From 0424e152c684a85f4b0691f1e84aec203115333d Mon Sep 17 00:00:00 2001 From: naruse Date: Fri, 17 Feb 2012 07:42:23 +0000 Subject: * Merge Onigmo-5.13.1. [ruby-dev:45057] [Feature #5820] https://github.com/k-takata/Onigmo cp reg{comp,enc,error,exec,parse,syntax}.c reg{enc,int,parse}.h cp oniguruma.h cp tool/enc-unicode.rb cp -r enc/ git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@34663 b2dd03c8-39d4-4d8f-98ff-823fe69b080e --- include/ruby/oniguruma.h | 90 ++++++++++++++++++++++++++++++++---------------- 1 file changed, 61 insertions(+), 29 deletions(-) (limited to 'include') diff --git a/include/ruby/oniguruma.h b/include/ruby/oniguruma.h index 67773aa482..c01b6f7bb4 100644 --- a/include/ruby/oniguruma.h +++ b/include/ruby/oniguruma.h @@ -1,10 +1,11 @@ #ifndef ONIGURUMA_H #define ONIGURUMA_H /********************************************************************** - oniguruma.h - Oniguruma (regular expression library) + oniguruma.h - Onigmo (Oniguruma-mod) (regular expression library) **********************************************************************/ /*- - * Copyright (c) 2002-2008 K.Kosako + * Copyright (c) 2002-2009 K.Kosako + * Copyright (c) 2011-2012 K.Takata * All rights reserved. * * Redistribution and use in source and binary forms, with or without @@ -38,8 +39,8 @@ extern "C" { #define ONIGURUMA #define ONIGURUMA_VERSION_MAJOR 5 -#define ONIGURUMA_VERSION_MINOR 9 -#define ONIGURUMA_VERSION_TEENY 2 +#define ONIGURUMA_VERSION_MINOR 13 +#define ONIGURUMA_VERSION_TEENY 1 #ifdef __cplusplus # ifndef HAVE_PROTOTYPES @@ -101,6 +102,8 @@ extern "C" { #pragma GCC visibility push(default) #endif +#include /* for size_t */ + /* PART: character encoding */ #ifndef ONIG_ESCAPE_UCHAR_COLLISION @@ -108,9 +111,10 @@ extern "C" { #endif typedef unsigned char OnigUChar; -typedef unsigned int OnigCodePoint; +typedef unsigned int OnigCodePoint; typedef unsigned int OnigCtype; typedef size_t OnigDistance; +typedef ptrdiff_t OnigPosition; #define ONIG_INFINITE_DISTANCE ~((OnigDistance )0) @@ -171,6 +175,7 @@ typedef struct OnigEncodingTypeST { OnigUChar* (*left_adjust_char_head)(const OnigUChar* start, const OnigUChar* p, const OnigUChar* end, struct OnigEncodingTypeST* enc); int (*is_allowed_reverse_match)(const OnigUChar* p, const OnigUChar* end, struct OnigEncodingTypeST* enc); int ruby_encoding_index; + unsigned int flags; } OnigEncodingType; typedef OnigEncodingType* OnigEncoding; @@ -204,17 +209,12 @@ ONIG_EXTERN OnigEncodingType OnigEncodingASCII; #define ONIGENC_CTYPE_ALNUM 13 /* alpha || digit */ #define ONIGENC_CTYPE_ASCII 14 #define ONIGENC_MAX_STD_CTYPE ONIGENC_CTYPE_ASCII -#define ONIGENC_CTYPE_SPECIAL_MASK 256 -#define ONIGENC_CTYPE_S /* [\t\n\v\f\r\s] */ \ - ONIGENC_CTYPE_SPECIAL_MASK | ONIGENC_CTYPE_SPACE -#define ONIGENC_CTYPE_D /* [0-9] */ \ - ONIGENC_CTYPE_SPECIAL_MASK | ONIGENC_CTYPE_DIGIT -#define ONIGENC_CTYPE_W /* [0-9A-Za-z_] */ \ - ONIGENC_CTYPE_SPECIAL_MASK | ONIGENC_CTYPE_WORD -#define ONIGENC_CTYPE_SPECIAL_P(ctype) ((ctype) & ONIGENC_CTYPE_SPECIAL_MASK) +/* flags */ +#define ONIGENC_FLAG_NONE 0U +#define ONIGENC_FLAG_UNICODE 1U -#define onig_enc_len(enc,p,e) ONIGENC_MBC_ENC_LEN(enc, p, e) +#define onig_enc_len(enc,p,e) ONIGENC_MBC_ENC_LEN(enc, p, e) #define ONIGENC_IS_UNDEF(enc) ((enc) == ONIG_ENCODING_UNDEF) #define ONIGENC_IS_SINGLEBYTE(enc) (ONIGENC_MBC_MAXLEN(enc) == 1) @@ -223,6 +223,10 @@ ONIG_EXTERN OnigEncodingType OnigEncodingASCII; #define ONIGENC_IS_CODE_ASCII(code) ((code) < 128) #define ONIGENC_IS_MBC_WORD(enc,s,end) \ ONIGENC_IS_CODE_WORD(enc,ONIGENC_MBC_TO_CODE(enc,s,end)) +#define ONIGENC_IS_MBC_ASCII_WORD(enc,s,end) \ + onigenc_ascii_is_code_ctype( \ + ONIGENC_MBC_TO_CODE(enc,s,end),ONIGENC_CTYPE_WORD,enc) +#define ONIGENC_IS_UNICODE(enc) ((enc)->flags & ONIGENC_FLAG_UNICODE) #define ONIGENC_NAME(enc) ((enc)->name) @@ -350,6 +354,7 @@ typedef unsigned int OnigOptionType; #define ONIG_OPTION_IGNORECASE 1U #define ONIG_OPTION_EXTEND (ONIG_OPTION_IGNORECASE << 1) #define ONIG_OPTION_MULTILINE (ONIG_OPTION_EXTEND << 1) +#define ONIG_OPTION_DOTALL ONIG_OPTION_MULTILINE #define ONIG_OPTION_SINGLELINE (ONIG_OPTION_MULTILINE << 1) #define ONIG_OPTION_FIND_LONGEST (ONIG_OPTION_SINGLELINE << 1) #define ONIG_OPTION_FIND_NOT_EMPTY (ONIG_OPTION_FIND_LONGEST << 1) @@ -360,7 +365,13 @@ typedef unsigned int OnigOptionType; #define ONIG_OPTION_NOTBOL (ONIG_OPTION_CAPTURE_GROUP << 1) #define ONIG_OPTION_NOTEOL (ONIG_OPTION_NOTBOL << 1) #define ONIG_OPTION_POSIX_REGION (ONIG_OPTION_NOTEOL << 1) -#define ONIG_OPTION_MAXBIT ONIG_OPTION_POSIX_REGION /* limit */ +/* options (ctype range) */ +#define ONIG_OPTION_ASCII_RANGE (ONIG_OPTION_POSIX_REGION << 1) +#define ONIG_OPTION_POSIX_BRACKET_ALL_RANGE (ONIG_OPTION_ASCII_RANGE << 1) +#define ONIG_OPTION_WORD_BOUND_ALL_RANGE (ONIG_OPTION_POSIX_BRACKET_ALL_RANGE << 1) +/* options (newline) */ +#define ONIG_OPTION_NEWLINE_CRLF (ONIG_OPTION_WORD_BOUND_ALL_RANGE << 1) +#define ONIG_OPTION_MAXBIT ONIG_OPTION_NEWLINE_CRLF /* limit */ #define ONIG_OPTION_ON(options,regopt) ((options) |= (regopt)) #define ONIG_OPTION_OFF(options,regopt) ((options) &= ~(regopt)) @@ -382,9 +393,11 @@ ONIG_EXTERN const OnigSyntaxType OnigSyntaxEmacs; ONIG_EXTERN const OnigSyntaxType OnigSyntaxGrep; ONIG_EXTERN const OnigSyntaxType OnigSyntaxGnuRegex; ONIG_EXTERN const OnigSyntaxType OnigSyntaxJava; +ONIG_EXTERN const OnigSyntaxType OnigSyntaxPerl58; +ONIG_EXTERN const OnigSyntaxType OnigSyntaxPerl58_NG; ONIG_EXTERN const OnigSyntaxType OnigSyntaxPerl; -ONIG_EXTERN const OnigSyntaxType OnigSyntaxPerl_NG; ONIG_EXTERN const OnigSyntaxType OnigSyntaxRuby; +ONIG_EXTERN const OnigSyntaxType OnigSyntaxPython; /* predefined syntaxes (see regsyntax.c) */ #define ONIG_SYNTAX_ASIS (&OnigSyntaxASIS) @@ -394,9 +407,11 @@ ONIG_EXTERN const OnigSyntaxType OnigSyntaxRuby; #define ONIG_SYNTAX_GREP (&OnigSyntaxGrep) #define ONIG_SYNTAX_GNU_REGEX (&OnigSyntaxGnuRegex) #define ONIG_SYNTAX_JAVA (&OnigSyntaxJava) +#define ONIG_SYNTAX_PERL58 (&OnigSyntaxPerl58) +#define ONIG_SYNTAX_PERL58_NG (&OnigSyntaxPerl58_NG) #define ONIG_SYNTAX_PERL (&OnigSyntaxPerl) -#define ONIG_SYNTAX_PERL_NG (&OnigSyntaxPerl_NG) #define ONIG_SYNTAX_RUBY (&OnigSyntaxRuby) +#define ONIG_SYNTAX_PYTHON (&OnigSyntaxPython) /* default syntax */ ONIG_EXTERN const OnigSyntaxType* OnigDefaultSyntax; @@ -434,11 +449,12 @@ ONIG_EXTERN const OnigSyntaxType* OnigDefaultSyntax; #define ONIG_SYN_OP_ESC_OCTAL3 (1U<<28) /* \OOO */ #define ONIG_SYN_OP_ESC_X_HEX2 (1U<<29) /* \xHH */ #define ONIG_SYN_OP_ESC_X_BRACE_HEX8 (1U<<30) /* \x{7HHHHHHH} */ +#define ONIG_SYN_OP_ESC_O_BRACE_OCTAL (1U<<31) /* \o{OOO} */ /* NOTIMPL */ #define ONIG_SYN_OP2_ESC_CAPITAL_Q_QUOTE (1U<<0) /* \Q...\E */ #define ONIG_SYN_OP2_QMARK_GROUP_EFFECT (1U<<1) /* (?...) */ -#define ONIG_SYN_OP2_OPTION_PERL (1U<<2) /* (?imsx),(?-imsx) */ -#define ONIG_SYN_OP2_OPTION_RUBY (1U<<3) /* (?imx), (?-imx) */ +#define ONIG_SYN_OP2_OPTION_PERL (1U<<2) /* (?imsxadlu), (?-imsx), (?^imsxalu) */ +#define ONIG_SYN_OP2_OPTION_RUBY (1U<<3) /* (?imxadu), (?-imx) */ #define ONIG_SYN_OP2_PLUS_POSSESSIVE_REPEAT (1U<<4) /* ?+,*+,++ */ #define ONIG_SYN_OP2_PLUS_POSSESSIVE_INTERVAL (1U<<5) /* {n,m}+ */ #define ONIG_SYN_OP2_CCLASS_SET_OP (1U<<6) /* [...&&..[..]..] */ @@ -456,6 +472,17 @@ ONIG_EXTERN const OnigSyntaxType* OnigDefaultSyntax; /* #define ONIG_SYN_OP2_CHAR_PROPERTY_PREFIX_IS (1U<<18) */ #define ONIG_SYN_OP2_ESC_H_XDIGIT (1U<<19) /* \h, \H */ #define ONIG_SYN_OP2_INEFFECTIVE_ESCAPE (1U<<20) /* \ */ +#define ONIG_SYN_OP2_ESC_CAPITAL_R_LINEBREAK (1U<<21) /* \R as (?>\x0D\x0A|[\x0A-\x0D\x{85}\x{2028}\x{2029}]) */ +#define ONIG_SYN_OP2_ESC_CAPITAL_X_EXTENDED_GRAPHEME_CLUSTER (1U<<22) /* \X as (?>\P{M}\p{M}*) */ +#define ONIG_SYN_OP2_ESC_V_VERTICAL_WHITESPACE (1U<<23) /* \v, \V -- Perl */ /* NOTIMPL */ +#define ONIG_SYN_OP2_ESC_H_HORIZONTAL_WHITESPACE (1U<<24) /* \h, \H -- Perl */ /* NOTIMPL */ +#define ONIG_SYN_OP2_ESC_CAPITAL_K_KEEP (1U<<25) /* \K */ +#define ONIG_SYN_OP2_ESC_G_BRACE_BACKREF (1U<<26) /* \g{name}, \g{n} */ +#define ONIG_SYN_OP2_QMARK_SUBEXP_CALL (1U<<27) /* (?&name), (?n), (?R), (?0) */ +#define ONIG_SYN_OP2_QMARK_VBAR_BRANCH_RESET (1U<<28) /* (?|...) */ /* NOTIMPL */ +#define ONIG_SYN_OP2_QMARK_LPAREN_CONDITION (1U<<29) /* (?(cond)yes...|no...) */ +#define ONIG_SYN_OP2_QMARK_CAPITAL_P_NAMED_GROUP (1U<<30) /* (?P...), (?P=name), (?P>name) -- Python/PCRE */ +#define ONIG_SYN_OP2_OPTION_JAVA (1U<<31) /* (?idmsux), (?-idmsux) */ /* NOTIMPL */ /* syntax (behavior) */ #define ONIG_SYN_CONTEXT_INDEP_ANCHORS (1U<<31) /* not implemented */ @@ -469,6 +496,7 @@ ONIG_EXTERN const OnigSyntaxType* OnigDefaultSyntax; #define ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP (1U<<7) /* see doc/RE */ #define ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME (1U<<8) /* (?)(?) */ #define ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY (1U<<9) /* a{n}?=(?:a{n})? */ +#define ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME_CALL (1U<<10) /* (?)(?)(?&x) */ /* syntax (behavior) in char class [...] */ #define ONIG_SYN_NOT_NEWLINE_IN_NEGATIVE_CC (1U<<20) /* [^...] */ @@ -505,7 +533,7 @@ ONIG_EXTERN const OnigSyntaxType* OnigDefaultSyntax; #define ONIGERR_UNDEFINED_BYTECODE -13 #define ONIGERR_UNEXPECTED_BYTECODE -14 #define ONIGERR_MATCH_STACK_LIMIT_OVER -15 -#define ONIGERR_DEFAULT_ENCODING_IS_NOT_SETTED -21 +#define ONIGERR_DEFAULT_ENCODING_IS_NOT_SET -21 #define ONIGERR_SPECIFIED_ENCODING_CANT_CONVERT_TO_WIDE_CHAR -22 /* general error */ #define ONIGERR_INVALID_ARGUMENT -30 @@ -532,6 +560,7 @@ ONIG_EXTERN const OnigSyntaxType* OnigDefaultSyntax; #define ONIGERR_INVALID_POSIX_BRACKET_TYPE -121 #define ONIGERR_INVALID_LOOK_BEHIND_PATTERN -122 #define ONIGERR_INVALID_REPEAT_RANGE_PATTERN -123 +#define ONIGERR_INVALID_CONDITION_PATTERN -124 /* values error (syntax error) */ #define ONIGERR_TOO_BIG_NUMBER -200 #define ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE -201 @@ -543,6 +572,7 @@ ONIG_EXTERN const OnigSyntaxType* OnigDefaultSyntax; #define ONIGERR_TOO_BIG_BACKREF_NUMBER -207 #define ONIGERR_INVALID_BACKREF -208 #define ONIGERR_NUMBERED_BACKREF_OR_CALL_NOT_ALLOWED -209 +#define ONIGERR_TOO_SHORT_DIGITS -210 #define ONIGERR_TOO_LONG_WIDE_CHAR_VALUE -212 #define ONIGERR_EMPTY_GROUP_NAME -214 #define ONIGERR_INVALID_GROUP_NAME -215 @@ -571,8 +601,8 @@ ONIG_EXTERN const OnigSyntaxType* OnigDefaultSyntax; typedef struct OnigCaptureTreeNodeStruct { int group; /* group number */ - int beg; - int end; + OnigPosition beg; + OnigPosition end; int allocated; int num_childs; struct OnigCaptureTreeNodeStruct** childs; @@ -582,8 +612,8 @@ typedef struct OnigCaptureTreeNodeStruct { struct re_registers { int allocated; int num_regs; - int* beg; - int* end; + OnigPosition* beg; + OnigPosition* end; /* extended */ OnigCaptureTreeNode* history_root; /* capture history tree root */ }; @@ -689,7 +719,7 @@ typedef struct { ONIG_EXTERN int onig_init P_((void)); ONIG_EXTERN -int onig_error_code_to_str PV_((OnigUChar* s, int err_code, ...)); +int onig_error_code_to_str PV_((OnigUChar* s, OnigPosition err_code, ...)); ONIG_EXTERN void onig_set_warn_func P_((OnigWarnFunc f)); ONIG_EXTERN @@ -697,7 +727,7 @@ void onig_set_verb_warn_func P_((OnigWarnFunc f)); ONIG_EXTERN int onig_new P_((OnigRegex*, const OnigUChar* pattern, const OnigUChar* pattern_end, OnigOptionType option, OnigEncoding enc, const OnigSyntaxType* syntax, OnigErrorInfo* einfo)); ONIG_EXTERN -int onig_reg_init P_((regex_t* reg, OnigOptionType option, OnigCaseFoldType case_fold_flag, OnigEncoding enc, const OnigSyntaxType* syntax)); +int onig_reg_init P_((OnigRegex reg, OnigOptionType option, OnigCaseFoldType case_fold_flag, OnigEncoding enc, const OnigSyntaxType* syntax)); ONIG_EXTERN int onig_new_without_alloc P_((OnigRegex, const OnigUChar* pattern, const OnigUChar* pattern_end, OnigOptionType option, OnigEncoding enc, OnigSyntaxType* syntax, OnigErrorInfo* einfo)); ONIG_EXTERN @@ -711,9 +741,11 @@ int onig_recompile P_((OnigRegex, const OnigUChar* pattern, const OnigUChar* pat ONIG_EXTERN int onig_recompile_deluxe P_((OnigRegex reg, const OnigUChar* pattern, const OnigUChar* pattern_end, OnigCompileInfo* ci, OnigErrorInfo* einfo)); ONIG_EXTERN -long onig_search P_((OnigRegex, const OnigUChar* str, const OnigUChar* end, const OnigUChar* start, const OnigUChar* range, OnigRegion* region, OnigOptionType option)); +OnigPosition onig_search P_((OnigRegex, const OnigUChar* str, const OnigUChar* end, const OnigUChar* start, const OnigUChar* range, OnigRegion* region, OnigOptionType option)); +ONIG_EXTERN +OnigPosition onig_search_gpos P_((OnigRegex, const OnigUChar* str, const OnigUChar* end, const OnigUChar* global_pos, const OnigUChar* start, const OnigUChar* range, OnigRegion* region, OnigOptionType option)); ONIG_EXTERN -long onig_match P_((OnigRegex, const OnigUChar* str, const OnigUChar* end, const OnigUChar* at, OnigRegion* region, OnigOptionType option)); +OnigPosition onig_match P_((OnigRegex, const OnigUChar* str, const OnigUChar* end, const OnigUChar* at, OnigRegion* region, OnigOptionType option)); ONIG_EXTERN OnigRegion* onig_region_new P_((void)); ONIG_EXTERN @@ -743,7 +775,7 @@ int onig_number_of_capture_histories P_((OnigRegex reg)); ONIG_EXTERN OnigCaptureTreeNode* onig_get_capture_tree P_((OnigRegion* region)); ONIG_EXTERN -int onig_capture_tree_traverse P_((OnigRegion* region, int at, int(*callback_func)(int,int,int,int,int,void*), void* arg)); +int onig_capture_tree_traverse P_((OnigRegion* region, int at, int(*callback_func)(int,OnigPosition,OnigPosition,int,int,void*), void* arg)); ONIG_EXTERN int onig_noname_group_capture_is_active P_((OnigRegex reg)); ONIG_EXTERN -- cgit v1.2.3