1 | /**********************************************************************
|
---|
2 | unicode.c - Oniguruma (regular expression library)
|
---|
3 | **********************************************************************/
|
---|
4 | /*-
|
---|
5 | * Copyright (c) 2002-2013 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
|
---|
6 | * All rights reserved.
|
---|
7 | *
|
---|
8 | * Redistribution and use in source and binary forms, with or without
|
---|
9 | * modification, are permitted provided that the following conditions
|
---|
10 | * are met:
|
---|
11 | * 1. Redistributions of source code must retain the above copyright
|
---|
12 | * notice, this list of conditions and the following disclaimer.
|
---|
13 | * 2. Redistributions in binary form must reproduce the above copyright
|
---|
14 | * notice, this list of conditions and the following disclaimer in the
|
---|
15 | * documentation and/or other materials provided with the distribution.
|
---|
16 | *
|
---|
17 | * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
|
---|
18 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
---|
19 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
---|
20 | * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
|
---|
21 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
---|
22 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
|
---|
23 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
|
---|
24 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
---|
25 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
|
---|
26 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
---|
27 | * SUCH DAMAGE.
|
---|
28 | */
|
---|
29 |
|
---|
30 | #include "regint.h"
|
---|
31 |
|
---|
32 | #define ONIGENC_IS_UNICODE_ISO_8859_1_CTYPE(code,ctype) \
|
---|
33 | ((EncUNICODE_ISO_8859_1_CtypeTable[code] & CTYPE_TO_BIT(ctype)) != 0)
|
---|
34 | #if 0
|
---|
35 | #define ONIGENC_IS_UNICODE_ISO_8859_1_BIT_CTYPE(code,cbit) \
|
---|
36 | ((EncUNICODE_ISO_8859_1_CtypeTable[code] & (cbit)) != 0)
|
---|
37 | #endif
|
---|
38 |
|
---|
39 | static const unsigned short EncUNICODE_ISO_8859_1_CtypeTable[256] = {
|
---|
40 | 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008,
|
---|
41 | 0x4008, 0x420c, 0x4209, 0x4208, 0x4208, 0x4208, 0x4008, 0x4008,
|
---|
42 | 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008,
|
---|
43 | 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008,
|
---|
44 | 0x4284, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0,
|
---|
45 | 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0,
|
---|
46 | 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0,
|
---|
47 | 0x78b0, 0x78b0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0,
|
---|
48 | 0x41a0, 0x7ca2, 0x7ca2, 0x7ca2, 0x7ca2, 0x7ca2, 0x7ca2, 0x74a2,
|
---|
49 | 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2,
|
---|
50 | 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2,
|
---|
51 | 0x74a2, 0x74a2, 0x74a2, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x51a0,
|
---|
52 | 0x41a0, 0x78e2, 0x78e2, 0x78e2, 0x78e2, 0x78e2, 0x78e2, 0x70e2,
|
---|
53 | 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2,
|
---|
54 | 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2,
|
---|
55 | 0x70e2, 0x70e2, 0x70e2, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x4008,
|
---|
56 | 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0288, 0x0008, 0x0008,
|
---|
57 | 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008,
|
---|
58 | 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008,
|
---|
59 | 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008,
|
---|
60 | 0x0284, 0x01a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0,
|
---|
61 | 0x00a0, 0x00a0, 0x30e2, 0x01a0, 0x00a0, 0x00a8, 0x00a0, 0x00a0,
|
---|
62 | 0x00a0, 0x00a0, 0x10a0, 0x10a0, 0x00a0, 0x30e2, 0x00a0, 0x01a0,
|
---|
63 | 0x00a0, 0x10a0, 0x30e2, 0x01a0, 0x10a0, 0x10a0, 0x10a0, 0x01a0,
|
---|
64 | 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2,
|
---|
65 | 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2,
|
---|
66 | 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x00a0,
|
---|
67 | 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x30e2,
|
---|
68 | 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2,
|
---|
69 | 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2,
|
---|
70 | 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x00a0,
|
---|
71 | 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2
|
---|
72 | };
|
---|
73 |
|
---|
74 | typedef struct {
|
---|
75 | int n;
|
---|
76 | OnigCodePoint code[3];
|
---|
77 | } CodePointList3;
|
---|
78 |
|
---|
79 | typedef struct {
|
---|
80 | OnigCodePoint from;
|
---|
81 | CodePointList3 to;
|
---|
82 | } CaseFold_11_Type;
|
---|
83 |
|
---|
84 | typedef struct {
|
---|
85 | OnigCodePoint from;
|
---|
86 | CodePointList3 to;
|
---|
87 | } CaseUnfold_11_Type;
|
---|
88 |
|
---|
89 | typedef struct {
|
---|
90 | int n;
|
---|
91 | OnigCodePoint code[2];
|
---|
92 | } CodePointList2;
|
---|
93 |
|
---|
94 | typedef struct {
|
---|
95 | OnigCodePoint from[2];
|
---|
96 | CodePointList2 to;
|
---|
97 | } CaseUnfold_12_Type;
|
---|
98 |
|
---|
99 | typedef struct {
|
---|
100 | OnigCodePoint from[3];
|
---|
101 | CodePointList2 to;
|
---|
102 | } CaseUnfold_13_Type;
|
---|
103 |
|
---|
104 | static inline int
|
---|
105 | bits_of(const OnigCodePoint c, const int n)
|
---|
106 | {
|
---|
107 | return (c >> (2 - n) * 7) & 127;
|
---|
108 | }
|
---|
109 |
|
---|
110 | static inline int
|
---|
111 | bits_at(const OnigCodePoint *c, const int n)
|
---|
112 | {
|
---|
113 | return bits_of(c[n / 3], n % 3);
|
---|
114 | }
|
---|
115 |
|
---|
116 | static int
|
---|
117 | code1_equal(const OnigCodePoint x, const OnigCodePoint y)
|
---|
118 | {
|
---|
119 | if (x != y) return 0;
|
---|
120 | return 1;
|
---|
121 | }
|
---|
122 |
|
---|
123 | static int
|
---|
124 | code2_equal(const OnigCodePoint *x, const OnigCodePoint *y)
|
---|
125 | {
|
---|
126 | if (x[0] != y[0]) return 0;
|
---|
127 | if (x[1] != y[1]) return 0;
|
---|
128 | return 1;
|
---|
129 | }
|
---|
130 |
|
---|
131 | static int
|
---|
132 | code3_equal(const OnigCodePoint *x, const OnigCodePoint *y)
|
---|
133 | {
|
---|
134 | if (x[0] != y[0]) return 0;
|
---|
135 | if (x[1] != y[1]) return 0;
|
---|
136 | if (x[2] != y[2]) return 0;
|
---|
137 | return 1;
|
---|
138 | }
|
---|
139 |
|
---|
140 | #include "enc/unicode/casefold.h"
|
---|
141 |
|
---|
142 | #include "enc/unicode/name2ctype.h"
|
---|
143 |
|
---|
144 | #define CODE_RANGES_NUM numberof(CodeRanges)
|
---|
145 |
|
---|
146 | extern int
|
---|
147 | onigenc_unicode_is_code_ctype(OnigCodePoint code, unsigned int ctype)
|
---|
148 | {
|
---|
149 | if (
|
---|
150 | #ifdef USE_UNICODE_PROPERTIES
|
---|
151 | ctype <= ONIGENC_MAX_STD_CTYPE &&
|
---|
152 | #endif
|
---|
153 | code < 256) {
|
---|
154 | return ONIGENC_IS_UNICODE_ISO_8859_1_CTYPE(code, ctype);
|
---|
155 | }
|
---|
156 |
|
---|
157 | if (ctype >= CODE_RANGES_NUM) {
|
---|
158 | return ONIGERR_TYPE_BUG;
|
---|
159 | }
|
---|
160 |
|
---|
161 | return onig_is_in_code_range((UChar* )CodeRanges[ctype], code);
|
---|
162 | }
|
---|
163 |
|
---|
164 |
|
---|
165 | extern int
|
---|
166 | onigenc_unicode_ctype_code_range(int ctype, const OnigCodePoint* ranges[])
|
---|
167 | {
|
---|
168 | if (ctype >= CODE_RANGES_NUM) {
|
---|
169 | return ONIGERR_TYPE_BUG;
|
---|
170 | }
|
---|
171 |
|
---|
172 | *ranges = CodeRanges[ctype];
|
---|
173 |
|
---|
174 | return 0;
|
---|
175 | }
|
---|
176 |
|
---|
177 | extern int
|
---|
178 | onigenc_utf16_32_get_ctype_code_range(OnigCtype ctype, OnigCodePoint* sb_out,
|
---|
179 | const OnigCodePoint* ranges[])
|
---|
180 | {
|
---|
181 | *sb_out = 0x00;
|
---|
182 | return onigenc_unicode_ctype_code_range(ctype, ranges);
|
---|
183 | }
|
---|
184 |
|
---|
185 | #include "st.h"
|
---|
186 |
|
---|
187 | #define PROPERTY_NAME_MAX_SIZE (MAX_WORD_LENGTH + 1)
|
---|
188 |
|
---|
189 | extern int
|
---|
190 | onigenc_unicode_property_name_to_ctype(OnigEncoding enc, UChar* name, UChar* end)
|
---|
191 | {
|
---|
192 | int len;
|
---|
193 | int ctype;
|
---|
194 | UChar buf[PROPERTY_NAME_MAX_SIZE];
|
---|
195 | UChar *p;
|
---|
196 | OnigCodePoint code;
|
---|
197 |
|
---|
198 | len = 0;
|
---|
199 | for (p = name; p < end; p += enclen(enc, p)) {
|
---|
200 | code = ONIGENC_MBC_TO_CODE(enc, p, end);
|
---|
201 | if (code == ' ' || code == '-' || code == '_')
|
---|
202 | continue;
|
---|
203 | if (code >= 0x80)
|
---|
204 | return ONIGERR_INVALID_CHAR_PROPERTY_NAME;
|
---|
205 |
|
---|
206 | buf[len++] = ONIGENC_ASCII_CODE_TO_LOWER_CASE(code);
|
---|
207 | if (len >= PROPERTY_NAME_MAX_SIZE)
|
---|
208 | return ONIGERR_INVALID_CHAR_PROPERTY_NAME;
|
---|
209 | }
|
---|
210 |
|
---|
211 | buf[len] = 0;
|
---|
212 |
|
---|
213 | if ((ctype = uniname2ctype(buf, len)) < 0) {
|
---|
214 | return ONIGERR_INVALID_CHAR_PROPERTY_NAME;
|
---|
215 | }
|
---|
216 |
|
---|
217 | return ctype;
|
---|
218 | }
|
---|
219 |
|
---|
220 | #define onigenc_unicode_fold_lookup onigenc_unicode_CaseFold_11_lookup
|
---|
221 | #define onigenc_unicode_unfold1_lookup onigenc_unicode_CaseUnfold_11_lookup
|
---|
222 | #define onigenc_unicode_unfold2_lookup onigenc_unicode_CaseUnfold_12_lookup
|
---|
223 | #define onigenc_unicode_unfold3_lookup onigenc_unicode_CaseUnfold_13_lookup
|
---|
224 |
|
---|
225 | extern int
|
---|
226 | onigenc_unicode_mbc_case_fold(OnigEncoding enc,
|
---|
227 | OnigCaseFoldType flag ARG_UNUSED, const UChar** pp, const UChar* end,
|
---|
228 | UChar* fold)
|
---|
229 | {
|
---|
230 | const CodePointList3 *to;
|
---|
231 | OnigCodePoint code;
|
---|
232 | int i, len, rlen;
|
---|
233 | const UChar *p = *pp;
|
---|
234 |
|
---|
235 | code = ONIGENC_MBC_TO_CODE(enc, p, end);
|
---|
236 | len = enclen(enc, p);
|
---|
237 | *pp += len;
|
---|
238 |
|
---|
239 | #ifdef USE_UNICODE_CASE_FOLD_TURKISH_AZERI
|
---|
240 | if ((flag & ONIGENC_CASE_FOLD_TURKISH_AZERI) != 0) {
|
---|
241 | if (code == 0x0049) {
|
---|
242 | return ONIGENC_CODE_TO_MBC(enc, 0x0131, fold);
|
---|
243 | }
|
---|
244 | else if (code == 0x0130) {
|
---|
245 | return ONIGENC_CODE_TO_MBC(enc, 0x0069, fold);
|
---|
246 | }
|
---|
247 | }
|
---|
248 | #endif
|
---|
249 |
|
---|
250 | if ((to = onigenc_unicode_fold_lookup(code)) != 0) {
|
---|
251 | if (to->n == 1) {
|
---|
252 | return ONIGENC_CODE_TO_MBC(enc, to->code[0], fold);
|
---|
253 | }
|
---|
254 | #if 0
|
---|
255 | /* NO NEEDS TO CHECK */
|
---|
256 | else if ((flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0)
|
---|
257 | #else
|
---|
258 | else
|
---|
259 | #endif
|
---|
260 | {
|
---|
261 | rlen = 0;
|
---|
262 | for (i = 0; i < to->n; i++) {
|
---|
263 | len = ONIGENC_CODE_TO_MBC(enc, to->code[i], fold);
|
---|
264 | fold += len;
|
---|
265 | rlen += len;
|
---|
266 | }
|
---|
267 | return rlen;
|
---|
268 | }
|
---|
269 | }
|
---|
270 |
|
---|
271 | for (i = 0; i < len; i++) {
|
---|
272 | *fold++ = *p++;
|
---|
273 | }
|
---|
274 | return len;
|
---|
275 | }
|
---|
276 |
|
---|
277 | extern int
|
---|
278 | onigenc_unicode_apply_all_case_fold(OnigCaseFoldType flag,
|
---|
279 | OnigApplyAllCaseFoldFunc f, void* arg)
|
---|
280 | {
|
---|
281 | const CaseUnfold_11_Type* p11;
|
---|
282 | OnigCodePoint code;
|
---|
283 | int i, j, k, r;
|
---|
284 |
|
---|
285 | for (i = 0; i < numberof(CaseUnfold_11); i++) {
|
---|
286 | p11 = &CaseUnfold_11[i];
|
---|
287 | for (j = 0; j < p11->to.n; j++) {
|
---|
288 | code = p11->from;
|
---|
289 | r = (*f)(p11->to.code[j], &code, 1, arg);
|
---|
290 | if (r != 0) return r;
|
---|
291 |
|
---|
292 | code = p11->to.code[j];
|
---|
293 | r = (*f)(p11->from, &code, 1, arg);
|
---|
294 | if (r != 0) return r;
|
---|
295 |
|
---|
296 | for (k = 0; k < j; k++) {
|
---|
297 | r = (*f)(p11->to.code[j], (OnigCodePoint* )(&p11->to.code[k]), 1, arg);
|
---|
298 | if (r != 0) return r;
|
---|
299 |
|
---|
300 | r = (*f)(p11->to.code[k], (OnigCodePoint* )(&p11->to.code[j]), 1, arg);
|
---|
301 | if (r != 0) return r;
|
---|
302 | }
|
---|
303 | }
|
---|
304 | }
|
---|
305 |
|
---|
306 | #ifdef USE_UNICODE_CASE_FOLD_TURKISH_AZERI
|
---|
307 | if ((flag & ONIGENC_CASE_FOLD_TURKISH_AZERI) != 0) {
|
---|
308 | code = 0x0131;
|
---|
309 | r = (*f)(0x0049, &code, 1, arg);
|
---|
310 | if (r != 0) return r;
|
---|
311 | code = 0x0049;
|
---|
312 | r = (*f)(0x0131, &code, 1, arg);
|
---|
313 | if (r != 0) return r;
|
---|
314 |
|
---|
315 | code = 0x0130;
|
---|
316 | r = (*f)(0x0069, &code, 1, arg);
|
---|
317 | if (r != 0) return r;
|
---|
318 | code = 0x0069;
|
---|
319 | r = (*f)(0x0130, &code, 1, arg);
|
---|
320 | if (r != 0) return r;
|
---|
321 | }
|
---|
322 | else {
|
---|
323 | #endif
|
---|
324 | for (i = 0; i < numberof(CaseUnfold_11_Locale); i++) {
|
---|
325 | p11 = &CaseUnfold_11_Locale[i];
|
---|
326 | for (j = 0; j < p11->to.n; j++) {
|
---|
327 | code = p11->from;
|
---|
328 | r = (*f)(p11->to.code[j], &code, 1, arg);
|
---|
329 | if (r != 0) return r;
|
---|
330 |
|
---|
331 | code = p11->to.code[j];
|
---|
332 | r = (*f)(p11->from, &code, 1, arg);
|
---|
333 | if (r != 0) return r;
|
---|
334 |
|
---|
335 | for (k = 0; k < j; k++) {
|
---|
336 | r = (*f)(p11->to.code[j], (OnigCodePoint* )(&p11->to.code[k]),
|
---|
337 | 1, arg);
|
---|
338 | if (r != 0) return r;
|
---|
339 |
|
---|
340 | r = (*f)(p11->to.code[k], (OnigCodePoint* )(&p11->to.code[j]),
|
---|
341 | 1, arg);
|
---|
342 | if (r != 0) return r;
|
---|
343 | }
|
---|
344 | }
|
---|
345 | }
|
---|
346 | #ifdef USE_UNICODE_CASE_FOLD_TURKISH_AZERI
|
---|
347 | }
|
---|
348 | #endif
|
---|
349 |
|
---|
350 | if ((flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0) {
|
---|
351 | for (i = 0; i < numberof(CaseUnfold_12); i++) {
|
---|
352 | for (j = 0; j < CaseUnfold_12[i].to.n; j++) {
|
---|
353 | r = (*f)(CaseUnfold_12[i].to.code[j],
|
---|
354 | (OnigCodePoint* )CaseUnfold_12[i].from, 2, arg);
|
---|
355 | if (r != 0) return r;
|
---|
356 |
|
---|
357 | for (k = 0; k < CaseUnfold_12[i].to.n; k++) {
|
---|
358 | if (k == j) continue;
|
---|
359 |
|
---|
360 | r = (*f)(CaseUnfold_12[i].to.code[j],
|
---|
361 | (OnigCodePoint* )(&CaseUnfold_12[i].to.code[k]), 1, arg);
|
---|
362 | if (r != 0) return r;
|
---|
363 | }
|
---|
364 | }
|
---|
365 | }
|
---|
366 |
|
---|
367 | #ifdef USE_UNICODE_CASE_FOLD_TURKISH_AZERI
|
---|
368 | if ((flag & ONIGENC_CASE_FOLD_TURKISH_AZERI) == 0) {
|
---|
369 | #endif
|
---|
370 | for (i = 0; i < numberof(CaseUnfold_12_Locale); i++) {
|
---|
371 | for (j = 0; j < CaseUnfold_12_Locale[i].to.n; j++) {
|
---|
372 | r = (*f)(CaseUnfold_12_Locale[i].to.code[j],
|
---|
373 | (OnigCodePoint* )CaseUnfold_12_Locale[i].from, 2, arg);
|
---|
374 | if (r != 0) return r;
|
---|
375 |
|
---|
376 | for (k = 0; k < CaseUnfold_12_Locale[i].to.n; k++) {
|
---|
377 | if (k == j) continue;
|
---|
378 |
|
---|
379 | r = (*f)(CaseUnfold_12_Locale[i].to.code[j],
|
---|
380 | (OnigCodePoint* )(&CaseUnfold_12_Locale[i].to.code[k]),
|
---|
381 | 1, arg);
|
---|
382 | if (r != 0) return r;
|
---|
383 | }
|
---|
384 | }
|
---|
385 | }
|
---|
386 | #ifdef USE_UNICODE_CASE_FOLD_TURKISH_AZERI
|
---|
387 | }
|
---|
388 | #endif
|
---|
389 |
|
---|
390 | for (i = 0; i < numberof(CaseUnfold_13); i++) {
|
---|
391 | for (j = 0; j < CaseUnfold_13[i].to.n; j++) {
|
---|
392 | r = (*f)(CaseUnfold_13[i].to.code[j],
|
---|
393 | (OnigCodePoint* )CaseUnfold_13[i].from, 3, arg);
|
---|
394 | if (r != 0) return r;
|
---|
395 |
|
---|
396 | for (k = 0; k < CaseUnfold_13[i].to.n; k++) {
|
---|
397 | if (k == j) continue;
|
---|
398 |
|
---|
399 | r = (*f)(CaseUnfold_13[i].to.code[j],
|
---|
400 | (OnigCodePoint* )(&CaseUnfold_13[i].to.code[k]), 1, arg);
|
---|
401 | if (r != 0) return r;
|
---|
402 | }
|
---|
403 | }
|
---|
404 | }
|
---|
405 | }
|
---|
406 |
|
---|
407 | return 0;
|
---|
408 | }
|
---|
409 |
|
---|
410 | extern int
|
---|
411 | onigenc_unicode_get_case_fold_codes_by_str(OnigEncoding enc,
|
---|
412 | OnigCaseFoldType flag, const OnigUChar* p, const OnigUChar* end,
|
---|
413 | OnigCaseFoldCodeItem items[])
|
---|
414 | {
|
---|
415 | int n, i, j, k, len;
|
---|
416 | OnigCodePoint code, codes[3];
|
---|
417 | const CodePointList3 *to, *z3;
|
---|
418 | const CodePointList2 *z2;
|
---|
419 |
|
---|
420 | n = 0;
|
---|
421 |
|
---|
422 | code = ONIGENC_MBC_TO_CODE(enc, p, end);
|
---|
423 | len = enclen(enc, p);
|
---|
424 |
|
---|
425 | #ifdef USE_UNICODE_CASE_FOLD_TURKISH_AZERI
|
---|
426 | if ((flag & ONIGENC_CASE_FOLD_TURKISH_AZERI) != 0) {
|
---|
427 | if (code == 0x0049) {
|
---|
428 | items[0].byte_len = len;
|
---|
429 | items[0].code_len = 1;
|
---|
430 | items[0].code[0] = 0x0131;
|
---|
431 | return 1;
|
---|
432 | }
|
---|
433 | else if (code == 0x0130) {
|
---|
434 | items[0].byte_len = len;
|
---|
435 | items[0].code_len = 1;
|
---|
436 | items[0].code[0] = 0x0069;
|
---|
437 | return 1;
|
---|
438 | }
|
---|
439 | else if (code == 0x0131) {
|
---|
440 | items[0].byte_len = len;
|
---|
441 | items[0].code_len = 1;
|
---|
442 | items[0].code[0] = 0x0049;
|
---|
443 | return 1;
|
---|
444 | }
|
---|
445 | else if (code == 0x0069) {
|
---|
446 | items[0].byte_len = len;
|
---|
447 | items[0].code_len = 1;
|
---|
448 | items[0].code[0] = 0x0130;
|
---|
449 | return 1;
|
---|
450 | }
|
---|
451 | }
|
---|
452 | #endif
|
---|
453 |
|
---|
454 | if ((to = onigenc_unicode_fold_lookup(code)) != 0) {
|
---|
455 | if (to->n == 1) {
|
---|
456 | OnigCodePoint orig_code = code;
|
---|
457 |
|
---|
458 | items[0].byte_len = len;
|
---|
459 | items[0].code_len = 1;
|
---|
460 | items[0].code[0] = to->code[0];
|
---|
461 | n++;
|
---|
462 |
|
---|
463 | code = to->code[0];
|
---|
464 | if ((to = onigenc_unicode_unfold1_lookup(code)) != 0) {
|
---|
465 | for (i = 0; i < to->n; i++) {
|
---|
466 | if (to->code[i] != orig_code) {
|
---|
467 | items[n].byte_len = len;
|
---|
468 | items[n].code_len = 1;
|
---|
469 | items[n].code[0] = to->code[i];
|
---|
470 | n++;
|
---|
471 | }
|
---|
472 | }
|
---|
473 | }
|
---|
474 | }
|
---|
475 | else if ((flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0) {
|
---|
476 | OnigCodePoint cs[3][4];
|
---|
477 | int fn, ncs[3];
|
---|
478 |
|
---|
479 | for (fn = 0; fn < to->n; fn++) {
|
---|
480 | cs[fn][0] = to->code[fn];
|
---|
481 | if ((z3 = onigenc_unicode_unfold1_lookup(cs[fn][0])) != 0) {
|
---|
482 | for (i = 0; i < z3->n; i++) {
|
---|
483 | cs[fn][i+1] = z3->code[i];
|
---|
484 | }
|
---|
485 | ncs[fn] = z3->n + 1;
|
---|
486 | }
|
---|
487 | else
|
---|
488 | ncs[fn] = 1;
|
---|
489 | }
|
---|
490 |
|
---|
491 | if (fn == 2) {
|
---|
492 | for (i = 0; i < ncs[0]; i++) {
|
---|
493 | for (j = 0; j < ncs[1]; j++) {
|
---|
494 | items[n].byte_len = len;
|
---|
495 | items[n].code_len = 2;
|
---|
496 | items[n].code[0] = cs[0][i];
|
---|
497 | items[n].code[1] = cs[1][j];
|
---|
498 | n++;
|
---|
499 | }
|
---|
500 | }
|
---|
501 |
|
---|
502 | if ((z2 = onigenc_unicode_unfold2_lookup(to->code)) != 0) {
|
---|
503 | for (i = 0; i < z2->n; i++) {
|
---|
504 | if (z2->code[i] == code) continue;
|
---|
505 |
|
---|
506 | items[n].byte_len = len;
|
---|
507 | items[n].code_len = 1;
|
---|
508 | items[n].code[0] = z2->code[i];
|
---|
509 | n++;
|
---|
510 | }
|
---|
511 | }
|
---|
512 | }
|
---|
513 | else {
|
---|
514 | for (i = 0; i < ncs[0]; i++) {
|
---|
515 | for (j = 0; j < ncs[1]; j++) {
|
---|
516 | for (k = 0; k < ncs[2]; k++) {
|
---|
517 | items[n].byte_len = len;
|
---|
518 | items[n].code_len = 3;
|
---|
519 | items[n].code[0] = cs[0][i];
|
---|
520 | items[n].code[1] = cs[1][j];
|
---|
521 | items[n].code[2] = cs[2][k];
|
---|
522 | n++;
|
---|
523 | }
|
---|
524 | }
|
---|
525 | }
|
---|
526 |
|
---|
527 | if ((z2 = onigenc_unicode_unfold3_lookup(to->code)) != 0) {
|
---|
528 | for (i = 0; i < z2->n; i++) {
|
---|
529 | if (z2->code[i] == code) continue;
|
---|
530 |
|
---|
531 | items[n].byte_len = len;
|
---|
532 | items[n].code_len = 1;
|
---|
533 | items[n].code[0] = z2->code[i];
|
---|
534 | n++;
|
---|
535 | }
|
---|
536 | }
|
---|
537 | }
|
---|
538 |
|
---|
539 | /* multi char folded code is not head of another folded multi char */
|
---|
540 | flag = 0; /* DISABLE_CASE_FOLD_MULTI_CHAR(flag); */
|
---|
541 | }
|
---|
542 | }
|
---|
543 | else {
|
---|
544 | if ((to = onigenc_unicode_unfold1_lookup(code)) != 0) {
|
---|
545 | for (i = 0; i < to->n; i++) {
|
---|
546 | items[n].byte_len = len;
|
---|
547 | items[n].code_len = 1;
|
---|
548 | items[n].code[0] = to->code[i];
|
---|
549 | n++;
|
---|
550 | }
|
---|
551 | }
|
---|
552 | }
|
---|
553 |
|
---|
554 |
|
---|
555 | if ((flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0) {
|
---|
556 | p += len;
|
---|
557 | if (p < end) {
|
---|
558 | int clen;
|
---|
559 |
|
---|
560 | codes[0] = code;
|
---|
561 | code = ONIGENC_MBC_TO_CODE(enc, p, end);
|
---|
562 | if ((to = onigenc_unicode_fold_lookup(code)) != 0
|
---|
563 | && to->n == 1) {
|
---|
564 | codes[1] = to->code[0];
|
---|
565 | }
|
---|
566 | else
|
---|
567 | codes[1] = code;
|
---|
568 |
|
---|
569 | clen = enclen(enc, p);
|
---|
570 | len += clen;
|
---|
571 | if ((z2 = onigenc_unicode_unfold2_lookup(codes)) != 0) {
|
---|
572 | for (i = 0; i < z2->n; i++) {
|
---|
573 | items[n].byte_len = len;
|
---|
574 | items[n].code_len = 1;
|
---|
575 | items[n].code[0] = z2->code[i];
|
---|
576 | n++;
|
---|
577 | }
|
---|
578 | }
|
---|
579 |
|
---|
580 | p += clen;
|
---|
581 | if (p < end) {
|
---|
582 | code = ONIGENC_MBC_TO_CODE(enc, p, end);
|
---|
583 | if ((to = onigenc_unicode_fold_lookup(code)) != 0
|
---|
584 | && to->n == 1) {
|
---|
585 | codes[2] = to->code[0];
|
---|
586 | }
|
---|
587 | else
|
---|
588 | codes[2] = code;
|
---|
589 |
|
---|
590 | clen = enclen(enc, p);
|
---|
591 | len += clen;
|
---|
592 | if ((z2 = onigenc_unicode_unfold3_lookup(codes)) != 0) {
|
---|
593 | for (i = 0; i < z2->n; i++) {
|
---|
594 | items[n].byte_len = len;
|
---|
595 | items[n].code_len = 1;
|
---|
596 | items[n].code[0] = z2->code[i];
|
---|
597 | n++;
|
---|
598 | }
|
---|
599 | }
|
---|
600 | }
|
---|
601 | }
|
---|
602 | }
|
---|
603 |
|
---|
604 | return n;
|
---|
605 | }
|
---|