Skip to content

Commit

Permalink
Updated spec.txt expands what's recognized as Unicode punctuation.
Browse files Browse the repository at this point in the history 
Namely all P and S general categories are now treated as punctuation.
  • Loading branch information
@mity
mity committed Jan 9, 2024
1 parent fc31a80 commit fbbdffa
Show file tree
Hide file tree
Showing 2 changed files with 60 additions and 35 deletions.
3 changes: 1 addition & 2 deletions scripts/build_punct_map.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,6 @@
f = open(self_path + "/unicode/DerivedGeneralCategory.txt", "r")

codepoint_list = []
category_list = [ "Pc", "Pd", "Pe", "Pf", "Pi", "Po", "Ps" ]

# Filter codepoints falling in the right category:
for line in f:
Expand All @@ -24,7 +23,7 @@
char_range = char_range.strip()
category = category.strip()

if not category in category_list:
if category[0] != 'P' and category[0] != 'S':
continue

delim_off = char_range.find("..")
Expand Down
92 changes: 59 additions & 33 deletions src/md4c.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -567,41 +567,67 @@ struct MD_UNICODE_FOLD_INFO_tag {
{
#define R(cp_min, cp_max) ((cp_min) | 0x40000000), ((cp_max) | 0x80000000)
#define S(cp) (cp)
/* Unicode "Pc", "Pd", "Pe", "Pf", "Pi", "Po", "Ps" categories.
/* Unicode general "P" and "S" categories.
* (generated by scripts/build_punct_map.py) */
static const unsigned PUNCT_MAP[] = {
R(0x0021,0x0023), R(0x0025,0x002a), R(0x002c,0x002f), R(0x003a,0x003b), R(0x003f,0x0040),
R(0x005b,0x005d), S(0x005f), S(0x007b), S(0x007d), S(0x00a1), S(0x00a7), S(0x00ab), R(0x00b6,0x00b7),
S(0x00bb), S(0x00bf), S(0x037e), S(0x0387), R(0x055a,0x055f), R(0x0589,0x058a), S(0x05be), S(0x05c0),
S(0x05c3), S(0x05c6), R(0x05f3,0x05f4), R(0x0609,0x060a), R(0x060c,0x060d), S(0x061b), R(0x061d,0x061f),
R(0x066a,0x066d), S(0x06d4), R(0x0700,0x070d), R(0x07f7,0x07f9), R(0x0830,0x083e), S(0x085e),
R(0x0964,0x0965), S(0x0970), S(0x09fd), S(0x0a76), S(0x0af0), S(0x0c77), S(0x0c84), S(0x0df4), S(0x0e4f),
R(0x0e5a,0x0e5b), R(0x0f04,0x0f12), S(0x0f14), R(0x0f3a,0x0f3d), S(0x0f85), R(0x0fd0,0x0fd4),
R(0x0fd9,0x0fda), R(0x104a,0x104f), S(0x10fb), R(0x1360,0x1368), S(0x1400), S(0x166e), R(0x169b,0x169c),
R(0x16eb,0x16ed), R(0x1735,0x1736), R(0x17d4,0x17d6), R(0x17d8,0x17da), R(0x1800,0x180a),
R(0x1944,0x1945), R(0x1a1e,0x1a1f), R(0x1aa0,0x1aa6), R(0x1aa8,0x1aad), R(0x1b5a,0x1b60),
R(0x1b7d,0x1b7e), R(0x1bfc,0x1bff), R(0x1c3b,0x1c3f), R(0x1c7e,0x1c7f), R(0x1cc0,0x1cc7), S(0x1cd3),
R(0x2010,0x2027), R(0x2030,0x2043), R(0x2045,0x2051), R(0x2053,0x205e), R(0x207d,0x207e),
R(0x208d,0x208e), R(0x2308,0x230b), R(0x2329,0x232a), R(0x2768,0x2775), R(0x27c5,0x27c6),
R(0x27e6,0x27ef), R(0x2983,0x2998), R(0x29d8,0x29db), R(0x29fc,0x29fd), R(0x2cf9,0x2cfc),
R(0x2cfe,0x2cff), S(0x2d70), R(0x2e00,0x2e2e), R(0x2e30,0x2e4f), R(0x2e52,0x2e5d), R(0x3001,0x3003),
R(0x3008,0x3011), R(0x3014,0x301f), S(0x3030), S(0x303d), S(0x30a0), S(0x30fb), R(0xa4fe,0xa4ff),
R(0xa60d,0xa60f), S(0xa673), S(0xa67e), R(0xa6f2,0xa6f7), R(0xa874,0xa877), R(0xa8ce,0xa8cf),
R(0xa8f8,0xa8fa), S(0xa8fc), R(0xa92e,0xa92f), S(0xa95f), R(0xa9c1,0xa9cd), R(0xa9de,0xa9df),
R(0xaa5c,0xaa5f), R(0xaade,0xaadf), R(0xaaf0,0xaaf1), S(0xabeb), R(0xfd3e,0xfd3f), R(0xfe10,0xfe19),
R(0xfe30,0xfe52), R(0xfe54,0xfe61), S(0xfe63), S(0xfe68), R(0xfe6a,0xfe6b), R(0xff01,0xff03),
R(0xff05,0xff0a), R(0xff0c,0xff0f), R(0xff1a,0xff1b), R(0xff1f,0xff20), R(0xff3b,0xff3d), S(0xff3f),
S(0xff5b), S(0xff5d), R(0xff5f,0xff65), R(0x10100,0x10102), S(0x1039f), S(0x103d0), S(0x1056f),
S(0x10857), S(0x1091f), S(0x1093f), R(0x10a50,0x10a58), S(0x10a7f), R(0x10af0,0x10af6),
R(0x10b39,0x10b3f), R(0x10b99,0x10b9c), S(0x10ead), R(0x10f55,0x10f59), R(0x10f86,0x10f89),
R(0x11047,0x1104d), R(0x110bb,0x110bc), R(0x110be,0x110c1), R(0x11140,0x11143), R(0x11174,0x11175),
R(0x111c5,0x111c8), S(0x111cd), S(0x111db), R(0x111dd,0x111df), R(0x11238,0x1123d), S(0x112a9),
R(0x1144b,0x1144f), R(0x1145a,0x1145b), S(0x1145d), S(0x114c6), R(0x115c1,0x115d7), R(0x11641,0x11643),
R(0x11660,0x1166c), S(0x116b9), R(0x1173c,0x1173e), S(0x1183b), R(0x11944,0x11946), S(0x119e2),
R(0x11a3f,0x11a46), R(0x11a9a,0x11a9c), R(0x11a9e,0x11aa2), R(0x11b00,0x11b09), R(0x11c41,0x11c45),
R(0x11c70,0x11c71), R(0x11ef7,0x11ef8), R(0x11f43,0x11f4f), S(0x11fff), R(0x12470,0x12474),
R(0x12ff1,0x12ff2), R(0x16a6e,0x16a6f), S(0x16af5), R(0x16b37,0x16b3b), S(0x16b44), R(0x16e97,0x16e9a),
S(0x16fe2), S(0x1bc9f), R(0x1da87,0x1da8b), R(0x1e95e,0x1e95f)
R(0x0021,0x002f), R(0x003a,0x0040), R(0x005b,0x0060), R(0x007b,0x007e), R(0x00a1,0x00a9),
R(0x00ab,0x00ac), R(0x00ae,0x00b1), S(0x00b4), R(0x00b6,0x00b8), S(0x00bb), S(0x00bf), S(0x00d7),
S(0x00f7), R(0x02c2,0x02c5), R(0x02d2,0x02df), R(0x02e5,0x02eb), S(0x02ed), R(0x02ef,0x02ff), S(0x0375),
S(0x037e), R(0x0384,0x0385), S(0x0387), S(0x03f6), S(0x0482), R(0x055a,0x055f), R(0x0589,0x058a),
R(0x058d,0x058f), S(0x05be), S(0x05c0), S(0x05c3), S(0x05c6), R(0x05f3,0x05f4), R(0x0606,0x060f),
S(0x061b), R(0x061d,0x061f), R(0x066a,0x066d), S(0x06d4), S(0x06de), S(0x06e9), R(0x06fd,0x06fe),
R(0x0700,0x070d), R(0x07f6,0x07f9), R(0x07fe,0x07ff), R(0x0830,0x083e), S(0x085e), S(0x0888),
R(0x0964,0x0965), S(0x0970), R(0x09f2,0x09f3), R(0x09fa,0x09fb), S(0x09fd), S(0x0a76), R(0x0af0,0x0af1),
S(0x0b70), R(0x0bf3,0x0bfa), S(0x0c77), S(0x0c7f), S(0x0c84), S(0x0d4f), S(0x0d79), S(0x0df4), S(0x0e3f),
S(0x0e4f), R(0x0e5a,0x0e5b), R(0x0f01,0x0f17), R(0x0f1a,0x0f1f), S(0x0f34), S(0x0f36), S(0x0f38),
R(0x0f3a,0x0f3d), S(0x0f85), R(0x0fbe,0x0fc5), R(0x0fc7,0x0fcc), R(0x0fce,0x0fda), R(0x104a,0x104f),
R(0x109e,0x109f), S(0x10fb), R(0x1360,0x1368), R(0x1390,0x1399), S(0x1400), R(0x166d,0x166e),
R(0x169b,0x169c), R(0x16eb,0x16ed), R(0x1735,0x1736), R(0x17d4,0x17d6), R(0x17d8,0x17db),
R(0x1800,0x180a), S(0x1940), R(0x1944,0x1945), R(0x19de,0x19ff), R(0x1a1e,0x1a1f), R(0x1aa0,0x1aa6),
R(0x1aa8,0x1aad), R(0x1b5a,0x1b6a), R(0x1b74,0x1b7e), R(0x1bfc,0x1bff), R(0x1c3b,0x1c3f),
R(0x1c7e,0x1c7f), R(0x1cc0,0x1cc7), S(0x1cd3), S(0x1fbd), R(0x1fbf,0x1fc1), R(0x1fcd,0x1fcf),
R(0x1fdd,0x1fdf), R(0x1fed,0x1fef), R(0x1ffd,0x1ffe), R(0x2010,0x2027), R(0x2030,0x205e),
R(0x207a,0x207e), R(0x208a,0x208e), R(0x20a0,0x20c0), R(0x2100,0x2101), R(0x2103,0x2106),
R(0x2108,0x2109), S(0x2114), R(0x2116,0x2118), R(0x211e,0x2123), S(0x2125), S(0x2127), S(0x2129),
S(0x212e), R(0x213a,0x213b), R(0x2140,0x2144), R(0x214a,0x214d), S(0x214f), R(0x218a,0x218b),
R(0x2190,0x2426), R(0x2440,0x244a), R(0x249c,0x24e9), R(0x2500,0x2775), R(0x2794,0x2b73),
R(0x2b76,0x2b95), R(0x2b97,0x2bff), R(0x2ce5,0x2cea), R(0x2cf9,0x2cfc), R(0x2cfe,0x2cff), S(0x2d70),
R(0x2e00,0x2e2e), R(0x2e30,0x2e5d), R(0x2e80,0x2e99), R(0x2e9b,0x2ef3), R(0x2f00,0x2fd5),
R(0x2ff0,0x2fff), R(0x3001,0x3004), R(0x3008,0x3020), S(0x3030), R(0x3036,0x3037), R(0x303d,0x303f),
R(0x309b,0x309c), S(0x30a0), S(0x30fb), R(0x3190,0x3191), R(0x3196,0x319f), R(0x31c0,0x31e3), S(0x31ef),
R(0x3200,0x321e), R(0x322a,0x3247), S(0x3250), R(0x3260,0x327f), R(0x328a,0x32b0), R(0x32c0,0x33ff),
R(0x4dc0,0x4dff), R(0xa490,0xa4c6), R(0xa4fe,0xa4ff), R(0xa60d,0xa60f), S(0xa673), S(0xa67e),
R(0xa6f2,0xa6f7), R(0xa700,0xa716), R(0xa720,0xa721), R(0xa789,0xa78a), R(0xa828,0xa82b),
R(0xa836,0xa839), R(0xa874,0xa877), R(0xa8ce,0xa8cf), R(0xa8f8,0xa8fa), S(0xa8fc), R(0xa92e,0xa92f),
S(0xa95f), R(0xa9c1,0xa9cd), R(0xa9de,0xa9df), R(0xaa5c,0xaa5f), R(0xaa77,0xaa79), R(0xaade,0xaadf),
R(0xaaf0,0xaaf1), S(0xab5b), R(0xab6a,0xab6b), S(0xabeb), S(0xfb29), R(0xfbb2,0xfbc2), R(0xfd3e,0xfd4f),
S(0xfdcf), R(0xfdfc,0xfdff), R(0xfe10,0xfe19), R(0xfe30,0xfe52), R(0xfe54,0xfe66), R(0xfe68,0xfe6b),
R(0xff01,0xff0f), R(0xff1a,0xff20), R(0xff3b,0xff40), R(0xff5b,0xff65), R(0xffe0,0xffe6),
R(0xffe8,0xffee), R(0xfffc,0xfffd), R(0x10100,0x10102), R(0x10137,0x1013f), R(0x10179,0x10189),
R(0x1018c,0x1018e), R(0x10190,0x1019c), S(0x101a0), R(0x101d0,0x101fc), S(0x1039f), S(0x103d0),
S(0x1056f), S(0x10857), R(0x10877,0x10878), S(0x1091f), S(0x1093f), R(0x10a50,0x10a58), S(0x10a7f),
S(0x10ac8), R(0x10af0,0x10af6), R(0x10b39,0x10b3f), R(0x10b99,0x10b9c), S(0x10ead), R(0x10f55,0x10f59),
R(0x10f86,0x10f89), R(0x11047,0x1104d), R(0x110bb,0x110bc), R(0x110be,0x110c1), R(0x11140,0x11143),
R(0x11174,0x11175), R(0x111c5,0x111c8), S(0x111cd), S(0x111db), R(0x111dd,0x111df), R(0x11238,0x1123d),
S(0x112a9), R(0x1144b,0x1144f), R(0x1145a,0x1145b), S(0x1145d), S(0x114c6), R(0x115c1,0x115d7),
R(0x11641,0x11643), R(0x11660,0x1166c), S(0x116b9), R(0x1173c,0x1173f), S(0x1183b), R(0x11944,0x11946),
S(0x119e2), R(0x11a3f,0x11a46), R(0x11a9a,0x11a9c), R(0x11a9e,0x11aa2), R(0x11b00,0x11b09),
R(0x11c41,0x11c45), R(0x11c70,0x11c71), R(0x11ef7,0x11ef8), R(0x11f43,0x11f4f), R(0x11fd5,0x11ff1),
S(0x11fff), R(0x12470,0x12474), R(0x12ff1,0x12ff2), R(0x16a6e,0x16a6f), S(0x16af5), R(0x16b37,0x16b3f),
R(0x16b44,0x16b45), R(0x16e97,0x16e9a), S(0x16fe2), S(0x1bc9c), S(0x1bc9f), R(0x1cf50,0x1cfc3),
R(0x1d000,0x1d0f5), R(0x1d100,0x1d126), R(0x1d129,0x1d164), R(0x1d16a,0x1d16c), R(0x1d183,0x1d184),
R(0x1d18c,0x1d1a9), R(0x1d1ae,0x1d1ea), R(0x1d200,0x1d241), S(0x1d245), R(0x1d300,0x1d356), S(0x1d6c1),
S(0x1d6db), S(0x1d6fb), S(0x1d715), S(0x1d735), S(0x1d74f), S(0x1d76f), S(0x1d789), S(0x1d7a9),
S(0x1d7c3), R(0x1d800,0x1d9ff), R(0x1da37,0x1da3a), R(0x1da6d,0x1da74), R(0x1da76,0x1da83),
R(0x1da85,0x1da8b), S(0x1e14f), S(0x1e2ff), R(0x1e95e,0x1e95f), S(0x1ecac), S(0x1ecb0), S(0x1ed2e),
R(0x1eef0,0x1eef1), R(0x1f000,0x1f02b), R(0x1f030,0x1f093), R(0x1f0a0,0x1f0ae), R(0x1f0b1,0x1f0bf),
R(0x1f0c1,0x1f0cf), R(0x1f0d1,0x1f0f5), R(0x1f10d,0x1f1ad), R(0x1f1e6,0x1f202), R(0x1f210,0x1f23b),
R(0x1f240,0x1f248), R(0x1f250,0x1f251), R(0x1f260,0x1f265), R(0x1f300,0x1f6d7), R(0x1f6dc,0x1f6ec),
R(0x1f6f0,0x1f6fc), R(0x1f700,0x1f776), R(0x1f77b,0x1f7d9), R(0x1f7e0,0x1f7eb), S(0x1f7f0),
R(0x1f800,0x1f80b), R(0x1f810,0x1f847), R(0x1f850,0x1f859), R(0x1f860,0x1f887), R(0x1f890,0x1f8ad),
R(0x1f8b0,0x1f8b1), R(0x1f900,0x1fa53), R(0x1fa60,0x1fa6d), R(0x1fa70,0x1fa7c), R(0x1fa80,0x1fa88),
R(0x1fa90,0x1fabd), R(0x1fabf,0x1fac5), R(0x1face,0x1fadb), R(0x1fae0,0x1fae8), R(0x1faf0,0x1faf8),
R(0x1fb00,0x1fb92), R(0x1fb94,0x1fbca)
};
#undef R
#undef S
Expand Down

0 comments on commit fbbdffa

Please sign in to comment.