Commit 72dad97ed6b7b4abef1079d51c16b9b706add99b

Martin Mitas 2020-05-20T16:44:07

scripts/build_folding_map.py: Handle properly "ranges" of length 2. Update the data structures in md_get_unicode_fold_info() to reflect the update in the script and handle the previously omitted characters. Fixes #113.

diff --git a/scripts/build_folding_map.py b/scripts/build_folding_map.py
index b27b480..4f6312c 100644
--- a/scripts/build_folding_map.py
+++ b/scripts/build_folding_map.py
@@ -97,12 +97,12 @@ for mapping_len in range(1, 4):
             records.append("R(0x{:04x},0x{:04x})".format(codepoint_list[index0], codepoint_list[index1-1]))
             data_records.append(mapping_str(data_records, folding[codepoint_list[index0]]))
             data_records.append(mapping_str(data_records, folding[codepoint_list[index1-1]]))
+            index0 = index1
         else:
             # Single codepoint
             records.append("S(0x{:04x})".format(codepoint_list[index0]))
             data_records.append(mapping_str(data_records, folding[codepoint_list[index0]]))
-
-        index0 = index1
+            index0 += 1
 
     sys.stdout.write("static const unsigned FOLD_MAP_{}[] = {{\n".format(mapping_len))
     sys.stdout.write("\n".join(textwrap.wrap(", ".join(records), 110,
diff --git a/src/md4c.c b/src/md4c.c
index f3a8df1..4abd7aa 100644
--- a/src/md4c.c
+++ b/src/md4c.c
@@ -585,52 +585,56 @@ struct MD_UNICODE_FOLD_INFO_tag {
         static const unsigned FOLD_MAP_1[] = {
             R(0x0041,0x005a), S(0x00b5), R(0x00c0,0x00d6), R(0x00d8,0x00de), R(0x0100,0x012e), R(0x0132,0x0136),
             R(0x0139,0x0147), R(0x014a,0x0176), S(0x0178), R(0x0179,0x017d), S(0x017f), S(0x0181), S(0x0182),
-            S(0x0186), S(0x0187), S(0x0189), S(0x018b), S(0x018e), S(0x018f), S(0x0190), S(0x0191), S(0x0193),
-            S(0x0194), S(0x0196), S(0x0197), S(0x0198), S(0x019c), S(0x019d), S(0x019f), R(0x01a0,0x01a4), S(0x01a6),
-            S(0x01a7), S(0x01a9), S(0x01ac), S(0x01ae), S(0x01af), S(0x01b1), S(0x01b3), S(0x01b7), S(0x01b8),
-            S(0x01bc), S(0x01c4), S(0x01c5), S(0x01c7), S(0x01c8), S(0x01ca), R(0x01cb,0x01db), R(0x01de,0x01ee),
-            S(0x01f1), S(0x01f2), S(0x01f6), S(0x01f7), R(0x01f8,0x021e), S(0x0220), R(0x0222,0x0232), S(0x023a),
-            S(0x023b), S(0x023d), S(0x023e), S(0x0241), S(0x0243), S(0x0244), S(0x0245), R(0x0246,0x024e), S(0x0345),
-            S(0x0370), S(0x0376), S(0x037f), S(0x0386), R(0x0388,0x038a), S(0x038c), S(0x038e), R(0x0391,0x03a1),
-            R(0x03a3,0x03ab), S(0x03c2), S(0x03cf), S(0x03d0), S(0x03d1), S(0x03d5), S(0x03d6), R(0x03d8,0x03ee),
-            S(0x03f0), S(0x03f1), S(0x03f4), S(0x03f5), S(0x03f7), S(0x03f9), S(0x03fa), R(0x03fd,0x03ff),
-            R(0x0400,0x040f), R(0x0410,0x042f), R(0x0460,0x0480), R(0x048a,0x04be), S(0x04c0), R(0x04c1,0x04cd),
-            R(0x04d0,0x052e), R(0x0531,0x0556), R(0x10a0,0x10c5), S(0x10c7), S(0x10cd), R(0x13f8,0x13fd), S(0x1c80),
-            S(0x1c81), S(0x1c82), S(0x1c83), S(0x1c85), S(0x1c86), S(0x1c87), S(0x1c88), R(0x1c90,0x1cba),
+            S(0x0184), S(0x0186), S(0x0187), S(0x0189), S(0x018a), S(0x018b), S(0x018e), S(0x018f), S(0x0190),
+            S(0x0191), S(0x0193), S(0x0194), S(0x0196), S(0x0197), S(0x0198), S(0x019c), S(0x019d), S(0x019f),
+            R(0x01a0,0x01a4), S(0x01a6), S(0x01a7), S(0x01a9), S(0x01ac), S(0x01ae), S(0x01af), S(0x01b1), S(0x01b2),
+            S(0x01b3), S(0x01b5), S(0x01b7), S(0x01b8), S(0x01bc), S(0x01c4), S(0x01c5), S(0x01c7), S(0x01c8),
+            S(0x01ca), R(0x01cb,0x01db), R(0x01de,0x01ee), S(0x01f1), S(0x01f2), S(0x01f4), S(0x01f6), S(0x01f7),
+            R(0x01f8,0x021e), S(0x0220), R(0x0222,0x0232), S(0x023a), S(0x023b), S(0x023d), S(0x023e), S(0x0241),
+            S(0x0243), S(0x0244), S(0x0245), R(0x0246,0x024e), S(0x0345), S(0x0370), S(0x0372), S(0x0376), S(0x037f),
+            S(0x0386), R(0x0388,0x038a), S(0x038c), S(0x038e), S(0x038f), R(0x0391,0x03a1), R(0x03a3,0x03ab),
+            S(0x03c2), S(0x03cf), S(0x03d0), S(0x03d1), S(0x03d5), S(0x03d6), R(0x03d8,0x03ee), S(0x03f0), S(0x03f1),
+            S(0x03f4), S(0x03f5), S(0x03f7), S(0x03f9), S(0x03fa), R(0x03fd,0x03ff), R(0x0400,0x040f),
+            R(0x0410,0x042f), R(0x0460,0x0480), R(0x048a,0x04be), S(0x04c0), R(0x04c1,0x04cd), R(0x04d0,0x052e),
+            R(0x0531,0x0556), R(0x10a0,0x10c5), S(0x10c7), S(0x10cd), R(0x13f8,0x13fd), S(0x1c80), S(0x1c81),
+            S(0x1c82), S(0x1c83), S(0x1c84), S(0x1c85), S(0x1c86), S(0x1c87), S(0x1c88), R(0x1c90,0x1cba),
             R(0x1cbd,0x1cbf), R(0x1e00,0x1e94), S(0x1e9b), R(0x1ea0,0x1efe), R(0x1f08,0x1f0f), R(0x1f18,0x1f1d),
             R(0x1f28,0x1f2f), R(0x1f38,0x1f3f), R(0x1f48,0x1f4d), S(0x1f59), S(0x1f5b), S(0x1f5d), S(0x1f5f),
-            R(0x1f68,0x1f6f), S(0x1fb8), S(0x1fba), S(0x1fbe), R(0x1fc8,0x1fcb), S(0x1fd8), S(0x1fda), S(0x1fe8),
-            S(0x1fea), S(0x1fec), S(0x1ff8), S(0x1ffa), S(0x2126), S(0x212a), S(0x212b), S(0x2132), R(0x2160,0x216f),
-            S(0x2183), R(0x24b6,0x24cf), R(0x2c00,0x2c2e), S(0x2c60), S(0x2c62), S(0x2c63), S(0x2c64),
-            R(0x2c67,0x2c6b), S(0x2c6d), S(0x2c6e), S(0x2c6f), S(0x2c70), S(0x2c72), S(0x2c75), S(0x2c7e),
-            R(0x2c80,0x2ce2), S(0x2ceb), S(0x2cf2), R(0xa640,0xa66c), R(0xa680,0xa69a), R(0xa722,0xa72e),
-            R(0xa732,0xa76e), S(0xa779), S(0xa77d), R(0xa77e,0xa786), S(0xa78b), S(0xa78d), S(0xa790),
+            R(0x1f68,0x1f6f), S(0x1fb8), S(0x1fb9), S(0x1fba), S(0x1fbb), S(0x1fbe), R(0x1fc8,0x1fcb), S(0x1fd8),
+            S(0x1fd9), S(0x1fda), S(0x1fdb), S(0x1fe8), S(0x1fe9), S(0x1fea), S(0x1feb), S(0x1fec), S(0x1ff8),
+            S(0x1ff9), S(0x1ffa), S(0x1ffb), S(0x2126), S(0x212a), S(0x212b), S(0x2132), R(0x2160,0x216f), S(0x2183),
+            R(0x24b6,0x24cf), R(0x2c00,0x2c2e), S(0x2c60), S(0x2c62), S(0x2c63), S(0x2c64), R(0x2c67,0x2c6b),
+            S(0x2c6d), S(0x2c6e), S(0x2c6f), S(0x2c70), S(0x2c72), S(0x2c75), S(0x2c7e), S(0x2c7f), R(0x2c80,0x2ce2),
+            S(0x2ceb), S(0x2ced), S(0x2cf2), R(0xa640,0xa66c), R(0xa680,0xa69a), R(0xa722,0xa72e), R(0xa732,0xa76e),
+            S(0xa779), S(0xa77b), S(0xa77d), R(0xa77e,0xa786), S(0xa78b), S(0xa78d), S(0xa790), S(0xa792),
             R(0xa796,0xa7a8), S(0xa7aa), S(0xa7ab), S(0xa7ac), S(0xa7ad), S(0xa7ae), S(0xa7b0), S(0xa7b1), S(0xa7b2),
-            S(0xa7b3), R(0xa7b4,0xa7be), S(0xa7c2), S(0xa7c4), S(0xa7c5), S(0xa7c6), S(0xa7c7), S(0xa7f5),
+            S(0xa7b3), R(0xa7b4,0xa7be), S(0xa7c2), S(0xa7c4), S(0xa7c5), S(0xa7c6), S(0xa7c7), S(0xa7c9), S(0xa7f5),
             R(0xab70,0xabbf), R(0xff21,0xff3a), R(0x10400,0x10427), R(0x104b0,0x104d3), R(0x10c80,0x10cb2),
             R(0x118a0,0x118bf), R(0x16e40,0x16e5f), R(0x1e900,0x1e921)
         };
         static const unsigned FOLD_MAP_1_DATA[] = {
             0x0061, 0x007a, 0x03bc, 0x00e0, 0x00f6, 0x00f8, 0x00fe, 0x0101, 0x012f, 0x0133, 0x0137, 0x013a, 0x0148,
-            0x014b, 0x0177, 0x00ff, 0x017a, 0x017e, 0x0073, 0x0253, 0x0183, 0x0254, 0x0188, 0x0256, 0x018c, 0x01dd,
-            0x0259, 0x025b, 0x0192, 0x0260, 0x0263, 0x0269, 0x0268, 0x0199, 0x026f, 0x0272, 0x0275, 0x01a1, 0x01a5,
-            0x0280, 0x01a8, 0x0283, 0x01ad, 0x0288, 0x01b0, 0x028a, 0x01b4, 0x0292, 0x01b9, 0x01bd, 0x01c6, 0x01c6,
-            0x01c9, 0x01c9, 0x01cc, 0x01cc, 0x01dc, 0x01df, 0x01ef, 0x01f3, 0x01f3, 0x0195, 0x01bf, 0x01f9, 0x021f,
-            0x019e, 0x0223, 0x0233, 0x2c65, 0x023c, 0x019a, 0x2c66, 0x0242, 0x0180, 0x0289, 0x028c, 0x0247, 0x024f,
-            0x03b9, 0x0371, 0x0377, 0x03f3, 0x03ac, 0x03ad, 0x03af, 0x03cc, 0x03cd, 0x03b1, 0x03c1, 0x03c3, 0x03cb,
-            0x03c3, 0x03d7, 0x03b2, 0x03b8, 0x03c6, 0x03c0, 0x03d9, 0x03ef, 0x03ba, 0x03c1, 0x03b8, 0x03b5, 0x03f8,
-            0x03f2, 0x03fb, 0x037b, 0x037d, 0x0450, 0x045f, 0x0430, 0x044f, 0x0461, 0x0481, 0x048b, 0x04bf, 0x04cf,
-            0x04c2, 0x04ce, 0x04d1, 0x052f, 0x0561, 0x0586, 0x2d00, 0x2d25, 0x2d27, 0x2d2d, 0x13f0, 0x13f5, 0x0432,
-            0x0434, 0x043e, 0x0441, 0x0442, 0x044a, 0x0463, 0xa64b, 0x10d0, 0x10fa, 0x10fd, 0x10ff, 0x1e01, 0x1e95,
-            0x1e61, 0x1ea1, 0x1eff, 0x1f00, 0x1f07, 0x1f10, 0x1f15, 0x1f20, 0x1f27, 0x1f30, 0x1f37, 0x1f40, 0x1f45,
-            0x1f51, 0x1f53, 0x1f55, 0x1f57, 0x1f60, 0x1f67, 0x1fb0, 0x1f70, 0x03b9, 0x1f72, 0x1f75, 0x1fd0, 0x1f76,
-            0x1fe0, 0x1f7a, 0x1fe5, 0x1f78, 0x1f7c, 0x03c9, 0x006b, 0x00e5, 0x214e, 0x2170, 0x217f, 0x2184, 0x24d0,
-            0x24e9, 0x2c30, 0x2c5e, 0x2c61, 0x026b, 0x1d7d, 0x027d, 0x2c68, 0x2c6c, 0x0251, 0x0271, 0x0250, 0x0252,
-            0x2c73, 0x2c76, 0x023f, 0x2c81, 0x2ce3, 0x2cec, 0x2cf3, 0xa641, 0xa66d, 0xa681, 0xa69b, 0xa723, 0xa72f,
-            0xa733, 0xa76f, 0xa77a, 0x1d79, 0xa77f, 0xa787, 0xa78c, 0x0265, 0xa791, 0xa797, 0xa7a9, 0x0266, 0x025c,
-            0x0261, 0x026c, 0x026a, 0x029e, 0x0287, 0x029d, 0xab53, 0xa7b5, 0xa7bf, 0xa7c3, 0xa794, 0x0282, 0x1d8e,
-            0xa7c8, 0xa7f6, 0x13a0, 0x13ef, 0xff41, 0xff5a, 0x10428, 0x1044f, 0x104d8, 0x104fb, 0x10cc0, 0x10cf2,
-            0x118c0, 0x118df, 0x16e60, 0x16e7f, 0x1e922, 0x1e943
+            0x014b, 0x0177, 0x00ff, 0x017a, 0x017e, 0x0073, 0x0253, 0x0183, 0x0185, 0x0254, 0x0188, 0x0256, 0x0257,
+            0x018c, 0x01dd, 0x0259, 0x025b, 0x0192, 0x0260, 0x0263, 0x0269, 0x0268, 0x0199, 0x026f, 0x0272, 0x0275,
+            0x01a1, 0x01a5, 0x0280, 0x01a8, 0x0283, 0x01ad, 0x0288, 0x01b0, 0x028a, 0x028b, 0x01b4, 0x01b6, 0x0292,
+            0x01b9, 0x01bd, 0x01c6, 0x01c6, 0x01c9, 0x01c9, 0x01cc, 0x01cc, 0x01dc, 0x01df, 0x01ef, 0x01f3, 0x01f3,
+            0x01f5, 0x0195, 0x01bf, 0x01f9, 0x021f, 0x019e, 0x0223, 0x0233, 0x2c65, 0x023c, 0x019a, 0x2c66, 0x0242,
+            0x0180, 0x0289, 0x028c, 0x0247, 0x024f, 0x03b9, 0x0371, 0x0373, 0x0377, 0x03f3, 0x03ac, 0x03ad, 0x03af,
+            0x03cc, 0x03cd, 0x03ce, 0x03b1, 0x03c1, 0x03c3, 0x03cb, 0x03c3, 0x03d7, 0x03b2, 0x03b8, 0x03c6, 0x03c0,
+            0x03d9, 0x03ef, 0x03ba, 0x03c1, 0x03b8, 0x03b5, 0x03f8, 0x03f2, 0x03fb, 0x037b, 0x037d, 0x0450, 0x045f,
+            0x0430, 0x044f, 0x0461, 0x0481, 0x048b, 0x04bf, 0x04cf, 0x04c2, 0x04ce, 0x04d1, 0x052f, 0x0561, 0x0586,
+            0x2d00, 0x2d25, 0x2d27, 0x2d2d, 0x13f0, 0x13f5, 0x0432, 0x0434, 0x043e, 0x0441, 0x0442, 0x0442, 0x044a,
+            0x0463, 0xa64b, 0x10d0, 0x10fa, 0x10fd, 0x10ff, 0x1e01, 0x1e95, 0x1e61, 0x1ea1, 0x1eff, 0x1f00, 0x1f07,
+            0x1f10, 0x1f15, 0x1f20, 0x1f27, 0x1f30, 0x1f37, 0x1f40, 0x1f45, 0x1f51, 0x1f53, 0x1f55, 0x1f57, 0x1f60,
+            0x1f67, 0x1fb0, 0x1fb1, 0x1f70, 0x1f71, 0x03b9, 0x1f72, 0x1f75, 0x1fd0, 0x1fd1, 0x1f76, 0x1f77, 0x1fe0,
+            0x1fe1, 0x1f7a, 0x1f7b, 0x1fe5, 0x1f78, 0x1f79, 0x1f7c, 0x1f7d, 0x03c9, 0x006b, 0x00e5, 0x214e, 0x2170,
+            0x217f, 0x2184, 0x24d0, 0x24e9, 0x2c30, 0x2c5e, 0x2c61, 0x026b, 0x1d7d, 0x027d, 0x2c68, 0x2c6c, 0x0251,
+            0x0271, 0x0250, 0x0252, 0x2c73, 0x2c76, 0x023f, 0x0240, 0x2c81, 0x2ce3, 0x2cec, 0x2cee, 0x2cf3, 0xa641,
+            0xa66d, 0xa681, 0xa69b, 0xa723, 0xa72f, 0xa733, 0xa76f, 0xa77a, 0xa77c, 0x1d79, 0xa77f, 0xa787, 0xa78c,
+            0x0265, 0xa791, 0xa793, 0xa797, 0xa7a9, 0x0266, 0x025c, 0x0261, 0x026c, 0x026a, 0x029e, 0x0287, 0x029d,
+            0xab53, 0xa7b5, 0xa7bf, 0xa7c3, 0xa794, 0x0282, 0x1d8e, 0xa7c8, 0xa7ca, 0xa7f6, 0x13a0, 0x13ef, 0xff41,
+            0xff5a, 0x10428, 0x1044f, 0x104d8, 0x104fb, 0x10cc0, 0x10cf2, 0x118c0, 0x118df, 0x16e60, 0x16e7f, 0x1e922,
+            0x1e943
         };
         static const unsigned FOLD_MAP_2[] = {
             S(0x00df), S(0x0130), S(0x0149), S(0x01f0), S(0x0587), S(0x1e96), S(0x1e97), S(0x1e98), S(0x1e99),