchars];
for(size_t c = 0; c < nchars; ++c)
{
nfa_state_t **end = buffer;
for (std::vector<nfa_state_t*>::const_iterator j = arcs[c].begin(); j != arcs[c].end(); ++j)
{
end = closure(end, *j);
}
s->arcs[c] = find_state(buffer, end, kernels);
}
for(size_t c = 0; c < nchars; ++c)
{
arcs[c].clear();
}
}
delete[] buffer;
const size_t count = states.size();
for (size_t i = 0; i < count; ++i)
{
dfa_state_t *s = states[i];
std::set<RuleOp*> &rs = s2rules[i];
// for each final state: choose the rule with the smallest rank
for (std::set<RuleOp*>::const_iterator j = rs.begin(); j != rs.end(); ++j)
{
RuleOp *rule = *j;
if (!s->rule || rule->rank < s->rule->rank)
{
s->rule = rule;
}
}
// other rules are shadowed by the chosen rule
for (std::set<RuleOp*>::const_iterator j = rs.begin(); j != rs.end(); ++j)
{
RuleOp *rule = *j;
if (s->rule != rule)
{
rules[rule->rank].shadow.insert(s->rule->rank);
}
}
}
}
dfa_t::~dfa_t()
{
std::vector<dfa_state_t*>::iterator
i = states.begin(),
e = states.end();
for (; i != e; ++i)
{
delete *i;
}
}
} // namespace re2c

58
tools/re2c/src/ir/dfa/dfa.h Normal file
View file

@ -0,0 +1,58 @@
#ifndef _RE2C_IR_DFA_DFA_
#define _RE2C_IR_DFA_DFA_
#include "src/util/c99_stdint.h"
#include <vector>
#include "src/ir/regexp/regexp.h"
#include "src/parse/rules.h"
#include "src/util/forbid_copy.h"
namespace re2c
{
struct nfa_t;
class RuleOp;
struct dfa_state_t
{
size_t *arcs;
RuleOp *rule;
bool ctx;
dfa_state_t()
: arcs(NULL)
, rule(NULL)
, ctx(false)
{}
~dfa_state_t()
{
delete[] arcs;
}
FORBID_COPY(dfa_state_t);
};
struct dfa_t
{
static const size_t NIL;
std::vector<dfa_state_t*> states;
const size_t nchars;
dfa_t(const nfa_t &nfa, const charset_t &charset, rules_t &rules);
~dfa_t();
};
enum dfa_minimization_t
{
DFA_MINIMIZATION_TABLE,
DFA_MINIMIZATION_MOORE
};
void minimization(dfa_t &dfa);
void fillpoints(const dfa_t &dfa, std::vector<size_t> &fill);
} // namespace re2c
#endif // _RE2C_IR_DFA_DFA_

154
tools/re2c/src/ir/dfa/fillpoints.cc Normal file
View file

@ -0,0 +1,154 @@
#include <limits>
#include <stack>
#include <vector>
#include "src/ir/dfa/dfa.h"
namespace re2c
{
static const size_t INFINITI = std::numeric_limits<size_t>::max();
static const size_t UNDEFINED = INFINITI - 1;
static bool loopback(size_t node, size_t narcs, const size_t *arcs)
{
for (size_t i = 0; i < narcs; ++i)
{
if (arcs[i] == node)
{
return true;
}
}
return false;
}
/*
* node [finding strongly connected components of DFA]
*
* A slight modification of Tarjan's algorithm.
*
* The algorithm walks graph in deep-first order. It maintains a stack
* of nodes that have already been visited but haven't been assigned to
* SCC yet. For each node the algorithm calculates 'lowlink': index of
* the highest ancestor node reachable in one step from a descendant of
* the node. Lowlink is used to determine when a set of nodes should be
* popped off the stack into a new SCC.
*
* We use lowlink to hold different kinds of information:
* - values in range [0 .. stack size] mean that this node is on stack
* (link to a node with the smallest index reachable from this one)
* - UNDEFINED means that this node has not been visited yet
* - INFINITI means that this node has already been popped off stack
*
* We use stack size (rather than topological sort index) as unique index
* of a node on stack. This is safe because indices of nodes on stack are
* still unique and less than indices of nodes that have been popped off
* stack (INFINITI).
*
*/
static void scc(
const dfa_t &dfa,
std::stack<size_t> &stack,
std::vector<size_t> &lowlink,
std::vector<bool> &trivial,
size_t i)
{
const size_t link = stack.size();
lowlink[i] = link;
stack.push(i);
const size_t *arcs = dfa.states[i]->arcs;
for (size_t c = 0; c < dfa.nchars; ++c)
{
const size_t j = arcs[c];
if (j != dfa_t::NIL)
{
if (lowlink[j] == UNDEFINED)
{
scc(dfa, stack, lowlink, trivial, j);
}
if (lowlink[j] < lowlink[i])
{
lowlink[i] = lowlink[j];
}
}
}
if (lowlink[i] == link)
{
// SCC is non-trivial (has loops) iff it either:
// - consists of multiple nodes (they all must be interconnected)
// - consists of single node which loops back to itself
trivial[i] = i == stack.top()
&& !loopback(i, dfa.nchars, arcs);
size_t j;
do
{
j = stack.top();
stack.pop();
lowlink[j] = INFINITI;
}
while (j != i);
}
}
static void calc_fill(
const dfa_t &dfa,
const std::vector<bool> &trivial,
std::vector<size_t> &fill,
size_t i)
{
if (fill[i] == UNDEFINED)
{
fill[i] = 0;
const size_t *arcs = dfa.states[i]->arcs;
for (size_t c = 0; c < dfa.nchars; ++c)
{
const size_t j = arcs[c];
if (j != dfa_t::NIL)
{
calc_fill(dfa, trivial, fill, j);
size_t max = 1;
if (trivial[j])
{
max += fill[j];
}
if (max > fill[i])
{
fill[i] = max;
}
}
}
}
}
void fillpoints(const dfa_t &dfa, std::vector<size_t> &fill)
{
const size_t size = dfa.states.size();
// find DFA states that belong to non-trivial SCC
std::stack<size_t> stack;
std::vector<size_t> lowlink(size, UNDEFINED);
std::vector<bool> trivial(size, false);
scc(dfa, stack, lowlink, trivial, 0);
// for each DFA state, calculate YYFILL argument:
// maximal path length to the next YYFILL state
fill.resize(size, UNDEFINED);
calc_fill(dfa, trivial, fill, 0);
// The following states must trigger YYFILL:
// - inital state
// - all states in non-trivial SCCs
// for other states, reset YYFILL argument to zero
for (size_t i = 1; i < size; ++i)
{
if (trivial[i])
{
fill[i] = 0;
}
}
}
} // namespace re2c

252
tools/re2c/src/ir/dfa/minimization.cc Normal file
View file

@ -0,0 +1,252 @@
#include <string.h>
#include <utility>
#include <vector>
#include "src/conf/opt.h"
#include "src/ir/dfa/dfa.h"
#include "src/globals.h"
namespace re2c
{
class RuleOp;
/*
* note [DFA minimization: table filling algorithm]
*
* This algorithm is simple and slow; it's a reference implementation.
*
* The algorithm constructs (strictly lower triangular) boolean matrix
* indexed by DFA states. Each matrix cell (S1,S2) indicates if states
* S1 and S2 are distinguishable. Initialy states are distinguished
* according to their rule and context. One step of the algorithm
* updates the matrix as follows: each pair of states S1 and S2 is
* marked as distinguishable iff exist transitions from S1 and S2 on
* the same symbol that go to distinguishable states. The algorithm
* loops until the matrix stops changing.
*/
static void minimization_table(
size_t *part,
const std::vector<dfa_state_t*> &states,
size_t nchars)
{
const size_t count = states.size();
bool **tbl = new bool*[count];
tbl[0] = new bool[count * (count - 1) / 2];
for (size_t i = 0; i < count - 1; ++i)
{
tbl[i + 1] = tbl[i] + i;
}
for (size_t i = 0; i < count; ++i)
{
dfa_state_t *s1 = states[i];
for (size_t j = 0; j < i; ++j)
{
dfa_state_t *s2 = states[j];
tbl[i][j] = s1->ctx != s2->ctx
|| s1->rule != s2->rule;
}
}
for (bool loop = true; loop;)
{
loop = false;
for (size_t i = 0; i < count; ++i)
{
for (size_t j = 0; j < i; ++j)
{
if (!tbl[i][j])
{
for (size_t k = 0; k < nchars; ++k)
{
size_t oi = states[i]->arcs[k];
size_t oj = states[j]->arcs[k];
if (oi < oj)
{
std::swap(oi, oj);
}
if (oi != oj &&
(oi == dfa_t::NIL ||
oj == dfa_t::NIL ||
tbl[oi][oj]))
{
tbl[i][j] = true;
loop = true;
break;
}
}
}
}
}
}
for (size_t i = 0; i < count; ++i)
{
part[i] = i;
for (size_t j = 0; j < i; ++j)
{
if (!tbl[i][j])
{
part[i] = j;
break;
}
}
}
delete[] tbl[0];
delete[] tbl;
}
/*
* note [DFA minimization: Moore algorithm]
*
* The algorithm maintains partition of DFA states.
* Initial partition is coarse: states are distinguished according
* to their rule and context. Partition is gradually refined: each
* set of states is split into minimal number of subsets such that
* for all states in a subset transitions on the same symbol go to
* the same set of states.
* The algorithm loops until partition stops changing.
*/
static void minimization_moore(
size_t *part,
const std::vector<dfa_state_t*> &states,
size_t nchars)
{
const size_t count = states.size();
size_t *next = new size_t[count];
std::map<std::pair<RuleOp*, bool>, size_t> init;
for (size_t i = 0; i < count; ++i)
{
dfa_state_t *s = states[i];
std::pair<RuleOp*, bool> key(s->rule, s->ctx);
if (init.insert(std::make_pair(key, i)).second)
{
part[i] = i;
next[i] = dfa_t::NIL;
}
else
{
const size_t j = init[key];
part[i] = j;
next[i] = next[j];
next[j] = i;
}
}
size_t *out = new size_t[nchars * count];
size_t *diff = new size_t[count];
for (bool loop = true; loop;)
{
loop = false;
for (size_t i = 0; i < count; ++i)
{
if (i != part[i] || next[i] == dfa_t::NIL)
{
continue;
}
for (size_t j = i; j != dfa_t::NIL; j = next[j])
{
size_t *o = &out[j * nchars];
size_t *a = states[j]->arcs;
for (size_t c = 0; c < nchars; ++c)
{
o[c] = a[c] == dfa_t::NIL
? dfa_t::NIL
: part[a[c]];
}
}
size_t diff_count = 0;
for (size_t j = i; j != dfa_t::NIL;)
{
const size_t j_next = next[j];
size_t n = 0;
for (; n < diff_count; ++n)
{
size_t k = diff[n];
if (memcmp(&out[j * nchars],
&out[k * nchars],
nchars * sizeof(size_t)) == 0)
{
part[j] = k;
next[j] = next[k];
next[k] = j;
break;
}
}
if (n == diff_count)
{
diff[diff_count++] = j;
part[j] = j;
next[j] = dfa_t::NIL;
}
j = j_next;
}
loop |= diff_count > 1;
}
}
delete[] out;
delete[] diff;
delete[] next;
}
void minimization(dfa_t &dfa)
{
const size_t count = dfa.states.size();
size_t *part = new size_t[count];
switch (opts->dfa_minimization)
{
case DFA_MINIMIZATION_TABLE:
minimization_table(part, dfa.states, dfa.nchars);
break;
case DFA_MINIMIZATION_MOORE:
minimization_moore(part, dfa.states, dfa.nchars);
break;
}
size_t *compact = new size_t[count];
for (size_t i = 0, j = 0; i < count; ++i)
{
if (i == part[i])
{
compact[i] = j++;
}
}
size_t new_count = 0;
for (size_t i = 0; i < count; ++i)
{
dfa_state_t *s = dfa.states[i];
if (i == part[i])
{
size_t *arcs = s->arcs;
for (size_t c = 0; c < dfa.nchars; ++c)
{
if (arcs[c] != dfa_t::NIL)
{
arcs[c] = compact[part[arcs[c]]];
}
}
dfa.states[new_count++] = s;
}
else
{
delete s;
}
}
dfa.states.resize(new_count);
delete[] compact;
delete[] part;
}
} // namespace re2c

50
tools/re2c/src/ir/nfa/calc_size.cc Normal file
View file

@ -0,0 +1,50 @@
#include "src/util/c99_stdint.h"
#include "src/ir/regexp/regexp.h"
#include "src/ir/regexp/regexp_alt.h"
#include "src/ir/regexp/regexp_cat.h"
#include "src/ir/regexp/regexp_close.h"
#include "src/ir/regexp/regexp_match.h"
#include "src/ir/regexp/regexp_null.h"
#include "src/ir/regexp/regexp_rule.h"
namespace re2c
{
uint32_t AltOp::calc_size() const
{
return exp1->calc_size()
+ exp2->calc_size()
+ 1;
}
uint32_t CatOp::calc_size() const
{
return exp1->calc_size()
+ exp2->calc_size();
}
uint32_t CloseOp::calc_size() const
{
return exp->calc_size() + 1;
}
uint32_t MatchOp::calc_size() const
{
return 1;
}
uint32_t NullOp::calc_size() const
{
return 0;
}
uint32_t RuleOp::calc_size() const
{
const uint32_t n = ctx->calc_size();
return exp->calc_size()
+ (n > 0 ? n + 1 : 0)
+ 1;
}
} // end namespace re2c

72
tools/re2c/src/ir/nfa/nfa.cc Normal file
View file

@ -0,0 +1,72 @@
#include "src/ir/nfa/nfa.h"
#include "src/ir/regexp/regexp.h"
#include "src/ir/regexp/regexp_alt.h"
#include "src/ir/regexp/regexp_cat.h"
#include "src/ir/regexp/regexp_close.h"
#include "src/ir/regexp/regexp_match.h"
#include "src/ir/regexp/regexp_null.h"
#include "src/ir/regexp/regexp_rule.h"
namespace re2c {
nfa_t::nfa_t(RegExp *re)
: max_size(re->calc_size())
, size(0)
, states(new nfa_state_t[max_size])
, root(re->compile(*this, NULL))
{}
nfa_t::~nfa_t()
{
delete[] states;
}
nfa_state_t *AltOp::compile(nfa_t &nfa, nfa_state_t *t)
{
nfa_state_t *s = &nfa.states[nfa.size++];
s->alt(exp1->compile(nfa, t),
exp2->compile(nfa, t));
return s;
}
nfa_state_t *CatOp::compile(nfa_t &nfa, nfa_state_t *t)
{
nfa_state_t *s2 = exp2->compile(nfa, t);
nfa_state_t *s1 = exp1->compile(nfa, s2);
return s1;
}
nfa_state_t *CloseOp::compile(nfa_t &nfa, nfa_state_t *t)
{
nfa_state_t *s = &nfa.states[nfa.size++];
s->alt(t, exp->compile(nfa, s));
return s;
}
nfa_state_t *MatchOp::compile(nfa_t &nfa, nfa_state_t *t)
{
nfa_state_t *s = &nfa.states[nfa.size++];
s->ran(t, match);
return s;
}
nfa_state_t *NullOp::compile(nfa_t &, nfa_state_t *t)
{
return t;
}
nfa_state_t *RuleOp::compile(nfa_t &nfa, nfa_state_t *)
{
nfa_state_t *s3 = &nfa.states[nfa.size++];
s3->fin(this);
if (ctx->calc_size() > 0)
{
nfa_state_t *s2 = &nfa.states[nfa.size++];
s2->ctx(ctx->compile(nfa, s3));
s3 = s2;
}
nfa_state_t *s1 = exp->compile(nfa, s3);
return s1;
}
} // namespace re2c

90
tools/re2c/src/ir/nfa/nfa.h Normal file
View file

@ -0,0 +1,90 @@
#ifndef _RE2C_IR_NFA_NFA_
#define _RE2C_IR_NFA_NFA_
#include "src/util/c99_stdint.h"
#include "src/util/forbid_copy.h"
namespace re2c
{
class Range;
class RegExp;
class RuleOp;
struct nfa_state_t
{
enum type_t
{
ALT,
RAN,
CTX,
FIN
} type;
union
{
struct
{
nfa_state_t *out1;
nfa_state_t *out2;
} alt;
struct
{
nfa_state_t *out;
Range *ran;
} ran;
struct
{
nfa_state_t *out;
} ctx;
struct
{
RuleOp *rule;
} fin;
} value;
bool mark;
void alt(nfa_state_t *s1, nfa_state_t *s2)
{
type = ALT;
value.alt.out1 = s1;
value.alt.out2 = s2;
mark = false;
}
void ran(nfa_state_t *s, Range *r)
{
type = RAN;
value.ran.out = s;
value.ran.ran = r;
mark = false;
}
void ctx(nfa_state_t *s)
{
type = CTX;
value.ctx.out = s;
mark = false;
}
void fin(RuleOp *r)
{
type = FIN;
value.fin.rule = r;
mark = false;
}
};
struct nfa_t
{
const uint32_t max_size;
uint32_t size;
nfa_state_t *states;
nfa_state_t *root;
nfa_t(RegExp *re);
~nfa_t();
FORBID_COPY(nfa_t);
};
} // namespace re2c
#endif // _RE2C_IR_NFA_NFA_

49
tools/re2c/src/ir/nfa/split.cc Normal file
View file

@ -0,0 +1,49 @@
#include "src/util/c99_stdint.h"
#include <set>
#include "src/ir/regexp/regexp.h"
#include "src/ir/regexp/regexp_alt.h"
#include "src/ir/regexp/regexp_cat.h"
#include "src/ir/regexp/regexp_close.h"
#include "src/ir/regexp/regexp_match.h"
#include "src/ir/regexp/regexp_null.h"
#include "src/ir/regexp/regexp_rule.h"
#include "src/util/range.h"
namespace re2c {
void AltOp::split (std::set<uint32_t> & cs)
{
exp1->split (cs);
exp2->split (cs);
}
void CatOp::split (std::set<uint32_t> & cs)
{
exp1->split (cs);
exp2->split (cs);
}
void CloseOp::split (std::set<uint32_t> & cs)
{
exp->split (cs);
}
void MatchOp::split (std::set<uint32_t> & cs)
{
for (Range *r = match; r; r = r->next ())
{
cs.insert (r->lower ());
cs.insert (r->upper ());
}
}
void NullOp::split (std::set<uint32_t> &) {}
void RuleOp::split (std::set<uint32_t> & cs)
{
exp->split (cs);
ctx->split (cs);
}
} // namespace re2c

51
tools/re2c/src/ir/regexp/display.cc Normal file
View file

@ -0,0 +1,51 @@
#include <iostream>
#include "src/ir/regexp/regexp.h"
#include "src/ir/regexp/regexp_alt.h"
#include "src/ir/regexp/regexp_cat.h"
#include "src/ir/regexp/regexp_close.h"
#include "src/ir/regexp/regexp_match.h"
#include "src/ir/regexp/regexp_null.h"
#include "src/ir/regexp/regexp_rule.h"
namespace re2c
{
std::ostream & operator << (std::ostream & o, const RegExp & re)
{
re.display (o);
return o;
}
void AltOp::display (std::ostream & o) const
{
o << exp1 << "|" << exp2;
}
void CatOp::display (std::ostream & o) const
{
o << exp1 << exp2;
}
void CloseOp::display (std::ostream & o) const
{
o << exp << "+";
}
void MatchOp::display (std::ostream & o) const
{
o << match;
}
void NullOp::display (std::ostream & o) const
{
o << "_";
}
void RuleOp::display (std::ostream & o) const
{
o << exp << "/" << ctx << ";";
}
} // end namespace re2c

15
tools/re2c/src/ir/regexp/empty_class_policy.h Normal file
View file

@ -0,0 +1,15 @@
#ifndef _RE2C_IR_REGEXP_EMPTY_CLASS_POLICY_
#define _RE2C_IR_REGEXP_EMPTY_CLASS_POLICY_
namespace re2c {
enum empty_class_policy_t
{
EMPTY_CLASS_MATCH_EMPTY, // match on empty input
EMPTY_CLASS_MATCH_NONE, // fail to match on any input
EMPTY_CLASS_ERROR // compilation error
};
} // namespace re2c
#endif // _RE2C_IR_REGEXP_EMPTY_CLASS_POLICY_

31
tools/re2c/src/ir/regexp/encoding/case.h Normal file
View file

@ -0,0 +1,31 @@
#ifndef _RE2C_IR_REGEXP_ENCODING_CASE_
#define _RE2C_IR_REGEXP_ENCODING_CASE_
#include "src/util/c99_stdint.h"
namespace re2c {
// TODO: support non-ASCII encodings
bool is_alpha (uint32_t c);
uint32_t to_lower_unsafe (uint32_t c);
uint32_t to_upper_unsafe (uint32_t c);
inline bool is_alpha (uint32_t c)
{
return (c >= 'a' && c <= 'z')
|| (c >= 'A' && c <= 'Z');
}
inline uint32_t to_lower_unsafe (uint32_t c)
{
return c | 0x20u;
}
inline uint32_t to_upper_unsafe (uint32_t c)
{
return c & ~0x20u;
}
}
#endif // _RE2C_IR_REGEXP_ENCODING_CASE_

206
tools/re2c/src/ir/regexp/encoding/enc.cc Normal file
View file

@ -0,0 +1,206 @@
#include "src/ir/regexp/encoding/enc.h"
#include "src/util/range.h"
namespace re2c {
const uint32_t Enc::SURR_MIN = 0xD800;
const uint32_t Enc::SURR_MAX = 0xDFFF;
const uint32_t Enc::UNICODE_ERROR = 0xFFFD;
const uint32_t Enc::asc2ebc[256] =
{ /* Based on ISO 8859/1 and Code Page 37 */
0x00, 0x01, 0x02, 0x03, 0x37, 0x2d, 0x2e, 0x2f, 0x16, 0x05, 0x25, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f,
0x10, 0x11, 0x12, 0x13, 0x3c, 0x3d, 0x32, 0x26, 0x18, 0x19, 0x3f, 0x27, 0x1c, 0x1d, 0x1e, 0x1f,
0x40, 0x5a, 0x7f, 0x7b, 0x5b, 0x6c, 0x50, 0x7d, 0x4d, 0x5d, 0x5c, 0x4e, 0x6b, 0x60, 0x4b, 0x61,
0xf0, 0xf1, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7, 0xf8, 0xf9, 0x7a, 0x5e, 0x4c, 0x7e, 0x6e, 0x6f,
0x7c, 0xc1, 0xc2, 0xc3, 0xc4, 0xc5, 0xc6, 0xc7, 0xc8, 0xc9, 0xd1, 0xd2, 0xd3, 0xd4, 0xd5, 0xd6,
0xd7, 0xd8, 0xd9, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 0xe7, 0xe8, 0xe9, 0xba, 0xe0, 0xbb, 0xb0, 0x6d,
0x79, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96,
0x97, 0x98, 0x99, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7, 0xa8, 0xa9, 0xc0, 0x4f, 0xd0, 0xa1, 0x07,
0x20, 0x21, 0x22, 0x23, 0x24, 0x15, 0x06, 0x17, 0x28, 0x29, 0x2a, 0x2b, 0x2c, 0x09, 0x0a, 0x1b,
0x30, 0x31, 0x1a, 0x33, 0x34, 0x35, 0x36, 0x08, 0x38, 0x39, 0x3a, 0x3b, 0x04, 0x14, 0x3e, 0xff,
0x41, 0xaa, 0x4a, 0xb1, 0x9f, 0xb2, 0x6a, 0xb5, 0xbd, 0xb4, 0x9a, 0x8a, 0x5f, 0xca, 0xaf, 0xbc,
0x90, 0x8f, 0xea, 0xfa, 0xbe, 0xa0, 0xb6, 0xb3, 0x9d, 0xda, 0x9b, 0x8b, 0xb7, 0xb8, 0xb9, 0xab,
0x64, 0x65, 0x62, 0x66, 0x63, 0x67, 0x9e, 0x68, 0x74, 0x71, 0x72, 0x73, 0x78, 0x75, 0x76, 0x77,
0xac, 0x69, 0xed, 0xee, 0xeb, 0xef, 0xec, 0xbf, 0x80, 0xfd, 0xfe, 0xfb, 0xfc, 0xad, 0x8e, 0x59,
0x44, 0x45, 0x42, 0x46, 0x43, 0x47, 0x9c, 0x48, 0x54, 0x51, 0x52, 0x53, 0x58, 0x55, 0x56, 0x57,
0x8c, 0x49, 0xcd, 0xce, 0xcb, 0xcf, 0xcc, 0xe1, 0x70, 0xdd, 0xde, 0xdb, 0xdc, 0x8d, 0xae, 0xdf
};
const uint32_t Enc::ebc2asc[256] =
{ /* Based on ISO 8859/1 and Code Page 37 */
0x00, 0x01, 0x02, 0x03, 0x9c, 0x09, 0x86, 0x7f, 0x97, 0x8d, 0x8e, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f,
0x10, 0x11, 0x12, 0x13, 0x9d, 0x85, 0x08, 0x87, 0x18, 0x19, 0x92, 0x8f, 0x1c, 0x1d, 0x1e, 0x1f,
0x80, 0x81, 0x82, 0x83, 0x84, 0x0a, 0x17, 0x1b, 0x88, 0x89, 0x8a, 0x8b, 0x8c, 0x05, 0x06, 0x07,
0x90, 0x91, 0x16, 0x93, 0x94, 0x95, 0x96, 0x04, 0x98, 0x99, 0x9a, 0x9b, 0x14, 0x15, 0x9e, 0x1a,
0x20, 0xa0, 0xe2, 0xe4, 0xe0, 0xe1, 0xe3, 0xe5, 0xe7, 0xf1, 0xa2, 0x2e, 0x3c, 0x28, 0x2b, 0x7c,
0x26, 0xe9, 0xea, 0xeb, 0xe8, 0xed, 0xee, 0xef, 0xec, 0xdf, 0x21, 0x24, 0x2a, 0x29, 0x3b, 0xac,
0x2d, 0x2f, 0xc2, 0xc4, 0xc0, 0xc1, 0xc3, 0xc5, 0xc7, 0xd1, 0xa6, 0x2c, 0x25, 0x5f, 0x3e, 0x3f,
0xf8, 0xc9, 0xca, 0xcb, 0xc8, 0xcd, 0xce, 0xcf, 0xcc, 0x60, 0x3a, 0x23, 0x40, 0x27, 0x3d, 0x22,
0xd8, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0xab, 0xbb, 0xf0, 0xfd, 0xde, 0xb1,
0xb0, 0x6a, 0x6b, 0x6c, 0x6d, 0x6e, 0x6f, 0x70, 0x71, 0x72, 0xaa, 0xba, 0xe6, 0xb8, 0xc6, 0xa4,
0xb5, 0x7e, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78, 0x79, 0x7a, 0xa1, 0xbf, 0xd0, 0xdd, 0xfe, 0xae,
0x5e, 0xa3, 0xa5, 0xb7, 0xa9, 0xa7, 0xb6, 0xbc, 0xbd, 0xbe, 0x5b, 0x5d, 0xaf, 0xa8, 0xb4, 0xd7,
0x7b, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47, 0x48, 0x49, 0xad, 0xf4, 0xf6, 0xf2, 0xf3, 0xf5,
0x7d, 0x4a, 0x4b, 0x4c, 0x4d, 0x4e, 0x4f, 0x50, 0x51, 0x52, 0xb9, 0xfb, 0xfc, 0xf9, 0xfa, 0xff,
0x5c, 0xf7, 0x53, 0x54, 0x55, 0x56, 0x57, 0x58, 0x59, 0x5a, 0xb2, 0xd4, 0xd6, 0xd2, 0xd3, 0xd5,
0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, 0x38, 0x39, 0xb3, 0xdb, 0xdc, 0xd9, 0xda, 0x9f
};
/*
* Returns code point representation for current
* encoding with regard to current policy.
*
* Since code point is exacly specified by user,
* it is assumed that user considers it to be valid.
* We must check it.
*
* Returns false if this code point exceeds maximum
* or is forbidden by current policy, otherwise
* returns true. Overwrites code point.
*/
bool Enc::encode(uint32_t & c) const
{
if (c >= nCodePoints ())
{
return false;
}
switch (type_)
{
case ASCII:
return true;
case EBCDIC:
c = asc2ebc[c];
return true;
case UCS2:
case UTF16:
case UTF32:
case UTF8:
if (c < SURR_MIN || c > SURR_MAX)
return true;
else
{
switch (policy_)
{
case POLICY_FAIL:
return false;
case POLICY_SUBSTITUTE:
c = UNICODE_ERROR;
return true;
case POLICY_IGNORE:
return true;
}
}
}
return false; // to silence gcc warning
}
/*
* Returns original representation of code point.
* Assumes code point is valid (hence 'unsafe').
*/
uint32_t Enc::decodeUnsafe(uint32_t c) const
{
switch (type_)
{
case EBCDIC:
c = ebc2asc[c & 0xFF];
break;
case ASCII:
case UCS2:
case UTF16:
case UTF32:
case UTF8:
break;
}
return c;
}
/*
* Returns [l - h] range representation for current
* encoding with regard to current policy.
*
* Since range borders are exacly specified by user,
* it is assumed that user considers that all code
* points from this range are valid. re2c must check it.
*
* Returns NULL if range contains code points that
* exceed maximum or are forbidden by current policy,
* otherwise returns pointer to newly constructed range.
*/
Range * Enc::encodeRange(uint32_t l, uint32_t h) const
{
if (l >= nCodePoints () || h >= nCodePoints ())
{
return NULL;
}
Range * r = NULL;
switch (type_)
{
case ASCII:
r = Range::ran (l, h + 1);
break;
case EBCDIC:
{
const uint32_t el = asc2ebc[l];
r = Range::sym (el);
for (uint32_t c = l + 1; c <= h; ++c)
{
const uint32_t ec = asc2ebc[c];
r = Range::add (r, Range::sym (ec));
}
break;
}
case UCS2:
case UTF16:
case UTF32:
case UTF8:
r = Range::ran (l, h + 1);
if (l <= SURR_MAX && h >= SURR_MIN)
{
switch (policy_)
{
case POLICY_FAIL:
r = NULL;
break;
case POLICY_SUBSTITUTE:
{
Range * surrs = Range::ran (SURR_MIN, SURR_MAX + 1);
Range * error = Range::sym (UNICODE_ERROR);
r = Range::sub (r, surrs);
r = Range::add (r, error);
break;
}
case POLICY_IGNORE:
break;
}
}
break;
}
return r;
}
/*
* Returns full range representation for current encoding
* with regard to current policy.
*
* Since range is defined declaratively, re2c does
* all the necessary corrections 'for free'.
*
* Always succeeds, returns pointer to newly constructed
* range.
*/
Range * Enc::fullRange() const
{
Range * r = Range::ran (0, nCodePoints());
if (policy_ != POLICY_IGNORE)
{
Range * surrs = Range::ran (SURR_MIN, SURR_MAX + 1);
r = Range::sub (r, surrs);
}
return r;
}
} // namespace re2c

197
tools/re2c/src/ir/regexp/encoding/enc.h Normal file
View file

@ -0,0 +1,197 @@
#ifndef _RE2C_IR_REGEXP_ENCODING_ENC_
#define _RE2C_IR_REGEXP_ENCODING_ENC_
#include "src/util/c99_stdint.h"
namespace re2c {
class Range;
/*
* note [encodings]
*
* Each encoding defines two concepts:
*
* 1) Code point -- abstract number, which represents single encoding symbol.
* E.g., Unicode defines code points in the range [0 - 0x10FFFF] , so each
* Unicode encoding must be capable of representing 0x110000 code points.
*
* 2) Code unit -- the smallest unit of memory, which is used in the encoded
* text. One or more code units can be needed to represent a single code
* point, depending on the encoding. For each encoding, all code points
* either are represented with equal number of code units (fixed-length
* encodings), or with variable number of code units (variable-length
* encodings).
*
* +----------+------------------+-----------------------+-----------------+----------------+
* | encoding | code point range | code point size | code unit range | code unit size |
* +----------+------------------+-----------------------+-----------------+----------------+
* | ASCII | 0 - 0xFF | fixed, 1 byte | 0 - 0xFF | 1 byte |
* | EBCDIC | 0 - 0xFF | fixed, 1 byte | 0 - 0xFF | 1 byte |
* | UCS2 | 0 - 0xFFFF | fixed, 2 bytes | 0 - 0xFFFF | 2 bytes |
* | UTF16 | 0 - 0x10FFFF | variable, 2 - 4 bytes | 0 - 0xFFFF | 2 bytes |
* | UTF32 | 0 - 0x10FFFF | fixed, 4 bytes | 0 - 0x10FFFF | 4 bytes |
* | UTF8 | 0 - 0x10FFFF | variable, 1 - 4 bytes | 0 - 0xFF | 1 byte |
* +----------+------------------+-----------------------+-----------------+----------------+
*/
class Enc
{
public:
// Supported encodings.
enum type_t
{ ASCII
, EBCDIC
, UCS2
, UTF16
, UTF32
, UTF8
};
// What to do with invalid code points
enum policy_t
{ POLICY_FAIL
, POLICY_SUBSTITUTE
, POLICY_IGNORE
};
private:
static const uint32_t asc2ebc[256];
static const uint32_t ebc2asc[256];
static const uint32_t SURR_MIN;
static const uint32_t SURR_MAX;
static const uint32_t UNICODE_ERROR;
type_t type_;
policy_t policy_;
public:
Enc()
: type_ (ASCII)
, policy_ (POLICY_IGNORE)
{ }
static const char * name (type_t t);
bool operator != (const Enc & e) const { return type_ != e.type_; }
inline uint32_t nCodePoints() const;
inline uint32_t nCodeUnits() const;
inline uint32_t szCodePoint() const;
inline uint32_t szCodeUnit() const;
inline bool set(type_t t);
inline void unset(type_t);
inline type_t type () const;
inline void setPolicy(policy_t t);
bool encode(uint32_t & c) const;
uint32_t decodeUnsafe(uint32_t c) const;
Range * encodeRange(uint32_t l, uint32_t h) const;
Range * fullRange() const;
};
inline const char * Enc::name (type_t t)
{
switch (t)
{
case ASCII: return "ASCII";
case EBCDIC: return "EBCDIC";
case UTF8: return "UTF8";
case UCS2: return "USC2";
case UTF16: return "UTF16";
case UTF32: return "UTF32";
default: return "<bad encoding>";
}
}
inline uint32_t Enc::nCodePoints() const
{
switch (type_)
{
case ASCII:
case EBCDIC: return 0x100;
case UCS2: return 0x10000;
case UTF16:
case UTF32:
case UTF8:
default: return 0x110000;
}
}
inline uint32_t Enc::nCodeUnits() const
{
switch (type_)
{
case ASCII:
case EBCDIC:
case UTF8: return 0x100;
case UCS2:
case UTF16: return 0x10000;
case UTF32:
default: return 0x110000;
}
}
// returns *maximal* code point size for encoding
inline uint32_t Enc::szCodePoint() const
{
switch (type_)
{
case ASCII:
case EBCDIC: return 1;
case UCS2: return 2;
case UTF16:
case UTF32:
case UTF8:
default: return 4;
}
}
inline uint32_t Enc::szCodeUnit() const
{
switch (type_)
{
case ASCII:
case EBCDIC:
case UTF8: return 1;
case UCS2:
case UTF16: return 2;
case UTF32:
default: return 4;
}
}
inline bool Enc::set(type_t t)
{
if (type_ == t)
return true;
else if (type_ != ASCII)
return false;
else
{
type_ = t;
return true;
}
}
inline void Enc::unset(type_t t)
{
if (type_ == t)
type_ = ASCII;
}
inline Enc::type_t Enc::type () const
{
return type_;
}
inline void Enc::setPolicy(policy_t t)
{
policy_ = t;
}
} // namespace re2c
#endif // _RE2C_IR_REGEXP_ENCODING_ENC_

38
tools/re2c/src/ir/regexp/encoding/range_suffix.cc Normal file
View file

@ -0,0 +1,38 @@
#include "src/ir/regexp/encoding/range_suffix.h"
#include "src/ir/regexp/regexp.h"
#include "src/ir/regexp/regexp_match.h"
#include "src/util/range.h"
namespace re2c {
static RegExp * emit (RangeSuffix * p, RegExp * re);
free_list<RangeSuffix *> RangeSuffix::freeList;
RegExp * to_regexp (RangeSuffix * p)
{
return p
? emit (p, NULL)
: new MatchOp (NULL);
}
/*
* Build regexp from suffix tree.
*/
RegExp * emit(RangeSuffix * p, RegExp * re)
{
if (p == NULL)
return re;
else
{
RegExp * regexp = NULL;
for (; p != NULL; p = p->next)
{
RegExp * re1 = doCat(new MatchOp(Range::ran (p->l, p->h + 1)), re);
regexp = doAlt(regexp, emit(p->child, re1));
}
return regexp;
}
}
} // namespace re2c

39
tools/re2c/src/ir/regexp/encoding/range_suffix.h Normal file
View file

@ -0,0 +1,39 @@
#ifndef _RE2C_IR_REGEXP_ENCODING_RANGE_SUFFIX_
#define _RE2C_IR_REGEXP_ENCODING_RANGE_SUFFIX_
#include "src/util/c99_stdint.h"
#include <stddef.h> // NULL
#include "src/util/forbid_copy.h"
#include "src/util/free_list.h"
namespace re2c {
class RegExp;
struct RangeSuffix
{
static free_list<RangeSuffix *> freeList;
uint32_t l;
uint32_t h;
RangeSuffix * next;
RangeSuffix * child;
RangeSuffix (uint32_t lo, uint32_t hi)
: l (lo)
, h (hi)
, next (NULL)
, child (NULL)
{
freeList.insert(this);
}
FORBID_COPY (RangeSuffix);
};
RegExp * to_regexp (RangeSuffix * p);
} // namespace re2c
#endif // _RE2C_IR_REGEXP_ENCODING_RANGE_SUFFIX_

10
tools/re2c/src/ir/regexp/encoding/utf16/utf16.cc Normal file
View file

@ -0,0 +1,10 @@
#include "src/ir/regexp/encoding/utf16/utf16.h"
namespace re2c {
const uint32_t utf16::MAX_1WORD_RUNE = 0xFFFFu;
const uint32_t utf16::MIN_LEAD_SURR = 0xD800u;
const uint32_t utf16::MIN_TRAIL_SURR = 0xDC00u;
const uint32_t utf16::MAX_TRAIL_SURR = 0xDFFFu;
} // namespace re2c

37
tools/re2c/src/ir/regexp/encoding/utf16/utf16.h Normal file
View file

@ -0,0 +1,37 @@
#ifndef _RE2C_IR_REGEXP_ENCODING_UTF16_UTF16_
#define _RE2C_IR_REGEXP_ENCODING_UTF16_UTF16_
#include "src/util/c99_stdint.h"
namespace re2c {
class utf16
{
public:
typedef uint32_t rune;
static const uint32_t MAX_1WORD_RUNE;
static const uint32_t MIN_LEAD_SURR;
static const uint32_t MIN_TRAIL_SURR;
static const uint32_t MAX_TRAIL_SURR;
/* leading surrogate of UTF-16 symbol */
static inline uint32_t lead_surr(rune r);
/* trailing surrogate of UTF-16 symbol */
static inline uint32_t trail_surr(rune r);
};
inline uint32_t utf16::lead_surr(rune r)
{
return ((r - 0x10000u) / 0x400u) + MIN_LEAD_SURR;
}
inline uint32_t utf16::trail_surr(rune r)
{
return ((r - 0x10000u) % 0x400u) + MIN_TRAIL_SURR;
}
} // namespace re2c
#endif // _RE2C_IR_REGEXP_ENCODING_UTF16_UTF16_

146
tools/re2c/src/ir/regexp/encoding/utf16/utf16_range.cc Normal file
View file

@ -0,0 +1,146 @@
#include "src/ir/regexp/encoding/utf16/utf16_range.h"
#include "src/ir/regexp/encoding/range_suffix.h"
namespace re2c {
/*
* Add word range [w1-w2].
*/
void UTF16addContinuous1(RangeSuffix * & root, uint32_t l, uint32_t h)
{
RangeSuffix ** p = &root;
for (;;)
{
if (*p == NULL)
{
*p = new RangeSuffix(l, h);
break;
}
else if ((*p)->l == l && (*p)->h == h)
{
break;
}
else
p = &(*p)->next;
}
}
/*
* Now that we have catenation of word ranges [l1-h1],[l2-h2],
* we want to add it to existing range, merging suffixes on the fly.
*/
void UTF16addContinuous2(RangeSuffix * & root, uint32_t l_ld, uint32_t h_ld, uint32_t l_tr, uint32_t h_tr)
{
RangeSuffix ** p = &root;
for (;;)
{
if (*p == NULL)
{
*p = new RangeSuffix(l_tr, h_tr);
p = &(*p)->child;
break;
}
else if ((*p)->l == l_tr && (*p)->h == h_tr)
{
p = &(*p)->child;
break;
}
else
p = &(*p)->next;
}
for (;;)
{
if (*p == NULL)
{
*p = new RangeSuffix(l_ld, h_ld);
break;
}
else if ((*p)->l == l_ld && (*p)->h == h_ld)
{
break;
}
else
p = &(*p)->next;
}
}
/*
* Split range into sub-ranges that agree on leading surrogates.
*
* We have two Unicode runes, L and H, both map to UTF-16
* surrogate pairs 'L1 L2' and 'H1 H2'.
* We want to represent Unicode range [L - H] as a catenation
* of word ranges [L1 - H1],[L2 - H2].
*
* This is only possible if the following condition holds:
* if L1 /= H1, then L2 == 0xdc00 and H2 == 0xdfff.
* This condition ensures that:
* 1) all possible UTF-16 sequences between L and H are allowed
* 2) no word ranges [w1 - w2] appear, such that w1 > w2
*
* E.g.:
* [\U00010001-\U00010400] => [d800-d801],[dc01-dc00].
* The last word range, [dc01-dc00], is incorrect: its lower bound
* is greater than its upper bound. To fix this, we must split
* the original range into two sub-ranges:
* [\U00010001-\U000103ff] => [d800-d800],[dc01-dfff]
* [\U00010400-\U00010400] => [d801-d801],[dc00-dc00]
*
* This function finds all such 'points of discontinuity'
* and represents original range as alternation of continuous
* sub-ranges.
*/
void UTF16splitByContinuity(RangeSuffix * & root, uint32_t l_ld, uint32_t h_ld, uint32_t l_tr, uint32_t h_tr)
{
if (l_ld != h_ld)
{
if (l_tr > utf16::MIN_TRAIL_SURR)
{
UTF16splitByContinuity(root, l_ld, l_ld, l_tr, utf16::MAX_TRAIL_SURR);
UTF16splitByContinuity(root, l_ld + 1, h_ld, utf16::MIN_TRAIL_SURR, h_tr);
return;
}
if (h_tr < utf16::MAX_TRAIL_SURR)
{
UTF16splitByContinuity(root, l_ld, h_ld - 1, l_tr, utf16::MAX_TRAIL_SURR);
UTF16splitByContinuity(root, h_ld, h_ld, utf16::MIN_TRAIL_SURR, h_tr);
return;
}
}
UTF16addContinuous2(root, l_ld, h_ld, l_tr, h_tr);
}
/*
* Split range into sub-ranges, so that all runes in the same
* sub-range have equal length of UTF-16 sequence. E.g., full
* Unicode range [0-0x10FFFF] gets split into sub-ranges:
* [0 - 0xFFFF] (2-byte UTF-16 sequences)
* [0x10000 - 0x10FFFF] (4-byte UTF-16 sequences)
*/
void UTF16splitByRuneLength(RangeSuffix * & root, utf16::rune l, utf16::rune h)
{
if (l <= utf16::MAX_1WORD_RUNE)
{
if (h <= utf16::MAX_1WORD_RUNE)
{
UTF16addContinuous1(root, l, h);
}
else
{
UTF16addContinuous1(root, l, utf16::MAX_1WORD_RUNE);
const uint32_t h_ld = utf16::lead_surr(h);
const uint32_t h_tr = utf16::trail_surr(h);
UTF16splitByContinuity(root, utf16::MIN_LEAD_SURR, h_ld, utf16::MIN_TRAIL_SURR, h_tr);
}
}
else
{
const uint32_t l_ld = utf16::lead_surr(l);
const uint32_t l_tr = utf16::trail_surr(l);
const uint32_t h_ld = utf16::lead_surr(h);
const uint32_t h_tr = utf16::trail_surr(h);
UTF16splitByContinuity(root, l_ld, h_ld, l_tr, h_tr);
}
}
} // namespace re2c

19
tools/re2c/src/ir/regexp/encoding/utf16/utf16_range.h Normal file
View file

@ -0,0 +1,19 @@
#ifndef _RE2C_IR_REGEXP_ENCODING_UTF16_RANGE_
#define _RE2C_IR_REGEXP_ENCODING_UTF16_RANGE_
#include "src/util/c99_stdint.h"
#include "src/ir/regexp/encoding/utf16/utf16.h"
namespace re2c {
struct RangeSuffix;
void UTF16addContinuous1(RangeSuffix * & root, uint32_t l, uint32_t h);
void UTF16addContinuous2(RangeSuffix * & root, uint32_t l_ld, uint32_t h_ld, uint32_t l_tr, uint32_t h_tr);
void UTF16splitByContinuity(RangeSuffix * & root, uint32_t l_ld, uint32_t h_ld, uint32_t l_tr, uint32_t h_tr);
void UTF16splitByRuneLength(RangeSuffix * & root, utf16::rune l, utf16::rune h);
} // namespace re2c
#endif // _RE2C_IR_REGEXP_ENCODING_UTF16_RANGE_

38
tools/re2c/src/ir/regexp/encoding/utf16/utf16_regexp.cc Normal file
View file

@ -0,0 +1,38 @@
#include "src/util/c99_stdint.h"
#include "src/ir/regexp/encoding/utf16/utf16_regexp.h"
#include "src/ir/regexp/encoding/range_suffix.h"
#include "src/ir/regexp/encoding/utf16/utf16_range.h"
#include "src/ir/regexp/regexp_cat.h"
#include "src/ir/regexp/regexp_match.h"
#include "src/util/range.h"
namespace re2c {
RegExp * UTF16Symbol(utf16::rune r)
{
if (r <= utf16::MAX_1WORD_RUNE)
return new MatchOp(Range::sym (r));
else
{
const uint32_t ld = utf16::lead_surr(r);
const uint32_t tr = utf16::trail_surr(r);
return new CatOp(new MatchOp(Range::sym (ld)), new MatchOp(Range::sym (tr)));
}
}
/*
* Split Unicode character class {[l1, h1), ..., [lN, hN)} into
* ranges [l1, h1-1], ..., [lN, hN-1] and return alternation of
* them. We store partially built range in suffix tree, which
* allows to eliminate common suffixes while building.
*/
RegExp * UTF16Range(const Range * r)
{
RangeSuffix * root = NULL;
for (; r != NULL; r = r->next ())
UTF16splitByRuneLength(root, r->lower (), r->upper () - 1);
return to_regexp (root);
}
} // namespace re2c

16
tools/re2c/src/ir/regexp/encoding/utf16/utf16_regexp.h Normal file
View file

@ -0,0 +1,16 @@
#ifndef _RE2C_IR_REGEXP_ENCODING_UTF16_REGEXP_
#define _RE2C_IR_REGEXP_ENCODING_UTF16_REGEXP_
#include "src/ir/regexp/encoding/utf16/utf16.h"
namespace re2c {
class Range;
class RegExp;
RegExp * UTF16Symbol(utf16::rune r);
RegExp * UTF16Range(const Range * r);
} // namespace re2c
#endif // _RE2C_IR_REGEXP_ENCODING_UTF16_REGEXP_

84
tools/re2c/src/ir/regexp/encoding/utf8/utf8.cc Normal file
View file

@ -0,0 +1,84 @@
#include "src/ir/regexp/encoding/utf8/utf8.h"
namespace re2c {
const uint32_t utf8::ERROR = 0xFFFDu;
const utf8::rune utf8::MAX_1BYTE_RUNE = 0x7Fu;
const utf8::rune utf8::MAX_2BYTE_RUNE = 0x7FFu;
const utf8::rune utf8::MAX_3BYTE_RUNE = 0xFFFFu;
const utf8::rune utf8::MAX_4BYTE_RUNE = 0x10FFFFu;
const utf8::rune utf8::MAX_RUNE = utf8::MAX_4BYTE_RUNE;
const uint32_t utf8::PREFIX_1BYTE = 0u; // 0000 0000
const uint32_t utf8::INFIX = 0x80u; // 1000 0000
const uint32_t utf8::PREFIX_2BYTE = 0xC0u; // 1100 0000
const uint32_t utf8::PREFIX_3BYTE = 0xE0u; // 1110 0000
const uint32_t utf8::PREFIX_4BYTE = 0xF0u; // 1111 0000
const uint32_t utf8::SHIFT = 6u;
const uint32_t utf8::MASK = 0x3Fu; // 0011 1111
uint32_t utf8::rune_to_bytes(uint32_t *str, rune c)
{
// one byte sequence: 0-0x7F => 0xxxxxxx
if (c <= MAX_1BYTE_RUNE)
{
str[0] = PREFIX_1BYTE | c;
return 1;
}
// two byte sequence: 0x80-0x7FF => 110xxxxx 10xxxxxx
if (c <= MAX_2BYTE_RUNE)
{
str[0] = PREFIX_2BYTE | (c >> 1*SHIFT);
str[1] = INFIX | (c & MASK);
return 2;
}
// If the Rune is out of range, convert it to the error rune.
// Do this test here because the error rune encodes to three bytes.
// Doing it earlier would duplicate work, since an out of range
// Rune wouldn't have fit in one or two bytes.
if (c > MAX_RUNE)
c = ERROR;
// three byte sequence: 0x800 - 0xFFFF => 1110xxxx 10xxxxxx 10xxxxxx
if (c <= MAX_3BYTE_RUNE)
{
str[0] = PREFIX_3BYTE | (c >> 2*SHIFT);
str[1] = INFIX | ((c >> 1*SHIFT) & MASK);
str[2] = INFIX | (c & MASK);
return 3;
}
// four byte sequence (21-bit value):
// 0x10000 - 0x1FFFFF => 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
str[0] = PREFIX_4BYTE | (c >> 3*SHIFT);
str[1] = INFIX | ((c >> 2*SHIFT) & MASK);
str[2] = INFIX | ((c >> 1*SHIFT) & MASK);
str[3] = INFIX | (c & MASK);
return 4;
}
uint32_t utf8::rune_length(rune r)
{
if (r <= MAX_2BYTE_RUNE)
return r <= MAX_1BYTE_RUNE ? 1 : 2;
else
return r <= MAX_3BYTE_RUNE ? 3 : 4;
}
utf8::rune utf8::max_rune(uint32_t i)
{
switch (i)
{
case 1: return MAX_1BYTE_RUNE;
case 2: return MAX_2BYTE_RUNE;
case 3: return MAX_3BYTE_RUNE;
case 4: return MAX_4BYTE_RUNE;
default: return ERROR;
}
}
} // namespace re2c

48
tools/re2c/src/ir/regexp/encoding/utf8/utf8.h Normal file
View file

@ -0,0 +1,48 @@
#ifndef _RE2C_IR_REGEXP_ENCODING_UTF8_UTF8_
#define _RE2C_IR_REGEXP_ENCODING_UTF8_UTF8_
#include "src/util/c99_stdint.h"
namespace re2c {
class utf8
{
public:
typedef uint32_t rune;
// maximum characters per rune
// enum instead of static const member because of [-Wvla]
enum { MAX_RUNE_LENGTH = 4u };
// decoding error
static const uint32_t ERROR;
// maximal runes for each rune length
static const rune MAX_1BYTE_RUNE;
static const rune MAX_2BYTE_RUNE;
static const rune MAX_3BYTE_RUNE;
static const rune MAX_4BYTE_RUNE;
static const rune MAX_RUNE;
static const uint32_t PREFIX_1BYTE;
static const uint32_t INFIX;
static const uint32_t PREFIX_2BYTE;
static const uint32_t PREFIX_3BYTE;
static const uint32_t PREFIX_4BYTE;
static const uint32_t SHIFT;
static const uint32_t MASK;
// UTF-8 bytestring for given Unicode rune
static uint32_t rune_to_bytes(uint32_t * s, rune r);
// length of UTF-8 bytestring for given Unicode rune
static uint32_t rune_length(rune r);
// maximal Unicode rune with given length of UTF-8 bytestring
static rune max_rune(uint32_t i);
};
} // namespace re2c
#endif // _RE2C_IR_REGEXP_ENCODING_UTF8_UTF8_

112
tools/re2c/src/ir/regexp/encoding/utf8/utf8_range.cc Normal file
View file

@ -0,0 +1,112 @@
#include "src/ir/regexp/encoding/utf8/utf8_range.h"
#include "src/ir/regexp/encoding/range_suffix.h"
namespace re2c {
/*
* Now that we have catenation of byte ranges [l1-h1]...[lN-hN],
* we want to add it to existing range, merging suffixes on the fly.
*/
void UTF8addContinuous(RangeSuffix * & root, utf8::rune l, utf8::rune h, uint32_t n)
{
uint32_t lcs[utf8::MAX_RUNE_LENGTH];
uint32_t hcs[utf8::MAX_RUNE_LENGTH];
utf8::rune_to_bytes(lcs, l);
utf8::rune_to_bytes(hcs, h);
RangeSuffix ** p = &root;
for (uint32_t i = 1; i <= n; ++i)
{
const uint32_t lc = lcs[n - i];
const uint32_t hc = hcs[n - i];
for (;;)
{
if (*p == NULL)
{
*p = new RangeSuffix(lc, hc);
p = &(*p)->child;
break;
}
else if ((*p)->l == lc && (*p)->h == hc)
{
p = &(*p)->child;
break;
}
else
p = &(*p)->next;
}
}
}
/*
* Split range into sub-ranges that agree on leading bytes.
*
* We have two Unicode runes of equal length, L and H, which
* map to UTF-8 sequences 'L_1 ... L_n' and 'H_1 ... H_n'.
* We want to represent Unicode range [L - H] as a catenation
* of byte ranges [L_1 - H_1], ..., [L_n - H_n].
*
* This is only possible if for all i > 1:
* if L_i /= H_i, then L_(i+1) == 0x80 and H_(i+1) == 0xbf.
* This condition ensures that:
* 1) all possible UTF-8 sequences between L and H are allowed
* 2) no byte ranges [b1 - b2] appear, such that b1 > b2
*
* E.g.:
* [\U000e0031-\U000e0043] => [f3-f3],[a0-a0],[80-81],[b1-83].
* The last byte range, [b1-83], is incorrect: its lower bound
* is greater than its upper bound. To fix this, we must split
* the original range into two sub-ranges:
* [\U000e0031-\U000e003f] => [f3-f3],[a0-a0],[80-80],[b1-bf]
* [\U000e0040-\U000e0043] => [f3-f3],[a0-a0],[81-81],[80-83]
*
* This function finds all such 'points of discontinuity'
* and represents original range as alternation of continuous
* sub-ranges.
*/
void UTF8splitByContinuity(RangeSuffix * & root, utf8::rune l, utf8::rune h, uint32_t n)
{
for (uint32_t i = 1; i < n; ++i)
{
uint32_t m = (1u << (6u * i)) - 1u; // last i bytes of a UTF-8 sequence
if ((l & ~m) != (h & ~m))
{
if ((l & m) != 0)
{
UTF8splitByContinuity(root, l, l | m, n);
UTF8splitByContinuity(root, (l | m) + 1, h, n);
return;
}
if ((h & m) != m)
{
UTF8splitByContinuity(root, l, (h & ~m) - 1, n);
UTF8splitByContinuity(root, h & ~m, h, n);
return;
}
}
}
UTF8addContinuous(root, l, h, n);
}
/*
* Split range into sub-ranges, so that all runes in the same
* sub-range have equal length of UTF-8 sequence. E.g., full
* Unicode range [0-0x10FFFF] gets split into sub-ranges:
* [0 - 0x7F] (1-byte UTF-8 sequences)
* [0x80 - 0x7FF] (2-byte UTF-8 sequences)
* [0x800 - 0xFFFF] (3-byte UTF-8 sequences)
* [0x10000 - 0x10FFFF] (4-byte UTF-8 sequences)
*/
void UTF8splitByRuneLength(RangeSuffix * & root, utf8::rune l, utf8::rune h)
{
const uint32_t nh = utf8::rune_length(h);
for (uint32_t nl = utf8::rune_length(l); nl < nh; ++nl)
{
utf8::rune r = utf8::max_rune(nl);
UTF8splitByContinuity(root, l, r, nl);
l = r + 1;
}
UTF8splitByContinuity(root, l, h, nh);
}
} // namespace re2c

18
tools/re2c/src/ir/regexp/encoding/utf8/utf8_range.h Normal file
View file

@ -0,0 +1,18 @@
#ifndef _RE2C_IR_REGEXP_ENCODING_UTF8_RANGE_
#define _RE2C_IR_REGEXP_ENCODING_UTF8_RANGE_
#include "src/util/c99_stdint.h"
#include "src/ir/regexp/encoding/utf8/utf8.h"
namespace re2c {
struct RangeSuffix;
void UTF8addContinuous(RangeSuffix * & p, utf8::rune l, utf8::rune h, uint32_t n);
void UTF8splitByContinuity(RangeSuffix * & p, utf8::rune l, utf8::rune h, uint32_t n);
void UTF8splitByRuneLength(RangeSuffix * & p, utf8::rune l, utf8::rune h);
} // namespace re2c
#endif // _RE2C_IR_REGEXP_ENCODING_UTF8_RANGE_

36
tools/re2c/src/ir/regexp/encoding/utf8/utf8_regexp.cc Normal file
View file

@ -0,0 +1,36 @@
#include "src/util/c99_stdint.h"
#include "src/ir/regexp/encoding/utf8/utf8_regexp.h"
#include "src/ir/regexp/encoding/range_suffix.h"
#include "src/ir/regexp/encoding/utf8/utf8_range.h"
#include "src/ir/regexp/regexp_cat.h"
#include "src/ir/regexp/regexp_match.h"
#include "src/util/range.h"
namespace re2c {
RegExp * UTF8Symbol(utf8::rune r)
{
uint32_t chars[utf8::MAX_RUNE_LENGTH];
const uint32_t chars_count = utf8::rune_to_bytes(chars, r);
RegExp * re = new MatchOp(Range::sym (chars[0]));
for (uint32_t i = 1; i < chars_count; ++i)
re = new CatOp(re, new MatchOp(Range::sym (chars[i])));
return re;
}
/*
* Split Unicode character class {[l1, h1), ..., [lN, hN)} into
* ranges [l1, h1-1], ..., [lN, hN-1] and return alternation of
* them. We store partially built range in suffix tree, which
* allows to eliminate common suffixes while building.
*/
RegExp * UTF8Range(const Range * r)
{
RangeSuffix * root = NULL;
for (; r != NULL; r = r->next ())
UTF8splitByRuneLength(root, r->lower (), r->upper () - 1);
return to_regexp (root);
}
} // namespace re2c

16
tools/re2c/src/ir/regexp/encoding/utf8/utf8_regexp.h Normal file
View file

@ -0,0 +1,16 @@
#ifndef _RE2C_IR_REGEXP_ENCODING_UTF8_REGEXP_
#define _RE2C_IR_REGEXP_ENCODING_UTF8_REGEXP_
#include "src/ir/regexp/encoding/utf8/utf8.h"
namespace re2c {
class Range;
class RegExp;
RegExp * UTF8Symbol(utf8::rune r);
RegExp * UTF8Range(const Range * r);
} // namespace re2c
#endif // _RE2C_IR_REGEXP_ENCODING_UTF8_REGEXP_

55
tools/re2c/src/ir/regexp/fixed_length.cc Normal file
View file

@ -0,0 +1,55 @@
#include "src/util/c99_stdint.h"
#include "src/ir/regexp/regexp.h"
#include "src/ir/regexp/regexp_alt.h"
#include "src/ir/regexp/regexp_cat.h"
#include "src/ir/regexp/regexp_match.h"
#include "src/ir/regexp/regexp_null.h"
namespace re2c
{
uint32_t RegExp::fixedLength ()
{
return ~0u;
}
uint32_t AltOp::fixedLength ()
{
uint32_t l1 = exp1->fixedLength ();
uint32_t l2 = exp1->fixedLength ();
if (l1 != l2 || l1 == ~0u)
{
return ~0u;
}
return l1;
}
uint32_t CatOp::fixedLength ()
{
const uint32_t l1 = exp1->fixedLength ();
if (l1 != ~0u)
{
const uint32_t l2 = exp2->fixedLength ();
if (l2 != ~0u)
{
return l1 + l2;
}
}
return ~0u;
}
uint32_t MatchOp::fixedLength ()
{
return 1;
}
uint32_t NullOp::fixedLength ()
{
return 0;
}
} // end namespace re2c

241
tools/re2c/src/ir/regexp/regexp.cc Normal file
View file

@ -0,0 +1,241 @@
#include <stddef.h>
#include "src/conf/opt.h"
#include "src/conf/warn.h"
#include "src/globals.h"
#include "src/ir/regexp/empty_class_policy.h"
#include "src/ir/regexp/encoding/case.h"
#include "src/ir/regexp/encoding/enc.h"
#include "src/ir/regexp/encoding/utf16/utf16_regexp.h"
#include "src/ir/regexp/encoding/utf8/utf8_regexp.h"
#include "src/ir/regexp/regexp.h"
#include "src/ir/regexp/regexp_alt.h"
#include "src/ir/regexp/regexp_cat.h"
#include "src/ir/regexp/regexp_close.h"
#include "src/ir/regexp/regexp_match.h"
#include "src/ir/regexp/regexp_null.h"
#include "src/parse/scanner.h"
#include "src/util/range.h"
namespace re2c
{
static MatchOp * merge (MatchOp * m1, MatchOp * m2);
free_list<RegExp*> RegExp::vFreeList;
RegExp * doAlt (RegExp * e1, RegExp * e2)
{
if (!e1)
{
return e2;
}
if (!e2)
{
return e1;
}
return new AltOp (e1, e2);
}
RegExp * mkAlt (RegExp * e1, RegExp * e2)
{
AltOp * a;
MatchOp * m1;
MatchOp * m2;
a = dynamic_cast<AltOp*> (e1);
if (a != NULL)
{
m1 = dynamic_cast<MatchOp*> (a->exp1);
if (m1 != NULL)
{
e1 = a->exp2;
}
}
else
{
m1 = dynamic_cast<MatchOp*> (e1);
if (m1 != NULL)
{
e1 = NULL;
}
}
a = dynamic_cast<AltOp*> (e2);
if (a != NULL)
{
m2 = dynamic_cast<MatchOp*> (a->exp1);
if (m2 != NULL)
{
e2 = a->exp2;
}
}
else
{
m2 = dynamic_cast<MatchOp*> (e2);
if (m2 != NULL)
{
e2 = NULL;
}
}
return doAlt (merge (m1, m2), doAlt (e1, e2));
}
MatchOp * merge (MatchOp * m1, MatchOp * m2)
{
if (!m1)
{
return m2;
}
if (!m2)
{
return m1;
}
MatchOp * m = new MatchOp (Range::add (m1->match, m2->match));
return m;
}
RegExp * doCat (RegExp * e1, RegExp * e2)
{
if (!e1)
{
return e2;
}
if (!e2)
{
return e1;
}
return new CatOp (e1, e2);
}
RegExp *Scanner::schr(uint32_t c) const
{
if (!opts->encoding.encode(c)) {
fatalf("Bad code point: '0x%X'", c);
}
switch (opts->encoding.type ()) {
case Enc::UTF16: return UTF16Symbol(c);
case Enc::UTF8: return UTF8Symbol(c);
default: return new MatchOp(Range::sym(c));
}
}
RegExp *Scanner::ichr(uint32_t c) const
{
if (is_alpha(c)) {
RegExp *l = schr(to_lower_unsafe(c));
RegExp *u = schr(to_upper_unsafe(c));
return mkAlt(l, u);
} else {
return schr(c);
}
}
RegExp *Scanner::cls(Range *r) const
{
if (!r)
{
switch (opts->empty_class_policy)
{
case EMPTY_CLASS_MATCH_EMPTY:
warn.empty_class (get_line ());
return new NullOp;
case EMPTY_CLASS_MATCH_NONE:
warn.empty_class (get_line ());
break;
case EMPTY_CLASS_ERROR:
fatal ("empty character class");
break;
}
}
switch (opts->encoding.type ())
{
case Enc::UTF16: return UTF16Range(r);
case Enc::UTF8: return UTF8Range(r);
default: return new MatchOp(r);
}
}
RegExp * Scanner::mkDiff (RegExp * e1, RegExp * e2) const
{
MatchOp * m1 = dynamic_cast<MatchOp *> (e1);
MatchOp * m2 = dynamic_cast<MatchOp *> (e2);
if (m1 == NULL || m2 == NULL)
{
fatal("can only difference char sets");
}
Range * r = Range::sub (m1->match, m2->match);
return cls(r);
}
RegExp * Scanner::mkDot() const
{
Range * full = opts->encoding.fullRange();
uint32_t c = '\n';
if (!opts->encoding.encode(c))
fatalf("Bad code point: '0x%X'", c);
Range * ran = Range::sym (c);
Range * inv = Range::sub (full, ran);
return cls(inv);
}
/*
* Create a byte range that includes all possible input characters.
* This may include characters, which do not map to any valid symbol
* in current encoding. For encodings, which directly map symbols to
* input characters (ASCII, EBCDIC, UTF-32), it equals [^]. For other
* encodings (UTF-16, UTF-8), [^] and this range are different.
*
* Also note that default range doesn't respect encoding policy
* (the way invalid code points are treated).
*/
RegExp * Scanner::mkDefault() const
{
Range * def = Range::ran (0, opts->encoding.nCodeUnits());
return new MatchOp(def);
}
/*
* note [counted repetition expansion]
*
* r{0} ;;= <empty regexp>
* r{n} ::= r{n-1} r
* r{n,m} ::= r{n} (r{0} | ... | r{m-n})
* r{n,} ::= r{n} r*
*/
// see note [counted repetition expansion]
RegExp * repeat (RegExp * e, uint32_t n)
{
RegExp * r = NULL;
for (uint32_t i = 0; i < n; ++i)
{
r = doCat (r, e);
}
return r;
}
// see note [counted repetition expansion]
RegExp * repeat_from_to (RegExp * e, uint32_t n, uint32_t m)
{
RegExp * r1 = repeat (e, n);
RegExp * r2 = NULL;
for (uint32_t i = n; i < m; ++i)
{
r2 = mkAlt (new NullOp, doCat (e, r2));
}
return doCat (r1, r2);
}
// see note [counted repetition expansion]
RegExp * repeat_from (RegExp * e, uint32_t n)
{
RegExp * r1 = repeat (e, n);
RegExp * r2 = new CloseOp (e);
return doCat (r1, r2);
}
} // namespace re2c

52
tools/re2c/src/ir/regexp/regexp.h Normal file
View file

@ -0,0 +1,52 @@
#ifndef _RE2C_IR_REGEXP_REGEXP_
#define _RE2C_IR_REGEXP_REGEXP_
#include "src/util/c99_stdint.h"
#include <iosfwd>
#include <set>
#include <vector>
#include "src/util/free_list.h"
#include "src/util/forbid_copy.h"
namespace re2c
{
struct nfa_state_t;
struct nfa_t;
typedef std::vector<uint32_t> charset_t;
class RegExp
{
public:
static free_list <RegExp *> vFreeList;
inline RegExp ()
{
vFreeList.insert (this);
}
inline virtual ~RegExp ()
{
vFreeList.erase (this);
}
virtual void split (std::set<uint32_t> &) = 0;
virtual uint32_t calc_size() const = 0;
virtual uint32_t fixedLength ();
virtual nfa_state_t *compile(nfa_t &nfa, nfa_state_t *n) = 0;
virtual void display (std::ostream &) const = 0;
friend std::ostream & operator << (std::ostream & o, const RegExp & re);
FORBID_COPY (RegExp);
};
RegExp * doAlt (RegExp * e1, RegExp * e2);
RegExp * mkAlt (RegExp * e1, RegExp * e2);
RegExp * doCat (RegExp * e1, RegExp * e2);
RegExp * repeat (RegExp * e, uint32_t n);
RegExp * repeat_from_to (RegExp * e, uint32_t n, uint32_t m);
RegExp * repeat_from (RegExp * e, uint32_t n);
} // end namespace re2c
#endif // _RE2C_IR_REGEXP_REGEXP_

31
tools/re2c/src/ir/regexp/regexp_alt.h Normal file
View file

@ -0,0 +1,31 @@
#ifndef _RE2C_IR_REGEXP_REGEXP_ALT_
#define _RE2C_IR_REGEXP_REGEXP_ALT_
#include "src/ir/regexp/regexp.h"
namespace re2c
{
class AltOp: public RegExp
{
RegExp * exp1;
RegExp * exp2;
public:
inline AltOp (RegExp * e1, RegExp * e2)
: exp1 (e1)
, exp2 (e2)
{}
void split (std::set<uint32_t> &);
uint32_t calc_size() const;
uint32_t fixedLength ();
nfa_state_t *compile(nfa_t &nfa, nfa_state_t *n);
void display (std::ostream & o) const;
friend RegExp * mkAlt (RegExp *, RegExp *);
FORBID_COPY (AltOp);
};
} // end namespace re2c
#endif // _RE2C_IR_REGEXP_REGEXP_ALT_

30
tools/re2c/src/ir/regexp/regexp_cat.h Normal file
View file

@ -0,0 +1,30 @@
#ifndef _RE2C_IR_REGEXP_REGEXP_CAT_
#define _RE2C_IR_REGEXP_REGEXP_CAT_
#include "src/ir/regexp/regexp.h"
namespace re2c
{
class CatOp: public RegExp
{
RegExp * exp1;
RegExp * exp2;
public:
inline CatOp (RegExp * e1, RegExp * e2)
: exp1 (e1)
, exp2 (e2)
{}
void split (std::set<uint32_t> &);
uint32_t calc_size() const;
uint32_t fixedLength ();
nfa_state_t *compile(nfa_t &nfa, nfa_state_t *n);
void display (std::ostream & o) const;
FORBID_COPY (CatOp);
};
} // end namespace re2c
#endif // _RE2C_IR_REGEXP_REGEXP_CAT_

27
tools/re2c/src/ir/regexp/regexp_close.h Normal file
View file

@ -0,0 +1,27 @@
#ifndef _RE2C_IR_REGEXP_REGEXP_CLOSE_
#define _RE2C_IR_REGEXP_REGEXP_CLOSE_
#include "src/ir/regexp/regexp.h"
namespace re2c
{
class CloseOp: public RegExp
{
RegExp * exp;
public:
inline CloseOp (RegExp * e)
: exp (e)
{}
void split (std::set<uint32_t> &);
uint32_t calc_size() const;
nfa_state_t *compile(nfa_t &nfa, nfa_state_t *n);
void display (std::ostream & o) const;
FORBID_COPY (CloseOp);
};
} // end namespace re2c
#endif // _RE2C_IR_REGEXP_REGEXP_CLOSE_

29
tools/re2c/src/ir/regexp/regexp_match.h Normal file
View file

@ -0,0 +1,29 @@
#ifndef _RE2C_IR_REGEXP_REGEXP_MATCH_
#define _RE2C_IR_REGEXP_REGEXP_MATCH_
#include "src/ir/regexp/regexp.h"
#include "src/util/range.h"
namespace re2c
{
class MatchOp: public RegExp
{
public:
Range * match;
inline MatchOp (Range * m)
: match (m)
{}
void split (std::set<uint32_t> &);
uint32_t calc_size() const;
uint32_t fixedLength ();
nfa_state_t *compile(nfa_t &nfa, nfa_state_t *n);
void display (std::ostream & o) const;
FORBID_COPY (MatchOp);
};
} // end namespace re2c
#endif // _RE2C_IR_REGEXP_REGEXP_MATCH_

21
tools/re2c/src/ir/regexp/regexp_null.h Normal file
View file

@ -0,0 +1,21 @@
#ifndef _RE2C_IR_REGEXP_REGEXP_NULL_
#define _RE2C_IR_REGEXP_REGEXP_NULL_
#include "src/ir/regexp/regexp.h"
namespace re2c
{
class NullOp: public RegExp
{
public:
void split (std::set<uint32_t> &);
uint32_t calc_size() const;
uint32_t fixedLength ();
nfa_state_t *compile(nfa_t &nfa, nfa_state_t *n);
void display (std::ostream & o) const;
};
} // end namespace re2c
#endif // _RE2C_IR_REGEXP_REGEXP_NULL_

Some files were not shown because too many files have changed in this diff Show more