diff options
Diffstat (limited to 'ext/mbstring/oniguruma/enc/iso8859_5.c')
-rw-r--r-- | ext/mbstring/oniguruma/enc/iso8859_5.c | 292 |
1 files changed, 111 insertions, 181 deletions
diff --git a/ext/mbstring/oniguruma/enc/iso8859_5.c b/ext/mbstring/oniguruma/enc/iso8859_5.c index 87b7fb8a29..1ca67e735f 100644 --- a/ext/mbstring/oniguruma/enc/iso8859_5.c +++ b/ext/mbstring/oniguruma/enc/iso8859_5.c @@ -2,7 +2,7 @@ iso8859_5.c - Oniguruma (regular expression library) **********************************************************************/ /*- - * Copyright (c) 2002-2005 K.Kosako <sndgk393 AT ybb DOT ne DOT jp> + * Copyright (c) 2002-2008 K.Kosako <sndgk393 AT ybb DOT ne DOT jp> * All rights reserved. * * Redistribution and use in source and binary forms, with or without @@ -31,7 +31,7 @@ #define ENC_ISO_8859_5_TO_LOWER_CASE(c) EncISO_8859_5_ToLowerCaseTable[c] #define ENC_IS_ISO_8859_5_CTYPE(code,ctype) \ - ((EncISO_8859_5_CtypeTable[code] & ctype) != 0) + ((EncISO_8859_5_CtypeTable[code] & CTYPE_TO_BIT(ctype)) != 0) static const UChar EncISO_8859_5_ToLowerCaseTable[256] = { '\000', '\001', '\002', '\003', '\004', '\005', '\006', '\007', @@ -69,80 +69,66 @@ static const UChar EncISO_8859_5_ToLowerCaseTable[256] = { }; static const unsigned short EncISO_8859_5_CtypeTable[256] = { - 0x2008, 0x2008, 0x2008, 0x2008, 0x2008, 0x2008, 0x2008, 0x2008, - 0x2008, 0x220c, 0x2209, 0x2208, 0x2208, 0x2208, 0x2008, 0x2008, - 0x2008, 0x2008, 0x2008, 0x2008, 0x2008, 0x2008, 0x2008, 0x2008, - 0x2008, 0x2008, 0x2008, 0x2008, 0x2008, 0x2008, 0x2008, 0x2008, - 0x2284, 0x21a0, 0x21a0, 0x21a0, 0x21a0, 0x21a0, 0x21a0, 0x21a0, - 0x21a0, 0x21a0, 0x21a0, 0x21a0, 0x21a0, 0x21a0, 0x21a0, 0x21a0, - 0x38b0, 0x38b0, 0x38b0, 0x38b0, 0x38b0, 0x38b0, 0x38b0, 0x38b0, - 0x38b0, 0x38b0, 0x21a0, 0x21a0, 0x21a0, 0x21a0, 0x21a0, 0x21a0, - 0x21a0, 0x3ca2, 0x3ca2, 0x3ca2, 0x3ca2, 0x3ca2, 0x3ca2, 0x34a2, - 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, - 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, - 0x34a2, 0x34a2, 0x34a2, 0x21a0, 0x21a0, 0x21a0, 0x21a0, 0x31a0, - 0x21a0, 0x38e2, 0x38e2, 0x38e2, 0x38e2, 0x38e2, 0x38e2, 0x30e2, - 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, - 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, - 0x30e2, 0x30e2, 0x30e2, 0x21a0, 0x21a0, 0x21a0, 0x21a0, 0x2008, + 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, + 0x4008, 0x420c, 0x4209, 0x4208, 0x4208, 0x4208, 0x4008, 0x4008, + 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, + 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, + 0x4284, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, + 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, + 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0, + 0x78b0, 0x78b0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, + 0x41a0, 0x7ca2, 0x7ca2, 0x7ca2, 0x7ca2, 0x7ca2, 0x7ca2, 0x74a2, + 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, + 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, + 0x74a2, 0x74a2, 0x74a2, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x51a0, + 0x41a0, 0x78e2, 0x78e2, 0x78e2, 0x78e2, 0x78e2, 0x78e2, 0x70e2, + 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, + 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, + 0x70e2, 0x70e2, 0x70e2, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x4008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, - 0x0284, 0x14a2, 0x14a2, 0x14a2, 0x14a2, 0x14a2, 0x14a2, 0x14a2, - 0x14a2, 0x14a2, 0x14a2, 0x14a2, 0x14a2, 0x01a0, 0x14a2, 0x14a2, - 0x14a2, 0x14a2, 0x14a2, 0x14a2, 0x14a2, 0x14a2, 0x14a2, 0x14a2, - 0x14a2, 0x14a2, 0x14a2, 0x14a2, 0x14a2, 0x14a2, 0x14a2, 0x14a2, - 0x14a2, 0x14a2, 0x14a2, 0x14a2, 0x14a2, 0x14a2, 0x14a2, 0x14a2, - 0x14a2, 0x14a2, 0x14a2, 0x14a2, 0x14a2, 0x14a2, 0x14a2, 0x14a2, - 0x10e2, 0x10e2, 0x10e2, 0x10e2, 0x10e2, 0x10e2, 0x10e2, 0x10e2, - 0x10e2, 0x10e2, 0x10e2, 0x10e2, 0x10e2, 0x10e2, 0x10e2, 0x10e2, - 0x10e2, 0x10e2, 0x10e2, 0x10e2, 0x10e2, 0x10e2, 0x10e2, 0x10e2, - 0x10e2, 0x10e2, 0x10e2, 0x10e2, 0x10e2, 0x10e2, 0x10e2, 0x10e2, - 0x00a0, 0x10e2, 0x10e2, 0x10e2, 0x10e2, 0x10e2, 0x10e2, 0x10e2, - 0x10e2, 0x10e2, 0x10e2, 0x10e2, 0x10e2, 0x00a0, 0x10e2, 0x10e2 + 0x0284, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, + 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x01a0, 0x34a2, 0x34a2, + 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, + 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, + 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, + 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, + 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, + 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, + 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, + 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, + 0x00a0, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, + 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x00a0, 0x30e2, 0x30e2 }; static int -iso_8859_5_mbc_to_normalize(OnigAmbigType flag, - const UChar** pp, const UChar* end, UChar* lower) +mbc_case_fold(OnigCaseFoldType flag ARG_UNUSED, + const UChar** pp, const UChar* end ARG_UNUSED, UChar* lower) { const UChar* p = *pp; - if (((flag & ONIGENC_AMBIGUOUS_MATCH_ASCII_CASE) != 0 && - ONIGENC_IS_MBC_ASCII(p)) || - ((flag & ONIGENC_AMBIGUOUS_MATCH_NONASCII_CASE) != 0 && - !ONIGENC_IS_MBC_ASCII(p))) { - *lower = ENC_ISO_8859_5_TO_LOWER_CASE(*p); - } - else { - *lower = *p; - } - + *lower = ENC_ISO_8859_5_TO_LOWER_CASE(*p); (*pp)++; - return 1; /* return byte length of converted char to lower */ + return 1; } +#if 0 static int -iso_8859_5_is_mbc_ambiguous(OnigAmbigType flag, - const UChar** pp, const UChar* end) +is_mbc_ambiguous(OnigCaseFoldType flag, const UChar** pp, const UChar* end) { + int v; const UChar* p = *pp; (*pp)++; - if (((flag & ONIGENC_AMBIGUOUS_MATCH_ASCII_CASE) != 0 && - ONIGENC_IS_MBC_ASCII(p)) || - ((flag & ONIGENC_AMBIGUOUS_MATCH_NONASCII_CASE) != 0 && - !ONIGENC_IS_MBC_ASCII(p))) { - int v = (EncISO_8859_5_CtypeTable[*p] & - (ONIGENC_CTYPE_UPPER | ONIGENC_CTYPE_LOWER)); - return (v != 0 ? TRUE : FALSE); - } - return FALSE; + v = (EncISO_8859_5_CtypeTable[*p] & (BIT_CTYPE_UPPER | BIT_CTYPE_LOWER)); + return (v != 0 ? TRUE : FALSE); } +#endif static int -iso_8859_5_is_code_ctype(OnigCodePoint code, unsigned int ctype) +is_code_ctype(OnigCodePoint code, unsigned int ctype) { if (code < 256) return ENC_IS_ISO_8859_5_CTYPE(code, ctype); @@ -150,120 +136,74 @@ iso_8859_5_is_code_ctype(OnigCodePoint code, unsigned int ctype) return FALSE; } +static const OnigPairCaseFoldCodes CaseFoldMap[] = { + { 0xa1, 0xf1 }, + { 0xa2, 0xf2 }, + { 0xa3, 0xf3 }, + { 0xa4, 0xf4 }, + { 0xa5, 0xf5 }, + { 0xa6, 0xf6 }, + { 0xa7, 0xf7 }, + { 0xa8, 0xf8 }, + { 0xa9, 0xf9 }, + { 0xaa, 0xfa }, + { 0xab, 0xfb }, + { 0xac, 0xfc }, + { 0xae, 0xfe }, + { 0xaf, 0xff }, + + { 0xb0, 0xd0 }, + { 0xb1, 0xd1 }, + { 0xb2, 0xd2 }, + { 0xb3, 0xd3 }, + { 0xb4, 0xd4 }, + { 0xb5, 0xd5 }, + { 0xb6, 0xd6 }, + { 0xb7, 0xd7 }, + { 0xb8, 0xd8 }, + { 0xb9, 0xd9 }, + { 0xba, 0xda }, + { 0xbb, 0xdb }, + { 0xbc, 0xdc }, + { 0xbd, 0xdd }, + { 0xbe, 0xde }, + { 0xbf, 0xdf }, + + { 0xc0, 0xe0 }, + { 0xc1, 0xe1 }, + { 0xc2, 0xe2 }, + { 0xc3, 0xe3 }, + { 0xc4, 0xe4 }, + { 0xc5, 0xe5 }, + { 0xc6, 0xe6 }, + { 0xc7, 0xe7 }, + { 0xc8, 0xe8 }, + { 0xc9, 0xe9 }, + { 0xca, 0xea }, + { 0xcb, 0xeb }, + { 0xcc, 0xec }, + { 0xcd, 0xed }, + { 0xce, 0xee }, + { 0xcf, 0xef } +}; + static int -iso_8859_5_get_all_pair_ambig_codes(OnigAmbigType flag, - const OnigPairAmbigCodes** ccs) +apply_all_case_fold(OnigCaseFoldType flag, + OnigApplyAllCaseFoldFunc f, void* arg) { - static const OnigPairAmbigCodes cc[] = { - { 0xa1, 0xf1 }, - { 0xa2, 0xf2 }, - { 0xa3, 0xf3 }, - { 0xa4, 0xf4 }, - { 0xa5, 0xf5 }, - { 0xa6, 0xf6 }, - { 0xa7, 0xf7 }, - { 0xa8, 0xf8 }, - { 0xa9, 0xf9 }, - { 0xaa, 0xfa }, - { 0xab, 0xfb }, - { 0xac, 0xfc }, - { 0xae, 0xfe }, - { 0xaf, 0xff }, - - { 0xb0, 0xd0 }, - { 0xb1, 0xd1 }, - { 0xb2, 0xd2 }, - { 0xb3, 0xd3 }, - { 0xb4, 0xd4 }, - { 0xb5, 0xd5 }, - { 0xb6, 0xd6 }, - { 0xb7, 0xd7 }, - { 0xb8, 0xd8 }, - { 0xb9, 0xd9 }, - { 0xba, 0xda }, - { 0xbb, 0xdb }, - { 0xbc, 0xdc }, - { 0xbd, 0xdd }, - { 0xbe, 0xdf }, - { 0xbf, 0xdf }, - - { 0xc0, 0xe0 }, - { 0xc1, 0xe1 }, - { 0xc2, 0xe2 }, - { 0xc3, 0xe3 }, - { 0xc4, 0xe4 }, - { 0xc5, 0xe5 }, - { 0xc6, 0xe6 }, - { 0xc7, 0xe7 }, - { 0xc8, 0xe8 }, - { 0xc9, 0xe9 }, - { 0xca, 0xea }, - { 0xcb, 0xeb }, - { 0xcc, 0xec }, - { 0xcd, 0xed }, - { 0xce, 0xee }, - { 0xcf, 0xef }, - - { 0xd0, 0xb0 }, - { 0xd1, 0xb1 }, - { 0xd2, 0xb2 }, - { 0xd3, 0xb3 }, - { 0xd4, 0xb4 }, - { 0xd5, 0xb5 }, - { 0xd6, 0xb6 }, - { 0xd7, 0xb7 }, - { 0xd8, 0xb8 }, - { 0xd9, 0xb9 }, - { 0xda, 0xba }, - { 0xdb, 0xbb }, - { 0xdc, 0xbc }, - { 0xdd, 0xbd }, - { 0xde, 0xbe }, - { 0xdf, 0xbf }, - - { 0xe0, 0xc0 }, - { 0xe1, 0xc1 }, - { 0xe2, 0xc2 }, - { 0xe3, 0xc3 }, - { 0xe4, 0xc4 }, - { 0xe5, 0xc5 }, - { 0xe6, 0xc6 }, - { 0xe7, 0xc7 }, - { 0xe8, 0xc8 }, - { 0xe9, 0xc9 }, - { 0xea, 0xca }, - { 0xeb, 0xcb }, - { 0xec, 0xcc }, - { 0xed, 0xcd }, - { 0xee, 0xce }, - { 0xef, 0xcf }, - - { 0xf1, 0xa1 }, - { 0xf2, 0xa2 }, - { 0xf3, 0xa3 }, - { 0xf4, 0xa4 }, - { 0xf5, 0xa5 }, - { 0xf6, 0xa6 }, - { 0xf7, 0xa7 }, - { 0xf8, 0xa8 }, - { 0xf9, 0xa9 }, - { 0xfa, 0xaa }, - { 0xfb, 0xab }, - { 0xfc, 0xac }, - { 0xfe, 0xae }, - { 0xff, 0xaf } - }; + return onigenc_apply_all_case_fold_with_map( + sizeof(CaseFoldMap)/sizeof(OnigPairCaseFoldCodes), CaseFoldMap, 0, + flag, f, arg); +} - if (flag == ONIGENC_AMBIGUOUS_MATCH_ASCII_CASE) { - *ccs = OnigAsciiPairAmbigCodes; - return 52; - } - if (flag == ONIGENC_AMBIGUOUS_MATCH_NONASCII_CASE) { - *ccs = cc; - return sizeof(cc) / sizeof(OnigPairAmbigCodes); - } - else - return 0; +static int +get_case_fold_codes_by_str(OnigCaseFoldType flag, + const OnigUChar* p, const OnigUChar* end, + OnigCaseFoldCodeItem items[]) +{ + return onigenc_get_case_fold_codes_by_str_with_map( + sizeof(CaseFoldMap)/sizeof(OnigPairCaseFoldCodes), CaseFoldMap, 0, + flag, p, end, items); } OnigEncodingType OnigEncodingISO_8859_5 = { @@ -271,25 +211,15 @@ OnigEncodingType OnigEncodingISO_8859_5 = { "ISO-8859-5", /* name */ 1, /* max enc length */ 1, /* min enc length */ - (ONIGENC_AMBIGUOUS_MATCH_ASCII_CASE | - ONIGENC_AMBIGUOUS_MATCH_NONASCII_CASE ), - { - (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 */ - }, onigenc_is_mbc_newline_0x0a, onigenc_single_byte_mbc_to_code, onigenc_single_byte_code_to_mbclen, onigenc_single_byte_code_to_mbc, - iso_8859_5_mbc_to_normalize, - iso_8859_5_is_mbc_ambiguous, - iso_8859_5_get_all_pair_ambig_codes, - onigenc_nothing_get_all_comp_ambig_codes, - iso_8859_5_is_code_ctype, + mbc_case_fold, + apply_all_case_fold, + get_case_fold_codes_by_str, + onigenc_minimum_property_name_to_ctype, + is_code_ctype, onigenc_not_support_get_ctype_code_range, onigenc_single_byte_left_adjust_char_head, onigenc_always_true_is_allowed_reverse_match |