diff options
-rw-r--r-- | charclass_invlists.h | 3224 | ||||
-rw-r--r-- | lib/unicore/mktables | 35 | ||||
-rw-r--r-- | perl.c | 2 | ||||
-rw-r--r-- | regcharclass.h | 2 | ||||
-rw-r--r-- | regen/mk_invlists.pl | 13 | ||||
-rw-r--r-- | regexec.c | 2 |
6 files changed, 1644 insertions, 1634 deletions
diff --git a/charclass_invlists.h b/charclass_invlists.h index 19c48ee0d8..f688eefe0a 100644 --- a/charclass_invlists.h +++ b/charclass_invlists.h @@ -312,7 +312,387 @@ static const UV Cased_invlist[] = { /* for ASCII/Latin1 */ 0x1F18A }; -static const UV Grapheme_Cluster_Break_invlist[] = { /* for ASCII/Latin1 */ +#endif /* defined(PERL_IN_PERL_C) */ + +#if defined(PERL_IN_REGCOMP_C) + +static const UV NonL1_Perl_Non_Final_Folds_invlist[] = { /* for ASCII/Latin1 */ + 45, /* Number of elements */ + 148565664, /* Version and data structure type */ + 1, /* 0 if the list starts at 0; + 1 if it starts at the element beyond 0 */ + 0x0, + 0x2BC, + 0x2BD, + 0x308, + 0x309, + 0x313, + 0x314, + 0x342, + 0x343, + 0x3AC, + 0x3AD, + 0x3AE, + 0x3AF, + 0x3B1, + 0x3B2, + 0x3B7, + 0x3B8, + 0x3B9, + 0x3BA, + 0x3C1, + 0x3C2, + 0x3C5, + 0x3C6, + 0x3C9, + 0x3CA, + 0x3CE, + 0x3CF, + 0x565, + 0x566, + 0x574, + 0x575, + 0x57E, + 0x57F, + 0x1F00, + 0x1F08, + 0x1F20, + 0x1F28, + 0x1F60, + 0x1F68, + 0x1F70, + 0x1F71, + 0x1F74, + 0x1F75, + 0x1F7C, + 0x1F7D +}; + +static const UV _Perl_Any_Folds_invlist[] = { /* for ASCII/Latin1 */ + 247, /* Number of elements */ + 148565664, /* Version and data structure type */ + 1, /* 0 if the list starts at 0; + 1 if it starts at the element beyond 0 */ + 0x0, + 0x41, + 0x5B, + 0x61, + 0x7B, + 0xB5, + 0xB6, + 0xC0, + 0xD7, + 0xD8, + 0xF7, + 0xF8, + 0x131, + 0x132, + 0x138, + 0x139, + 0x18D, + 0x18E, + 0x19B, + 0x19C, + 0x1AA, + 0x1AC, + 0x1BA, + 0x1BC, + 0x1BE, + 0x1BF, + 0x1C0, + 0x1C4, + 0x221, + 0x222, + 0x234, + 0x23A, + 0x255, + 0x256, + 0x258, + 0x259, + 0x25A, + 0x25B, + 0x25D, + 0x260, + 0x262, + 0x263, + 0x264, + 0x265, + 0x267, + 0x268, + 0x26A, + 0x26B, + 0x26D, + 0x26F, + 0x270, + 0x271, + 0x273, + 0x275, + 0x276, + 0x27D, + 0x27E, + 0x280, + 0x281, + 0x283, + 0x284, + 0x287, + 0x28D, + 0x292, + 0x293, + 0x29D, + 0x29F, + 0x2BC, + 0x2BD, + 0x2BE, + 0x2BF, + 0x300, + 0x302, + 0x307, + 0x309, + 0x30A, + 0x30B, + 0x30C, + 0x30D, + 0x313, + 0x314, + 0x331, + 0x332, + 0x342, + 0x343, + 0x345, + 0x346, + 0x370, + 0x374, + 0x376, + 0x378, + 0x37B, + 0x37E, + 0x37F, + 0x380, + 0x386, + 0x387, + 0x388, + 0x38B, + 0x38C, + 0x38D, + 0x38E, + 0x3A2, + 0x3A3, + 0x3D2, + 0x3D5, + 0x3F6, + 0x3F7, + 0x3FC, + 0x3FD, + 0x482, + 0x48A, + 0x530, + 0x531, + 0x557, + 0x561, + 0x588, + 0x10A0, + 0x10C6, + 0x10C7, + 0x10C8, + 0x10CD, + 0x10CE, + 0x13A0, + 0x13F6, + 0x13F8, + 0x13FE, + 0x1D79, + 0x1D7A, + 0x1D7D, + 0x1D7E, + 0x1E00, + 0x1E9C, + 0x1E9E, + 0x1E9F, + 0x1EA0, + 0x1F16, + 0x1F18, + 0x1F1E, + 0x1F20, + 0x1F46, + 0x1F48, + 0x1F4E, + 0x1F50, + 0x1F58, + 0x1F59, + 0x1F5A, + 0x1F5B, + 0x1F5C, + 0x1F5D, + 0x1F5E, + 0x1F5F, + 0x1F7E, + 0x1F80, + 0x1FB5, + 0x1FB6, + 0x1FBD, + 0x1FBE, + 0x1FBF, + 0x1FC2, + 0x1FC5, + 0x1FC6, + 0x1FCD, + 0x1FD0, + 0x1FD4, + 0x1FD6, + 0x1FDC, + 0x1FE0, + 0x1FED, + 0x1FF2, + 0x1FF5, + 0x1FF6, + 0x1FFD, + 0x2126, + 0x2127, + 0x212A, + 0x212C, + 0x2132, + 0x2133, + 0x214E, + 0x214F, + 0x2160, + 0x2180, + 0x2183, + 0x2185, + 0x24B6, + 0x24EA, + 0x2C00, + 0x2C2F, + 0x2C30, + 0x2C5F, + 0x2C60, + 0x2C71, + 0x2C72, + 0x2C74, + 0x2C75, + 0x2C77, + 0x2C7E, + 0x2CE4, + 0x2CEB, + 0x2CEF, + 0x2CF2, + 0x2CF4, + 0x2D00, + 0x2D26, + 0x2D27, + 0x2D28, + 0x2D2D, + 0x2D2E, + 0xA640, + 0xA66E, + 0xA680, + 0xA69C, + 0xA722, + 0xA730, + 0xA732, + 0xA770, + 0xA779, + 0xA788, + 0xA78B, + 0xA78E, + 0xA790, + 0xA794, + 0xA796, + 0xA7AE, + 0xA7B0, + 0xA7B8, + 0xAB53, + 0xAB54, + 0xAB70, + 0xABC0, + 0xFB00, + 0xFB07, + 0xFB13, + 0xFB18, + 0xFF21, + 0xFF3B, + 0xFF41, + 0xFF5B, + 0x10400, + 0x10450, + 0x10C80, + 0x10CB3, + 0x10CC0, + 0x10CF3, + 0x118A0, + 0x118E0 +}; + +static const UV _Perl_Folds_To_Multi_Char_invlist[] = { /* for ASCII/Latin1 */ + 59, /* Number of elements */ + 148565664, /* Version and data structure type */ + 1, /* 0 if the list starts at 0; + 1 if it starts at the element beyond 0 */ + 0x0, + 0xDF, + 0xE0, + 0x130, + 0x131, + 0x149, + 0x14A, + 0x1F0, + 0x1F1, + 0x390, + 0x391, + 0x3B0, + 0x3B1, + 0x587, + 0x588, + 0x1E96, + 0x1E9B, + 0x1E9E, + 0x1E9F, + 0x1F50, + 0x1F51, + 0x1F52, + 0x1F53, + 0x1F54, + 0x1F55, + 0x1F56, + 0x1F57, + 0x1F80, + 0x1FB0, + 0x1FB2, + 0x1FB5, + 0x1FB6, + 0x1FB8, + 0x1FBC, + 0x1FBD, + 0x1FC2, + 0x1FC5, + 0x1FC6, + 0x1FC8, + 0x1FCC, + 0x1FCD, + 0x1FD2, + 0x1FD4, + 0x1FD6, + 0x1FD8, + 0x1FE2, + 0x1FE5, + 0x1FE6, + 0x1FE8, + 0x1FF2, + 0x1FF5, + 0x1FF6, + 0x1FF8, + 0x1FFC, + 0x1FFD, + 0xFB00, + 0xFB07, + 0xFB13, + 0xFB18 +}; + +#endif /* defined(PERL_IN_REGCOMP_C) */ + +#if defined(PERL_IN_PERL_C) + +static const UV _Perl_GCB_invlist[] = { /* for ASCII/Latin1 */ 1502, /* Number of elements */ 148565664, /* Version and data structure type */ 0, /* 0 if the list starts at 0; @@ -1844,7 +2224,7 @@ typedef enum { GCB_V = 13 } GCB_enum; -static const GCB_enum Grapheme_Cluster_Break_invmap[] = { /* for ASCII/Latin1 */ +static const GCB_enum _Perl_GCB_invmap[] = { /* for ASCII/Latin1 */ GCB_Control, GCB_LF, GCB_Control, @@ -3351,382 +3731,6 @@ static const GCB_enum Grapheme_Cluster_Break_invmap[] = { /* for ASCII/Latin1 */ #endif /* defined(PERL_IN_REGEXEC_C) */ -#if defined(PERL_IN_REGCOMP_C) - -static const UV NonL1_Perl_Non_Final_Folds_invlist[] = { /* for ASCII/Latin1 */ - 45, /* Number of elements */ - 148565664, /* Version and data structure type */ - 1, /* 0 if the list starts at 0; - 1 if it starts at the element beyond 0 */ - 0x0, - 0x2BC, - 0x2BD, - 0x308, - 0x309, - 0x313, - 0x314, - 0x342, - 0x343, - 0x3AC, - 0x3AD, - 0x3AE, - 0x3AF, - 0x3B1, - 0x3B2, - 0x3B7, - 0x3B8, - 0x3B9, - 0x3BA, - 0x3C1, - 0x3C2, - 0x3C5, - 0x3C6, - 0x3C9, - 0x3CA, - 0x3CE, - 0x3CF, - 0x565, - 0x566, - 0x574, - 0x575, - 0x57E, - 0x57F, - 0x1F00, - 0x1F08, - 0x1F20, - 0x1F28, - 0x1F60, - 0x1F68, - 0x1F70, - 0x1F71, - 0x1F74, - 0x1F75, - 0x1F7C, - 0x1F7D -}; - -static const UV _Perl_Any_Folds_invlist[] = { /* for ASCII/Latin1 */ - 247, /* Number of elements */ - 148565664, /* Version and data structure type */ - 1, /* 0 if the list starts at 0; - 1 if it starts at the element beyond 0 */ - 0x0, - 0x41, - 0x5B, - 0x61, - 0x7B, - 0xB5, - 0xB6, - 0xC0, - 0xD7, - 0xD8, - 0xF7, - 0xF8, - 0x131, - 0x132, - 0x138, - 0x139, - 0x18D, - 0x18E, - 0x19B, - 0x19C, - 0x1AA, - 0x1AC, - 0x1BA, - 0x1BC, - 0x1BE, - 0x1BF, - 0x1C0, - 0x1C4, - 0x221, - 0x222, - 0x234, - 0x23A, - 0x255, - 0x256, - 0x258, - 0x259, - 0x25A, - 0x25B, - 0x25D, - 0x260, - 0x262, - 0x263, - 0x264, - 0x265, - 0x267, - 0x268, - 0x26A, - 0x26B, - 0x26D, - 0x26F, - 0x270, - 0x271, - 0x273, - 0x275, - 0x276, - 0x27D, - 0x27E, - 0x280, - 0x281, - 0x283, - 0x284, - 0x287, - 0x28D, - 0x292, - 0x293, - 0x29D, - 0x29F, - 0x2BC, - 0x2BD, - 0x2BE, - 0x2BF, - 0x300, - 0x302, - 0x307, - 0x309, - 0x30A, - 0x30B, - 0x30C, - 0x30D, - 0x313, - 0x314, - 0x331, - 0x332, - 0x342, - 0x343, - 0x345, - 0x346, - 0x370, - 0x374, - 0x376, - 0x378, - 0x37B, - 0x37E, - 0x37F, - 0x380, - 0x386, - 0x387, - 0x388, - 0x38B, - 0x38C, - 0x38D, - 0x38E, - 0x3A2, - 0x3A3, - 0x3D2, - 0x3D5, - 0x3F6, - 0x3F7, - 0x3FC, - 0x3FD, - 0x482, - 0x48A, - 0x530, - 0x531, - 0x557, - 0x561, - 0x588, - 0x10A0, - 0x10C6, - 0x10C7, - 0x10C8, - 0x10CD, - 0x10CE, - 0x13A0, - 0x13F6, - 0x13F8, - 0x13FE, - 0x1D79, - 0x1D7A, - 0x1D7D, - 0x1D7E, - 0x1E00, - 0x1E9C, - 0x1E9E, - 0x1E9F, - 0x1EA0, - 0x1F16, - 0x1F18, - 0x1F1E, - 0x1F20, - 0x1F46, - 0x1F48, - 0x1F4E, - 0x1F50, - 0x1F58, - 0x1F59, - 0x1F5A, - 0x1F5B, - 0x1F5C, - 0x1F5D, - 0x1F5E, - 0x1F5F, - 0x1F7E, - 0x1F80, - 0x1FB5, - 0x1FB6, - 0x1FBD, - 0x1FBE, - 0x1FBF, - 0x1FC2, - 0x1FC5, - 0x1FC6, - 0x1FCD, - 0x1FD0, - 0x1FD4, - 0x1FD6, - 0x1FDC, - 0x1FE0, - 0x1FED, - 0x1FF2, - 0x1FF5, - 0x1FF6, - 0x1FFD, - 0x2126, - 0x2127, - 0x212A, - 0x212C, - 0x2132, - 0x2133, - 0x214E, - 0x214F, - 0x2160, - 0x2180, - 0x2183, - 0x2185, - 0x24B6, - 0x24EA, - 0x2C00, - 0x2C2F, - 0x2C30, - 0x2C5F, - 0x2C60, - 0x2C71, - 0x2C72, - 0x2C74, - 0x2C75, - 0x2C77, - 0x2C7E, - 0x2CE4, - 0x2CEB, - 0x2CEF, - 0x2CF2, - 0x2CF4, - 0x2D00, - 0x2D26, - 0x2D27, - 0x2D28, - 0x2D2D, - 0x2D2E, - 0xA640, - 0xA66E, - 0xA680, - 0xA69C, - 0xA722, - 0xA730, - 0xA732, - 0xA770, - 0xA779, - 0xA788, - 0xA78B, - 0xA78E, - 0xA790, - 0xA794, - 0xA796, - 0xA7AE, - 0xA7B0, - 0xA7B8, - 0xAB53, - 0xAB54, - 0xAB70, - 0xABC0, - 0xFB00, - 0xFB07, - 0xFB13, - 0xFB18, - 0xFF21, - 0xFF3B, - 0xFF41, - 0xFF5B, - 0x10400, - 0x10450, - 0x10C80, - 0x10CB3, - 0x10CC0, - 0x10CF3, - 0x118A0, - 0x118E0 -}; - -static const UV _Perl_Folds_To_Multi_Char_invlist[] = { /* for ASCII/Latin1 */ - 59, /* Number of elements */ - 148565664, /* Version and data structure type */ - 1, /* 0 if the list starts at 0; - 1 if it starts at the element beyond 0 */ - 0x0, - 0xDF, - 0xE0, - 0x130, - 0x131, - 0x149, - 0x14A, - 0x1F0, - 0x1F1, - 0x390, - 0x391, - 0x3B0, - 0x3B1, - 0x587, - 0x588, - 0x1E96, - 0x1E9B, - 0x1E9E, - 0x1E9F, - 0x1F50, - 0x1F51, - 0x1F52, - 0x1F53, - 0x1F54, - 0x1F55, - 0x1F56, - 0x1F57, - 0x1F80, - 0x1FB0, - 0x1FB2, - 0x1FB5, - 0x1FB6, - 0x1FB8, - 0x1FBC, - 0x1FBD, - 0x1FC2, - 0x1FC5, - 0x1FC6, - 0x1FC8, - 0x1FCC, - 0x1FCD, - 0x1FD2, - 0x1FD4, - 0x1FD6, - 0x1FD8, - 0x1FE2, - 0x1FE5, - 0x1FE6, - 0x1FE8, - 0x1FF2, - 0x1FF5, - 0x1FF6, - 0x1FF8, - 0x1FFC, - 0x1FFD, - 0xFB00, - 0xFB07, - 0xFB13, - 0xFB18 -}; - -#endif /* defined(PERL_IN_REGCOMP_C) */ - #if defined(PERL_IN_UTF8_C) static const UV _Perl_IDCont_invlist[] = { /* for ASCII/Latin1 */ @@ -24947,7 +24951,419 @@ static const UV Cased_invlist[] = { /* for EBCDIC 1047 */ 0x1F18A }; -static const UV Grapheme_Cluster_Break_invlist[] = { /* for EBCDIC 1047 */ +#endif /* defined(PERL_IN_PERL_C) */ + +#if defined(PERL_IN_REGCOMP_C) + +static const UV NonL1_Perl_Non_Final_Folds_invlist[] = { /* for EBCDIC 1047 */ + 45, /* Number of elements */ + 148565664, /* Version and data structure type */ + 1, /* 0 if the list starts at 0; + 1 if it starts at the element beyond 0 */ + 0x0, + 0x2BC, + 0x2BD, + 0x308, + 0x309, + 0x313, + 0x314, + 0x342, + 0x343, + 0x3AC, + 0x3AD, + 0x3AE, + 0x3AF, + 0x3B1, + 0x3B2, + 0x3B7, + 0x3B8, + 0x3B9, + 0x3BA, + 0x3C1, + 0x3C2, + 0x3C5, + 0x3C6, + 0x3C9, + 0x3CA, + 0x3CE, + 0x3CF, + 0x565, + 0x566, + 0x574, + 0x575, + 0x57E, + 0x57F, + 0x1F00, + 0x1F08, + 0x1F20, + 0x1F28, + 0x1F60, + 0x1F68, + 0x1F70, + 0x1F71, + 0x1F74, + 0x1F75, + 0x1F7C, + 0x1F7D +}; + +static const UV _Perl_Any_Folds_invlist[] = { /* for EBCDIC 1047 */ + 279, /* Number of elements */ + 148565664, /* Version and data structure type */ + 1, /* 0 if the list starts at 0; + 1 if it starts at the element beyond 0 */ + 0x0, + 0x42, + 0x4A, + 0x51, + 0x5A, + 0x62, + 0x6A, + 0x70, + 0x79, + 0x80, + 0x8A, + 0x8C, + 0x8F, + 0x91, + 0x9A, + 0x9C, + 0x9D, + 0x9E, + 0x9F, + 0xA0, + 0xA1, + 0xA2, + 0xAA, + 0xAC, + 0xAD, + 0xAE, + 0xAF, + 0xBA, + 0xBB, + 0xC1, + 0xCA, + 0xCB, + 0xD0, + 0xD1, + 0xDA, + 0xDB, + 0xE0, + 0xE2, + 0xEA, + 0xEB, + 0xF0, + 0xFB, + 0xFF, + 0x100, + 0x131, + 0x132, + 0x138, + 0x139, + 0x18D, + 0x18E, + 0x19B, + 0x19C, + 0x1AA, + 0x1AC, + 0x1BA, + 0x1BC, + 0x1BE, + 0x1BF, + 0x1C0, + 0x1C4, + 0x221, + 0x222, + 0x234, + 0x23A, + 0x255, + 0x256, + 0x258, + 0x259, + 0x25A, + 0x25B, + 0x25D, + 0x260, + 0x262, + 0x263, + 0x264, + 0x265, + 0x267, + 0x268, + 0x26A, + 0x26B, + 0x26D, + 0x26F, + 0x270, + 0x271, + 0x273, + 0x275, + 0x276, + 0x27D, + 0x27E, + 0x280, + 0x281, + 0x283, + 0x284, + 0x287, + 0x28D, + 0x292, + 0x293, + 0x29D, + 0x29F, + 0x2BC, + 0x2BD, + 0x2BE, + 0x2BF, + 0x300, + 0x302, + 0x307, + 0x309, + 0x30A, + 0x30B, + 0x30C, + 0x30D, + 0x313, + 0x314, + 0x331, + 0x332, + 0x342, + 0x343, + 0x345, + 0x346, + 0x370, + 0x374, + 0x376, + 0x378, + 0x37B, + 0x37E, + 0x37F, + 0x380, + 0x386, + 0x387, + 0x388, + 0x38B, + 0x38C, + 0x38D, + 0x38E, + 0x3A2, + 0x3A3, + 0x3D2, + 0x3D5, + 0x3F6, + 0x3F7, + 0x3FC, + 0x3FD, + 0x482, + 0x48A, + 0x530, + 0x531, + 0x557, + 0x561, + 0x588, + 0x10A0, + 0x10C6, + 0x10C7, + 0x10C8, + 0x10CD, + 0x10CE, + 0x13A0, + 0x13F6, + 0x13F8, + 0x13FE, + 0x1D79, + 0x1D7A, + 0x1D7D, + 0x1D7E, + 0x1E00, + 0x1E9C, + 0x1E9E, + 0x1E9F, + 0x1EA0, + 0x1F16, + 0x1F18, + 0x1F1E, + 0x1F20, + 0x1F46, + 0x1F48, + 0x1F4E, + 0x1F50, + 0x1F58, + 0x1F59, + 0x1F5A, + 0x1F5B, + 0x1F5C, + 0x1F5D, + 0x1F5E, + 0x1F5F, + 0x1F7E, + 0x1F80, + 0x1FB5, + 0x1FB6, + 0x1FBD, + 0x1FBE, + 0x1FBF, + 0x1FC2, + 0x1FC5, + 0x1FC6, + 0x1FCD, + 0x1FD0, + 0x1FD4, + 0x1FD6, + 0x1FDC, + 0x1FE0, + 0x1FED, + 0x1FF2, + 0x1FF5, + 0x1FF6, + 0x1FFD, + 0x2126, + 0x2127, + 0x212A, + 0x212C, + 0x2132, + 0x2133, + 0x214E, + 0x214F, + 0x2160, + 0x2180, + 0x2183, + 0x2185, + 0x24B6, + 0x24EA, + 0x2C00, + 0x2C2F, + 0x2C30, + 0x2C5F, + 0x2C60, + 0x2C71, + 0x2C72, + 0x2C74, + 0x2C75, + 0x2C77, + 0x2C7E, + 0x2CE4, + 0x2CEB, + 0x2CEF, + 0x2CF2, + 0x2CF4, + 0x2D00, + 0x2D26, + 0x2D27, + 0x2D28, + 0x2D2D, + 0x2D2E, + 0xA640, + 0xA66E, + 0xA680, + 0xA69C, + 0xA722, + 0xA730, + 0xA732, + 0xA770, + 0xA779, + 0xA788, + 0xA78B, + 0xA78E, + 0xA790, + 0xA794, + 0xA796, + 0xA7AE, + 0xA7B0, + 0xA7B8, + 0xAB53, + 0xAB54, + 0xAB70, + 0xABC0, + 0xFB00, + 0xFB07, + 0xFB13, + 0xFB18, + 0xFF21, + 0xFF3B, + 0xFF41, + 0xFF5B, + 0x10400, + 0x10450, + 0x10C80, + 0x10CB3, + 0x10CC0, + 0x10CF3, + 0x118A0, + 0x118E0 +}; + +static const UV _Perl_Folds_To_Multi_Char_invlist[] = { /* for EBCDIC 1047 */ + 59, /* Number of elements */ + 148565664, /* Version and data structure type */ + 1, /* 0 if the list starts at 0; + 1 if it starts at the element beyond 0 */ + 0x0, + 0x59, + 0x5A, + 0x130, + 0x131, + 0x149, + 0x14A, + 0x1F0, + 0x1F1, + 0x390, + 0x391, + 0x3B0, + 0x3B1, + 0x587, + 0x588, + 0x1E96, + 0x1E9B, + 0x1E9E, + 0x1E9F, + 0x1F50, + 0x1F51, + 0x1F52, + 0x1F53, + 0x1F54, + 0x1F55, + 0x1F56, + 0x1F57, + 0x1F80, + 0x1FB0, + 0x1FB2, + 0x1FB5, + 0x1FB6, + 0x1FB8, + 0x1FBC, + 0x1FBD, + 0x1FC2, + 0x1FC5, + 0x1FC6, + 0x1FC8, + 0x1FCC, + 0x1FCD, + 0x1FD2, + 0x1FD4, + 0x1FD6, + 0x1FD8, + 0x1FE2, + 0x1FE5, + 0x1FE6, + 0x1FE8, + 0x1FF2, + 0x1FF5, + 0x1FF6, + 0x1FF8, + 0x1FFC, + 0x1FFD, + 0xFB00, + 0xFB07, + 0xFB13, + 0xFB18 +}; + +#endif /* defined(PERL_IN_REGCOMP_C) */ + +#if defined(PERL_IN_PERL_C) + +static const UV _Perl_GCB_invlist[] = { /* for EBCDIC 1047 */ 1502, /* Number of elements */ 148565664, /* Version and data structure type */ 0, /* 0 if the list starts at 0; @@ -26479,7 +26895,7 @@ typedef enum { GCB_V = 13 } GCB_enum; -static const GCB_enum Grapheme_Cluster_Break_invmap[] = { /* for EBCDIC 1047 */ +static const GCB_enum _Perl_GCB_invmap[] = { /* for EBCDIC 1047 */ GCB_Control, GCB_CR, GCB_Control, @@ -27986,414 +28402,6 @@ static const GCB_enum Grapheme_Cluster_Break_invmap[] = { /* for EBCDIC 1047 */ #endif /* defined(PERL_IN_REGEXEC_C) */ -#if defined(PERL_IN_REGCOMP_C) - -static const UV NonL1_Perl_Non_Final_Folds_invlist[] = { /* for EBCDIC 1047 */ - 45, /* Number of elements */ - 148565664, /* Version and data structure type */ - 1, /* 0 if the list starts at 0; - 1 if it starts at the element beyond 0 */ - 0x0, - 0x2BC, - 0x2BD, - 0x308, - 0x309, - 0x313, - 0x314, - 0x342, - 0x343, - 0x3AC, - 0x3AD, - 0x3AE, - 0x3AF, - 0x3B1, - 0x3B2, - 0x3B7, - 0x3B8, - 0x3B9, - 0x3BA, - 0x3C1, - 0x3C2, - 0x3C5, - 0x3C6, - 0x3C9, - 0x3CA, - 0x3CE, - 0x3CF, - 0x565, - 0x566, - 0x574, - 0x575, - 0x57E, - 0x57F, - 0x1F00, - 0x1F08, - 0x1F20, - 0x1F28, - 0x1F60, - 0x1F68, - 0x1F70, - 0x1F71, - 0x1F74, - 0x1F75, - 0x1F7C, - 0x1F7D -}; - -static const UV _Perl_Any_Folds_invlist[] = { /* for EBCDIC 1047 */ - 279, /* Number of elements */ - 148565664, /* Version and data structure type */ - 1, /* 0 if the list starts at 0; - 1 if it starts at the element beyond 0 */ - 0x0, - 0x42, - 0x4A, - 0x51, - 0x5A, - 0x62, - 0x6A, - 0x70, - 0x79, - 0x80, - 0x8A, - 0x8C, - 0x8F, - 0x91, - 0x9A, - 0x9C, - 0x9D, - 0x9E, - 0x9F, - 0xA0, - 0xA1, - 0xA2, - 0xAA, - 0xAC, - 0xAD, - 0xAE, - 0xAF, - 0xBA, - 0xBB, - 0xC1, - 0xCA, - 0xCB, - 0xD0, - 0xD1, - 0xDA, - 0xDB, - 0xE0, - 0xE2, - 0xEA, - 0xEB, - 0xF0, - 0xFB, - 0xFF, - 0x100, - 0x131, - 0x132, - 0x138, - 0x139, - 0x18D, - 0x18E, - 0x19B, - 0x19C, - 0x1AA, - 0x1AC, - 0x1BA, - 0x1BC, - 0x1BE, - 0x1BF, - 0x1C0, - 0x1C4, - 0x221, - 0x222, - 0x234, - 0x23A, - 0x255, - 0x256, - 0x258, - 0x259, - 0x25A, - 0x25B, - 0x25D, - 0x260, - 0x262, - 0x263, - 0x264, - 0x265, - 0x267, - 0x268, - 0x26A, - 0x26B, - 0x26D, - 0x26F, - 0x270, - 0x271, - 0x273, - 0x275, - 0x276, - 0x27D, - 0x27E, - 0x280, - 0x281, - 0x283, - 0x284, - 0x287, - 0x28D, - 0x292, - 0x293, - 0x29D, - 0x29F, - 0x2BC, - 0x2BD, - 0x2BE, - 0x2BF, - 0x300, - 0x302, - 0x307, - 0x309, - 0x30A, - 0x30B, - 0x30C, - 0x30D, - 0x313, - 0x314, - 0x331, - 0x332, - 0x342, - 0x343, - 0x345, - 0x346, - 0x370, - 0x374, - 0x376, - 0x378, - 0x37B, - 0x37E, - 0x37F, - 0x380, - 0x386, - 0x387, - 0x388, - 0x38B, - 0x38C, - 0x38D, - 0x38E, - 0x3A2, - 0x3A3, - 0x3D2, - 0x3D5, - 0x3F6, - 0x3F7, - 0x3FC, - 0x3FD, - 0x482, - 0x48A, - 0x530, - 0x531, - 0x557, - 0x561, - 0x588, - 0x10A0, - 0x10C6, - 0x10C7, - 0x10C8, - 0x10CD, - 0x10CE, - 0x13A0, - 0x13F6, - 0x13F8, - 0x13FE, - 0x1D79, - 0x1D7A, - 0x1D7D, - 0x1D7E, - 0x1E00, - 0x1E9C, - 0x1E9E, - 0x1E9F, - 0x1EA0, - 0x1F16, - 0x1F18, - 0x1F1E, - 0x1F20, - 0x1F46, - 0x1F48, - 0x1F4E, - 0x1F50, - 0x1F58, - 0x1F59, - 0x1F5A, - 0x1F5B, - 0x1F5C, - 0x1F5D, - 0x1F5E, - 0x1F5F, - 0x1F7E, - 0x1F80, - 0x1FB5, - 0x1FB6, - 0x1FBD, - 0x1FBE, - 0x1FBF, - 0x1FC2, - 0x1FC5, - 0x1FC6, - 0x1FCD, - 0x1FD0, - 0x1FD4, - 0x1FD6, - 0x1FDC, - 0x1FE0, - 0x1FED, - 0x1FF2, - 0x1FF5, - 0x1FF6, - 0x1FFD, - 0x2126, - 0x2127, - 0x212A, - 0x212C, - 0x2132, - 0x2133, - 0x214E, - 0x214F, - 0x2160, - 0x2180, - 0x2183, - 0x2185, - 0x24B6, - 0x24EA, - 0x2C00, - 0x2C2F, - 0x2C30, - 0x2C5F, - 0x2C60, - 0x2C71, - 0x2C72, - 0x2C74, - 0x2C75, - 0x2C77, - 0x2C7E, - 0x2CE4, - 0x2CEB, - 0x2CEF, - 0x2CF2, - 0x2CF4, - 0x2D00, - 0x2D26, - 0x2D27, - 0x2D28, - 0x2D2D, - 0x2D2E, - 0xA640, - 0xA66E, - 0xA680, - 0xA69C, - 0xA722, - 0xA730, - 0xA732, - 0xA770, - 0xA779, - 0xA788, - 0xA78B, - 0xA78E, - 0xA790, - 0xA794, - 0xA796, - 0xA7AE, - 0xA7B0, - 0xA7B8, - 0xAB53, - 0xAB54, - 0xAB70, - 0xABC0, - 0xFB00, - 0xFB07, - 0xFB13, - 0xFB18, - 0xFF21, - 0xFF3B, - 0xFF41, - 0xFF5B, - 0x10400, - 0x10450, - 0x10C80, - 0x10CB3, - 0x10CC0, - 0x10CF3, - 0x118A0, - 0x118E0 -}; - -static const UV _Perl_Folds_To_Multi_Char_invlist[] = { /* for EBCDIC 1047 */ - 59, /* Number of elements */ - 148565664, /* Version and data structure type */ - 1, /* 0 if the list starts at 0; - 1 if it starts at the element beyond 0 */ - 0x0, - 0x59, - 0x5A, - 0x130, - 0x131, - 0x149, - 0x14A, - 0x1F0, - 0x1F1, - 0x390, - 0x391, - 0x3B0, - 0x3B1, - 0x587, - 0x588, - 0x1E96, - 0x1E9B, - 0x1E9E, - 0x1E9F, - 0x1F50, - 0x1F51, - 0x1F52, - 0x1F53, - 0x1F54, - 0x1F55, - 0x1F56, - 0x1F57, - 0x1F80, - 0x1FB0, - 0x1FB2, - 0x1FB5, - 0x1FB6, - 0x1FB8, - 0x1FBC, - 0x1FBD, - 0x1FC2, - 0x1FC5, - 0x1FC6, - 0x1FC8, - 0x1FCC, - 0x1FCD, - 0x1FD2, - 0x1FD4, - 0x1FD6, - 0x1FD8, - 0x1FE2, - 0x1FE5, - 0x1FE6, - 0x1FE8, - 0x1FF2, - 0x1FF5, - 0x1FF6, - 0x1FF8, - 0x1FFC, - 0x1FFD, - 0xFB00, - 0xFB07, - 0xFB13, - 0xFB18 -}; - -#endif /* defined(PERL_IN_REGCOMP_C) */ - #if defined(PERL_IN_UTF8_C) static const UV _Perl_IDCont_invlist[] = { /* for EBCDIC 1047 */ @@ -49924,7 +49932,415 @@ static const UV Cased_invlist[] = { /* for EBCDIC 037 */ 0x1F18A }; -static const UV Grapheme_Cluster_Break_invlist[] = { /* for EBCDIC 037 */ +#endif /* defined(PERL_IN_PERL_C) */ + +#if defined(PERL_IN_REGCOMP_C) + +static const UV NonL1_Perl_Non_Final_Folds_invlist[] = { /* for EBCDIC 037 */ + 45, /* Number of elements */ + 148565664, /* Version and data structure type */ + 1, /* 0 if the list starts at 0; + 1 if it starts at the element beyond 0 */ + 0x0, + 0x2BC, + 0x2BD, + 0x308, + 0x309, + 0x313, + 0x314, + 0x342, + 0x343, + 0x3AC, + 0x3AD, + 0x3AE, + 0x3AF, + 0x3B1, + 0x3B2, + 0x3B7, + 0x3B8, + 0x3B9, + 0x3BA, + 0x3C1, + 0x3C2, + 0x3C5, + 0x3C6, + 0x3C9, + 0x3CA, + 0x3CE, + 0x3CF, + 0x565, + 0x566, + 0x574, + 0x575, + 0x57E, + 0x57F, + 0x1F00, + 0x1F08, + 0x1F20, + 0x1F28, + 0x1F60, + 0x1F68, + 0x1F70, + 0x1F71, + 0x1F74, + 0x1F75, + 0x1F7C, + 0x1F7D +}; + +static const UV _Perl_Any_Folds_invlist[] = { /* for EBCDIC 037 */ + 275, /* Number of elements */ + 148565664, /* Version and data structure type */ + 1, /* 0 if the list starts at 0; + 1 if it starts at the element beyond 0 */ + 0x0, + 0x42, + 0x4A, + 0x51, + 0x5A, + 0x62, + 0x6A, + 0x70, + 0x79, + 0x80, + 0x8A, + 0x8C, + 0x8F, + 0x91, + 0x9A, + 0x9C, + 0x9D, + 0x9E, + 0x9F, + 0xA0, + 0xA1, + 0xA2, + 0xAA, + 0xAC, + 0xAF, + 0xC1, + 0xCA, + 0xCB, + 0xD0, + 0xD1, + 0xDA, + 0xDB, + 0xE0, + 0xE2, + 0xEA, + 0xEB, + 0xF0, + 0xFB, + 0xFF, + 0x100, + 0x131, + 0x132, + 0x138, + 0x139, + 0x18D, + 0x18E, + 0x19B, + 0x19C, + 0x1AA, + 0x1AC, + 0x1BA, + 0x1BC, + 0x1BE, + 0x1BF, + 0x1C0, + 0x1C4, + 0x221, + 0x222, + 0x234, + 0x23A, + 0x255, + 0x256, + 0x258, + 0x259, + 0x25A, + 0x25B, + 0x25D, + 0x260, + 0x262, + 0x263, + 0x264, + 0x265, + 0x267, + 0x268, + 0x26A, + 0x26B, + 0x26D, + 0x26F, + 0x270, + 0x271, + 0x273, + 0x275, + 0x276, + 0x27D, + 0x27E, + 0x280, + 0x281, + 0x283, + 0x284, + 0x287, + 0x28D, + 0x292, + 0x293, + 0x29D, + 0x29F, + 0x2BC, + 0x2BD, + 0x2BE, + 0x2BF, + 0x300, + 0x302, + 0x307, + 0x309, + 0x30A, + 0x30B, + 0x30C, + 0x30D, + 0x313, + 0x314, + 0x331, + 0x332, + 0x342, + 0x343, + 0x345, + 0x346, + 0x370, + 0x374, + 0x376, + 0x378, + 0x37B, + 0x37E, + 0x37F, + 0x380, + 0x386, + 0x387, + 0x388, + 0x38B, + 0x38C, + 0x38D, + 0x38E, + 0x3A2, + 0x3A3, + 0x3D2, + 0x3D5, + 0x3F6, + 0x3F7, + 0x3FC, + 0x3FD, + 0x482, + 0x48A, + 0x530, + 0x531, + 0x557, + 0x561, + 0x588, + 0x10A0, + 0x10C6, + 0x10C7, + 0x10C8, + 0x10CD, + 0x10CE, + 0x13A0, + 0x13F6, + 0x13F8, + 0x13FE, + 0x1D79, + 0x1D7A, + 0x1D7D, + 0x1D7E, + 0x1E00, + 0x1E9C, + 0x1E9E, + 0x1E9F, + 0x1EA0, + 0x1F16, + 0x1F18, + 0x1F1E, + 0x1F20, + 0x1F46, + 0x1F48, + 0x1F4E, + 0x1F50, + 0x1F58, + 0x1F59, + 0x1F5A, + 0x1F5B, + 0x1F5C, + 0x1F5D, + 0x1F5E, + 0x1F5F, + 0x1F7E, + 0x1F80, + 0x1FB5, + 0x1FB6, + 0x1FBD, + 0x1FBE, + 0x1FBF, + 0x1FC2, + 0x1FC5, + 0x1FC6, + 0x1FCD, + 0x1FD0, + 0x1FD4, + 0x1FD6, + 0x1FDC, + 0x1FE0, + 0x1FED, + 0x1FF2, + 0x1FF5, + 0x1FF6, + 0x1FFD, + 0x2126, + 0x2127, + 0x212A, + 0x212C, + 0x2132, + 0x2133, + 0x214E, + 0x214F, + 0x2160, + 0x2180, + 0x2183, + 0x2185, + 0x24B6, + 0x24EA, + 0x2C00, + 0x2C2F, + 0x2C30, + 0x2C5F, + 0x2C60, + 0x2C71, + 0x2C72, + 0x2C74, + 0x2C75, + 0x2C77, + 0x2C7E, + 0x2CE4, + 0x2CEB, + 0x2CEF, + 0x2CF2, + 0x2CF4, + 0x2D00, + 0x2D26, + 0x2D27, + 0x2D28, + 0x2D2D, + 0x2D2E, + 0xA640, + 0xA66E, + 0xA680, + 0xA69C, + 0xA722, + 0xA730, + 0xA732, + 0xA770, + 0xA779, + 0xA788, + 0xA78B, + 0xA78E, + 0xA790, + 0xA794, + 0xA796, + 0xA7AE, + 0xA7B0, + 0xA7B8, + 0xAB53, + 0xAB54, + 0xAB70, + 0xABC0, + 0xFB00, + 0xFB07, + 0xFB13, + 0xFB18, + 0xFF21, + 0xFF3B, + 0xFF41, + 0xFF5B, + 0x10400, + 0x10450, + 0x10C80, + 0x10CB3, + 0x10CC0, + 0x10CF3, + 0x118A0, + 0x118E0 +}; + +static const UV _Perl_Folds_To_Multi_Char_invlist[] = { /* for EBCDIC 037 */ + 59, /* Number of elements */ + 148565664, /* Version and data structure type */ + 1, /* 0 if the list starts at 0; + 1 if it starts at the element beyond 0 */ + 0x0, + 0x59, + 0x5A, + 0x130, + 0x131, + 0x149, + 0x14A, + 0x1F0, + 0x1F1, + 0x390, + 0x391, + 0x3B0, + 0x3B1, + 0x587, + 0x588, + 0x1E96, + 0x1E9B, + 0x1E9E, + 0x1E9F, + 0x1F50, + 0x1F51, + 0x1F52, + 0x1F53, + 0x1F54, + 0x1F55, + 0x1F56, + 0x1F57, + 0x1F80, + 0x1FB0, + 0x1FB2, + 0x1FB5, + 0x1FB6, + 0x1FB8, + 0x1FBC, + 0x1FBD, + 0x1FC2, + 0x1FC5, + 0x1FC6, + 0x1FC8, + 0x1FCC, + 0x1FCD, + 0x1FD2, + 0x1FD4, + 0x1FD6, + 0x1FD8, + 0x1FE2, + 0x1FE5, + 0x1FE6, + 0x1FE8, + 0x1FF2, + 0x1FF5, + 0x1FF6, + 0x1FF8, + 0x1FFC, + 0x1FFD, + 0xFB00, + 0xFB07, + 0xFB13, + 0xFB18 +}; + +#endif /* defined(PERL_IN_REGCOMP_C) */ + +#if defined(PERL_IN_PERL_C) + +static const UV _Perl_GCB_invlist[] = { /* for EBCDIC 037 */ 1502, /* Number of elements */ 148565664, /* Version and data structure type */ 0, /* 0 if the list starts at 0; @@ -51456,7 +51872,7 @@ typedef enum { GCB_V = 13 } GCB_enum; -static const GCB_enum Grapheme_Cluster_Break_invmap[] = { /* for EBCDIC 037 */ +static const GCB_enum _Perl_GCB_invmap[] = { /* for EBCDIC 037 */ GCB_Control, GCB_CR, GCB_Control, @@ -52963,410 +53379,6 @@ static const GCB_enum Grapheme_Cluster_Break_invmap[] = { /* for EBCDIC 037 */ #endif /* defined(PERL_IN_REGEXEC_C) */ -#if defined(PERL_IN_REGCOMP_C) - -static const UV NonL1_Perl_Non_Final_Folds_invlist[] = { /* for EBCDIC 037 */ - 45, /* Number of elements */ - 148565664, /* Version and data structure type */ - 1, /* 0 if the list starts at 0; - 1 if it starts at the element beyond 0 */ - 0x0, - 0x2BC, - 0x2BD, - 0x308, - 0x309, - 0x313, - 0x314, - 0x342, - 0x343, - 0x3AC, - 0x3AD, - 0x3AE, - 0x3AF, - 0x3B1, - 0x3B2, - 0x3B7, - 0x3B8, - 0x3B9, - 0x3BA, - 0x3C1, - 0x3C2, - 0x3C5, - 0x3C6, - 0x3C9, - 0x3CA, - 0x3CE, - 0x3CF, - 0x565, - 0x566, - 0x574, - 0x575, - 0x57E, - 0x57F, - 0x1F00, - 0x1F08, - 0x1F20, - 0x1F28, - 0x1F60, - 0x1F68, - 0x1F70, - 0x1F71, - 0x1F74, - 0x1F75, - 0x1F7C, - 0x1F7D -}; - -static const UV _Perl_Any_Folds_invlist[] = { /* for EBCDIC 037 */ - 275, /* Number of elements */ - 148565664, /* Version and data structure type */ - 1, /* 0 if the list starts at 0; - 1 if it starts at the element beyond 0 */ - 0x0, - 0x42, - 0x4A, - 0x51, - 0x5A, - 0x62, - 0x6A, - 0x70, - 0x79, - 0x80, - 0x8A, - 0x8C, - 0x8F, - 0x91, - 0x9A, - 0x9C, - 0x9D, - 0x9E, - 0x9F, - 0xA0, - 0xA1, - 0xA2, - 0xAA, - 0xAC, - 0xAF, - 0xC1, - 0xCA, - 0xCB, - 0xD0, - 0xD1, - 0xDA, - 0xDB, - 0xE0, - 0xE2, - 0xEA, - 0xEB, - 0xF0, - 0xFB, - 0xFF, - 0x100, - 0x131, - 0x132, - 0x138, - 0x139, - 0x18D, - 0x18E, - 0x19B, - 0x19C, - 0x1AA, - 0x1AC, - 0x1BA, - 0x1BC, - 0x1BE, - 0x1BF, - 0x1C0, - 0x1C4, - 0x221, - 0x222, - 0x234, - 0x23A, - 0x255, - 0x256, - 0x258, - 0x259, - 0x25A, - 0x25B, - 0x25D, - 0x260, - 0x262, - 0x263, - 0x264, - 0x265, - 0x267, - 0x268, - 0x26A, - 0x26B, - 0x26D, - 0x26F, - 0x270, - 0x271, - 0x273, - 0x275, - 0x276, - 0x27D, - 0x27E, - 0x280, - 0x281, - 0x283, - 0x284, - 0x287, - 0x28D, - 0x292, - 0x293, - 0x29D, - 0x29F, - 0x2BC, - 0x2BD, - 0x2BE, - 0x2BF, - 0x300, - 0x302, - 0x307, - 0x309, - 0x30A, - 0x30B, - 0x30C, - 0x30D, - 0x313, - 0x314, - 0x331, - 0x332, - 0x342, - 0x343, - 0x345, - 0x346, - 0x370, - 0x374, - 0x376, - 0x378, - 0x37B, - 0x37E, - 0x37F, - 0x380, - 0x386, - 0x387, - 0x388, - 0x38B, - 0x38C, - 0x38D, - 0x38E, - 0x3A2, - 0x3A3, - 0x3D2, - 0x3D5, - 0x3F6, - 0x3F7, - 0x3FC, - 0x3FD, - 0x482, - 0x48A, - 0x530, - 0x531, - 0x557, - 0x561, - 0x588, - 0x10A0, - 0x10C6, - 0x10C7, - 0x10C8, - 0x10CD, - 0x10CE, - 0x13A0, - 0x13F6, - 0x13F8, - 0x13FE, - 0x1D79, - 0x1D7A, - 0x1D7D, - 0x1D7E, - 0x1E00, - 0x1E9C, - 0x1E9E, - 0x1E9F, - 0x1EA0, - 0x1F16, - 0x1F18, - 0x1F1E, - 0x1F20, - 0x1F46, - 0x1F48, - 0x1F4E, - 0x1F50, - 0x1F58, - 0x1F59, - 0x1F5A, - 0x1F5B, - 0x1F5C, - 0x1F5D, - 0x1F5E, - 0x1F5F, - 0x1F7E, - 0x1F80, - 0x1FB5, - 0x1FB6, - 0x1FBD, - 0x1FBE, - 0x1FBF, - 0x1FC2, - 0x1FC5, - 0x1FC6, - 0x1FCD, - 0x1FD0, - 0x1FD4, - 0x1FD6, - 0x1FDC, - 0x1FE0, - 0x1FED, - 0x1FF2, - 0x1FF5, - 0x1FF6, - 0x1FFD, - 0x2126, - 0x2127, - 0x212A, - 0x212C, - 0x2132, - 0x2133, - 0x214E, - 0x214F, - 0x2160, - 0x2180, - 0x2183, - 0x2185, - 0x24B6, - 0x24EA, - 0x2C00, - 0x2C2F, - 0x2C30, - 0x2C5F, - 0x2C60, - 0x2C71, - 0x2C72, - 0x2C74, - 0x2C75, - 0x2C77, - 0x2C7E, - 0x2CE4, - 0x2CEB, - 0x2CEF, - 0x2CF2, - 0x2CF4, - 0x2D00, - 0x2D26, - 0x2D27, - 0x2D28, - 0x2D2D, - 0x2D2E, - 0xA640, - 0xA66E, - 0xA680, - 0xA69C, - 0xA722, - 0xA730, - 0xA732, - 0xA770, - 0xA779, - 0xA788, - 0xA78B, - 0xA78E, - 0xA790, - 0xA794, - 0xA796, - 0xA7AE, - 0xA7B0, - 0xA7B8, - 0xAB53, - 0xAB54, - 0xAB70, - 0xABC0, - 0xFB00, - 0xFB07, - 0xFB13, - 0xFB18, - 0xFF21, - 0xFF3B, - 0xFF41, - 0xFF5B, - 0x10400, - 0x10450, - 0x10C80, - 0x10CB3, - 0x10CC0, - 0x10CF3, - 0x118A0, - 0x118E0 -}; - -static const UV _Perl_Folds_To_Multi_Char_invlist[] = { /* for EBCDIC 037 */ - 59, /* Number of elements */ - 148565664, /* Version and data structure type */ - 1, /* 0 if the list starts at 0; - 1 if it starts at the element beyond 0 */ - 0x0, - 0x59, - 0x5A, - 0x130, - 0x131, - 0x149, - 0x14A, - 0x1F0, - 0x1F1, - 0x390, - 0x391, - 0x3B0, - 0x3B1, - 0x587, - 0x588, - 0x1E96, - 0x1E9B, - 0x1E9E, - 0x1E9F, - 0x1F50, - 0x1F51, - 0x1F52, - 0x1F53, - 0x1F54, - 0x1F55, - 0x1F56, - 0x1F57, - 0x1F80, - 0x1FB0, - 0x1FB2, - 0x1FB5, - 0x1FB6, - 0x1FB8, - 0x1FBC, - 0x1FBD, - 0x1FC2, - 0x1FC5, - 0x1FC6, - 0x1FC8, - 0x1FCC, - 0x1FCD, - 0x1FD2, - 0x1FD4, - 0x1FD6, - 0x1FD8, - 0x1FE2, - 0x1FE5, - 0x1FE6, - 0x1FE8, - 0x1FF2, - 0x1FF5, - 0x1FF6, - 0x1FF8, - 0x1FFC, - 0x1FFD, - 0xFB00, - 0xFB07, - 0xFB13, - 0xFB18 -}; - -#endif /* defined(PERL_IN_REGCOMP_C) */ - #if defined(PERL_IN_UTF8_C) static const UV _Perl_IDCont_invlist[] = { /* for EBCDIC 037 */ @@ -74865,7 +74877,417 @@ static const UV Cased_invlist[] = { /* for EBCDIC POSIX-BC */ 0x1F18A }; -static const UV Grapheme_Cluster_Break_invlist[] = { /* for EBCDIC POSIX-BC */ +#endif /* defined(PERL_IN_PERL_C) */ + +#if defined(PERL_IN_REGCOMP_C) + +static const UV NonL1_Perl_Non_Final_Folds_invlist[] = { /* for EBCDIC POSIX-BC */ + 45, /* Number of elements */ + 148565664, /* Version and data structure type */ + 1, /* 0 if the list starts at 0; + 1 if it starts at the element beyond 0 */ + 0x0, + 0x2BC, + 0x2BD, + 0x308, + 0x309, + 0x313, + 0x314, + 0x342, + 0x343, + 0x3AC, + 0x3AD, + 0x3AE, + 0x3AF, + 0x3B1, + 0x3B2, + 0x3B7, + 0x3B8, + 0x3B9, + 0x3BA, + 0x3C1, + 0x3C2, + 0x3C5, + 0x3C6, + 0x3C9, + 0x3CA, + 0x3CE, + 0x3CF, + 0x565, + 0x566, + 0x574, + 0x575, + 0x57E, + 0x57F, + 0x1F00, + 0x1F08, + 0x1F20, + 0x1F28, + 0x1F60, + 0x1F68, + 0x1F70, + 0x1F71, + 0x1F74, + 0x1F75, + 0x1F7C, + 0x1F7D +}; + +static const UV _Perl_Any_Folds_invlist[] = { /* for EBCDIC POSIX-BC */ + 277, /* Number of elements */ + 148565664, /* Version and data structure type */ + 1, /* 0 if the list starts at 0; + 1 if it starts at the element beyond 0 */ + 0x0, + 0x42, + 0x4A, + 0x51, + 0x5A, + 0x62, + 0x6A, + 0x70, + 0x79, + 0x80, + 0x8A, + 0x8C, + 0x8F, + 0x91, + 0x9A, + 0x9C, + 0x9D, + 0x9E, + 0x9F, + 0xA0, + 0xA1, + 0xA2, + 0xAA, + 0xAC, + 0xAF, + 0xC0, + 0xCA, + 0xCB, + 0xD0, + 0xD1, + 0xDA, + 0xDB, + 0xE1, + 0xE2, + 0xEA, + 0xEB, + 0xF0, + 0xFC, + 0xFD, + 0xFE, + 0xFF, + 0x100, + 0x131, + 0x132, + 0x138, + 0x139, + 0x18D, + 0x18E, + 0x19B, + 0x19C, + 0x1AA, + 0x1AC, + 0x1BA, + 0x1BC, + 0x1BE, + 0x1BF, + 0x1C0, + 0x1C4, + 0x221, + 0x222, + 0x234, + 0x23A, + 0x255, + 0x256, + 0x258, + 0x259, + 0x25A, + 0x25B, + 0x25D, + 0x260, + 0x262, + 0x263, + 0x264, + 0x265, + 0x267, + 0x268, + 0x26A, + 0x26B, + 0x26D, + 0x26F, + 0x270, + 0x271, + 0x273, + 0x275, + 0x276, + 0x27D, + 0x27E, + 0x280, + 0x281, + 0x283, + 0x284, + 0x287, + 0x28D, + 0x292, + 0x293, + 0x29D, + 0x29F, + 0x2BC, + 0x2BD, + 0x2BE, + 0x2BF, + 0x300, + 0x302, + 0x307, + 0x309, + 0x30A, + 0x30B, + 0x30C, + 0x30D, + 0x313, + 0x314, + 0x331, + 0x332, + 0x342, + 0x343, + 0x345, + 0x346, + 0x370, + 0x374, + 0x376, + 0x378, + 0x37B, + 0x37E, + 0x37F, + 0x380, + 0x386, + 0x387, + 0x388, + 0x38B, + 0x38C, + 0x38D, + 0x38E, + 0x3A2, + 0x3A3, + 0x3D2, + 0x3D5, + 0x3F6, + 0x3F7, + 0x3FC, + 0x3FD, + 0x482, + 0x48A, + 0x530, + 0x531, + 0x557, + 0x561, + 0x588, + 0x10A0, + 0x10C6, + 0x10C7, + 0x10C8, + 0x10CD, + 0x10CE, + 0x13A0, + 0x13F6, + 0x13F8, + 0x13FE, + 0x1D79, + 0x1D7A, + 0x1D7D, + 0x1D7E, + 0x1E00, + 0x1E9C, + 0x1E9E, + 0x1E9F, + 0x1EA0, + 0x1F16, + 0x1F18, + 0x1F1E, + 0x1F20, + 0x1F46, + 0x1F48, + 0x1F4E, + 0x1F50, + 0x1F58, + 0x1F59, + 0x1F5A, + 0x1F5B, + 0x1F5C, + 0x1F5D, + 0x1F5E, + 0x1F5F, + 0x1F7E, + 0x1F80, + 0x1FB5, + 0x1FB6, + 0x1FBD, + 0x1FBE, + 0x1FBF, + 0x1FC2, + 0x1FC5, + 0x1FC6, + 0x1FCD, + 0x1FD0, + 0x1FD4, + 0x1FD6, + 0x1FDC, + 0x1FE0, + 0x1FED, + 0x1FF2, + 0x1FF5, + 0x1FF6, + 0x1FFD, + 0x2126, + 0x2127, + 0x212A, + 0x212C, + 0x2132, + 0x2133, + 0x214E, + 0x214F, + 0x2160, + 0x2180, + 0x2183, + 0x2185, + 0x24B6, + 0x24EA, + 0x2C00, + 0x2C2F, + 0x2C30, + 0x2C5F, + 0x2C60, + 0x2C71, + 0x2C72, + 0x2C74, + 0x2C75, + 0x2C77, + 0x2C7E, + 0x2CE4, + 0x2CEB, + 0x2CEF, + 0x2CF2, + 0x2CF4, + 0x2D00, + 0x2D26, + 0x2D27, + 0x2D28, + 0x2D2D, + 0x2D2E, + 0xA640, + 0xA66E, + 0xA680, + 0xA69C, + 0xA722, + 0xA730, + 0xA732, + 0xA770, + 0xA779, + 0xA788, + 0xA78B, + 0xA78E, + 0xA790, + 0xA794, + 0xA796, + 0xA7AE, + 0xA7B0, + 0xA7B8, + 0xAB53, + 0xAB54, + 0xAB70, + 0xABC0, + 0xFB00, + 0xFB07, + 0xFB13, + 0xFB18, + 0xFF21, + 0xFF3B, + 0xFF41, + 0xFF5B, + 0x10400, + 0x10450, + 0x10C80, + 0x10CB3, + 0x10CC0, + 0x10CF3, + 0x118A0, + 0x118E0 +}; + +static const UV _Perl_Folds_To_Multi_Char_invlist[] = { /* for EBCDIC POSIX-BC */ + 59, /* Number of elements */ + 148565664, /* Version and data structure type */ + 1, /* 0 if the list starts at 0; + 1 if it starts at the element beyond 0 */ + 0x0, + 0x59, + 0x5A, + 0x130, + 0x131, + 0x149, + 0x14A, + 0x1F0, + 0x1F1, + 0x390, + 0x391, + 0x3B0, + 0x3B1, + 0x587, + 0x588, + 0x1E96, + 0x1E9B, + 0x1E9E, + 0x1E9F, + 0x1F50, + 0x1F51, + 0x1F52, + 0x1F53, + 0x1F54, + 0x1F55, + 0x1F56, + 0x1F57, + 0x1F80, + 0x1FB0, + 0x1FB2, + 0x1FB5, + 0x1FB6, + 0x1FB8, + 0x1FBC, + 0x1FBD, + 0x1FC2, + 0x1FC5, + 0x1FC6, + 0x1FC8, + 0x1FCC, + 0x1FCD, + 0x1FD2, + 0x1FD4, + 0x1FD6, + 0x1FD8, + 0x1FE2, + 0x1FE5, + 0x1FE6, + 0x1FE8, + 0x1FF2, + 0x1FF5, + 0x1FF6, + 0x1FF8, + 0x1FFC, + 0x1FFD, + 0xFB00, + 0xFB07, + 0xFB13, + 0xFB18 +}; + +#endif /* defined(PERL_IN_REGCOMP_C) */ + +#if defined(PERL_IN_PERL_C) + +static const UV _Perl_GCB_invlist[] = { /* for EBCDIC POSIX-BC */ 1502, /* Number of elements */ 148565664, /* Version and data structure type */ 0, /* 0 if the list starts at 0; @@ -76397,7 +76819,7 @@ typedef enum { GCB_V = 13 } GCB_enum; -static const GCB_enum Grapheme_Cluster_Break_invmap[] = { /* for EBCDIC POSIX-BC */ +static const GCB_enum _Perl_GCB_invmap[] = { /* for EBCDIC POSIX-BC */ GCB_Control, GCB_CR, GCB_Control, @@ -77904,412 +78326,6 @@ static const GCB_enum Grapheme_Cluster_Break_invmap[] = { /* for EBCDIC POSIX-BC #endif /* defined(PERL_IN_REGEXEC_C) */ -#if defined(PERL_IN_REGCOMP_C) - -static const UV NonL1_Perl_Non_Final_Folds_invlist[] = { /* for EBCDIC POSIX-BC */ - 45, /* Number of elements */ - 148565664, /* Version and data structure type */ - 1, /* 0 if the list starts at 0; - 1 if it starts at the element beyond 0 */ - 0x0, - 0x2BC, - 0x2BD, - 0x308, - 0x309, - 0x313, - 0x314, - 0x342, - 0x343, - 0x3AC, - 0x3AD, - 0x3AE, - 0x3AF, - 0x3B1, - 0x3B2, - 0x3B7, - 0x3B8, - 0x3B9, - 0x3BA, - 0x3C1, - 0x3C2, - 0x3C5, - 0x3C6, - 0x3C9, - 0x3CA, - 0x3CE, - 0x3CF, - 0x565, - 0x566, - 0x574, - 0x575, - 0x57E, - 0x57F, - 0x1F00, - 0x1F08, - 0x1F20, - 0x1F28, - 0x1F60, - 0x1F68, - 0x1F70, - 0x1F71, - 0x1F74, - 0x1F75, - 0x1F7C, - 0x1F7D -}; - -static const UV _Perl_Any_Folds_invlist[] = { /* for EBCDIC POSIX-BC */ - 277, /* Number of elements */ - 148565664, /* Version and data structure type */ - 1, /* 0 if the list starts at 0; - 1 if it starts at the element beyond 0 */ - 0x0, - 0x42, - 0x4A, - 0x51, - 0x5A, - 0x62, - 0x6A, - 0x70, - 0x79, - 0x80, - 0x8A, - 0x8C, - 0x8F, - 0x91, - 0x9A, - 0x9C, - 0x9D, - 0x9E, - 0x9F, - 0xA0, - 0xA1, - 0xA2, - 0xAA, - 0xAC, - 0xAF, - 0xC0, - 0xCA, - 0xCB, - 0xD0, - 0xD1, - 0xDA, - 0xDB, - 0xE1, - 0xE2, - 0xEA, - 0xEB, - 0xF0, - 0xFC, - 0xFD, - 0xFE, - 0xFF, - 0x100, - 0x131, - 0x132, - 0x138, - 0x139, - 0x18D, - 0x18E, - 0x19B, - 0x19C, - 0x1AA, - 0x1AC, - 0x1BA, - 0x1BC, - 0x1BE, - 0x1BF, - 0x1C0, - 0x1C4, - 0x221, - 0x222, - 0x234, - 0x23A, - 0x255, - 0x256, - 0x258, - 0x259, - 0x25A, - 0x25B, - 0x25D, - 0x260, - 0x262, - 0x263, - 0x264, - 0x265, - 0x267, - 0x268, - 0x26A, - 0x26B, - 0x26D, - 0x26F, - 0x270, - 0x271, - 0x273, - 0x275, - 0x276, - 0x27D, - 0x27E, - 0x280, - 0x281, - 0x283, - 0x284, - 0x287, - 0x28D, - 0x292, - 0x293, - 0x29D, - 0x29F, - 0x2BC, - 0x2BD, - 0x2BE, - 0x2BF, - 0x300, - 0x302, - 0x307, - 0x309, - 0x30A, - 0x30B, - 0x30C, - 0x30D, - 0x313, - 0x314, - 0x331, - 0x332, - 0x342, - 0x343, - 0x345, - 0x346, - 0x370, - 0x374, - 0x376, - 0x378, - 0x37B, - 0x37E, - 0x37F, - 0x380, - 0x386, - 0x387, - 0x388, - 0x38B, - 0x38C, - 0x38D, - 0x38E, - 0x3A2, - 0x3A3, - 0x3D2, - 0x3D5, - 0x3F6, - 0x3F7, - 0x3FC, - 0x3FD, - 0x482, - 0x48A, - 0x530, - 0x531, - 0x557, - 0x561, - 0x588, - 0x10A0, - 0x10C6, - 0x10C7, - 0x10C8, - 0x10CD, - 0x10CE, - 0x13A0, - 0x13F6, - 0x13F8, - 0x13FE, - 0x1D79, - 0x1D7A, - 0x1D7D, - 0x1D7E, - 0x1E00, - 0x1E9C, - 0x1E9E, - 0x1E9F, - 0x1EA0, - 0x1F16, - 0x1F18, - 0x1F1E, - 0x1F20, - 0x1F46, - 0x1F48, - 0x1F4E, - 0x1F50, - 0x1F58, - 0x1F59, - 0x1F5A, - 0x1F5B, - 0x1F5C, - 0x1F5D, - 0x1F5E, - 0x1F5F, - 0x1F7E, - 0x1F80, - 0x1FB5, - 0x1FB6, - 0x1FBD, - 0x1FBE, - 0x1FBF, - 0x1FC2, - 0x1FC5, - 0x1FC6, - 0x1FCD, - 0x1FD0, - 0x1FD4, - 0x1FD6, - 0x1FDC, - 0x1FE0, - 0x1FED, - 0x1FF2, - 0x1FF5, - 0x1FF6, - 0x1FFD, - 0x2126, - 0x2127, - 0x212A, - 0x212C, - 0x2132, - 0x2133, - 0x214E, - 0x214F, - 0x2160, - 0x2180, - 0x2183, - 0x2185, - 0x24B6, - 0x24EA, - 0x2C00, - 0x2C2F, - 0x2C30, - 0x2C5F, - 0x2C60, - 0x2C71, - 0x2C72, - 0x2C74, - 0x2C75, - 0x2C77, - 0x2C7E, - 0x2CE4, - 0x2CEB, - 0x2CEF, - 0x2CF2, - 0x2CF4, - 0x2D00, - 0x2D26, - 0x2D27, - 0x2D28, - 0x2D2D, - 0x2D2E, - 0xA640, - 0xA66E, - 0xA680, - 0xA69C, - 0xA722, - 0xA730, - 0xA732, - 0xA770, - 0xA779, - 0xA788, - 0xA78B, - 0xA78E, - 0xA790, - 0xA794, - 0xA796, - 0xA7AE, - 0xA7B0, - 0xA7B8, - 0xAB53, - 0xAB54, - 0xAB70, - 0xABC0, - 0xFB00, - 0xFB07, - 0xFB13, - 0xFB18, - 0xFF21, - 0xFF3B, - 0xFF41, - 0xFF5B, - 0x10400, - 0x10450, - 0x10C80, - 0x10CB3, - 0x10CC0, - 0x10CF3, - 0x118A0, - 0x118E0 -}; - -static const UV _Perl_Folds_To_Multi_Char_invlist[] = { /* for EBCDIC POSIX-BC */ - 59, /* Number of elements */ - 148565664, /* Version and data structure type */ - 1, /* 0 if the list starts at 0; - 1 if it starts at the element beyond 0 */ - 0x0, - 0x59, - 0x5A, - 0x130, - 0x131, - 0x149, - 0x14A, - 0x1F0, - 0x1F1, - 0x390, - 0x391, - 0x3B0, - 0x3B1, - 0x587, - 0x588, - 0x1E96, - 0x1E9B, - 0x1E9E, - 0x1E9F, - 0x1F50, - 0x1F51, - 0x1F52, - 0x1F53, - 0x1F54, - 0x1F55, - 0x1F56, - 0x1F57, - 0x1F80, - 0x1FB0, - 0x1FB2, - 0x1FB5, - 0x1FB6, - 0x1FB8, - 0x1FBC, - 0x1FBD, - 0x1FC2, - 0x1FC5, - 0x1FC6, - 0x1FC8, - 0x1FCC, - 0x1FCD, - 0x1FD2, - 0x1FD4, - 0x1FD6, - 0x1FD8, - 0x1FE2, - 0x1FE5, - 0x1FE6, - 0x1FE8, - 0x1FF2, - 0x1FF5, - 0x1FF6, - 0x1FF8, - 0x1FFC, - 0x1FFD, - 0xFB00, - 0xFB07, - 0xFB13, - 0xFB18 -}; - -#endif /* defined(PERL_IN_REGCOMP_C) */ - #if defined(PERL_IN_UTF8_C) static const UV _Perl_IDCont_invlist[] = { /* for EBCDIC POSIX-BC */ @@ -99521,8 +99537,8 @@ static const UV XPosixXDigit_invlist[] = { /* for EBCDIC POSIX-BC */ * 1a0687fb9c6c4567e853913549df0944fe40821279a3e9cdaa6ab8679bc286fd lib/unicore/extracted/DLineBreak.txt * 40bcfed3ca727c19e1331f6c33806231d5f7eeeabd2e6a9e06a3740c85d0c250 lib/unicore/extracted/DNumType.txt * a18d502bad39d527ac5586d7bc93e29f565859e3bcc24ada627eff606d6f5fed lib/unicore/extracted/DNumValues.txt - * 445c2d31d4731031ab802208b4b827b140c8db348c23d36094050d3607c9339f lib/unicore/mktables + * 96205213df4d25074497ca2449802bd4fb51e2a31cfff389fd5197a933bef455 lib/unicore/mktables * 462c9aaa608fb2014cd9649af1c5c009485c60b9c8b15b89401fdc10cf6161c6 lib/unicore/version * c6884f4d629f04d1316f3476cb1050b6a1b98ca30c903262955d4eae337c6b1e regen/charset_translations.pl - * 7b6f61662df48e0cbfb234a926e02e5cb9468af052f7f9feb84285996f30df09 regen/mk_invlists.pl + * 95c35beb704819594c061545c9fac9868baaf44dbf01d648657395ab846cd14f regen/mk_invlists.pl * ex: set ro: */ diff --git a/lib/unicore/mktables b/lib/unicore/mktables index 7507e08155..e285a34bb4 100644 --- a/lib/unicore/mktables +++ b/lib/unicore/mktables @@ -9878,13 +9878,6 @@ sub finish_property_setup { Property->new('JSN', Full_Name => 'Jamo_Short_Name'); } - # This property must be defined in all releases so we can generate - # the tables from it to make regex \X work, but suppress its output so - # isn't application visible prior to releases where it should be - if (! defined property_ref('GCB')) { - Property->new('GCB', Full_Name => 'Grapheme_Cluster_Break', - Fate => $PLACEHOLDER); - } # These are used so much, that we set globals for them. $gc = property_ref('General_Category'); $block = property_ref('Block'); @@ -10214,16 +10207,16 @@ END } if ($v_version lt 4.1.0) { $file->insert_lines(split /\n/, <<'END' -GCB; CN ; Control -GCB; CR ; CR -GCB; EX ; Extend -GCB; L ; L -GCB; LF ; LF -GCB; LV ; LV -GCB; LVT ; LVT -GCB; T ; T -GCB; V ; V -GCB; XX ; Other +_Perl_GCB; CN ; Control +_Perl_GCB; CR ; CR +_Perl_GCB; EX ; Extend +_Perl_GCB; L ; L +_Perl_GCB; LF ; LF +_Perl_GCB; LV ; LV +_Perl_GCB; LVT ; LVT +_Perl_GCB; T ; T +_Perl_GCB; V ; V +_Perl_GCB; XX ; Other END ); } @@ -13105,7 +13098,7 @@ sub generate_GCB { generate_hst($file); } - return; + main::process_generic_property_file($file); } sub setup_early_name_alias { @@ -18810,12 +18803,10 @@ my @input_file_objects = ( Property => 'Word_Break', Has_Missings_Defaults => $NOT_IGNORED, ), - Input_file->new("$AUXILIARY/GraphemeBreakProperty.txt", v0, + Input_file->new("$AUXILIARY/GraphemeBreakProperty.txt", v4.1, + Early => [ \&generate_GCB, '_Perl_GCB' ], Property => 'Grapheme_Cluster_Break', Has_Missings_Defaults => $NOT_IGNORED, - Pre_Handler => ($v_version lt v4.1.0) - ? \&generate_GCB - : undef, ), Input_file->new("$AUXILIARY/GCBTest.txt", v4.1.0, Handler => \&process_GCB_test, @@ -389,7 +389,7 @@ perl_construct(pTHXx) PL_XPosix_ptrs[_CC_VERTSPACE] = _new_invlist_C_array(VertSpace_invlist); PL_XPosix_ptrs[_CC_WORDCHAR] = _new_invlist_C_array(XPosixWord_invlist); PL_XPosix_ptrs[_CC_XDIGIT] = _new_invlist_C_array(XPosixXDigit_invlist); - PL_GCB_invlist = _new_invlist_C_array(Grapheme_Cluster_Break_invlist); + PL_GCB_invlist = _new_invlist_C_array(_Perl_GCB_invlist); PL_SB_invlist = _new_invlist_C_array(Sentence_Break_invlist); PL_WB_invlist = _new_invlist_C_array(Word_Break_invlist); diff --git a/regcharclass.h b/regcharclass.h index cee7ef6557..c4515fc327 100644 --- a/regcharclass.h +++ b/regcharclass.h @@ -2514,7 +2514,7 @@ * 1a0687fb9c6c4567e853913549df0944fe40821279a3e9cdaa6ab8679bc286fd lib/unicore/extracted/DLineBreak.txt * 40bcfed3ca727c19e1331f6c33806231d5f7eeeabd2e6a9e06a3740c85d0c250 lib/unicore/extracted/DNumType.txt * a18d502bad39d527ac5586d7bc93e29f565859e3bcc24ada627eff606d6f5fed lib/unicore/extracted/DNumValues.txt - * 445c2d31d4731031ab802208b4b827b140c8db348c23d36094050d3607c9339f lib/unicore/mktables + * 96205213df4d25074497ca2449802bd4fb51e2a31cfff389fd5197a933bef455 lib/unicore/mktables * 462c9aaa608fb2014cd9649af1c5c009485c60b9c8b15b89401fdc10cf6161c6 lib/unicore/version * c6884f4d629f04d1316f3476cb1050b6a1b98ca30c903262955d4eae337c6b1e regen/charset_translations.pl * d9c04ac46bdd81bb3e26519f2b8eb6242cb12337205add3f7cf092b0c58dccc4 regen/regcharclass.pl diff --git a/regen/mk_invlists.pl b/regen/mk_invlists.pl index 3c7e42cbc5..8d691eeaf6 100644 --- a/regen/mk_invlists.pl +++ b/regen/mk_invlists.pl @@ -234,8 +234,8 @@ sub output_invmap ($$$$$$$) { my $name_prefix; if ($input_format eq 's') { - $prop_name = (prop_aliases($prop_name))[1]; # Get full name - my $short_name = (prop_aliases($prop_name))[0]; + $prop_name = (prop_aliases($prop_name))[1] // $prop_name =~ s/^_Perl_//r; # Get full name + my $short_name = (prop_aliases($prop_name))[0] // $prop_name; my @enums = prop_values($prop_name); if (! @enums) { die "Only enum properties are currently handled; '$prop_name' isn't one"; @@ -266,7 +266,8 @@ sub output_invmap ($$$$$$$) { # Assign a value to each element of the enum. The default # value always gets 0; the others are arbitrarily assigned. my $enum_val = 0; - $default = prop_value_aliases($prop_name, $default); + my $canonical_default = prop_value_aliases($prop_name, $default); + $default = $canonical_default if defined $canonical_default; $enums{$default} = $enum_val++; for my $enum (@enums) { $enums{$enum} = $enum_val++ unless exists $enums{$enum}; @@ -314,7 +315,9 @@ sub output_invmap ($$$$$$$) { # The main body are the scalars passed in to this routine. for my $i (0 .. $count - 1) { my $element = $invmap->[$i]; - $element = $name_prefix . prop_value_aliases($prop_name, $element); + my $full_element_name = prop_value_aliases($prop_name, $element); + $element = $full_element_name if defined $full_element_name; + $element = $name_prefix . $element; print $out_fh "\t$element"; print $out_fh "," if $i < $count - 1; print $out_fh "\n"; @@ -447,7 +450,7 @@ for my $charset (get_supported_code_pages()) { &UpperLatin1 _Perl_IDStart _Perl_IDCont - Grapheme_Cluster_Break,EDGE + _Perl_GCB,EDGE Word_Break,EDGE,UNKNOWN Sentence_Break,EDGE ) @@ -1753,7 +1753,7 @@ REXEC_FBC_SCAN( /* Loops while (s < strend) */ \ #define getGCB_VAL_CP(cp) \ _generic_GET_BREAK_VAL_CP( \ PL_GCB_invlist, \ - Grapheme_Cluster_Break_invmap, \ + _Perl_GCB_invmap, \ (cp)) /* Returns the GCB value for the first code point in the UTF-8 encoded string |