?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Linux: Fix detection of register_sysctl_sz
[zfs.git] / module / unicode / u8_textprep.c
blob281396678ec22ca048f79c01d1b07744c6276fd8
1 /*
2 * CDDL HEADER START
3 *
4 * The contents of this file are subject to the terms of the
5 * Common Development and Distribution License (the "License").
6 * You may not use this file except in compliance with the License.
7 *
8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9 * or https://opensource.org/licenses/CDDL-1.0.
10 * See the License for the specific language governing permissions
11 * and limitations under the License.
12 *
13 * When distributing Covered Code, include this CDDL HEADER in each
14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15 * If applicable, add the following below this CDDL HEADER, with the
16 * fields enclosed by brackets "[]" replaced with your own identifying
17 * information: Portions Copyright [yyyy] [name of copyright owner]
18 *
19 * CDDL HEADER END
20 */
21 /*
22 * Copyright 2008 Sun Microsystems, Inc. All rights reserved.
23 * Use is subject to license terms.
24 */
25
26 /*
27 * Copyright 2022 MNX Cloud, Inc.
28 */
29
30
31
32 /*
33 * UTF-8 text preparation functions (PSARC/2007/149, PSARC/2007/458).
34 *
35 * Man pages: u8_textprep_open(9F), u8_textprep_buf(9F), u8_textprep_close(9F),
36 * u8_textprep_str(9F), u8_strcmp(9F), and u8_validate(9F). See also
37 * the section 3C man pages.
38 * Interface stability: Committed.
39 */
40
41 #include <sys/types.h>
42 #include <sys/string.h>
43 #include <sys/param.h>
44 #include <sys/sysmacros.h>
45 #include <sys/debug.h>
46 #include <sys/kmem.h>
47 #include <sys/sunddi.h>
48 #include <sys/u8_textprep.h>
49 #include <sys/byteorder.h>
50 #include <sys/errno.h>
51 #include <sys/u8_textprep_data.h>
52 #include <sys/mod.h>
53
54 /* The maximum possible number of bytes in a UTF-8 character. */
55 #define U8_MB_CUR_MAX (4)
56
57 /*
58 * The maximum number of bytes needed for a UTF-8 character to cover
59 * U+0000 - U+FFFF, i.e., the coding space of now deprecated UCS-2.
60 */
61 #define U8_MAX_BYTES_UCS2 (3)
62
63 /* The maximum possible number of bytes in a Stream-Safe Text. */
64 #define U8_STREAM_SAFE_TEXT_MAX (128)
65
66 /*
67 * The maximum number of characters in a combining/conjoining sequence and
68 * the actual upperbound limit of a combining/conjoining sequence.
69 */
70 #define U8_MAX_CHARS_A_SEQ (32)
71 #define U8_UPPER_LIMIT_IN_A_SEQ (31)
72
73 /* The combining class value for Starter. */
74 #define U8_COMBINING_CLASS_STARTER (0)
75
76 /*
77 * Some Hangul related macros at below.
78 *
79 * The first and the last of Hangul syllables, Hangul Jamo Leading consonants,
80 * Vowels, and optional Trailing consonants in Unicode scalar values.
81 *
82 * Please be noted that the U8_HANGUL_JAMO_T_FIRST is 0x11A7 at below not
83 * the actual U+11A8. This is due to that the trailing consonant is optional
84 * and thus we are doing a pre-calculation of subtracting one.
85 *
86 * Each of 19 modern leading consonants has total 588 possible syllables since
87 * Hangul has 21 modern vowels and 27 modern trailing consonants plus 1 for
88 * no trailing consonant case, i.e., 21 x 28 = 588.
89 *
90 * We also have bunch of Hangul related macros at below. Please bear in mind
91 * that the U8_HANGUL_JAMO_1ST_BYTE can be used to check whether it is
92 * a Hangul Jamo or not but the value does not guarantee that it is a Hangul
93 * Jamo; it just guarantee that it will be most likely.
94 */
95 #define U8_HANGUL_SYL_FIRST (0xAC00U)
96 #define U8_HANGUL_SYL_LAST (0xD7A3U)
97
98 #define U8_HANGUL_JAMO_L_FIRST (0x1100U)
99 #define U8_HANGUL_JAMO_L_LAST (0x1112U)
100 #define U8_HANGUL_JAMO_V_FIRST (0x1161U)
101 #define U8_HANGUL_JAMO_V_LAST (0x1175U)
102 #define U8_HANGUL_JAMO_T_FIRST (0x11A7U)
103 #define U8_HANGUL_JAMO_T_LAST (0x11C2U)
104
105 #define U8_HANGUL_V_COUNT (21)
106 #define U8_HANGUL_VT_COUNT (588)
107 #define U8_HANGUL_T_COUNT (28)
108
109 #define U8_HANGUL_JAMO_1ST_BYTE (0xE1U)
110
111 #define U8_SAVE_HANGUL_AS_UTF8(s, i, j, k, b) \
112 (s)[(i)] = (uchar_t)(0xE0U | ((uint32_t)(b) & 0xF000U) >> 12); \
113 (s)[(j)] = (uchar_t)(0x80U | ((uint32_t)(b) & 0x0FC0U) >> 6); \
114 (s)[(k)] = (uchar_t)(0x80U | ((uint32_t)(b) & 0x003FU));
115
116 #define U8_HANGUL_JAMO_L(u) \
117 ((u) >= U8_HANGUL_JAMO_L_FIRST && (u) <= U8_HANGUL_JAMO_L_LAST)
118
119 #define U8_HANGUL_JAMO_V(u) \
120 ((u) >= U8_HANGUL_JAMO_V_FIRST && (u) <= U8_HANGUL_JAMO_V_LAST)
121
122 #define U8_HANGUL_JAMO_T(u) \
123 ((u) > U8_HANGUL_JAMO_T_FIRST && (u) <= U8_HANGUL_JAMO_T_LAST)
124
125 #define U8_HANGUL_JAMO(u) \
126 ((u) >= U8_HANGUL_JAMO_L_FIRST && (u) <= U8_HANGUL_JAMO_T_LAST)
127
128 #define U8_HANGUL_SYLLABLE(u) \
129 ((u) >= U8_HANGUL_SYL_FIRST && (u) <= U8_HANGUL_SYL_LAST)
130
131 #define U8_HANGUL_COMPOSABLE_L_V(s, u) \
132 ((s) == U8_STATE_HANGUL_L && U8_HANGUL_JAMO_V((u)))
133
134 #define U8_HANGUL_COMPOSABLE_LV_T(s, u) \
135 ((s) == U8_STATE_HANGUL_LV && U8_HANGUL_JAMO_T((u)))
136
137 /* The types of decomposition mappings. */
138 #define U8_DECOMP_BOTH (0xF5U)
139 #define U8_DECOMP_CANONICAL (0xF6U)
140
141 /* The indicator for 16-bit table. */
142 #define U8_16BIT_TABLE_INDICATOR (0x8000U)
143
144 /* The following are some convenience macros. */
145 #define U8_PUT_3BYTES_INTO_UTF32(u, b1, b2, b3) \
146 (u) = ((((uint32_t)(b1) & 0x0F) << 12) | \
147 (((uint32_t)(b2) & 0x3F) << 6) | \
148 ((uint32_t)(b3) & 0x3F));
149
150 #define U8_SIMPLE_SWAP(a, b, t) \
151 (t) = (a); \
152 (a) = (b); \
153 (b) = (t);
154
155 #define U8_ASCII_TOUPPER(c) \
156 (((c) >= 'a' && (c) <= 'z') ? (c) - 'a' + 'A' : (c))
157
158 #define U8_ASCII_TOLOWER(c) \
159 (((c) >= 'A' && (c) <= 'Z') ? (c) - 'A' + 'a' : (c))
160
161 #define U8_ISASCII(c) (((uchar_t)(c)) < 0x80U)
162 /*
163 * The following macro assumes that the two characters that are to be
164 * swapped are adjacent to each other and 'a' comes before 'b'.
165 *
166 * If the assumptions are not met, then, the macro will fail.
167 */
168 #define U8_SWAP_COMB_MARKS(a, b) \
169 for (k = 0; k < disp[(a)]; k++) \
170 u8t[k] = u8s[start[(a)] + k]; \
171 for (k = 0; k < disp[(b)]; k++) \
172 u8s[start[(a)] + k] = u8s[start[(b)] + k]; \
173 start[(b)] = start[(a)] + disp[(b)]; \
174 for (k = 0; k < disp[(a)]; k++) \
175 u8s[start[(b)] + k] = u8t[k]; \
176 U8_SIMPLE_SWAP(comb_class[(a)], comb_class[(b)], tc); \
177 U8_SIMPLE_SWAP(disp[(a)], disp[(b)], tc);
178
179 /* The possible states during normalization. */
180 typedef enum {
181 U8_STATE_START = 0,
182 U8_STATE_HANGUL_L = 1,
183 U8_STATE_HANGUL_LV = 2,
184 U8_STATE_HANGUL_LVT = 3,
185 U8_STATE_HANGUL_V = 4,
186 U8_STATE_HANGUL_T = 5,
187 U8_STATE_COMBINING_MARK = 6
188 } u8_normalization_states_t;
189
190 /*
191 * The three vectors at below are used to check bytes of a given UTF-8
192 * character are valid and not containing any malformed byte values.
193 *
194 * We used to have a quite relaxed UTF-8 binary representation but then there
195 * was some security related issues and so the Unicode Consortium defined
196 * and announced the UTF-8 Corrigendum at Unicode 3.1 and then refined it
197 * one more time at the Unicode 3.2. The following three tables are based on
198 * that.
199 */
200
201 #define U8_ILLEGAL_NEXT_BYTE_COMMON(c) ((c) < 0x80 || (c) > 0xBF)
202
203 #define I_ U8_ILLEGAL_CHAR
204 #define O_ U8_OUT_OF_RANGE_CHAR
205
206 static const int8_t u8_number_of_bytes[0x100] = {
207 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
208 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
209 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
210 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
211 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
212 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
213 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
214 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
215
216 /* 80 81 82 83 84 85 86 87 88 89 8A 8B 8C 8D 8E 8F */
217 I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_,
218
219 /* 90 91 92 93 94 95 96 97 98 99 9A 9B 9C 9D 9E 9F */
220 I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_,
221
222 /* A0 A1 A2 A3 A4 A5 A6 A7 A8 A9 AA AB AC AD AE AF */
223 I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_,
224
225 /* B0 B1 B2 B3 B4 B5 B6 B7 B8 B9 BA BB BC BD BE BF */
226 I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_, I_,
227
228 /* C0 C1 C2 C3 C4 C5 C6 C7 C8 C9 CA CB CC CD CE CF */
229 I_, I_, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
230
231 /* D0 D1 D2 D3 D4 D5 D6 D7 D8 D9 DA DB DC DD DE DF */
232 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
233
234 /* E0 E1 E2 E3 E4 E5 E6 E7 E8 E9 EA EB EC ED EE EF */
235 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,
236
237 /* F0 F1 F2 F3 F4 F5 F6 F7 F8 F9 FA FB FC FD FE FF */
238 4, 4, 4, 4, 4, O_, O_, O_, O_, O_, O_, O_, O_, O_, O_, O_,
239 };
240
241 #undef I_
242 #undef O_
243
244 static const uint8_t u8_valid_min_2nd_byte[0x100] = {
245 0, 0, 0, 0, 0, 0, 0, 0,
246 0, 0, 0, 0, 0, 0, 0, 0,
247 0, 0, 0, 0, 0, 0, 0, 0,
248 0, 0, 0, 0, 0, 0, 0, 0,
249 0, 0, 0, 0, 0, 0, 0, 0,
250 0, 0, 0, 0, 0, 0, 0, 0,
251 0, 0, 0, 0, 0, 0, 0, 0,
252 0, 0, 0, 0, 0, 0, 0, 0,
253 0, 0, 0, 0, 0, 0, 0, 0,
254 0, 0, 0, 0, 0, 0, 0, 0,
255 0, 0, 0, 0, 0, 0, 0, 0,
256 0, 0, 0, 0, 0, 0, 0, 0,
257 0, 0, 0, 0, 0, 0, 0, 0,
258 0, 0, 0, 0, 0, 0, 0, 0,
259 0, 0, 0, 0, 0, 0, 0, 0,
260 0, 0, 0, 0, 0, 0, 0, 0,
261 0, 0, 0, 0, 0, 0, 0, 0,
262 0, 0, 0, 0, 0, 0, 0, 0,
263 0, 0, 0, 0, 0, 0, 0, 0,
264 0, 0, 0, 0, 0, 0, 0, 0,
265 0, 0, 0, 0, 0, 0, 0, 0,
266 0, 0, 0, 0, 0, 0, 0, 0,
267 0, 0, 0, 0, 0, 0, 0, 0,
268 0, 0, 0, 0, 0, 0, 0, 0,
269 /* C0 C1 C2 C3 C4 C5 C6 C7 */
270 0, 0, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
271 /* C8 C9 CA CB CC CD CE CF */
272 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
273 /* D0 D1 D2 D3 D4 D5 D6 D7 */
274 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
275 /* D8 D9 DA DB DC DD DE DF */
276 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
277 /* E0 E1 E2 E3 E4 E5 E6 E7 */
278 0xa0, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
279 /* E8 E9 EA EB EC ED EE EF */
280 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
281 /* F0 F1 F2 F3 F4 F5 F6 F7 */
282 0x90, 0x80, 0x80, 0x80, 0x80, 0, 0, 0,
283 0, 0, 0, 0, 0, 0, 0, 0,
284 };
285
286 static const uint8_t u8_valid_max_2nd_byte[0x100] = {
287 0, 0, 0, 0, 0, 0, 0, 0,
288 0, 0, 0, 0, 0, 0, 0, 0,
289 0, 0, 0, 0, 0, 0, 0, 0,
290 0, 0, 0, 0, 0, 0, 0, 0,
291 0, 0, 0, 0, 0, 0, 0, 0,
292 0, 0, 0, 0, 0, 0, 0, 0,
293 0, 0, 0, 0, 0, 0, 0, 0,
294 0, 0, 0, 0, 0, 0, 0, 0,
295 0, 0, 0, 0, 0, 0, 0, 0,
296 0, 0, 0, 0, 0, 0, 0, 0,
297 0, 0, 0, 0, 0, 0, 0, 0,
298 0, 0, 0, 0, 0, 0, 0, 0,
299 0, 0, 0, 0, 0, 0, 0, 0,
300 0, 0, 0, 0, 0, 0, 0, 0,
301 0, 0, 0, 0, 0, 0, 0, 0,
302 0, 0, 0, 0, 0, 0, 0, 0,
303 0, 0, 0, 0, 0, 0, 0, 0,
304 0, 0, 0, 0, 0, 0, 0, 0,
305 0, 0, 0, 0, 0, 0, 0, 0,
306 0, 0, 0, 0, 0, 0, 0, 0,
307 0, 0, 0, 0, 0, 0, 0, 0,
308 0, 0, 0, 0, 0, 0, 0, 0,
309 0, 0, 0, 0, 0, 0, 0, 0,
310 0, 0, 0, 0, 0, 0, 0, 0,
311 /* C0 C1 C2 C3 C4 C5 C6 C7 */
312 0, 0, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf,
313 /* C8 C9 CA CB CC CD CE CF */
314 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf,
315 /* D0 D1 D2 D3 D4 D5 D6 D7 */
316 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf,
317 /* D8 D9 DA DB DC DD DE DF */
318 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf,
319 /* E0 E1 E2 E3 E4 E5 E6 E7 */
320 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf,
321 /* E8 E9 EA EB EC ED EE EF */
322 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, 0x9f, 0xbf, 0xbf,
323 /* F0 F1 F2 F3 F4 F5 F6 F7 */
324 0xbf, 0xbf, 0xbf, 0xbf, 0x8f, 0, 0, 0,
325 0, 0, 0, 0, 0, 0, 0, 0,
326 };
327
328
329 /*
330 * The u8_validate() validates on the given UTF-8 character string and
331 * calculate the byte length. It is quite similar to mblen(3C) except that
332 * this will validate against the list of characters if required and
333 * specific to UTF-8 and Unicode.
334 */
335 int
336 u8_validate(const char *u8str, size_t n, char **list, int flag, int *errnum)
337 {
338 uchar_t *ib;
339 uchar_t *ibtail;
340 uchar_t **p;
341 uchar_t *s1;
342 uchar_t *s2;
343 uchar_t f;
344 int sz;
345 size_t i;
346 int ret_val;
347 boolean_t second;
348 boolean_t no_need_to_validate_entire;
349 boolean_t check_additional;
350 boolean_t validate_ucs2_range_only;
351
352 if (! u8str)
353 return (0);
354
355 ib = (uchar_t *)u8str;
356 ibtail = ib + n;
357
358 ret_val = 0;
359
360 no_need_to_validate_entire = ! (flag & U8_VALIDATE_ENTIRE);
361 check_additional = flag & U8_VALIDATE_CHECK_ADDITIONAL;
362 validate_ucs2_range_only = flag & U8_VALIDATE_UCS2_RANGE;
363
364 while (ib < ibtail) {
365 /*
366 * The first byte of a UTF-8 character tells how many
367 * bytes will follow for the character. If the first byte
368 * is an illegal byte value or out of range value, we just
369 * return -1 with an appropriate error number.
370 */
371 sz = u8_number_of_bytes[*ib];
372 if (sz == U8_ILLEGAL_CHAR) {
373 *errnum = EILSEQ;
374 return (-1);
375 }
376
377 if (sz == U8_OUT_OF_RANGE_CHAR ||
378 (validate_ucs2_range_only && sz > U8_MAX_BYTES_UCS2)) {
379 *errnum = ERANGE;
380 return (-1);
381 }
382
383 /*
384 * If we don't have enough bytes to check on, that's also
385 * an error. As you can see, we give illegal byte sequence
386 * checking higher priority then EINVAL cases.
387 */
388 if ((ibtail - ib) < sz) {
389 *errnum = EINVAL;
390 return (-1);
391 }
392
393 if (sz == 1) {
394 ib++;
395 ret_val++;
396 } else {
397 /*
398 * Check on the multi-byte UTF-8 character. For more
399 * details on this, see comment added for the used
400 * data structures at the beginning of the file.
401 */
402 f = *ib++;
403 ret_val++;
404 second = B_TRUE;
405 for (i = 1; i < sz; i++) {
406 if (second) {
407 if (*ib < u8_valid_min_2nd_byte[f] ||
408 *ib > u8_valid_max_2nd_byte[f]) {
409 *errnum = EILSEQ;
410 return (-1);
411 }
412 second = B_FALSE;
413 } else if (U8_ILLEGAL_NEXT_BYTE_COMMON(*ib)) {
414 *errnum = EILSEQ;
415 return (-1);
416 }
417 ib++;
418 ret_val++;
419 }
420 }
421
422 if (check_additional) {
423 for (p = (uchar_t **)list, i = 0; p[i]; i++) {
424 s1 = ib - sz;
425 s2 = p[i];
426 while (s1 < ib) {
427 if (*s1 != *s2 || *s2 == '\0')
428 break;
429 s1++;
430 s2++;
431 }
432
433 if (s1 >= ib && *s2 == '\0') {
434 *errnum = EBADF;
435 return (-1);
436 }
437 }
438 }
439
440 if (no_need_to_validate_entire)
441 break;
442 }
443
444 return (ret_val);
445 }
446
447 /*
448 * The do_case_conv() looks at the mapping tables and returns found
449 * bytes if any. If not found, the input bytes are returned. The function
450 * always terminate the return bytes with a null character assuming that
451 * there are plenty of room to do so.
452 *
453 * The case conversions are simple case conversions mapping a character to
454 * another character as specified in the Unicode data. The byte size of
455 * the mapped character could be different from that of the input character.
456 *
457 * The return value is the byte length of the returned character excluding
458 * the terminating null byte.
459 */
460 static size_t
461 do_case_conv(int uv, uchar_t *u8s, uchar_t *s, int sz, boolean_t is_it_toupper)
462 {
463 size_t i;
464 uint16_t b1 = 0;
465 uint16_t b2 = 0;
466 uint16_t b3 = 0;
467 uint16_t b3_tbl;
468 uint16_t b3_base;
469 uint16_t b4 = 0;
470 size_t start_id;
471 size_t end_id;
472
473 /*
474 * At this point, the only possible values for sz are 2, 3, and 4.
475 * The u8s should point to a vector that is well beyond the size of
476 * 5 bytes.
477 */
478 if (sz == 2) {
479 b3 = u8s[0] = s[0];
480 b4 = u8s[1] = s[1];
481 } else if (sz == 3) {
482 b2 = u8s[0] = s[0];
483 b3 = u8s[1] = s[1];
484 b4 = u8s[2] = s[2];
485 } else if (sz == 4) {
486 b1 = u8s[0] = s[0];
487 b2 = u8s[1] = s[1];
488 b3 = u8s[2] = s[2];
489 b4 = u8s[3] = s[3];
490 } else {
491 /* This is not possible but just in case as a fallback. */
492 if (is_it_toupper)
493 *u8s = U8_ASCII_TOUPPER(*s);
494 else
495 *u8s = U8_ASCII_TOLOWER(*s);
496 u8s[1] = '\0';
497
498 return (1);
499 }
500 u8s[sz] = '\0';
501
502 /*
503 * Let's find out if we have a corresponding character.
504 */
505 b1 = u8_common_b1_tbl[uv][b1];
506 if (b1 == U8_TBL_ELEMENT_NOT_DEF)
507 return ((size_t)sz);
508
509 b2 = u8_case_common_b2_tbl[uv][b1][b2];
510 if (b2 == U8_TBL_ELEMENT_NOT_DEF)
511 return ((size_t)sz);
512
513 if (is_it_toupper) {
514 b3_tbl = u8_toupper_b3_tbl[uv][b2][b3].tbl_id;
515 if (b3_tbl == U8_TBL_ELEMENT_NOT_DEF)
516 return ((size_t)sz);
517
518 start_id = u8_toupper_b4_tbl[uv][b3_tbl][b4];
519 end_id = u8_toupper_b4_tbl[uv][b3_tbl][b4 + 1];
520
521 /* Either there is no match or an error at the table. */
522 if (start_id >= end_id || (end_id - start_id) > U8_MB_CUR_MAX)
523 return ((size_t)sz);
524
525 b3_base = u8_toupper_b3_tbl[uv][b2][b3].base;
526
527 for (i = 0; start_id < end_id; start_id++)
528 u8s[i++] = u8_toupper_final_tbl[uv][b3_base + start_id];
529 } else {
530 #ifdef U8_STRCMP_CI_LOWER
531 b3_tbl = u8_tolower_b3_tbl[uv][b2][b3].tbl_id;
532 if (b3_tbl == U8_TBL_ELEMENT_NOT_DEF)
533 return ((size_t)sz);
534
535 start_id = u8_tolower_b4_tbl[uv][b3_tbl][b4];
536 end_id = u8_tolower_b4_tbl[uv][b3_tbl][b4 + 1];
537
538 if (start_id >= end_id || (end_id - start_id) > U8_MB_CUR_MAX)
539 return ((size_t)sz);
540
541 b3_base = u8_tolower_b3_tbl[uv][b2][b3].base;
542
543 for (i = 0; start_id < end_id; start_id++)
544 u8s[i++] = u8_tolower_final_tbl[uv][b3_base + start_id];
545 #else
546 __builtin_unreachable();
547 #endif
548 }
549
550 /*
551 * If i is still zero, that means there is no corresponding character.
552 */
553 if (i == 0)
554 return ((size_t)sz);
555
556 u8s[i] = '\0';
557
558 return (i);
559 }
560
561 /*
562 * The do_case_compare() function compares the two input strings, s1 and s2,
563 * one character at a time doing case conversions if applicable and return
564 * the comparison result as like strcmp().
565 *
566 * Since, in empirical sense, most of text data are 7-bit ASCII characters,
567 * we treat the 7-bit ASCII characters as a special case trying to yield
568 * faster processing time.
569 */
570 static int
571 do_case_compare(size_t uv, uchar_t *s1, uchar_t *s2, size_t n1,
572 size_t n2, boolean_t is_it_toupper, int *errnum)
573 {
574 int f;
575 int sz1;
576 int sz2;
577 size_t j;
578 size_t i1;
579 size_t i2;
580 uchar_t u8s1[U8_MB_CUR_MAX + 1];
581 uchar_t u8s2[U8_MB_CUR_MAX + 1];
582
583 i1 = i2 = 0;
584 while (i1 < n1 && i2 < n2) {
585 /*
586 * Find out what would be the byte length for this UTF-8
587 * character at string s1 and also find out if this is
588 * an illegal start byte or not and if so, issue a proper
589 * error number and yet treat this byte as a character.
590 */
591 sz1 = u8_number_of_bytes[*s1];
592 if (sz1 < 0) {
593 *errnum = EILSEQ;
594 sz1 = 1;
595 }
596
597 /*
598 * For 7-bit ASCII characters mainly, we do a quick case
599 * conversion right at here.
600 *
601 * If we don't have enough bytes for this character, issue
602 * an EINVAL error and use what are available.
603 *
604 * If we have enough bytes, find out if there is
605 * a corresponding uppercase character and if so, copy over
606 * the bytes for a comparison later. If there is no
607 * corresponding uppercase character, then, use what we have
608 * for the comparison.
609 */
610 if (sz1 == 1) {
611 if (is_it_toupper)
612 u8s1[0] = U8_ASCII_TOUPPER(*s1);
613 else
614 u8s1[0] = U8_ASCII_TOLOWER(*s1);
615 s1++;
616 u8s1[1] = '\0';
617 } else if ((i1 + sz1) > n1) {
618 *errnum = EINVAL;
619 for (j = 0; (i1 + j) < n1; )
620 u8s1[j++] = *s1++;
621 u8s1[j] = '\0';
622 } else {
623 (void) do_case_conv(uv, u8s1, s1, sz1, is_it_toupper);
624 s1 += sz1;
625 }
626
627 /* Do the same for the string s2. */
628 sz2 = u8_number_of_bytes[*s2];
629 if (sz2 < 0) {
630 *errnum = EILSEQ;
631 sz2 = 1;
632 }
633
634 if (sz2 == 1) {
635 if (is_it_toupper)
636 u8s2[0] = U8_ASCII_TOUPPER(*s2);
637 else
638 u8s2[0] = U8_ASCII_TOLOWER(*s2);
639 s2++;
640 u8s2[1] = '\0';
641 } else if ((i2 + sz2) > n2) {
642 *errnum = EINVAL;
643 for (j = 0; (i2 + j) < n2; )
644 u8s2[j++] = *s2++;
645 u8s2[j] = '\0';
646 } else {
647 (void) do_case_conv(uv, u8s2, s2, sz2, is_it_toupper);
648 s2 += sz2;
649 }
650
651 /* Now compare the two characters. */
652 if (sz1 == 1 && sz2 == 1) {
653 if (*u8s1 > *u8s2)
654 return (1);
655 if (*u8s1 < *u8s2)
656 return (-1);
657 } else {
658 f = strcmp((const char *)u8s1, (const char *)u8s2);
659 if (f != 0)
660 return (f);
661 }
662
663 /*
664 * They were the same. Let's move on to the next
665 * characters then.
666 */
667 i1 += sz1;
668 i2 += sz2;
669 }
670
671 /*
672 * We compared until the end of either or both strings.
673 *
674 * If we reached to or went over the ends for the both, that means
675 * they are the same.
676 *
677 * If we reached only one of the two ends, that means the other string
678 * has something which then the fact can be used to determine
679 * the return value.
680 */
681 if (i1 >= n1) {
682 if (i2 >= n2)
683 return (0);
684 return (-1);
685 }
686 return (1);
687 }
688
689 /*
690 * The combining_class() function checks on the given bytes and find out
691 * the corresponding Unicode combining class value. The return value 0 means
692 * it is a Starter. Any illegal UTF-8 character will also be treated as
693 * a Starter.
694 */
695 static uchar_t
696 combining_class(size_t uv, uchar_t *s, size_t sz)
697 {
698 uint16_t b1 = 0;
699 uint16_t b2 = 0;
700 uint16_t b3 = 0;
701 uint16_t b4 = 0;
702
703 if (sz == 1 || sz > 4)
704 return (0);
705
706 if (sz == 2) {
707 b3 = s[0];
708 b4 = s[1];
709 } else if (sz == 3) {
710 b2 = s[0];
711 b3 = s[1];
712 b4 = s[2];
713 } else if (sz == 4) {
714 b1 = s[0];
715 b2 = s[1];
716 b3 = s[2];
717 b4 = s[3];
718 }
719
720 b1 = u8_common_b1_tbl[uv][b1];
721 if (b1 == U8_TBL_ELEMENT_NOT_DEF)
722 return (0);
723
724 b2 = u8_combining_class_b2_tbl[uv][b1][b2];
725 if (b2 == U8_TBL_ELEMENT_NOT_DEF)
726 return (0);
727
728 b3 = u8_combining_class_b3_tbl[uv][b2][b3];
729 if (b3 == U8_TBL_ELEMENT_NOT_DEF)
730 return (0);
731
732 return (u8_combining_class_b4_tbl[uv][b3][b4]);
733 }
734
735 /*
736 * The do_decomp() function finds out a matching decomposition if any
737 * and return. If there is no match, the input bytes are copied and returned.
738 * The function also checks if there is a Hangul, decomposes it if necessary
739 * and returns.
740 *
741 * To save time, a single byte 7-bit ASCII character should be handled by
742 * the caller.
743 *
744 * The function returns the number of bytes returned sans always terminating
745 * the null byte. It will also return a state that will tell if there was
746 * a Hangul character decomposed which then will be used by the caller.
747 */
748 static size_t
749 do_decomp(size_t uv, uchar_t *u8s, uchar_t *s, int sz,
750 boolean_t canonical_decomposition, u8_normalization_states_t *state)
751 {
752 uint16_t b1 = 0;
753 uint16_t b2 = 0;
754 uint16_t b3 = 0;
755 uint16_t b3_tbl;
756 uint16_t b3_base;
757 uint16_t b4 = 0;
758 size_t start_id;
759 size_t end_id;
760 size_t i;
761 uint32_t u1;
762
763 if (sz == 2) {
764 b3 = u8s[0] = s[0];
765 b4 = u8s[1] = s[1];
766 u8s[2] = '\0';
767 } else if (sz == 3) {
768 /* Convert it to a Unicode scalar value. */
769 U8_PUT_3BYTES_INTO_UTF32(u1, s[0], s[1], s[2]);
770
771 /*
772 * If this is a Hangul syllable, we decompose it into
773 * a leading consonant, a vowel, and an optional trailing
774 * consonant and then return.
775 */
776 if (U8_HANGUL_SYLLABLE(u1)) {
777 u1 -= U8_HANGUL_SYL_FIRST;
778
779 b1 = U8_HANGUL_JAMO_L_FIRST + u1 / U8_HANGUL_VT_COUNT;
780 b2 = U8_HANGUL_JAMO_V_FIRST + (u1 % U8_HANGUL_VT_COUNT)
781 / U8_HANGUL_T_COUNT;
782 b3 = u1 % U8_HANGUL_T_COUNT;
783
784 U8_SAVE_HANGUL_AS_UTF8(u8s, 0, 1, 2, b1);
785 U8_SAVE_HANGUL_AS_UTF8(u8s, 3, 4, 5, b2);
786 if (b3) {
787 b3 += U8_HANGUL_JAMO_T_FIRST;
788 U8_SAVE_HANGUL_AS_UTF8(u8s, 6, 7, 8, b3);
789
790 u8s[9] = '\0';
791 *state = U8_STATE_HANGUL_LVT;
792 return (9);
793 }
794
795 u8s[6] = '\0';
796 *state = U8_STATE_HANGUL_LV;
797 return (6);
798 }
799
800 b2 = u8s[0] = s[0];
801 b3 = u8s[1] = s[1];
802 b4 = u8s[2] = s[2];
803 u8s[3] = '\0';
804
805 /*
806 * If this is a Hangul Jamo, we know there is nothing
807 * further that we can decompose.
808 */
809 if (U8_HANGUL_JAMO_L(u1)) {
810 *state = U8_STATE_HANGUL_L;
811 return (3);
812 }
813
814 if (U8_HANGUL_JAMO_V(u1)) {
815 if (*state == U8_STATE_HANGUL_L)
816 *state = U8_STATE_HANGUL_LV;
817 else
818 *state = U8_STATE_HANGUL_V;
819 return (3);
820 }
821
822 if (U8_HANGUL_JAMO_T(u1)) {
823 if (*state == U8_STATE_HANGUL_LV)
824 *state = U8_STATE_HANGUL_LVT;
825 else
826 *state = U8_STATE_HANGUL_T;
827 return (3);
828 }
829 } else if (sz == 4) {
830 b1 = u8s[0] = s[0];
831 b2 = u8s[1] = s[1];
832 b3 = u8s[2] = s[2];
833 b4 = u8s[3] = s[3];
834 u8s[4] = '\0';
835 } else {
836 /*
837 * This is a fallback and should not happen if the function
838 * was called properly.
839 */
840 u8s[0] = s[0];
841 u8s[1] = '\0';
842 *state = U8_STATE_START;
843 return (1);
844 }
845
846 /*
847 * At this point, this routine does not know what it would get.
848 * The caller should sort it out if the state isn't a Hangul one.
849 */
850 *state = U8_STATE_START;
851
852 /* Try to find matching decomposition mapping byte sequence. */
853 b1 = u8_common_b1_tbl[uv][b1];
854 if (b1 == U8_TBL_ELEMENT_NOT_DEF)
855 return ((size_t)sz);
856
857 b2 = u8_decomp_b2_tbl[uv][b1][b2];
858 if (b2 == U8_TBL_ELEMENT_NOT_DEF)
859 return ((size_t)sz);
860
861 b3_tbl = u8_decomp_b3_tbl[uv][b2][b3].tbl_id;
862 if (b3_tbl == U8_TBL_ELEMENT_NOT_DEF)
863 return ((size_t)sz);
864
865 /*
866 * If b3_tbl is bigger than or equal to U8_16BIT_TABLE_INDICATOR
867 * which is 0x8000, this means we couldn't fit the mappings into
868 * the cardinality of a unsigned byte.
869 */
870 if (b3_tbl >= U8_16BIT_TABLE_INDICATOR) {
871 b3_tbl -= U8_16BIT_TABLE_INDICATOR;
872 start_id = u8_decomp_b4_16bit_tbl[uv][b3_tbl][b4];
873 end_id = u8_decomp_b4_16bit_tbl[uv][b3_tbl][b4 + 1];
874 } else {
875 // cppcheck-suppress arrayIndexOutOfBoundsCond
876 start_id = u8_decomp_b4_tbl[uv][b3_tbl][b4];
877 // cppcheck-suppress arrayIndexOutOfBoundsCond
878 end_id = u8_decomp_b4_tbl[uv][b3_tbl][b4 + 1];
879 }
880
881 /* This also means there wasn't any matching decomposition. */
882 if (start_id >= end_id)
883 return ((size_t)sz);
884
885 /*
886 * The final table for decomposition mappings has three types of
887 * byte sequences depending on whether a mapping is for compatibility
888 * decomposition, canonical decomposition, or both like the following:
889 *
890 * (1) Compatibility decomposition mappings:
891 *
892 * +---+---+-...-+---+
893 * | B0| B1| ... | Bm|
894 * +---+---+-...-+---+
895 *
896 * The first byte, B0, is always less than 0xF5 (U8_DECOMP_BOTH).
897 *
898 * (2) Canonical decomposition mappings:
899 *
900 * +---+---+---+-...-+---+
901 * | T | b0| b1| ... | bn|
902 * +---+---+---+-...-+---+
903 *
904 * where the first byte, T, is 0xF6 (U8_DECOMP_CANONICAL).
905 *
906 * (3) Both mappings:
907 *
908 * +---+---+---+---+-...-+---+---+---+-...-+---+
909 * | T | D | b0| b1| ... | bn| B0| B1| ... | Bm|
910 * +---+---+---+---+-...-+---+---+---+-...-+---+
911 *
912 * where T is 0xF5 (U8_DECOMP_BOTH) and D is a displacement
913 * byte, b0 to bn are canonical mapping bytes and B0 to Bm are
914 * compatibility mapping bytes.
915 *
916 * Note that compatibility decomposition means doing recursive
917 * decompositions using both compatibility decomposition mappings and
918 * canonical decomposition mappings. On the other hand, canonical
919 * decomposition means doing recursive decompositions using only
920 * canonical decomposition mappings. Since the table we have has gone
921 * through the recursions already, we do not need to do so during
922 * runtime, i.e., the table has been completely flattened out
923 * already.
924 */
925
926 b3_base = u8_decomp_b3_tbl[uv][b2][b3].base;
927
928 /* Get the type, T, of the byte sequence. */
929 b1 = u8_decomp_final_tbl[uv][b3_base + start_id];
930
931 /*
932 * If necessary, adjust start_id, end_id, or both. Note that if
933 * this is compatibility decomposition mapping, there is no
934 * adjustment.
935 */
936 if (canonical_decomposition) {
937 /* Is the mapping only for compatibility decomposition? */
938 if (b1 < U8_DECOMP_BOTH)
939 return ((size_t)sz);
940
941 start_id++;
942
943 if (b1 == U8_DECOMP_BOTH) {
944 end_id = start_id +
945 u8_decomp_final_tbl[uv][b3_base + start_id];
946 start_id++;
947 }
948 } else {
949 /*
950 * Unless this is a compatibility decomposition mapping,
951 * we adjust the start_id.
952 */
953 if (b1 == U8_DECOMP_BOTH) {
954 start_id++;
955 start_id += u8_decomp_final_tbl[uv][b3_base + start_id];
956 } else if (b1 == U8_DECOMP_CANONICAL) {
957 start_id++;
958 }
959 }
960
961 for (i = 0; start_id < end_id; start_id++)
962 u8s[i++] = u8_decomp_final_tbl[uv][b3_base + start_id];
963 u8s[i] = '\0';
964
965 return (i);
966 }
967
968 /*
969 * The find_composition_start() function uses the character bytes given and
970 * find out the matching composition mappings if any and return the address
971 * to the composition mappings as explained in the do_composition().
972 */
973 static uchar_t *
974 find_composition_start(size_t uv, uchar_t *s, size_t sz)
975 {
976 uint16_t b1 = 0;
977 uint16_t b2 = 0;
978 uint16_t b3 = 0;
979 uint16_t b3_tbl;
980 uint16_t b3_base;
981 uint16_t b4 = 0;
982 size_t start_id;
983 size_t end_id;
984
985 if (sz == 1) {
986 b4 = s[0];
987 } else if (sz == 2) {
988 b3 = s[0];
989 b4 = s[1];
990 } else if (sz == 3) {
991 b2 = s[0];
992 b3 = s[1];
993 b4 = s[2];
994 } else if (sz == 4) {
995 b1 = s[0];
996 b2 = s[1];
997 b3 = s[2];
998 b4 = s[3];
999 } else {
1000 /*
1001 * This is a fallback and should not happen if the function
1002 * was called properly.
1003 */
1004 return (NULL);
1005 }
1006
1007 b1 = u8_composition_b1_tbl[uv][b1];
1008 if (b1 == U8_TBL_ELEMENT_NOT_DEF)
1009 return (NULL);
1010
1011 b2 = u8_composition_b2_tbl[uv][b1][b2];
1012 if (b2 == U8_TBL_ELEMENT_NOT_DEF)
1013 return (NULL);
1014
1015 b3_tbl = u8_composition_b3_tbl[uv][b2][b3].tbl_id;
1016 if (b3_tbl == U8_TBL_ELEMENT_NOT_DEF)
1017 return (NULL);
1018
1019 if (b3_tbl >= U8_16BIT_TABLE_INDICATOR) {
1020 b3_tbl -= U8_16BIT_TABLE_INDICATOR;
1021 start_id = u8_composition_b4_16bit_tbl[uv][b3_tbl][b4];
1022 end_id = u8_composition_b4_16bit_tbl[uv][b3_tbl][b4 + 1];
1023 } else {
1024 // cppcheck-suppress arrayIndexOutOfBoundsCond
1025 start_id = u8_composition_b4_tbl[uv][b3_tbl][b4];
1026 // cppcheck-suppress arrayIndexOutOfBoundsCond
1027 end_id = u8_composition_b4_tbl[uv][b3_tbl][b4 + 1];
1028 }
1029
1030 if (start_id >= end_id)
1031 return (NULL);
1032
1033 b3_base = u8_composition_b3_tbl[uv][b2][b3].base;
1034
1035 return ((uchar_t *)&(u8_composition_final_tbl[uv][b3_base + start_id]));
1036 }
1037
1038 /*
1039 * The blocked() function checks on the combining class values of previous
1040 * characters in this sequence and return whether it is blocked or not.
1041 */
1042 static boolean_t
1043 blocked(uchar_t *comb_class, size_t last)
1044 {
1045 uchar_t my_comb_class;
1046 size_t i;
1047
1048 my_comb_class = comb_class[last];
1049 for (i = 1; i < last; i++)
1050 if (comb_class[i] >= my_comb_class ||
1051 comb_class[i] == U8_COMBINING_CLASS_STARTER)
1052 return (B_TRUE);
1053
1054 return (B_FALSE);
1055 }
1056
1057 /*
1058 * The do_composition() reads the character string pointed by 's' and
1059 * do necessary canonical composition and then copy over the result back to
1060 * the 's'.
1061 *
1062 * The input argument 's' cannot contain more than 32 characters.
1063 */
1064 static size_t
1065 do_composition(size_t uv, uchar_t *s, uchar_t *comb_class, uchar_t *start,
1066 uchar_t *disp, size_t last, uchar_t **os, uchar_t *oslast)
1067 {
1068 uchar_t t[U8_STREAM_SAFE_TEXT_MAX + 1];
1069 uchar_t tc[U8_MB_CUR_MAX] = { '\0' };
1070 uint8_t saved_marks[U8_MAX_CHARS_A_SEQ];
1071 size_t saved_marks_count;
1072 uchar_t *p;
1073 uchar_t *saved_p;
1074 uchar_t *q;
1075 size_t i;
1076 size_t saved_i;
1077 size_t j;
1078 size_t k;
1079 size_t l;
1080 size_t C;
1081 size_t saved_l;
1082 size_t size;
1083 uint32_t u1;
1084 uint32_t u2;
1085 boolean_t match_not_found = B_TRUE;
1086
1087 /*
1088 * This should never happen unless the callers are doing some strange
1089 * and unexpected things.
1090 *
1091 * The "last" is the index pointing to the last character not last + 1.
1092 */
1093 if (last >= U8_MAX_CHARS_A_SEQ)
1094 last = U8_UPPER_LIMIT_IN_A_SEQ;
1095
1096 for (i = l = 0; i <= last; i++) {
1097 /*
1098 * The last or any non-Starters at the beginning, we don't
1099 * have any chance to do composition and so we just copy them
1100 * to the temporary buffer.
1101 */
1102 if (i >= last || comb_class[i] != U8_COMBINING_CLASS_STARTER) {
1103 SAVE_THE_CHAR:
1104 p = s + start[i];
1105 size = disp[i];
1106 for (k = 0; k < size; k++)
1107 t[l++] = *p++;
1108 continue;
1109 }
1110
1111 /*
1112 * If this could be a start of Hangul Jamos, then, we try to
1113 * conjoin them.
1114 */
1115 if (s[start[i]] == U8_HANGUL_JAMO_1ST_BYTE) {
1116 U8_PUT_3BYTES_INTO_UTF32(u1, s[start[i]],
1117 s[start[i] + 1], s[start[i] + 2]);
1118 U8_PUT_3BYTES_INTO_UTF32(u2, s[start[i] + 3],
1119 s[start[i] + 4], s[start[i] + 5]);
1120
1121 if (U8_HANGUL_JAMO_L(u1) && U8_HANGUL_JAMO_V(u2)) {
1122 u1 -= U8_HANGUL_JAMO_L_FIRST;
1123 u2 -= U8_HANGUL_JAMO_V_FIRST;
1124 u1 = U8_HANGUL_SYL_FIRST +
1125 (u1 * U8_HANGUL_V_COUNT + u2) *
1126 U8_HANGUL_T_COUNT;
1127
1128 i += 2;
1129 if (i <= last) {
1130 U8_PUT_3BYTES_INTO_UTF32(u2,
1131 s[start[i]], s[start[i] + 1],
1132 s[start[i] + 2]);
1133
1134 if (U8_HANGUL_JAMO_T(u2)) {
1135 u1 += u2 -
1136 U8_HANGUL_JAMO_T_FIRST;
1137 i++;
1138 }
1139 }
1140
1141 U8_SAVE_HANGUL_AS_UTF8(t + l, 0, 1, 2, u1);
1142 i--;
1143 l += 3;
1144 continue;
1145 }
1146 }
1147
1148 /*
1149 * Let's then find out if this Starter has composition
1150 * mapping.
1151 */
1152 p = find_composition_start(uv, s + start[i], disp[i]);
1153 if (p == NULL)
1154 goto SAVE_THE_CHAR;
1155
1156 /*
1157 * We have a Starter with composition mapping and the next
1158 * character is a non-Starter. Let's try to find out if
1159 * we can do composition.
1160 */
1161
1162 saved_p = p;
1163 saved_i = i;
1164 saved_l = l;
1165 saved_marks_count = 0;
1166
1167 TRY_THE_NEXT_MARK:
1168 q = s + start[++i];
1169 size = disp[i];
1170
1171 /*
1172 * The next for() loop compares the non-Starter pointed by
1173 * 'q' with the possible (joinable) characters pointed by 'p'.
1174 *
1175 * The composition final table entry pointed by the 'p'
1176 * looks like the following:
1177 *
1178 * +---+---+---+-...-+---+---+---+---+-...-+---+---+
1179 * | C | b0| b2| ... | bn| F | B0| B1| ... | Bm| F |
1180 * +---+---+---+-...-+---+---+---+---+-...-+---+---+
1181 *
1182 * where C is the count byte indicating the number of
1183 * mapping pairs where each pair would be look like
1184 * (b0-bn F, B0-Bm F). The b0-bn are the bytes of the second
1185 * character of a canonical decomposition and the B0-Bm are
1186 * the bytes of a matching composite character. The F is
1187 * a filler byte after each character as the separator.
1188 */
1189
1190 match_not_found = B_TRUE;
1191
1192 for (C = *p++; C > 0; C--) {
1193 for (k = 0; k < size; p++, k++)
1194 if (*p != q[k])
1195 break;
1196
1197 /* Have we found it? */
1198 if (k >= size && *p == U8_TBL_ELEMENT_FILLER) {
1199 match_not_found = B_FALSE;
1200
1201 l = saved_l;
1202
1203 while (*++p != U8_TBL_ELEMENT_FILLER)
1204 t[l++] = *p;
1205
1206 break;
1207 }
1208
1209 /* We didn't find; skip to the next pair. */
1210 if (*p != U8_TBL_ELEMENT_FILLER)
1211 while (*++p != U8_TBL_ELEMENT_FILLER)
1212 ;
1213 while (*++p != U8_TBL_ELEMENT_FILLER)
1214 ;
1215 p++;
1216 }
1217
1218 /*
1219 * If there was no match, we will need to save the combining
1220 * mark for later appending. After that, if the next one
1221 * is a non-Starter and not blocked, then, we try once
1222 * again to do composition with the next non-Starter.
1223 *
1224 * If there was no match and this was a Starter, then,
1225 * this is a new start.
1226 *
1227 * If there was a match and a composition done and we have
1228 * more to check on, then, we retrieve a new composition final
1229 * table entry for the composite and then try to do the
1230 * composition again.
1231 */
1232
1233 if (match_not_found) {
1234 if (comb_class[i] == U8_COMBINING_CLASS_STARTER) {
1235 i--;
1236 goto SAVE_THE_CHAR;
1237 }
1238
1239 saved_marks[saved_marks_count++] = i;
1240 }
1241
1242 if (saved_l == l) {
1243 while (i < last) {
1244 if (blocked(comb_class, i + 1))
1245 saved_marks[saved_marks_count++] = ++i;
1246 else
1247 break;
1248 }
1249 if (i < last) {
1250 p = saved_p;
1251 goto TRY_THE_NEXT_MARK;
1252 }
1253 } else if (i < last) {
1254 p = find_composition_start(uv, t + saved_l,
1255 l - saved_l);
1256 if (p != NULL) {
1257 saved_p = p;
1258 goto TRY_THE_NEXT_MARK;
1259 }
1260 }
1261
1262 /*
1263 * There is no more composition possible.
1264 *
1265 * If there was no composition what so ever then we copy
1266 * over the original Starter and then append any non-Starters
1267 * remaining at the target string sequentially after that.
1268 */
1269
1270 if (saved_l == l) {
1271 p = s + start[saved_i];
1272 size = disp[saved_i];
1273 for (j = 0; j < size; j++)
1274 t[l++] = *p++;
1275 }
1276
1277 for (k = 0; k < saved_marks_count; k++) {
1278 p = s + start[saved_marks[k]];
1279 size = disp[saved_marks[k]];
1280 for (j = 0; j < size; j++)
1281 t[l++] = *p++;
1282 }
1283 }
1284
1285 /*
1286 * If the last character is a Starter and if we have a character
1287 * (possibly another Starter) that can be turned into a composite,
1288 * we do so and we do so until there is no more of composition
1289 * possible.
1290 */
1291 if (comb_class[last] == U8_COMBINING_CLASS_STARTER) {
1292 p = *os;
1293 saved_l = l - disp[last];
1294
1295 while (p < oslast) {
1296 int8_t number_of_bytes = u8_number_of_bytes[*p];
1297
1298 if (number_of_bytes <= 1)
1299 break;
1300 size = number_of_bytes;
1301 if ((p + size) > oslast)
1302 break;
1303
1304 saved_p = p;
1305
1306 for (i = 0; i < size; i++)
1307 tc[i] = *p++;
1308
1309 q = find_composition_start(uv, t + saved_l,
1310 l - saved_l);
1311 if (q == NULL) {
1312 p = saved_p;
1313 break;
1314 }
1315
1316 match_not_found = B_TRUE;
1317
1318 for (C = *q++; C > 0; C--) {
1319 for (k = 0; k < size; q++, k++)
1320 if (*q != tc[k])
1321 break;
1322
1323 if (k >= size && *q == U8_TBL_ELEMENT_FILLER) {
1324 match_not_found = B_FALSE;
1325
1326 l = saved_l;
1327
1328 while (*++q != U8_TBL_ELEMENT_FILLER) {
1329 /*
1330 * This is practically
1331 * impossible but we don't
1332 * want to take any chances.
1333 */
1334 if (l >=
1335 U8_STREAM_SAFE_TEXT_MAX) {
1336 p = saved_p;
1337 goto SAFE_RETURN;
1338 }
1339 t[l++] = *q;
1340 }
1341
1342 break;
1343 }
1344
1345 if (*q != U8_TBL_ELEMENT_FILLER)
1346 while (*++q != U8_TBL_ELEMENT_FILLER)
1347 ;
1348 while (*++q != U8_TBL_ELEMENT_FILLER)
1349 ;
1350 q++;
1351 }
1352
1353 if (match_not_found) {
1354 p = saved_p;
1355 break;
1356 }
1357 }
1358 SAFE_RETURN:
1359 *os = p;
1360 }
1361
1362 /*
1363 * Now we copy over the temporary string to the target string.
1364 * Since composition always reduces the number of characters or
1365 * the number of characters stay, we don't need to worry about
1366 * the buffer overflow here.
1367 */
1368 for (i = 0; i < l; i++)
1369 s[i] = t[i];
1370 s[l] = '\0';
1371
1372 return (l);
1373 }
1374
1375 /*
1376 * The collect_a_seq() function checks on the given string s, collect
1377 * a sequence of characters at u8s, and return the sequence. While it collects
1378 * a sequence, it also applies case conversion, canonical or compatibility
1379 * decomposition, canonical decomposition, or some or all of them and
1380 * in that order.
1381 *
1382 * The collected sequence cannot be bigger than 32 characters since if
1383 * it is having more than 31 characters, the sequence will be terminated
1384 * with a U+034F COMBINING GRAPHEME JOINER (CGJ) character and turned into
1385 * a Stream-Safe Text. The collected sequence is always terminated with
1386 * a null byte and the return value is the byte length of the sequence
1387 * including 0. The return value does not include the terminating
1388 * null byte.
1389 */
1390 static size_t
1391 collect_a_seq(size_t uv, uchar_t *u8s, uchar_t **source, uchar_t *slast,
1392 boolean_t is_it_toupper,
1393 boolean_t is_it_tolower,
1394 boolean_t canonical_decomposition,
1395 boolean_t compatibility_decomposition,
1396 boolean_t canonical_composition,
1397 int *errnum, u8_normalization_states_t *state)
1398 {
1399 uchar_t *s;
1400 int sz;
1401 int saved_sz;
1402 size_t i;
1403 size_t j;
1404 size_t k;
1405 size_t l;
1406 uchar_t comb_class[U8_MAX_CHARS_A_SEQ];
1407 uchar_t disp[U8_MAX_CHARS_A_SEQ];
1408 uchar_t start[U8_MAX_CHARS_A_SEQ];
1409 uchar_t u8t[U8_MB_CUR_MAX] = { '\0' };
1410 uchar_t uts[U8_STREAM_SAFE_TEXT_MAX + 1];
1411 uchar_t tc;
1412 size_t last;
1413 size_t saved_last;
1414 uint32_t u1;
1415
1416 /*
1417 * Save the source string pointer which we will return a changed
1418 * pointer if we do processing.
1419 */
1420 s = *source;
1421
1422 /*
1423 * The following is a fallback for just in case callers are not
1424 * checking the string boundaries before the calling.
1425 */
1426 if (s >= slast) {
1427 u8s[0] = '\0';
1428
1429 return (0);
1430 }
1431
1432 /*
1433 * As the first thing, let's collect a character and do case
1434 * conversion if necessary.
1435 */
1436
1437 sz = u8_number_of_bytes[*s];
1438
1439 if (sz < 0) {
1440 *errnum = EILSEQ;
1441
1442 u8s[0] = *s++;
1443 u8s[1] = '\0';
1444
1445 *source = s;
1446
1447 return (1);
1448 }
1449
1450 if (sz == 1) {
1451 if (is_it_toupper)
1452 u8s[0] = U8_ASCII_TOUPPER(*s);
1453 else if (is_it_tolower)
1454 u8s[0] = U8_ASCII_TOLOWER(*s);
1455 else
1456 u8s[0] = *s;
1457 s++;
1458 u8s[1] = '\0';
1459 } else if ((s + sz) > slast) {
1460 *errnum = EINVAL;
1461
1462 for (i = 0; s < slast; )
1463 u8s[i++] = *s++;
1464 u8s[i] = '\0';
1465
1466 *source = s;
1467
1468 return (i);
1469 } else {
1470 if (is_it_toupper || is_it_tolower) {
1471 i = do_case_conv(uv, u8s, s, sz, is_it_toupper);
1472 s += sz;
1473 sz = i;
1474 } else {
1475 for (i = 0; i < sz; )
1476 u8s[i++] = *s++;
1477 u8s[i] = '\0';
1478 }
1479 }
1480
1481 /*
1482 * And then canonical/compatibility decomposition followed by
1483 * an optional canonical composition. Please be noted that
1484 * canonical composition is done only when a decomposition is
1485 * done.
1486 */
1487 if (canonical_decomposition || compatibility_decomposition) {
1488 if (sz == 1) {
1489 *state = U8_STATE_START;
1490
1491 saved_sz = 1;
1492
1493 comb_class[0] = 0;
1494 start[0] = 0;
1495 disp[0] = 1;
1496
1497 last = 1;
1498 } else {
1499 saved_sz = do_decomp(uv, u8s, u8s, sz,
1500 canonical_decomposition, state);
1501
1502 last = 0;
1503
1504 for (i = 0; i < saved_sz; ) {
1505 sz = u8_number_of_bytes[u8s[i]];
1506
1507 comb_class[last] = combining_class(uv,
1508 u8s + i, sz);
1509 start[last] = i;
1510 disp[last] = sz;
1511
1512 last++;
1513 i += sz;
1514 }
1515
1516 /*
1517 * Decomposition yields various Hangul related
1518 * states but not on combining marks. We need to
1519 * find out at here by checking on the last
1520 * character.
1521 */
1522 if (*state == U8_STATE_START) {
1523 if (comb_class[last - 1])
1524 *state = U8_STATE_COMBINING_MARK;
1525 }
1526 }
1527
1528 saved_last = last;
1529
1530 while (s < slast) {
1531 sz = u8_number_of_bytes[*s];
1532
1533 /*
1534 * If this is an illegal character, an incomplete
1535 * character, or an 7-bit ASCII Starter character,
1536 * then we have collected a sequence; break and let
1537 * the next call deal with the two cases.
1538 *
1539 * Note that this is okay only if you are using this
1540 * function with a fixed length string, not on
1541 * a buffer with multiple calls of one chunk at a time.
1542 */
1543 if (sz <= 1) {
1544 break;
1545 } else if ((s + sz) > slast) {
1546 break;
1547 } else {
1548 /*
1549 * If the previous character was a Hangul Jamo
1550 * and this character is a Hangul Jamo that
1551 * can be conjoined, we collect the Jamo.
1552 */
1553 if (*s == U8_HANGUL_JAMO_1ST_BYTE) {
1554 U8_PUT_3BYTES_INTO_UTF32(u1,
1555 *s, *(s + 1), *(s + 2));
1556
1557 if (U8_HANGUL_COMPOSABLE_L_V(*state,
1558 u1)) {
1559 i = 0;
1560 *state = U8_STATE_HANGUL_LV;
1561 goto COLLECT_A_HANGUL;
1562 }
1563
1564 if (U8_HANGUL_COMPOSABLE_LV_T(*state,
1565 u1)) {
1566 i = 0;
1567 *state = U8_STATE_HANGUL_LVT;
1568 goto COLLECT_A_HANGUL;
1569 }
1570 }
1571
1572 /*
1573 * Regardless of whatever it was, if this is
1574 * a Starter, we don't collect the character
1575 * since that's a new start and we will deal
1576 * with it at the next time.
1577 */
1578 i = combining_class(uv, s, sz);
1579 if (i == U8_COMBINING_CLASS_STARTER)
1580 break;
1581
1582 /*
1583 * We know the current character is a combining
1584 * mark. If the previous character wasn't
1585 * a Starter (not Hangul) or a combining mark,
1586 * then, we don't collect this combining mark.
1587 */
1588 if (*state != U8_STATE_START &&
1589 *state != U8_STATE_COMBINING_MARK)
1590 break;
1591
1592 *state = U8_STATE_COMBINING_MARK;
1593 COLLECT_A_HANGUL:
1594 /*
1595 * If we collected a Starter and combining
1596 * marks up to 30, i.e., total 31 characters,
1597 * then, we terminate this degenerately long
1598 * combining sequence with a U+034F COMBINING
1599 * GRAPHEME JOINER (CGJ) which is 0xCD 0x8F in
1600 * UTF-8 and turn this into a Stream-Safe
1601 * Text. This will be extremely rare but
1602 * possible.
1603 *
1604 * The following will also guarantee that
1605 * we are not writing more than 32 characters
1606 * plus a NULL at u8s[].
1607 */
1608 if (last >= U8_UPPER_LIMIT_IN_A_SEQ) {
1609 TURN_STREAM_SAFE:
1610 *state = U8_STATE_START;
1611 comb_class[last] = 0;
1612 start[last] = saved_sz;
1613 disp[last] = 2;
1614 last++;
1615
1616 u8s[saved_sz++] = 0xCD;
1617 u8s[saved_sz++] = 0x8F;
1618
1619 break;
1620 }
1621
1622 /*
1623 * Some combining marks also do decompose into
1624 * another combining mark or marks.
1625 */
1626 if (*state == U8_STATE_COMBINING_MARK) {
1627 k = last;
1628 l = sz;
1629 i = do_decomp(uv, uts, s, sz,
1630 canonical_decomposition, state);
1631 for (j = 0; j < i; ) {
1632 sz = u8_number_of_bytes[uts[j]];
1633
1634 comb_class[last] =
1635 combining_class(uv,
1636 uts + j, sz);
1637 start[last] = saved_sz + j;
1638 disp[last] = sz;
1639
1640 last++;
1641 if (last >=
1642 U8_UPPER_LIMIT_IN_A_SEQ) {
1643 last = k;
1644 goto TURN_STREAM_SAFE;
1645 }
1646 j += sz;
1647 }
1648
1649 *state = U8_STATE_COMBINING_MARK;
1650 sz = i;
1651 s += l;
1652
1653 for (i = 0; i < sz; i++)
1654 u8s[saved_sz++] = uts[i];
1655 } else {
1656 comb_class[last] = i;
1657 start[last] = saved_sz;
1658 disp[last] = sz;
1659 last++;
1660
1661 for (i = 0; i < sz; i++)
1662 u8s[saved_sz++] = *s++;
1663 }
1664
1665 /*
1666 * If this is U+0345 COMBINING GREEK
1667 * YPOGEGRAMMENI (0xCD 0x85 in UTF-8), a.k.a.,
1668 * iota subscript, and need to be converted to
1669 * uppercase letter, convert it to U+0399 GREEK
1670 * CAPITAL LETTER IOTA (0xCE 0x99 in UTF-8),
1671 * i.e., convert to capital adscript form as
1672 * specified in the Unicode standard.
1673 *
1674 * This is the only special case of (ambiguous)
1675 * case conversion at combining marks and
1676 * probably the standard will never have
1677 * anything similar like this in future.
1678 */
1679 if (is_it_toupper && sz >= 2 &&
1680 u8s[saved_sz - 2] == 0xCD &&
1681 u8s[saved_sz - 1] == 0x85) {
1682 u8s[saved_sz - 2] = 0xCE;
1683 u8s[saved_sz - 1] = 0x99;
1684 }
1685 }
1686 }
1687
1688 /*
1689 * Let's try to ensure a canonical ordering for the collected
1690 * combining marks. We do this only if we have collected
1691 * at least one more non-Starter. (The decomposition mapping
1692 * data tables have fully (and recursively) expanded and
1693 * canonically ordered decompositions.)
1694 *
1695 * The U8_SWAP_COMB_MARKS() convenience macro has some
1696 * assumptions and we are meeting the assumptions.
1697 */
1698 last--;
1699 if (last >= saved_last) {
1700 for (i = 0; i < last; i++)
1701 for (j = last; j > i; j--)
1702 if (comb_class[j] &&
1703 comb_class[j - 1] > comb_class[j]) {
1704 U8_SWAP_COMB_MARKS(j - 1, j);
1705 }
1706 }
1707
1708 *source = s;
1709
1710 if (! canonical_composition) {
1711 u8s[saved_sz] = '\0';
1712 return (saved_sz);
1713 }
1714
1715 /*
1716 * Now do the canonical composition. Note that we do this
1717 * only after a canonical or compatibility decomposition to
1718 * finish up NFC or NFKC.
1719 */
1720 sz = do_composition(uv, u8s, comb_class, start, disp, last,
1721 &s, slast);
1722 }
1723
1724 *source = s;
1725
1726 return ((size_t)sz);
1727 }
1728
1729 /*
1730 * The do_norm_compare() function does string comparison based on Unicode
1731 * simple case mappings and Unicode Normalization definitions.
1732 *
1733 * It does so by collecting a sequence of character at a time and comparing
1734 * the collected sequences from the strings.
1735 *
1736 * The meanings on the return values are the same as the usual strcmp().
1737 */
1738 static int
1739 do_norm_compare(size_t uv, uchar_t *s1, uchar_t *s2, size_t n1, size_t n2,
1740 int flag, int *errnum)
1741 {
1742 int result;
1743 size_t sz1;
1744 size_t sz2;
1745 uchar_t u8s1[U8_STREAM_SAFE_TEXT_MAX + 1];
1746 uchar_t u8s2[U8_STREAM_SAFE_TEXT_MAX + 1];
1747 uchar_t *s1last;
1748 uchar_t *s2last;
1749 boolean_t is_it_toupper;
1750 boolean_t is_it_tolower;
1751 boolean_t canonical_decomposition;
1752 boolean_t compatibility_decomposition;
1753 boolean_t canonical_composition;
1754 u8_normalization_states_t state;
1755
1756 s1last = s1 + n1;
1757 s2last = s2 + n2;
1758
1759 is_it_toupper = flag & U8_TEXTPREP_TOUPPER;
1760 #ifdef U8_STRCMP_CI_LOWER
1761 is_it_tolower = flag & U8_TEXTPREP_TOLOWER;
1762 #else
1763 is_it_tolower = 0;
1764 #endif
1765 canonical_decomposition = flag & U8_CANON_DECOMP;
1766 compatibility_decomposition = flag & U8_COMPAT_DECOMP;
1767 canonical_composition = flag & U8_CANON_COMP;
1768
1769 while (s1 < s1last && s2 < s2last) {
1770 /*
1771 * If the current character is a 7-bit ASCII and the last
1772 * character, or, if the current character and the next
1773 * character are both some 7-bit ASCII characters then
1774 * we treat the current character as a sequence.
1775 *
1776 * In any other cases, we need to call collect_a_seq().
1777 */
1778
1779 if (U8_ISASCII(*s1) && ((s1 + 1) >= s1last ||
1780 ((s1 + 1) < s1last && U8_ISASCII(*(s1 + 1))))) {
1781 if (is_it_toupper)
1782 u8s1[0] = U8_ASCII_TOUPPER(*s1);
1783 else if (is_it_tolower)
1784 u8s1[0] = U8_ASCII_TOLOWER(*s1);
1785 else
1786 u8s1[0] = *s1;
1787 u8s1[1] = '\0';
1788 sz1 = 1;
1789 s1++;
1790 } else {
1791 state = U8_STATE_START;
1792 sz1 = collect_a_seq(uv, u8s1, &s1, s1last,
1793 is_it_toupper, is_it_tolower,
1794 canonical_decomposition,
1795 compatibility_decomposition,
1796 canonical_composition, errnum, &state);
1797 }
1798
1799 if (U8_ISASCII(*s2) && ((s2 + 1) >= s2last ||
1800 ((s2 + 1) < s2last && U8_ISASCII(*(s2 + 1))))) {
1801 if (is_it_toupper)
1802 u8s2[0] = U8_ASCII_TOUPPER(*s2);
1803 else if (is_it_tolower)
1804 u8s2[0] = U8_ASCII_TOLOWER(*s2);
1805 else
1806 u8s2[0] = *s2;
1807 u8s2[1] = '\0';
1808 sz2 = 1;
1809 s2++;
1810 } else {
1811 state = U8_STATE_START;
1812 sz2 = collect_a_seq(uv, u8s2, &s2, s2last,
1813 is_it_toupper, is_it_tolower,
1814 canonical_decomposition,
1815 compatibility_decomposition,
1816 canonical_composition, errnum, &state);
1817 }
1818
1819 /*
1820 * Now compare the two characters. If they are the same,
1821 * we move on to the next character sequences.
1822 */
1823 if (sz1 == 1 && sz2 == 1) {
1824 if (*u8s1 > *u8s2)
1825 return (1);
1826 if (*u8s1 < *u8s2)
1827 return (-1);
1828 } else {
1829 result = strcmp((const char *)u8s1, (const char *)u8s2);
1830 if (result != 0)
1831 return (result);
1832 }
1833 }
1834
1835 /*
1836 * We compared until the end of either or both strings.
1837 *
1838 * If we reached to or went over the ends for the both, that means
1839 * they are the same.
1840 *
1841 * If we reached only one end, that means the other string has
1842 * something which then can be used to determine the return value.
1843 */
1844 if (s1 >= s1last) {
1845 if (s2 >= s2last)
1846 return (0);
1847 return (-1);
1848 }
1849 return (1);
1850 }
1851
1852 /*
1853 * The u8_strcmp() function compares two UTF-8 strings quite similar to
1854 * the strcmp(). For the comparison, however, Unicode Normalization specific
1855 * equivalency and Unicode simple case conversion mappings based equivalency
1856 * can be requested and checked against.
1857 */
1858 int
1859 u8_strcmp(const char *s1, const char *s2, size_t n, int flag, size_t uv,
1860 int *errnum)
1861 {
1862 int f;
1863 size_t n1;
1864 size_t n2;
1865
1866 *errnum = 0;
1867
1868 /*
1869 * Check on the requested Unicode version, case conversion, and
1870 * normalization flag values.
1871 */
1872
1873 if (uv > U8_UNICODE_LATEST) {
1874 *errnum = ERANGE;
1875 uv = U8_UNICODE_LATEST;
1876 }
1877
1878 if (flag == 0) {
1879 flag = U8_STRCMP_CS;
1880 } else {
1881 #ifdef U8_STRCMP_CI_LOWER
1882 f = flag & (U8_STRCMP_CS | U8_STRCMP_CI_UPPER
1883 | U8_STRCMP_CI_LOWER);
1884 #else
1885 f = flag & (U8_STRCMP_CS | U8_STRCMP_CI_UPPER);
1886 #endif
1887 if (f == 0) {
1888 flag |= U8_STRCMP_CS;
1889 }
1890 #ifdef U8_STRCMP_CI_LOWER
1891 else if (f != U8_STRCMP_CS && f != U8_STRCMP_CI_UPPER &&
1892 f != U8_STRCMP_CI_LOWER)
1893 #else
1894 else if (f != U8_STRCMP_CS && f != U8_STRCMP_CI_UPPER)
1895 #endif
1896 {
1897 *errnum = EBADF;
1898 flag = U8_STRCMP_CS;
1899 }
1900
1901 f = flag & (U8_CANON_DECOMP | U8_COMPAT_DECOMP | U8_CANON_COMP);
1902 if (f && f != U8_STRCMP_NFD && f != U8_STRCMP_NFC &&
1903 f != U8_STRCMP_NFKD && f != U8_STRCMP_NFKC) {
1904 *errnum = EBADF;
1905 flag = U8_STRCMP_CS;
1906 }
1907 }
1908
1909 if (flag == U8_STRCMP_CS) {
1910 return (n == 0 ? strcmp(s1, s2) : strncmp(s1, s2, n));
1911 }
1912
1913 n1 = strlen(s1);
1914 n2 = strlen(s2);
1915 if (n != 0) {
1916 if (n < n1)
1917 n1 = n;
1918 if (n < n2)
1919 n2 = n;
1920 }
1921
1922 /*
1923 * Simple case conversion can be done much faster and so we do
1924 * them separately here.
1925 */
1926 if (flag == U8_STRCMP_CI_UPPER) {
1927 return (do_case_compare(uv, (uchar_t *)s1, (uchar_t *)s2,
1928 n1, n2, B_TRUE, errnum));
1929 }
1930 #ifdef U8_STRCMP_CI_LOWER
1931 else if (flag == U8_STRCMP_CI_LOWER) {
1932 return (do_case_compare(uv, (uchar_t *)s1, (uchar_t *)s2,
1933 n1, n2, B_FALSE, errnum));
1934 }
1935 #endif
1936
1937 return (do_norm_compare(uv, (uchar_t *)s1, (uchar_t *)s2, n1, n2,
1938 flag, errnum));
1939 }
1940
1941 size_t
1942 u8_textprep_str(char *inarray, size_t *inlen, char *outarray, size_t *outlen,
1943 int flag, size_t unicode_version, int *errnum)
1944 {
1945 int f;
1946 int sz;
1947 uchar_t *ib;
1948 uchar_t *ibtail;
1949 uchar_t *ob;
1950 uchar_t *obtail;
1951 boolean_t do_not_ignore_null;
1952 boolean_t do_not_ignore_invalid;
1953 boolean_t is_it_toupper;
1954 boolean_t is_it_tolower;
1955 boolean_t canonical_decomposition;
1956 boolean_t compatibility_decomposition;
1957 boolean_t canonical_composition;
1958 size_t ret_val;
1959 size_t i;
1960 size_t j;
1961 uchar_t u8s[U8_STREAM_SAFE_TEXT_MAX + 1];
1962 u8_normalization_states_t state;
1963
1964 if (unicode_version > U8_UNICODE_LATEST) {
1965 *errnum = ERANGE;
1966 return ((size_t)-1);
1967 }
1968
1969 #ifdef U8_TEXTPREP_TOLOWER
1970 f = flag & (U8_TEXTPREP_TOUPPER | U8_TEXTPREP_TOLOWER);
1971 if (f == (U8_TEXTPREP_TOUPPER | U8_TEXTPREP_TOLOWER)) {
1972 *errnum = EBADF;
1973 return ((size_t)-1);
1974 }
1975 #endif
1976
1977 f = flag & (U8_CANON_DECOMP | U8_COMPAT_DECOMP | U8_CANON_COMP);
1978 if (f && f != U8_TEXTPREP_NFD && f != U8_TEXTPREP_NFC &&
1979 f != U8_TEXTPREP_NFKD && f != U8_TEXTPREP_NFKC) {
1980 *errnum = EBADF;
1981 return ((size_t)-1);
1982 }
1983
1984 if (inarray == NULL || *inlen == 0)
1985 return (0);
1986
1987 if (outarray == NULL) {
1988 *errnum = E2BIG;
1989 return ((size_t)-1);
1990 }
1991
1992 ib = (uchar_t *)inarray;
1993 ob = (uchar_t *)outarray;
1994 ibtail = ib + *inlen;
1995 obtail = ob + *outlen;
1996
1997 do_not_ignore_null = !(flag & U8_TEXTPREP_IGNORE_NULL);
1998 do_not_ignore_invalid = !(flag & U8_TEXTPREP_IGNORE_INVALID);
1999 is_it_toupper = flag & U8_TEXTPREP_TOUPPER;
2000 #ifdef U8_TEXTPREP_TOLOWER
2001 is_it_tolower = flag & U8_TEXTPREP_TOLOWER;
2002 #else
2003 is_it_tolower = 0;
2004 #endif
2005
2006 ret_val = 0;
2007
2008 /*
2009 * If we don't have a normalization flag set, we do the simple case
2010 * conversion based text preparation separately below. Text
2011 * preparation involving Normalization will be done in the false task
2012 * block, again, separately since it will take much more time and
2013 * resource than doing simple case conversions.
2014 */
2015 if (f == 0) {
2016 while (ib < ibtail) {
2017 if (*ib == '\0' && do_not_ignore_null)
2018 break;
2019
2020 sz = u8_number_of_bytes[*ib];
2021
2022 if (sz < 0) {
2023 if (do_not_ignore_invalid) {
2024 *errnum = EILSEQ;
2025 ret_val = (size_t)-1;
2026 break;
2027 }
2028
2029 sz = 1;
2030 ret_val++;
2031 }
2032
2033 if (sz == 1) {
2034 if (ob >= obtail) {
2035 *errnum = E2BIG;
2036 ret_val = (size_t)-1;
2037 break;
2038 }
2039
2040 if (is_it_toupper)
2041 *ob = U8_ASCII_TOUPPER(*ib);
2042 else if (is_it_tolower)
2043 *ob = U8_ASCII_TOLOWER(*ib);
2044 else
2045 *ob = *ib;
2046 ib++;
2047 ob++;
2048 } else if ((ib + sz) > ibtail) {
2049 if (do_not_ignore_invalid) {
2050 *errnum = EINVAL;
2051 ret_val = (size_t)-1;
2052 break;
2053 }
2054
2055 if ((obtail - ob) < (ibtail - ib)) {
2056 *errnum = E2BIG;
2057 ret_val = (size_t)-1;
2058 break;
2059 }
2060
2061 /*
2062 * We treat the remaining incomplete character
2063 * bytes as a character.
2064 */
2065 ret_val++;
2066
2067 while (ib < ibtail)
2068 *ob++ = *ib++;
2069 } else {
2070 if (is_it_toupper || is_it_tolower) {
2071 i = do_case_conv(unicode_version, u8s,
2072 ib, sz, is_it_toupper);
2073
2074 if ((obtail - ob) < i) {
2075 *errnum = E2BIG;
2076 ret_val = (size_t)-1;
2077 break;
2078 }
2079
2080 ib += sz;
2081
2082 for (sz = 0; sz < i; sz++)
2083 *ob++ = u8s[sz];
2084 } else {
2085 if ((obtail - ob) < sz) {
2086 *errnum = E2BIG;
2087 ret_val = (size_t)-1;
2088 break;
2089 }
2090
2091 for (i = 0; i < sz; i++)
2092 *ob++ = *ib++;
2093 }
2094 }
2095 }
2096 } else {
2097 canonical_decomposition = flag & U8_CANON_DECOMP;
2098 compatibility_decomposition = flag & U8_COMPAT_DECOMP;
2099 canonical_composition = flag & U8_CANON_COMP;
2100
2101 while (ib < ibtail) {
2102 if (*ib == '\0' && do_not_ignore_null)
2103 break;
2104
2105 /*
2106 * If the current character is a 7-bit ASCII
2107 * character and it is the last character, or,
2108 * if the current character is a 7-bit ASCII
2109 * character and the next character is also a 7-bit
2110 * ASCII character, then, we copy over this
2111 * character without going through collect_a_seq().
2112 *
2113 * In any other cases, we need to look further with
2114 * the collect_a_seq() function.
2115 */
2116 if (U8_ISASCII(*ib) && ((ib + 1) >= ibtail ||
2117 ((ib + 1) < ibtail && U8_ISASCII(*(ib + 1))))) {
2118 if (ob >= obtail) {
2119 *errnum = E2BIG;
2120 ret_val = (size_t)-1;
2121 break;
2122 }
2123
2124 if (is_it_toupper)
2125 *ob = U8_ASCII_TOUPPER(*ib);
2126 else if (is_it_tolower)
2127 *ob = U8_ASCII_TOLOWER(*ib);
2128 else
2129 *ob = *ib;
2130 ib++;
2131 ob++;
2132 } else {
2133 *errnum = 0;
2134 state = U8_STATE_START;
2135
2136 j = collect_a_seq(unicode_version, u8s,
2137 &ib, ibtail,
2138 is_it_toupper,
2139 is_it_tolower,
2140 canonical_decomposition,
2141 compatibility_decomposition,
2142 canonical_composition,
2143 errnum, &state);
2144
2145 if (*errnum && do_not_ignore_invalid) {
2146 ret_val = (size_t)-1;
2147 break;
2148 }
2149
2150 if ((obtail - ob) < j) {
2151 *errnum = E2BIG;
2152 ret_val = (size_t)-1;
2153 break;
2154 }
2155
2156 for (i = 0; i < j; i++)
2157 *ob++ = u8s[i];
2158 }
2159 }
2160 }
2161
2162 *inlen = ibtail - ib;
2163 *outlen = obtail - ob;
2164
2165 return (ret_val);
2166 }
2167
2168 EXPORT_SYMBOL(u8_validate);
2169 EXPORT_SYMBOL(u8_strcmp);
2170 EXPORT_SYMBOL(u8_textprep_str);
OpenZFS on Linux and FreeBSD