diff options
Diffstat (limited to 'regsyntax.c')
| -rw-r--r-- | regsyntax.c | 115 |
1 files changed, 94 insertions, 21 deletions
diff --git a/regsyntax.c b/regsyntax.c index ade5b55f77..657ffcd0f3 100644 --- a/regsyntax.c +++ b/regsyntax.c @@ -1,8 +1,9 @@ /********************************************************************** - regsyntax.c - Oniguruma (regular expression library) + regsyntax.c - Onigmo (Oniguruma-mod) (regular expression library) **********************************************************************/ /*- * Copyright (c) 2002-2006 K.Kosako <sndgk393 AT ybb DOT ne DOT jp> + * Copyright (c) 2011-2016 K.Takata <kentkt AT csc DOT jp> * All rights reserved. * * Redistribution and use in source and binary forms, with or without @@ -29,7 +30,7 @@ #include "regint.h" -OnigSyntaxType OnigSyntaxASIS = { +const OnigSyntaxType OnigSyntaxASIS = { 0 , ONIG_SYN_OP2_INEFFECTIVE_ESCAPE , 0 @@ -45,7 +46,7 @@ OnigSyntaxType OnigSyntaxASIS = { } }; -OnigSyntaxType OnigSyntaxPosixBasic = { +const OnigSyntaxType OnigSyntaxPosixBasic = { ( SYN_POSIX_COMMON_OP | ONIG_SYN_OP_ESC_LPAREN_SUBEXP | ONIG_SYN_OP_ESC_BRACE_INTERVAL ) , 0 @@ -62,13 +63,13 @@ OnigSyntaxType OnigSyntaxPosixBasic = { } }; -OnigSyntaxType OnigSyntaxPosixExtended = { +const OnigSyntaxType OnigSyntaxPosixExtended = { ( SYN_POSIX_COMMON_OP | ONIG_SYN_OP_LPAREN_SUBEXP | ONIG_SYN_OP_BRACE_INTERVAL | ONIG_SYN_OP_PLUS_ONE_INF | ONIG_SYN_OP_QMARK_ZERO_ONE | ONIG_SYN_OP_VBAR_ALT ) , 0 - , ( ONIG_SYN_CONTEXT_INDEP_ANCHORS | - ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS | ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS | + , ( ONIG_SYN_CONTEXT_INDEP_ANCHORS | + ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS | ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS | ONIG_SYN_ALLOW_UNMATCHED_CLOSE_SUBEXP | ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC ) , ( ONIG_OPTION_SINGLELINE | ONIG_OPTION_MULTILINE ) @@ -83,7 +84,7 @@ OnigSyntaxType OnigSyntaxPosixExtended = { } }; -OnigSyntaxType OnigSyntaxEmacs = { +const OnigSyntaxType OnigSyntaxEmacs = { ( ONIG_SYN_OP_DOT_ANYCHAR | ONIG_SYN_OP_BRACKET_CC | ONIG_SYN_OP_ESC_BRACE_INTERVAL | ONIG_SYN_OP_ESC_LPAREN_SUBEXP | ONIG_SYN_OP_ESC_VBAR_ALT | @@ -104,7 +105,7 @@ OnigSyntaxType OnigSyntaxEmacs = { } }; -OnigSyntaxType OnigSyntaxGrep = { +const OnigSyntaxType OnigSyntaxGrep = { ( ONIG_SYN_OP_DOT_ANYCHAR | ONIG_SYN_OP_BRACKET_CC | ONIG_SYN_OP_POSIX_BRACKET | ONIG_SYN_OP_ESC_BRACE_INTERVAL | ONIG_SYN_OP_ESC_LPAREN_SUBEXP | ONIG_SYN_OP_ESC_VBAR_ALT | @@ -126,7 +127,7 @@ OnigSyntaxType OnigSyntaxGrep = { } }; -OnigSyntaxType OnigSyntaxGnuRegex = { +const OnigSyntaxType OnigSyntaxGnuRegex = { SYN_GNU_REGEX_OP , 0 , SYN_GNU_REGEX_BV @@ -142,7 +143,7 @@ OnigSyntaxType OnigSyntaxGnuRegex = { } }; -OnigSyntaxType OnigSyntaxJava = { +const OnigSyntaxType OnigSyntaxJava = { (( SYN_GNU_REGEX_OP | ONIG_SYN_OP_QMARK_NON_GREEDY | ONIG_SYN_OP_ESC_CONTROL_CHARS | ONIG_SYN_OP_ESC_C_CONTROL | ONIG_SYN_OP_ESC_OCTAL3 | ONIG_SYN_OP_ESC_X_HEX2 ) @@ -153,7 +154,8 @@ OnigSyntaxType OnigSyntaxJava = { ONIG_SYN_OP2_ESC_V_VTAB | ONIG_SYN_OP2_ESC_U_HEX4 | ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY ) , ( SYN_GNU_REGEX_BV | ONIG_SYN_DIFFERENT_LEN_ALT_LOOK_BEHIND ) - , ONIG_OPTION_SINGLELINE + , ( ONIG_OPTION_SINGLELINE | ONIG_OPTION_ASCII_RANGE | + ONIG_OPTION_WORD_BOUND_ALL_RANGE ) , { (OnigCodePoint )'\\' /* esc */ @@ -165,7 +167,8 @@ OnigSyntaxType OnigSyntaxJava = { } }; -OnigSyntaxType OnigSyntaxPerl = { +/* Perl 5.8 */ +const OnigSyntaxType OnigSyntaxPerl58 = { (( SYN_GNU_REGEX_OP | ONIG_SYN_OP_QMARK_NON_GREEDY | ONIG_SYN_OP_ESC_OCTAL3 | ONIG_SYN_OP_ESC_X_HEX2 | ONIG_SYN_OP_ESC_X_BRACE_HEX8 | ONIG_SYN_OP_ESC_CONTROL_CHARS | @@ -174,7 +177,9 @@ OnigSyntaxType OnigSyntaxPerl = { , ( ONIG_SYN_OP2_ESC_CAPITAL_Q_QUOTE | ONIG_SYN_OP2_QMARK_GROUP_EFFECT | ONIG_SYN_OP2_OPTION_PERL | ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY | - ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT ) + ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT | + ONIG_SYN_OP2_ESC_CAPITAL_X_EXTENDED_GRAPHEME_CLUSTER | + ONIG_SYN_OP2_QMARK_LPAREN_CONDITION) , SYN_GNU_REGEX_BV , ONIG_OPTION_SINGLELINE , @@ -188,8 +193,8 @@ OnigSyntaxType OnigSyntaxPerl = { } }; -/* Perl + named group */ -OnigSyntaxType OnigSyntaxPerl_NG = { +/* Perl 5.8 + named group */ +const OnigSyntaxType OnigSyntaxPerl58_NG = { (( SYN_GNU_REGEX_OP | ONIG_SYN_OP_QMARK_NON_GREEDY | ONIG_SYN_OP_ESC_OCTAL3 | ONIG_SYN_OP_ESC_X_HEX2 | ONIG_SYN_OP_ESC_X_BRACE_HEX8 | ONIG_SYN_OP_ESC_CONTROL_CHARS | @@ -199,6 +204,8 @@ OnigSyntaxType OnigSyntaxPerl_NG = { ONIG_SYN_OP2_QMARK_GROUP_EFFECT | ONIG_SYN_OP2_OPTION_PERL | ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY | ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT | + ONIG_SYN_OP2_ESC_CAPITAL_X_EXTENDED_GRAPHEME_CLUSTER | + ONIG_SYN_OP2_QMARK_LPAREN_CONDITION | ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP | ONIG_SYN_OP2_ESC_K_NAMED_BACKREF | ONIG_SYN_OP2_ESC_G_SUBEXP_CALL ) @@ -217,10 +224,76 @@ OnigSyntaxType OnigSyntaxPerl_NG = { } }; +/* Perl 5.10+ */ +const OnigSyntaxType OnigSyntaxPerl = { + (( SYN_GNU_REGEX_OP | ONIG_SYN_OP_QMARK_NON_GREEDY | + ONIG_SYN_OP_ESC_OCTAL3 | ONIG_SYN_OP_ESC_X_HEX2 | + ONIG_SYN_OP_ESC_X_BRACE_HEX8 | ONIG_SYN_OP_ESC_CONTROL_CHARS | + ONIG_SYN_OP_ESC_O_BRACE_OCTAL | ONIG_SYN_OP_ESC_C_CONTROL ) + & ~ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END ) + , ( ONIG_SYN_OP2_ESC_CAPITAL_Q_QUOTE | + ONIG_SYN_OP2_QMARK_GROUP_EFFECT | ONIG_SYN_OP2_OPTION_PERL | + ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY | + ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT | + ONIG_SYN_OP2_ESC_CAPITAL_X_EXTENDED_GRAPHEME_CLUSTER | + ONIG_SYN_OP2_QMARK_LPAREN_CONDITION | + ONIG_SYN_OP2_PLUS_POSSESSIVE_REPEAT | + ONIG_SYN_OP2_PLUS_POSSESSIVE_INTERVAL | + ONIG_SYN_OP2_ESC_CAPITAL_R_LINEBREAK | + ONIG_SYN_OP2_ESC_CAPITAL_K_KEEP | + ONIG_SYN_OP2_QMARK_SUBEXP_CALL | + ONIG_SYN_OP2_ESC_G_BRACE_BACKREF | + ONIG_SYN_OP2_QMARK_CAPITAL_P_NAMED_GROUP | + ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP | + ONIG_SYN_OP2_ESC_K_NAMED_BACKREF ) + , ( SYN_GNU_REGEX_BV | + ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME | + ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME_CALL | + ONIG_SYN_USE_LEFT_MOST_NAMED_GROUP ) + , ( ONIG_OPTION_SINGLELINE | ONIG_OPTION_CAPTURE_GROUP ) + , + { + (OnigCodePoint )'\\' /* esc */ + , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* anychar '.' */ + , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* anytime '*' */ + , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* zero or one time '?' */ + , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* one or more time '+' */ + , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* anychar anytime */ + } +}; + +const OnigSyntaxType OnigSyntaxPython = { + (( SYN_GNU_REGEX_OP | ONIG_SYN_OP_QMARK_NON_GREEDY | + ONIG_SYN_OP_ESC_OCTAL3 | ONIG_SYN_OP_ESC_X_HEX2 | + ONIG_SYN_OP_ESC_X_BRACE_HEX8 | ONIG_SYN_OP_ESC_CONTROL_CHARS | + ONIG_SYN_OP_ESC_C_CONTROL ) + & ~ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END ) + , ( ONIG_SYN_OP2_QMARK_GROUP_EFFECT | ONIG_SYN_OP2_OPTION_PERL | + ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY | + ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT | + ONIG_SYN_OP2_PLUS_POSSESSIVE_REPEAT | + ONIG_SYN_OP2_ESC_V_VTAB | + ONIG_SYN_OP2_ESC_U_HEX4 | + ONIG_SYN_OP2_QMARK_LPAREN_CONDITION | + ONIG_SYN_OP2_QMARK_CAPITAL_P_NAMED_GROUP ) + , ( SYN_GNU_REGEX_BV | + ONIG_SYN_ALLOW_INTERVAL_LOW_ABBREV ) + , ( ONIG_OPTION_SINGLELINE | ONIG_OPTION_ASCII_RANGE ) + , + { + (OnigCodePoint )'\\' /* esc */ + , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* anychar '.' */ + , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* anytime '*' */ + , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* zero or one time '?' */ + , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* one or more time '+' */ + , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* anychar anytime */ + } +}; + extern int -onig_set_default_syntax(OnigSyntaxType* syntax) +onig_set_default_syntax(const OnigSyntaxType* syntax) { if (IS_NULL(syntax)) syntax = ONIG_SYNTAX_RUBY; @@ -230,7 +303,7 @@ onig_set_default_syntax(OnigSyntaxType* syntax) } extern void -onig_copy_syntax(OnigSyntaxType* to, OnigSyntaxType* from) +onig_copy_syntax(OnigSyntaxType* to, const OnigSyntaxType* from) { *to = *from; } @@ -260,25 +333,25 @@ onig_set_syntax_options(OnigSyntaxType* syntax, OnigOptionType options) } extern unsigned int -onig_get_syntax_op(OnigSyntaxType* syntax) +onig_get_syntax_op(const OnigSyntaxType* syntax) { return syntax->op; } extern unsigned int -onig_get_syntax_op2(OnigSyntaxType* syntax) +onig_get_syntax_op2(const OnigSyntaxType* syntax) { return syntax->op2; } extern unsigned int -onig_get_syntax_behavior(OnigSyntaxType* syntax) +onig_get_syntax_behavior(const OnigSyntaxType* syntax) { return syntax->behavior; } extern OnigOptionType -onig_get_syntax_options(OnigSyntaxType* syntax) +onig_get_syntax_options(const OnigSyntaxType* syntax) { return syntax->options; } |
