|
1 # This file is generated by mkstringprep.py. DO NOT EDIT. |
|
2 """Library that exposes various tables found in the StringPrep RFC 3454. |
|
3 |
|
4 There are two kinds of tables: sets, for which a member test is provided, |
|
5 and mappings, for which a mapping function is provided. |
|
6 """ |
|
7 |
|
8 from unicodedata import ucd_3_2_0 as unicodedata |
|
9 |
|
10 assert unicodedata.unidata_version == '3.2.0' |
|
11 |
|
12 def in_table_a1(code): |
|
13 if unicodedata.category(code) != 'Cn': return False |
|
14 c = ord(code) |
|
15 if 0xFDD0 <= c < 0xFDF0: return False |
|
16 return (c & 0xFFFF) not in (0xFFFE, 0xFFFF) |
|
17 |
|
18 |
|
19 b1_set = set([173, 847, 6150, 6155, 6156, 6157, 8203, 8204, 8205, 8288, 65279] + range(65024,65040)) |
|
20 def in_table_b1(code): |
|
21 return ord(code) in b1_set |
|
22 |
|
23 |
|
24 b3_exceptions = { |
|
25 0xb5:u'\u03bc', 0xdf:u'ss', 0x130:u'i\u0307', 0x149:u'\u02bcn', |
|
26 0x17f:u's', 0x1f0:u'j\u030c', 0x345:u'\u03b9', 0x37a:u' \u03b9', |
|
27 0x390:u'\u03b9\u0308\u0301', 0x3b0:u'\u03c5\u0308\u0301', 0x3c2:u'\u03c3', 0x3d0:u'\u03b2', |
|
28 0x3d1:u'\u03b8', 0x3d2:u'\u03c5', 0x3d3:u'\u03cd', 0x3d4:u'\u03cb', |
|
29 0x3d5:u'\u03c6', 0x3d6:u'\u03c0', 0x3f0:u'\u03ba', 0x3f1:u'\u03c1', |
|
30 0x3f2:u'\u03c3', 0x3f5:u'\u03b5', 0x587:u'\u0565\u0582', 0x1e96:u'h\u0331', |
|
31 0x1e97:u't\u0308', 0x1e98:u'w\u030a', 0x1e99:u'y\u030a', 0x1e9a:u'a\u02be', |
|
32 0x1e9b:u'\u1e61', 0x1f50:u'\u03c5\u0313', 0x1f52:u'\u03c5\u0313\u0300', 0x1f54:u'\u03c5\u0313\u0301', |
|
33 0x1f56:u'\u03c5\u0313\u0342', 0x1f80:u'\u1f00\u03b9', 0x1f81:u'\u1f01\u03b9', 0x1f82:u'\u1f02\u03b9', |
|
34 0x1f83:u'\u1f03\u03b9', 0x1f84:u'\u1f04\u03b9', 0x1f85:u'\u1f05\u03b9', 0x1f86:u'\u1f06\u03b9', |
|
35 0x1f87:u'\u1f07\u03b9', 0x1f88:u'\u1f00\u03b9', 0x1f89:u'\u1f01\u03b9', 0x1f8a:u'\u1f02\u03b9', |
|
36 0x1f8b:u'\u1f03\u03b9', 0x1f8c:u'\u1f04\u03b9', 0x1f8d:u'\u1f05\u03b9', 0x1f8e:u'\u1f06\u03b9', |
|
37 0x1f8f:u'\u1f07\u03b9', 0x1f90:u'\u1f20\u03b9', 0x1f91:u'\u1f21\u03b9', 0x1f92:u'\u1f22\u03b9', |
|
38 0x1f93:u'\u1f23\u03b9', 0x1f94:u'\u1f24\u03b9', 0x1f95:u'\u1f25\u03b9', 0x1f96:u'\u1f26\u03b9', |
|
39 0x1f97:u'\u1f27\u03b9', 0x1f98:u'\u1f20\u03b9', 0x1f99:u'\u1f21\u03b9', 0x1f9a:u'\u1f22\u03b9', |
|
40 0x1f9b:u'\u1f23\u03b9', 0x1f9c:u'\u1f24\u03b9', 0x1f9d:u'\u1f25\u03b9', 0x1f9e:u'\u1f26\u03b9', |
|
41 0x1f9f:u'\u1f27\u03b9', 0x1fa0:u'\u1f60\u03b9', 0x1fa1:u'\u1f61\u03b9', 0x1fa2:u'\u1f62\u03b9', |
|
42 0x1fa3:u'\u1f63\u03b9', 0x1fa4:u'\u1f64\u03b9', 0x1fa5:u'\u1f65\u03b9', 0x1fa6:u'\u1f66\u03b9', |
|
43 0x1fa7:u'\u1f67\u03b9', 0x1fa8:u'\u1f60\u03b9', 0x1fa9:u'\u1f61\u03b9', 0x1faa:u'\u1f62\u03b9', |
|
44 0x1fab:u'\u1f63\u03b9', 0x1fac:u'\u1f64\u03b9', 0x1fad:u'\u1f65\u03b9', 0x1fae:u'\u1f66\u03b9', |
|
45 0x1faf:u'\u1f67\u03b9', 0x1fb2:u'\u1f70\u03b9', 0x1fb3:u'\u03b1\u03b9', 0x1fb4:u'\u03ac\u03b9', |
|
46 0x1fb6:u'\u03b1\u0342', 0x1fb7:u'\u03b1\u0342\u03b9', 0x1fbc:u'\u03b1\u03b9', 0x1fbe:u'\u03b9', |
|
47 0x1fc2:u'\u1f74\u03b9', 0x1fc3:u'\u03b7\u03b9', 0x1fc4:u'\u03ae\u03b9', 0x1fc6:u'\u03b7\u0342', |
|
48 0x1fc7:u'\u03b7\u0342\u03b9', 0x1fcc:u'\u03b7\u03b9', 0x1fd2:u'\u03b9\u0308\u0300', 0x1fd3:u'\u03b9\u0308\u0301', |
|
49 0x1fd6:u'\u03b9\u0342', 0x1fd7:u'\u03b9\u0308\u0342', 0x1fe2:u'\u03c5\u0308\u0300', 0x1fe3:u'\u03c5\u0308\u0301', |
|
50 0x1fe4:u'\u03c1\u0313', 0x1fe6:u'\u03c5\u0342', 0x1fe7:u'\u03c5\u0308\u0342', 0x1ff2:u'\u1f7c\u03b9', |
|
51 0x1ff3:u'\u03c9\u03b9', 0x1ff4:u'\u03ce\u03b9', 0x1ff6:u'\u03c9\u0342', 0x1ff7:u'\u03c9\u0342\u03b9', |
|
52 0x1ffc:u'\u03c9\u03b9', 0x20a8:u'rs', 0x2102:u'c', 0x2103:u'\xb0c', |
|
53 0x2107:u'\u025b', 0x2109:u'\xb0f', 0x210b:u'h', 0x210c:u'h', |
|
54 0x210d:u'h', 0x2110:u'i', 0x2111:u'i', 0x2112:u'l', |
|
55 0x2115:u'n', 0x2116:u'no', 0x2119:u'p', 0x211a:u'q', |
|
56 0x211b:u'r', 0x211c:u'r', 0x211d:u'r', 0x2120:u'sm', |
|
57 0x2121:u'tel', 0x2122:u'tm', 0x2124:u'z', 0x2128:u'z', |
|
58 0x212c:u'b', 0x212d:u'c', 0x2130:u'e', 0x2131:u'f', |
|
59 0x2133:u'm', 0x213e:u'\u03b3', 0x213f:u'\u03c0', 0x2145:u'd', |
|
60 0x3371:u'hpa', 0x3373:u'au', 0x3375:u'ov', 0x3380:u'pa', |
|
61 0x3381:u'na', 0x3382:u'\u03bca', 0x3383:u'ma', 0x3384:u'ka', |
|
62 0x3385:u'kb', 0x3386:u'mb', 0x3387:u'gb', 0x338a:u'pf', |
|
63 0x338b:u'nf', 0x338c:u'\u03bcf', 0x3390:u'hz', 0x3391:u'khz', |
|
64 0x3392:u'mhz', 0x3393:u'ghz', 0x3394:u'thz', 0x33a9:u'pa', |
|
65 0x33aa:u'kpa', 0x33ab:u'mpa', 0x33ac:u'gpa', 0x33b4:u'pv', |
|
66 0x33b5:u'nv', 0x33b6:u'\u03bcv', 0x33b7:u'mv', 0x33b8:u'kv', |
|
67 0x33b9:u'mv', 0x33ba:u'pw', 0x33bb:u'nw', 0x33bc:u'\u03bcw', |
|
68 0x33bd:u'mw', 0x33be:u'kw', 0x33bf:u'mw', 0x33c0:u'k\u03c9', |
|
69 0x33c1:u'm\u03c9', 0x33c3:u'bq', 0x33c6:u'c\u2215kg', 0x33c7:u'co.', |
|
70 0x33c8:u'db', 0x33c9:u'gy', 0x33cb:u'hp', 0x33cd:u'kk', |
|
71 0x33ce:u'km', 0x33d7:u'ph', 0x33d9:u'ppm', 0x33da:u'pr', |
|
72 0x33dc:u'sv', 0x33dd:u'wb', 0xfb00:u'ff', 0xfb01:u'fi', |
|
73 0xfb02:u'fl', 0xfb03:u'ffi', 0xfb04:u'ffl', 0xfb05:u'st', |
|
74 0xfb06:u'st', 0xfb13:u'\u0574\u0576', 0xfb14:u'\u0574\u0565', 0xfb15:u'\u0574\u056b', |
|
75 0xfb16:u'\u057e\u0576', 0xfb17:u'\u0574\u056d', 0x1d400:u'a', 0x1d401:u'b', |
|
76 0x1d402:u'c', 0x1d403:u'd', 0x1d404:u'e', 0x1d405:u'f', |
|
77 0x1d406:u'g', 0x1d407:u'h', 0x1d408:u'i', 0x1d409:u'j', |
|
78 0x1d40a:u'k', 0x1d40b:u'l', 0x1d40c:u'm', 0x1d40d:u'n', |
|
79 0x1d40e:u'o', 0x1d40f:u'p', 0x1d410:u'q', 0x1d411:u'r', |
|
80 0x1d412:u's', 0x1d413:u't', 0x1d414:u'u', 0x1d415:u'v', |
|
81 0x1d416:u'w', 0x1d417:u'x', 0x1d418:u'y', 0x1d419:u'z', |
|
82 0x1d434:u'a', 0x1d435:u'b', 0x1d436:u'c', 0x1d437:u'd', |
|
83 0x1d438:u'e', 0x1d439:u'f', 0x1d43a:u'g', 0x1d43b:u'h', |
|
84 0x1d43c:u'i', 0x1d43d:u'j', 0x1d43e:u'k', 0x1d43f:u'l', |
|
85 0x1d440:u'm', 0x1d441:u'n', 0x1d442:u'o', 0x1d443:u'p', |
|
86 0x1d444:u'q', 0x1d445:u'r', 0x1d446:u's', 0x1d447:u't', |
|
87 0x1d448:u'u', 0x1d449:u'v', 0x1d44a:u'w', 0x1d44b:u'x', |
|
88 0x1d44c:u'y', 0x1d44d:u'z', 0x1d468:u'a', 0x1d469:u'b', |
|
89 0x1d46a:u'c', 0x1d46b:u'd', 0x1d46c:u'e', 0x1d46d:u'f', |
|
90 0x1d46e:u'g', 0x1d46f:u'h', 0x1d470:u'i', 0x1d471:u'j', |
|
91 0x1d472:u'k', 0x1d473:u'l', 0x1d474:u'm', 0x1d475:u'n', |
|
92 0x1d476:u'o', 0x1d477:u'p', 0x1d478:u'q', 0x1d479:u'r', |
|
93 0x1d47a:u's', 0x1d47b:u't', 0x1d47c:u'u', 0x1d47d:u'v', |
|
94 0x1d47e:u'w', 0x1d47f:u'x', 0x1d480:u'y', 0x1d481:u'z', |
|
95 0x1d49c:u'a', 0x1d49e:u'c', 0x1d49f:u'd', 0x1d4a2:u'g', |
|
96 0x1d4a5:u'j', 0x1d4a6:u'k', 0x1d4a9:u'n', 0x1d4aa:u'o', |
|
97 0x1d4ab:u'p', 0x1d4ac:u'q', 0x1d4ae:u's', 0x1d4af:u't', |
|
98 0x1d4b0:u'u', 0x1d4b1:u'v', 0x1d4b2:u'w', 0x1d4b3:u'x', |
|
99 0x1d4b4:u'y', 0x1d4b5:u'z', 0x1d4d0:u'a', 0x1d4d1:u'b', |
|
100 0x1d4d2:u'c', 0x1d4d3:u'd', 0x1d4d4:u'e', 0x1d4d5:u'f', |
|
101 0x1d4d6:u'g', 0x1d4d7:u'h', 0x1d4d8:u'i', 0x1d4d9:u'j', |
|
102 0x1d4da:u'k', 0x1d4db:u'l', 0x1d4dc:u'm', 0x1d4dd:u'n', |
|
103 0x1d4de:u'o', 0x1d4df:u'p', 0x1d4e0:u'q', 0x1d4e1:u'r', |
|
104 0x1d4e2:u's', 0x1d4e3:u't', 0x1d4e4:u'u', 0x1d4e5:u'v', |
|
105 0x1d4e6:u'w', 0x1d4e7:u'x', 0x1d4e8:u'y', 0x1d4e9:u'z', |
|
106 0x1d504:u'a', 0x1d505:u'b', 0x1d507:u'd', 0x1d508:u'e', |
|
107 0x1d509:u'f', 0x1d50a:u'g', 0x1d50d:u'j', 0x1d50e:u'k', |
|
108 0x1d50f:u'l', 0x1d510:u'm', 0x1d511:u'n', 0x1d512:u'o', |
|
109 0x1d513:u'p', 0x1d514:u'q', 0x1d516:u's', 0x1d517:u't', |
|
110 0x1d518:u'u', 0x1d519:u'v', 0x1d51a:u'w', 0x1d51b:u'x', |
|
111 0x1d51c:u'y', 0x1d538:u'a', 0x1d539:u'b', 0x1d53b:u'd', |
|
112 0x1d53c:u'e', 0x1d53d:u'f', 0x1d53e:u'g', 0x1d540:u'i', |
|
113 0x1d541:u'j', 0x1d542:u'k', 0x1d543:u'l', 0x1d544:u'm', |
|
114 0x1d546:u'o', 0x1d54a:u's', 0x1d54b:u't', 0x1d54c:u'u', |
|
115 0x1d54d:u'v', 0x1d54e:u'w', 0x1d54f:u'x', 0x1d550:u'y', |
|
116 0x1d56c:u'a', 0x1d56d:u'b', 0x1d56e:u'c', 0x1d56f:u'd', |
|
117 0x1d570:u'e', 0x1d571:u'f', 0x1d572:u'g', 0x1d573:u'h', |
|
118 0x1d574:u'i', 0x1d575:u'j', 0x1d576:u'k', 0x1d577:u'l', |
|
119 0x1d578:u'm', 0x1d579:u'n', 0x1d57a:u'o', 0x1d57b:u'p', |
|
120 0x1d57c:u'q', 0x1d57d:u'r', 0x1d57e:u's', 0x1d57f:u't', |
|
121 0x1d580:u'u', 0x1d581:u'v', 0x1d582:u'w', 0x1d583:u'x', |
|
122 0x1d584:u'y', 0x1d585:u'z', 0x1d5a0:u'a', 0x1d5a1:u'b', |
|
123 0x1d5a2:u'c', 0x1d5a3:u'd', 0x1d5a4:u'e', 0x1d5a5:u'f', |
|
124 0x1d5a6:u'g', 0x1d5a7:u'h', 0x1d5a8:u'i', 0x1d5a9:u'j', |
|
125 0x1d5aa:u'k', 0x1d5ab:u'l', 0x1d5ac:u'm', 0x1d5ad:u'n', |
|
126 0x1d5ae:u'o', 0x1d5af:u'p', 0x1d5b0:u'q', 0x1d5b1:u'r', |
|
127 0x1d5b2:u's', 0x1d5b3:u't', 0x1d5b4:u'u', 0x1d5b5:u'v', |
|
128 0x1d5b6:u'w', 0x1d5b7:u'x', 0x1d5b8:u'y', 0x1d5b9:u'z', |
|
129 0x1d5d4:u'a', 0x1d5d5:u'b', 0x1d5d6:u'c', 0x1d5d7:u'd', |
|
130 0x1d5d8:u'e', 0x1d5d9:u'f', 0x1d5da:u'g', 0x1d5db:u'h', |
|
131 0x1d5dc:u'i', 0x1d5dd:u'j', 0x1d5de:u'k', 0x1d5df:u'l', |
|
132 0x1d5e0:u'm', 0x1d5e1:u'n', 0x1d5e2:u'o', 0x1d5e3:u'p', |
|
133 0x1d5e4:u'q', 0x1d5e5:u'r', 0x1d5e6:u's', 0x1d5e7:u't', |
|
134 0x1d5e8:u'u', 0x1d5e9:u'v', 0x1d5ea:u'w', 0x1d5eb:u'x', |
|
135 0x1d5ec:u'y', 0x1d5ed:u'z', 0x1d608:u'a', 0x1d609:u'b', |
|
136 0x1d60a:u'c', 0x1d60b:u'd', 0x1d60c:u'e', 0x1d60d:u'f', |
|
137 0x1d60e:u'g', 0x1d60f:u'h', 0x1d610:u'i', 0x1d611:u'j', |
|
138 0x1d612:u'k', 0x1d613:u'l', 0x1d614:u'm', 0x1d615:u'n', |
|
139 0x1d616:u'o', 0x1d617:u'p', 0x1d618:u'q', 0x1d619:u'r', |
|
140 0x1d61a:u's', 0x1d61b:u't', 0x1d61c:u'u', 0x1d61d:u'v', |
|
141 0x1d61e:u'w', 0x1d61f:u'x', 0x1d620:u'y', 0x1d621:u'z', |
|
142 0x1d63c:u'a', 0x1d63d:u'b', 0x1d63e:u'c', 0x1d63f:u'd', |
|
143 0x1d640:u'e', 0x1d641:u'f', 0x1d642:u'g', 0x1d643:u'h', |
|
144 0x1d644:u'i', 0x1d645:u'j', 0x1d646:u'k', 0x1d647:u'l', |
|
145 0x1d648:u'm', 0x1d649:u'n', 0x1d64a:u'o', 0x1d64b:u'p', |
|
146 0x1d64c:u'q', 0x1d64d:u'r', 0x1d64e:u's', 0x1d64f:u't', |
|
147 0x1d650:u'u', 0x1d651:u'v', 0x1d652:u'w', 0x1d653:u'x', |
|
148 0x1d654:u'y', 0x1d655:u'z', 0x1d670:u'a', 0x1d671:u'b', |
|
149 0x1d672:u'c', 0x1d673:u'd', 0x1d674:u'e', 0x1d675:u'f', |
|
150 0x1d676:u'g', 0x1d677:u'h', 0x1d678:u'i', 0x1d679:u'j', |
|
151 0x1d67a:u'k', 0x1d67b:u'l', 0x1d67c:u'm', 0x1d67d:u'n', |
|
152 0x1d67e:u'o', 0x1d67f:u'p', 0x1d680:u'q', 0x1d681:u'r', |
|
153 0x1d682:u's', 0x1d683:u't', 0x1d684:u'u', 0x1d685:u'v', |
|
154 0x1d686:u'w', 0x1d687:u'x', 0x1d688:u'y', 0x1d689:u'z', |
|
155 0x1d6a8:u'\u03b1', 0x1d6a9:u'\u03b2', 0x1d6aa:u'\u03b3', 0x1d6ab:u'\u03b4', |
|
156 0x1d6ac:u'\u03b5', 0x1d6ad:u'\u03b6', 0x1d6ae:u'\u03b7', 0x1d6af:u'\u03b8', |
|
157 0x1d6b0:u'\u03b9', 0x1d6b1:u'\u03ba', 0x1d6b2:u'\u03bb', 0x1d6b3:u'\u03bc', |
|
158 0x1d6b4:u'\u03bd', 0x1d6b5:u'\u03be', 0x1d6b6:u'\u03bf', 0x1d6b7:u'\u03c0', |
|
159 0x1d6b8:u'\u03c1', 0x1d6b9:u'\u03b8', 0x1d6ba:u'\u03c3', 0x1d6bb:u'\u03c4', |
|
160 0x1d6bc:u'\u03c5', 0x1d6bd:u'\u03c6', 0x1d6be:u'\u03c7', 0x1d6bf:u'\u03c8', |
|
161 0x1d6c0:u'\u03c9', 0x1d6d3:u'\u03c3', 0x1d6e2:u'\u03b1', 0x1d6e3:u'\u03b2', |
|
162 0x1d6e4:u'\u03b3', 0x1d6e5:u'\u03b4', 0x1d6e6:u'\u03b5', 0x1d6e7:u'\u03b6', |
|
163 0x1d6e8:u'\u03b7', 0x1d6e9:u'\u03b8', 0x1d6ea:u'\u03b9', 0x1d6eb:u'\u03ba', |
|
164 0x1d6ec:u'\u03bb', 0x1d6ed:u'\u03bc', 0x1d6ee:u'\u03bd', 0x1d6ef:u'\u03be', |
|
165 0x1d6f0:u'\u03bf', 0x1d6f1:u'\u03c0', 0x1d6f2:u'\u03c1', 0x1d6f3:u'\u03b8', |
|
166 0x1d6f4:u'\u03c3', 0x1d6f5:u'\u03c4', 0x1d6f6:u'\u03c5', 0x1d6f7:u'\u03c6', |
|
167 0x1d6f8:u'\u03c7', 0x1d6f9:u'\u03c8', 0x1d6fa:u'\u03c9', 0x1d70d:u'\u03c3', |
|
168 0x1d71c:u'\u03b1', 0x1d71d:u'\u03b2', 0x1d71e:u'\u03b3', 0x1d71f:u'\u03b4', |
|
169 0x1d720:u'\u03b5', 0x1d721:u'\u03b6', 0x1d722:u'\u03b7', 0x1d723:u'\u03b8', |
|
170 0x1d724:u'\u03b9', 0x1d725:u'\u03ba', 0x1d726:u'\u03bb', 0x1d727:u'\u03bc', |
|
171 0x1d728:u'\u03bd', 0x1d729:u'\u03be', 0x1d72a:u'\u03bf', 0x1d72b:u'\u03c0', |
|
172 0x1d72c:u'\u03c1', 0x1d72d:u'\u03b8', 0x1d72e:u'\u03c3', 0x1d72f:u'\u03c4', |
|
173 0x1d730:u'\u03c5', 0x1d731:u'\u03c6', 0x1d732:u'\u03c7', 0x1d733:u'\u03c8', |
|
174 0x1d734:u'\u03c9', 0x1d747:u'\u03c3', 0x1d756:u'\u03b1', 0x1d757:u'\u03b2', |
|
175 0x1d758:u'\u03b3', 0x1d759:u'\u03b4', 0x1d75a:u'\u03b5', 0x1d75b:u'\u03b6', |
|
176 0x1d75c:u'\u03b7', 0x1d75d:u'\u03b8', 0x1d75e:u'\u03b9', 0x1d75f:u'\u03ba', |
|
177 0x1d760:u'\u03bb', 0x1d761:u'\u03bc', 0x1d762:u'\u03bd', 0x1d763:u'\u03be', |
|
178 0x1d764:u'\u03bf', 0x1d765:u'\u03c0', 0x1d766:u'\u03c1', 0x1d767:u'\u03b8', |
|
179 0x1d768:u'\u03c3', 0x1d769:u'\u03c4', 0x1d76a:u'\u03c5', 0x1d76b:u'\u03c6', |
|
180 0x1d76c:u'\u03c7', 0x1d76d:u'\u03c8', 0x1d76e:u'\u03c9', 0x1d781:u'\u03c3', |
|
181 0x1d790:u'\u03b1', 0x1d791:u'\u03b2', 0x1d792:u'\u03b3', 0x1d793:u'\u03b4', |
|
182 0x1d794:u'\u03b5', 0x1d795:u'\u03b6', 0x1d796:u'\u03b7', 0x1d797:u'\u03b8', |
|
183 0x1d798:u'\u03b9', 0x1d799:u'\u03ba', 0x1d79a:u'\u03bb', 0x1d79b:u'\u03bc', |
|
184 0x1d79c:u'\u03bd', 0x1d79d:u'\u03be', 0x1d79e:u'\u03bf', 0x1d79f:u'\u03c0', |
|
185 0x1d7a0:u'\u03c1', 0x1d7a1:u'\u03b8', 0x1d7a2:u'\u03c3', 0x1d7a3:u'\u03c4', |
|
186 0x1d7a4:u'\u03c5', 0x1d7a5:u'\u03c6', 0x1d7a6:u'\u03c7', 0x1d7a7:u'\u03c8', |
|
187 0x1d7a8:u'\u03c9', 0x1d7bb:u'\u03c3', } |
|
188 |
|
189 def map_table_b3(code): |
|
190 r = b3_exceptions.get(ord(code)) |
|
191 if r is not None: return r |
|
192 return code.lower() |
|
193 |
|
194 |
|
195 def map_table_b2(a): |
|
196 al = map_table_b3(a) |
|
197 b = unicodedata.normalize("NFKC", al) |
|
198 bl = u"".join([map_table_b3(ch) for ch in b]) |
|
199 c = unicodedata.normalize("NFKC", bl) |
|
200 if b != c: |
|
201 return c |
|
202 else: |
|
203 return al |
|
204 |
|
205 |
|
206 def in_table_c11(code): |
|
207 return code == u" " |
|
208 |
|
209 |
|
210 def in_table_c12(code): |
|
211 return unicodedata.category(code) == "Zs" and code != u" " |
|
212 |
|
213 def in_table_c11_c12(code): |
|
214 return unicodedata.category(code) == "Zs" |
|
215 |
|
216 |
|
217 def in_table_c21(code): |
|
218 return ord(code) < 128 and unicodedata.category(code) == "Cc" |
|
219 |
|
220 c22_specials = set([1757, 1807, 6158, 8204, 8205, 8232, 8233, 65279] + range(8288,8292) + range(8298,8304) + range(65529,65533) + range(119155,119163)) |
|
221 def in_table_c22(code): |
|
222 c = ord(code) |
|
223 if c < 128: return False |
|
224 if unicodedata.category(code) == "Cc": return True |
|
225 return c in c22_specials |
|
226 |
|
227 def in_table_c21_c22(code): |
|
228 return unicodedata.category(code) == "Cc" or \ |
|
229 ord(code) in c22_specials |
|
230 |
|
231 |
|
232 def in_table_c3(code): |
|
233 return unicodedata.category(code) == "Co" |
|
234 |
|
235 |
|
236 def in_table_c4(code): |
|
237 c = ord(code) |
|
238 if c < 0xFDD0: return False |
|
239 if c < 0xFDF0: return True |
|
240 return (ord(code) & 0xFFFF) in (0xFFFE, 0xFFFF) |
|
241 |
|
242 |
|
243 def in_table_c5(code): |
|
244 return unicodedata.category(code) == "Cs" |
|
245 |
|
246 |
|
247 c6_set = set(range(65529,65534)) |
|
248 def in_table_c6(code): |
|
249 return ord(code) in c6_set |
|
250 |
|
251 |
|
252 c7_set = set(range(12272,12284)) |
|
253 def in_table_c7(code): |
|
254 return ord(code) in c7_set |
|
255 |
|
256 |
|
257 c8_set = set([832, 833, 8206, 8207] + range(8234,8239) + range(8298,8304)) |
|
258 def in_table_c8(code): |
|
259 return ord(code) in c8_set |
|
260 |
|
261 |
|
262 c9_set = set([917505] + range(917536,917632)) |
|
263 def in_table_c9(code): |
|
264 return ord(code) in c9_set |
|
265 |
|
266 |
|
267 def in_table_d1(code): |
|
268 return unicodedata.bidirectional(code) in ("R","AL") |
|
269 |
|
270 |
|
271 def in_table_d2(code): |
|
272 return unicodedata.bidirectional(code) == "L" |