]>
Commit | Line | Data |
---|---|---|
c2f9875c GKB |
1 | /* |
2 | * Copyright (c) 2014 SGI. | |
3 | * All rights reserved. | |
4 | * | |
5 | * This program is free software; you can redistribute it and/or | |
6 | * modify it under the terms of the GNU General Public License as | |
7 | * published by the Free Software Foundation. | |
8 | * | |
9 | * This program is distributed in the hope that it would be useful, | |
10 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
11 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
12 | * GNU General Public License for more details. | |
13 | * | |
14 | */ | |
15 | ||
16 | /* | |
17 | * This code is adapted from the Linux Kernel. We have a | |
18 | * userspace version here such that the hashes will match that | |
19 | * implementation. | |
20 | */ | |
21 | ||
22 | #include "utf8n.h" | |
23 | ||
24 | struct utf8data { | |
25 | unsigned int maxage; | |
26 | unsigned int offset; | |
27 | }; | |
28 | ||
29 | #define __INCLUDED_FROM_UTF8NORM_C__ | |
30 | #include "utf8data.h" | |
31 | #undef __INCLUDED_FROM_UTF8NORM_C__ | |
32 | ||
33 | #define ARRAY_SIZE(array) \ | |
34 | (sizeof(array) / sizeof(array[0])) | |
35 | ||
36 | int utf8version_is_supported(uint8_t maj, uint8_t min, uint8_t rev) | |
37 | { | |
38 | int i = ARRAY_SIZE(utf8agetab) - 1; | |
39 | unsigned int sb_utf8version = UNICODE_AGE(maj, min, rev); | |
40 | ||
41 | while (i >= 0 && utf8agetab[i] != 0) { | |
42 | if (sb_utf8version == utf8agetab[i]) | |
43 | return 1; | |
44 | i--; | |
45 | } | |
46 | return 0; | |
47 | } | |
48 | ||
6790d66c | 49 | int utf8version_latest(void) |
c2f9875c GKB |
50 | { |
51 | return utf8vers; | |
52 | } | |
53 | ||
54 | /* | |
55 | * UTF-8 valid ranges. | |
56 | * | |
57 | * The UTF-8 encoding spreads the bits of a 32bit word over several | |
58 | * bytes. This table gives the ranges that can be held and how they'd | |
59 | * be represented. | |
60 | * | |
61 | * 0x00000000 0x0000007F: 0xxxxxxx | |
62 | * 0x00000000 0x000007FF: 110xxxxx 10xxxxxx | |
63 | * 0x00000000 0x0000FFFF: 1110xxxx 10xxxxxx 10xxxxxx | |
64 | * 0x00000000 0x001FFFFF: 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx | |
65 | * 0x00000000 0x03FFFFFF: 111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx | |
66 | * 0x00000000 0x7FFFFFFF: 1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx | |
67 | * | |
68 | * There is an additional requirement on UTF-8, in that only the | |
69 | * shortest representation of a 32bit value is to be used. A decoder | |
70 | * must not decode sequences that do not satisfy this requirement. | |
71 | * Thus the allowed ranges have a lower bound. | |
72 | * | |
73 | * 0x00000000 0x0000007F: 0xxxxxxx | |
74 | * 0x00000080 0x000007FF: 110xxxxx 10xxxxxx | |
75 | * 0x00000800 0x0000FFFF: 1110xxxx 10xxxxxx 10xxxxxx | |
76 | * 0x00010000 0x001FFFFF: 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx | |
77 | * 0x00200000 0x03FFFFFF: 111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx | |
78 | * 0x04000000 0x7FFFFFFF: 1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx | |
79 | * | |
80 | * Actual unicode characters are limited to the range 0x0 - 0x10FFFF, | |
81 | * 17 planes of 65536 values. This limits the sequences actually seen | |
82 | * even more, to just the following. | |
83 | * | |
84 | * 0 - 0x7F: 0 - 0x7F | |
85 | * 0x80 - 0x7FF: 0xC2 0x80 - 0xDF 0xBF | |
86 | * 0x800 - 0xFFFF: 0xE0 0xA0 0x80 - 0xEF 0xBF 0xBF | |
87 | * 0x10000 - 0x10FFFF: 0xF0 0x90 0x80 0x80 - 0xF4 0x8F 0xBF 0xBF | |
88 | * | |
89 | * Within those ranges the surrogates 0xD800 - 0xDFFF are not allowed. | |
90 | * | |
91 | * Note that the longest sequence seen with valid usage is 4 bytes, | |
92 | * the same a single UTF-32 character. This makes the UTF-8 | |
93 | * representation of Unicode strictly smaller than UTF-32. | |
94 | * | |
95 | * The shortest sequence requirement was introduced by: | |
96 | * Corrigendum #1: UTF-8 Shortest Form | |
97 | * It can be found here: | |
98 | * http://www.unicode.org/versions/corrigendum1.html | |
99 | * | |
100 | */ | |
101 | ||
102 | /* | |
103 | * Return the number of bytes used by the current UTF-8 sequence. | |
104 | * Assumes the input points to the first byte of a valid UTF-8 | |
105 | * sequence. | |
106 | */ | |
107 | static inline int utf8clen(const char *s) | |
108 | { | |
109 | unsigned char c = *s; | |
110 | ||
111 | return 1 + (c >= 0xC0) + (c >= 0xE0) + (c >= 0xF0); | |
112 | } | |
113 | ||
114 | /* | |
115 | * Decode a 3-byte UTF-8 sequence. | |
116 | */ | |
117 | static unsigned int | |
118 | utf8decode3(const char *str) | |
119 | { | |
120 | unsigned int uc; | |
121 | ||
122 | uc = *str++ & 0x0F; | |
123 | uc <<= 6; | |
124 | uc |= *str++ & 0x3F; | |
125 | uc <<= 6; | |
126 | uc |= *str++ & 0x3F; | |
127 | ||
128 | return uc; | |
129 | } | |
130 | ||
131 | /* | |
132 | * Encode a 3-byte UTF-8 sequence. | |
133 | */ | |
134 | static int | |
135 | utf8encode3(char *str, unsigned int val) | |
136 | { | |
137 | str[2] = (val & 0x3F) | 0x80; | |
138 | val >>= 6; | |
139 | str[1] = (val & 0x3F) | 0x80; | |
140 | val >>= 6; | |
141 | str[0] = val | 0xE0; | |
142 | ||
143 | return 3; | |
144 | } | |
145 | ||
146 | /* | |
147 | * utf8trie_t | |
148 | * | |
149 | * A compact binary tree, used to decode UTF-8 characters. | |
150 | * | |
151 | * Internal nodes are one byte for the node itself, and up to three | |
152 | * bytes for an offset into the tree. The first byte contains the | |
153 | * following information: | |
154 | * NEXTBYTE - flag - advance to next byte if set | |
155 | * BITNUM - 3 bit field - the bit number to tested | |
156 | * OFFLEN - 2 bit field - number of bytes in the offset | |
157 | * if offlen == 0 (non-branching node) | |
158 | * RIGHTPATH - 1 bit field - set if the following node is for the | |
159 | * right-hand path (tested bit is set) | |
160 | * TRIENODE - 1 bit field - set if the following node is an internal | |
161 | * node, otherwise it is a leaf node | |
162 | * if offlen != 0 (branching node) | |
163 | * LEFTNODE - 1 bit field - set if the left-hand node is internal | |
164 | * RIGHTNODE - 1 bit field - set if the right-hand node is internal | |
165 | * | |
166 | * Due to the way utf8 works, there cannot be branching nodes with | |
167 | * NEXTBYTE set, and moreover those nodes always have a righthand | |
168 | * descendant. | |
169 | */ | |
170 | typedef const unsigned char utf8trie_t; | |
171 | #define BITNUM 0x07 | |
172 | #define NEXTBYTE 0x08 | |
173 | #define OFFLEN 0x30 | |
174 | #define OFFLEN_SHIFT 4 | |
175 | #define RIGHTPATH 0x40 | |
176 | #define TRIENODE 0x80 | |
177 | #define RIGHTNODE 0x40 | |
178 | #define LEFTNODE 0x80 | |
179 | ||
180 | /* | |
181 | * utf8leaf_t | |
182 | * | |
183 | * The leaves of the trie are embedded in the trie, and so the same | |
184 | * underlying datatype: unsigned char. | |
185 | * | |
186 | * leaf[0]: The unicode version, stored as a generation number that is | |
187 | * an index into utf8agetab[]. With this we can filter code | |
188 | * points based on the unicode version in which they were | |
189 | * defined. The CCC of a non-defined code point is 0. | |
190 | * leaf[1]: Canonical Combining Class. During normalization, we need | |
191 | * to do a stable sort into ascending order of all characters | |
192 | * with a non-zero CCC that occur between two characters with | |
193 | * a CCC of 0, or at the begin or end of a string. | |
194 | * The unicode standard guarantees that all CCC values are | |
195 | * between 0 and 254 inclusive, which leaves 255 available as | |
196 | * a special value. | |
197 | * Code points with CCC 0 are known as stoppers. | |
198 | * leaf[2]: Decomposition. If leaf[1] == 255, then leaf[2] is the | |
199 | * start of a NUL-terminated string that is the decomposition | |
200 | * of the character. | |
201 | * The CCC of a decomposable character is the same as the CCC | |
202 | * of the first character of its decomposition. | |
203 | * Some characters decompose as the empty string: these are | |
204 | * characters with the Default_Ignorable_Code_Point property. | |
205 | * These do affect normalization, as they all have CCC 0. | |
206 | * | |
207 | * The decompositions in the trie have been fully expanded, with the | |
208 | * exception of Hangul syllables, which are decomposed algorithmically. | |
209 | * | |
210 | * Casefolding, if applicable, is also done using decompositions. | |
211 | * | |
212 | * The trie is constructed in such a way that leaves exist for all | |
213 | * UTF-8 sequences that match the criteria from the "UTF-8 valid | |
214 | * ranges" comment above, and only for those sequences. Therefore a | |
215 | * lookup in the trie can be used to validate the UTF-8 input. | |
216 | */ | |
217 | typedef const unsigned char utf8leaf_t; | |
218 | ||
219 | #define LEAF_GEN(LEAF) ((LEAF)[0]) | |
220 | #define LEAF_CCC(LEAF) ((LEAF)[1]) | |
221 | #define LEAF_STR(LEAF) ((const char *)((LEAF) + 2)) | |
222 | ||
223 | #define MINCCC (0) | |
224 | #define MAXCCC (254) | |
225 | #define STOPPER (0) | |
226 | #define DECOMPOSE (255) | |
227 | ||
228 | /* Marker for hangul syllable decomposition. */ | |
229 | #define HANGUL ((char)(255)) | |
230 | /* Size of the synthesized leaf used for Hangul syllable decomposition. */ | |
231 | #define UTF8HANGULLEAF (12) | |
232 | ||
233 | /* | |
234 | * Hangul decomposition (algorithm from Section 3.12 of Unicode 6.3.0) | |
235 | * | |
236 | * AC00;<Hangul Syllable, First>;Lo;0;L;;;;;N;;;;; | |
237 | * D7A3;<Hangul Syllable, Last>;Lo;0;L;;;;;N;;;;; | |
238 | * | |
239 | * SBase = 0xAC00 | |
240 | * LBase = 0x1100 | |
241 | * VBase = 0x1161 | |
242 | * TBase = 0x11A7 | |
243 | * LCount = 19 | |
244 | * VCount = 21 | |
245 | * TCount = 28 | |
246 | * NCount = 588 (VCount * TCount) | |
247 | * SCount = 11172 (LCount * NCount) | |
248 | * | |
249 | * Decomposition: | |
250 | * SIndex = s - SBase | |
251 | * | |
252 | * LV (Canonical/Full) | |
253 | * LIndex = SIndex / NCount | |
254 | * VIndex = (Sindex % NCount) / TCount | |
255 | * LPart = LBase + LIndex | |
256 | * VPart = VBase + VIndex | |
257 | * | |
258 | * LVT (Canonical) | |
259 | * LVIndex = (SIndex / TCount) * TCount | |
260 | * TIndex = (Sindex % TCount) | |
261 | * LVPart = SBase + LVIndex | |
262 | * TPart = TBase + TIndex | |
263 | * | |
264 | * LVT (Full) | |
265 | * LIndex = SIndex / NCount | |
266 | * VIndex = (Sindex % NCount) / TCount | |
267 | * TIndex = (Sindex % TCount) | |
268 | * LPart = LBase + LIndex | |
269 | * VPart = VBase + VIndex | |
270 | * if (TIndex == 0) { | |
271 | * d = <LPart, VPart> | |
272 | * } else { | |
273 | * TPart = TBase + TIndex | |
274 | * d = <LPart, TPart, VPart> | |
275 | * } | |
276 | */ | |
277 | ||
278 | /* Constants */ | |
279 | #define SB (0xAC00) | |
280 | #define LB (0x1100) | |
281 | #define VB (0x1161) | |
282 | #define TB (0x11A7) | |
283 | #define LC (19) | |
284 | #define VC (21) | |
285 | #define TC (28) | |
286 | #define NC (VC * TC) | |
287 | #define SC (LC * NC) | |
288 | ||
289 | /* Algorithmic decomposition of hangul syllable. */ | |
290 | static utf8leaf_t * | |
291 | utf8hangul(const char *str, unsigned char *hangul) | |
292 | { | |
293 | unsigned int si; | |
294 | unsigned int li; | |
295 | unsigned int vi; | |
296 | unsigned int ti; | |
297 | unsigned char *h; | |
298 | ||
299 | /* Calculate the SI, LI, VI, and TI values. */ | |
300 | si = utf8decode3(str) - SB; | |
301 | li = si / NC; | |
302 | vi = (si % NC) / TC; | |
303 | ti = si % TC; | |
304 | ||
305 | /* Fill in base of leaf. */ | |
306 | h = hangul; | |
307 | LEAF_GEN(h) = 2; | |
308 | LEAF_CCC(h) = DECOMPOSE; | |
309 | h += 2; | |
310 | ||
311 | /* Add LPart, a 3-byte UTF-8 sequence. */ | |
312 | h += utf8encode3((char *)h, li + LB); | |
313 | ||
314 | /* Add VPart, a 3-byte UTF-8 sequence. */ | |
315 | h += utf8encode3((char *)h, vi + VB); | |
316 | ||
317 | /* Add TPart if required, also a 3-byte UTF-8 sequence. */ | |
318 | if (ti) | |
319 | h += utf8encode3((char *)h, ti + TB); | |
320 | ||
321 | /* Terminate string. */ | |
322 | h[0] = '\0'; | |
323 | ||
324 | return hangul; | |
325 | } | |
326 | ||
327 | /* | |
328 | * Use trie to scan s, touching at most len bytes. | |
329 | * Returns the leaf if one exists, NULL otherwise. | |
330 | * | |
331 | * A non-NULL return guarantees that the UTF-8 sequence starting at s | |
332 | * is well-formed and corresponds to a known unicode code point. The | |
333 | * shorthand for this will be "is valid UTF-8 unicode". | |
334 | */ | |
335 | static utf8leaf_t *utf8nlookup(const struct utf8data *data, | |
336 | unsigned char *hangul, const char *s, size_t len) | |
337 | { | |
dea1e527 | 338 | utf8trie_t *trie; |
c2f9875c GKB |
339 | int offlen; |
340 | int offset; | |
341 | int mask; | |
342 | int node; | |
343 | ||
344 | if (!data) | |
345 | return NULL; | |
346 | if (len == 0) | |
347 | return NULL; | |
dea1e527 GKB |
348 | |
349 | trie = utf8data + data->offset; | |
c2f9875c GKB |
350 | node = 1; |
351 | while (node) { | |
352 | offlen = (*trie & OFFLEN) >> OFFLEN_SHIFT; | |
353 | if (*trie & NEXTBYTE) { | |
354 | if (--len == 0) | |
355 | return NULL; | |
356 | s++; | |
357 | } | |
358 | mask = 1 << (*trie & BITNUM); | |
359 | if (*s & mask) { | |
360 | /* Right leg */ | |
361 | if (offlen) { | |
362 | /* Right node at offset of trie */ | |
363 | node = (*trie & RIGHTNODE); | |
364 | offset = trie[offlen]; | |
365 | while (--offlen) { | |
366 | offset <<= 8; | |
367 | offset |= trie[offlen]; | |
368 | } | |
369 | trie += offset; | |
370 | } else if (*trie & RIGHTPATH) { | |
371 | /* Right node after this node */ | |
372 | node = (*trie & TRIENODE); | |
373 | trie++; | |
374 | } else { | |
375 | /* No right node. */ | |
376 | return NULL; | |
377 | } | |
378 | } else { | |
379 | /* Left leg */ | |
380 | if (offlen) { | |
381 | /* Left node after this node. */ | |
382 | node = (*trie & LEFTNODE); | |
383 | trie += offlen + 1; | |
384 | } else if (*trie & RIGHTPATH) { | |
385 | /* No left node. */ | |
386 | return NULL; | |
387 | } else { | |
388 | /* Left node after this node */ | |
389 | node = (*trie & TRIENODE); | |
390 | trie++; | |
391 | } | |
392 | } | |
393 | } | |
394 | /* | |
395 | * Hangul decomposition is done algorithmically. These are the | |
396 | * codepoints >= 0xAC00 and <= 0xD7A3. Their UTF-8 encoding is | |
397 | * always 3 bytes long, so s has been advanced twice, and the | |
398 | * start of the sequence is at s-2. | |
399 | */ | |
400 | if (LEAF_CCC(trie) == DECOMPOSE && LEAF_STR(trie)[0] == HANGUL) | |
401 | trie = utf8hangul(s - 2, hangul); | |
402 | return trie; | |
403 | } | |
404 | ||
405 | /* | |
406 | * Use trie to scan s. | |
407 | * Returns the leaf if one exists, NULL otherwise. | |
408 | * | |
409 | * Forwards to utf8nlookup(). | |
410 | */ | |
411 | static utf8leaf_t *utf8lookup(const struct utf8data *data, | |
412 | unsigned char *hangul, const char *s) | |
413 | { | |
414 | return utf8nlookup(data, hangul, s, (size_t)-1); | |
415 | } | |
416 | ||
417 | /* | |
418 | * Maximum age of any character in s. | |
419 | * Return -1 if s is not valid UTF-8 unicode. | |
420 | * Return 0 if only non-assigned code points are used. | |
421 | */ | |
422 | int utf8agemax(const struct utf8data *data, const char *s) | |
423 | { | |
424 | utf8leaf_t *leaf; | |
425 | int age = 0; | |
426 | int leaf_age; | |
427 | unsigned char hangul[UTF8HANGULLEAF]; | |
428 | ||
429 | if (!data) | |
430 | return -1; | |
431 | ||
432 | while (*s) { | |
433 | leaf = utf8lookup(data, hangul, s); | |
434 | if (!leaf) | |
435 | return -1; | |
436 | ||
437 | leaf_age = utf8agetab[LEAF_GEN(leaf)]; | |
438 | if (leaf_age <= data->maxage && leaf_age > age) | |
439 | age = leaf_age; | |
440 | s += utf8clen(s); | |
441 | } | |
442 | return age; | |
443 | } | |
444 | ||
445 | /* | |
446 | * Minimum age of any character in s. | |
447 | * Return -1 if s is not valid UTF-8 unicode. | |
448 | * Return 0 if non-assigned code points are used. | |
449 | */ | |
450 | int utf8agemin(const struct utf8data *data, const char *s) | |
451 | { | |
452 | utf8leaf_t *leaf; | |
453 | int age; | |
454 | int leaf_age; | |
455 | unsigned char hangul[UTF8HANGULLEAF]; | |
456 | ||
457 | if (!data) | |
458 | return -1; | |
459 | age = data->maxage; | |
460 | while (*s) { | |
461 | leaf = utf8lookup(data, hangul, s); | |
462 | if (!leaf) | |
463 | return -1; | |
464 | leaf_age = utf8agetab[LEAF_GEN(leaf)]; | |
465 | if (leaf_age <= data->maxage && leaf_age < age) | |
466 | age = leaf_age; | |
467 | s += utf8clen(s); | |
468 | } | |
469 | return age; | |
470 | } | |
471 | ||
472 | /* | |
473 | * Maximum age of any character in s, touch at most len bytes. | |
474 | * Return -1 if s is not valid UTF-8 unicode. | |
475 | */ | |
476 | int utf8nagemax(const struct utf8data *data, const char *s, size_t len) | |
477 | { | |
478 | utf8leaf_t *leaf; | |
479 | int age = 0; | |
480 | int leaf_age; | |
481 | unsigned char hangul[UTF8HANGULLEAF]; | |
482 | ||
483 | if (!data) | |
484 | return -1; | |
485 | ||
486 | while (len && *s) { | |
487 | leaf = utf8nlookup(data, hangul, s, len); | |
488 | if (!leaf) | |
489 | return -1; | |
490 | leaf_age = utf8agetab[LEAF_GEN(leaf)]; | |
491 | if (leaf_age <= data->maxage && leaf_age > age) | |
492 | age = leaf_age; | |
493 | len -= utf8clen(s); | |
494 | s += utf8clen(s); | |
495 | } | |
496 | return age; | |
497 | } | |
498 | ||
499 | /* | |
500 | * Maximum age of any character in s, touch at most len bytes. | |
501 | * Return -1 if s is not valid UTF-8 unicode. | |
502 | */ | |
503 | int utf8nagemin(const struct utf8data *data, const char *s, size_t len) | |
504 | { | |
505 | utf8leaf_t *leaf; | |
506 | int leaf_age; | |
507 | int age; | |
508 | unsigned char hangul[UTF8HANGULLEAF]; | |
509 | ||
510 | if (!data) | |
511 | return -1; | |
512 | age = data->maxage; | |
513 | while (len && *s) { | |
514 | leaf = utf8nlookup(data, hangul, s, len); | |
515 | if (!leaf) | |
516 | return -1; | |
517 | leaf_age = utf8agetab[LEAF_GEN(leaf)]; | |
518 | if (leaf_age <= data->maxage && leaf_age < age) | |
519 | age = leaf_age; | |
520 | len -= utf8clen(s); | |
521 | s += utf8clen(s); | |
522 | } | |
523 | return age; | |
524 | } | |
525 | ||
526 | /* | |
527 | * Length of the normalization of s. | |
528 | * Return -1 if s is not valid UTF-8 unicode. | |
529 | * | |
530 | * A string of Default_Ignorable_Code_Point has length 0. | |
531 | */ | |
532 | ssize_t utf8len(const struct utf8data *data, const char *s) | |
533 | { | |
534 | utf8leaf_t *leaf; | |
535 | size_t ret = 0; | |
536 | unsigned char hangul[UTF8HANGULLEAF]; | |
537 | ||
538 | if (!data) | |
539 | return -1; | |
540 | while (*s) { | |
541 | leaf = utf8lookup(data, hangul, s); | |
542 | if (!leaf) | |
543 | return -1; | |
544 | if (utf8agetab[LEAF_GEN(leaf)] > data->maxage) | |
545 | ret += utf8clen(s); | |
546 | else if (LEAF_CCC(leaf) == DECOMPOSE) | |
547 | ret += strlen(LEAF_STR(leaf)); | |
548 | else | |
549 | ret += utf8clen(s); | |
550 | s += utf8clen(s); | |
551 | } | |
552 | return ret; | |
553 | } | |
554 | ||
555 | /* | |
556 | * Length of the normalization of s, touch at most len bytes. | |
557 | * Return -1 if s is not valid UTF-8 unicode. | |
558 | */ | |
559 | ssize_t utf8nlen(const struct utf8data *data, const char *s, size_t len) | |
560 | { | |
561 | utf8leaf_t *leaf; | |
562 | size_t ret = 0; | |
563 | unsigned char hangul[UTF8HANGULLEAF]; | |
564 | ||
565 | if (!data) | |
566 | return -1; | |
567 | while (len && *s) { | |
568 | leaf = utf8nlookup(data, hangul, s, len); | |
569 | if (!leaf) | |
570 | return -1; | |
571 | if (utf8agetab[LEAF_GEN(leaf)] > data->maxage) | |
572 | ret += utf8clen(s); | |
573 | else if (LEAF_CCC(leaf) == DECOMPOSE) | |
574 | ret += strlen(LEAF_STR(leaf)); | |
575 | else | |
576 | ret += utf8clen(s); | |
577 | len -= utf8clen(s); | |
578 | s += utf8clen(s); | |
579 | } | |
580 | return ret; | |
581 | } | |
582 | ||
583 | /* | |
584 | * Set up an utf8cursor for use by utf8byte(). | |
585 | * | |
586 | * u8c : pointer to cursor. | |
587 | * data : const struct utf8data to use for normalization. | |
588 | * s : string. | |
589 | * len : length of s. | |
590 | * | |
591 | * Returns -1 on error, 0 on success. | |
592 | */ | |
593 | int utf8ncursor(struct utf8cursor *u8c, const struct utf8data *data, | |
594 | const char *s, size_t len) | |
595 | { | |
596 | if (!data) | |
597 | return -1; | |
598 | if (!s) | |
599 | return -1; | |
600 | u8c->data = data; | |
601 | u8c->s = s; | |
602 | u8c->p = NULL; | |
603 | u8c->ss = NULL; | |
604 | u8c->sp = NULL; | |
605 | u8c->len = len; | |
606 | u8c->slen = 0; | |
607 | u8c->ccc = STOPPER; | |
608 | u8c->nccc = STOPPER; | |
609 | /* Check we didn't clobber the maximum length. */ | |
610 | if (u8c->len != len) | |
611 | return -1; | |
612 | /* The first byte of s may not be an utf8 continuation. */ | |
613 | if (len > 0 && (*s & 0xC0) == 0x80) | |
614 | return -1; | |
615 | return 0; | |
616 | } | |
617 | ||
618 | /* | |
619 | * Set up an utf8cursor for use by utf8byte(). | |
620 | * | |
621 | * u8c : pointer to cursor. | |
622 | * data : const struct utf8data to use for normalization. | |
623 | * s : NUL-terminated string. | |
624 | * | |
625 | * Returns -1 on error, 0 on success. | |
626 | */ | |
627 | int utf8cursor(struct utf8cursor *u8c, const struct utf8data *data, | |
628 | const char *s) | |
629 | { | |
630 | return utf8ncursor(u8c, data, s, (unsigned int)-1); | |
631 | } | |
632 | ||
633 | /* | |
634 | * Get one byte from the normalized form of the string described by u8c. | |
635 | * | |
636 | * Returns the byte cast to an unsigned char on succes, and -1 on failure. | |
637 | * | |
638 | * The cursor keeps track of the location in the string in u8c->s. | |
639 | * When a character is decomposed, the current location is stored in | |
640 | * u8c->p, and u8c->s is set to the start of the decomposition. Note | |
641 | * that bytes from a decomposition do not count against u8c->len. | |
642 | * | |
643 | * Characters are emitted if they match the current CCC in u8c->ccc. | |
644 | * Hitting end-of-string while u8c->ccc == STOPPER means we're done, | |
645 | * and the function returns 0 in that case. | |
646 | * | |
647 | * Sorting by CCC is done by repeatedly scanning the string. The | |
648 | * values of u8c->s and u8c->p are stored in u8c->ss and u8c->sp at | |
649 | * the start of the scan. The first pass finds the lowest CCC to be | |
650 | * emitted and stores it in u8c->nccc, the second pass emits the | |
651 | * characters with this CCC and finds the next lowest CCC. This limits | |
652 | * the number of passes to 1 + the number of different CCCs in the | |
653 | * sequence being scanned. | |
654 | * | |
655 | * Therefore: | |
656 | * u8c->p != NULL -> a decomposition is being scanned. | |
657 | * u8c->ss != NULL -> this is a repeating scan. | |
658 | * u8c->ccc == -1 -> this is the first scan of a repeating scan. | |
659 | */ | |
660 | int utf8byte(struct utf8cursor *u8c) | |
661 | { | |
662 | utf8leaf_t *leaf; | |
663 | int ccc; | |
664 | ||
665 | for (;;) { | |
666 | /* Check for the end of a decomposed character. */ | |
667 | if (u8c->p && *u8c->s == '\0') { | |
668 | u8c->s = u8c->p; | |
669 | u8c->p = NULL; | |
670 | } | |
671 | ||
672 | /* Check for end-of-string. */ | |
673 | if (!u8c->p && (u8c->len == 0 || *u8c->s == '\0')) { | |
674 | /* There is no next byte. */ | |
675 | if (u8c->ccc == STOPPER) | |
676 | return 0; | |
677 | /* End-of-string during a scan counts as a stopper. */ | |
678 | ccc = STOPPER; | |
679 | goto ccc_mismatch; | |
680 | } else if ((*u8c->s & 0xC0) == 0x80) { | |
681 | /* This is a continuation of the current character. */ | |
682 | if (!u8c->p) | |
683 | u8c->len--; | |
684 | return (unsigned char)*u8c->s++; | |
685 | } | |
686 | ||
687 | /* Look up the data for the current character. */ | |
688 | if (u8c->p) { | |
689 | leaf = utf8lookup(u8c->data, u8c->hangul, u8c->s); | |
690 | } else { | |
691 | leaf = utf8nlookup(u8c->data, u8c->hangul, | |
692 | u8c->s, u8c->len); | |
693 | } | |
694 | ||
695 | /* No leaf found implies that the input is a binary blob. */ | |
696 | if (!leaf) | |
697 | return -1; | |
698 | ||
699 | ccc = LEAF_CCC(leaf); | |
700 | /* Characters that are too new have CCC 0. */ | |
701 | if (utf8agetab[LEAF_GEN(leaf)] > u8c->data->maxage) { | |
702 | ccc = STOPPER; | |
703 | } else if (ccc == DECOMPOSE) { | |
704 | u8c->len -= utf8clen(u8c->s); | |
705 | u8c->p = u8c->s + utf8clen(u8c->s); | |
706 | u8c->s = LEAF_STR(leaf); | |
707 | /* Empty decomposition implies CCC 0. */ | |
708 | if (*u8c->s == '\0') { | |
709 | if (u8c->ccc == STOPPER) | |
710 | continue; | |
711 | ccc = STOPPER; | |
712 | goto ccc_mismatch; | |
713 | } | |
714 | ||
715 | leaf = utf8lookup(u8c->data, u8c->hangul, u8c->s); | |
716 | ccc = LEAF_CCC(leaf); | |
717 | } | |
718 | ||
719 | /* | |
720 | * If this is not a stopper, then see if it updates | |
721 | * the next canonical class to be emitted. | |
722 | */ | |
723 | if (ccc != STOPPER && u8c->ccc < ccc && ccc < u8c->nccc) | |
724 | u8c->nccc = ccc; | |
725 | ||
726 | /* | |
727 | * Return the current byte if this is the current | |
728 | * combining class. | |
729 | */ | |
730 | if (ccc == u8c->ccc) { | |
731 | if (!u8c->p) | |
732 | u8c->len--; | |
733 | return (unsigned char)*u8c->s++; | |
734 | } | |
735 | ||
736 | /* Current combining class mismatch. */ | |
737 | ccc_mismatch: | |
738 | if (u8c->nccc == STOPPER) { | |
739 | /* | |
740 | * Scan forward for the first canonical class | |
741 | * to be emitted. Save the position from | |
742 | * which to restart. | |
743 | */ | |
744 | u8c->ccc = MINCCC - 1; | |
745 | u8c->nccc = ccc; | |
746 | u8c->sp = u8c->p; | |
747 | u8c->ss = u8c->s; | |
748 | u8c->slen = u8c->len; | |
749 | if (!u8c->p) | |
750 | u8c->len -= utf8clen(u8c->s); | |
751 | u8c->s += utf8clen(u8c->s); | |
752 | } else if (ccc != STOPPER) { | |
753 | /* Not a stopper, and not the ccc we're emitting. */ | |
754 | if (!u8c->p) | |
755 | u8c->len -= utf8clen(u8c->s); | |
756 | u8c->s += utf8clen(u8c->s); | |
757 | } else if (u8c->nccc != MAXCCC + 1) { | |
758 | /* At a stopper, restart for next ccc. */ | |
759 | u8c->ccc = u8c->nccc; | |
760 | u8c->nccc = MAXCCC + 1; | |
761 | u8c->s = u8c->ss; | |
762 | u8c->p = u8c->sp; | |
763 | u8c->len = u8c->slen; | |
764 | } else { | |
765 | /* All done, proceed from here. */ | |
766 | u8c->ccc = STOPPER; | |
767 | u8c->nccc = STOPPER; | |
768 | u8c->sp = NULL; | |
769 | u8c->ss = NULL; | |
770 | u8c->slen = 0; | |
771 | } | |
772 | } | |
773 | } | |
774 | ||
3e1c513a | 775 | const struct utf8data *utf8nfdi(unsigned int maxage) |
c2f9875c | 776 | { |
3e1c513a | 777 | int i = ARRAY_SIZE(utf8nfdidata) - 1; |
c2f9875c | 778 | |
3e1c513a | 779 | while (maxage < utf8nfdidata[i].maxage) |
c2f9875c | 780 | i--; |
3e1c513a | 781 | if (maxage > utf8nfdidata[i].maxage) |
c2f9875c | 782 | return NULL; |
3e1c513a | 783 | return &utf8nfdidata[i]; |
c2f9875c GKB |
784 | } |
785 | ||
3e1c513a | 786 | const struct utf8data *utf8nfdicf(unsigned int maxage) |
c2f9875c | 787 | { |
3e1c513a | 788 | int i = ARRAY_SIZE(utf8nfdicfdata) - 1; |
c2f9875c | 789 | |
3e1c513a | 790 | while (maxage < utf8nfdicfdata[i].maxage) |
c2f9875c | 791 | i--; |
3e1c513a | 792 | if (maxage > utf8nfdicfdata[i].maxage) |
c2f9875c | 793 | return NULL; |
3e1c513a | 794 | return &utf8nfdicfdata[i]; |
c2f9875c | 795 | } |