cchardet 0.2.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.gitignore +8 -0
- data/.gitmodules +3 -0
- data/.rubocop.yml +11 -0
- data/CHANGELOG.md +5 -0
- data/Gemfile +10 -0
- data/README.md +35 -0
- data/Rakefile +15 -0
- data/cchardet.gemspec +30 -0
- data/ext/cchardet/extconf.rb +26 -0
- data/ext/uchardet/.gitignore +1 -0
- data/ext/uchardet/.gitlab-ci.yml +106 -0
- data/ext/uchardet/AUTHORS +16 -0
- data/ext/uchardet/CMakeLists.txt +74 -0
- data/ext/uchardet/COPYING +1316 -0
- data/ext/uchardet/INSTALL +26 -0
- data/ext/uchardet/README.md +295 -0
- data/ext/uchardet/build-mac/uchardet.cpp +7 -0
- data/ext/uchardet/build-mac/uchardet.xcodeproj/project.pbxproj +543 -0
- data/ext/uchardet/build-mac/uchardet.xcodeproj/project.xcworkspace/contents.xcworkspacedata +7 -0
- data/ext/uchardet/build-mac/uchardet.xcodeproj/project.xcworkspace/xcshareddata/uchardet.xccheckout +41 -0
- data/ext/uchardet/build-mac/uchardet.xcodeproj/xcshareddata/xcschemes/uchardet-ios.xcscheme +77 -0
- data/ext/uchardet/build-mac/uchardet.xcodeproj/xcshareddata/xcschemes/uchardet.xcscheme +77 -0
- data/ext/uchardet/doc/CMakeLists.txt +6 -0
- data/ext/uchardet/doc/README.maintainer +59 -0
- data/ext/uchardet/doc/uchardet.1 +18 -0
- data/ext/uchardet/script/BuildLangModel.py +533 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangArabicModel.log +142 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangCroatianModel.log +157 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangCzechModel.log +161 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangDanishModel.log +158 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangEsperantoModel.log +110 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangEstonianModel.log +159 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangFinnishModel.log +156 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangFrenchModel.log +116 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangGermanModel.log +159 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangGreekModel.log +272 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangHungarianModel.log +109 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangIrishModel.log +156 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangItalianModel.log +162 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangLatvianModel.log +162 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangLithuanianModel.log +162 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangMalteseModel.log +147 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangPolishModel.log +154 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangPortugueseModel.log +166 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangRomanianModel.log +153 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangSlovakModel.log +158 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangSloveneModel.log +148 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangSpanishModel.log +109 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangSwedishModel.log +151 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangThaiModel.log +141 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangTurkishModel.log +113 -0
- data/ext/uchardet/script/BuildLangModelLogs/LangVietnameseModel.log +121 -0
- data/ext/uchardet/script/README +63 -0
- data/ext/uchardet/script/charsets/codepoints.py +53 -0
- data/ext/uchardet/script/charsets/db.py +73 -0
- data/ext/uchardet/script/charsets/ibm852.py +72 -0
- data/ext/uchardet/script/charsets/iso-8859-1.py +85 -0
- data/ext/uchardet/script/charsets/iso-8859-10.py +73 -0
- data/ext/uchardet/script/charsets/iso-8859-11.py +77 -0
- data/ext/uchardet/script/charsets/iso-8859-13.py +72 -0
- data/ext/uchardet/script/charsets/iso-8859-15.py +80 -0
- data/ext/uchardet/script/charsets/iso-8859-16.py +83 -0
- data/ext/uchardet/script/charsets/iso-8859-2.py +73 -0
- data/ext/uchardet/script/charsets/iso-8859-3.py +75 -0
- data/ext/uchardet/script/charsets/iso-8859-4.py +73 -0
- data/ext/uchardet/script/charsets/iso-8859-6.py +73 -0
- data/ext/uchardet/script/charsets/iso-8859-7.py +73 -0
- data/ext/uchardet/script/charsets/iso-8859-9.py +76 -0
- data/ext/uchardet/script/charsets/mac-centraleurope.py +72 -0
- data/ext/uchardet/script/charsets/tis-620.py +77 -0
- data/ext/uchardet/script/charsets/viscii.py +72 -0
- data/ext/uchardet/script/charsets/windows-1250.py +75 -0
- data/ext/uchardet/script/charsets/windows-1252.py +76 -0
- data/ext/uchardet/script/charsets/windows-1253.py +72 -0
- data/ext/uchardet/script/charsets/windows-1256.py +75 -0
- data/ext/uchardet/script/charsets/windows-1257.py +72 -0
- data/ext/uchardet/script/charsets/windows-1258.py +72 -0
- data/ext/uchardet/script/debug.sh +9 -0
- data/ext/uchardet/script/header-template.cpp +38 -0
- data/ext/uchardet/script/langs/ar.py +59 -0
- data/ext/uchardet/script/langs/cs.py +80 -0
- data/ext/uchardet/script/langs/da.py +69 -0
- data/ext/uchardet/script/langs/de.py +69 -0
- data/ext/uchardet/script/langs/el.py +55 -0
- data/ext/uchardet/script/langs/eo.py +67 -0
- data/ext/uchardet/script/langs/es.py +69 -0
- data/ext/uchardet/script/langs/et.py +57 -0
- data/ext/uchardet/script/langs/fi.py +60 -0
- data/ext/uchardet/script/langs/fr.py +79 -0
- data/ext/uchardet/script/langs/ga.py +60 -0
- data/ext/uchardet/script/langs/hr.py +59 -0
- data/ext/uchardet/script/langs/hu.py +66 -0
- data/ext/uchardet/script/langs/it.py +56 -0
- data/ext/uchardet/script/langs/lt.py +70 -0
- data/ext/uchardet/script/langs/lv.py +69 -0
- data/ext/uchardet/script/langs/mt.py +80 -0
- data/ext/uchardet/script/langs/pl.py +81 -0
- data/ext/uchardet/script/langs/pt.py +80 -0
- data/ext/uchardet/script/langs/ro.py +65 -0
- data/ext/uchardet/script/langs/sk.py +80 -0
- data/ext/uchardet/script/langs/sl.py +59 -0
- data/ext/uchardet/script/langs/sv.py +56 -0
- data/ext/uchardet/script/langs/th.py +55 -0
- data/ext/uchardet/script/langs/tr.py +67 -0
- data/ext/uchardet/script/langs/vi.py +64 -0
- data/ext/uchardet/script/release.sh +8 -0
- data/ext/uchardet/script/win32.sh +7 -0
- data/ext/uchardet/src/Big5Freq.tab +943 -0
- data/ext/uchardet/src/CMakeLists.txt +160 -0
- data/ext/uchardet/src/CharDistribution.cpp +109 -0
- data/ext/uchardet/src/CharDistribution.h +242 -0
- data/ext/uchardet/src/EUCKRFreq.tab +614 -0
- data/ext/uchardet/src/EUCTWFreq.tab +447 -0
- data/ext/uchardet/src/GB2312Freq.tab +491 -0
- data/ext/uchardet/src/JISFreq.tab +589 -0
- data/ext/uchardet/src/JpCntx.cpp +230 -0
- data/ext/uchardet/src/JpCntx.h +140 -0
- data/ext/uchardet/src/LangModels/LangArabicModel.cpp +267 -0
- data/ext/uchardet/src/LangModels/LangBulgarianModel.cpp +249 -0
- data/ext/uchardet/src/LangModels/LangCroatianModel.cpp +298 -0
- data/ext/uchardet/src/LangModels/LangCzechModel.cpp +285 -0
- data/ext/uchardet/src/LangModels/LangDanishModel.cpp +201 -0
- data/ext/uchardet/src/LangModels/LangEsperantoModel.cpp +142 -0
- data/ext/uchardet/src/LangModels/LangEstonianModel.cpp +268 -0
- data/ext/uchardet/src/LangModels/LangFinnishModel.cpp +297 -0
- data/ext/uchardet/src/LangModels/LangFrenchModel.cpp +209 -0
- data/ext/uchardet/src/LangModels/LangGermanModel.cpp +170 -0
- data/ext/uchardet/src/LangModels/LangGreekModel.cpp +231 -0
- data/ext/uchardet/src/LangModels/LangHebrewModel.cpp +220 -0
- data/ext/uchardet/src/LangModels/LangHungarianModel.cpp +171 -0
- data/ext/uchardet/src/LangModels/LangIrishModel.cpp +234 -0
- data/ext/uchardet/src/LangModels/LangItalianModel.cpp +269 -0
- data/ext/uchardet/src/LangModels/LangLatvianModel.cpp +210 -0
- data/ext/uchardet/src/LangModels/LangLithuanianModel.cpp +209 -0
- data/ext/uchardet/src/LangModels/LangMalteseModel.cpp +138 -0
- data/ext/uchardet/src/LangModels/LangPolishModel.cpp +304 -0
- data/ext/uchardet/src/LangModels/LangPortugueseModel.cpp +241 -0
- data/ext/uchardet/src/LangModels/LangRomanianModel.cpp +236 -0
- data/ext/uchardet/src/LangModels/LangRussianModel.cpp +367 -0
- data/ext/uchardet/src/LangModels/LangSlovakModel.cpp +293 -0
- data/ext/uchardet/src/LangModels/LangSloveneModel.cpp +264 -0
- data/ext/uchardet/src/LangModels/LangSpanishModel.cpp +204 -0
- data/ext/uchardet/src/LangModels/LangSwedishModel.cpp +266 -0
- data/ext/uchardet/src/LangModels/LangThaiModel.cpp +267 -0
- data/ext/uchardet/src/LangModels/LangTurkishModel.cpp +175 -0
- data/ext/uchardet/src/LangModels/LangVietnameseModel.cpp +249 -0
- data/ext/uchardet/src/nsBig5Prober.cpp +88 -0
- data/ext/uchardet/src/nsBig5Prober.h +75 -0
- data/ext/uchardet/src/nsCharSetProber.cpp +125 -0
- data/ext/uchardet/src/nsCharSetProber.h +77 -0
- data/ext/uchardet/src/nsCodingStateMachine.h +105 -0
- data/ext/uchardet/src/nsEUCJPProber.cpp +99 -0
- data/ext/uchardet/src/nsEUCJPProber.h +79 -0
- data/ext/uchardet/src/nsEUCKRProber.cpp +91 -0
- data/ext/uchardet/src/nsEUCKRProber.h +81 -0
- data/ext/uchardet/src/nsEUCTWProber.cpp +91 -0
- data/ext/uchardet/src/nsEUCTWProber.h +75 -0
- data/ext/uchardet/src/nsEscCharsetProber.cpp +101 -0
- data/ext/uchardet/src/nsEscCharsetProber.h +70 -0
- data/ext/uchardet/src/nsEscSM.cpp +267 -0
- data/ext/uchardet/src/nsGB2312Prober.cpp +96 -0
- data/ext/uchardet/src/nsGB2312Prober.h +77 -0
- data/ext/uchardet/src/nsHebrewProber.cpp +194 -0
- data/ext/uchardet/src/nsHebrewProber.h +177 -0
- data/ext/uchardet/src/nsLatin1Prober.cpp +182 -0
- data/ext/uchardet/src/nsLatin1Prober.h +73 -0
- data/ext/uchardet/src/nsMBCSGroupProber.cpp +242 -0
- data/ext/uchardet/src/nsMBCSGroupProber.h +81 -0
- data/ext/uchardet/src/nsMBCSSM.cpp +513 -0
- data/ext/uchardet/src/nsPkgInt.h +89 -0
- data/ext/uchardet/src/nsSBCSGroupProber.cpp +343 -0
- data/ext/uchardet/src/nsSBCSGroupProber.h +71 -0
- data/ext/uchardet/src/nsSBCharSetProber.cpp +160 -0
- data/ext/uchardet/src/nsSBCharSetProber.h +258 -0
- data/ext/uchardet/src/nsSJISProber.cpp +98 -0
- data/ext/uchardet/src/nsSJISProber.h +81 -0
- data/ext/uchardet/src/nsUTF8Prober.cpp +87 -0
- data/ext/uchardet/src/nsUTF8Prober.h +66 -0
- data/ext/uchardet/src/nsUniversalDetector.cpp +339 -0
- data/ext/uchardet/src/nsUniversalDetector.h +91 -0
- data/ext/uchardet/src/nscore.h +59 -0
- data/ext/uchardet/src/prmem.h +49 -0
- data/ext/uchardet/src/symbols.cmake +41 -0
- data/ext/uchardet/src/tools/CMakeLists.txt +23 -0
- data/ext/uchardet/src/tools/uchardet.cpp +254 -0
- data/ext/uchardet/src/uchardet.cpp +274 -0
- data/ext/uchardet/src/uchardet.h +136 -0
- data/ext/uchardet/test/CMakeLists.txt +47 -0
- data/ext/uchardet/test/ar/iso-8859-6.txt +3 -0
- data/ext/uchardet/test/ar/utf-8.txt +3 -0
- data/ext/uchardet/test/ar/windows-1256.txt +3 -0
- data/ext/uchardet/test/bg/windows-1251.txt +3 -0
- data/ext/uchardet/test/cs/ibm852.txt +4 -0
- data/ext/uchardet/test/cs/iso-8859-2.txt +4 -0
- data/ext/uchardet/test/cs/mac-centraleurope.txt +4 -0
- data/ext/uchardet/test/cs/utf-8.txt +4 -0
- data/ext/uchardet/test/cs/windows-1250.txt +4 -0
- data/ext/uchardet/test/da/iso-8859-1.txt +7 -0
- data/ext/uchardet/test/da/iso-8859-15.txt +10 -0
- data/ext/uchardet/test/da/utf-8.txt +10 -0
- data/ext/uchardet/test/da/windows-1252.txt +10 -0
- data/ext/uchardet/test/de/iso-8859-1.txt +11 -0
- data/ext/uchardet/test/de/windows-1252.txt +11 -0
- data/ext/uchardet/test/el/iso-8859-7.txt +3 -0
- data/ext/uchardet/test/el/utf-8.txt +3 -0
- data/ext/uchardet/test/el/windows-1253.txt +5 -0
- data/ext/uchardet/test/en/ascii.txt +4 -0
- data/ext/uchardet/test/eo/iso-8859-3.txt +7 -0
- data/ext/uchardet/test/es/iso-8859-1.txt +5 -0
- data/ext/uchardet/test/es/iso-8859-15.txt +5 -0
- data/ext/uchardet/test/es/utf-8.txt +5 -0
- data/ext/uchardet/test/es/windows-1252.txt +5 -0
- data/ext/uchardet/test/et/iso-8859-13.txt +6 -0
- data/ext/uchardet/test/et/iso-8859-15.txt +6 -0
- data/ext/uchardet/test/et/iso-8859-4.txt +6 -0
- data/ext/uchardet/test/et/utf-8.txt +6 -0
- data/ext/uchardet/test/et/windows-1252.txt +6 -0
- data/ext/uchardet/test/et/windows-1257.txt +6 -0
- data/ext/uchardet/test/fi/iso-8859-1.txt +8 -0
- data/ext/uchardet/test/fi/utf-8.txt +8 -0
- data/ext/uchardet/test/fr/iso-8859-1.txt +5 -0
- data/ext/uchardet/test/fr/iso-8859-15.txt +16 -0
- data/ext/uchardet/test/fr/utf-16.be +0 -0
- data/ext/uchardet/test/fr/utf-32.le +0 -0
- data/ext/uchardet/test/fr/utf-8.txt +14 -0
- data/ext/uchardet/test/fr/windows-1252.txt +3 -0
- data/ext/uchardet/test/ga/iso-8859-1.txt +6 -0
- data/ext/uchardet/test/ga/utf-8.txt +6 -0
- data/ext/uchardet/test/ga/windows-1252.txt +6 -0
- data/ext/uchardet/test/he/iso-8859-8.txt +2 -0
- data/ext/uchardet/test/he/utf-8.txt +3 -0
- data/ext/uchardet/test/he/windows-1255.txt +1 -0
- data/ext/uchardet/test/hr/ibm852.txt +4 -0
- data/ext/uchardet/test/hr/iso-8859-13.txt +4 -0
- data/ext/uchardet/test/hr/iso-8859-16.txt +4 -0
- data/ext/uchardet/test/hr/iso-8859-2.txt +4 -0
- data/ext/uchardet/test/hr/mac-centraleurope.txt +4 -0
- data/ext/uchardet/test/hr/utf-8.txt +4 -0
- data/ext/uchardet/test/hr/windows-1250.txt +4 -0
- data/ext/uchardet/test/hu/iso-8859-2.txt +3 -0
- data/ext/uchardet/test/hu/windows-1250.txt +1 -0
- data/ext/uchardet/test/it/iso-8859-1.txt +18 -0
- data/ext/uchardet/test/it/utf-8.txt +18 -0
- data/ext/uchardet/test/ja/euc-jp.txt +10 -0
- data/ext/uchardet/test/ja/iso-2022-jp.txt +8 -0
- data/ext/uchardet/test/ja/shift_jis.txt +1 -0
- data/ext/uchardet/test/ja/utf-16be.txt +0 -0
- data/ext/uchardet/test/ja/utf-16le.txt +0 -0
- data/ext/uchardet/test/ja/utf-8.txt +9 -0
- data/ext/uchardet/test/ko/iso-2022-kr.txt +8 -0
- data/ext/uchardet/test/ko/uhc.smi +16 -0
- data/ext/uchardet/test/ko/utf-16.le +0 -0
- data/ext/uchardet/test/ko/utf-32.be +0 -0
- data/ext/uchardet/test/ko/utf-8.txt +3 -0
- data/ext/uchardet/test/lt/iso-8859-10.txt +3 -0
- data/ext/uchardet/test/lt/iso-8859-13.txt +3 -0
- data/ext/uchardet/test/lt/iso-8859-4.txt +3 -0
- data/ext/uchardet/test/lt/utf-8.txt +3 -0
- data/ext/uchardet/test/lv/iso-8859-10.txt +6 -0
- data/ext/uchardet/test/lv/iso-8859-13.txt +6 -0
- data/ext/uchardet/test/lv/iso-8859-4.txt +6 -0
- data/ext/uchardet/test/lv/utf-8.txt +6 -0
- data/ext/uchardet/test/mt/iso-8859-3.txt +4 -0
- data/ext/uchardet/test/mt/utf-8.txt +4 -0
- data/ext/uchardet/test/pl/ibm852.txt +3 -0
- data/ext/uchardet/test/pl/iso-8859-13.txt +3 -0
- data/ext/uchardet/test/pl/iso-8859-16.txt +3 -0
- data/ext/uchardet/test/pl/iso-8859-2.txt +3 -0
- data/ext/uchardet/test/pl/mac-centraleurope.txt +3 -0
- data/ext/uchardet/test/pl/utf-8.txt +3 -0
- data/ext/uchardet/test/pl/windows-1250.txt +3 -0
- data/ext/uchardet/test/pt/iso-8859-1.txt +6 -0
- data/ext/uchardet/test/pt/utf-8.txt +6 -0
- data/ext/uchardet/test/ro/ibm852.txt +9 -0
- data/ext/uchardet/test/ro/iso-8859-16.txt +9 -0
- data/ext/uchardet/test/ro/utf-8.txt +9 -0
- data/ext/uchardet/test/ro/windows-1250.txt +9 -0
- data/ext/uchardet/test/ru/ibm855.txt +5 -0
- data/ext/uchardet/test/ru/ibm866.txt +11 -0
- data/ext/uchardet/test/ru/iso-8859-5.txt +3 -0
- data/ext/uchardet/test/ru/koi8-r.txt +1 -0
- data/ext/uchardet/test/ru/mac-cyrillic.txt +9 -0
- data/ext/uchardet/test/ru/windows-1251.txt +4 -0
- data/ext/uchardet/test/sk/ibm852.txt +3 -0
- data/ext/uchardet/test/sk/iso-8859-2.txt +3 -0
- data/ext/uchardet/test/sk/mac-centraleurope.txt +3 -0
- data/ext/uchardet/test/sk/utf-8.txt +3 -0
- data/ext/uchardet/test/sk/windows-1250.txt +3 -0
- data/ext/uchardet/test/sl/ibm852.txt +9 -0
- data/ext/uchardet/test/sl/iso-8859-16.txt +9 -0
- data/ext/uchardet/test/sl/iso-8859-2.txt +9 -0
- data/ext/uchardet/test/sl/mac-centraleurope.txt +9 -0
- data/ext/uchardet/test/sl/utf-8.txt +9 -0
- data/ext/uchardet/test/sl/windows-1250.txt +9 -0
- data/ext/uchardet/test/sv/iso-8859-1.txt +10 -0
- data/ext/uchardet/test/sv/utf-8.txt +10 -0
- data/ext/uchardet/test/sv/windows-1252.txt +10 -0
- data/ext/uchardet/test/th/iso-8859-11.txt +5 -0
- data/ext/uchardet/test/th/tis-620.txt +5 -0
- data/ext/uchardet/test/th/utf-8.txt +1 -0
- data/ext/uchardet/test/tr/iso-8859-3.txt +13 -0
- data/ext/uchardet/test/tr/iso-8859-9.txt +13 -0
- data/ext/uchardet/test/uchardet-tests.c +130 -0
- data/ext/uchardet/test/vi/utf-8.txt +4 -0
- data/ext/uchardet/test/vi/viscii.txt +4 -0
- data/ext/uchardet/test/vi/windows-1258.txt +4 -0
- data/ext/uchardet/test/zh/big5.txt +1 -0
- data/ext/uchardet/test/zh/euc-tw.txt +1 -0
- data/ext/uchardet/test/zh/gb18030.txt +1 -0
- data/ext/uchardet/test/zh/utf-8.txt +1 -0
- data/ext/uchardet/uchardet.doap +51 -0
- data/ext/uchardet/uchardet.pc.in +10 -0
- data/lib/cchardet.rb +56 -0
- data/lib/cchardet/lib_finder.rb +32 -0
- data/lib/cchardet/version.rb +5 -0
- metadata +362 -0
@@ -0,0 +1,73 @@
|
|
1
|
+
/* -*- Mode: C; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
|
2
|
+
/* ***** BEGIN LICENSE BLOCK *****
|
3
|
+
* Version: MPL 1.1/GPL 2.0/LGPL 2.1
|
4
|
+
*
|
5
|
+
* The contents of this file are subject to the Mozilla Public License Version
|
6
|
+
* 1.1 (the "License"); you may not use this file except in compliance with
|
7
|
+
* the License. You may obtain a copy of the License at
|
8
|
+
* http://www.mozilla.org/MPL/
|
9
|
+
*
|
10
|
+
* Software distributed under the License is distributed on an "AS IS" basis,
|
11
|
+
* WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License
|
12
|
+
* for the specific language governing rights and limitations under the
|
13
|
+
* License.
|
14
|
+
*
|
15
|
+
* The Original Code is Mozilla Universal charset detector code.
|
16
|
+
*
|
17
|
+
* The Initial Developer of the Original Code is
|
18
|
+
* Netscape Communications Corporation.
|
19
|
+
* Portions created by the Initial Developer are Copyright (C) 2001
|
20
|
+
* the Initial Developer. All Rights Reserved.
|
21
|
+
*
|
22
|
+
* Contributor(s):
|
23
|
+
* Shy Shalom <shooshX@gmail.com>
|
24
|
+
*
|
25
|
+
* Alternatively, the contents of this file may be used under the terms of
|
26
|
+
* either the GNU General Public License Version 2 or later (the "GPL"), or
|
27
|
+
* the GNU Lesser General Public License Version 2.1 or later (the "LGPL"),
|
28
|
+
* in which case the provisions of the GPL or the LGPL are applicable instead
|
29
|
+
* of those above. If you wish to allow use of your version of this file only
|
30
|
+
* under the terms of either the GPL or the LGPL, and not to allow others to
|
31
|
+
* use your version of this file under the terms of the MPL, indicate your
|
32
|
+
* decision by deleting the provisions above and replace them with the notice
|
33
|
+
* and other provisions required by the GPL or the LGPL. If you do not delete
|
34
|
+
* the provisions above, a recipient may use your version of this file under
|
35
|
+
* the terms of any one of the MPL, the GPL or the LGPL.
|
36
|
+
*
|
37
|
+
* ***** END LICENSE BLOCK ***** */
|
38
|
+
|
39
|
+
#ifndef nsLatin1Prober_h__
|
40
|
+
#define nsLatin1Prober_h__
|
41
|
+
|
42
|
+
#include <cstddef>
|
43
|
+
|
44
|
+
#include "nsCharSetProber.h"
|
45
|
+
|
46
|
+
#define FREQ_CAT_NUM 4
|
47
|
+
|
48
|
+
class nsLatin1Prober: public nsCharSetProber {
|
49
|
+
public:
|
50
|
+
nsLatin1Prober(void){Reset();}
|
51
|
+
virtual ~nsLatin1Prober(void){}
|
52
|
+
nsProbingState HandleData(const char* aBuf, PRUint32 aLen);
|
53
|
+
const char* GetCharSetName() {return "WINDOWS-1252";}
|
54
|
+
const char* GetLanguage() {return NULL;}
|
55
|
+
nsProbingState GetState(void) {return mState;}
|
56
|
+
void Reset(void);
|
57
|
+
float GetConfidence(void);
|
58
|
+
void SetOpion() {}
|
59
|
+
|
60
|
+
#ifdef DEBUG_chardet
|
61
|
+
virtual void DumpStatus();
|
62
|
+
#endif
|
63
|
+
|
64
|
+
protected:
|
65
|
+
|
66
|
+
nsProbingState mState;
|
67
|
+
char mLastCharClass;
|
68
|
+
PRUint32 mFreqCounter[FREQ_CAT_NUM];
|
69
|
+
};
|
70
|
+
|
71
|
+
|
72
|
+
#endif /* nsLatin1Prober_h__ */
|
73
|
+
|
@@ -0,0 +1,242 @@
|
|
1
|
+
/* -*- Mode: C; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
|
2
|
+
/* ***** BEGIN LICENSE BLOCK *****
|
3
|
+
* Version: MPL 1.1/GPL 2.0/LGPL 2.1
|
4
|
+
*
|
5
|
+
* The contents of this file are subject to the Mozilla Public License Version
|
6
|
+
* 1.1 (the "License"); you may not use this file except in compliance with
|
7
|
+
* the License. You may obtain a copy of the License at
|
8
|
+
* http://www.mozilla.org/MPL/
|
9
|
+
*
|
10
|
+
* Software distributed under the License is distributed on an "AS IS" basis,
|
11
|
+
* WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License
|
12
|
+
* for the specific language governing rights and limitations under the
|
13
|
+
* License.
|
14
|
+
*
|
15
|
+
* The Original Code is Mozilla Universal charset detector code.
|
16
|
+
*
|
17
|
+
* The Initial Developer of the Original Code is
|
18
|
+
* Netscape Communications Corporation.
|
19
|
+
* Portions created by the Initial Developer are Copyright (C) 2001
|
20
|
+
* the Initial Developer. All Rights Reserved.
|
21
|
+
*
|
22
|
+
* Contributor(s):
|
23
|
+
* Shy Shalom <shooshX@gmail.com>
|
24
|
+
* Proofpoint, Inc.
|
25
|
+
*
|
26
|
+
* Alternatively, the contents of this file may be used under the terms of
|
27
|
+
* either the GNU General Public License Version 2 or later (the "GPL"), or
|
28
|
+
* the GNU Lesser General Public License Version 2.1 or later (the "LGPL"),
|
29
|
+
* in which case the provisions of the GPL or the LGPL are applicable instead
|
30
|
+
* of those above. If you wish to allow use of your version of this file only
|
31
|
+
* under the terms of either the GPL or the LGPL, and not to allow others to
|
32
|
+
* use your version of this file under the terms of the MPL, indicate your
|
33
|
+
* decision by deleting the provisions above and replace them with the notice
|
34
|
+
* and other provisions required by the GPL or the LGPL. If you do not delete
|
35
|
+
* the provisions above, a recipient may use your version of this file under
|
36
|
+
* the terms of any one of the MPL, the GPL or the LGPL.
|
37
|
+
*
|
38
|
+
* ***** END LICENSE BLOCK ***** */
|
39
|
+
#include <stdio.h>
|
40
|
+
|
41
|
+
#include "nsMBCSGroupProber.h"
|
42
|
+
#include "nsUniversalDetector.h"
|
43
|
+
|
44
|
+
#if defined(DEBUG_chardet) || defined(DEBUG_jgmyers)
|
45
|
+
const char *ProberName[] =
|
46
|
+
{
|
47
|
+
"UTF-8",
|
48
|
+
"SJIS",
|
49
|
+
"EUC-JP",
|
50
|
+
"GB18030",
|
51
|
+
"EUC-KR",
|
52
|
+
"Big5",
|
53
|
+
"EUC-TW",
|
54
|
+
};
|
55
|
+
|
56
|
+
#endif
|
57
|
+
|
58
|
+
nsMBCSGroupProber::nsMBCSGroupProber(PRUint32 aLanguageFilter)
|
59
|
+
{
|
60
|
+
for (PRUint32 i = 0; i < NUM_OF_PROBERS; i++)
|
61
|
+
mProbers[i] = nsnull;
|
62
|
+
|
63
|
+
mProbers[0] = new nsUTF8Prober();
|
64
|
+
if (aLanguageFilter & NS_FILTER_JAPANESE)
|
65
|
+
{
|
66
|
+
mProbers[1] = new nsSJISProber(aLanguageFilter == NS_FILTER_JAPANESE);
|
67
|
+
mProbers[2] = new nsEUCJPProber(aLanguageFilter == NS_FILTER_JAPANESE);
|
68
|
+
}
|
69
|
+
if (aLanguageFilter & NS_FILTER_CHINESE_SIMPLIFIED)
|
70
|
+
mProbers[3] = new nsGB18030Prober(aLanguageFilter == NS_FILTER_CHINESE_SIMPLIFIED);
|
71
|
+
if (aLanguageFilter & NS_FILTER_KOREAN)
|
72
|
+
mProbers[4] = new nsEUCKRProber(aLanguageFilter == NS_FILTER_KOREAN);
|
73
|
+
if (aLanguageFilter & NS_FILTER_CHINESE_TRADITIONAL)
|
74
|
+
{
|
75
|
+
mProbers[5] = new nsBig5Prober(aLanguageFilter == NS_FILTER_CHINESE_TRADITIONAL);
|
76
|
+
mProbers[6] = new nsEUCTWProber(aLanguageFilter == NS_FILTER_CHINESE_TRADITIONAL);
|
77
|
+
}
|
78
|
+
Reset();
|
79
|
+
}
|
80
|
+
|
81
|
+
nsMBCSGroupProber::~nsMBCSGroupProber()
|
82
|
+
{
|
83
|
+
for (PRUint32 i = 0; i < NUM_OF_PROBERS; i++)
|
84
|
+
{
|
85
|
+
delete mProbers[i];
|
86
|
+
}
|
87
|
+
}
|
88
|
+
|
89
|
+
const char* nsMBCSGroupProber::GetCharSetName()
|
90
|
+
{
|
91
|
+
if (mBestGuess == -1)
|
92
|
+
{
|
93
|
+
GetConfidence();
|
94
|
+
if (mBestGuess == -1)
|
95
|
+
mBestGuess = 0;
|
96
|
+
}
|
97
|
+
return mProbers[mBestGuess]->GetCharSetName();
|
98
|
+
}
|
99
|
+
|
100
|
+
const char* nsMBCSGroupProber::GetLanguage(void)
|
101
|
+
{
|
102
|
+
if (mBestGuess == -1)
|
103
|
+
{
|
104
|
+
GetConfidence();
|
105
|
+
}
|
106
|
+
if (mBestGuess == -1)
|
107
|
+
return NULL;
|
108
|
+
else
|
109
|
+
return mProbers[mBestGuess]->GetLanguage();
|
110
|
+
}
|
111
|
+
|
112
|
+
void nsMBCSGroupProber::Reset(void)
|
113
|
+
{
|
114
|
+
mActiveNum = 0;
|
115
|
+
for (PRUint32 i = 0; i < NUM_OF_PROBERS; i++)
|
116
|
+
{
|
117
|
+
if (mProbers[i])
|
118
|
+
{
|
119
|
+
mProbers[i]->Reset();
|
120
|
+
mIsActive[i] = PR_TRUE;
|
121
|
+
++mActiveNum;
|
122
|
+
}
|
123
|
+
else
|
124
|
+
mIsActive[i] = PR_FALSE;
|
125
|
+
}
|
126
|
+
mBestGuess = -1;
|
127
|
+
mState = eDetecting;
|
128
|
+
mKeepNext = 0;
|
129
|
+
}
|
130
|
+
|
131
|
+
nsProbingState nsMBCSGroupProber::HandleData(const char* aBuf, PRUint32 aLen)
|
132
|
+
{
|
133
|
+
nsProbingState st;
|
134
|
+
PRUint32 start = 0;
|
135
|
+
PRUint32 keepNext = mKeepNext;
|
136
|
+
|
137
|
+
//do filtering to reduce load to probers
|
138
|
+
for (PRUint32 pos = 0; pos < aLen; ++pos)
|
139
|
+
{
|
140
|
+
if (aBuf[pos] & 0x80)
|
141
|
+
{
|
142
|
+
if (!keepNext)
|
143
|
+
start = pos;
|
144
|
+
keepNext = 2;
|
145
|
+
}
|
146
|
+
else if (keepNext)
|
147
|
+
{
|
148
|
+
if (--keepNext == 0)
|
149
|
+
{
|
150
|
+
for (PRUint32 i = 0; i < NUM_OF_PROBERS; i++)
|
151
|
+
{
|
152
|
+
if (!mIsActive[i])
|
153
|
+
continue;
|
154
|
+
st = mProbers[i]->HandleData(aBuf + start, pos + 1 - start);
|
155
|
+
if (st == eFoundIt)
|
156
|
+
{
|
157
|
+
mBestGuess = i;
|
158
|
+
mState = eFoundIt;
|
159
|
+
return mState;
|
160
|
+
}
|
161
|
+
}
|
162
|
+
}
|
163
|
+
}
|
164
|
+
}
|
165
|
+
|
166
|
+
if (keepNext) {
|
167
|
+
for (PRUint32 i = 0; i < NUM_OF_PROBERS; i++)
|
168
|
+
{
|
169
|
+
if (!mIsActive[i])
|
170
|
+
continue;
|
171
|
+
st = mProbers[i]->HandleData(aBuf + start, aLen - start);
|
172
|
+
if (st == eFoundIt)
|
173
|
+
{
|
174
|
+
mBestGuess = i;
|
175
|
+
mState = eFoundIt;
|
176
|
+
return mState;
|
177
|
+
}
|
178
|
+
}
|
179
|
+
}
|
180
|
+
mKeepNext = keepNext;
|
181
|
+
|
182
|
+
return mState;
|
183
|
+
}
|
184
|
+
|
185
|
+
float nsMBCSGroupProber::GetConfidence(void)
|
186
|
+
{
|
187
|
+
PRUint32 i;
|
188
|
+
float bestConf = 0.0, cf;
|
189
|
+
|
190
|
+
switch (mState)
|
191
|
+
{
|
192
|
+
case eFoundIt:
|
193
|
+
return (float)0.99;
|
194
|
+
case eNotMe:
|
195
|
+
return (float)0.01;
|
196
|
+
default:
|
197
|
+
for (i = 0; i < NUM_OF_PROBERS; i++)
|
198
|
+
{
|
199
|
+
if (!mIsActive[i])
|
200
|
+
continue;
|
201
|
+
cf = mProbers[i]->GetConfidence();
|
202
|
+
if (bestConf < cf)
|
203
|
+
{
|
204
|
+
bestConf = cf;
|
205
|
+
mBestGuess = i;
|
206
|
+
}
|
207
|
+
}
|
208
|
+
}
|
209
|
+
return bestConf;
|
210
|
+
}
|
211
|
+
|
212
|
+
#ifdef DEBUG_chardet
|
213
|
+
void nsMBCSGroupProber::DumpStatus()
|
214
|
+
{
|
215
|
+
PRUint32 i;
|
216
|
+
float cf;
|
217
|
+
|
218
|
+
GetConfidence();
|
219
|
+
for (i = 0; i < NUM_OF_PROBERS; i++)
|
220
|
+
{
|
221
|
+
if (!mIsActive[i])
|
222
|
+
printf(" MBCS inactive: [%s] (confidence is too low).\r\n", ProberName[i]);
|
223
|
+
else
|
224
|
+
{
|
225
|
+
cf = mProbers[i]->GetConfidence();
|
226
|
+
printf(" MBCS %1.3f: [%s]\r\n", cf, ProberName[i]);
|
227
|
+
}
|
228
|
+
}
|
229
|
+
}
|
230
|
+
#endif
|
231
|
+
|
232
|
+
#ifdef DEBUG_jgmyers
|
233
|
+
void nsMBCSGroupProber::GetDetectorState(nsUniversalDetector::DetectorState (&states)[nsUniversalDetector::NumDetectors], PRUint32 &offset)
|
234
|
+
{
|
235
|
+
for (PRUint32 i = 0; i < NUM_OF_PROBERS; ++i) {
|
236
|
+
states[offset].name = ProberName[i];
|
237
|
+
states[offset].isActive = mIsActive[i];
|
238
|
+
states[offset].confidence = mIsActive[i] ? mProbers[i]->GetConfidence() : 0.0;
|
239
|
+
++offset;
|
240
|
+
}
|
241
|
+
}
|
242
|
+
#endif /* DEBUG_jgmyers */
|
@@ -0,0 +1,81 @@
|
|
1
|
+
/* -*- Mode: C; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
|
2
|
+
/* ***** BEGIN LICENSE BLOCK *****
|
3
|
+
* Version: MPL 1.1/GPL 2.0/LGPL 2.1
|
4
|
+
*
|
5
|
+
* The contents of this file are subject to the Mozilla Public License Version
|
6
|
+
* 1.1 (the "License"); you may not use this file except in compliance with
|
7
|
+
* the License. You may obtain a copy of the License at
|
8
|
+
* http://www.mozilla.org/MPL/
|
9
|
+
*
|
10
|
+
* Software distributed under the License is distributed on an "AS IS" basis,
|
11
|
+
* WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License
|
12
|
+
* for the specific language governing rights and limitations under the
|
13
|
+
* License.
|
14
|
+
*
|
15
|
+
* The Original Code is mozilla.org code.
|
16
|
+
*
|
17
|
+
* The Initial Developer of the Original Code is
|
18
|
+
* Netscape Communications Corporation.
|
19
|
+
* Portions created by the Initial Developer are Copyright (C) 1998
|
20
|
+
* the Initial Developer. All Rights Reserved.
|
21
|
+
*
|
22
|
+
* Contributor(s):
|
23
|
+
* Proofpoint, Inc.
|
24
|
+
*
|
25
|
+
* Alternatively, the contents of this file may be used under the terms of
|
26
|
+
* either the GNU General Public License Version 2 or later (the "GPL"), or
|
27
|
+
* the GNU Lesser General Public License Version 2.1 or later (the "LGPL"),
|
28
|
+
* in which case the provisions of the GPL or the LGPL are applicable instead
|
29
|
+
* of those above. If you wish to allow use of your version of this file only
|
30
|
+
* under the terms of either the GPL or the LGPL, and not to allow others to
|
31
|
+
* use your version of this file under the terms of the MPL, indicate your
|
32
|
+
* decision by deleting the provisions above and replace them with the notice
|
33
|
+
* and other provisions required by the GPL or the LGPL. If you do not delete
|
34
|
+
* the provisions above, a recipient may use your version of this file under
|
35
|
+
* the terms of any one of the MPL, the GPL or the LGPL.
|
36
|
+
*
|
37
|
+
* ***** END LICENSE BLOCK ***** */
|
38
|
+
|
39
|
+
#ifndef nsMBCSGroupProber_h__
|
40
|
+
#define nsMBCSGroupProber_h__
|
41
|
+
|
42
|
+
#include "nsSJISProber.h"
|
43
|
+
#include "nsUTF8Prober.h"
|
44
|
+
#include "nsEUCJPProber.h"
|
45
|
+
#include "nsGB2312Prober.h"
|
46
|
+
#include "nsEUCKRProber.h"
|
47
|
+
#include "nsBig5Prober.h"
|
48
|
+
#include "nsEUCTWProber.h"
|
49
|
+
|
50
|
+
#define NUM_OF_PROBERS 7
|
51
|
+
|
52
|
+
class nsMBCSGroupProber: public nsCharSetProber {
|
53
|
+
public:
|
54
|
+
nsMBCSGroupProber(PRUint32 aLanguageFilter);
|
55
|
+
virtual ~nsMBCSGroupProber();
|
56
|
+
nsProbingState HandleData(const char* aBuf, PRUint32 aLen);
|
57
|
+
const char* GetCharSetName();
|
58
|
+
const char* GetLanguage();
|
59
|
+
nsProbingState GetState(void) {return mState;}
|
60
|
+
void Reset(void);
|
61
|
+
float GetConfidence(void);
|
62
|
+
void SetOpion() {}
|
63
|
+
|
64
|
+
#ifdef DEBUG_chardet
|
65
|
+
void DumpStatus();
|
66
|
+
#endif
|
67
|
+
#ifdef DEBUG_jgmyers
|
68
|
+
void GetDetectorState(nsUniversalDetector::DetectorState (&states)[nsUniversalDetector::NumDetectors], PRUint32 &offset);
|
69
|
+
#endif
|
70
|
+
|
71
|
+
protected:
|
72
|
+
nsProbingState mState;
|
73
|
+
nsCharSetProber* mProbers[NUM_OF_PROBERS];
|
74
|
+
PRBool mIsActive[NUM_OF_PROBERS];
|
75
|
+
PRInt32 mBestGuess;
|
76
|
+
PRUint32 mActiveNum;
|
77
|
+
PRUint32 mKeepNext;
|
78
|
+
};
|
79
|
+
|
80
|
+
#endif /* nsMBCSGroupProber_h__ */
|
81
|
+
|
@@ -0,0 +1,513 @@
|
|
1
|
+
/* -*- Mode: C; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
|
2
|
+
/* ***** BEGIN LICENSE BLOCK *****
|
3
|
+
* Version: MPL 1.1/GPL 2.0/LGPL 2.1
|
4
|
+
*
|
5
|
+
* The contents of this file are subject to the Mozilla Public License Version
|
6
|
+
* 1.1 (the "License"); you may not use this file except in compliance with
|
7
|
+
* the License. You may obtain a copy of the License at
|
8
|
+
* http://www.mozilla.org/MPL/
|
9
|
+
*
|
10
|
+
* Software distributed under the License is distributed on an "AS IS" basis,
|
11
|
+
* WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License
|
12
|
+
* for the specific language governing rights and limitations under the
|
13
|
+
* License.
|
14
|
+
*
|
15
|
+
* The Original Code is mozilla.org code.
|
16
|
+
*
|
17
|
+
* The Initial Developer of the Original Code is
|
18
|
+
* Netscape Communications Corporation.
|
19
|
+
* Portions created by the Initial Developer are Copyright (C) 1998
|
20
|
+
* the Initial Developer. All Rights Reserved.
|
21
|
+
*
|
22
|
+
* Contributor(s):
|
23
|
+
*
|
24
|
+
* Alternatively, the contents of this file may be used under the terms of
|
25
|
+
* either the GNU General Public License Version 2 or later (the "GPL"), or
|
26
|
+
* the GNU Lesser General Public License Version 2.1 or later (the "LGPL"),
|
27
|
+
* in which case the provisions of the GPL or the LGPL are applicable instead
|
28
|
+
* of those above. If you wish to allow use of your version of this file only
|
29
|
+
* under the terms of either the GPL or the LGPL, and not to allow others to
|
30
|
+
* use your version of this file under the terms of the MPL, indicate your
|
31
|
+
* decision by deleting the provisions above and replace them with the notice
|
32
|
+
* and other provisions required by the GPL or the LGPL. If you do not delete
|
33
|
+
* the provisions above, a recipient may use your version of this file under
|
34
|
+
* the terms of any one of the MPL, the GPL or the LGPL.
|
35
|
+
*
|
36
|
+
* ***** END LICENSE BLOCK ***** */
|
37
|
+
#include "nsCodingStateMachine.h"
|
38
|
+
|
39
|
+
/*
|
40
|
+
Modification from frank tang's original work:
|
41
|
+
. 0x00 is allowed as a legal character. Since some web pages contains this char in
|
42
|
+
text stream.
|
43
|
+
*/
|
44
|
+
|
45
|
+
// BIG5
|
46
|
+
|
47
|
+
static const PRUint32 BIG5_cls [ 256 / 8 ] = {
|
48
|
+
//PCK4BITS(0,1,1,1,1,1,1,1), // 00 - 07
|
49
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 00 - 07 //allow 0x00 as legal value
|
50
|
+
PCK4BITS(1,1,1,1,1,1,0,0), // 08 - 0f
|
51
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 10 - 17
|
52
|
+
PCK4BITS(1,1,1,0,1,1,1,1), // 18 - 1f
|
53
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 20 - 27
|
54
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 28 - 2f
|
55
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 30 - 37
|
56
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 38 - 3f
|
57
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 40 - 47
|
58
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 48 - 4f
|
59
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 50 - 57
|
60
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 58 - 5f
|
61
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 60 - 67
|
62
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 68 - 6f
|
63
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 70 - 77
|
64
|
+
PCK4BITS(2,2,2,2,2,2,2,1), // 78 - 7f
|
65
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 80 - 87
|
66
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 88 - 8f
|
67
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 90 - 97
|
68
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 98 - 9f
|
69
|
+
PCK4BITS(4,3,3,3,3,3,3,3), // a0 - a7
|
70
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // a8 - af
|
71
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // b0 - b7
|
72
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // b8 - bf
|
73
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // c0 - c7
|
74
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // c8 - cf
|
75
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // d0 - d7
|
76
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // d8 - df
|
77
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // e0 - e7
|
78
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // e8 - ef
|
79
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // f0 - f7
|
80
|
+
PCK4BITS(3,3,3,3,3,3,3,0) // f8 - ff
|
81
|
+
};
|
82
|
+
|
83
|
+
|
84
|
+
static const PRUint32 BIG5_st [ 3] = {
|
85
|
+
PCK4BITS(eError,eStart,eStart, 3,eError,eError,eError,eError),//00-07
|
86
|
+
PCK4BITS(eError,eError,eItsMe,eItsMe,eItsMe,eItsMe,eItsMe,eError),//08-0f
|
87
|
+
PCK4BITS(eError,eStart,eStart,eStart,eStart,eStart,eStart,eStart) //10-17
|
88
|
+
};
|
89
|
+
|
90
|
+
static const PRUint32 Big5CharLenTable[] = {0, 1, 1, 2, 0};
|
91
|
+
|
92
|
+
SMModel const Big5SMModel = {
|
93
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, BIG5_cls },
|
94
|
+
5,
|
95
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, BIG5_st },
|
96
|
+
Big5CharLenTable,
|
97
|
+
"BIG5",
|
98
|
+
};
|
99
|
+
|
100
|
+
static const PRUint32 EUCJP_cls [ 256 / 8 ] = {
|
101
|
+
//PCK4BITS(5,4,4,4,4,4,4,4), // 00 - 07
|
102
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 00 - 07
|
103
|
+
PCK4BITS(4,4,4,4,4,4,5,5), // 08 - 0f
|
104
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 10 - 17
|
105
|
+
PCK4BITS(4,4,4,5,4,4,4,4), // 18 - 1f
|
106
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 20 - 27
|
107
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 28 - 2f
|
108
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 30 - 37
|
109
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 38 - 3f
|
110
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 40 - 47
|
111
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 48 - 4f
|
112
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 50 - 57
|
113
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 58 - 5f
|
114
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 60 - 67
|
115
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 68 - 6f
|
116
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 70 - 77
|
117
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 78 - 7f
|
118
|
+
PCK4BITS(5,5,5,5,5,5,5,5), // 80 - 87
|
119
|
+
PCK4BITS(5,5,5,5,5,5,1,3), // 88 - 8f
|
120
|
+
PCK4BITS(5,5,5,5,5,5,5,5), // 90 - 97
|
121
|
+
PCK4BITS(5,5,5,5,5,5,5,5), // 98 - 9f
|
122
|
+
PCK4BITS(5,2,2,2,2,2,2,2), // a0 - a7
|
123
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // a8 - af
|
124
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // b0 - b7
|
125
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // b8 - bf
|
126
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // c0 - c7
|
127
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // c8 - cf
|
128
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // d0 - d7
|
129
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // d8 - df
|
130
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // e0 - e7
|
131
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // e8 - ef
|
132
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // f0 - f7
|
133
|
+
PCK4BITS(0,0,0,0,0,0,0,5) // f8 - ff
|
134
|
+
};
|
135
|
+
|
136
|
+
|
137
|
+
static const PRUint32 EUCJP_st [ 5] = {
|
138
|
+
PCK4BITS( 3, 4, 3, 5,eStart,eError,eError,eError),//00-07
|
139
|
+
PCK4BITS(eError,eError,eError,eError,eItsMe,eItsMe,eItsMe,eItsMe),//08-0f
|
140
|
+
PCK4BITS(eItsMe,eItsMe,eStart,eError,eStart,eError,eError,eError),//10-17
|
141
|
+
PCK4BITS(eError,eError,eStart,eError,eError,eError, 3,eError),//18-1f
|
142
|
+
PCK4BITS( 3,eError,eError,eError,eStart,eStart,eStart,eStart) //20-27
|
143
|
+
};
|
144
|
+
|
145
|
+
static const PRUint32 EUCJPCharLenTable[] = {2, 2, 2, 3, 1, 0};
|
146
|
+
|
147
|
+
const SMModel EUCJPSMModel = {
|
148
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, EUCJP_cls },
|
149
|
+
6,
|
150
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, EUCJP_st },
|
151
|
+
EUCJPCharLenTable,
|
152
|
+
"EUC-JP",
|
153
|
+
};
|
154
|
+
|
155
|
+
static const PRUint32 EUCKR_cls [ 256 / 8 ] = {
|
156
|
+
//PCK4BITS(0,1,1,1,1,1,1,1), // 00 - 07
|
157
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 00 - 07
|
158
|
+
PCK4BITS(1,1,1,1,1,1,0,0), // 08 - 0f
|
159
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 10 - 17
|
160
|
+
PCK4BITS(1,1,1,0,1,1,1,1), // 18 - 1f
|
161
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 20 - 27
|
162
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 28 - 2f
|
163
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 30 - 37
|
164
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 38 - 3f
|
165
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 40 - 47
|
166
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 48 - 4f
|
167
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 50 - 57
|
168
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 58 - 5f
|
169
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 60 - 67
|
170
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 68 - 6f
|
171
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 70 - 77
|
172
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 78 - 7f
|
173
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 80 - 87
|
174
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 88 - 8f
|
175
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 90 - 97
|
176
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 98 - 9f
|
177
|
+
PCK4BITS(0,2,2,2,2,2,2,2), // a0 - a7
|
178
|
+
PCK4BITS(2,2,2,2,2,3,3,3), // a8 - af
|
179
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // b0 - b7
|
180
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // b8 - bf
|
181
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // c0 - c7
|
182
|
+
PCK4BITS(2,3,2,2,2,2,2,2), // c8 - cf
|
183
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // d0 - d7
|
184
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // d8 - df
|
185
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // e0 - e7
|
186
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // e8 - ef
|
187
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // f0 - f7
|
188
|
+
PCK4BITS(2,2,2,2,2,2,2,0) // f8 - ff
|
189
|
+
};
|
190
|
+
|
191
|
+
|
192
|
+
static const PRUint32 EUCKR_st [ 2] = {
|
193
|
+
PCK4BITS(eError,eStart, 3,eError,eError,eError,eError,eError),//00-07
|
194
|
+
PCK4BITS(eItsMe,eItsMe,eItsMe,eItsMe,eError,eError,eStart,eStart) //08-0f
|
195
|
+
};
|
196
|
+
|
197
|
+
static const PRUint32 EUCKRCharLenTable[] = {0, 1, 2, 0};
|
198
|
+
|
199
|
+
const SMModel EUCKRSMModel = {
|
200
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, EUCKR_cls },
|
201
|
+
4,
|
202
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, EUCKR_st },
|
203
|
+
EUCKRCharLenTable,
|
204
|
+
"EUC-KR",
|
205
|
+
};
|
206
|
+
|
207
|
+
static const PRUint32 EUCTW_cls [ 256 / 8 ] = {
|
208
|
+
//PCK4BITS(0,2,2,2,2,2,2,2), // 00 - 07
|
209
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 00 - 07
|
210
|
+
PCK4BITS(2,2,2,2,2,2,0,0), // 08 - 0f
|
211
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 10 - 17
|
212
|
+
PCK4BITS(2,2,2,0,2,2,2,2), // 18 - 1f
|
213
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 20 - 27
|
214
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 28 - 2f
|
215
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 30 - 37
|
216
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 38 - 3f
|
217
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 40 - 47
|
218
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 48 - 4f
|
219
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 50 - 57
|
220
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 58 - 5f
|
221
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 60 - 67
|
222
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 68 - 6f
|
223
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 70 - 77
|
224
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 78 - 7f
|
225
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 80 - 87
|
226
|
+
PCK4BITS(0,0,0,0,0,0,6,0), // 88 - 8f
|
227
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 90 - 97
|
228
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 98 - 9f
|
229
|
+
PCK4BITS(0,3,4,4,4,4,4,4), // a0 - a7
|
230
|
+
PCK4BITS(5,5,1,1,1,1,1,1), // a8 - af
|
231
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // b0 - b7
|
232
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // b8 - bf
|
233
|
+
PCK4BITS(1,1,3,1,3,3,3,3), // c0 - c7
|
234
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // c8 - cf
|
235
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // d0 - d7
|
236
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // d8 - df
|
237
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // e0 - e7
|
238
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // e8 - ef
|
239
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // f0 - f7
|
240
|
+
PCK4BITS(3,3,3,3,3,3,3,0) // f8 - ff
|
241
|
+
};
|
242
|
+
|
243
|
+
|
244
|
+
static const PRUint32 EUCTW_st [ 6] = {
|
245
|
+
PCK4BITS(eError,eError,eStart, 3, 3, 3, 4,eError),//00-07
|
246
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eItsMe,eItsMe),//08-0f
|
247
|
+
PCK4BITS(eItsMe,eItsMe,eItsMe,eItsMe,eItsMe,eError,eStart,eError),//10-17
|
248
|
+
PCK4BITS(eStart,eStart,eStart,eError,eError,eError,eError,eError),//18-1f
|
249
|
+
PCK4BITS( 5,eError,eError,eError,eStart,eError,eStart,eStart),//20-27
|
250
|
+
PCK4BITS(eStart,eError,eStart,eStart,eStart,eStart,eStart,eStart) //28-2f
|
251
|
+
};
|
252
|
+
|
253
|
+
static const PRUint32 EUCTWCharLenTable[] = {0, 0, 1, 2, 2, 2, 3};
|
254
|
+
|
255
|
+
const SMModel EUCTWSMModel = {
|
256
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, EUCTW_cls },
|
257
|
+
7,
|
258
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, EUCTW_st },
|
259
|
+
EUCTWCharLenTable,
|
260
|
+
"EUC-TW",
|
261
|
+
};
|
262
|
+
|
263
|
+
/* obsolete GB2312 by GB18030
|
264
|
+
static PRUint32 GB2312_cls [ 256 / 8 ] = {
|
265
|
+
//PCK4BITS(0,1,1,1,1,1,1,1), // 00 - 07
|
266
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 00 - 07
|
267
|
+
PCK4BITS(1,1,1,1,1,1,0,0), // 08 - 0f
|
268
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 10 - 17
|
269
|
+
PCK4BITS(1,1,1,0,1,1,1,1), // 18 - 1f
|
270
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 20 - 27
|
271
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 28 - 2f
|
272
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 30 - 37
|
273
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 38 - 3f
|
274
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 40 - 47
|
275
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 48 - 4f
|
276
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 50 - 57
|
277
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 58 - 5f
|
278
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 60 - 67
|
279
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 68 - 6f
|
280
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 70 - 77
|
281
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 78 - 7f
|
282
|
+
PCK4BITS(1,0,0,0,0,0,0,0), // 80 - 87
|
283
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 88 - 8f
|
284
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 90 - 97
|
285
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 98 - 9f
|
286
|
+
PCK4BITS(0,2,2,2,2,2,2,2), // a0 - a7
|
287
|
+
PCK4BITS(2,2,3,3,3,3,3,3), // a8 - af
|
288
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // b0 - b7
|
289
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // b8 - bf
|
290
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // c0 - c7
|
291
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // c8 - cf
|
292
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // d0 - d7
|
293
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // d8 - df
|
294
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // e0 - e7
|
295
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // e8 - ef
|
296
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // f0 - f7
|
297
|
+
PCK4BITS(2,2,2,2,2,2,2,0) // f8 - ff
|
298
|
+
};
|
299
|
+
|
300
|
+
|
301
|
+
static PRUint32 GB2312_st [ 2] = {
|
302
|
+
PCK4BITS(eError,eStart, 3,eError,eError,eError,eError,eError),//00-07
|
303
|
+
PCK4BITS(eItsMe,eItsMe,eItsMe,eItsMe,eError,eError,eStart,eStart) //08-0f
|
304
|
+
};
|
305
|
+
|
306
|
+
static const PRUint32 GB2312CharLenTable[] = {0, 1, 2, 0};
|
307
|
+
|
308
|
+
SMModel GB2312SMModel = {
|
309
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, GB2312_cls },
|
310
|
+
4,
|
311
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, GB2312_st },
|
312
|
+
GB2312CharLenTable,
|
313
|
+
"GB2312",
|
314
|
+
};
|
315
|
+
*/
|
316
|
+
|
317
|
+
// the following state machine data was created by perl script in
|
318
|
+
// intl/chardet/tools. It should be the same as in PSM detector.
|
319
|
+
static const PRUint32 GB18030_cls [ 256 / 8 ] = {
|
320
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 00 - 07
|
321
|
+
PCK4BITS(1,1,1,1,1,1,0,0), // 08 - 0f
|
322
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 10 - 17
|
323
|
+
PCK4BITS(1,1,1,0,1,1,1,1), // 18 - 1f
|
324
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 20 - 27
|
325
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 28 - 2f
|
326
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // 30 - 37
|
327
|
+
PCK4BITS(3,3,1,1,1,1,1,1), // 38 - 3f
|
328
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 40 - 47
|
329
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 48 - 4f
|
330
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 50 - 57
|
331
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 58 - 5f
|
332
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 60 - 67
|
333
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 68 - 6f
|
334
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 70 - 77
|
335
|
+
PCK4BITS(2,2,2,2,2,2,2,4), // 78 - 7f
|
336
|
+
PCK4BITS(5,6,6,6,6,6,6,6), // 80 - 87
|
337
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // 88 - 8f
|
338
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // 90 - 97
|
339
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // 98 - 9f
|
340
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // a0 - a7
|
341
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // a8 - af
|
342
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // b0 - b7
|
343
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // b8 - bf
|
344
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // c0 - c7
|
345
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // c8 - cf
|
346
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // d0 - d7
|
347
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // d8 - df
|
348
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // e0 - e7
|
349
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // e8 - ef
|
350
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // f0 - f7
|
351
|
+
PCK4BITS(6,6,6,6,6,6,6,0) // f8 - ff
|
352
|
+
};
|
353
|
+
|
354
|
+
|
355
|
+
static const PRUint32 GB18030_st [ 6] = {
|
356
|
+
PCK4BITS(eError,eStart,eStart,eStart,eStart,eStart, 3,eError),//00-07
|
357
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eItsMe,eItsMe),//08-0f
|
358
|
+
PCK4BITS(eItsMe,eItsMe,eItsMe,eItsMe,eItsMe,eError,eError,eStart),//10-17
|
359
|
+
PCK4BITS( 4,eError,eStart,eStart,eError,eError,eError,eError),//18-1f
|
360
|
+
PCK4BITS(eError,eError, 5,eError,eError,eError,eItsMe,eError),//20-27
|
361
|
+
PCK4BITS(eError,eError,eStart,eStart,eStart,eStart,eStart,eStart) //28-2f
|
362
|
+
};
|
363
|
+
|
364
|
+
// To be accurate, the length of class 6 can be either 2 or 4.
|
365
|
+
// But it is not necessary to discriminate between the two since
|
366
|
+
// it is used for frequency analysis only, and we are validing
|
367
|
+
// each code range there as well. So it is safe to set it to be
|
368
|
+
// 2 here.
|
369
|
+
static const PRUint32 GB18030CharLenTable[] = {0, 1, 1, 1, 1, 1, 2};
|
370
|
+
|
371
|
+
const SMModel GB18030SMModel = {
|
372
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, GB18030_cls },
|
373
|
+
7,
|
374
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, GB18030_st },
|
375
|
+
GB18030CharLenTable,
|
376
|
+
"GB18030",
|
377
|
+
};
|
378
|
+
|
379
|
+
// sjis
|
380
|
+
|
381
|
+
static const PRUint32 SJIS_cls [ 256 / 8 ] = {
|
382
|
+
//PCK4BITS(0,1,1,1,1,1,1,1), // 00 - 07
|
383
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 00 - 07
|
384
|
+
PCK4BITS(1,1,1,1,1,1,0,0), // 08 - 0f
|
385
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 10 - 17
|
386
|
+
PCK4BITS(1,1,1,0,1,1,1,1), // 18 - 1f
|
387
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 20 - 27
|
388
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 28 - 2f
|
389
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 30 - 37
|
390
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 38 - 3f
|
391
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 40 - 47
|
392
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 48 - 4f
|
393
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 50 - 57
|
394
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 58 - 5f
|
395
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 60 - 67
|
396
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 68 - 6f
|
397
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 70 - 77
|
398
|
+
PCK4BITS(2,2,2,2,2,2,2,1), // 78 - 7f
|
399
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // 80 - 87
|
400
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // 88 - 8f
|
401
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // 90 - 97
|
402
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // 98 - 9f
|
403
|
+
//0xa0 is illegal in sjis encoding, but some pages does
|
404
|
+
//contain such byte. We need to be more error forgiven.
|
405
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // a0 - a7
|
406
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // a8 - af
|
407
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // b0 - b7
|
408
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // b8 - bf
|
409
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // c0 - c7
|
410
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // c8 - cf
|
411
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // d0 - d7
|
412
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // d8 - df
|
413
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // e0 - e7
|
414
|
+
PCK4BITS(3,3,3,3,3,4,4,4), // e8 - ef
|
415
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // f0 - f7
|
416
|
+
PCK4BITS(4,4,4,4,4,0,0,0) // f8 - ff
|
417
|
+
};
|
418
|
+
|
419
|
+
|
420
|
+
static const PRUint32 SJIS_st [ 3] = {
|
421
|
+
PCK4BITS(eError,eStart,eStart, 3,eError,eError,eError,eError),//00-07
|
422
|
+
PCK4BITS(eError,eError,eError,eError,eItsMe,eItsMe,eItsMe,eItsMe),//08-0f
|
423
|
+
PCK4BITS(eItsMe,eItsMe,eError,eError,eStart,eStart,eStart,eStart) //10-17
|
424
|
+
};
|
425
|
+
|
426
|
+
static const PRUint32 SJISCharLenTable[] = {0, 1, 1, 2, 0, 0};
|
427
|
+
|
428
|
+
const SMModel SJISSMModel = {
|
429
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, SJIS_cls },
|
430
|
+
6,
|
431
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, SJIS_st },
|
432
|
+
SJISCharLenTable,
|
433
|
+
"SHIFT_JIS",
|
434
|
+
};
|
435
|
+
|
436
|
+
|
437
|
+
static const PRUint32 UTF8_cls [ 256 / 8 ] = {
|
438
|
+
//PCK4BITS(0,1,1,1,1,1,1,1), // 00 - 07
|
439
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 00 - 07 //allow 0x00 as a legal value
|
440
|
+
PCK4BITS(1,1,1,1,1,1,0,0), // 08 - 0f
|
441
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 10 - 17
|
442
|
+
PCK4BITS(1,1,1,0,1,1,1,1), // 18 - 1f
|
443
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 20 - 27
|
444
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 28 - 2f
|
445
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 30 - 37
|
446
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 38 - 3f
|
447
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 40 - 47
|
448
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 48 - 4f
|
449
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 50 - 57
|
450
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 58 - 5f
|
451
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 60 - 67
|
452
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 68 - 6f
|
453
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 70 - 77
|
454
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 78 - 7f
|
455
|
+
PCK4BITS(2,2,2,2,3,3,3,3), // 80 - 87
|
456
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 88 - 8f
|
457
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 90 - 97
|
458
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 98 - 9f
|
459
|
+
PCK4BITS(5,5,5,5,5,5,5,5), // a0 - a7
|
460
|
+
PCK4BITS(5,5,5,5,5,5,5,5), // a8 - af
|
461
|
+
PCK4BITS(5,5,5,5,5,5,5,5), // b0 - b7
|
462
|
+
PCK4BITS(5,5,5,5,5,5,5,5), // b8 - bf
|
463
|
+
PCK4BITS(0,0,6,6,6,6,6,6), // c0 - c7
|
464
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // c8 - cf
|
465
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // d0 - d7
|
466
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // d8 - df
|
467
|
+
PCK4BITS(7,8,8,8,8,8,8,8), // e0 - e7
|
468
|
+
PCK4BITS(8,8,8,8,8,9,8,8), // e8 - ef
|
469
|
+
PCK4BITS(10,11,11,11,11,11,11,11), // f0 - f7
|
470
|
+
PCK4BITS(12,13,13,13,14,15,0,0) // f8 - ff
|
471
|
+
};
|
472
|
+
|
473
|
+
|
474
|
+
static const PRUint32 UTF8_st [ 26] = {
|
475
|
+
PCK4BITS(eError,eStart,eError,eError,eError,eError, 12, 10),//00-07
|
476
|
+
PCK4BITS( 9, 11, 8, 7, 6, 5, 4, 3),//08-0f
|
477
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError),//10-17
|
478
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError),//18-1f
|
479
|
+
PCK4BITS(eItsMe,eItsMe,eItsMe,eItsMe,eItsMe,eItsMe,eItsMe,eItsMe),//20-27
|
480
|
+
PCK4BITS(eItsMe,eItsMe,eItsMe,eItsMe,eItsMe,eItsMe,eItsMe,eItsMe),//28-2f
|
481
|
+
PCK4BITS(eError,eError, 5, 5, 5, 5,eError,eError),//30-37
|
482
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError),//38-3f
|
483
|
+
PCK4BITS(eError,eError,eError, 5, 5, 5,eError,eError),//40-47
|
484
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError),//48-4f
|
485
|
+
PCK4BITS(eError,eError, 7, 7, 7, 7,eError,eError),//50-57
|
486
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError),//58-5f
|
487
|
+
PCK4BITS(eError,eError,eError,eError, 7, 7,eError,eError),//60-67
|
488
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError),//68-6f
|
489
|
+
PCK4BITS(eError,eError, 9, 9, 9, 9,eError,eError),//70-77
|
490
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError),//78-7f
|
491
|
+
PCK4BITS(eError,eError,eError,eError,eError, 9,eError,eError),//80-87
|
492
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError),//88-8f
|
493
|
+
PCK4BITS(eError,eError, 12, 12, 12, 12,eError,eError),//90-97
|
494
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError),//98-9f
|
495
|
+
PCK4BITS(eError,eError,eError,eError,eError, 12,eError,eError),//a0-a7
|
496
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError),//a8-af
|
497
|
+
PCK4BITS(eError,eError, 12, 12, 12,eError,eError,eError),//b0-b7
|
498
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError),//b8-bf
|
499
|
+
PCK4BITS(eError,eError,eStart,eStart,eStart,eStart,eError,eError),//c0-c7
|
500
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError) //c8-cf
|
501
|
+
};
|
502
|
+
|
503
|
+
static const PRUint32 UTF8CharLenTable[] = {0, 1, 0, 0, 0, 0, 2, 3,
|
504
|
+
3, 3, 4, 4, 5, 5, 6, 6 };
|
505
|
+
|
506
|
+
const SMModel UTF8SMModel = {
|
507
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, UTF8_cls },
|
508
|
+
16,
|
509
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, UTF8_st },
|
510
|
+
UTF8CharLenTable,
|
511
|
+
"UTF-8",
|
512
|
+
};
|
513
|
+
|