@port-labs/jq-node-bindings 0.0.4 → 0.0.6
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/binding.gyp +3 -3
- package/configure +8 -6
- package/deps/jq/.travis.yml +184 -35
- package/deps/jq/AUTHORS +31 -30
- package/deps/jq/COPYING +2 -2
- package/deps/jq/KEYS +8 -0
- package/deps/jq/Makefile.am +85 -70
- package/deps/jq/NEWS +5 -4
- package/deps/jq/README.md +18 -9
- package/deps/jq/appveyor.yml +53 -0
- package/deps/jq/compile-ios.sh +5 -5
- package/deps/jq/config/m4/check-math-func.m4 +7 -2
- package/deps/jq/configure.ac +130 -68
- package/deps/jq/docs/Rakefile +14 -135
- package/deps/jq/docs/Rakefile.manual +49 -0
- package/deps/jq/docs/Rakefile.website +76 -0
- package/deps/jq/docs/content/2.download/default.yml +115 -39
- package/deps/jq/docs/content/3.manual/manual.yml +641 -231
- package/deps/jq/docs/content/3.manual/v1.3/manual.yml +130 -130
- package/deps/jq/docs/content/3.manual/v1.4/manual.yml +160 -160
- package/deps/jq/docs/content/3.manual/v1.5/manual.yml +2878 -0
- package/deps/jq/docs/content/3.manual/v1.6/manual.yml +3287 -0
- package/deps/jq/docs/content/index/index.yml +28 -6
- package/deps/jq/docs/public/.htaccess +2 -2
- package/deps/jq/docs/public/css/base.scss +141 -59
- package/deps/jq/docs/public/js/manual-search.js +52 -0
- package/deps/jq/docs/site.yml +1 -10
- package/deps/jq/docs/templates/default.liquid +10 -10
- package/deps/jq/docs/templates/index.liquid +45 -32
- package/deps/jq/docs/templates/manual.liquid +67 -90
- package/deps/jq/docs/templates/shared/_footer.liquid +10 -5
- package/deps/jq/docs/templates/shared/_head.liquid +17 -12
- package/deps/jq/docs/templates/shared/_navbar.liquid +27 -0
- package/deps/jq/jq.1.prebuilt +523 -139
- package/deps/jq/modules/oniguruma/AUTHORS +1 -0
- package/deps/jq/modules/oniguruma/CMakeLists.txt +86 -0
- package/deps/jq/modules/oniguruma/COPYING +28 -0
- package/deps/jq/modules/oniguruma/ChangeLog +0 -0
- package/deps/jq/modules/oniguruma/HISTORY +2138 -0
- package/deps/jq/modules/oniguruma/Makefile.am +33 -0
- package/deps/jq/modules/oniguruma/NEWS +0 -0
- package/deps/jq/modules/oniguruma/README +189 -0
- package/deps/jq/modules/oniguruma/README.ja +195 -0
- package/deps/jq/modules/oniguruma/README.md +203 -0
- package/deps/jq/modules/oniguruma/cmake/dist.cmake +321 -0
- package/deps/jq/modules/oniguruma/compile +348 -0
- package/deps/jq/modules/oniguruma/config.guess +1754 -0
- package/deps/jq/modules/oniguruma/config.sub +1890 -0
- package/deps/jq/modules/oniguruma/configure.ac +87 -0
- package/deps/jq/modules/oniguruma/contributed/libfuzzer-onig.cpp +31 -0
- package/deps/jq/modules/oniguruma/depcomp +791 -0
- package/deps/jq/modules/oniguruma/dist.info +10 -0
- package/deps/jq/modules/oniguruma/doc/API +660 -0
- package/deps/jq/modules/oniguruma/doc/API.ja +667 -0
- package/deps/jq/modules/oniguruma/doc/FAQ +12 -0
- package/deps/jq/modules/oniguruma/doc/FAQ.ja +22 -0
- package/deps/jq/modules/oniguruma/doc/RE +436 -0
- package/deps/jq/modules/oniguruma/doc/RE.ja +448 -0
- package/deps/jq/modules/oniguruma/doc/UNICODE_PROPERTIES +698 -0
- package/deps/jq/modules/oniguruma/index.html +181 -0
- package/deps/jq/modules/oniguruma/index_ja.html +184 -0
- package/deps/jq/modules/oniguruma/install-sh +541 -0
- package/deps/jq/modules/oniguruma/m4/.whatever +0 -0
- package/deps/jq/modules/oniguruma/make_win32.bat +3 -0
- package/deps/jq/modules/oniguruma/make_win64.bat +3 -0
- package/deps/jq/modules/oniguruma/missing +215 -0
- package/deps/jq/modules/oniguruma/onig-config.in +78 -0
- package/deps/jq/modules/oniguruma/oniguruma.pc.cmake.in +13 -0
- package/deps/jq/modules/oniguruma/oniguruma.pc.in +14 -0
- package/deps/jq/modules/oniguruma/sample/CMakeLists.txt +28 -0
- package/deps/jq/modules/oniguruma/sample/Makefile.am +34 -0
- package/deps/jq/modules/oniguruma/sample/bug_fix.c +131 -0
- package/deps/jq/modules/oniguruma/sample/crnl.c +127 -0
- package/deps/jq/modules/oniguruma/sample/encode.c +308 -0
- package/deps/jq/modules/oniguruma/sample/listcap.c +110 -0
- package/deps/jq/modules/oniguruma/sample/names.c +75 -0
- package/deps/jq/modules/oniguruma/sample/posix.c +96 -0
- package/deps/jq/modules/oniguruma/sample/scan.c +88 -0
- package/deps/jq/modules/oniguruma/sample/simple.c +59 -0
- package/deps/jq/modules/oniguruma/sample/sql.c +76 -0
- package/deps/jq/modules/oniguruma/sample/syntax.c +76 -0
- package/deps/jq/modules/oniguruma/sample/user_property.c +87 -0
- package/deps/jq/modules/oniguruma/src/Makefile.am +58 -0
- package/deps/jq/modules/oniguruma/src/Makefile.windows +183 -0
- package/deps/jq/modules/oniguruma/src/ascii.c +61 -0
- package/deps/jq/modules/oniguruma/src/big5.c +190 -0
- package/deps/jq/modules/oniguruma/src/config.h.cmake.in +71 -0
- package/deps/jq/modules/oniguruma/src/config.h.win32 +84 -0
- package/deps/jq/modules/oniguruma/src/config.h.win64 +84 -0
- package/deps/jq/modules/oniguruma/src/cp1251.c +203 -0
- package/deps/jq/modules/oniguruma/src/euc_jp.c +310 -0
- package/deps/jq/modules/oniguruma/src/euc_jp_prop.c +158 -0
- package/deps/jq/modules/oniguruma/src/euc_jp_prop.gperf +27 -0
- package/deps/jq/modules/oniguruma/src/euc_kr.c +188 -0
- package/deps/jq/modules/oniguruma/src/euc_tw.c +171 -0
- package/deps/jq/modules/oniguruma/src/gb18030.c +538 -0
- package/deps/jq/modules/oniguruma/src/gperf_fold_key_conv.py +67 -0
- package/deps/jq/modules/oniguruma/src/gperf_unfold_key_conv.py +55 -0
- package/deps/jq/modules/oniguruma/src/iso8859_1.c +275 -0
- package/deps/jq/modules/oniguruma/src/iso8859_10.c +242 -0
- package/deps/jq/modules/oniguruma/src/iso8859_11.c +99 -0
- package/deps/jq/modules/oniguruma/src/iso8859_13.c +231 -0
- package/deps/jq/modules/oniguruma/src/iso8859_14.c +244 -0
- package/deps/jq/modules/oniguruma/src/iso8859_15.c +238 -0
- package/deps/jq/modules/oniguruma/src/iso8859_16.c +240 -0
- package/deps/jq/modules/oniguruma/src/iso8859_2.c +238 -0
- package/deps/jq/modules/oniguruma/src/iso8859_3.c +238 -0
- package/deps/jq/modules/oniguruma/src/iso8859_4.c +240 -0
- package/deps/jq/modules/oniguruma/src/iso8859_5.c +229 -0
- package/deps/jq/modules/oniguruma/src/iso8859_6.c +99 -0
- package/deps/jq/modules/oniguruma/src/iso8859_7.c +225 -0
- package/deps/jq/modules/oniguruma/src/iso8859_8.c +99 -0
- package/deps/jq/modules/oniguruma/src/iso8859_9.c +231 -0
- package/deps/jq/modules/oniguruma/src/koi8.c +253 -0
- package/deps/jq/modules/oniguruma/src/koi8_r.c +215 -0
- package/deps/jq/modules/oniguruma/src/make_unicode_fold.sh +22 -0
- package/deps/jq/modules/oniguruma/src/make_unicode_fold_data.py +306 -0
- package/deps/jq/modules/oniguruma/src/make_unicode_property.sh +18 -0
- package/deps/jq/modules/oniguruma/src/make_unicode_property_data.py +545 -0
- package/deps/jq/modules/oniguruma/src/mktable.c +1184 -0
- package/deps/jq/modules/oniguruma/src/onig_init.c +45 -0
- package/deps/jq/modules/oniguruma/src/oniggnu.h +85 -0
- package/deps/jq/modules/oniguruma/src/onigposix.h +169 -0
- package/deps/jq/modules/oniguruma/src/oniguruma.h +841 -0
- package/deps/jq/modules/oniguruma/src/regcomp.c +6300 -0
- package/deps/jq/modules/oniguruma/src/regenc.c +917 -0
- package/deps/jq/modules/oniguruma/src/regenc.h +243 -0
- package/deps/jq/modules/oniguruma/src/regerror.c +393 -0
- package/deps/jq/modules/oniguruma/src/regexec.c +3856 -0
- package/deps/jq/modules/oniguruma/src/regext.c +202 -0
- package/deps/jq/modules/oniguruma/src/reggnu.c +147 -0
- package/deps/jq/modules/oniguruma/src/regint.h +779 -0
- package/deps/jq/modules/oniguruma/src/regparse.c +5403 -0
- package/deps/jq/modules/oniguruma/src/regparse.h +351 -0
- package/deps/jq/modules/oniguruma/src/regposerr.c +105 -0
- package/deps/jq/modules/oniguruma/src/regposix.c +306 -0
- package/deps/jq/modules/oniguruma/src/regsyntax.c +315 -0
- package/deps/jq/modules/oniguruma/src/regtrav.c +76 -0
- package/deps/jq/modules/oniguruma/src/regversion.c +57 -0
- package/deps/jq/modules/oniguruma/src/sjis.c +341 -0
- package/deps/jq/modules/oniguruma/src/sjis_prop.c +158 -0
- package/deps/jq/modules/oniguruma/src/sjis_prop.gperf +27 -0
- package/deps/jq/modules/oniguruma/src/st.c +589 -0
- package/deps/jq/modules/oniguruma/src/st.h +68 -0
- package/deps/jq/modules/oniguruma/src/unicode-7.0/unicode_fold1_key.c +2250 -0
- package/deps/jq/modules/oniguruma/src/unicode-7.0/unicode_fold2_key.c +203 -0
- package/deps/jq/modules/oniguruma/src/unicode-7.0/unicode_fold3_key.c +113 -0
- package/deps/jq/modules/oniguruma/src/unicode-7.0/unicode_fold_data.c +1225 -0
- package/deps/jq/modules/oniguruma/src/unicode-7.0/unicode_property_data.c +24742 -0
- package/deps/jq/modules/oniguruma/src/unicode-7.0/unicode_property_data_posix.c +4846 -0
- package/deps/jq/modules/oniguruma/src/unicode-7.0/unicode_unfold_key.c +2571 -0
- package/deps/jq/modules/oniguruma/src/unicode.c +664 -0
- package/deps/jq/modules/oniguruma/src/unicode_fold1_key.c +2548 -0
- package/deps/jq/modules/oniguruma/src/unicode_fold2_key.c +203 -0
- package/deps/jq/modules/oniguruma/src/unicode_fold3_key.c +113 -0
- package/deps/jq/modules/oniguruma/src/unicode_fold_data.c +1366 -0
- package/deps/jq/modules/oniguruma/src/unicode_property_data.c +25306 -0
- package/deps/jq/modules/oniguruma/src/unicode_property_data_posix.c +4940 -0
- package/deps/jq/modules/oniguruma/src/unicode_unfold_key.c +2854 -0
- package/deps/jq/modules/oniguruma/src/utf16_be.c +235 -0
- package/deps/jq/modules/oniguruma/src/utf16_le.c +245 -0
- package/deps/jq/modules/oniguruma/src/utf32_be.c +194 -0
- package/deps/jq/modules/oniguruma/src/utf32_le.c +194 -0
- package/deps/jq/modules/oniguruma/src/utf8.c +335 -0
- package/deps/jq/modules/oniguruma/test/Makefile.am +28 -0
- package/deps/jq/modules/oniguruma/test/testc.c +874 -0
- package/deps/jq/modules/oniguruma/test/testu.c +916 -0
- package/deps/jq/modules/oniguruma/test-driver +153 -0
- package/deps/jq/modules/oniguruma/windows/testc.c +869 -0
- package/deps/jq/scripts/crosscompile +2 -1
- package/deps/jq/scripts/gen_utf8_tables.py +2 -3
- package/deps/jq/scripts/update-website +29 -0
- package/deps/jq/scripts/version +7 -2
- package/deps/jq/sig/jq-release.key +41 -0
- package/deps/jq/sig/v1.3/jq-linux-x86.asc +17 -0
- package/deps/jq/sig/v1.3/jq-linux-x86_64.asc +17 -0
- package/deps/jq/sig/v1.3/jq-osx-x86.asc +17 -0
- package/deps/jq/sig/v1.3/jq-osx-x86_64.asc +17 -0
- package/deps/jq/sig/v1.3/jq-win32.exe.asc +17 -0
- package/deps/jq/sig/v1.3/jq-win64.exe.asc +17 -0
- package/deps/jq/sig/v1.3/sha256sum.txt +6 -0
- package/deps/jq/sig/v1.4/jq-linux-x86.asc +17 -0
- package/deps/jq/sig/v1.4/jq-linux-x86_64.asc +17 -0
- package/deps/jq/sig/v1.4/jq-osx-x86.asc +17 -0
- package/deps/jq/sig/v1.4/jq-osx-x86_64.asc +17 -0
- package/deps/jq/sig/v1.4/jq-solaris11-32.asc +17 -0
- package/deps/jq/sig/v1.4/jq-solaris11-64.asc +17 -0
- package/deps/jq/sig/v1.4/jq-win32.exe.asc +17 -0
- package/deps/jq/sig/v1.4/jq-win64.exe.asc +17 -0
- package/deps/jq/sig/v1.4/sha256sum.txt +8 -0
- package/deps/jq/sig/v1.5/jq-linux32-no-oniguruma.asc +17 -0
- package/deps/jq/sig/v1.5/jq-linux32.asc +17 -0
- package/deps/jq/sig/v1.5/jq-linux64.asc +17 -0
- package/deps/jq/sig/v1.5/jq-osx-amd64.asc +17 -0
- package/deps/jq/sig/v1.5/jq-win32.exe.asc +17 -0
- package/deps/jq/sig/v1.5/jq-win64.exe.asc +17 -0
- package/deps/jq/sig/v1.5/sha256sum.txt +5 -0
- package/deps/jq/sig/v1.5rc1/jq-linux-x86_64-static.asc +17 -0
- package/deps/jq/sig/v1.5rc1/jq-win32.exe.asc +17 -0
- package/deps/jq/sig/v1.5rc1/jq-win64.exe.asc +17 -0
- package/deps/jq/sig/v1.5rc1/sha256sum.txt +3 -0
- package/deps/jq/sig/v1.5rc2/jq-linux-x86.asc +17 -0
- package/deps/jq/sig/v1.5rc2/jq-linux-x86_64.asc +17 -0
- package/deps/jq/sig/v1.5rc2/jq-osx-x86_64.asc +17 -0
- package/deps/jq/sig/v1.5rc2/jq-win32.exe.asc +17 -0
- package/deps/jq/sig/v1.5rc2/jq-win64.exe.asc +17 -0
- package/deps/jq/sig/v1.5rc2/sha256sum.txt +5 -0
- package/deps/jq/sig/v1.6/jq-linux32.asc +16 -0
- package/deps/jq/sig/v1.6/jq-linux64.asc +16 -0
- package/deps/jq/sig/v1.6/jq-osx-amd64.asc +16 -0
- package/deps/jq/sig/v1.6/jq-win32.exe.asc +16 -0
- package/deps/jq/sig/v1.6/jq-win64.exe.asc +16 -0
- package/deps/jq/sig/v1.6/sha256sum.txt +5 -0
- package/deps/jq/{builtin.c → src/builtin.c} +449 -344
- package/deps/jq/{builtin.h → src/builtin.h} +0 -0
- package/deps/jq/src/builtin.jq +311 -0
- package/deps/jq/{bytecode.c → src/bytecode.c} +0 -0
- package/deps/jq/{bytecode.h → src/bytecode.h} +0 -0
- package/deps/jq/{compile.c → src/compile.c} +213 -37
- package/deps/jq/{compile.h → src/compile.h} +7 -2
- package/deps/jq/{exec_stack.h → src/exec_stack.h} +0 -0
- package/deps/jq/{execute.c → src/execute.c} +127 -24
- package/deps/jq/{inject_errors.c → src/inject_errors.c} +0 -0
- package/deps/jq/{jq.h → src/jq.h} +13 -2
- package/deps/jq/{jq_parser.h → src/jq_parser.h} +0 -0
- package/deps/jq/{jq_test.c → src/jq_test.c} +1 -1
- package/deps/jq/{jv.c → src/jv.c} +24 -5
- package/deps/jq/{jv.h → src/jv.h} +46 -11
- package/deps/jq/{jv_alloc.c → src/jv_alloc.c} +0 -0
- package/deps/jq/{jv_alloc.h → src/jv_alloc.h} +0 -0
- package/deps/jq/{jv_aux.c → src/jv_aux.c} +38 -14
- package/deps/jq/{jv_dtoa.c → src/jv_dtoa.c} +4 -5
- package/deps/jq/{jv_dtoa.h → src/jv_dtoa.h} +0 -0
- package/deps/jq/src/jv_file.c +81 -0
- package/deps/jq/{jv_parse.c → src/jv_parse.c} +8 -2
- package/deps/jq/{jv_print.c → src/jv_print.c} +80 -27
- package/deps/jq/{jv_unicode.c → src/jv_unicode.c} +28 -4
- package/deps/jq/{jv_unicode.h → src/jv_unicode.h} +1 -0
- package/deps/jq/{jv_utf8_tables.h → src/jv_utf8_tables.h} +0 -0
- package/deps/jq/{lexer.c → src/lexer.c} +317 -292
- package/deps/jq/{lexer.h → src/lexer.h} +16 -12
- package/deps/jq/{lexer.l → src/lexer.l} +5 -4
- package/deps/jq/src/libm.h +291 -0
- package/deps/jq/{linker.c → src/linker.c} +5 -3
- package/deps/jq/{linker.h → src/linker.h} +0 -0
- package/deps/jq/{locfile.c → src/locfile.c} +1 -1
- package/deps/jq/{locfile.h → src/locfile.h} +0 -0
- package/deps/jq/{main.c → src/main.c} +202 -100
- package/deps/jq/{opcode_list.h → src/opcode_list.h} +4 -0
- package/deps/jq/{parser.c → src/parser.c} +1149 -1069
- package/deps/jq/{parser.h → src/parser.h} +28 -24
- package/deps/jq/{parser.y → src/parser.y} +108 -52
- package/deps/jq/{util.c → src/util.c} +12 -9
- package/deps/jq/{util.h → src/util.h} +0 -0
- package/deps/jq/tests/base64.test +35 -0
- package/deps/jq/tests/base64test +5 -0
- package/deps/jq/tests/jq-f-test.sh +4 -0
- package/deps/jq/tests/jq.test +305 -21
- package/deps/jq/tests/jqtest +1 -1
- package/deps/jq/tests/mantest +1 -1
- package/deps/jq/tests/onig.test +13 -0
- package/deps/jq/tests/onigtest +1 -1
- package/deps/jq/tests/optional.test +20 -0
- package/deps/jq/tests/optionaltest +5 -0
- package/deps/jq/tests/setup +7 -1
- package/deps/jq/tests/shtest +133 -29
- package/deps/jq/tests/utf8-truncate.jq +3 -0
- package/deps/jq/tests/utf8test +10 -0
- package/lib/index.js +7 -1
- package/package.json +3 -3
- package/test/santiy.test.js +24 -0
- package/deps/jq/docs/default_manpage.md +0 -22
- package/deps/jq/docs/public/bootstrap/css/bootstrap-responsive.css +0 -1058
- package/deps/jq/docs/public/bootstrap/css/bootstrap-responsive.min.css +0 -9
- package/deps/jq/docs/public/bootstrap/css/bootstrap.css +0 -5224
- package/deps/jq/docs/public/bootstrap/css/bootstrap.min.css +0 -9
- package/deps/jq/docs/public/bootstrap/img/glyphicons-halflings-white.png +0 -0
- package/deps/jq/docs/public/bootstrap/img/glyphicons-halflings.png +0 -0
- package/deps/jq/docs/public/bootstrap/js/bootstrap.js +0 -2027
- package/deps/jq/docs/public/bootstrap/js/bootstrap.min.js +0 -6
- package/deps/jq/docs/templates/shared/_header.liquid +0 -26
- package/deps/jq/jq.1.default +0 -39
- package/deps/jq/jv_file.c +0 -49
- package/deps/jq/libm.h +0 -160
- package/deps/jq/setup.sh +0 -33
- package/reports/jest-port-api.xml +0 -35
|
@@ -0,0 +1,3856 @@
|
|
|
1
|
+
/**********************************************************************
|
|
2
|
+
regexec.c - Oniguruma (regular expression library)
|
|
3
|
+
**********************************************************************/
|
|
4
|
+
/*-
|
|
5
|
+
* Copyright (c) 2002-2016 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
|
|
6
|
+
* All rights reserved.
|
|
7
|
+
*
|
|
8
|
+
* Redistribution and use in source and binary forms, with or without
|
|
9
|
+
* modification, are permitted provided that the following conditions
|
|
10
|
+
* are met:
|
|
11
|
+
* 1. Redistributions of source code must retain the above copyright
|
|
12
|
+
* notice, this list of conditions and the following disclaimer.
|
|
13
|
+
* 2. Redistributions in binary form must reproduce the above copyright
|
|
14
|
+
* notice, this list of conditions and the following disclaimer in the
|
|
15
|
+
* documentation and/or other materials provided with the distribution.
|
|
16
|
+
*
|
|
17
|
+
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
|
|
18
|
+
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
|
19
|
+
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
|
20
|
+
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
|
|
21
|
+
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
|
22
|
+
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
|
|
23
|
+
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
|
|
24
|
+
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
|
25
|
+
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
|
|
26
|
+
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
|
27
|
+
* SUCH DAMAGE.
|
|
28
|
+
*/
|
|
29
|
+
|
|
30
|
+
#include "regint.h"
|
|
31
|
+
|
|
32
|
+
#define USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
|
|
33
|
+
|
|
34
|
+
#ifdef USE_CRNL_AS_LINE_TERMINATOR
|
|
35
|
+
#define ONIGENC_IS_MBC_CRNL(enc,p,end) \
|
|
36
|
+
(ONIGENC_MBC_TO_CODE(enc,p,end) == 13 && \
|
|
37
|
+
ONIGENC_IS_MBC_NEWLINE(enc,(p+enclen(enc,p)),end))
|
|
38
|
+
#endif
|
|
39
|
+
|
|
40
|
+
#ifdef USE_CAPTURE_HISTORY
|
|
41
|
+
static void history_tree_free(OnigCaptureTreeNode* node);
|
|
42
|
+
|
|
43
|
+
static void
|
|
44
|
+
history_tree_clear(OnigCaptureTreeNode* node)
|
|
45
|
+
{
|
|
46
|
+
int i;
|
|
47
|
+
|
|
48
|
+
if (IS_NOT_NULL(node)) {
|
|
49
|
+
for (i = 0; i < node->num_childs; i++) {
|
|
50
|
+
if (IS_NOT_NULL(node->childs[i])) {
|
|
51
|
+
history_tree_free(node->childs[i]);
|
|
52
|
+
}
|
|
53
|
+
}
|
|
54
|
+
for (i = 0; i < node->allocated; i++) {
|
|
55
|
+
node->childs[i] = (OnigCaptureTreeNode* )0;
|
|
56
|
+
}
|
|
57
|
+
node->num_childs = 0;
|
|
58
|
+
node->beg = ONIG_REGION_NOTPOS;
|
|
59
|
+
node->end = ONIG_REGION_NOTPOS;
|
|
60
|
+
node->group = -1;
|
|
61
|
+
}
|
|
62
|
+
}
|
|
63
|
+
|
|
64
|
+
static void
|
|
65
|
+
history_tree_free(OnigCaptureTreeNode* node)
|
|
66
|
+
{
|
|
67
|
+
history_tree_clear(node);
|
|
68
|
+
xfree(node);
|
|
69
|
+
}
|
|
70
|
+
|
|
71
|
+
static void
|
|
72
|
+
history_root_free(OnigRegion* r)
|
|
73
|
+
{
|
|
74
|
+
if (IS_NOT_NULL(r->history_root)) {
|
|
75
|
+
history_tree_free(r->history_root);
|
|
76
|
+
r->history_root = (OnigCaptureTreeNode* )0;
|
|
77
|
+
}
|
|
78
|
+
}
|
|
79
|
+
|
|
80
|
+
static OnigCaptureTreeNode*
|
|
81
|
+
history_node_new(void)
|
|
82
|
+
{
|
|
83
|
+
OnigCaptureTreeNode* node;
|
|
84
|
+
|
|
85
|
+
node = (OnigCaptureTreeNode* )xmalloc(sizeof(OnigCaptureTreeNode));
|
|
86
|
+
CHECK_NULL_RETURN(node);
|
|
87
|
+
node->childs = (OnigCaptureTreeNode** )0;
|
|
88
|
+
node->allocated = 0;
|
|
89
|
+
node->num_childs = 0;
|
|
90
|
+
node->group = -1;
|
|
91
|
+
node->beg = ONIG_REGION_NOTPOS;
|
|
92
|
+
node->end = ONIG_REGION_NOTPOS;
|
|
93
|
+
|
|
94
|
+
return node;
|
|
95
|
+
}
|
|
96
|
+
|
|
97
|
+
static int
|
|
98
|
+
history_tree_add_child(OnigCaptureTreeNode* parent, OnigCaptureTreeNode* child)
|
|
99
|
+
{
|
|
100
|
+
#define HISTORY_TREE_INIT_ALLOC_SIZE 8
|
|
101
|
+
|
|
102
|
+
if (parent->num_childs >= parent->allocated) {
|
|
103
|
+
int n, i;
|
|
104
|
+
|
|
105
|
+
if (IS_NULL(parent->childs)) {
|
|
106
|
+
n = HISTORY_TREE_INIT_ALLOC_SIZE;
|
|
107
|
+
parent->childs =
|
|
108
|
+
(OnigCaptureTreeNode** )xmalloc(sizeof(OnigCaptureTreeNode*) * n);
|
|
109
|
+
}
|
|
110
|
+
else {
|
|
111
|
+
n = parent->allocated * 2;
|
|
112
|
+
parent->childs =
|
|
113
|
+
(OnigCaptureTreeNode** )xrealloc(parent->childs,
|
|
114
|
+
sizeof(OnigCaptureTreeNode*) * n);
|
|
115
|
+
}
|
|
116
|
+
CHECK_NULL_RETURN_MEMERR(parent->childs);
|
|
117
|
+
for (i = parent->allocated; i < n; i++) {
|
|
118
|
+
parent->childs[i] = (OnigCaptureTreeNode* )0;
|
|
119
|
+
}
|
|
120
|
+
parent->allocated = n;
|
|
121
|
+
}
|
|
122
|
+
|
|
123
|
+
parent->childs[parent->num_childs] = child;
|
|
124
|
+
parent->num_childs++;
|
|
125
|
+
return 0;
|
|
126
|
+
}
|
|
127
|
+
|
|
128
|
+
static OnigCaptureTreeNode*
|
|
129
|
+
history_tree_clone(OnigCaptureTreeNode* node)
|
|
130
|
+
{
|
|
131
|
+
int i;
|
|
132
|
+
OnigCaptureTreeNode *clone, *child;
|
|
133
|
+
|
|
134
|
+
clone = history_node_new();
|
|
135
|
+
CHECK_NULL_RETURN(clone);
|
|
136
|
+
|
|
137
|
+
clone->beg = node->beg;
|
|
138
|
+
clone->end = node->end;
|
|
139
|
+
for (i = 0; i < node->num_childs; i++) {
|
|
140
|
+
child = history_tree_clone(node->childs[i]);
|
|
141
|
+
if (IS_NULL(child)) {
|
|
142
|
+
history_tree_free(clone);
|
|
143
|
+
return (OnigCaptureTreeNode* )0;
|
|
144
|
+
}
|
|
145
|
+
history_tree_add_child(clone, child);
|
|
146
|
+
}
|
|
147
|
+
|
|
148
|
+
return clone;
|
|
149
|
+
}
|
|
150
|
+
|
|
151
|
+
extern OnigCaptureTreeNode*
|
|
152
|
+
onig_get_capture_tree(OnigRegion* region)
|
|
153
|
+
{
|
|
154
|
+
return region->history_root;
|
|
155
|
+
}
|
|
156
|
+
#endif /* USE_CAPTURE_HISTORY */
|
|
157
|
+
|
|
158
|
+
extern void
|
|
159
|
+
onig_region_clear(OnigRegion* region)
|
|
160
|
+
{
|
|
161
|
+
int i;
|
|
162
|
+
|
|
163
|
+
for (i = 0; i < region->num_regs; i++) {
|
|
164
|
+
region->beg[i] = region->end[i] = ONIG_REGION_NOTPOS;
|
|
165
|
+
}
|
|
166
|
+
#ifdef USE_CAPTURE_HISTORY
|
|
167
|
+
history_root_free(region);
|
|
168
|
+
#endif
|
|
169
|
+
}
|
|
170
|
+
|
|
171
|
+
extern int
|
|
172
|
+
onig_region_resize(OnigRegion* region, int n)
|
|
173
|
+
{
|
|
174
|
+
region->num_regs = n;
|
|
175
|
+
|
|
176
|
+
if (n < ONIG_NREGION)
|
|
177
|
+
n = ONIG_NREGION;
|
|
178
|
+
|
|
179
|
+
if (region->allocated == 0) {
|
|
180
|
+
region->beg = (int* )xmalloc(n * sizeof(int));
|
|
181
|
+
region->end = (int* )xmalloc(n * sizeof(int));
|
|
182
|
+
|
|
183
|
+
if (region->beg == 0 || region->end == 0)
|
|
184
|
+
return ONIGERR_MEMORY;
|
|
185
|
+
|
|
186
|
+
region->allocated = n;
|
|
187
|
+
}
|
|
188
|
+
else if (region->allocated < n) {
|
|
189
|
+
region->beg = (int* )xrealloc(region->beg, n * sizeof(int));
|
|
190
|
+
region->end = (int* )xrealloc(region->end, n * sizeof(int));
|
|
191
|
+
|
|
192
|
+
if (region->beg == 0 || region->end == 0)
|
|
193
|
+
return ONIGERR_MEMORY;
|
|
194
|
+
|
|
195
|
+
region->allocated = n;
|
|
196
|
+
}
|
|
197
|
+
|
|
198
|
+
return 0;
|
|
199
|
+
}
|
|
200
|
+
|
|
201
|
+
static int
|
|
202
|
+
onig_region_resize_clear(OnigRegion* region, int n)
|
|
203
|
+
{
|
|
204
|
+
int r;
|
|
205
|
+
|
|
206
|
+
r = onig_region_resize(region, n);
|
|
207
|
+
if (r != 0) return r;
|
|
208
|
+
onig_region_clear(region);
|
|
209
|
+
return 0;
|
|
210
|
+
}
|
|
211
|
+
|
|
212
|
+
extern int
|
|
213
|
+
onig_region_set(OnigRegion* region, int at, int beg, int end)
|
|
214
|
+
{
|
|
215
|
+
if (at < 0) return ONIGERR_INVALID_ARGUMENT;
|
|
216
|
+
|
|
217
|
+
if (at >= region->allocated) {
|
|
218
|
+
int r = onig_region_resize(region, at + 1);
|
|
219
|
+
if (r < 0) return r;
|
|
220
|
+
}
|
|
221
|
+
|
|
222
|
+
region->beg[at] = beg;
|
|
223
|
+
region->end[at] = end;
|
|
224
|
+
return 0;
|
|
225
|
+
}
|
|
226
|
+
|
|
227
|
+
extern void
|
|
228
|
+
onig_region_init(OnigRegion* region)
|
|
229
|
+
{
|
|
230
|
+
region->num_regs = 0;
|
|
231
|
+
region->allocated = 0;
|
|
232
|
+
region->beg = (int* )0;
|
|
233
|
+
region->end = (int* )0;
|
|
234
|
+
region->history_root = (OnigCaptureTreeNode* )0;
|
|
235
|
+
}
|
|
236
|
+
|
|
237
|
+
extern OnigRegion*
|
|
238
|
+
onig_region_new(void)
|
|
239
|
+
{
|
|
240
|
+
OnigRegion* r;
|
|
241
|
+
|
|
242
|
+
r = (OnigRegion* )xmalloc(sizeof(OnigRegion));
|
|
243
|
+
onig_region_init(r);
|
|
244
|
+
return r;
|
|
245
|
+
}
|
|
246
|
+
|
|
247
|
+
extern void
|
|
248
|
+
onig_region_free(OnigRegion* r, int free_self)
|
|
249
|
+
{
|
|
250
|
+
if (r) {
|
|
251
|
+
if (r->allocated > 0) {
|
|
252
|
+
if (r->beg) xfree(r->beg);
|
|
253
|
+
if (r->end) xfree(r->end);
|
|
254
|
+
r->allocated = 0;
|
|
255
|
+
}
|
|
256
|
+
#ifdef USE_CAPTURE_HISTORY
|
|
257
|
+
history_root_free(r);
|
|
258
|
+
#endif
|
|
259
|
+
if (free_self) xfree(r);
|
|
260
|
+
}
|
|
261
|
+
}
|
|
262
|
+
|
|
263
|
+
extern void
|
|
264
|
+
onig_region_copy(OnigRegion* to, OnigRegion* from)
|
|
265
|
+
{
|
|
266
|
+
#define RREGC_SIZE (sizeof(int) * from->num_regs)
|
|
267
|
+
int i;
|
|
268
|
+
|
|
269
|
+
if (to == from) return;
|
|
270
|
+
|
|
271
|
+
if (to->allocated == 0) {
|
|
272
|
+
if (from->num_regs > 0) {
|
|
273
|
+
to->beg = (int* )xmalloc(RREGC_SIZE);
|
|
274
|
+
to->end = (int* )xmalloc(RREGC_SIZE);
|
|
275
|
+
to->allocated = from->num_regs;
|
|
276
|
+
}
|
|
277
|
+
}
|
|
278
|
+
else if (to->allocated < from->num_regs) {
|
|
279
|
+
to->beg = (int* )xrealloc(to->beg, RREGC_SIZE);
|
|
280
|
+
to->end = (int* )xrealloc(to->end, RREGC_SIZE);
|
|
281
|
+
to->allocated = from->num_regs;
|
|
282
|
+
}
|
|
283
|
+
|
|
284
|
+
for (i = 0; i < from->num_regs; i++) {
|
|
285
|
+
to->beg[i] = from->beg[i];
|
|
286
|
+
to->end[i] = from->end[i];
|
|
287
|
+
}
|
|
288
|
+
to->num_regs = from->num_regs;
|
|
289
|
+
|
|
290
|
+
#ifdef USE_CAPTURE_HISTORY
|
|
291
|
+
history_root_free(to);
|
|
292
|
+
|
|
293
|
+
if (IS_NOT_NULL(from->history_root)) {
|
|
294
|
+
to->history_root = history_tree_clone(from->history_root);
|
|
295
|
+
}
|
|
296
|
+
#endif
|
|
297
|
+
}
|
|
298
|
+
|
|
299
|
+
|
|
300
|
+
/** stack **/
|
|
301
|
+
#define INVALID_STACK_INDEX -1
|
|
302
|
+
|
|
303
|
+
/* stack type */
|
|
304
|
+
/* used by normal-POP */
|
|
305
|
+
#define STK_ALT 0x0001
|
|
306
|
+
#define STK_LOOK_BEHIND_NOT 0x0002
|
|
307
|
+
#define STK_POS_NOT 0x0003
|
|
308
|
+
/* handled by normal-POP */
|
|
309
|
+
#define STK_MEM_START 0x0100
|
|
310
|
+
#define STK_MEM_END 0x8200
|
|
311
|
+
#define STK_REPEAT_INC 0x0300
|
|
312
|
+
#define STK_STATE_CHECK_MARK 0x1000
|
|
313
|
+
/* avoided by normal-POP */
|
|
314
|
+
#define STK_NULL_CHECK_START 0x3000
|
|
315
|
+
#define STK_NULL_CHECK_END 0x5000 /* for recursive call */
|
|
316
|
+
#define STK_MEM_END_MARK 0x8400
|
|
317
|
+
#define STK_POS 0x0500 /* used when POP-POS */
|
|
318
|
+
#define STK_STOP_BT 0x0600 /* mark for "(?>...)" */
|
|
319
|
+
#define STK_REPEAT 0x0700
|
|
320
|
+
#define STK_CALL_FRAME 0x0800
|
|
321
|
+
#define STK_RETURN 0x0900
|
|
322
|
+
#define STK_VOID 0x0a00 /* for fill a blank */
|
|
323
|
+
|
|
324
|
+
/* stack type check mask */
|
|
325
|
+
#define STK_MASK_POP_USED 0x00ff
|
|
326
|
+
#define STK_MASK_TO_VOID_TARGET 0x10ff
|
|
327
|
+
#define STK_MASK_MEM_END_OR_MARK 0x8000 /* MEM_END or MEM_END_MARK */
|
|
328
|
+
|
|
329
|
+
#ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
|
|
330
|
+
#define MATCH_ARG_INIT(msa, reg, arg_option, arg_region, arg_start) do {\
|
|
331
|
+
(msa).stack_p = (void* )0;\
|
|
332
|
+
(msa).options = (arg_option);\
|
|
333
|
+
(msa).region = (arg_region);\
|
|
334
|
+
(msa).start = (arg_start);\
|
|
335
|
+
(msa).best_len = ONIG_MISMATCH;\
|
|
336
|
+
(msa).ptr_num = (reg)->num_repeat + (reg)->num_mem * 2;\
|
|
337
|
+
} while(0)
|
|
338
|
+
#else
|
|
339
|
+
#define MATCH_ARG_INIT(msa, reg, arg_option, arg_region, arg_start) do {\
|
|
340
|
+
(msa).stack_p = (void* )0;\
|
|
341
|
+
(msa).options = (arg_option);\
|
|
342
|
+
(msa).region = (arg_region);\
|
|
343
|
+
(msa).start = (arg_start);\
|
|
344
|
+
(msa).ptr_num = (reg)->num_repeat + (reg)->num_mem * 2;\
|
|
345
|
+
} while(0)
|
|
346
|
+
#endif
|
|
347
|
+
|
|
348
|
+
#ifdef USE_COMBINATION_EXPLOSION_CHECK
|
|
349
|
+
|
|
350
|
+
#define STATE_CHECK_BUFF_MALLOC_THRESHOLD_SIZE 16
|
|
351
|
+
|
|
352
|
+
#define STATE_CHECK_BUFF_INIT(msa, str_len, offset, state_num) do { \
|
|
353
|
+
if ((state_num) > 0 && str_len >= STATE_CHECK_STRING_THRESHOLD_LEN) {\
|
|
354
|
+
unsigned int size = (unsigned int )(((str_len) + 1) * (state_num) + 7) >> 3;\
|
|
355
|
+
offset = ((offset) * (state_num)) >> 3;\
|
|
356
|
+
if (size > 0 && offset < size && size < STATE_CHECK_BUFF_MAX_SIZE) {\
|
|
357
|
+
if (size >= STATE_CHECK_BUFF_MALLOC_THRESHOLD_SIZE) \
|
|
358
|
+
(msa).state_check_buff = (void* )xmalloc(size);\
|
|
359
|
+
else \
|
|
360
|
+
(msa).state_check_buff = (void* )xalloca(size);\
|
|
361
|
+
xmemset(((char* )((msa).state_check_buff)+(offset)), 0, \
|
|
362
|
+
(size_t )(size - (offset))); \
|
|
363
|
+
(msa).state_check_buff_size = size;\
|
|
364
|
+
}\
|
|
365
|
+
else {\
|
|
366
|
+
(msa).state_check_buff = (void* )0;\
|
|
367
|
+
(msa).state_check_buff_size = 0;\
|
|
368
|
+
}\
|
|
369
|
+
}\
|
|
370
|
+
else {\
|
|
371
|
+
(msa).state_check_buff = (void* )0;\
|
|
372
|
+
(msa).state_check_buff_size = 0;\
|
|
373
|
+
}\
|
|
374
|
+
} while(0)
|
|
375
|
+
|
|
376
|
+
#define MATCH_ARG_FREE(msa) do {\
|
|
377
|
+
if ((msa).stack_p) xfree((msa).stack_p);\
|
|
378
|
+
if ((msa).state_check_buff_size >= STATE_CHECK_BUFF_MALLOC_THRESHOLD_SIZE) { \
|
|
379
|
+
if ((msa).state_check_buff) xfree((msa).state_check_buff);\
|
|
380
|
+
}\
|
|
381
|
+
} while(0)
|
|
382
|
+
#else
|
|
383
|
+
#define STATE_CHECK_BUFF_INIT(msa, str_len, offset, state_num)
|
|
384
|
+
#define MATCH_ARG_FREE(msa) if ((msa).stack_p) xfree((msa).stack_p)
|
|
385
|
+
#endif
|
|
386
|
+
|
|
387
|
+
|
|
388
|
+
#define ALLOCA_PTR_NUM_LIMIT 50
|
|
389
|
+
|
|
390
|
+
#define STACK_INIT(stack_num) do {\
|
|
391
|
+
if (msa->stack_p) {\
|
|
392
|
+
is_alloca = 0;\
|
|
393
|
+
alloc_base = msa->stack_p;\
|
|
394
|
+
stk_base = (OnigStackType* )(alloc_base\
|
|
395
|
+
+ (sizeof(OnigStackIndex) * msa->ptr_num));\
|
|
396
|
+
stk = stk_base;\
|
|
397
|
+
stk_end = stk_base + msa->stack_n;\
|
|
398
|
+
}\
|
|
399
|
+
else if (msa->ptr_num > ALLOCA_PTR_NUM_LIMIT) {\
|
|
400
|
+
is_alloca = 0;\
|
|
401
|
+
alloc_base = (char* )xmalloc(sizeof(OnigStackIndex) * msa->ptr_num\
|
|
402
|
+
+ sizeof(OnigStackType) * (stack_num));\
|
|
403
|
+
stk_base = (OnigStackType* )(alloc_base\
|
|
404
|
+
+ (sizeof(OnigStackIndex) * msa->ptr_num));\
|
|
405
|
+
stk = stk_base;\
|
|
406
|
+
stk_end = stk_base + (stack_num);\
|
|
407
|
+
}\
|
|
408
|
+
else {\
|
|
409
|
+
is_alloca = 1;\
|
|
410
|
+
alloc_base = (char* )xalloca(sizeof(OnigStackIndex) * msa->ptr_num\
|
|
411
|
+
+ sizeof(OnigStackType) * (stack_num));\
|
|
412
|
+
stk_base = (OnigStackType* )(alloc_base\
|
|
413
|
+
+ (sizeof(OnigStackIndex) * msa->ptr_num));\
|
|
414
|
+
stk = stk_base;\
|
|
415
|
+
stk_end = stk_base + (stack_num);\
|
|
416
|
+
}\
|
|
417
|
+
} while(0);
|
|
418
|
+
|
|
419
|
+
|
|
420
|
+
#define STACK_SAVE do{\
|
|
421
|
+
msa->stack_n = stk_end - stk_base;\
|
|
422
|
+
if (is_alloca != 0) {\
|
|
423
|
+
size_t size = sizeof(OnigStackIndex) * msa->ptr_num \
|
|
424
|
+
+ sizeof(OnigStackType) * msa->stack_n;\
|
|
425
|
+
msa->stack_p = xmalloc(size);\
|
|
426
|
+
xmemcpy(msa->stack_p, alloc_base, size);\
|
|
427
|
+
}\
|
|
428
|
+
else {\
|
|
429
|
+
msa->stack_p = alloc_base;\
|
|
430
|
+
};\
|
|
431
|
+
} while(0)
|
|
432
|
+
|
|
433
|
+
#define UPDATE_FOR_STACK_REALLOC do{\
|
|
434
|
+
repeat_stk = (OnigStackIndex* )alloc_base;\
|
|
435
|
+
mem_start_stk = (OnigStackIndex* )(repeat_stk + reg->num_repeat);\
|
|
436
|
+
mem_end_stk = mem_start_stk + num_mem;\
|
|
437
|
+
mem_start_stk--; /* for index start from 1 */\
|
|
438
|
+
mem_end_stk--; /* for index start from 1 */\
|
|
439
|
+
} while(0)
|
|
440
|
+
|
|
441
|
+
static unsigned int MatchStackLimitSize = DEFAULT_MATCH_STACK_LIMIT_SIZE;
|
|
442
|
+
|
|
443
|
+
extern unsigned int
|
|
444
|
+
onig_get_match_stack_limit_size(void)
|
|
445
|
+
{
|
|
446
|
+
return MatchStackLimitSize;
|
|
447
|
+
}
|
|
448
|
+
|
|
449
|
+
extern int
|
|
450
|
+
onig_set_match_stack_limit_size(unsigned int size)
|
|
451
|
+
{
|
|
452
|
+
MatchStackLimitSize = size;
|
|
453
|
+
return 0;
|
|
454
|
+
}
|
|
455
|
+
|
|
456
|
+
static int
|
|
457
|
+
stack_double(int is_alloca, char** arg_alloc_base,
|
|
458
|
+
OnigStackType** arg_stk_base,
|
|
459
|
+
OnigStackType** arg_stk_end, OnigStackType** arg_stk,
|
|
460
|
+
OnigMatchArg* msa)
|
|
461
|
+
{
|
|
462
|
+
unsigned int n;
|
|
463
|
+
int used;
|
|
464
|
+
size_t size;
|
|
465
|
+
size_t new_size;
|
|
466
|
+
char* alloc_base;
|
|
467
|
+
char* new_alloc_base;
|
|
468
|
+
OnigStackType *stk_base, *stk_end, *stk;
|
|
469
|
+
|
|
470
|
+
alloc_base = *arg_alloc_base;
|
|
471
|
+
stk_base = *arg_stk_base;
|
|
472
|
+
stk_end = *arg_stk_end;
|
|
473
|
+
stk = *arg_stk;
|
|
474
|
+
|
|
475
|
+
n = stk_end - stk_base;
|
|
476
|
+
size = sizeof(OnigStackIndex) * msa->ptr_num + sizeof(OnigStackType) * n;
|
|
477
|
+
n *= 2;
|
|
478
|
+
new_size = sizeof(OnigStackIndex) * msa->ptr_num + sizeof(OnigStackType) * n;
|
|
479
|
+
if (is_alloca != 0) {
|
|
480
|
+
new_alloc_base = (char* )xmalloc(new_size);
|
|
481
|
+
if (IS_NULL(new_alloc_base)) {
|
|
482
|
+
STACK_SAVE;
|
|
483
|
+
return ONIGERR_MEMORY;
|
|
484
|
+
}
|
|
485
|
+
xmemcpy(new_alloc_base, alloc_base, size);
|
|
486
|
+
}
|
|
487
|
+
else {
|
|
488
|
+
if (MatchStackLimitSize != 0 && n > MatchStackLimitSize) {
|
|
489
|
+
if ((unsigned int )(stk_end - stk_base) == MatchStackLimitSize)
|
|
490
|
+
return ONIGERR_MATCH_STACK_LIMIT_OVER;
|
|
491
|
+
else
|
|
492
|
+
n = MatchStackLimitSize;
|
|
493
|
+
}
|
|
494
|
+
new_alloc_base = (char* )xrealloc(alloc_base, new_size);
|
|
495
|
+
if (IS_NULL(new_alloc_base)) {
|
|
496
|
+
STACK_SAVE;
|
|
497
|
+
return ONIGERR_MEMORY;
|
|
498
|
+
}
|
|
499
|
+
}
|
|
500
|
+
|
|
501
|
+
alloc_base = new_alloc_base;
|
|
502
|
+
used = stk - stk_base;
|
|
503
|
+
*arg_alloc_base = alloc_base;
|
|
504
|
+
*arg_stk_base = (OnigStackType* )(alloc_base
|
|
505
|
+
+ (sizeof(OnigStackIndex) * msa->ptr_num));
|
|
506
|
+
*arg_stk = *arg_stk_base + used;
|
|
507
|
+
*arg_stk_end = *arg_stk_base + n;
|
|
508
|
+
return 0;
|
|
509
|
+
}
|
|
510
|
+
|
|
511
|
+
#define STACK_ENSURE(n) do {\
|
|
512
|
+
if (stk_end - stk < (n)) {\
|
|
513
|
+
int r = stack_double(is_alloca, &alloc_base, &stk_base, &stk_end, &stk,\
|
|
514
|
+
msa);\
|
|
515
|
+
if (r != 0) { STACK_SAVE; return r; } \
|
|
516
|
+
is_alloca = 0;\
|
|
517
|
+
UPDATE_FOR_STACK_REALLOC;\
|
|
518
|
+
}\
|
|
519
|
+
} while(0)
|
|
520
|
+
|
|
521
|
+
#define STACK_AT(index) (stk_base + (index))
|
|
522
|
+
#define GET_STACK_INDEX(stk) ((stk) - stk_base)
|
|
523
|
+
|
|
524
|
+
#define STACK_PUSH_TYPE(stack_type) do {\
|
|
525
|
+
STACK_ENSURE(1);\
|
|
526
|
+
stk->type = (stack_type);\
|
|
527
|
+
STACK_INC;\
|
|
528
|
+
} while(0)
|
|
529
|
+
|
|
530
|
+
#define IS_TO_VOID_TARGET(stk) (((stk)->type & STK_MASK_TO_VOID_TARGET) != 0)
|
|
531
|
+
|
|
532
|
+
#ifdef USE_COMBINATION_EXPLOSION_CHECK
|
|
533
|
+
#define STATE_CHECK_POS(s,snum) \
|
|
534
|
+
(((s) - str) * num_comb_exp_check + ((snum) - 1))
|
|
535
|
+
#define STATE_CHECK_VAL(v,snum) do {\
|
|
536
|
+
if (state_check_buff != NULL) {\
|
|
537
|
+
int x = STATE_CHECK_POS(s,snum);\
|
|
538
|
+
(v) = state_check_buff[x/8] & (1<<(x%8));\
|
|
539
|
+
}\
|
|
540
|
+
else (v) = 0;\
|
|
541
|
+
} while(0)
|
|
542
|
+
|
|
543
|
+
|
|
544
|
+
#define ELSE_IF_STATE_CHECK_MARK(stk) \
|
|
545
|
+
else if ((stk)->type == STK_STATE_CHECK_MARK) { \
|
|
546
|
+
int x = STATE_CHECK_POS(stk->u.state.pstr, stk->u.state.state_check);\
|
|
547
|
+
state_check_buff[x/8] |= (1<<(x%8)); \
|
|
548
|
+
}
|
|
549
|
+
|
|
550
|
+
#define STACK_PUSH(stack_type,pat,s,sprev) do {\
|
|
551
|
+
STACK_ENSURE(1);\
|
|
552
|
+
stk->type = (stack_type);\
|
|
553
|
+
stk->u.state.pcode = (pat);\
|
|
554
|
+
stk->u.state.pstr = (s);\
|
|
555
|
+
stk->u.state.pstr_prev = (sprev);\
|
|
556
|
+
stk->u.state.state_check = 0;\
|
|
557
|
+
STACK_INC;\
|
|
558
|
+
} while(0)
|
|
559
|
+
|
|
560
|
+
#define STACK_PUSH_ENSURED(stack_type,pat) do {\
|
|
561
|
+
stk->type = (stack_type);\
|
|
562
|
+
stk->u.state.pcode = (pat);\
|
|
563
|
+
stk->u.state.state_check = 0;\
|
|
564
|
+
STACK_INC;\
|
|
565
|
+
} while(0)
|
|
566
|
+
|
|
567
|
+
#define STACK_PUSH_ALT_WITH_STATE_CHECK(pat,s,sprev,snum) do {\
|
|
568
|
+
STACK_ENSURE(1);\
|
|
569
|
+
stk->type = STK_ALT;\
|
|
570
|
+
stk->u.state.pcode = (pat);\
|
|
571
|
+
stk->u.state.pstr = (s);\
|
|
572
|
+
stk->u.state.pstr_prev = (sprev);\
|
|
573
|
+
stk->u.state.state_check = ((state_check_buff != NULL) ? (snum) : 0);\
|
|
574
|
+
STACK_INC;\
|
|
575
|
+
} while(0)
|
|
576
|
+
|
|
577
|
+
#define STACK_PUSH_STATE_CHECK(s,snum) do {\
|
|
578
|
+
if (state_check_buff != NULL) {\
|
|
579
|
+
STACK_ENSURE(1);\
|
|
580
|
+
stk->type = STK_STATE_CHECK_MARK;\
|
|
581
|
+
stk->u.state.pstr = (s);\
|
|
582
|
+
stk->u.state.state_check = (snum);\
|
|
583
|
+
STACK_INC;\
|
|
584
|
+
}\
|
|
585
|
+
} while(0)
|
|
586
|
+
|
|
587
|
+
#else /* USE_COMBINATION_EXPLOSION_CHECK */
|
|
588
|
+
|
|
589
|
+
#define ELSE_IF_STATE_CHECK_MARK(stk)
|
|
590
|
+
|
|
591
|
+
#define STACK_PUSH(stack_type,pat,s,sprev) do {\
|
|
592
|
+
STACK_ENSURE(1);\
|
|
593
|
+
stk->type = (stack_type);\
|
|
594
|
+
stk->u.state.pcode = (pat);\
|
|
595
|
+
stk->u.state.pstr = (s);\
|
|
596
|
+
stk->u.state.pstr_prev = (sprev);\
|
|
597
|
+
STACK_INC;\
|
|
598
|
+
} while(0)
|
|
599
|
+
|
|
600
|
+
#define STACK_PUSH_ENSURED(stack_type,pat) do {\
|
|
601
|
+
stk->type = (stack_type);\
|
|
602
|
+
stk->u.state.pcode = (pat);\
|
|
603
|
+
STACK_INC;\
|
|
604
|
+
} while(0)
|
|
605
|
+
#endif /* USE_COMBINATION_EXPLOSION_CHECK */
|
|
606
|
+
|
|
607
|
+
#define STACK_PUSH_ALT(pat,s,sprev) STACK_PUSH(STK_ALT,pat,s,sprev)
|
|
608
|
+
#define STACK_PUSH_POS(s,sprev) STACK_PUSH(STK_POS,NULL_UCHARP,s,sprev)
|
|
609
|
+
#define STACK_PUSH_POS_NOT(pat,s,sprev) STACK_PUSH(STK_POS_NOT,pat,s,sprev)
|
|
610
|
+
#define STACK_PUSH_STOP_BT STACK_PUSH_TYPE(STK_STOP_BT)
|
|
611
|
+
#define STACK_PUSH_LOOK_BEHIND_NOT(pat,s,sprev) \
|
|
612
|
+
STACK_PUSH(STK_LOOK_BEHIND_NOT,pat,s,sprev)
|
|
613
|
+
|
|
614
|
+
#define STACK_PUSH_REPEAT(id, pat) do {\
|
|
615
|
+
STACK_ENSURE(1);\
|
|
616
|
+
stk->type = STK_REPEAT;\
|
|
617
|
+
stk->u.repeat.num = (id);\
|
|
618
|
+
stk->u.repeat.pcode = (pat);\
|
|
619
|
+
stk->u.repeat.count = 0;\
|
|
620
|
+
STACK_INC;\
|
|
621
|
+
} while(0)
|
|
622
|
+
|
|
623
|
+
#define STACK_PUSH_REPEAT_INC(sindex) do {\
|
|
624
|
+
STACK_ENSURE(1);\
|
|
625
|
+
stk->type = STK_REPEAT_INC;\
|
|
626
|
+
stk->u.repeat_inc.si = (sindex);\
|
|
627
|
+
STACK_INC;\
|
|
628
|
+
} while(0)
|
|
629
|
+
|
|
630
|
+
#define STACK_PUSH_MEM_START(mnum, s) do {\
|
|
631
|
+
STACK_ENSURE(1);\
|
|
632
|
+
stk->type = STK_MEM_START;\
|
|
633
|
+
stk->u.mem.num = (mnum);\
|
|
634
|
+
stk->u.mem.pstr = (s);\
|
|
635
|
+
stk->u.mem.start = mem_start_stk[mnum];\
|
|
636
|
+
stk->u.mem.end = mem_end_stk[mnum];\
|
|
637
|
+
mem_start_stk[mnum] = GET_STACK_INDEX(stk);\
|
|
638
|
+
mem_end_stk[mnum] = INVALID_STACK_INDEX;\
|
|
639
|
+
STACK_INC;\
|
|
640
|
+
} while(0)
|
|
641
|
+
|
|
642
|
+
#define STACK_PUSH_MEM_END(mnum, s) do {\
|
|
643
|
+
STACK_ENSURE(1);\
|
|
644
|
+
stk->type = STK_MEM_END;\
|
|
645
|
+
stk->u.mem.num = (mnum);\
|
|
646
|
+
stk->u.mem.pstr = (s);\
|
|
647
|
+
stk->u.mem.start = mem_start_stk[mnum];\
|
|
648
|
+
stk->u.mem.end = mem_end_stk[mnum];\
|
|
649
|
+
mem_end_stk[mnum] = GET_STACK_INDEX(stk);\
|
|
650
|
+
STACK_INC;\
|
|
651
|
+
} while(0)
|
|
652
|
+
|
|
653
|
+
#define STACK_PUSH_MEM_END_MARK(mnum) do {\
|
|
654
|
+
STACK_ENSURE(1);\
|
|
655
|
+
stk->type = STK_MEM_END_MARK;\
|
|
656
|
+
stk->u.mem.num = (mnum);\
|
|
657
|
+
STACK_INC;\
|
|
658
|
+
} while(0)
|
|
659
|
+
|
|
660
|
+
#define STACK_GET_MEM_START(mnum, k) do {\
|
|
661
|
+
int level = 0;\
|
|
662
|
+
k = stk;\
|
|
663
|
+
while (k > stk_base) {\
|
|
664
|
+
k--;\
|
|
665
|
+
if ((k->type & STK_MASK_MEM_END_OR_MARK) != 0 \
|
|
666
|
+
&& k->u.mem.num == (mnum)) {\
|
|
667
|
+
level++;\
|
|
668
|
+
}\
|
|
669
|
+
else if (k->type == STK_MEM_START && k->u.mem.num == (mnum)) {\
|
|
670
|
+
if (level == 0) break;\
|
|
671
|
+
level--;\
|
|
672
|
+
}\
|
|
673
|
+
}\
|
|
674
|
+
} while(0)
|
|
675
|
+
|
|
676
|
+
#define STACK_GET_MEM_RANGE(k, mnum, start, end) do {\
|
|
677
|
+
int level = 0;\
|
|
678
|
+
while (k < stk) {\
|
|
679
|
+
if (k->type == STK_MEM_START && k->u.mem.num == (mnum)) {\
|
|
680
|
+
if (level == 0) (start) = k->u.mem.pstr;\
|
|
681
|
+
level++;\
|
|
682
|
+
}\
|
|
683
|
+
else if (k->type == STK_MEM_END && k->u.mem.num == (mnum)) {\
|
|
684
|
+
level--;\
|
|
685
|
+
if (level == 0) {\
|
|
686
|
+
(end) = k->u.mem.pstr;\
|
|
687
|
+
break;\
|
|
688
|
+
}\
|
|
689
|
+
}\
|
|
690
|
+
k++;\
|
|
691
|
+
}\
|
|
692
|
+
} while(0)
|
|
693
|
+
|
|
694
|
+
#define STACK_PUSH_NULL_CHECK_START(cnum, s) do {\
|
|
695
|
+
STACK_ENSURE(1);\
|
|
696
|
+
stk->type = STK_NULL_CHECK_START;\
|
|
697
|
+
stk->u.null_check.num = (cnum);\
|
|
698
|
+
stk->u.null_check.pstr = (s);\
|
|
699
|
+
STACK_INC;\
|
|
700
|
+
} while(0)
|
|
701
|
+
|
|
702
|
+
#define STACK_PUSH_NULL_CHECK_END(cnum) do {\
|
|
703
|
+
STACK_ENSURE(1);\
|
|
704
|
+
stk->type = STK_NULL_CHECK_END;\
|
|
705
|
+
stk->u.null_check.num = (cnum);\
|
|
706
|
+
STACK_INC;\
|
|
707
|
+
} while(0)
|
|
708
|
+
|
|
709
|
+
#define STACK_PUSH_CALL_FRAME(pat) do {\
|
|
710
|
+
STACK_ENSURE(1);\
|
|
711
|
+
stk->type = STK_CALL_FRAME;\
|
|
712
|
+
stk->u.call_frame.ret_addr = (pat);\
|
|
713
|
+
STACK_INC;\
|
|
714
|
+
} while(0)
|
|
715
|
+
|
|
716
|
+
#define STACK_PUSH_RETURN do {\
|
|
717
|
+
STACK_ENSURE(1);\
|
|
718
|
+
stk->type = STK_RETURN;\
|
|
719
|
+
STACK_INC;\
|
|
720
|
+
} while(0)
|
|
721
|
+
|
|
722
|
+
|
|
723
|
+
#ifdef ONIG_DEBUG
|
|
724
|
+
#define STACK_BASE_CHECK(p, at) \
|
|
725
|
+
if ((p) < stk_base) {\
|
|
726
|
+
fprintf(stderr, "at %s\n", at);\
|
|
727
|
+
goto stack_error;\
|
|
728
|
+
}
|
|
729
|
+
#else
|
|
730
|
+
#define STACK_BASE_CHECK(p, at)
|
|
731
|
+
#endif
|
|
732
|
+
|
|
733
|
+
#define STACK_POP_ONE do {\
|
|
734
|
+
stk--;\
|
|
735
|
+
STACK_BASE_CHECK(stk, "STACK_POP_ONE"); \
|
|
736
|
+
} while(0)
|
|
737
|
+
|
|
738
|
+
#define STACK_POP do {\
|
|
739
|
+
switch (pop_level) {\
|
|
740
|
+
case STACK_POP_LEVEL_FREE:\
|
|
741
|
+
while (1) {\
|
|
742
|
+
stk--;\
|
|
743
|
+
STACK_BASE_CHECK(stk, "STACK_POP"); \
|
|
744
|
+
if ((stk->type & STK_MASK_POP_USED) != 0) break;\
|
|
745
|
+
ELSE_IF_STATE_CHECK_MARK(stk);\
|
|
746
|
+
}\
|
|
747
|
+
break;\
|
|
748
|
+
case STACK_POP_LEVEL_MEM_START:\
|
|
749
|
+
while (1) {\
|
|
750
|
+
stk--;\
|
|
751
|
+
STACK_BASE_CHECK(stk, "STACK_POP 2"); \
|
|
752
|
+
if ((stk->type & STK_MASK_POP_USED) != 0) break;\
|
|
753
|
+
else if (stk->type == STK_MEM_START) {\
|
|
754
|
+
mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
|
755
|
+
mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
|
756
|
+
}\
|
|
757
|
+
ELSE_IF_STATE_CHECK_MARK(stk);\
|
|
758
|
+
}\
|
|
759
|
+
break;\
|
|
760
|
+
default:\
|
|
761
|
+
while (1) {\
|
|
762
|
+
stk--;\
|
|
763
|
+
STACK_BASE_CHECK(stk, "STACK_POP 3"); \
|
|
764
|
+
if ((stk->type & STK_MASK_POP_USED) != 0) break;\
|
|
765
|
+
else if (stk->type == STK_MEM_START) {\
|
|
766
|
+
mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
|
767
|
+
mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
|
768
|
+
}\
|
|
769
|
+
else if (stk->type == STK_REPEAT_INC) {\
|
|
770
|
+
STACK_AT(stk->u.repeat_inc.si)->u.repeat.count--;\
|
|
771
|
+
}\
|
|
772
|
+
else if (stk->type == STK_MEM_END) {\
|
|
773
|
+
mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
|
774
|
+
mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
|
775
|
+
}\
|
|
776
|
+
ELSE_IF_STATE_CHECK_MARK(stk);\
|
|
777
|
+
}\
|
|
778
|
+
break;\
|
|
779
|
+
}\
|
|
780
|
+
} while(0)
|
|
781
|
+
|
|
782
|
+
#define STACK_POP_TIL_POS_NOT do {\
|
|
783
|
+
while (1) {\
|
|
784
|
+
stk--;\
|
|
785
|
+
STACK_BASE_CHECK(stk, "STACK_POP_TIL_POS_NOT"); \
|
|
786
|
+
if (stk->type == STK_POS_NOT) break;\
|
|
787
|
+
else if (stk->type == STK_MEM_START) {\
|
|
788
|
+
mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
|
789
|
+
mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
|
790
|
+
}\
|
|
791
|
+
else if (stk->type == STK_REPEAT_INC) {\
|
|
792
|
+
STACK_AT(stk->u.repeat_inc.si)->u.repeat.count--;\
|
|
793
|
+
}\
|
|
794
|
+
else if (stk->type == STK_MEM_END) {\
|
|
795
|
+
mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
|
796
|
+
mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
|
797
|
+
}\
|
|
798
|
+
ELSE_IF_STATE_CHECK_MARK(stk);\
|
|
799
|
+
}\
|
|
800
|
+
} while(0)
|
|
801
|
+
|
|
802
|
+
#define STACK_POP_TIL_LOOK_BEHIND_NOT do {\
|
|
803
|
+
while (1) {\
|
|
804
|
+
stk--;\
|
|
805
|
+
STACK_BASE_CHECK(stk, "STACK_POP_TIL_LOOK_BEHIND_NOT"); \
|
|
806
|
+
if (stk->type == STK_LOOK_BEHIND_NOT) break;\
|
|
807
|
+
else if (stk->type == STK_MEM_START) {\
|
|
808
|
+
mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
|
809
|
+
mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
|
810
|
+
}\
|
|
811
|
+
else if (stk->type == STK_REPEAT_INC) {\
|
|
812
|
+
STACK_AT(stk->u.repeat_inc.si)->u.repeat.count--;\
|
|
813
|
+
}\
|
|
814
|
+
else if (stk->type == STK_MEM_END) {\
|
|
815
|
+
mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
|
816
|
+
mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
|
817
|
+
}\
|
|
818
|
+
ELSE_IF_STATE_CHECK_MARK(stk);\
|
|
819
|
+
}\
|
|
820
|
+
} while(0)
|
|
821
|
+
|
|
822
|
+
#define STACK_POS_END(k) do {\
|
|
823
|
+
k = stk;\
|
|
824
|
+
while (1) {\
|
|
825
|
+
k--;\
|
|
826
|
+
STACK_BASE_CHECK(k, "STACK_POS_END"); \
|
|
827
|
+
if (IS_TO_VOID_TARGET(k)) {\
|
|
828
|
+
k->type = STK_VOID;\
|
|
829
|
+
}\
|
|
830
|
+
else if (k->type == STK_POS) {\
|
|
831
|
+
k->type = STK_VOID;\
|
|
832
|
+
break;\
|
|
833
|
+
}\
|
|
834
|
+
}\
|
|
835
|
+
} while(0)
|
|
836
|
+
|
|
837
|
+
#define STACK_STOP_BT_END do {\
|
|
838
|
+
OnigStackType *k = stk;\
|
|
839
|
+
while (1) {\
|
|
840
|
+
k--;\
|
|
841
|
+
STACK_BASE_CHECK(k, "STACK_STOP_BT_END"); \
|
|
842
|
+
if (IS_TO_VOID_TARGET(k)) {\
|
|
843
|
+
k->type = STK_VOID;\
|
|
844
|
+
}\
|
|
845
|
+
else if (k->type == STK_STOP_BT) {\
|
|
846
|
+
k->type = STK_VOID;\
|
|
847
|
+
break;\
|
|
848
|
+
}\
|
|
849
|
+
}\
|
|
850
|
+
} while(0)
|
|
851
|
+
|
|
852
|
+
#define STACK_NULL_CHECK(isnull,id,s) do {\
|
|
853
|
+
OnigStackType* k = stk;\
|
|
854
|
+
while (1) {\
|
|
855
|
+
k--;\
|
|
856
|
+
STACK_BASE_CHECK(k, "STACK_NULL_CHECK"); \
|
|
857
|
+
if (k->type == STK_NULL_CHECK_START) {\
|
|
858
|
+
if (k->u.null_check.num == (id)) {\
|
|
859
|
+
(isnull) = (k->u.null_check.pstr == (s));\
|
|
860
|
+
break;\
|
|
861
|
+
}\
|
|
862
|
+
}\
|
|
863
|
+
}\
|
|
864
|
+
} while(0)
|
|
865
|
+
|
|
866
|
+
#define STACK_NULL_CHECK_REC(isnull,id,s) do {\
|
|
867
|
+
int level = 0;\
|
|
868
|
+
OnigStackType* k = stk;\
|
|
869
|
+
while (1) {\
|
|
870
|
+
k--;\
|
|
871
|
+
STACK_BASE_CHECK(k, "STACK_NULL_CHECK_REC"); \
|
|
872
|
+
if (k->type == STK_NULL_CHECK_START) {\
|
|
873
|
+
if (k->u.null_check.num == (id)) {\
|
|
874
|
+
if (level == 0) {\
|
|
875
|
+
(isnull) = (k->u.null_check.pstr == (s));\
|
|
876
|
+
break;\
|
|
877
|
+
}\
|
|
878
|
+
else level--;\
|
|
879
|
+
}\
|
|
880
|
+
}\
|
|
881
|
+
else if (k->type == STK_NULL_CHECK_END) {\
|
|
882
|
+
level++;\
|
|
883
|
+
}\
|
|
884
|
+
}\
|
|
885
|
+
} while(0)
|
|
886
|
+
|
|
887
|
+
#define STACK_NULL_CHECK_MEMST(isnull,id,s,reg) do {\
|
|
888
|
+
OnigStackType* k = stk;\
|
|
889
|
+
while (1) {\
|
|
890
|
+
k--;\
|
|
891
|
+
STACK_BASE_CHECK(k, "STACK_NULL_CHECK_MEMST"); \
|
|
892
|
+
if (k->type == STK_NULL_CHECK_START) {\
|
|
893
|
+
if (k->u.null_check.num == (id)) {\
|
|
894
|
+
if (k->u.null_check.pstr != (s)) {\
|
|
895
|
+
(isnull) = 0;\
|
|
896
|
+
break;\
|
|
897
|
+
}\
|
|
898
|
+
else {\
|
|
899
|
+
UChar* endp;\
|
|
900
|
+
(isnull) = 1;\
|
|
901
|
+
while (k < stk) {\
|
|
902
|
+
if (k->type == STK_MEM_START) {\
|
|
903
|
+
if (k->u.mem.end == INVALID_STACK_INDEX) {\
|
|
904
|
+
(isnull) = 0; break;\
|
|
905
|
+
}\
|
|
906
|
+
if (BIT_STATUS_AT(reg->bt_mem_end, k->u.mem.num))\
|
|
907
|
+
endp = STACK_AT(k->u.mem.end)->u.mem.pstr;\
|
|
908
|
+
else\
|
|
909
|
+
endp = (UChar* )k->u.mem.end;\
|
|
910
|
+
if (STACK_AT(k->u.mem.start)->u.mem.pstr != endp) {\
|
|
911
|
+
(isnull) = 0; break;\
|
|
912
|
+
}\
|
|
913
|
+
else if (endp != s) {\
|
|
914
|
+
(isnull) = -1; /* empty, but position changed */ \
|
|
915
|
+
}\
|
|
916
|
+
}\
|
|
917
|
+
k++;\
|
|
918
|
+
}\
|
|
919
|
+
break;\
|
|
920
|
+
}\
|
|
921
|
+
}\
|
|
922
|
+
}\
|
|
923
|
+
}\
|
|
924
|
+
} while(0)
|
|
925
|
+
|
|
926
|
+
#define STACK_NULL_CHECK_MEMST_REC(isnull,id,s,reg) do {\
|
|
927
|
+
int level = 0;\
|
|
928
|
+
OnigStackType* k = stk;\
|
|
929
|
+
while (1) {\
|
|
930
|
+
k--;\
|
|
931
|
+
STACK_BASE_CHECK(k, "STACK_NULL_CHECK_MEMST_REC"); \
|
|
932
|
+
if (k->type == STK_NULL_CHECK_START) {\
|
|
933
|
+
if (k->u.null_check.num == (id)) {\
|
|
934
|
+
if (level == 0) {\
|
|
935
|
+
if (k->u.null_check.pstr != (s)) {\
|
|
936
|
+
(isnull) = 0;\
|
|
937
|
+
break;\
|
|
938
|
+
}\
|
|
939
|
+
else {\
|
|
940
|
+
UChar* endp;\
|
|
941
|
+
(isnull) = 1;\
|
|
942
|
+
while (k < stk) {\
|
|
943
|
+
if (k->type == STK_MEM_START) {\
|
|
944
|
+
if (k->u.mem.end == INVALID_STACK_INDEX) {\
|
|
945
|
+
(isnull) = 0; break;\
|
|
946
|
+
}\
|
|
947
|
+
if (BIT_STATUS_AT(reg->bt_mem_end, k->u.mem.num))\
|
|
948
|
+
endp = STACK_AT(k->u.mem.end)->u.mem.pstr;\
|
|
949
|
+
else\
|
|
950
|
+
endp = (UChar* )k->u.mem.end;\
|
|
951
|
+
if (STACK_AT(k->u.mem.start)->u.mem.pstr != endp) {\
|
|
952
|
+
(isnull) = 0; break;\
|
|
953
|
+
}\
|
|
954
|
+
else if (endp != s) {\
|
|
955
|
+
(isnull) = -1; /* empty, but position changed */ \
|
|
956
|
+
}\
|
|
957
|
+
}\
|
|
958
|
+
k++;\
|
|
959
|
+
}\
|
|
960
|
+
break;\
|
|
961
|
+
}\
|
|
962
|
+
}\
|
|
963
|
+
else {\
|
|
964
|
+
level--;\
|
|
965
|
+
}\
|
|
966
|
+
}\
|
|
967
|
+
}\
|
|
968
|
+
else if (k->type == STK_NULL_CHECK_END) {\
|
|
969
|
+
if (k->u.null_check.num == (id)) level++;\
|
|
970
|
+
}\
|
|
971
|
+
}\
|
|
972
|
+
} while(0)
|
|
973
|
+
|
|
974
|
+
#define STACK_GET_REPEAT(id, k) do {\
|
|
975
|
+
int level = 0;\
|
|
976
|
+
k = stk;\
|
|
977
|
+
while (1) {\
|
|
978
|
+
k--;\
|
|
979
|
+
STACK_BASE_CHECK(k, "STACK_GET_REPEAT"); \
|
|
980
|
+
if (k->type == STK_REPEAT) {\
|
|
981
|
+
if (level == 0) {\
|
|
982
|
+
if (k->u.repeat.num == (id)) {\
|
|
983
|
+
break;\
|
|
984
|
+
}\
|
|
985
|
+
}\
|
|
986
|
+
}\
|
|
987
|
+
else if (k->type == STK_CALL_FRAME) level--;\
|
|
988
|
+
else if (k->type == STK_RETURN) level++;\
|
|
989
|
+
}\
|
|
990
|
+
} while(0)
|
|
991
|
+
|
|
992
|
+
#define STACK_RETURN(addr) do {\
|
|
993
|
+
int level = 0;\
|
|
994
|
+
OnigStackType* k = stk;\
|
|
995
|
+
while (1) {\
|
|
996
|
+
k--;\
|
|
997
|
+
STACK_BASE_CHECK(k, "STACK_RETURN"); \
|
|
998
|
+
if (k->type == STK_CALL_FRAME) {\
|
|
999
|
+
if (level == 0) {\
|
|
1000
|
+
(addr) = k->u.call_frame.ret_addr;\
|
|
1001
|
+
break;\
|
|
1002
|
+
}\
|
|
1003
|
+
else level--;\
|
|
1004
|
+
}\
|
|
1005
|
+
else if (k->type == STK_RETURN)\
|
|
1006
|
+
level++;\
|
|
1007
|
+
}\
|
|
1008
|
+
} while(0)
|
|
1009
|
+
|
|
1010
|
+
|
|
1011
|
+
#define STRING_CMP(s1,s2,len) do {\
|
|
1012
|
+
while (len-- > 0) {\
|
|
1013
|
+
if (*s1++ != *s2++) goto fail;\
|
|
1014
|
+
}\
|
|
1015
|
+
} while(0)
|
|
1016
|
+
|
|
1017
|
+
#define STRING_CMP_IC(case_fold_flag,s1,ps2,len) do {\
|
|
1018
|
+
if (string_cmp_ic(encode, case_fold_flag, s1, ps2, len) == 0) \
|
|
1019
|
+
goto fail; \
|
|
1020
|
+
} while(0)
|
|
1021
|
+
|
|
1022
|
+
static int string_cmp_ic(OnigEncoding enc, int case_fold_flag,
|
|
1023
|
+
UChar* s1, UChar** ps2, int mblen)
|
|
1024
|
+
{
|
|
1025
|
+
UChar buf1[ONIGENC_MBC_CASE_FOLD_MAXLEN];
|
|
1026
|
+
UChar buf2[ONIGENC_MBC_CASE_FOLD_MAXLEN];
|
|
1027
|
+
UChar *p1, *p2, *end1, *s2, *end2;
|
|
1028
|
+
int len1, len2;
|
|
1029
|
+
|
|
1030
|
+
s2 = *ps2;
|
|
1031
|
+
end1 = s1 + mblen;
|
|
1032
|
+
end2 = s2 + mblen;
|
|
1033
|
+
while (s1 < end1) {
|
|
1034
|
+
len1 = ONIGENC_MBC_CASE_FOLD(enc, case_fold_flag, &s1, end1, buf1);
|
|
1035
|
+
len2 = ONIGENC_MBC_CASE_FOLD(enc, case_fold_flag, &s2, end2, buf2);
|
|
1036
|
+
if (len1 != len2) return 0;
|
|
1037
|
+
p1 = buf1;
|
|
1038
|
+
p2 = buf2;
|
|
1039
|
+
while (len1-- > 0) {
|
|
1040
|
+
if (*p1 != *p2) return 0;
|
|
1041
|
+
p1++;
|
|
1042
|
+
p2++;
|
|
1043
|
+
}
|
|
1044
|
+
}
|
|
1045
|
+
|
|
1046
|
+
*ps2 = s2;
|
|
1047
|
+
return 1;
|
|
1048
|
+
}
|
|
1049
|
+
|
|
1050
|
+
#define STRING_CMP_VALUE(s1,s2,len,is_fail) do {\
|
|
1051
|
+
is_fail = 0;\
|
|
1052
|
+
while (len-- > 0) {\
|
|
1053
|
+
if (*s1++ != *s2++) {\
|
|
1054
|
+
is_fail = 1; break;\
|
|
1055
|
+
}\
|
|
1056
|
+
}\
|
|
1057
|
+
} while(0)
|
|
1058
|
+
|
|
1059
|
+
#define STRING_CMP_VALUE_IC(case_fold_flag,s1,ps2,len,is_fail) do {\
|
|
1060
|
+
if (string_cmp_ic(encode, case_fold_flag, s1, ps2, len) == 0) \
|
|
1061
|
+
is_fail = 1; \
|
|
1062
|
+
else \
|
|
1063
|
+
is_fail = 0; \
|
|
1064
|
+
} while(0)
|
|
1065
|
+
|
|
1066
|
+
|
|
1067
|
+
#define IS_EMPTY_STR (str == end)
|
|
1068
|
+
#define ON_STR_BEGIN(s) ((s) == str)
|
|
1069
|
+
#define ON_STR_END(s) ((s) == end)
|
|
1070
|
+
#ifdef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
|
|
1071
|
+
#define DATA_ENSURE_CHECK1 (s < right_range)
|
|
1072
|
+
#define DATA_ENSURE_CHECK(n) (s + (n) <= right_range)
|
|
1073
|
+
#define DATA_ENSURE(n) if (s + (n) > right_range) goto fail
|
|
1074
|
+
#else
|
|
1075
|
+
#define DATA_ENSURE_CHECK1 (s < end)
|
|
1076
|
+
#define DATA_ENSURE_CHECK(n) (s + (n) <= end)
|
|
1077
|
+
#define DATA_ENSURE(n) if (s + (n) > end) goto fail
|
|
1078
|
+
#endif /* USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE */
|
|
1079
|
+
|
|
1080
|
+
|
|
1081
|
+
#ifdef USE_CAPTURE_HISTORY
|
|
1082
|
+
static int
|
|
1083
|
+
make_capture_history_tree(OnigCaptureTreeNode* node, OnigStackType** kp,
|
|
1084
|
+
OnigStackType* stk_top, UChar* str, regex_t* reg)
|
|
1085
|
+
{
|
|
1086
|
+
int n, r;
|
|
1087
|
+
OnigCaptureTreeNode* child;
|
|
1088
|
+
OnigStackType* k = *kp;
|
|
1089
|
+
|
|
1090
|
+
while (k < stk_top) {
|
|
1091
|
+
if (k->type == STK_MEM_START) {
|
|
1092
|
+
n = k->u.mem.num;
|
|
1093
|
+
if (n <= ONIG_MAX_CAPTURE_HISTORY_GROUP &&
|
|
1094
|
+
BIT_STATUS_AT(reg->capture_history, n) != 0) {
|
|
1095
|
+
child = history_node_new();
|
|
1096
|
+
CHECK_NULL_RETURN_MEMERR(child);
|
|
1097
|
+
child->group = n;
|
|
1098
|
+
child->beg = (int )(k->u.mem.pstr - str);
|
|
1099
|
+
r = history_tree_add_child(node, child);
|
|
1100
|
+
if (r != 0) return r;
|
|
1101
|
+
*kp = (k + 1);
|
|
1102
|
+
r = make_capture_history_tree(child, kp, stk_top, str, reg);
|
|
1103
|
+
if (r != 0) return r;
|
|
1104
|
+
|
|
1105
|
+
k = *kp;
|
|
1106
|
+
child->end = (int )(k->u.mem.pstr - str);
|
|
1107
|
+
}
|
|
1108
|
+
}
|
|
1109
|
+
else if (k->type == STK_MEM_END) {
|
|
1110
|
+
if (k->u.mem.num == node->group) {
|
|
1111
|
+
node->end = (int )(k->u.mem.pstr - str);
|
|
1112
|
+
*kp = k;
|
|
1113
|
+
return 0;
|
|
1114
|
+
}
|
|
1115
|
+
}
|
|
1116
|
+
k++;
|
|
1117
|
+
}
|
|
1118
|
+
|
|
1119
|
+
return 1; /* 1: root node ending. */
|
|
1120
|
+
}
|
|
1121
|
+
#endif
|
|
1122
|
+
|
|
1123
|
+
#ifdef USE_BACKREF_WITH_LEVEL
|
|
1124
|
+
static int mem_is_in_memp(int mem, int num, UChar* memp)
|
|
1125
|
+
{
|
|
1126
|
+
int i;
|
|
1127
|
+
MemNumType m;
|
|
1128
|
+
|
|
1129
|
+
for (i = 0; i < num; i++) {
|
|
1130
|
+
GET_MEMNUM_INC(m, memp);
|
|
1131
|
+
if (mem == (int )m) return 1;
|
|
1132
|
+
}
|
|
1133
|
+
return 0;
|
|
1134
|
+
}
|
|
1135
|
+
|
|
1136
|
+
static int backref_match_at_nested_level(regex_t* reg
|
|
1137
|
+
, OnigStackType* top, OnigStackType* stk_base
|
|
1138
|
+
, int ignore_case, int case_fold_flag
|
|
1139
|
+
, int nest, int mem_num, UChar* memp, UChar** s, const UChar* send)
|
|
1140
|
+
{
|
|
1141
|
+
UChar *ss, *p, *pstart, *pend = NULL_UCHARP;
|
|
1142
|
+
int level;
|
|
1143
|
+
OnigStackType* k;
|
|
1144
|
+
|
|
1145
|
+
level = 0;
|
|
1146
|
+
k = top;
|
|
1147
|
+
k--;
|
|
1148
|
+
while (k >= stk_base) {
|
|
1149
|
+
if (k->type == STK_CALL_FRAME) {
|
|
1150
|
+
level--;
|
|
1151
|
+
}
|
|
1152
|
+
else if (k->type == STK_RETURN) {
|
|
1153
|
+
level++;
|
|
1154
|
+
}
|
|
1155
|
+
else if (level == nest) {
|
|
1156
|
+
if (k->type == STK_MEM_START) {
|
|
1157
|
+
if (mem_is_in_memp(k->u.mem.num, mem_num, memp)) {
|
|
1158
|
+
pstart = k->u.mem.pstr;
|
|
1159
|
+
if (pend != NULL_UCHARP) {
|
|
1160
|
+
if (pend - pstart > send - *s) return 0; /* or goto next_mem; */
|
|
1161
|
+
p = pstart;
|
|
1162
|
+
ss = *s;
|
|
1163
|
+
|
|
1164
|
+
if (ignore_case != 0) {
|
|
1165
|
+
if (string_cmp_ic(reg->enc, case_fold_flag,
|
|
1166
|
+
pstart, &ss, (int )(pend - pstart)) == 0)
|
|
1167
|
+
return 0; /* or goto next_mem; */
|
|
1168
|
+
}
|
|
1169
|
+
else {
|
|
1170
|
+
while (p < pend) {
|
|
1171
|
+
if (*p++ != *ss++) return 0; /* or goto next_mem; */
|
|
1172
|
+
}
|
|
1173
|
+
}
|
|
1174
|
+
|
|
1175
|
+
*s = ss;
|
|
1176
|
+
return 1;
|
|
1177
|
+
}
|
|
1178
|
+
}
|
|
1179
|
+
}
|
|
1180
|
+
else if (k->type == STK_MEM_END) {
|
|
1181
|
+
if (mem_is_in_memp(k->u.mem.num, mem_num, memp)) {
|
|
1182
|
+
pend = k->u.mem.pstr;
|
|
1183
|
+
}
|
|
1184
|
+
}
|
|
1185
|
+
}
|
|
1186
|
+
k--;
|
|
1187
|
+
}
|
|
1188
|
+
|
|
1189
|
+
return 0;
|
|
1190
|
+
}
|
|
1191
|
+
#endif /* USE_BACKREF_WITH_LEVEL */
|
|
1192
|
+
|
|
1193
|
+
|
|
1194
|
+
#ifdef ONIG_DEBUG_STATISTICS
|
|
1195
|
+
|
|
1196
|
+
#define USE_TIMEOFDAY
|
|
1197
|
+
|
|
1198
|
+
#ifdef USE_TIMEOFDAY
|
|
1199
|
+
#ifdef HAVE_SYS_TIME_H
|
|
1200
|
+
#include <sys/time.h>
|
|
1201
|
+
#endif
|
|
1202
|
+
#ifdef HAVE_UNISTD_H
|
|
1203
|
+
#include <unistd.h>
|
|
1204
|
+
#endif
|
|
1205
|
+
static struct timeval ts, te;
|
|
1206
|
+
#define GETTIME(t) gettimeofday(&(t), (struct timezone* )0)
|
|
1207
|
+
#define TIMEDIFF(te,ts) (((te).tv_usec - (ts).tv_usec) + \
|
|
1208
|
+
(((te).tv_sec - (ts).tv_sec)*1000000))
|
|
1209
|
+
#else
|
|
1210
|
+
#ifdef HAVE_SYS_TIMES_H
|
|
1211
|
+
#include <sys/times.h>
|
|
1212
|
+
#endif
|
|
1213
|
+
static struct tms ts, te;
|
|
1214
|
+
#define GETTIME(t) times(&(t))
|
|
1215
|
+
#define TIMEDIFF(te,ts) ((te).tms_utime - (ts).tms_utime)
|
|
1216
|
+
#endif
|
|
1217
|
+
|
|
1218
|
+
static int OpCounter[256];
|
|
1219
|
+
static int OpPrevCounter[256];
|
|
1220
|
+
static unsigned long OpTime[256];
|
|
1221
|
+
static int OpCurr = OP_FINISH;
|
|
1222
|
+
static int OpPrevTarget = OP_FAIL;
|
|
1223
|
+
static int MaxStackDepth = 0;
|
|
1224
|
+
|
|
1225
|
+
#define MOP_IN(opcode) do {\
|
|
1226
|
+
if (opcode == OpPrevTarget) OpPrevCounter[OpCurr]++;\
|
|
1227
|
+
OpCurr = opcode;\
|
|
1228
|
+
OpCounter[opcode]++;\
|
|
1229
|
+
GETTIME(ts);\
|
|
1230
|
+
} while(0)
|
|
1231
|
+
|
|
1232
|
+
#define MOP_OUT do {\
|
|
1233
|
+
GETTIME(te);\
|
|
1234
|
+
OpTime[OpCurr] += TIMEDIFF(te, ts);\
|
|
1235
|
+
} while(0)
|
|
1236
|
+
|
|
1237
|
+
extern void
|
|
1238
|
+
onig_statistics_init(void)
|
|
1239
|
+
{
|
|
1240
|
+
int i;
|
|
1241
|
+
for (i = 0; i < 256; i++) {
|
|
1242
|
+
OpCounter[i] = OpPrevCounter[i] = 0; OpTime[i] = 0;
|
|
1243
|
+
}
|
|
1244
|
+
MaxStackDepth = 0;
|
|
1245
|
+
}
|
|
1246
|
+
|
|
1247
|
+
extern void
|
|
1248
|
+
onig_print_statistics(FILE* f)
|
|
1249
|
+
{
|
|
1250
|
+
int i;
|
|
1251
|
+
fprintf(f, " count prev time\n");
|
|
1252
|
+
for (i = 0; OnigOpInfo[i].opcode >= 0; i++) {
|
|
1253
|
+
fprintf(f, "%8d: %8d: %10ld: %s\n",
|
|
1254
|
+
OpCounter[i], OpPrevCounter[i], OpTime[i], OnigOpInfo[i].name);
|
|
1255
|
+
}
|
|
1256
|
+
fprintf(f, "\nmax stack depth: %d\n", MaxStackDepth);
|
|
1257
|
+
}
|
|
1258
|
+
|
|
1259
|
+
#define STACK_INC do {\
|
|
1260
|
+
stk++;\
|
|
1261
|
+
if (stk - stk_base > MaxStackDepth) \
|
|
1262
|
+
MaxStackDepth = stk - stk_base;\
|
|
1263
|
+
} while(0)
|
|
1264
|
+
|
|
1265
|
+
#else
|
|
1266
|
+
#define STACK_INC stk++
|
|
1267
|
+
|
|
1268
|
+
#define MOP_IN(opcode)
|
|
1269
|
+
#define MOP_OUT
|
|
1270
|
+
#endif
|
|
1271
|
+
|
|
1272
|
+
|
|
1273
|
+
/* matching region of POSIX API */
|
|
1274
|
+
typedef int regoff_t;
|
|
1275
|
+
|
|
1276
|
+
typedef struct {
|
|
1277
|
+
regoff_t rm_so;
|
|
1278
|
+
regoff_t rm_eo;
|
|
1279
|
+
} posix_regmatch_t;
|
|
1280
|
+
|
|
1281
|
+
/* match data(str - end) from position (sstart). */
|
|
1282
|
+
/* if sstart == str then set sprev to NULL. */
|
|
1283
|
+
static int
|
|
1284
|
+
match_at(regex_t* reg, const UChar* str, const UChar* end,
|
|
1285
|
+
#ifdef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
|
|
1286
|
+
const UChar* right_range,
|
|
1287
|
+
#endif
|
|
1288
|
+
const UChar* sstart, UChar* sprev, OnigMatchArg* msa)
|
|
1289
|
+
{
|
|
1290
|
+
static UChar FinishCode[] = { OP_FINISH };
|
|
1291
|
+
|
|
1292
|
+
int i, n, num_mem, best_len, pop_level;
|
|
1293
|
+
LengthType tlen, tlen2;
|
|
1294
|
+
MemNumType mem;
|
|
1295
|
+
RelAddrType addr;
|
|
1296
|
+
UChar *s, *q, *sbegin;
|
|
1297
|
+
int is_alloca;
|
|
1298
|
+
char *alloc_base;
|
|
1299
|
+
OnigStackType *stk_base, *stk, *stk_end;
|
|
1300
|
+
OnigStackType *stkp; /* used as any purpose. */
|
|
1301
|
+
OnigStackIndex si;
|
|
1302
|
+
OnigStackIndex *repeat_stk;
|
|
1303
|
+
OnigStackIndex *mem_start_stk, *mem_end_stk;
|
|
1304
|
+
#ifdef USE_COMBINATION_EXPLOSION_CHECK
|
|
1305
|
+
int scv;
|
|
1306
|
+
unsigned char* state_check_buff = msa->state_check_buff;
|
|
1307
|
+
int num_comb_exp_check = reg->num_comb_exp_check;
|
|
1308
|
+
#endif
|
|
1309
|
+
UChar *p = reg->p;
|
|
1310
|
+
OnigOptionType option = reg->options;
|
|
1311
|
+
OnigEncoding encode = reg->enc;
|
|
1312
|
+
OnigCaseFoldType case_fold_flag = reg->case_fold_flag;
|
|
1313
|
+
|
|
1314
|
+
//n = reg->num_repeat + reg->num_mem * 2;
|
|
1315
|
+
pop_level = reg->stack_pop_level;
|
|
1316
|
+
num_mem = reg->num_mem;
|
|
1317
|
+
STACK_INIT(INIT_MATCH_STACK_SIZE);
|
|
1318
|
+
UPDATE_FOR_STACK_REALLOC;
|
|
1319
|
+
for (i = 1; i <= num_mem; i++) {
|
|
1320
|
+
mem_start_stk[i] = mem_end_stk[i] = INVALID_STACK_INDEX;
|
|
1321
|
+
}
|
|
1322
|
+
|
|
1323
|
+
#ifdef ONIG_DEBUG_MATCH
|
|
1324
|
+
fprintf(stderr, "match_at: str: %d, end: %d, start: %d, sprev: %d\n",
|
|
1325
|
+
(int )str, (int )end, (int )sstart, (int )sprev);
|
|
1326
|
+
fprintf(stderr, "size: %d, start offset: %d\n",
|
|
1327
|
+
(int )(end - str), (int )(sstart - str));
|
|
1328
|
+
#endif
|
|
1329
|
+
|
|
1330
|
+
STACK_PUSH_ENSURED(STK_ALT, FinishCode); /* bottom stack */
|
|
1331
|
+
best_len = ONIG_MISMATCH;
|
|
1332
|
+
s = (UChar* )sstart;
|
|
1333
|
+
while (1) {
|
|
1334
|
+
#ifdef ONIG_DEBUG_MATCH
|
|
1335
|
+
{
|
|
1336
|
+
UChar *q, *bp, buf[50];
|
|
1337
|
+
int len;
|
|
1338
|
+
fprintf(stderr, "%4d> \"", (int )(s - str));
|
|
1339
|
+
bp = buf;
|
|
1340
|
+
for (i = 0, q = s; i < 7 && q < end; i++) {
|
|
1341
|
+
len = enclen(encode, q);
|
|
1342
|
+
while (len-- > 0) *bp++ = *q++;
|
|
1343
|
+
}
|
|
1344
|
+
if (q < end) { xmemcpy(bp, "...\"", 4); bp += 4; }
|
|
1345
|
+
else { xmemcpy(bp, "\"", 1); bp += 1; }
|
|
1346
|
+
*bp = 0;
|
|
1347
|
+
fputs((char* )buf, stderr);
|
|
1348
|
+
for (i = 0; i < 20 - (bp - buf); i++) fputc(' ', stderr);
|
|
1349
|
+
onig_print_compiled_byte_code(stderr, p, NULL, encode);
|
|
1350
|
+
fprintf(stderr, "\n");
|
|
1351
|
+
}
|
|
1352
|
+
#endif
|
|
1353
|
+
|
|
1354
|
+
sbegin = s;
|
|
1355
|
+
switch (*p++) {
|
|
1356
|
+
case OP_END: MOP_IN(OP_END);
|
|
1357
|
+
n = s - sstart;
|
|
1358
|
+
if (n > best_len) {
|
|
1359
|
+
OnigRegion* region;
|
|
1360
|
+
#ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
|
|
1361
|
+
if (IS_FIND_LONGEST(option)) {
|
|
1362
|
+
if (n > msa->best_len) {
|
|
1363
|
+
msa->best_len = n;
|
|
1364
|
+
msa->best_s = (UChar* )sstart;
|
|
1365
|
+
}
|
|
1366
|
+
else
|
|
1367
|
+
goto end_best_len;
|
|
1368
|
+
}
|
|
1369
|
+
#endif
|
|
1370
|
+
best_len = n;
|
|
1371
|
+
region = msa->region;
|
|
1372
|
+
if (region) {
|
|
1373
|
+
#ifdef USE_POSIX_API_REGION_OPTION
|
|
1374
|
+
if (IS_POSIX_REGION(msa->options)) {
|
|
1375
|
+
posix_regmatch_t* rmt = (posix_regmatch_t* )region;
|
|
1376
|
+
|
|
1377
|
+
rmt[0].rm_so = sstart - str;
|
|
1378
|
+
rmt[0].rm_eo = s - str;
|
|
1379
|
+
for (i = 1; i <= num_mem; i++) {
|
|
1380
|
+
if (mem_end_stk[i] != INVALID_STACK_INDEX) {
|
|
1381
|
+
if (BIT_STATUS_AT(reg->bt_mem_start, i))
|
|
1382
|
+
rmt[i].rm_so = STACK_AT(mem_start_stk[i])->u.mem.pstr - str;
|
|
1383
|
+
else
|
|
1384
|
+
rmt[i].rm_so = (UChar* )((void* )(mem_start_stk[i])) - str;
|
|
1385
|
+
|
|
1386
|
+
rmt[i].rm_eo = (BIT_STATUS_AT(reg->bt_mem_end, i)
|
|
1387
|
+
? STACK_AT(mem_end_stk[i])->u.mem.pstr
|
|
1388
|
+
: (UChar* )((void* )mem_end_stk[i])) - str;
|
|
1389
|
+
}
|
|
1390
|
+
else {
|
|
1391
|
+
rmt[i].rm_so = rmt[i].rm_eo = ONIG_REGION_NOTPOS;
|
|
1392
|
+
}
|
|
1393
|
+
}
|
|
1394
|
+
}
|
|
1395
|
+
else {
|
|
1396
|
+
#endif /* USE_POSIX_API_REGION_OPTION */
|
|
1397
|
+
region->beg[0] = sstart - str;
|
|
1398
|
+
region->end[0] = s - str;
|
|
1399
|
+
for (i = 1; i <= num_mem; i++) {
|
|
1400
|
+
if (mem_end_stk[i] != INVALID_STACK_INDEX) {
|
|
1401
|
+
if (BIT_STATUS_AT(reg->bt_mem_start, i))
|
|
1402
|
+
region->beg[i] = STACK_AT(mem_start_stk[i])->u.mem.pstr - str;
|
|
1403
|
+
else
|
|
1404
|
+
region->beg[i] = (UChar* )((void* )mem_start_stk[i]) - str;
|
|
1405
|
+
|
|
1406
|
+
region->end[i] = (BIT_STATUS_AT(reg->bt_mem_end, i)
|
|
1407
|
+
? STACK_AT(mem_end_stk[i])->u.mem.pstr
|
|
1408
|
+
: (UChar* )((void* )mem_end_stk[i])) - str;
|
|
1409
|
+
}
|
|
1410
|
+
else {
|
|
1411
|
+
region->beg[i] = region->end[i] = ONIG_REGION_NOTPOS;
|
|
1412
|
+
}
|
|
1413
|
+
}
|
|
1414
|
+
|
|
1415
|
+
#ifdef USE_CAPTURE_HISTORY
|
|
1416
|
+
if (reg->capture_history != 0) {
|
|
1417
|
+
int r;
|
|
1418
|
+
OnigCaptureTreeNode* node;
|
|
1419
|
+
|
|
1420
|
+
if (IS_NULL(region->history_root)) {
|
|
1421
|
+
region->history_root = node = history_node_new();
|
|
1422
|
+
CHECK_NULL_RETURN_MEMERR(node);
|
|
1423
|
+
}
|
|
1424
|
+
else {
|
|
1425
|
+
node = region->history_root;
|
|
1426
|
+
history_tree_clear(node);
|
|
1427
|
+
}
|
|
1428
|
+
|
|
1429
|
+
node->group = 0;
|
|
1430
|
+
node->beg = sstart - str;
|
|
1431
|
+
node->end = s - str;
|
|
1432
|
+
|
|
1433
|
+
stkp = stk_base;
|
|
1434
|
+
r = make_capture_history_tree(region->history_root, &stkp,
|
|
1435
|
+
stk, (UChar* )str, reg);
|
|
1436
|
+
if (r < 0) {
|
|
1437
|
+
best_len = r; /* error code */
|
|
1438
|
+
goto finish;
|
|
1439
|
+
}
|
|
1440
|
+
}
|
|
1441
|
+
#endif /* USE_CAPTURE_HISTORY */
|
|
1442
|
+
#ifdef USE_POSIX_API_REGION_OPTION
|
|
1443
|
+
} /* else IS_POSIX_REGION() */
|
|
1444
|
+
#endif
|
|
1445
|
+
} /* if (region) */
|
|
1446
|
+
} /* n > best_len */
|
|
1447
|
+
|
|
1448
|
+
#ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
|
|
1449
|
+
end_best_len:
|
|
1450
|
+
#endif
|
|
1451
|
+
MOP_OUT;
|
|
1452
|
+
|
|
1453
|
+
if (IS_FIND_CONDITION(option)) {
|
|
1454
|
+
if (IS_FIND_NOT_EMPTY(option) && s == sstart) {
|
|
1455
|
+
best_len = ONIG_MISMATCH;
|
|
1456
|
+
goto fail; /* for retry */
|
|
1457
|
+
}
|
|
1458
|
+
if (IS_FIND_LONGEST(option) && DATA_ENSURE_CHECK1) {
|
|
1459
|
+
goto fail; /* for retry */
|
|
1460
|
+
}
|
|
1461
|
+
}
|
|
1462
|
+
|
|
1463
|
+
/* default behavior: return first-matching result. */
|
|
1464
|
+
goto finish;
|
|
1465
|
+
break;
|
|
1466
|
+
|
|
1467
|
+
case OP_EXACT1: MOP_IN(OP_EXACT1);
|
|
1468
|
+
#if 0
|
|
1469
|
+
DATA_ENSURE(1);
|
|
1470
|
+
if (*p != *s) goto fail;
|
|
1471
|
+
p++; s++;
|
|
1472
|
+
#endif
|
|
1473
|
+
if (*p != *s++) goto fail;
|
|
1474
|
+
DATA_ENSURE(0);
|
|
1475
|
+
p++;
|
|
1476
|
+
MOP_OUT;
|
|
1477
|
+
break;
|
|
1478
|
+
|
|
1479
|
+
case OP_EXACT1_IC: MOP_IN(OP_EXACT1_IC);
|
|
1480
|
+
{
|
|
1481
|
+
int len;
|
|
1482
|
+
UChar *q, lowbuf[ONIGENC_MBC_CASE_FOLD_MAXLEN];
|
|
1483
|
+
|
|
1484
|
+
DATA_ENSURE(1);
|
|
1485
|
+
len = ONIGENC_MBC_CASE_FOLD(encode,
|
|
1486
|
+
/* DISABLE_CASE_FOLD_MULTI_CHAR(case_fold_flag), */
|
|
1487
|
+
case_fold_flag,
|
|
1488
|
+
&s, end, lowbuf);
|
|
1489
|
+
DATA_ENSURE(0);
|
|
1490
|
+
q = lowbuf;
|
|
1491
|
+
while (len-- > 0) {
|
|
1492
|
+
if (*p != *q) {
|
|
1493
|
+
goto fail;
|
|
1494
|
+
}
|
|
1495
|
+
p++; q++;
|
|
1496
|
+
}
|
|
1497
|
+
}
|
|
1498
|
+
MOP_OUT;
|
|
1499
|
+
break;
|
|
1500
|
+
|
|
1501
|
+
case OP_EXACT2: MOP_IN(OP_EXACT2);
|
|
1502
|
+
DATA_ENSURE(2);
|
|
1503
|
+
if (*p != *s) goto fail;
|
|
1504
|
+
p++; s++;
|
|
1505
|
+
if (*p != *s) goto fail;
|
|
1506
|
+
sprev = s;
|
|
1507
|
+
p++; s++;
|
|
1508
|
+
MOP_OUT;
|
|
1509
|
+
continue;
|
|
1510
|
+
break;
|
|
1511
|
+
|
|
1512
|
+
case OP_EXACT3: MOP_IN(OP_EXACT3);
|
|
1513
|
+
DATA_ENSURE(3);
|
|
1514
|
+
if (*p != *s) goto fail;
|
|
1515
|
+
p++; s++;
|
|
1516
|
+
if (*p != *s) goto fail;
|
|
1517
|
+
p++; s++;
|
|
1518
|
+
if (*p != *s) goto fail;
|
|
1519
|
+
sprev = s;
|
|
1520
|
+
p++; s++;
|
|
1521
|
+
MOP_OUT;
|
|
1522
|
+
continue;
|
|
1523
|
+
break;
|
|
1524
|
+
|
|
1525
|
+
case OP_EXACT4: MOP_IN(OP_EXACT4);
|
|
1526
|
+
DATA_ENSURE(4);
|
|
1527
|
+
if (*p != *s) goto fail;
|
|
1528
|
+
p++; s++;
|
|
1529
|
+
if (*p != *s) goto fail;
|
|
1530
|
+
p++; s++;
|
|
1531
|
+
if (*p != *s) goto fail;
|
|
1532
|
+
p++; s++;
|
|
1533
|
+
if (*p != *s) goto fail;
|
|
1534
|
+
sprev = s;
|
|
1535
|
+
p++; s++;
|
|
1536
|
+
MOP_OUT;
|
|
1537
|
+
continue;
|
|
1538
|
+
break;
|
|
1539
|
+
|
|
1540
|
+
case OP_EXACT5: MOP_IN(OP_EXACT5);
|
|
1541
|
+
DATA_ENSURE(5);
|
|
1542
|
+
if (*p != *s) goto fail;
|
|
1543
|
+
p++; s++;
|
|
1544
|
+
if (*p != *s) goto fail;
|
|
1545
|
+
p++; s++;
|
|
1546
|
+
if (*p != *s) goto fail;
|
|
1547
|
+
p++; s++;
|
|
1548
|
+
if (*p != *s) goto fail;
|
|
1549
|
+
p++; s++;
|
|
1550
|
+
if (*p != *s) goto fail;
|
|
1551
|
+
sprev = s;
|
|
1552
|
+
p++; s++;
|
|
1553
|
+
MOP_OUT;
|
|
1554
|
+
continue;
|
|
1555
|
+
break;
|
|
1556
|
+
|
|
1557
|
+
case OP_EXACTN: MOP_IN(OP_EXACTN);
|
|
1558
|
+
GET_LENGTH_INC(tlen, p);
|
|
1559
|
+
DATA_ENSURE(tlen);
|
|
1560
|
+
while (tlen-- > 0) {
|
|
1561
|
+
if (*p++ != *s++) goto fail;
|
|
1562
|
+
}
|
|
1563
|
+
sprev = s - 1;
|
|
1564
|
+
MOP_OUT;
|
|
1565
|
+
continue;
|
|
1566
|
+
break;
|
|
1567
|
+
|
|
1568
|
+
case OP_EXACTN_IC: MOP_IN(OP_EXACTN_IC);
|
|
1569
|
+
{
|
|
1570
|
+
int len;
|
|
1571
|
+
UChar *q, *endp, lowbuf[ONIGENC_MBC_CASE_FOLD_MAXLEN];
|
|
1572
|
+
|
|
1573
|
+
GET_LENGTH_INC(tlen, p);
|
|
1574
|
+
endp = p + tlen;
|
|
1575
|
+
|
|
1576
|
+
while (p < endp) {
|
|
1577
|
+
sprev = s;
|
|
1578
|
+
DATA_ENSURE(1);
|
|
1579
|
+
len = ONIGENC_MBC_CASE_FOLD(encode,
|
|
1580
|
+
/* DISABLE_CASE_FOLD_MULTI_CHAR(case_fold_flag), */
|
|
1581
|
+
case_fold_flag,
|
|
1582
|
+
&s, end, lowbuf);
|
|
1583
|
+
DATA_ENSURE(0);
|
|
1584
|
+
q = lowbuf;
|
|
1585
|
+
while (len-- > 0) {
|
|
1586
|
+
if (*p != *q) goto fail;
|
|
1587
|
+
p++; q++;
|
|
1588
|
+
}
|
|
1589
|
+
}
|
|
1590
|
+
}
|
|
1591
|
+
|
|
1592
|
+
MOP_OUT;
|
|
1593
|
+
continue;
|
|
1594
|
+
break;
|
|
1595
|
+
|
|
1596
|
+
case OP_EXACTMB2N1: MOP_IN(OP_EXACTMB2N1);
|
|
1597
|
+
DATA_ENSURE(2);
|
|
1598
|
+
if (*p != *s) goto fail;
|
|
1599
|
+
p++; s++;
|
|
1600
|
+
if (*p != *s) goto fail;
|
|
1601
|
+
p++; s++;
|
|
1602
|
+
MOP_OUT;
|
|
1603
|
+
break;
|
|
1604
|
+
|
|
1605
|
+
case OP_EXACTMB2N2: MOP_IN(OP_EXACTMB2N2);
|
|
1606
|
+
DATA_ENSURE(4);
|
|
1607
|
+
if (*p != *s) goto fail;
|
|
1608
|
+
p++; s++;
|
|
1609
|
+
if (*p != *s) goto fail;
|
|
1610
|
+
p++; s++;
|
|
1611
|
+
sprev = s;
|
|
1612
|
+
if (*p != *s) goto fail;
|
|
1613
|
+
p++; s++;
|
|
1614
|
+
if (*p != *s) goto fail;
|
|
1615
|
+
p++; s++;
|
|
1616
|
+
MOP_OUT;
|
|
1617
|
+
continue;
|
|
1618
|
+
break;
|
|
1619
|
+
|
|
1620
|
+
case OP_EXACTMB2N3: MOP_IN(OP_EXACTMB2N3);
|
|
1621
|
+
DATA_ENSURE(6);
|
|
1622
|
+
if (*p != *s) goto fail;
|
|
1623
|
+
p++; s++;
|
|
1624
|
+
if (*p != *s) goto fail;
|
|
1625
|
+
p++; s++;
|
|
1626
|
+
if (*p != *s) goto fail;
|
|
1627
|
+
p++; s++;
|
|
1628
|
+
if (*p != *s) goto fail;
|
|
1629
|
+
p++; s++;
|
|
1630
|
+
sprev = s;
|
|
1631
|
+
if (*p != *s) goto fail;
|
|
1632
|
+
p++; s++;
|
|
1633
|
+
if (*p != *s) goto fail;
|
|
1634
|
+
p++; s++;
|
|
1635
|
+
MOP_OUT;
|
|
1636
|
+
continue;
|
|
1637
|
+
break;
|
|
1638
|
+
|
|
1639
|
+
case OP_EXACTMB2N: MOP_IN(OP_EXACTMB2N);
|
|
1640
|
+
GET_LENGTH_INC(tlen, p);
|
|
1641
|
+
DATA_ENSURE(tlen * 2);
|
|
1642
|
+
while (tlen-- > 0) {
|
|
1643
|
+
if (*p != *s) goto fail;
|
|
1644
|
+
p++; s++;
|
|
1645
|
+
if (*p != *s) goto fail;
|
|
1646
|
+
p++; s++;
|
|
1647
|
+
}
|
|
1648
|
+
sprev = s - 2;
|
|
1649
|
+
MOP_OUT;
|
|
1650
|
+
continue;
|
|
1651
|
+
break;
|
|
1652
|
+
|
|
1653
|
+
case OP_EXACTMB3N: MOP_IN(OP_EXACTMB3N);
|
|
1654
|
+
GET_LENGTH_INC(tlen, p);
|
|
1655
|
+
DATA_ENSURE(tlen * 3);
|
|
1656
|
+
while (tlen-- > 0) {
|
|
1657
|
+
if (*p != *s) goto fail;
|
|
1658
|
+
p++; s++;
|
|
1659
|
+
if (*p != *s) goto fail;
|
|
1660
|
+
p++; s++;
|
|
1661
|
+
if (*p != *s) goto fail;
|
|
1662
|
+
p++; s++;
|
|
1663
|
+
}
|
|
1664
|
+
sprev = s - 3;
|
|
1665
|
+
MOP_OUT;
|
|
1666
|
+
continue;
|
|
1667
|
+
break;
|
|
1668
|
+
|
|
1669
|
+
case OP_EXACTMBN: MOP_IN(OP_EXACTMBN);
|
|
1670
|
+
GET_LENGTH_INC(tlen, p); /* mb-len */
|
|
1671
|
+
GET_LENGTH_INC(tlen2, p); /* string len */
|
|
1672
|
+
tlen2 *= tlen;
|
|
1673
|
+
DATA_ENSURE(tlen2);
|
|
1674
|
+
while (tlen2-- > 0) {
|
|
1675
|
+
if (*p != *s) goto fail;
|
|
1676
|
+
p++; s++;
|
|
1677
|
+
}
|
|
1678
|
+
sprev = s - tlen;
|
|
1679
|
+
MOP_OUT;
|
|
1680
|
+
continue;
|
|
1681
|
+
break;
|
|
1682
|
+
|
|
1683
|
+
case OP_CCLASS: MOP_IN(OP_CCLASS);
|
|
1684
|
+
DATA_ENSURE(1);
|
|
1685
|
+
if (BITSET_AT(((BitSetRef )p), *s) == 0) goto fail;
|
|
1686
|
+
p += SIZE_BITSET;
|
|
1687
|
+
s += enclen(encode, s); /* OP_CCLASS can match mb-code. \D, \S */
|
|
1688
|
+
MOP_OUT;
|
|
1689
|
+
break;
|
|
1690
|
+
|
|
1691
|
+
case OP_CCLASS_MB: MOP_IN(OP_CCLASS_MB);
|
|
1692
|
+
if (! ONIGENC_IS_MBC_HEAD(encode, s)) goto fail;
|
|
1693
|
+
|
|
1694
|
+
cclass_mb:
|
|
1695
|
+
GET_LENGTH_INC(tlen, p);
|
|
1696
|
+
{
|
|
1697
|
+
OnigCodePoint code;
|
|
1698
|
+
UChar *ss;
|
|
1699
|
+
int mb_len;
|
|
1700
|
+
|
|
1701
|
+
DATA_ENSURE(1);
|
|
1702
|
+
mb_len = enclen(encode, s);
|
|
1703
|
+
DATA_ENSURE(mb_len);
|
|
1704
|
+
ss = s;
|
|
1705
|
+
s += mb_len;
|
|
1706
|
+
code = ONIGENC_MBC_TO_CODE(encode, ss, s);
|
|
1707
|
+
|
|
1708
|
+
#ifdef PLATFORM_UNALIGNED_WORD_ACCESS
|
|
1709
|
+
if (! onig_is_in_code_range(p, code)) goto fail;
|
|
1710
|
+
#else
|
|
1711
|
+
q = p;
|
|
1712
|
+
ALIGNMENT_RIGHT(q);
|
|
1713
|
+
if (! onig_is_in_code_range(q, code)) goto fail;
|
|
1714
|
+
#endif
|
|
1715
|
+
}
|
|
1716
|
+
p += tlen;
|
|
1717
|
+
MOP_OUT;
|
|
1718
|
+
break;
|
|
1719
|
+
|
|
1720
|
+
case OP_CCLASS_MIX: MOP_IN(OP_CCLASS_MIX);
|
|
1721
|
+
DATA_ENSURE(1);
|
|
1722
|
+
if (ONIGENC_IS_MBC_HEAD(encode, s)) {
|
|
1723
|
+
p += SIZE_BITSET;
|
|
1724
|
+
goto cclass_mb;
|
|
1725
|
+
}
|
|
1726
|
+
else {
|
|
1727
|
+
if (BITSET_AT(((BitSetRef )p), *s) == 0)
|
|
1728
|
+
goto fail;
|
|
1729
|
+
|
|
1730
|
+
p += SIZE_BITSET;
|
|
1731
|
+
GET_LENGTH_INC(tlen, p);
|
|
1732
|
+
p += tlen;
|
|
1733
|
+
s++;
|
|
1734
|
+
}
|
|
1735
|
+
MOP_OUT;
|
|
1736
|
+
break;
|
|
1737
|
+
|
|
1738
|
+
case OP_CCLASS_NOT: MOP_IN(OP_CCLASS_NOT);
|
|
1739
|
+
DATA_ENSURE(1);
|
|
1740
|
+
if (BITSET_AT(((BitSetRef )p), *s) != 0) goto fail;
|
|
1741
|
+
p += SIZE_BITSET;
|
|
1742
|
+
s += enclen(encode, s);
|
|
1743
|
+
MOP_OUT;
|
|
1744
|
+
break;
|
|
1745
|
+
|
|
1746
|
+
case OP_CCLASS_MB_NOT: MOP_IN(OP_CCLASS_MB_NOT);
|
|
1747
|
+
DATA_ENSURE(1);
|
|
1748
|
+
if (! ONIGENC_IS_MBC_HEAD(encode, s)) {
|
|
1749
|
+
s++;
|
|
1750
|
+
GET_LENGTH_INC(tlen, p);
|
|
1751
|
+
p += tlen;
|
|
1752
|
+
goto cc_mb_not_success;
|
|
1753
|
+
}
|
|
1754
|
+
|
|
1755
|
+
cclass_mb_not:
|
|
1756
|
+
GET_LENGTH_INC(tlen, p);
|
|
1757
|
+
{
|
|
1758
|
+
OnigCodePoint code;
|
|
1759
|
+
UChar *ss;
|
|
1760
|
+
int mb_len = enclen(encode, s);
|
|
1761
|
+
|
|
1762
|
+
if (! DATA_ENSURE_CHECK(mb_len)) {
|
|
1763
|
+
DATA_ENSURE(1);
|
|
1764
|
+
s = (UChar* )end;
|
|
1765
|
+
p += tlen;
|
|
1766
|
+
goto cc_mb_not_success;
|
|
1767
|
+
}
|
|
1768
|
+
|
|
1769
|
+
ss = s;
|
|
1770
|
+
s += mb_len;
|
|
1771
|
+
code = ONIGENC_MBC_TO_CODE(encode, ss, s);
|
|
1772
|
+
|
|
1773
|
+
#ifdef PLATFORM_UNALIGNED_WORD_ACCESS
|
|
1774
|
+
if (onig_is_in_code_range(p, code)) goto fail;
|
|
1775
|
+
#else
|
|
1776
|
+
q = p;
|
|
1777
|
+
ALIGNMENT_RIGHT(q);
|
|
1778
|
+
if (onig_is_in_code_range(q, code)) goto fail;
|
|
1779
|
+
#endif
|
|
1780
|
+
}
|
|
1781
|
+
p += tlen;
|
|
1782
|
+
|
|
1783
|
+
cc_mb_not_success:
|
|
1784
|
+
MOP_OUT;
|
|
1785
|
+
break;
|
|
1786
|
+
|
|
1787
|
+
case OP_CCLASS_MIX_NOT: MOP_IN(OP_CCLASS_MIX_NOT);
|
|
1788
|
+
DATA_ENSURE(1);
|
|
1789
|
+
if (ONIGENC_IS_MBC_HEAD(encode, s)) {
|
|
1790
|
+
p += SIZE_BITSET;
|
|
1791
|
+
goto cclass_mb_not;
|
|
1792
|
+
}
|
|
1793
|
+
else {
|
|
1794
|
+
if (BITSET_AT(((BitSetRef )p), *s) != 0)
|
|
1795
|
+
goto fail;
|
|
1796
|
+
|
|
1797
|
+
p += SIZE_BITSET;
|
|
1798
|
+
GET_LENGTH_INC(tlen, p);
|
|
1799
|
+
p += tlen;
|
|
1800
|
+
s++;
|
|
1801
|
+
}
|
|
1802
|
+
MOP_OUT;
|
|
1803
|
+
break;
|
|
1804
|
+
|
|
1805
|
+
case OP_CCLASS_NODE: MOP_IN(OP_CCLASS_NODE);
|
|
1806
|
+
{
|
|
1807
|
+
OnigCodePoint code;
|
|
1808
|
+
void *node;
|
|
1809
|
+
int mb_len;
|
|
1810
|
+
UChar *ss;
|
|
1811
|
+
|
|
1812
|
+
DATA_ENSURE(1);
|
|
1813
|
+
GET_POINTER_INC(node, p);
|
|
1814
|
+
mb_len = enclen(encode, s);
|
|
1815
|
+
ss = s;
|
|
1816
|
+
s += mb_len;
|
|
1817
|
+
DATA_ENSURE(0);
|
|
1818
|
+
code = ONIGENC_MBC_TO_CODE(encode, ss, s);
|
|
1819
|
+
if (onig_is_code_in_cc_len(mb_len, code, node) == 0) goto fail;
|
|
1820
|
+
}
|
|
1821
|
+
MOP_OUT;
|
|
1822
|
+
break;
|
|
1823
|
+
|
|
1824
|
+
case OP_ANYCHAR: MOP_IN(OP_ANYCHAR);
|
|
1825
|
+
DATA_ENSURE(1);
|
|
1826
|
+
n = enclen(encode, s);
|
|
1827
|
+
DATA_ENSURE(n);
|
|
1828
|
+
if (ONIGENC_IS_MBC_NEWLINE(encode, s, end)) goto fail;
|
|
1829
|
+
s += n;
|
|
1830
|
+
MOP_OUT;
|
|
1831
|
+
break;
|
|
1832
|
+
|
|
1833
|
+
case OP_ANYCHAR_ML: MOP_IN(OP_ANYCHAR_ML);
|
|
1834
|
+
DATA_ENSURE(1);
|
|
1835
|
+
n = enclen(encode, s);
|
|
1836
|
+
DATA_ENSURE(n);
|
|
1837
|
+
s += n;
|
|
1838
|
+
MOP_OUT;
|
|
1839
|
+
break;
|
|
1840
|
+
|
|
1841
|
+
case OP_ANYCHAR_STAR: MOP_IN(OP_ANYCHAR_STAR);
|
|
1842
|
+
while (DATA_ENSURE_CHECK1) {
|
|
1843
|
+
STACK_PUSH_ALT(p, s, sprev);
|
|
1844
|
+
n = enclen(encode, s);
|
|
1845
|
+
DATA_ENSURE(n);
|
|
1846
|
+
if (ONIGENC_IS_MBC_NEWLINE(encode, s, end)) goto fail;
|
|
1847
|
+
sprev = s;
|
|
1848
|
+
s += n;
|
|
1849
|
+
}
|
|
1850
|
+
MOP_OUT;
|
|
1851
|
+
break;
|
|
1852
|
+
|
|
1853
|
+
case OP_ANYCHAR_ML_STAR: MOP_IN(OP_ANYCHAR_ML_STAR);
|
|
1854
|
+
while (DATA_ENSURE_CHECK1) {
|
|
1855
|
+
STACK_PUSH_ALT(p, s, sprev);
|
|
1856
|
+
n = enclen(encode, s);
|
|
1857
|
+
if (n > 1) {
|
|
1858
|
+
DATA_ENSURE(n);
|
|
1859
|
+
sprev = s;
|
|
1860
|
+
s += n;
|
|
1861
|
+
}
|
|
1862
|
+
else {
|
|
1863
|
+
sprev = s;
|
|
1864
|
+
s++;
|
|
1865
|
+
}
|
|
1866
|
+
}
|
|
1867
|
+
MOP_OUT;
|
|
1868
|
+
break;
|
|
1869
|
+
|
|
1870
|
+
case OP_ANYCHAR_STAR_PEEK_NEXT: MOP_IN(OP_ANYCHAR_STAR_PEEK_NEXT);
|
|
1871
|
+
while (DATA_ENSURE_CHECK1) {
|
|
1872
|
+
if (*p == *s) {
|
|
1873
|
+
STACK_PUSH_ALT(p + 1, s, sprev);
|
|
1874
|
+
}
|
|
1875
|
+
n = enclen(encode, s);
|
|
1876
|
+
DATA_ENSURE(n);
|
|
1877
|
+
if (ONIGENC_IS_MBC_NEWLINE(encode, s, end)) goto fail;
|
|
1878
|
+
sprev = s;
|
|
1879
|
+
s += n;
|
|
1880
|
+
}
|
|
1881
|
+
p++;
|
|
1882
|
+
MOP_OUT;
|
|
1883
|
+
break;
|
|
1884
|
+
|
|
1885
|
+
case OP_ANYCHAR_ML_STAR_PEEK_NEXT:MOP_IN(OP_ANYCHAR_ML_STAR_PEEK_NEXT);
|
|
1886
|
+
while (DATA_ENSURE_CHECK1) {
|
|
1887
|
+
if (*p == *s) {
|
|
1888
|
+
STACK_PUSH_ALT(p + 1, s, sprev);
|
|
1889
|
+
}
|
|
1890
|
+
n = enclen(encode, s);
|
|
1891
|
+
if (n > 1) {
|
|
1892
|
+
DATA_ENSURE(n);
|
|
1893
|
+
sprev = s;
|
|
1894
|
+
s += n;
|
|
1895
|
+
}
|
|
1896
|
+
else {
|
|
1897
|
+
sprev = s;
|
|
1898
|
+
s++;
|
|
1899
|
+
}
|
|
1900
|
+
}
|
|
1901
|
+
p++;
|
|
1902
|
+
MOP_OUT;
|
|
1903
|
+
break;
|
|
1904
|
+
|
|
1905
|
+
#ifdef USE_COMBINATION_EXPLOSION_CHECK
|
|
1906
|
+
case OP_STATE_CHECK_ANYCHAR_STAR: MOP_IN(OP_STATE_CHECK_ANYCHAR_STAR);
|
|
1907
|
+
GET_STATE_CHECK_NUM_INC(mem, p);
|
|
1908
|
+
while (DATA_ENSURE_CHECK1) {
|
|
1909
|
+
STATE_CHECK_VAL(scv, mem);
|
|
1910
|
+
if (scv) goto fail;
|
|
1911
|
+
|
|
1912
|
+
STACK_PUSH_ALT_WITH_STATE_CHECK(p, s, sprev, mem);
|
|
1913
|
+
n = enclen(encode, s);
|
|
1914
|
+
DATA_ENSURE(n);
|
|
1915
|
+
if (ONIGENC_IS_MBC_NEWLINE(encode, s, end)) goto fail;
|
|
1916
|
+
sprev = s;
|
|
1917
|
+
s += n;
|
|
1918
|
+
}
|
|
1919
|
+
MOP_OUT;
|
|
1920
|
+
break;
|
|
1921
|
+
|
|
1922
|
+
case OP_STATE_CHECK_ANYCHAR_ML_STAR:
|
|
1923
|
+
MOP_IN(OP_STATE_CHECK_ANYCHAR_ML_STAR);
|
|
1924
|
+
|
|
1925
|
+
GET_STATE_CHECK_NUM_INC(mem, p);
|
|
1926
|
+
while (DATA_ENSURE_CHECK1) {
|
|
1927
|
+
STATE_CHECK_VAL(scv, mem);
|
|
1928
|
+
if (scv) goto fail;
|
|
1929
|
+
|
|
1930
|
+
STACK_PUSH_ALT_WITH_STATE_CHECK(p, s, sprev, mem);
|
|
1931
|
+
n = enclen(encode, s);
|
|
1932
|
+
if (n > 1) {
|
|
1933
|
+
DATA_ENSURE(n);
|
|
1934
|
+
sprev = s;
|
|
1935
|
+
s += n;
|
|
1936
|
+
}
|
|
1937
|
+
else {
|
|
1938
|
+
sprev = s;
|
|
1939
|
+
s++;
|
|
1940
|
+
}
|
|
1941
|
+
}
|
|
1942
|
+
MOP_OUT;
|
|
1943
|
+
break;
|
|
1944
|
+
#endif /* USE_COMBINATION_EXPLOSION_CHECK */
|
|
1945
|
+
|
|
1946
|
+
case OP_WORD: MOP_IN(OP_WORD);
|
|
1947
|
+
DATA_ENSURE(1);
|
|
1948
|
+
if (! ONIGENC_IS_MBC_WORD(encode, s, end))
|
|
1949
|
+
goto fail;
|
|
1950
|
+
|
|
1951
|
+
s += enclen(encode, s);
|
|
1952
|
+
MOP_OUT;
|
|
1953
|
+
break;
|
|
1954
|
+
|
|
1955
|
+
case OP_NOT_WORD: MOP_IN(OP_NOT_WORD);
|
|
1956
|
+
DATA_ENSURE(1);
|
|
1957
|
+
if (ONIGENC_IS_MBC_WORD(encode, s, end))
|
|
1958
|
+
goto fail;
|
|
1959
|
+
|
|
1960
|
+
s += enclen(encode, s);
|
|
1961
|
+
MOP_OUT;
|
|
1962
|
+
break;
|
|
1963
|
+
|
|
1964
|
+
case OP_WORD_BOUND: MOP_IN(OP_WORD_BOUND);
|
|
1965
|
+
if (ON_STR_BEGIN(s)) {
|
|
1966
|
+
DATA_ENSURE(1);
|
|
1967
|
+
if (! ONIGENC_IS_MBC_WORD(encode, s, end))
|
|
1968
|
+
goto fail;
|
|
1969
|
+
}
|
|
1970
|
+
else if (ON_STR_END(s)) {
|
|
1971
|
+
if (! ONIGENC_IS_MBC_WORD(encode, sprev, end))
|
|
1972
|
+
goto fail;
|
|
1973
|
+
}
|
|
1974
|
+
else {
|
|
1975
|
+
if (ONIGENC_IS_MBC_WORD(encode, s, end)
|
|
1976
|
+
== ONIGENC_IS_MBC_WORD(encode, sprev, end))
|
|
1977
|
+
goto fail;
|
|
1978
|
+
}
|
|
1979
|
+
MOP_OUT;
|
|
1980
|
+
continue;
|
|
1981
|
+
break;
|
|
1982
|
+
|
|
1983
|
+
case OP_NOT_WORD_BOUND: MOP_IN(OP_NOT_WORD_BOUND);
|
|
1984
|
+
if (ON_STR_BEGIN(s)) {
|
|
1985
|
+
if (DATA_ENSURE_CHECK1 && ONIGENC_IS_MBC_WORD(encode, s, end))
|
|
1986
|
+
goto fail;
|
|
1987
|
+
}
|
|
1988
|
+
else if (ON_STR_END(s)) {
|
|
1989
|
+
if (ONIGENC_IS_MBC_WORD(encode, sprev, end))
|
|
1990
|
+
goto fail;
|
|
1991
|
+
}
|
|
1992
|
+
else {
|
|
1993
|
+
if (ONIGENC_IS_MBC_WORD(encode, s, end)
|
|
1994
|
+
!= ONIGENC_IS_MBC_WORD(encode, sprev, end))
|
|
1995
|
+
goto fail;
|
|
1996
|
+
}
|
|
1997
|
+
MOP_OUT;
|
|
1998
|
+
continue;
|
|
1999
|
+
break;
|
|
2000
|
+
|
|
2001
|
+
#ifdef USE_WORD_BEGIN_END
|
|
2002
|
+
case OP_WORD_BEGIN: MOP_IN(OP_WORD_BEGIN);
|
|
2003
|
+
if (DATA_ENSURE_CHECK1 && ONIGENC_IS_MBC_WORD(encode, s, end)) {
|
|
2004
|
+
if (ON_STR_BEGIN(s) || !ONIGENC_IS_MBC_WORD(encode, sprev, end)) {
|
|
2005
|
+
MOP_OUT;
|
|
2006
|
+
continue;
|
|
2007
|
+
}
|
|
2008
|
+
}
|
|
2009
|
+
goto fail;
|
|
2010
|
+
break;
|
|
2011
|
+
|
|
2012
|
+
case OP_WORD_END: MOP_IN(OP_WORD_END);
|
|
2013
|
+
if (!ON_STR_BEGIN(s) && ONIGENC_IS_MBC_WORD(encode, sprev, end)) {
|
|
2014
|
+
if (ON_STR_END(s) || !ONIGENC_IS_MBC_WORD(encode, s, end)) {
|
|
2015
|
+
MOP_OUT;
|
|
2016
|
+
continue;
|
|
2017
|
+
}
|
|
2018
|
+
}
|
|
2019
|
+
goto fail;
|
|
2020
|
+
break;
|
|
2021
|
+
#endif
|
|
2022
|
+
|
|
2023
|
+
case OP_BEGIN_BUF: MOP_IN(OP_BEGIN_BUF);
|
|
2024
|
+
if (! ON_STR_BEGIN(s)) goto fail;
|
|
2025
|
+
|
|
2026
|
+
MOP_OUT;
|
|
2027
|
+
continue;
|
|
2028
|
+
break;
|
|
2029
|
+
|
|
2030
|
+
case OP_END_BUF: MOP_IN(OP_END_BUF);
|
|
2031
|
+
if (! ON_STR_END(s)) goto fail;
|
|
2032
|
+
|
|
2033
|
+
MOP_OUT;
|
|
2034
|
+
continue;
|
|
2035
|
+
break;
|
|
2036
|
+
|
|
2037
|
+
case OP_BEGIN_LINE: MOP_IN(OP_BEGIN_LINE);
|
|
2038
|
+
if (ON_STR_BEGIN(s)) {
|
|
2039
|
+
if (IS_NOTBOL(msa->options)) goto fail;
|
|
2040
|
+
MOP_OUT;
|
|
2041
|
+
continue;
|
|
2042
|
+
}
|
|
2043
|
+
else if (ONIGENC_IS_MBC_NEWLINE(encode, sprev, end) && !ON_STR_END(s)) {
|
|
2044
|
+
MOP_OUT;
|
|
2045
|
+
continue;
|
|
2046
|
+
}
|
|
2047
|
+
goto fail;
|
|
2048
|
+
break;
|
|
2049
|
+
|
|
2050
|
+
case OP_END_LINE: MOP_IN(OP_END_LINE);
|
|
2051
|
+
if (ON_STR_END(s)) {
|
|
2052
|
+
#ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE
|
|
2053
|
+
if (IS_EMPTY_STR || !ONIGENC_IS_MBC_NEWLINE(encode, sprev, end)) {
|
|
2054
|
+
#endif
|
|
2055
|
+
if (IS_NOTEOL(msa->options)) goto fail;
|
|
2056
|
+
MOP_OUT;
|
|
2057
|
+
continue;
|
|
2058
|
+
#ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE
|
|
2059
|
+
}
|
|
2060
|
+
#endif
|
|
2061
|
+
}
|
|
2062
|
+
else if (ONIGENC_IS_MBC_NEWLINE(encode, s, end)) {
|
|
2063
|
+
MOP_OUT;
|
|
2064
|
+
continue;
|
|
2065
|
+
}
|
|
2066
|
+
#ifdef USE_CRNL_AS_LINE_TERMINATOR
|
|
2067
|
+
else if (ONIGENC_IS_MBC_CRNL(encode, s, end)) {
|
|
2068
|
+
MOP_OUT;
|
|
2069
|
+
continue;
|
|
2070
|
+
}
|
|
2071
|
+
#endif
|
|
2072
|
+
goto fail;
|
|
2073
|
+
break;
|
|
2074
|
+
|
|
2075
|
+
case OP_SEMI_END_BUF: MOP_IN(OP_SEMI_END_BUF);
|
|
2076
|
+
if (ON_STR_END(s)) {
|
|
2077
|
+
#ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE
|
|
2078
|
+
if (IS_EMPTY_STR || !ONIGENC_IS_MBC_NEWLINE(encode, sprev, end)) {
|
|
2079
|
+
#endif
|
|
2080
|
+
if (IS_NOTEOL(msa->options)) goto fail;
|
|
2081
|
+
MOP_OUT;
|
|
2082
|
+
continue;
|
|
2083
|
+
#ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE
|
|
2084
|
+
}
|
|
2085
|
+
#endif
|
|
2086
|
+
}
|
|
2087
|
+
else if (ONIGENC_IS_MBC_NEWLINE(encode, s, end) &&
|
|
2088
|
+
ON_STR_END(s + enclen(encode, s))) {
|
|
2089
|
+
MOP_OUT;
|
|
2090
|
+
continue;
|
|
2091
|
+
}
|
|
2092
|
+
#ifdef USE_CRNL_AS_LINE_TERMINATOR
|
|
2093
|
+
else if (ONIGENC_IS_MBC_CRNL(encode, s, end)) {
|
|
2094
|
+
UChar* ss = s + enclen(encode, s);
|
|
2095
|
+
ss += enclen(encode, ss);
|
|
2096
|
+
if (ON_STR_END(ss)) {
|
|
2097
|
+
MOP_OUT;
|
|
2098
|
+
continue;
|
|
2099
|
+
}
|
|
2100
|
+
}
|
|
2101
|
+
#endif
|
|
2102
|
+
goto fail;
|
|
2103
|
+
break;
|
|
2104
|
+
|
|
2105
|
+
case OP_BEGIN_POSITION: MOP_IN(OP_BEGIN_POSITION);
|
|
2106
|
+
if (s != msa->start)
|
|
2107
|
+
goto fail;
|
|
2108
|
+
|
|
2109
|
+
MOP_OUT;
|
|
2110
|
+
continue;
|
|
2111
|
+
break;
|
|
2112
|
+
|
|
2113
|
+
case OP_MEMORY_START_PUSH: MOP_IN(OP_MEMORY_START_PUSH);
|
|
2114
|
+
GET_MEMNUM_INC(mem, p);
|
|
2115
|
+
STACK_PUSH_MEM_START(mem, s);
|
|
2116
|
+
MOP_OUT;
|
|
2117
|
+
continue;
|
|
2118
|
+
break;
|
|
2119
|
+
|
|
2120
|
+
case OP_MEMORY_START: MOP_IN(OP_MEMORY_START);
|
|
2121
|
+
GET_MEMNUM_INC(mem, p);
|
|
2122
|
+
mem_start_stk[mem] = (OnigStackIndex )((void* )s);
|
|
2123
|
+
MOP_OUT;
|
|
2124
|
+
continue;
|
|
2125
|
+
break;
|
|
2126
|
+
|
|
2127
|
+
case OP_MEMORY_END_PUSH: MOP_IN(OP_MEMORY_END_PUSH);
|
|
2128
|
+
GET_MEMNUM_INC(mem, p);
|
|
2129
|
+
STACK_PUSH_MEM_END(mem, s);
|
|
2130
|
+
MOP_OUT;
|
|
2131
|
+
continue;
|
|
2132
|
+
break;
|
|
2133
|
+
|
|
2134
|
+
case OP_MEMORY_END: MOP_IN(OP_MEMORY_END);
|
|
2135
|
+
GET_MEMNUM_INC(mem, p);
|
|
2136
|
+
mem_end_stk[mem] = (OnigStackIndex )((void* )s);
|
|
2137
|
+
MOP_OUT;
|
|
2138
|
+
continue;
|
|
2139
|
+
break;
|
|
2140
|
+
|
|
2141
|
+
#ifdef USE_SUBEXP_CALL
|
|
2142
|
+
case OP_MEMORY_END_PUSH_REC: MOP_IN(OP_MEMORY_END_PUSH_REC);
|
|
2143
|
+
GET_MEMNUM_INC(mem, p);
|
|
2144
|
+
STACK_GET_MEM_START(mem, stkp); /* should be before push mem-end. */
|
|
2145
|
+
STACK_PUSH_MEM_END(mem, s);
|
|
2146
|
+
mem_start_stk[mem] = GET_STACK_INDEX(stkp);
|
|
2147
|
+
MOP_OUT;
|
|
2148
|
+
continue;
|
|
2149
|
+
break;
|
|
2150
|
+
|
|
2151
|
+
case OP_MEMORY_END_REC: MOP_IN(OP_MEMORY_END_REC);
|
|
2152
|
+
GET_MEMNUM_INC(mem, p);
|
|
2153
|
+
mem_end_stk[mem] = (OnigStackIndex )((void* )s);
|
|
2154
|
+
STACK_GET_MEM_START(mem, stkp);
|
|
2155
|
+
|
|
2156
|
+
if (BIT_STATUS_AT(reg->bt_mem_start, mem))
|
|
2157
|
+
mem_start_stk[mem] = GET_STACK_INDEX(stkp);
|
|
2158
|
+
else
|
|
2159
|
+
mem_start_stk[mem] = (OnigStackIndex )((void* )stkp->u.mem.pstr);
|
|
2160
|
+
|
|
2161
|
+
STACK_PUSH_MEM_END_MARK(mem);
|
|
2162
|
+
MOP_OUT;
|
|
2163
|
+
continue;
|
|
2164
|
+
break;
|
|
2165
|
+
#endif
|
|
2166
|
+
|
|
2167
|
+
case OP_BACKREF1: MOP_IN(OP_BACKREF1);
|
|
2168
|
+
mem = 1;
|
|
2169
|
+
goto backref;
|
|
2170
|
+
break;
|
|
2171
|
+
|
|
2172
|
+
case OP_BACKREF2: MOP_IN(OP_BACKREF2);
|
|
2173
|
+
mem = 2;
|
|
2174
|
+
goto backref;
|
|
2175
|
+
break;
|
|
2176
|
+
|
|
2177
|
+
case OP_BACKREFN: MOP_IN(OP_BACKREFN);
|
|
2178
|
+
GET_MEMNUM_INC(mem, p);
|
|
2179
|
+
backref:
|
|
2180
|
+
{
|
|
2181
|
+
int len;
|
|
2182
|
+
UChar *pstart, *pend;
|
|
2183
|
+
|
|
2184
|
+
/* if you want to remove following line,
|
|
2185
|
+
you should check in parse and compile time. */
|
|
2186
|
+
if (mem > num_mem) goto fail;
|
|
2187
|
+
if (mem_end_stk[mem] == INVALID_STACK_INDEX) goto fail;
|
|
2188
|
+
if (mem_start_stk[mem] == INVALID_STACK_INDEX) goto fail;
|
|
2189
|
+
|
|
2190
|
+
if (BIT_STATUS_AT(reg->bt_mem_start, mem))
|
|
2191
|
+
pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr;
|
|
2192
|
+
else
|
|
2193
|
+
pstart = (UChar* )((void* )mem_start_stk[mem]);
|
|
2194
|
+
|
|
2195
|
+
pend = (BIT_STATUS_AT(reg->bt_mem_end, mem)
|
|
2196
|
+
? STACK_AT(mem_end_stk[mem])->u.mem.pstr
|
|
2197
|
+
: (UChar* )((void* )mem_end_stk[mem]));
|
|
2198
|
+
n = pend - pstart;
|
|
2199
|
+
DATA_ENSURE(n);
|
|
2200
|
+
sprev = s;
|
|
2201
|
+
STRING_CMP(pstart, s, n);
|
|
2202
|
+
while (sprev + (len = enclen(encode, sprev)) < s)
|
|
2203
|
+
sprev += len;
|
|
2204
|
+
|
|
2205
|
+
MOP_OUT;
|
|
2206
|
+
continue;
|
|
2207
|
+
}
|
|
2208
|
+
break;
|
|
2209
|
+
|
|
2210
|
+
case OP_BACKREFN_IC: MOP_IN(OP_BACKREFN_IC);
|
|
2211
|
+
GET_MEMNUM_INC(mem, p);
|
|
2212
|
+
{
|
|
2213
|
+
int len;
|
|
2214
|
+
UChar *pstart, *pend;
|
|
2215
|
+
|
|
2216
|
+
/* if you want to remove following line,
|
|
2217
|
+
you should check in parse and compile time. */
|
|
2218
|
+
if (mem > num_mem) goto fail;
|
|
2219
|
+
if (mem_end_stk[mem] == INVALID_STACK_INDEX) goto fail;
|
|
2220
|
+
if (mem_start_stk[mem] == INVALID_STACK_INDEX) goto fail;
|
|
2221
|
+
|
|
2222
|
+
if (BIT_STATUS_AT(reg->bt_mem_start, mem))
|
|
2223
|
+
pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr;
|
|
2224
|
+
else
|
|
2225
|
+
pstart = (UChar* )((void* )mem_start_stk[mem]);
|
|
2226
|
+
|
|
2227
|
+
pend = (BIT_STATUS_AT(reg->bt_mem_end, mem)
|
|
2228
|
+
? STACK_AT(mem_end_stk[mem])->u.mem.pstr
|
|
2229
|
+
: (UChar* )((void* )mem_end_stk[mem]));
|
|
2230
|
+
n = pend - pstart;
|
|
2231
|
+
DATA_ENSURE(n);
|
|
2232
|
+
sprev = s;
|
|
2233
|
+
STRING_CMP_IC(case_fold_flag, pstart, &s, n);
|
|
2234
|
+
while (sprev + (len = enclen(encode, sprev)) < s)
|
|
2235
|
+
sprev += len;
|
|
2236
|
+
|
|
2237
|
+
MOP_OUT;
|
|
2238
|
+
continue;
|
|
2239
|
+
}
|
|
2240
|
+
break;
|
|
2241
|
+
|
|
2242
|
+
case OP_BACKREF_MULTI: MOP_IN(OP_BACKREF_MULTI);
|
|
2243
|
+
{
|
|
2244
|
+
int len, is_fail;
|
|
2245
|
+
UChar *pstart, *pend, *swork;
|
|
2246
|
+
|
|
2247
|
+
GET_LENGTH_INC(tlen, p);
|
|
2248
|
+
for (i = 0; i < tlen; i++) {
|
|
2249
|
+
GET_MEMNUM_INC(mem, p);
|
|
2250
|
+
|
|
2251
|
+
if (mem_end_stk[mem] == INVALID_STACK_INDEX) continue;
|
|
2252
|
+
if (mem_start_stk[mem] == INVALID_STACK_INDEX) continue;
|
|
2253
|
+
|
|
2254
|
+
if (BIT_STATUS_AT(reg->bt_mem_start, mem))
|
|
2255
|
+
pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr;
|
|
2256
|
+
else
|
|
2257
|
+
pstart = (UChar* )((void* )mem_start_stk[mem]);
|
|
2258
|
+
|
|
2259
|
+
pend = (BIT_STATUS_AT(reg->bt_mem_end, mem)
|
|
2260
|
+
? STACK_AT(mem_end_stk[mem])->u.mem.pstr
|
|
2261
|
+
: (UChar* )((void* )mem_end_stk[mem]));
|
|
2262
|
+
n = pend - pstart;
|
|
2263
|
+
DATA_ENSURE(n);
|
|
2264
|
+
sprev = s;
|
|
2265
|
+
swork = s;
|
|
2266
|
+
STRING_CMP_VALUE(pstart, swork, n, is_fail);
|
|
2267
|
+
if (is_fail) continue;
|
|
2268
|
+
s = swork;
|
|
2269
|
+
while (sprev + (len = enclen(encode, sprev)) < s)
|
|
2270
|
+
sprev += len;
|
|
2271
|
+
|
|
2272
|
+
p += (SIZE_MEMNUM * (tlen - i - 1));
|
|
2273
|
+
break; /* success */
|
|
2274
|
+
}
|
|
2275
|
+
if (i == tlen) goto fail;
|
|
2276
|
+
MOP_OUT;
|
|
2277
|
+
continue;
|
|
2278
|
+
}
|
|
2279
|
+
break;
|
|
2280
|
+
|
|
2281
|
+
case OP_BACKREF_MULTI_IC: MOP_IN(OP_BACKREF_MULTI_IC);
|
|
2282
|
+
{
|
|
2283
|
+
int len, is_fail;
|
|
2284
|
+
UChar *pstart, *pend, *swork;
|
|
2285
|
+
|
|
2286
|
+
GET_LENGTH_INC(tlen, p);
|
|
2287
|
+
for (i = 0; i < tlen; i++) {
|
|
2288
|
+
GET_MEMNUM_INC(mem, p);
|
|
2289
|
+
|
|
2290
|
+
if (mem_end_stk[mem] == INVALID_STACK_INDEX) continue;
|
|
2291
|
+
if (mem_start_stk[mem] == INVALID_STACK_INDEX) continue;
|
|
2292
|
+
|
|
2293
|
+
if (BIT_STATUS_AT(reg->bt_mem_start, mem))
|
|
2294
|
+
pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr;
|
|
2295
|
+
else
|
|
2296
|
+
pstart = (UChar* )((void* )mem_start_stk[mem]);
|
|
2297
|
+
|
|
2298
|
+
pend = (BIT_STATUS_AT(reg->bt_mem_end, mem)
|
|
2299
|
+
? STACK_AT(mem_end_stk[mem])->u.mem.pstr
|
|
2300
|
+
: (UChar* )((void* )mem_end_stk[mem]));
|
|
2301
|
+
n = pend - pstart;
|
|
2302
|
+
DATA_ENSURE(n);
|
|
2303
|
+
sprev = s;
|
|
2304
|
+
swork = s;
|
|
2305
|
+
STRING_CMP_VALUE_IC(case_fold_flag, pstart, &swork, n, is_fail);
|
|
2306
|
+
if (is_fail) continue;
|
|
2307
|
+
s = swork;
|
|
2308
|
+
while (sprev + (len = enclen(encode, sprev)) < s)
|
|
2309
|
+
sprev += len;
|
|
2310
|
+
|
|
2311
|
+
p += (SIZE_MEMNUM * (tlen - i - 1));
|
|
2312
|
+
break; /* success */
|
|
2313
|
+
}
|
|
2314
|
+
if (i == tlen) goto fail;
|
|
2315
|
+
MOP_OUT;
|
|
2316
|
+
continue;
|
|
2317
|
+
}
|
|
2318
|
+
break;
|
|
2319
|
+
|
|
2320
|
+
#ifdef USE_BACKREF_WITH_LEVEL
|
|
2321
|
+
case OP_BACKREF_WITH_LEVEL:
|
|
2322
|
+
{
|
|
2323
|
+
int len;
|
|
2324
|
+
OnigOptionType ic;
|
|
2325
|
+
LengthType level;
|
|
2326
|
+
|
|
2327
|
+
GET_OPTION_INC(ic, p);
|
|
2328
|
+
GET_LENGTH_INC(level, p);
|
|
2329
|
+
GET_LENGTH_INC(tlen, p);
|
|
2330
|
+
|
|
2331
|
+
sprev = s;
|
|
2332
|
+
if (backref_match_at_nested_level(reg, stk, stk_base, ic
|
|
2333
|
+
, case_fold_flag, (int )level, (int )tlen, p, &s, end)) {
|
|
2334
|
+
while (sprev + (len = enclen(encode, sprev)) < s)
|
|
2335
|
+
sprev += len;
|
|
2336
|
+
|
|
2337
|
+
p += (SIZE_MEMNUM * tlen);
|
|
2338
|
+
}
|
|
2339
|
+
else
|
|
2340
|
+
goto fail;
|
|
2341
|
+
|
|
2342
|
+
MOP_OUT;
|
|
2343
|
+
continue;
|
|
2344
|
+
}
|
|
2345
|
+
break;
|
|
2346
|
+
#endif
|
|
2347
|
+
|
|
2348
|
+
#if 0 /* no need: IS_DYNAMIC_OPTION() == 0 */
|
|
2349
|
+
case OP_SET_OPTION_PUSH: MOP_IN(OP_SET_OPTION_PUSH);
|
|
2350
|
+
GET_OPTION_INC(option, p);
|
|
2351
|
+
STACK_PUSH_ALT(p, s, sprev);
|
|
2352
|
+
p += SIZE_OP_SET_OPTION + SIZE_OP_FAIL;
|
|
2353
|
+
MOP_OUT;
|
|
2354
|
+
continue;
|
|
2355
|
+
break;
|
|
2356
|
+
|
|
2357
|
+
case OP_SET_OPTION: MOP_IN(OP_SET_OPTION);
|
|
2358
|
+
GET_OPTION_INC(option, p);
|
|
2359
|
+
MOP_OUT;
|
|
2360
|
+
continue;
|
|
2361
|
+
break;
|
|
2362
|
+
#endif
|
|
2363
|
+
|
|
2364
|
+
case OP_NULL_CHECK_START: MOP_IN(OP_NULL_CHECK_START);
|
|
2365
|
+
GET_MEMNUM_INC(mem, p); /* mem: null check id */
|
|
2366
|
+
STACK_PUSH_NULL_CHECK_START(mem, s);
|
|
2367
|
+
MOP_OUT;
|
|
2368
|
+
continue;
|
|
2369
|
+
break;
|
|
2370
|
+
|
|
2371
|
+
case OP_NULL_CHECK_END: MOP_IN(OP_NULL_CHECK_END);
|
|
2372
|
+
{
|
|
2373
|
+
int isnull;
|
|
2374
|
+
|
|
2375
|
+
GET_MEMNUM_INC(mem, p); /* mem: null check id */
|
|
2376
|
+
STACK_NULL_CHECK(isnull, mem, s);
|
|
2377
|
+
if (isnull) {
|
|
2378
|
+
#ifdef ONIG_DEBUG_MATCH
|
|
2379
|
+
fprintf(stderr, "NULL_CHECK_END: skip id:%d, s:%d\n",
|
|
2380
|
+
(int )mem, (int )s);
|
|
2381
|
+
#endif
|
|
2382
|
+
null_check_found:
|
|
2383
|
+
/* empty loop founded, skip next instruction */
|
|
2384
|
+
switch (*p++) {
|
|
2385
|
+
case OP_JUMP:
|
|
2386
|
+
case OP_PUSH:
|
|
2387
|
+
p += SIZE_RELADDR;
|
|
2388
|
+
break;
|
|
2389
|
+
case OP_REPEAT_INC:
|
|
2390
|
+
case OP_REPEAT_INC_NG:
|
|
2391
|
+
case OP_REPEAT_INC_SG:
|
|
2392
|
+
case OP_REPEAT_INC_NG_SG:
|
|
2393
|
+
p += SIZE_MEMNUM;
|
|
2394
|
+
break;
|
|
2395
|
+
default:
|
|
2396
|
+
goto unexpected_bytecode_error;
|
|
2397
|
+
break;
|
|
2398
|
+
}
|
|
2399
|
+
}
|
|
2400
|
+
}
|
|
2401
|
+
MOP_OUT;
|
|
2402
|
+
continue;
|
|
2403
|
+
break;
|
|
2404
|
+
|
|
2405
|
+
#ifdef USE_MONOMANIAC_CHECK_CAPTURES_IN_ENDLESS_REPEAT
|
|
2406
|
+
case OP_NULL_CHECK_END_MEMST: MOP_IN(OP_NULL_CHECK_END_MEMST);
|
|
2407
|
+
{
|
|
2408
|
+
int isnull;
|
|
2409
|
+
|
|
2410
|
+
GET_MEMNUM_INC(mem, p); /* mem: null check id */
|
|
2411
|
+
STACK_NULL_CHECK_MEMST(isnull, mem, s, reg);
|
|
2412
|
+
if (isnull) {
|
|
2413
|
+
#ifdef ONIG_DEBUG_MATCH
|
|
2414
|
+
fprintf(stderr, "NULL_CHECK_END_MEMST: skip id:%d, s:%d\n",
|
|
2415
|
+
(int )mem, (int )s);
|
|
2416
|
+
#endif
|
|
2417
|
+
if (isnull == -1) goto fail;
|
|
2418
|
+
goto null_check_found;
|
|
2419
|
+
}
|
|
2420
|
+
}
|
|
2421
|
+
MOP_OUT;
|
|
2422
|
+
continue;
|
|
2423
|
+
break;
|
|
2424
|
+
#endif
|
|
2425
|
+
|
|
2426
|
+
#ifdef USE_SUBEXP_CALL
|
|
2427
|
+
case OP_NULL_CHECK_END_MEMST_PUSH:
|
|
2428
|
+
MOP_IN(OP_NULL_CHECK_END_MEMST_PUSH);
|
|
2429
|
+
{
|
|
2430
|
+
int isnull;
|
|
2431
|
+
|
|
2432
|
+
GET_MEMNUM_INC(mem, p); /* mem: null check id */
|
|
2433
|
+
#ifdef USE_MONOMANIAC_CHECK_CAPTURES_IN_ENDLESS_REPEAT
|
|
2434
|
+
STACK_NULL_CHECK_MEMST_REC(isnull, mem, s, reg);
|
|
2435
|
+
#else
|
|
2436
|
+
STACK_NULL_CHECK_REC(isnull, mem, s);
|
|
2437
|
+
#endif
|
|
2438
|
+
if (isnull) {
|
|
2439
|
+
#ifdef ONIG_DEBUG_MATCH
|
|
2440
|
+
fprintf(stderr, "NULL_CHECK_END_MEMST_PUSH: skip id:%d, s:%d\n",
|
|
2441
|
+
(int )mem, (int )s);
|
|
2442
|
+
#endif
|
|
2443
|
+
if (isnull == -1) goto fail;
|
|
2444
|
+
goto null_check_found;
|
|
2445
|
+
}
|
|
2446
|
+
else {
|
|
2447
|
+
STACK_PUSH_NULL_CHECK_END(mem);
|
|
2448
|
+
}
|
|
2449
|
+
}
|
|
2450
|
+
MOP_OUT;
|
|
2451
|
+
continue;
|
|
2452
|
+
break;
|
|
2453
|
+
#endif
|
|
2454
|
+
|
|
2455
|
+
case OP_JUMP: MOP_IN(OP_JUMP);
|
|
2456
|
+
GET_RELADDR_INC(addr, p);
|
|
2457
|
+
p += addr;
|
|
2458
|
+
MOP_OUT;
|
|
2459
|
+
CHECK_INTERRUPT_IN_MATCH_AT;
|
|
2460
|
+
continue;
|
|
2461
|
+
break;
|
|
2462
|
+
|
|
2463
|
+
case OP_PUSH: MOP_IN(OP_PUSH);
|
|
2464
|
+
GET_RELADDR_INC(addr, p);
|
|
2465
|
+
STACK_PUSH_ALT(p + addr, s, sprev);
|
|
2466
|
+
MOP_OUT;
|
|
2467
|
+
continue;
|
|
2468
|
+
break;
|
|
2469
|
+
|
|
2470
|
+
#ifdef USE_COMBINATION_EXPLOSION_CHECK
|
|
2471
|
+
case OP_STATE_CHECK_PUSH: MOP_IN(OP_STATE_CHECK_PUSH);
|
|
2472
|
+
GET_STATE_CHECK_NUM_INC(mem, p);
|
|
2473
|
+
STATE_CHECK_VAL(scv, mem);
|
|
2474
|
+
if (scv) goto fail;
|
|
2475
|
+
|
|
2476
|
+
GET_RELADDR_INC(addr, p);
|
|
2477
|
+
STACK_PUSH_ALT_WITH_STATE_CHECK(p + addr, s, sprev, mem);
|
|
2478
|
+
MOP_OUT;
|
|
2479
|
+
continue;
|
|
2480
|
+
break;
|
|
2481
|
+
|
|
2482
|
+
case OP_STATE_CHECK_PUSH_OR_JUMP: MOP_IN(OP_STATE_CHECK_PUSH_OR_JUMP);
|
|
2483
|
+
GET_STATE_CHECK_NUM_INC(mem, p);
|
|
2484
|
+
GET_RELADDR_INC(addr, p);
|
|
2485
|
+
STATE_CHECK_VAL(scv, mem);
|
|
2486
|
+
if (scv) {
|
|
2487
|
+
p += addr;
|
|
2488
|
+
}
|
|
2489
|
+
else {
|
|
2490
|
+
STACK_PUSH_ALT_WITH_STATE_CHECK(p + addr, s, sprev, mem);
|
|
2491
|
+
}
|
|
2492
|
+
MOP_OUT;
|
|
2493
|
+
continue;
|
|
2494
|
+
break;
|
|
2495
|
+
|
|
2496
|
+
case OP_STATE_CHECK: MOP_IN(OP_STATE_CHECK);
|
|
2497
|
+
GET_STATE_CHECK_NUM_INC(mem, p);
|
|
2498
|
+
STATE_CHECK_VAL(scv, mem);
|
|
2499
|
+
if (scv) goto fail;
|
|
2500
|
+
|
|
2501
|
+
STACK_PUSH_STATE_CHECK(s, mem);
|
|
2502
|
+
MOP_OUT;
|
|
2503
|
+
continue;
|
|
2504
|
+
break;
|
|
2505
|
+
#endif /* USE_COMBINATION_EXPLOSION_CHECK */
|
|
2506
|
+
|
|
2507
|
+
case OP_POP: MOP_IN(OP_POP);
|
|
2508
|
+
STACK_POP_ONE;
|
|
2509
|
+
MOP_OUT;
|
|
2510
|
+
continue;
|
|
2511
|
+
break;
|
|
2512
|
+
|
|
2513
|
+
case OP_PUSH_OR_JUMP_EXACT1: MOP_IN(OP_PUSH_OR_JUMP_EXACT1);
|
|
2514
|
+
GET_RELADDR_INC(addr, p);
|
|
2515
|
+
if (*p == *s && DATA_ENSURE_CHECK1) {
|
|
2516
|
+
p++;
|
|
2517
|
+
STACK_PUSH_ALT(p + addr, s, sprev);
|
|
2518
|
+
MOP_OUT;
|
|
2519
|
+
continue;
|
|
2520
|
+
}
|
|
2521
|
+
p += (addr + 1);
|
|
2522
|
+
MOP_OUT;
|
|
2523
|
+
continue;
|
|
2524
|
+
break;
|
|
2525
|
+
|
|
2526
|
+
case OP_PUSH_IF_PEEK_NEXT: MOP_IN(OP_PUSH_IF_PEEK_NEXT);
|
|
2527
|
+
GET_RELADDR_INC(addr, p);
|
|
2528
|
+
if (*p == *s) {
|
|
2529
|
+
p++;
|
|
2530
|
+
STACK_PUSH_ALT(p + addr, s, sprev);
|
|
2531
|
+
MOP_OUT;
|
|
2532
|
+
continue;
|
|
2533
|
+
}
|
|
2534
|
+
p++;
|
|
2535
|
+
MOP_OUT;
|
|
2536
|
+
continue;
|
|
2537
|
+
break;
|
|
2538
|
+
|
|
2539
|
+
case OP_REPEAT: MOP_IN(OP_REPEAT);
|
|
2540
|
+
{
|
|
2541
|
+
GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */
|
|
2542
|
+
GET_RELADDR_INC(addr, p);
|
|
2543
|
+
|
|
2544
|
+
STACK_ENSURE(1);
|
|
2545
|
+
repeat_stk[mem] = GET_STACK_INDEX(stk);
|
|
2546
|
+
STACK_PUSH_REPEAT(mem, p);
|
|
2547
|
+
|
|
2548
|
+
if (reg->repeat_range[mem].lower == 0) {
|
|
2549
|
+
STACK_PUSH_ALT(p + addr, s, sprev);
|
|
2550
|
+
}
|
|
2551
|
+
}
|
|
2552
|
+
MOP_OUT;
|
|
2553
|
+
continue;
|
|
2554
|
+
break;
|
|
2555
|
+
|
|
2556
|
+
case OP_REPEAT_NG: MOP_IN(OP_REPEAT_NG);
|
|
2557
|
+
{
|
|
2558
|
+
GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */
|
|
2559
|
+
GET_RELADDR_INC(addr, p);
|
|
2560
|
+
|
|
2561
|
+
STACK_ENSURE(1);
|
|
2562
|
+
repeat_stk[mem] = GET_STACK_INDEX(stk);
|
|
2563
|
+
STACK_PUSH_REPEAT(mem, p);
|
|
2564
|
+
|
|
2565
|
+
if (reg->repeat_range[mem].lower == 0) {
|
|
2566
|
+
STACK_PUSH_ALT(p, s, sprev);
|
|
2567
|
+
p += addr;
|
|
2568
|
+
}
|
|
2569
|
+
}
|
|
2570
|
+
MOP_OUT;
|
|
2571
|
+
continue;
|
|
2572
|
+
break;
|
|
2573
|
+
|
|
2574
|
+
case OP_REPEAT_INC: MOP_IN(OP_REPEAT_INC);
|
|
2575
|
+
GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */
|
|
2576
|
+
si = repeat_stk[mem];
|
|
2577
|
+
stkp = STACK_AT(si);
|
|
2578
|
+
|
|
2579
|
+
repeat_inc:
|
|
2580
|
+
stkp->u.repeat.count++;
|
|
2581
|
+
if (stkp->u.repeat.count >= reg->repeat_range[mem].upper) {
|
|
2582
|
+
/* end of repeat. Nothing to do. */
|
|
2583
|
+
}
|
|
2584
|
+
else if (stkp->u.repeat.count >= reg->repeat_range[mem].lower) {
|
|
2585
|
+
STACK_PUSH_ALT(p, s, sprev);
|
|
2586
|
+
p = STACK_AT(si)->u.repeat.pcode; /* Don't use stkp after PUSH. */
|
|
2587
|
+
}
|
|
2588
|
+
else {
|
|
2589
|
+
p = stkp->u.repeat.pcode;
|
|
2590
|
+
}
|
|
2591
|
+
STACK_PUSH_REPEAT_INC(si);
|
|
2592
|
+
MOP_OUT;
|
|
2593
|
+
CHECK_INTERRUPT_IN_MATCH_AT;
|
|
2594
|
+
continue;
|
|
2595
|
+
break;
|
|
2596
|
+
|
|
2597
|
+
case OP_REPEAT_INC_SG: MOP_IN(OP_REPEAT_INC_SG);
|
|
2598
|
+
GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */
|
|
2599
|
+
STACK_GET_REPEAT(mem, stkp);
|
|
2600
|
+
si = GET_STACK_INDEX(stkp);
|
|
2601
|
+
goto repeat_inc;
|
|
2602
|
+
break;
|
|
2603
|
+
|
|
2604
|
+
case OP_REPEAT_INC_NG: MOP_IN(OP_REPEAT_INC_NG);
|
|
2605
|
+
GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */
|
|
2606
|
+
si = repeat_stk[mem];
|
|
2607
|
+
stkp = STACK_AT(si);
|
|
2608
|
+
|
|
2609
|
+
repeat_inc_ng:
|
|
2610
|
+
stkp->u.repeat.count++;
|
|
2611
|
+
if (stkp->u.repeat.count < reg->repeat_range[mem].upper) {
|
|
2612
|
+
if (stkp->u.repeat.count >= reg->repeat_range[mem].lower) {
|
|
2613
|
+
UChar* pcode = stkp->u.repeat.pcode;
|
|
2614
|
+
|
|
2615
|
+
STACK_PUSH_REPEAT_INC(si);
|
|
2616
|
+
STACK_PUSH_ALT(pcode, s, sprev);
|
|
2617
|
+
}
|
|
2618
|
+
else {
|
|
2619
|
+
p = stkp->u.repeat.pcode;
|
|
2620
|
+
STACK_PUSH_REPEAT_INC(si);
|
|
2621
|
+
}
|
|
2622
|
+
}
|
|
2623
|
+
else if (stkp->u.repeat.count == reg->repeat_range[mem].upper) {
|
|
2624
|
+
STACK_PUSH_REPEAT_INC(si);
|
|
2625
|
+
}
|
|
2626
|
+
MOP_OUT;
|
|
2627
|
+
CHECK_INTERRUPT_IN_MATCH_AT;
|
|
2628
|
+
continue;
|
|
2629
|
+
break;
|
|
2630
|
+
|
|
2631
|
+
case OP_REPEAT_INC_NG_SG: MOP_IN(OP_REPEAT_INC_NG_SG);
|
|
2632
|
+
GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */
|
|
2633
|
+
STACK_GET_REPEAT(mem, stkp);
|
|
2634
|
+
si = GET_STACK_INDEX(stkp);
|
|
2635
|
+
goto repeat_inc_ng;
|
|
2636
|
+
break;
|
|
2637
|
+
|
|
2638
|
+
case OP_PUSH_POS: MOP_IN(OP_PUSH_POS);
|
|
2639
|
+
STACK_PUSH_POS(s, sprev);
|
|
2640
|
+
MOP_OUT;
|
|
2641
|
+
continue;
|
|
2642
|
+
break;
|
|
2643
|
+
|
|
2644
|
+
case OP_POP_POS: MOP_IN(OP_POP_POS);
|
|
2645
|
+
{
|
|
2646
|
+
STACK_POS_END(stkp);
|
|
2647
|
+
s = stkp->u.state.pstr;
|
|
2648
|
+
sprev = stkp->u.state.pstr_prev;
|
|
2649
|
+
}
|
|
2650
|
+
MOP_OUT;
|
|
2651
|
+
continue;
|
|
2652
|
+
break;
|
|
2653
|
+
|
|
2654
|
+
case OP_PUSH_POS_NOT: MOP_IN(OP_PUSH_POS_NOT);
|
|
2655
|
+
GET_RELADDR_INC(addr, p);
|
|
2656
|
+
STACK_PUSH_POS_NOT(p + addr, s, sprev);
|
|
2657
|
+
MOP_OUT;
|
|
2658
|
+
continue;
|
|
2659
|
+
break;
|
|
2660
|
+
|
|
2661
|
+
case OP_FAIL_POS: MOP_IN(OP_FAIL_POS);
|
|
2662
|
+
STACK_POP_TIL_POS_NOT;
|
|
2663
|
+
goto fail;
|
|
2664
|
+
break;
|
|
2665
|
+
|
|
2666
|
+
case OP_PUSH_STOP_BT: MOP_IN(OP_PUSH_STOP_BT);
|
|
2667
|
+
STACK_PUSH_STOP_BT;
|
|
2668
|
+
MOP_OUT;
|
|
2669
|
+
continue;
|
|
2670
|
+
break;
|
|
2671
|
+
|
|
2672
|
+
case OP_POP_STOP_BT: MOP_IN(OP_POP_STOP_BT);
|
|
2673
|
+
STACK_STOP_BT_END;
|
|
2674
|
+
MOP_OUT;
|
|
2675
|
+
continue;
|
|
2676
|
+
break;
|
|
2677
|
+
|
|
2678
|
+
case OP_LOOK_BEHIND: MOP_IN(OP_LOOK_BEHIND);
|
|
2679
|
+
GET_LENGTH_INC(tlen, p);
|
|
2680
|
+
s = (UChar* )ONIGENC_STEP_BACK(encode, str, s, (int )tlen);
|
|
2681
|
+
if (IS_NULL(s)) goto fail;
|
|
2682
|
+
sprev = (UChar* )onigenc_get_prev_char_head(encode, str, s);
|
|
2683
|
+
MOP_OUT;
|
|
2684
|
+
continue;
|
|
2685
|
+
break;
|
|
2686
|
+
|
|
2687
|
+
case OP_PUSH_LOOK_BEHIND_NOT: MOP_IN(OP_PUSH_LOOK_BEHIND_NOT);
|
|
2688
|
+
GET_RELADDR_INC(addr, p);
|
|
2689
|
+
GET_LENGTH_INC(tlen, p);
|
|
2690
|
+
q = (UChar* )ONIGENC_STEP_BACK(encode, str, s, (int )tlen);
|
|
2691
|
+
if (IS_NULL(q)) {
|
|
2692
|
+
/* too short case -> success. ex. /(?<!XXX)a/.match("a")
|
|
2693
|
+
If you want to change to fail, replace following line. */
|
|
2694
|
+
p += addr;
|
|
2695
|
+
/* goto fail; */
|
|
2696
|
+
}
|
|
2697
|
+
else {
|
|
2698
|
+
STACK_PUSH_LOOK_BEHIND_NOT(p + addr, s, sprev);
|
|
2699
|
+
s = q;
|
|
2700
|
+
sprev = (UChar* )onigenc_get_prev_char_head(encode, str, s);
|
|
2701
|
+
}
|
|
2702
|
+
MOP_OUT;
|
|
2703
|
+
continue;
|
|
2704
|
+
break;
|
|
2705
|
+
|
|
2706
|
+
case OP_FAIL_LOOK_BEHIND_NOT: MOP_IN(OP_FAIL_LOOK_BEHIND_NOT);
|
|
2707
|
+
STACK_POP_TIL_LOOK_BEHIND_NOT;
|
|
2708
|
+
goto fail;
|
|
2709
|
+
break;
|
|
2710
|
+
|
|
2711
|
+
#ifdef USE_SUBEXP_CALL
|
|
2712
|
+
case OP_CALL: MOP_IN(OP_CALL);
|
|
2713
|
+
GET_ABSADDR_INC(addr, p);
|
|
2714
|
+
STACK_PUSH_CALL_FRAME(p);
|
|
2715
|
+
p = reg->p + addr;
|
|
2716
|
+
MOP_OUT;
|
|
2717
|
+
continue;
|
|
2718
|
+
break;
|
|
2719
|
+
|
|
2720
|
+
case OP_RETURN: MOP_IN(OP_RETURN);
|
|
2721
|
+
STACK_RETURN(p);
|
|
2722
|
+
STACK_PUSH_RETURN;
|
|
2723
|
+
MOP_OUT;
|
|
2724
|
+
continue;
|
|
2725
|
+
break;
|
|
2726
|
+
#endif
|
|
2727
|
+
|
|
2728
|
+
case OP_FINISH:
|
|
2729
|
+
goto finish;
|
|
2730
|
+
break;
|
|
2731
|
+
|
|
2732
|
+
fail:
|
|
2733
|
+
MOP_OUT;
|
|
2734
|
+
/* fall */
|
|
2735
|
+
case OP_FAIL: MOP_IN(OP_FAIL);
|
|
2736
|
+
STACK_POP;
|
|
2737
|
+
p = stk->u.state.pcode;
|
|
2738
|
+
s = stk->u.state.pstr;
|
|
2739
|
+
sprev = stk->u.state.pstr_prev;
|
|
2740
|
+
|
|
2741
|
+
#ifdef USE_COMBINATION_EXPLOSION_CHECK
|
|
2742
|
+
if (stk->u.state.state_check != 0) {
|
|
2743
|
+
stk->type = STK_STATE_CHECK_MARK;
|
|
2744
|
+
stk++;
|
|
2745
|
+
}
|
|
2746
|
+
#endif
|
|
2747
|
+
|
|
2748
|
+
MOP_OUT;
|
|
2749
|
+
continue;
|
|
2750
|
+
break;
|
|
2751
|
+
|
|
2752
|
+
default:
|
|
2753
|
+
goto bytecode_error;
|
|
2754
|
+
|
|
2755
|
+
} /* end of switch */
|
|
2756
|
+
sprev = sbegin;
|
|
2757
|
+
} /* end of while(1) */
|
|
2758
|
+
|
|
2759
|
+
finish:
|
|
2760
|
+
STACK_SAVE;
|
|
2761
|
+
return best_len;
|
|
2762
|
+
|
|
2763
|
+
#ifdef ONIG_DEBUG
|
|
2764
|
+
stack_error:
|
|
2765
|
+
STACK_SAVE;
|
|
2766
|
+
return ONIGERR_STACK_BUG;
|
|
2767
|
+
#endif
|
|
2768
|
+
|
|
2769
|
+
bytecode_error:
|
|
2770
|
+
STACK_SAVE;
|
|
2771
|
+
return ONIGERR_UNDEFINED_BYTECODE;
|
|
2772
|
+
|
|
2773
|
+
unexpected_bytecode_error:
|
|
2774
|
+
STACK_SAVE;
|
|
2775
|
+
return ONIGERR_UNEXPECTED_BYTECODE;
|
|
2776
|
+
}
|
|
2777
|
+
|
|
2778
|
+
|
|
2779
|
+
static UChar*
|
|
2780
|
+
slow_search(OnigEncoding enc, UChar* target, UChar* target_end,
|
|
2781
|
+
const UChar* text, const UChar* text_end, UChar* text_range)
|
|
2782
|
+
{
|
|
2783
|
+
UChar *t, *p, *s, *end;
|
|
2784
|
+
|
|
2785
|
+
end = (UChar* )text_end;
|
|
2786
|
+
end -= target_end - target - 1;
|
|
2787
|
+
if (end > text_range)
|
|
2788
|
+
end = text_range;
|
|
2789
|
+
|
|
2790
|
+
s = (UChar* )text;
|
|
2791
|
+
|
|
2792
|
+
while (s < end) {
|
|
2793
|
+
if (*s == *target) {
|
|
2794
|
+
p = s + 1;
|
|
2795
|
+
t = target + 1;
|
|
2796
|
+
while (t < target_end) {
|
|
2797
|
+
if (*t != *p++)
|
|
2798
|
+
break;
|
|
2799
|
+
t++;
|
|
2800
|
+
}
|
|
2801
|
+
if (t == target_end)
|
|
2802
|
+
return s;
|
|
2803
|
+
}
|
|
2804
|
+
s += enclen(enc, s);
|
|
2805
|
+
}
|
|
2806
|
+
|
|
2807
|
+
return (UChar* )NULL;
|
|
2808
|
+
}
|
|
2809
|
+
|
|
2810
|
+
static int
|
|
2811
|
+
str_lower_case_match(OnigEncoding enc, int case_fold_flag,
|
|
2812
|
+
const UChar* t, const UChar* tend,
|
|
2813
|
+
const UChar* p, const UChar* end)
|
|
2814
|
+
{
|
|
2815
|
+
int lowlen;
|
|
2816
|
+
UChar *q, lowbuf[ONIGENC_MBC_CASE_FOLD_MAXLEN];
|
|
2817
|
+
|
|
2818
|
+
while (t < tend) {
|
|
2819
|
+
lowlen = ONIGENC_MBC_CASE_FOLD(enc, case_fold_flag, &p, end, lowbuf);
|
|
2820
|
+
q = lowbuf;
|
|
2821
|
+
while (lowlen > 0) {
|
|
2822
|
+
if (*t++ != *q++) return 0;
|
|
2823
|
+
lowlen--;
|
|
2824
|
+
}
|
|
2825
|
+
}
|
|
2826
|
+
|
|
2827
|
+
return 1;
|
|
2828
|
+
}
|
|
2829
|
+
|
|
2830
|
+
static UChar*
|
|
2831
|
+
slow_search_ic(OnigEncoding enc, int case_fold_flag,
|
|
2832
|
+
UChar* target, UChar* target_end,
|
|
2833
|
+
const UChar* text, const UChar* text_end, UChar* text_range)
|
|
2834
|
+
{
|
|
2835
|
+
UChar *s, *end;
|
|
2836
|
+
|
|
2837
|
+
end = (UChar* )text_end;
|
|
2838
|
+
end -= target_end - target - 1;
|
|
2839
|
+
if (end > text_range)
|
|
2840
|
+
end = text_range;
|
|
2841
|
+
|
|
2842
|
+
s = (UChar* )text;
|
|
2843
|
+
|
|
2844
|
+
while (s < end) {
|
|
2845
|
+
if (str_lower_case_match(enc, case_fold_flag, target, target_end,
|
|
2846
|
+
s, text_end))
|
|
2847
|
+
return s;
|
|
2848
|
+
|
|
2849
|
+
s += enclen(enc, s);
|
|
2850
|
+
}
|
|
2851
|
+
|
|
2852
|
+
return (UChar* )NULL;
|
|
2853
|
+
}
|
|
2854
|
+
|
|
2855
|
+
static UChar*
|
|
2856
|
+
slow_search_backward(OnigEncoding enc, UChar* target, UChar* target_end,
|
|
2857
|
+
const UChar* text, const UChar* adjust_text,
|
|
2858
|
+
const UChar* text_end, const UChar* text_start)
|
|
2859
|
+
{
|
|
2860
|
+
UChar *t, *p, *s;
|
|
2861
|
+
|
|
2862
|
+
s = (UChar* )text_end;
|
|
2863
|
+
s -= (target_end - target);
|
|
2864
|
+
if (s > text_start)
|
|
2865
|
+
s = (UChar* )text_start;
|
|
2866
|
+
else
|
|
2867
|
+
s = ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc, adjust_text, s);
|
|
2868
|
+
|
|
2869
|
+
while (s >= text) {
|
|
2870
|
+
if (*s == *target) {
|
|
2871
|
+
p = s + 1;
|
|
2872
|
+
t = target + 1;
|
|
2873
|
+
while (t < target_end) {
|
|
2874
|
+
if (*t != *p++)
|
|
2875
|
+
break;
|
|
2876
|
+
t++;
|
|
2877
|
+
}
|
|
2878
|
+
if (t == target_end)
|
|
2879
|
+
return s;
|
|
2880
|
+
}
|
|
2881
|
+
s = (UChar* )onigenc_get_prev_char_head(enc, adjust_text, s);
|
|
2882
|
+
}
|
|
2883
|
+
|
|
2884
|
+
return (UChar* )NULL;
|
|
2885
|
+
}
|
|
2886
|
+
|
|
2887
|
+
static UChar*
|
|
2888
|
+
slow_search_backward_ic(OnigEncoding enc, int case_fold_flag,
|
|
2889
|
+
UChar* target, UChar* target_end,
|
|
2890
|
+
const UChar* text, const UChar* adjust_text,
|
|
2891
|
+
const UChar* text_end, const UChar* text_start)
|
|
2892
|
+
{
|
|
2893
|
+
UChar *s;
|
|
2894
|
+
|
|
2895
|
+
s = (UChar* )text_end;
|
|
2896
|
+
s -= (target_end - target);
|
|
2897
|
+
if (s > text_start)
|
|
2898
|
+
s = (UChar* )text_start;
|
|
2899
|
+
else
|
|
2900
|
+
s = ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc, adjust_text, s);
|
|
2901
|
+
|
|
2902
|
+
while (s >= text) {
|
|
2903
|
+
if (str_lower_case_match(enc, case_fold_flag,
|
|
2904
|
+
target, target_end, s, text_end))
|
|
2905
|
+
return s;
|
|
2906
|
+
|
|
2907
|
+
s = (UChar* )onigenc_get_prev_char_head(enc, adjust_text, s);
|
|
2908
|
+
}
|
|
2909
|
+
|
|
2910
|
+
return (UChar* )NULL;
|
|
2911
|
+
}
|
|
2912
|
+
|
|
2913
|
+
static UChar*
|
|
2914
|
+
bm_search_notrev(regex_t* reg, const UChar* target, const UChar* target_end,
|
|
2915
|
+
const UChar* text, const UChar* text_end,
|
|
2916
|
+
const UChar* text_range)
|
|
2917
|
+
{
|
|
2918
|
+
const UChar *s, *se, *t, *p, *end;
|
|
2919
|
+
const UChar *tail;
|
|
2920
|
+
int skip, tlen1;
|
|
2921
|
+
|
|
2922
|
+
#ifdef ONIG_DEBUG_SEARCH
|
|
2923
|
+
fprintf(stderr, "bm_search_notrev: text: %d, text_end: %d, text_range: %d\n",
|
|
2924
|
+
(int )text, (int )text_end, (int )text_range);
|
|
2925
|
+
#endif
|
|
2926
|
+
|
|
2927
|
+
tail = target_end - 1;
|
|
2928
|
+
tlen1 = tail - target;
|
|
2929
|
+
end = text_range;
|
|
2930
|
+
if (end + tlen1 > text_end)
|
|
2931
|
+
end = text_end - tlen1;
|
|
2932
|
+
|
|
2933
|
+
s = text;
|
|
2934
|
+
|
|
2935
|
+
if (IS_NULL(reg->int_map)) {
|
|
2936
|
+
while (s < end) {
|
|
2937
|
+
p = se = s + tlen1;
|
|
2938
|
+
t = tail;
|
|
2939
|
+
while (*p == *t) {
|
|
2940
|
+
if (t == target) return (UChar* )s;
|
|
2941
|
+
p--; t--;
|
|
2942
|
+
}
|
|
2943
|
+
skip = reg->map[*se];
|
|
2944
|
+
t = s;
|
|
2945
|
+
do {
|
|
2946
|
+
s += enclen(reg->enc, s);
|
|
2947
|
+
} while ((s - t) < skip && s < end);
|
|
2948
|
+
}
|
|
2949
|
+
}
|
|
2950
|
+
else {
|
|
2951
|
+
while (s < end) {
|
|
2952
|
+
p = se = s + tlen1;
|
|
2953
|
+
t = tail;
|
|
2954
|
+
while (*p == *t) {
|
|
2955
|
+
if (t == target) return (UChar* )s;
|
|
2956
|
+
p--; t--;
|
|
2957
|
+
}
|
|
2958
|
+
skip = reg->int_map[*se];
|
|
2959
|
+
t = s;
|
|
2960
|
+
do {
|
|
2961
|
+
s += enclen(reg->enc, s);
|
|
2962
|
+
} while ((s - t) < skip && s < end);
|
|
2963
|
+
}
|
|
2964
|
+
}
|
|
2965
|
+
|
|
2966
|
+
return (UChar* )NULL;
|
|
2967
|
+
}
|
|
2968
|
+
|
|
2969
|
+
static UChar*
|
|
2970
|
+
bm_search(regex_t* reg, const UChar* target, const UChar* target_end,
|
|
2971
|
+
const UChar* text, const UChar* text_end, const UChar* text_range)
|
|
2972
|
+
{
|
|
2973
|
+
const UChar *s, *t, *p, *end;
|
|
2974
|
+
const UChar *tail;
|
|
2975
|
+
|
|
2976
|
+
end = text_range + (target_end - target) - 1;
|
|
2977
|
+
if (end > text_end)
|
|
2978
|
+
end = text_end;
|
|
2979
|
+
|
|
2980
|
+
tail = target_end - 1;
|
|
2981
|
+
s = text + (target_end - target) - 1;
|
|
2982
|
+
if (IS_NULL(reg->int_map)) {
|
|
2983
|
+
while (s < end) {
|
|
2984
|
+
p = s;
|
|
2985
|
+
t = tail;
|
|
2986
|
+
while (*p == *t) {
|
|
2987
|
+
if (t == target) return (UChar* )p;
|
|
2988
|
+
p--; t--;
|
|
2989
|
+
}
|
|
2990
|
+
s += reg->map[*s];
|
|
2991
|
+
}
|
|
2992
|
+
}
|
|
2993
|
+
else { /* see int_map[] */
|
|
2994
|
+
while (s < end) {
|
|
2995
|
+
p = s;
|
|
2996
|
+
t = tail;
|
|
2997
|
+
while (*p == *t) {
|
|
2998
|
+
if (t == target) return (UChar* )p;
|
|
2999
|
+
p--; t--;
|
|
3000
|
+
}
|
|
3001
|
+
s += reg->int_map[*s];
|
|
3002
|
+
}
|
|
3003
|
+
}
|
|
3004
|
+
return (UChar* )NULL;
|
|
3005
|
+
}
|
|
3006
|
+
|
|
3007
|
+
#ifdef USE_INT_MAP_BACKWARD
|
|
3008
|
+
static int
|
|
3009
|
+
set_bm_backward_skip(UChar* s, UChar* end, OnigEncoding enc ARG_UNUSED,
|
|
3010
|
+
int** skip)
|
|
3011
|
+
|
|
3012
|
+
{
|
|
3013
|
+
int i, len;
|
|
3014
|
+
|
|
3015
|
+
if (IS_NULL(*skip)) {
|
|
3016
|
+
*skip = (int* )xmalloc(sizeof(int) * ONIG_CHAR_TABLE_SIZE);
|
|
3017
|
+
if (IS_NULL(*skip)) return ONIGERR_MEMORY;
|
|
3018
|
+
}
|
|
3019
|
+
|
|
3020
|
+
len = end - s;
|
|
3021
|
+
for (i = 0; i < ONIG_CHAR_TABLE_SIZE; i++)
|
|
3022
|
+
(*skip)[i] = len;
|
|
3023
|
+
|
|
3024
|
+
for (i = len - 1; i > 0; i--)
|
|
3025
|
+
(*skip)[s[i]] = i;
|
|
3026
|
+
|
|
3027
|
+
return 0;
|
|
3028
|
+
}
|
|
3029
|
+
|
|
3030
|
+
static UChar*
|
|
3031
|
+
bm_search_backward(regex_t* reg, const UChar* target, const UChar* target_end,
|
|
3032
|
+
const UChar* text, const UChar* adjust_text,
|
|
3033
|
+
const UChar* text_end, const UChar* text_start)
|
|
3034
|
+
{
|
|
3035
|
+
const UChar *s, *t, *p;
|
|
3036
|
+
|
|
3037
|
+
s = text_end - (target_end - target);
|
|
3038
|
+
if (text_start < s)
|
|
3039
|
+
s = text_start;
|
|
3040
|
+
else
|
|
3041
|
+
s = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, adjust_text, s);
|
|
3042
|
+
|
|
3043
|
+
while (s >= text) {
|
|
3044
|
+
p = s;
|
|
3045
|
+
t = target;
|
|
3046
|
+
while (t < target_end && *p == *t) {
|
|
3047
|
+
p++; t++;
|
|
3048
|
+
}
|
|
3049
|
+
if (t == target_end)
|
|
3050
|
+
return (UChar* )s;
|
|
3051
|
+
|
|
3052
|
+
s -= reg->int_map_backward[*s];
|
|
3053
|
+
s = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, adjust_text, s);
|
|
3054
|
+
}
|
|
3055
|
+
|
|
3056
|
+
return (UChar* )NULL;
|
|
3057
|
+
}
|
|
3058
|
+
#endif
|
|
3059
|
+
|
|
3060
|
+
static UChar*
|
|
3061
|
+
map_search(OnigEncoding enc, UChar map[],
|
|
3062
|
+
const UChar* text, const UChar* text_range)
|
|
3063
|
+
{
|
|
3064
|
+
const UChar *s = text;
|
|
3065
|
+
|
|
3066
|
+
while (s < text_range) {
|
|
3067
|
+
if (map[*s]) return (UChar* )s;
|
|
3068
|
+
|
|
3069
|
+
s += enclen(enc, s);
|
|
3070
|
+
}
|
|
3071
|
+
return (UChar* )NULL;
|
|
3072
|
+
}
|
|
3073
|
+
|
|
3074
|
+
static UChar*
|
|
3075
|
+
map_search_backward(OnigEncoding enc, UChar map[],
|
|
3076
|
+
const UChar* text, const UChar* adjust_text,
|
|
3077
|
+
const UChar* text_start)
|
|
3078
|
+
{
|
|
3079
|
+
const UChar *s = text_start;
|
|
3080
|
+
|
|
3081
|
+
while (s >= text) {
|
|
3082
|
+
if (map[*s]) return (UChar* )s;
|
|
3083
|
+
|
|
3084
|
+
s = onigenc_get_prev_char_head(enc, adjust_text, s);
|
|
3085
|
+
}
|
|
3086
|
+
return (UChar* )NULL;
|
|
3087
|
+
}
|
|
3088
|
+
|
|
3089
|
+
extern int
|
|
3090
|
+
onig_match(regex_t* reg, const UChar* str, const UChar* end, const UChar* at, OnigRegion* region,
|
|
3091
|
+
OnigOptionType option)
|
|
3092
|
+
{
|
|
3093
|
+
int r;
|
|
3094
|
+
UChar *prev;
|
|
3095
|
+
OnigMatchArg msa;
|
|
3096
|
+
|
|
3097
|
+
MATCH_ARG_INIT(msa, reg, option, region, at);
|
|
3098
|
+
#ifdef USE_COMBINATION_EXPLOSION_CHECK
|
|
3099
|
+
{
|
|
3100
|
+
int offset = at - str;
|
|
3101
|
+
STATE_CHECK_BUFF_INIT(msa, end - str, offset, reg->num_comb_exp_check);
|
|
3102
|
+
}
|
|
3103
|
+
#endif
|
|
3104
|
+
|
|
3105
|
+
if (region
|
|
3106
|
+
#ifdef USE_POSIX_API_REGION_OPTION
|
|
3107
|
+
&& !IS_POSIX_REGION(option)
|
|
3108
|
+
#endif
|
|
3109
|
+
) {
|
|
3110
|
+
r = onig_region_resize_clear(region, reg->num_mem + 1);
|
|
3111
|
+
}
|
|
3112
|
+
else
|
|
3113
|
+
r = 0;
|
|
3114
|
+
|
|
3115
|
+
if (r == 0) {
|
|
3116
|
+
if (ONIG_IS_OPTION_ON(option, ONIG_OPTION_CHECK_VALIDITY_OF_STRING)) {
|
|
3117
|
+
if (! ONIGENC_IS_VALID_MBC_STRING(reg->enc, str, end)) {
|
|
3118
|
+
r = ONIGERR_INVALID_WIDE_CHAR_VALUE;
|
|
3119
|
+
goto end;
|
|
3120
|
+
}
|
|
3121
|
+
}
|
|
3122
|
+
|
|
3123
|
+
prev = (UChar* )onigenc_get_prev_char_head(reg->enc, str, at);
|
|
3124
|
+
r = match_at(reg, str, end,
|
|
3125
|
+
#ifdef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
|
|
3126
|
+
end,
|
|
3127
|
+
#endif
|
|
3128
|
+
at, prev, &msa);
|
|
3129
|
+
}
|
|
3130
|
+
|
|
3131
|
+
end:
|
|
3132
|
+
MATCH_ARG_FREE(msa);
|
|
3133
|
+
return r;
|
|
3134
|
+
}
|
|
3135
|
+
|
|
3136
|
+
static int
|
|
3137
|
+
forward_search_range(regex_t* reg, const UChar* str, const UChar* end, UChar* s,
|
|
3138
|
+
UChar* range, UChar** low, UChar** high, UChar** low_prev)
|
|
3139
|
+
{
|
|
3140
|
+
UChar *p, *pprev = (UChar* )NULL;
|
|
3141
|
+
|
|
3142
|
+
#ifdef ONIG_DEBUG_SEARCH
|
|
3143
|
+
fprintf(stderr, "forward_search_range: str: %d, end: %d, s: %d, range: %d\n",
|
|
3144
|
+
(int )str, (int )end, (int )s, (int )range);
|
|
3145
|
+
#endif
|
|
3146
|
+
|
|
3147
|
+
p = s;
|
|
3148
|
+
if (reg->dmin > 0) {
|
|
3149
|
+
if (ONIGENC_IS_SINGLEBYTE(reg->enc)) {
|
|
3150
|
+
p += reg->dmin;
|
|
3151
|
+
}
|
|
3152
|
+
else {
|
|
3153
|
+
UChar *q = p + reg->dmin;
|
|
3154
|
+
while (p < q) p += enclen(reg->enc, p);
|
|
3155
|
+
}
|
|
3156
|
+
}
|
|
3157
|
+
|
|
3158
|
+
retry:
|
|
3159
|
+
switch (reg->optimize) {
|
|
3160
|
+
case ONIG_OPTIMIZE_EXACT:
|
|
3161
|
+
p = slow_search(reg->enc, reg->exact, reg->exact_end, p, end, range);
|
|
3162
|
+
break;
|
|
3163
|
+
case ONIG_OPTIMIZE_EXACT_IC:
|
|
3164
|
+
p = slow_search_ic(reg->enc, reg->case_fold_flag,
|
|
3165
|
+
reg->exact, reg->exact_end, p, end, range);
|
|
3166
|
+
break;
|
|
3167
|
+
|
|
3168
|
+
case ONIG_OPTIMIZE_EXACT_BM:
|
|
3169
|
+
p = bm_search(reg, reg->exact, reg->exact_end, p, end, range);
|
|
3170
|
+
break;
|
|
3171
|
+
|
|
3172
|
+
case ONIG_OPTIMIZE_EXACT_BM_NOT_REV:
|
|
3173
|
+
p = bm_search_notrev(reg, reg->exact, reg->exact_end, p, end, range);
|
|
3174
|
+
break;
|
|
3175
|
+
|
|
3176
|
+
case ONIG_OPTIMIZE_MAP:
|
|
3177
|
+
p = map_search(reg->enc, reg->map, p, range);
|
|
3178
|
+
break;
|
|
3179
|
+
}
|
|
3180
|
+
|
|
3181
|
+
if (p && p < range) {
|
|
3182
|
+
if (p - reg->dmin < s) {
|
|
3183
|
+
retry_gate:
|
|
3184
|
+
pprev = p;
|
|
3185
|
+
p += enclen(reg->enc, p);
|
|
3186
|
+
goto retry;
|
|
3187
|
+
}
|
|
3188
|
+
|
|
3189
|
+
if (reg->sub_anchor) {
|
|
3190
|
+
UChar* prev;
|
|
3191
|
+
|
|
3192
|
+
switch (reg->sub_anchor) {
|
|
3193
|
+
case ANCHOR_BEGIN_LINE:
|
|
3194
|
+
if (!ON_STR_BEGIN(p)) {
|
|
3195
|
+
prev = onigenc_get_prev_char_head(reg->enc,
|
|
3196
|
+
(pprev ? pprev : str), p);
|
|
3197
|
+
if (!ONIGENC_IS_MBC_NEWLINE(reg->enc, prev, end))
|
|
3198
|
+
goto retry_gate;
|
|
3199
|
+
}
|
|
3200
|
+
break;
|
|
3201
|
+
|
|
3202
|
+
case ANCHOR_END_LINE:
|
|
3203
|
+
if (ON_STR_END(p)) {
|
|
3204
|
+
#ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE
|
|
3205
|
+
prev = (UChar* )onigenc_get_prev_char_head(reg->enc,
|
|
3206
|
+
(pprev ? pprev : str), p);
|
|
3207
|
+
if (prev && ONIGENC_IS_MBC_NEWLINE(reg->enc, prev, end))
|
|
3208
|
+
goto retry_gate;
|
|
3209
|
+
#endif
|
|
3210
|
+
}
|
|
3211
|
+
else if (! ONIGENC_IS_MBC_NEWLINE(reg->enc, p, end)
|
|
3212
|
+
#ifdef USE_CRNL_AS_LINE_TERMINATOR
|
|
3213
|
+
&& ! ONIGENC_IS_MBC_CRNL(reg->enc, p, end)
|
|
3214
|
+
#endif
|
|
3215
|
+
)
|
|
3216
|
+
goto retry_gate;
|
|
3217
|
+
break;
|
|
3218
|
+
}
|
|
3219
|
+
}
|
|
3220
|
+
|
|
3221
|
+
if (reg->dmax == 0) {
|
|
3222
|
+
*low = p;
|
|
3223
|
+
if (low_prev) {
|
|
3224
|
+
if (*low > s)
|
|
3225
|
+
*low_prev = onigenc_get_prev_char_head(reg->enc, s, p);
|
|
3226
|
+
else
|
|
3227
|
+
*low_prev = onigenc_get_prev_char_head(reg->enc,
|
|
3228
|
+
(pprev ? pprev : str), p);
|
|
3229
|
+
}
|
|
3230
|
+
}
|
|
3231
|
+
else {
|
|
3232
|
+
if (reg->dmax != ONIG_INFINITE_DISTANCE) {
|
|
3233
|
+
*low = p - reg->dmax;
|
|
3234
|
+
if (*low > s) {
|
|
3235
|
+
*low = onigenc_get_right_adjust_char_head_with_prev(reg->enc, s,
|
|
3236
|
+
*low, (const UChar** )low_prev);
|
|
3237
|
+
if (low_prev && IS_NULL(*low_prev))
|
|
3238
|
+
*low_prev = onigenc_get_prev_char_head(reg->enc,
|
|
3239
|
+
(pprev ? pprev : s), *low);
|
|
3240
|
+
}
|
|
3241
|
+
else {
|
|
3242
|
+
if (low_prev)
|
|
3243
|
+
*low_prev = onigenc_get_prev_char_head(reg->enc,
|
|
3244
|
+
(pprev ? pprev : str), *low);
|
|
3245
|
+
}
|
|
3246
|
+
}
|
|
3247
|
+
}
|
|
3248
|
+
/* no needs to adjust *high, *high is used as range check only */
|
|
3249
|
+
*high = p - reg->dmin;
|
|
3250
|
+
|
|
3251
|
+
#ifdef ONIG_DEBUG_SEARCH
|
|
3252
|
+
fprintf(stderr,
|
|
3253
|
+
"forward_search_range success: low: %d, high: %d, dmin: %d, dmax: %d\n",
|
|
3254
|
+
(int )(*low - str), (int )(*high - str), reg->dmin, reg->dmax);
|
|
3255
|
+
#endif
|
|
3256
|
+
return 1; /* success */
|
|
3257
|
+
}
|
|
3258
|
+
|
|
3259
|
+
return 0; /* fail */
|
|
3260
|
+
}
|
|
3261
|
+
|
|
3262
|
+
|
|
3263
|
+
#define BM_BACKWARD_SEARCH_LENGTH_THRESHOLD 100
|
|
3264
|
+
|
|
3265
|
+
static int
|
|
3266
|
+
backward_search_range(regex_t* reg, const UChar* str, const UChar* end,
|
|
3267
|
+
UChar* s, const UChar* range, UChar* adjrange,
|
|
3268
|
+
UChar** low, UChar** high)
|
|
3269
|
+
{
|
|
3270
|
+
UChar *p;
|
|
3271
|
+
|
|
3272
|
+
range += reg->dmin;
|
|
3273
|
+
p = s;
|
|
3274
|
+
|
|
3275
|
+
retry:
|
|
3276
|
+
switch (reg->optimize) {
|
|
3277
|
+
case ONIG_OPTIMIZE_EXACT:
|
|
3278
|
+
exact_method:
|
|
3279
|
+
p = slow_search_backward(reg->enc, reg->exact, reg->exact_end,
|
|
3280
|
+
range, adjrange, end, p);
|
|
3281
|
+
break;
|
|
3282
|
+
|
|
3283
|
+
case ONIG_OPTIMIZE_EXACT_IC:
|
|
3284
|
+
p = slow_search_backward_ic(reg->enc, reg->case_fold_flag,
|
|
3285
|
+
reg->exact, reg->exact_end,
|
|
3286
|
+
range, adjrange, end, p);
|
|
3287
|
+
break;
|
|
3288
|
+
|
|
3289
|
+
case ONIG_OPTIMIZE_EXACT_BM:
|
|
3290
|
+
case ONIG_OPTIMIZE_EXACT_BM_NOT_REV:
|
|
3291
|
+
#ifdef USE_INT_MAP_BACKWARD
|
|
3292
|
+
if (IS_NULL(reg->int_map_backward)) {
|
|
3293
|
+
int r;
|
|
3294
|
+
|
|
3295
|
+
if (s - range < BM_BACKWARD_SEARCH_LENGTH_THRESHOLD)
|
|
3296
|
+
goto exact_method;
|
|
3297
|
+
|
|
3298
|
+
r = set_bm_backward_skip(reg->exact, reg->exact_end, reg->enc,
|
|
3299
|
+
&(reg->int_map_backward));
|
|
3300
|
+
if (r) return r;
|
|
3301
|
+
}
|
|
3302
|
+
p = bm_search_backward(reg, reg->exact, reg->exact_end, range, adjrange,
|
|
3303
|
+
end, p);
|
|
3304
|
+
#else
|
|
3305
|
+
goto exact_method;
|
|
3306
|
+
#endif
|
|
3307
|
+
break;
|
|
3308
|
+
|
|
3309
|
+
case ONIG_OPTIMIZE_MAP:
|
|
3310
|
+
p = map_search_backward(reg->enc, reg->map, range, adjrange, p);
|
|
3311
|
+
break;
|
|
3312
|
+
}
|
|
3313
|
+
|
|
3314
|
+
if (p) {
|
|
3315
|
+
if (reg->sub_anchor) {
|
|
3316
|
+
UChar* prev;
|
|
3317
|
+
|
|
3318
|
+
switch (reg->sub_anchor) {
|
|
3319
|
+
case ANCHOR_BEGIN_LINE:
|
|
3320
|
+
if (!ON_STR_BEGIN(p)) {
|
|
3321
|
+
prev = onigenc_get_prev_char_head(reg->enc, str, p);
|
|
3322
|
+
if (!ONIGENC_IS_MBC_NEWLINE(reg->enc, prev, end)) {
|
|
3323
|
+
p = prev;
|
|
3324
|
+
goto retry;
|
|
3325
|
+
}
|
|
3326
|
+
}
|
|
3327
|
+
break;
|
|
3328
|
+
|
|
3329
|
+
case ANCHOR_END_LINE:
|
|
3330
|
+
if (ON_STR_END(p)) {
|
|
3331
|
+
#ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE
|
|
3332
|
+
prev = onigenc_get_prev_char_head(reg->enc, adjrange, p);
|
|
3333
|
+
if (IS_NULL(prev)) goto fail;
|
|
3334
|
+
if (ONIGENC_IS_MBC_NEWLINE(reg->enc, prev, end)) {
|
|
3335
|
+
p = prev;
|
|
3336
|
+
goto retry;
|
|
3337
|
+
}
|
|
3338
|
+
#endif
|
|
3339
|
+
}
|
|
3340
|
+
else if (! ONIGENC_IS_MBC_NEWLINE(reg->enc, p, end)
|
|
3341
|
+
#ifdef USE_CRNL_AS_LINE_TERMINATOR
|
|
3342
|
+
&& ! ONIGENC_IS_MBC_CRNL(reg->enc, p, end)
|
|
3343
|
+
#endif
|
|
3344
|
+
) {
|
|
3345
|
+
p = onigenc_get_prev_char_head(reg->enc, adjrange, p);
|
|
3346
|
+
if (IS_NULL(p)) goto fail;
|
|
3347
|
+
goto retry;
|
|
3348
|
+
}
|
|
3349
|
+
break;
|
|
3350
|
+
}
|
|
3351
|
+
}
|
|
3352
|
+
|
|
3353
|
+
/* no needs to adjust *high, *high is used as range check only */
|
|
3354
|
+
if (reg->dmax != ONIG_INFINITE_DISTANCE) {
|
|
3355
|
+
*low = p - reg->dmax;
|
|
3356
|
+
*high = p - reg->dmin;
|
|
3357
|
+
*high = onigenc_get_right_adjust_char_head(reg->enc, adjrange, *high);
|
|
3358
|
+
}
|
|
3359
|
+
|
|
3360
|
+
#ifdef ONIG_DEBUG_SEARCH
|
|
3361
|
+
fprintf(stderr, "backward_search_range: low: %d, high: %d\n",
|
|
3362
|
+
(int )(*low - str), (int )(*high - str));
|
|
3363
|
+
#endif
|
|
3364
|
+
return 1; /* success */
|
|
3365
|
+
}
|
|
3366
|
+
|
|
3367
|
+
fail:
|
|
3368
|
+
#ifdef ONIG_DEBUG_SEARCH
|
|
3369
|
+
fprintf(stderr, "backward_search_range: fail.\n");
|
|
3370
|
+
#endif
|
|
3371
|
+
return 0; /* fail */
|
|
3372
|
+
}
|
|
3373
|
+
|
|
3374
|
+
|
|
3375
|
+
extern int
|
|
3376
|
+
onig_search(regex_t* reg, const UChar* str, const UChar* end,
|
|
3377
|
+
const UChar* start, const UChar* range, OnigRegion* region, OnigOptionType option)
|
|
3378
|
+
{
|
|
3379
|
+
int r;
|
|
3380
|
+
UChar *s, *prev;
|
|
3381
|
+
OnigMatchArg msa;
|
|
3382
|
+
const UChar *orig_start = start;
|
|
3383
|
+
#ifdef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
|
|
3384
|
+
const UChar *orig_range = range;
|
|
3385
|
+
#endif
|
|
3386
|
+
|
|
3387
|
+
#ifdef ONIG_DEBUG_SEARCH
|
|
3388
|
+
fprintf(stderr,
|
|
3389
|
+
"onig_search (entry point): str: %d, end: %d, start: %d, range: %d\n",
|
|
3390
|
+
(int )str, (int )(end - str), (int )(start - str), (int )(range - str));
|
|
3391
|
+
#endif
|
|
3392
|
+
|
|
3393
|
+
if (region
|
|
3394
|
+
#ifdef USE_POSIX_API_REGION_OPTION
|
|
3395
|
+
&& !IS_POSIX_REGION(option)
|
|
3396
|
+
#endif
|
|
3397
|
+
) {
|
|
3398
|
+
r = onig_region_resize_clear(region, reg->num_mem + 1);
|
|
3399
|
+
if (r) goto finish_no_msa;
|
|
3400
|
+
}
|
|
3401
|
+
|
|
3402
|
+
if (start > end || start < str) goto mismatch_no_msa;
|
|
3403
|
+
|
|
3404
|
+
if (ONIG_IS_OPTION_ON(option, ONIG_OPTION_CHECK_VALIDITY_OF_STRING)) {
|
|
3405
|
+
if (! ONIGENC_IS_VALID_MBC_STRING(reg->enc, str, end)) {
|
|
3406
|
+
r = ONIGERR_INVALID_WIDE_CHAR_VALUE;
|
|
3407
|
+
goto finish_no_msa;
|
|
3408
|
+
}
|
|
3409
|
+
}
|
|
3410
|
+
|
|
3411
|
+
|
|
3412
|
+
#ifdef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
|
|
3413
|
+
#ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
|
|
3414
|
+
#define MATCH_AND_RETURN_CHECK(upper_range) \
|
|
3415
|
+
r = match_at(reg, str, end, (upper_range), s, prev, &msa); \
|
|
3416
|
+
if (r != ONIG_MISMATCH) {\
|
|
3417
|
+
if (r >= 0) {\
|
|
3418
|
+
if (! IS_FIND_LONGEST(reg->options)) {\
|
|
3419
|
+
goto match;\
|
|
3420
|
+
}\
|
|
3421
|
+
}\
|
|
3422
|
+
else goto finish; /* error */ \
|
|
3423
|
+
}
|
|
3424
|
+
#else
|
|
3425
|
+
#define MATCH_AND_RETURN_CHECK(upper_range) \
|
|
3426
|
+
r = match_at(reg, str, end, (upper_range), s, prev, &msa); \
|
|
3427
|
+
if (r != ONIG_MISMATCH) {\
|
|
3428
|
+
if (r >= 0) {\
|
|
3429
|
+
goto match;\
|
|
3430
|
+
}\
|
|
3431
|
+
else goto finish; /* error */ \
|
|
3432
|
+
}
|
|
3433
|
+
#endif /* USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE */
|
|
3434
|
+
#else
|
|
3435
|
+
#ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
|
|
3436
|
+
#define MATCH_AND_RETURN_CHECK(none) \
|
|
3437
|
+
r = match_at(reg, str, end, s, prev, &msa);\
|
|
3438
|
+
if (r != ONIG_MISMATCH) {\
|
|
3439
|
+
if (r >= 0) {\
|
|
3440
|
+
if (! IS_FIND_LONGEST(reg->options)) {\
|
|
3441
|
+
goto match;\
|
|
3442
|
+
}\
|
|
3443
|
+
}\
|
|
3444
|
+
else goto finish; /* error */ \
|
|
3445
|
+
}
|
|
3446
|
+
#else
|
|
3447
|
+
#define MATCH_AND_RETURN_CHECK(none) \
|
|
3448
|
+
r = match_at(reg, str, end, s, prev, &msa);\
|
|
3449
|
+
if (r != ONIG_MISMATCH) {\
|
|
3450
|
+
if (r >= 0) {\
|
|
3451
|
+
goto match;\
|
|
3452
|
+
}\
|
|
3453
|
+
else goto finish; /* error */ \
|
|
3454
|
+
}
|
|
3455
|
+
#endif /* USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE */
|
|
3456
|
+
#endif /* USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE */
|
|
3457
|
+
|
|
3458
|
+
|
|
3459
|
+
/* anchor optimize: resume search range */
|
|
3460
|
+
if (reg->anchor != 0 && str < end) {
|
|
3461
|
+
UChar *min_semi_end, *max_semi_end;
|
|
3462
|
+
|
|
3463
|
+
if (reg->anchor & ANCHOR_BEGIN_POSITION) {
|
|
3464
|
+
/* search start-position only */
|
|
3465
|
+
begin_position:
|
|
3466
|
+
if (range > start)
|
|
3467
|
+
range = start + 1;
|
|
3468
|
+
else
|
|
3469
|
+
range = start;
|
|
3470
|
+
}
|
|
3471
|
+
else if (reg->anchor & ANCHOR_BEGIN_BUF) {
|
|
3472
|
+
/* search str-position only */
|
|
3473
|
+
if (range > start) {
|
|
3474
|
+
if (start != str) goto mismatch_no_msa;
|
|
3475
|
+
range = str + 1;
|
|
3476
|
+
}
|
|
3477
|
+
else {
|
|
3478
|
+
if (range <= str) {
|
|
3479
|
+
start = str;
|
|
3480
|
+
range = str;
|
|
3481
|
+
}
|
|
3482
|
+
else
|
|
3483
|
+
goto mismatch_no_msa;
|
|
3484
|
+
}
|
|
3485
|
+
}
|
|
3486
|
+
else if (reg->anchor & ANCHOR_END_BUF) {
|
|
3487
|
+
min_semi_end = max_semi_end = (UChar* )end;
|
|
3488
|
+
|
|
3489
|
+
end_buf:
|
|
3490
|
+
if ((OnigLen )(max_semi_end - str) < reg->anchor_dmin)
|
|
3491
|
+
goto mismatch_no_msa;
|
|
3492
|
+
|
|
3493
|
+
if (range > start) {
|
|
3494
|
+
if ((OnigLen )(min_semi_end - start) > reg->anchor_dmax) {
|
|
3495
|
+
start = min_semi_end - reg->anchor_dmax;
|
|
3496
|
+
if (start < end)
|
|
3497
|
+
start = onigenc_get_right_adjust_char_head(reg->enc, str, start);
|
|
3498
|
+
}
|
|
3499
|
+
if ((OnigLen )(max_semi_end - (range - 1)) < reg->anchor_dmin) {
|
|
3500
|
+
range = max_semi_end - reg->anchor_dmin + 1;
|
|
3501
|
+
}
|
|
3502
|
+
|
|
3503
|
+
if (start > range) goto mismatch_no_msa;
|
|
3504
|
+
/* If start == range, match with empty at end.
|
|
3505
|
+
Backward search is used. */
|
|
3506
|
+
}
|
|
3507
|
+
else {
|
|
3508
|
+
if ((OnigLen )(min_semi_end - range) > reg->anchor_dmax) {
|
|
3509
|
+
range = min_semi_end - reg->anchor_dmax;
|
|
3510
|
+
}
|
|
3511
|
+
if ((OnigLen )(max_semi_end - start) < reg->anchor_dmin) {
|
|
3512
|
+
start = max_semi_end - reg->anchor_dmin;
|
|
3513
|
+
start = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, str, start);
|
|
3514
|
+
}
|
|
3515
|
+
if (range > start) goto mismatch_no_msa;
|
|
3516
|
+
}
|
|
3517
|
+
}
|
|
3518
|
+
else if (reg->anchor & ANCHOR_SEMI_END_BUF) {
|
|
3519
|
+
UChar* pre_end = ONIGENC_STEP_BACK(reg->enc, str, end, 1);
|
|
3520
|
+
|
|
3521
|
+
max_semi_end = (UChar* )end;
|
|
3522
|
+
if (ONIGENC_IS_MBC_NEWLINE(reg->enc, pre_end, end)) {
|
|
3523
|
+
min_semi_end = pre_end;
|
|
3524
|
+
|
|
3525
|
+
#ifdef USE_CRNL_AS_LINE_TERMINATOR
|
|
3526
|
+
pre_end = ONIGENC_STEP_BACK(reg->enc, str, pre_end, 1);
|
|
3527
|
+
if (IS_NOT_NULL(pre_end) &&
|
|
3528
|
+
ONIGENC_IS_MBC_CRNL(reg->enc, pre_end, end)) {
|
|
3529
|
+
min_semi_end = pre_end;
|
|
3530
|
+
}
|
|
3531
|
+
#endif
|
|
3532
|
+
if (min_semi_end > str && start <= min_semi_end) {
|
|
3533
|
+
goto end_buf;
|
|
3534
|
+
}
|
|
3535
|
+
}
|
|
3536
|
+
else {
|
|
3537
|
+
min_semi_end = (UChar* )end;
|
|
3538
|
+
goto end_buf;
|
|
3539
|
+
}
|
|
3540
|
+
}
|
|
3541
|
+
else if ((reg->anchor & ANCHOR_ANYCHAR_STAR_ML)) {
|
|
3542
|
+
goto begin_position;
|
|
3543
|
+
}
|
|
3544
|
+
}
|
|
3545
|
+
else if (str == end) { /* empty string */
|
|
3546
|
+
static const UChar* address_for_empty_string = (UChar* )"";
|
|
3547
|
+
|
|
3548
|
+
#ifdef ONIG_DEBUG_SEARCH
|
|
3549
|
+
fprintf(stderr, "onig_search: empty string.\n");
|
|
3550
|
+
#endif
|
|
3551
|
+
|
|
3552
|
+
if (reg->threshold_len == 0) {
|
|
3553
|
+
start = end = str = address_for_empty_string;
|
|
3554
|
+
s = (UChar* )start;
|
|
3555
|
+
prev = (UChar* )NULL;
|
|
3556
|
+
|
|
3557
|
+
MATCH_ARG_INIT(msa, reg, option, region, start);
|
|
3558
|
+
#ifdef USE_COMBINATION_EXPLOSION_CHECK
|
|
3559
|
+
msa.state_check_buff = (void* )0;
|
|
3560
|
+
msa.state_check_buff_size = 0; /* NO NEED, for valgrind */
|
|
3561
|
+
#endif
|
|
3562
|
+
MATCH_AND_RETURN_CHECK(end);
|
|
3563
|
+
goto mismatch;
|
|
3564
|
+
}
|
|
3565
|
+
goto mismatch_no_msa;
|
|
3566
|
+
}
|
|
3567
|
+
|
|
3568
|
+
#ifdef ONIG_DEBUG_SEARCH
|
|
3569
|
+
fprintf(stderr, "onig_search(apply anchor): end: %d, start: %d, range: %d\n",
|
|
3570
|
+
(int )(end - str), (int )(start - str), (int )(range - str));
|
|
3571
|
+
#endif
|
|
3572
|
+
|
|
3573
|
+
MATCH_ARG_INIT(msa, reg, option, region, orig_start);
|
|
3574
|
+
#ifdef USE_COMBINATION_EXPLOSION_CHECK
|
|
3575
|
+
{
|
|
3576
|
+
int offset = (MIN(start, range) - str);
|
|
3577
|
+
STATE_CHECK_BUFF_INIT(msa, end - str, offset, reg->num_comb_exp_check);
|
|
3578
|
+
}
|
|
3579
|
+
#endif
|
|
3580
|
+
|
|
3581
|
+
s = (UChar* )start;
|
|
3582
|
+
if (range > start) { /* forward search */
|
|
3583
|
+
if (s > str)
|
|
3584
|
+
prev = onigenc_get_prev_char_head(reg->enc, str, s);
|
|
3585
|
+
else
|
|
3586
|
+
prev = (UChar* )NULL;
|
|
3587
|
+
|
|
3588
|
+
if (reg->optimize != ONIG_OPTIMIZE_NONE) {
|
|
3589
|
+
UChar *sch_range, *low, *high, *low_prev;
|
|
3590
|
+
|
|
3591
|
+
sch_range = (UChar* )range;
|
|
3592
|
+
if (reg->dmax != 0) {
|
|
3593
|
+
if (reg->dmax == ONIG_INFINITE_DISTANCE)
|
|
3594
|
+
sch_range = (UChar* )end;
|
|
3595
|
+
else {
|
|
3596
|
+
sch_range += reg->dmax;
|
|
3597
|
+
if (sch_range > end) sch_range = (UChar* )end;
|
|
3598
|
+
}
|
|
3599
|
+
}
|
|
3600
|
+
|
|
3601
|
+
if ((end - start) < reg->threshold_len)
|
|
3602
|
+
goto mismatch;
|
|
3603
|
+
|
|
3604
|
+
if (reg->dmax != ONIG_INFINITE_DISTANCE) {
|
|
3605
|
+
do {
|
|
3606
|
+
if (! forward_search_range(reg, str, end, s, sch_range,
|
|
3607
|
+
&low, &high, &low_prev)) goto mismatch;
|
|
3608
|
+
if (s < low) {
|
|
3609
|
+
s = low;
|
|
3610
|
+
prev = low_prev;
|
|
3611
|
+
}
|
|
3612
|
+
while (s <= high) {
|
|
3613
|
+
MATCH_AND_RETURN_CHECK(orig_range);
|
|
3614
|
+
prev = s;
|
|
3615
|
+
s += enclen(reg->enc, s);
|
|
3616
|
+
}
|
|
3617
|
+
} while (s < range);
|
|
3618
|
+
goto mismatch;
|
|
3619
|
+
}
|
|
3620
|
+
else { /* check only. */
|
|
3621
|
+
if (! forward_search_range(reg, str, end, s, sch_range,
|
|
3622
|
+
&low, &high, (UChar** )NULL)) goto mismatch;
|
|
3623
|
+
|
|
3624
|
+
if ((reg->anchor & ANCHOR_ANYCHAR_STAR) != 0) {
|
|
3625
|
+
do {
|
|
3626
|
+
MATCH_AND_RETURN_CHECK(orig_range);
|
|
3627
|
+
prev = s;
|
|
3628
|
+
s += enclen(reg->enc, s);
|
|
3629
|
+
|
|
3630
|
+
if ((reg->anchor & (ANCHOR_LOOK_BEHIND | ANCHOR_PREC_READ_NOT)) == 0) {
|
|
3631
|
+
while (!ONIGENC_IS_MBC_NEWLINE(reg->enc, prev, end) && s < range) {
|
|
3632
|
+
prev = s;
|
|
3633
|
+
s += enclen(reg->enc, s);
|
|
3634
|
+
}
|
|
3635
|
+
}
|
|
3636
|
+
} while (s < range);
|
|
3637
|
+
goto mismatch;
|
|
3638
|
+
}
|
|
3639
|
+
}
|
|
3640
|
+
}
|
|
3641
|
+
|
|
3642
|
+
do {
|
|
3643
|
+
MATCH_AND_RETURN_CHECK(orig_range);
|
|
3644
|
+
prev = s;
|
|
3645
|
+
s += enclen(reg->enc, s);
|
|
3646
|
+
} while (s < range);
|
|
3647
|
+
|
|
3648
|
+
if (s == range) { /* because empty match with /$/. */
|
|
3649
|
+
MATCH_AND_RETURN_CHECK(orig_range);
|
|
3650
|
+
}
|
|
3651
|
+
}
|
|
3652
|
+
else { /* backward search */
|
|
3653
|
+
#ifdef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
|
|
3654
|
+
if (orig_start < end)
|
|
3655
|
+
orig_start += enclen(reg->enc, orig_start); /* is upper range */
|
|
3656
|
+
#endif
|
|
3657
|
+
|
|
3658
|
+
if (reg->optimize != ONIG_OPTIMIZE_NONE) {
|
|
3659
|
+
UChar *low, *high, *adjrange, *sch_start;
|
|
3660
|
+
|
|
3661
|
+
if (range < end)
|
|
3662
|
+
adjrange = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, str, range);
|
|
3663
|
+
else
|
|
3664
|
+
adjrange = (UChar* )end;
|
|
3665
|
+
|
|
3666
|
+
if (reg->dmax != ONIG_INFINITE_DISTANCE &&
|
|
3667
|
+
(end - range) >= reg->threshold_len) {
|
|
3668
|
+
do {
|
|
3669
|
+
sch_start = s + reg->dmax;
|
|
3670
|
+
if (sch_start > end) sch_start = (UChar* )end;
|
|
3671
|
+
if (backward_search_range(reg, str, end, sch_start, range, adjrange,
|
|
3672
|
+
&low, &high) <= 0)
|
|
3673
|
+
goto mismatch;
|
|
3674
|
+
|
|
3675
|
+
if (s > high)
|
|
3676
|
+
s = high;
|
|
3677
|
+
|
|
3678
|
+
while (s >= low) {
|
|
3679
|
+
prev = onigenc_get_prev_char_head(reg->enc, str, s);
|
|
3680
|
+
MATCH_AND_RETURN_CHECK(orig_start);
|
|
3681
|
+
s = prev;
|
|
3682
|
+
}
|
|
3683
|
+
} while (s >= range);
|
|
3684
|
+
goto mismatch;
|
|
3685
|
+
}
|
|
3686
|
+
else { /* check only. */
|
|
3687
|
+
if ((end - range) < reg->threshold_len) goto mismatch;
|
|
3688
|
+
|
|
3689
|
+
sch_start = s;
|
|
3690
|
+
if (reg->dmax != 0) {
|
|
3691
|
+
if (reg->dmax == ONIG_INFINITE_DISTANCE)
|
|
3692
|
+
sch_start = (UChar* )end;
|
|
3693
|
+
else {
|
|
3694
|
+
sch_start += reg->dmax;
|
|
3695
|
+
if (sch_start > end) sch_start = (UChar* )end;
|
|
3696
|
+
else
|
|
3697
|
+
sch_start = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc,
|
|
3698
|
+
start, sch_start);
|
|
3699
|
+
}
|
|
3700
|
+
}
|
|
3701
|
+
if (backward_search_range(reg, str, end, sch_start, range, adjrange,
|
|
3702
|
+
&low, &high) <= 0) goto mismatch;
|
|
3703
|
+
}
|
|
3704
|
+
}
|
|
3705
|
+
|
|
3706
|
+
do {
|
|
3707
|
+
prev = onigenc_get_prev_char_head(reg->enc, str, s);
|
|
3708
|
+
MATCH_AND_RETURN_CHECK(orig_start);
|
|
3709
|
+
s = prev;
|
|
3710
|
+
} while (s >= range);
|
|
3711
|
+
}
|
|
3712
|
+
|
|
3713
|
+
mismatch:
|
|
3714
|
+
#ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
|
|
3715
|
+
if (IS_FIND_LONGEST(reg->options)) {
|
|
3716
|
+
if (msa.best_len >= 0) {
|
|
3717
|
+
s = msa.best_s;
|
|
3718
|
+
goto match;
|
|
3719
|
+
}
|
|
3720
|
+
}
|
|
3721
|
+
#endif
|
|
3722
|
+
r = ONIG_MISMATCH;
|
|
3723
|
+
|
|
3724
|
+
finish:
|
|
3725
|
+
MATCH_ARG_FREE(msa);
|
|
3726
|
+
|
|
3727
|
+
/* If result is mismatch and no FIND_NOT_EMPTY option,
|
|
3728
|
+
then the region is not set in match_at(). */
|
|
3729
|
+
if (IS_FIND_NOT_EMPTY(reg->options) && region
|
|
3730
|
+
#ifdef USE_POSIX_API_REGION_OPTION
|
|
3731
|
+
&& !IS_POSIX_REGION(option)
|
|
3732
|
+
#endif
|
|
3733
|
+
) {
|
|
3734
|
+
onig_region_clear(region);
|
|
3735
|
+
}
|
|
3736
|
+
|
|
3737
|
+
#ifdef ONIG_DEBUG
|
|
3738
|
+
if (r != ONIG_MISMATCH)
|
|
3739
|
+
fprintf(stderr, "onig_search: error %d\n", r);
|
|
3740
|
+
#endif
|
|
3741
|
+
return r;
|
|
3742
|
+
|
|
3743
|
+
mismatch_no_msa:
|
|
3744
|
+
r = ONIG_MISMATCH;
|
|
3745
|
+
finish_no_msa:
|
|
3746
|
+
#ifdef ONIG_DEBUG
|
|
3747
|
+
if (r != ONIG_MISMATCH)
|
|
3748
|
+
fprintf(stderr, "onig_search: error %d\n", r);
|
|
3749
|
+
#endif
|
|
3750
|
+
return r;
|
|
3751
|
+
|
|
3752
|
+
match:
|
|
3753
|
+
MATCH_ARG_FREE(msa);
|
|
3754
|
+
return s - str;
|
|
3755
|
+
}
|
|
3756
|
+
|
|
3757
|
+
extern int
|
|
3758
|
+
onig_scan(regex_t* reg, const UChar* str, const UChar* end,
|
|
3759
|
+
OnigRegion* region, OnigOptionType option,
|
|
3760
|
+
int (*scan_callback)(int, int, OnigRegion*, void*),
|
|
3761
|
+
void* callback_arg)
|
|
3762
|
+
{
|
|
3763
|
+
int r;
|
|
3764
|
+
int n;
|
|
3765
|
+
int rs;
|
|
3766
|
+
const UChar* start;
|
|
3767
|
+
|
|
3768
|
+
if (ONIG_IS_OPTION_ON(option, ONIG_OPTION_CHECK_VALIDITY_OF_STRING)) {
|
|
3769
|
+
if (! ONIGENC_IS_VALID_MBC_STRING(reg->enc, str, end))
|
|
3770
|
+
return ONIGERR_INVALID_WIDE_CHAR_VALUE;
|
|
3771
|
+
|
|
3772
|
+
ONIG_OPTION_OFF(option, ONIG_OPTION_CHECK_VALIDITY_OF_STRING);
|
|
3773
|
+
}
|
|
3774
|
+
|
|
3775
|
+
n = 0;
|
|
3776
|
+
start = str;
|
|
3777
|
+
while (1) {
|
|
3778
|
+
r = onig_search(reg, str, end, start, end, region, option);
|
|
3779
|
+
if (r >= 0) {
|
|
3780
|
+
rs = scan_callback(n, r, region, callback_arg);
|
|
3781
|
+
n++;
|
|
3782
|
+
if (rs != 0)
|
|
3783
|
+
return rs;
|
|
3784
|
+
|
|
3785
|
+
if (region->end[0] == start - str)
|
|
3786
|
+
start++;
|
|
3787
|
+
else
|
|
3788
|
+
start = str + region->end[0];
|
|
3789
|
+
|
|
3790
|
+
if (start > end)
|
|
3791
|
+
break;
|
|
3792
|
+
}
|
|
3793
|
+
else if (r == ONIG_MISMATCH) {
|
|
3794
|
+
break;
|
|
3795
|
+
}
|
|
3796
|
+
else { /* error */
|
|
3797
|
+
return r;
|
|
3798
|
+
}
|
|
3799
|
+
}
|
|
3800
|
+
|
|
3801
|
+
return n;
|
|
3802
|
+
}
|
|
3803
|
+
|
|
3804
|
+
extern OnigEncoding
|
|
3805
|
+
onig_get_encoding(regex_t* reg)
|
|
3806
|
+
{
|
|
3807
|
+
return reg->enc;
|
|
3808
|
+
}
|
|
3809
|
+
|
|
3810
|
+
extern OnigOptionType
|
|
3811
|
+
onig_get_options(regex_t* reg)
|
|
3812
|
+
{
|
|
3813
|
+
return reg->options;
|
|
3814
|
+
}
|
|
3815
|
+
|
|
3816
|
+
extern OnigCaseFoldType
|
|
3817
|
+
onig_get_case_fold_flag(regex_t* reg)
|
|
3818
|
+
{
|
|
3819
|
+
return reg->case_fold_flag;
|
|
3820
|
+
}
|
|
3821
|
+
|
|
3822
|
+
extern OnigSyntaxType*
|
|
3823
|
+
onig_get_syntax(regex_t* reg)
|
|
3824
|
+
{
|
|
3825
|
+
return reg->syntax;
|
|
3826
|
+
}
|
|
3827
|
+
|
|
3828
|
+
extern int
|
|
3829
|
+
onig_number_of_captures(regex_t* reg)
|
|
3830
|
+
{
|
|
3831
|
+
return reg->num_mem;
|
|
3832
|
+
}
|
|
3833
|
+
|
|
3834
|
+
extern int
|
|
3835
|
+
onig_number_of_capture_histories(regex_t* reg)
|
|
3836
|
+
{
|
|
3837
|
+
#ifdef USE_CAPTURE_HISTORY
|
|
3838
|
+
int i, n;
|
|
3839
|
+
|
|
3840
|
+
n = 0;
|
|
3841
|
+
for (i = 0; i <= ONIG_MAX_CAPTURE_HISTORY_GROUP; i++) {
|
|
3842
|
+
if (BIT_STATUS_AT(reg->capture_history, i) != 0)
|
|
3843
|
+
n++;
|
|
3844
|
+
}
|
|
3845
|
+
return n;
|
|
3846
|
+
#else
|
|
3847
|
+
return 0;
|
|
3848
|
+
#endif
|
|
3849
|
+
}
|
|
3850
|
+
|
|
3851
|
+
extern void
|
|
3852
|
+
onig_copy_encoding(OnigEncoding to, OnigEncoding from)
|
|
3853
|
+
{
|
|
3854
|
+
*to = *from;
|
|
3855
|
+
}
|
|
3856
|
+
|