compute Sentence_Break as well
[disorder] / lib / unidata.h
CommitLineData
e5a5a138 1/* Automatically generated file, see scripts/make-unidata */
61507e3c
RK
2#ifndef UNIDATA_H
3#define UNIDATA_H
4enum unicode_gc_cat {
5 unicode_gc_Cc,
6 unicode_gc_Cf,
7 unicode_gc_Cn,
8 unicode_gc_Co,
9 unicode_gc_Cs,
10 unicode_gc_Ll,
11 unicode_gc_Lm,
12 unicode_gc_Lo,
13 unicode_gc_Lt,
14 unicode_gc_Lu,
15 unicode_gc_Mc,
16 unicode_gc_Me,
17 unicode_gc_Mn,
18 unicode_gc_Nd,
19 unicode_gc_Nl,
20 unicode_gc_No,
21 unicode_gc_Pc,
22 unicode_gc_Pd,
23 unicode_gc_Pe,
24 unicode_gc_Pf,
25 unicode_gc_Pi,
26 unicode_gc_Po,
27 unicode_gc_Ps,
28 unicode_gc_Sc,
29 unicode_gc_Sk,
30 unicode_gc_Sm,
31 unicode_gc_So,
32 unicode_gc_Zl,
33 unicode_gc_Zp,
34 unicode_gc_Zs
35};
349b7b74
RK
36enum unicode_Grapheme_Break {
37 unicode_Grapheme_Break_CR,
38 unicode_Grapheme_Break_Control,
39 unicode_Grapheme_Break_Extend,
40 unicode_Grapheme_Break_L,
41 unicode_Grapheme_Break_LF,
42 unicode_Grapheme_Break_LV,
43 unicode_Grapheme_Break_LVT,
44 unicode_Grapheme_Break_Other,
45 unicode_Grapheme_Break_T,
46 unicode_Grapheme_Break_V
47};
48extern const char *const unicode_Grapheme_Break_names[];
0b7052da
RK
49enum unicode_Word_Break {
50 unicode_Word_Break_ALetter,
51 unicode_Word_Break_Extend,
52 unicode_Word_Break_ExtendNumLet,
53 unicode_Word_Break_Format,
54 unicode_Word_Break_Katakana,
55 unicode_Word_Break_MidLetter,
56 unicode_Word_Break_MidNum,
57 unicode_Word_Break_Numeric,
58 unicode_Word_Break_Other
59};
bb48024f 60extern const char *const unicode_Word_Break_names[];
349b7b74
RK
61enum unicode_Sentence_Break {
62 unicode_Sentence_Break_ATerm,
63 unicode_Sentence_Break_Close,
64 unicode_Sentence_Break_Extend,
65 unicode_Sentence_Break_Format,
66 unicode_Sentence_Break_Lower,
67 unicode_Sentence_Break_Numeric,
68 unicode_Sentence_Break_OLetter,
69 unicode_Sentence_Break_Other,
70 unicode_Sentence_Break_STerm,
71 unicode_Sentence_Break_Sep,
72 unicode_Sentence_Break_Sp,
73 unicode_Sentence_Break_Upper
74};
75extern const char *const unicode_Sentence_Break_names[];
e5a5a138 76enum unicode_flags {
0b7052da 77 unicode_normalize_before_casefold = 1
e5a5a138
RK
78};
79
61507e3c 80struct unidata {
e5a5a138
RK
81 const uint32_t *compat;
82 const uint32_t *canon;
83 const uint32_t *casefold;
84 int16_t upper_offset;
85 int16_t lower_offset;
86 unsigned char ccc;
87 char gc;
88 uint8_t flags;
349b7b74 89 char grapheme_break;
0b7052da 90 char word_break;
349b7b74 91 char sentence_break;
61507e3c
RK
92};
93extern const struct unidata *const unidata[];
e5a5a138
RK
94#define UNICODE_NCHARS 195200
95#define UNICODE_MODULUS 128
61507e3c 96#endif