intl/icu/source/common/unames.cpp@fc2d59ddac77
intl/icu/source/common/unames.cpp
Wed, 31 Dec 2014 07:22:50 +0100
- author
- Michael Schloh von Bennewitz <michael@schloh.com>
- date
- Wed, 31 Dec 2014 07:22:50 +0100
- branch
- TOR_BUG_3246
- changeset 4
- fc2d59ddac77
- permissions
- -rw-r--r--
Correct previous dual key logic pending first delivery installment.
1 /*
2 ******************************************************************************
3 *
4 * Copyright (C) 1999-2013, International Business Machines
5 * Corporation and others. All Rights Reserved.
6 *
7 ******************************************************************************
8 * file name: unames.c
9 * encoding: US-ASCII
10 * tab size: 8 (not used)
11 * indentation:4
12 *
13 * created on: 1999oct04
14 * created by: Markus W. Scherer
15 */
17 #include "unicode/utypes.h"
18 #include "unicode/putil.h"
19 #include "unicode/uchar.h"
20 #include "unicode/udata.h"
21 #include "unicode/utf.h"
22 #include "unicode/utf16.h"
23 #include "uassert.h"
24 #include "ustr_imp.h"
25 #include "umutex.h"
26 #include "cmemory.h"
27 #include "cstring.h"
28 #include "ucln_cmn.h"
29 #include "udataswp.h"
30 #include "uprops.h"
32 U_NAMESPACE_BEGIN
34 /* prototypes ------------------------------------------------------------- */
36 #define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0]))
38 static const char DATA_NAME[] = "unames";
39 static const char DATA_TYPE[] = "icu";
41 #define GROUP_SHIFT 5
42 #define LINES_PER_GROUP (1L<<GROUP_SHIFT)
43 #define GROUP_MASK (LINES_PER_GROUP-1)
45 /*
46 * This struct was replaced by explicitly accessing equivalent
47 * fields from triples of uint16_t.
48 * The Group struct was padded to 8 bytes on compilers for early ARM CPUs,
49 * which broke the assumption that sizeof(Group)==6 and that the ++ operator
50 * would advance by 6 bytes (3 uint16_t).
51 *
52 * We can't just change the data structure because it's loaded from a data file,
53 * and we don't want to make it less compact, so we changed the access code.
54 *
55 * For details see ICU tickets 6331 and 6008.
56 typedef struct {
57 uint16_t groupMSB,
58 offsetHigh, offsetLow; / * avoid padding * /
59 } Group;
60 */
61 enum {
62 GROUP_MSB,
63 GROUP_OFFSET_HIGH,
64 GROUP_OFFSET_LOW,
65 GROUP_LENGTH
66 };
68 /*
69 * Get the 32-bit group offset.
70 * @param group (const uint16_t *) pointer to a Group triple of uint16_t
71 * @return group offset (int32_t)
72 */
73 #define GET_GROUP_OFFSET(group) ((int32_t)(group)[GROUP_OFFSET_HIGH]<<16|(group)[GROUP_OFFSET_LOW])
75 #define NEXT_GROUP(group) ((group)+GROUP_LENGTH)
76 #define PREV_GROUP(group) ((group)-GROUP_LENGTH)
78 typedef struct {
79 uint32_t start, end;
80 uint8_t type, variant;
81 uint16_t size;
82 } AlgorithmicRange;
84 typedef struct {
85 uint32_t tokenStringOffset, groupsOffset, groupStringOffset, algNamesOffset;
86 } UCharNames;
88 /*
89 * Get the groups table from a UCharNames struct.
90 * The groups table consists of one uint16_t groupCount followed by
91 * groupCount groups. Each group is a triple of uint16_t, see GROUP_LENGTH
92 * and the comment for the old struct Group above.
93 *
94 * @param names (const UCharNames *) pointer to the UCharNames indexes
95 * @return (const uint16_t *) pointer to the groups table
96 */
97 #define GET_GROUPS(names) (const uint16_t *)((const char *)names+names->groupsOffset)
99 typedef struct {
100 const char *otherName;
101 UChar32 code;
102 } FindName;
104 #define DO_FIND_NAME NULL
106 static UDataMemory *uCharNamesData=NULL;
107 static UCharNames *uCharNames=NULL;
108 static icu::UInitOnce gCharNamesInitOnce = U_INITONCE_INITIALIZER;
110 /*
111 * Maximum length of character names (regular & 1.0).
112 */
113 static int32_t gMaxNameLength=0;
115 /*
116 * Set of chars used in character names (regular & 1.0).
117 * Chars are platform-dependent (can be EBCDIC).
118 */
119 static uint32_t gNameSet[8]={ 0 };
121 #define U_NONCHARACTER_CODE_POINT U_CHAR_CATEGORY_COUNT
122 #define U_LEAD_SURROGATE U_CHAR_CATEGORY_COUNT + 1
123 #define U_TRAIL_SURROGATE U_CHAR_CATEGORY_COUNT + 2
125 #define U_CHAR_EXTENDED_CATEGORY_COUNT (U_CHAR_CATEGORY_COUNT + 3)
127 static const char * const charCatNames[U_CHAR_EXTENDED_CATEGORY_COUNT] = {
128 "unassigned",
129 "uppercase letter",
130 "lowercase letter",
131 "titlecase letter",
132 "modifier letter",
133 "other letter",
134 "non spacing mark",
135 "enclosing mark",
136 "combining spacing mark",
137 "decimal digit number",
138 "letter number",
139 "other number",
140 "space separator",
141 "line separator",
142 "paragraph separator",
143 "control",
144 "format",
145 "private use area",
146 "surrogate",
147 "dash punctuation",
148 "start punctuation",
149 "end punctuation",
150 "connector punctuation",
151 "other punctuation",
152 "math symbol",
153 "currency symbol",
154 "modifier symbol",
155 "other symbol",
156 "initial punctuation",
157 "final punctuation",
158 "noncharacter",
159 "lead surrogate",
160 "trail surrogate"
161 };
163 /* implementation ----------------------------------------------------------- */
165 static UBool U_CALLCONV unames_cleanup(void)
166 {
167 if(uCharNamesData) {
168 udata_close(uCharNamesData);
169 uCharNamesData = NULL;
170 }
171 if(uCharNames) {
172 uCharNames = NULL;
173 }
174 gCharNamesInitOnce.reset();
175 gMaxNameLength=0;
176 return TRUE;
177 }
179 static UBool U_CALLCONV
180 isAcceptable(void * /*context*/,
181 const char * /*type*/, const char * /*name*/,
182 const UDataInfo *pInfo) {
183 return (UBool)(
184 pInfo->size>=20 &&
185 pInfo->isBigEndian==U_IS_BIG_ENDIAN &&
186 pInfo->charsetFamily==U_CHARSET_FAMILY &&
187 pInfo->dataFormat[0]==0x75 && /* dataFormat="unam" */
188 pInfo->dataFormat[1]==0x6e &&
189 pInfo->dataFormat[2]==0x61 &&
190 pInfo->dataFormat[3]==0x6d &&
191 pInfo->formatVersion[0]==1);
192 }
194 static void U_CALLCONV
195 loadCharNames(UErrorCode &status) {
196 U_ASSERT(uCharNamesData == NULL);
197 U_ASSERT(uCharNames == NULL);
199 uCharNamesData = udata_openChoice(NULL, DATA_TYPE, DATA_NAME, isAcceptable, NULL, &status);
200 if(U_FAILURE(status)) {
201 uCharNamesData = NULL;
202 } else {
203 uCharNames = (UCharNames *)udata_getMemory(uCharNamesData);
204 }
205 ucln_common_registerCleanup(UCLN_COMMON_UNAMES, unames_cleanup);
206 }
209 static UBool
210 isDataLoaded(UErrorCode *pErrorCode) {
211 umtx_initOnce(gCharNamesInitOnce, &loadCharNames, *pErrorCode);
212 return U_SUCCESS(*pErrorCode);
213 }
215 #define WRITE_CHAR(buffer, bufferLength, bufferPos, c) { \
216 if((bufferLength)>0) { \
217 *(buffer)++=c; \
218 --(bufferLength); \
219 } \
220 ++(bufferPos); \
221 }
223 #define U_ISO_COMMENT U_CHAR_NAME_CHOICE_COUNT
225 /*
226 * Important: expandName() and compareName() are almost the same -
227 * apply fixes to both.
228 *
229 * UnicodeData.txt uses ';' as a field separator, so no
230 * field can contain ';' as part of its contents.
231 * In unames.dat, it is marked as token[';']==-1 only if the
232 * semicolon is used in the data file - which is iff we
233 * have Unicode 1.0 names or ISO comments or aliases.
234 * So, it will be token[';']==-1 if we store U1.0 names/ISO comments/aliases
235 * although we know that it will never be part of a name.
236 */
237 static uint16_t
238 expandName(UCharNames *names,
239 const uint8_t *name, uint16_t nameLength, UCharNameChoice nameChoice,
240 char *buffer, uint16_t bufferLength) {
241 uint16_t *tokens=(uint16_t *)names+8;
242 uint16_t token, tokenCount=*tokens++, bufferPos=0;
243 uint8_t *tokenStrings=(uint8_t *)names+names->tokenStringOffset;
244 uint8_t c;
246 if(nameChoice!=U_UNICODE_CHAR_NAME && nameChoice!=U_EXTENDED_CHAR_NAME) {
247 /*
248 * skip the modern name if it is not requested _and_
249 * if the semicolon byte value is a character, not a token number
250 */
251 if((uint8_t)';'>=tokenCount || tokens[(uint8_t)';']==(uint16_t)(-1)) {
252 int fieldIndex= nameChoice==U_ISO_COMMENT ? 2 : nameChoice;
253 do {
254 while(nameLength>0) {
255 --nameLength;
256 if(*name++==';') {
257 break;
258 }
259 }
260 } while(--fieldIndex>0);
261 } else {
262 /*
263 * the semicolon byte value is a token number, therefore
264 * only modern names are stored in unames.dat and there is no
265 * such requested alternate name here
266 */
267 nameLength=0;
268 }
269 }
271 /* write each letter directly, and write a token word per token */
272 while(nameLength>0) {
273 --nameLength;
274 c=*name++;
276 if(c>=tokenCount) {
277 if(c!=';') {
278 /* implicit letter */
279 WRITE_CHAR(buffer, bufferLength, bufferPos, c);
280 } else {
281 /* finished */
282 break;
283 }
284 } else {
285 token=tokens[c];
286 if(token==(uint16_t)(-2)) {
287 /* this is a lead byte for a double-byte token */
288 token=tokens[c<<8|*name++];
289 --nameLength;
290 }
291 if(token==(uint16_t)(-1)) {
292 if(c!=';') {
293 /* explicit letter */
294 WRITE_CHAR(buffer, bufferLength, bufferPos, c);
295 } else {
296 /* stop, but skip the semicolon if we are seeking
297 extended names and there was no 2.0 name but there
298 is a 1.0 name. */
299 if(!bufferPos && nameChoice == U_EXTENDED_CHAR_NAME) {
300 if ((uint8_t)';'>=tokenCount || tokens[(uint8_t)';']==(uint16_t)(-1)) {
301 continue;
302 }
303 }
304 /* finished */
305 break;
306 }
307 } else {
308 /* write token word */
309 uint8_t *tokenString=tokenStrings+token;
310 while((c=*tokenString++)!=0) {
311 WRITE_CHAR(buffer, bufferLength, bufferPos, c);
312 }
313 }
314 }
315 }
317 /* zero-terminate */
318 if(bufferLength>0) {
319 *buffer=0;
320 }
322 return bufferPos;
323 }
325 /*
326 * compareName() is almost the same as expandName() except that it compares
327 * the currently expanded name to an input name.
328 * It returns the match/no match result as soon as possible.
329 */
330 static UBool
331 compareName(UCharNames *names,
332 const uint8_t *name, uint16_t nameLength, UCharNameChoice nameChoice,
333 const char *otherName) {
334 uint16_t *tokens=(uint16_t *)names+8;
335 uint16_t token, tokenCount=*tokens++;
336 uint8_t *tokenStrings=(uint8_t *)names+names->tokenStringOffset;
337 uint8_t c;
338 const char *origOtherName = otherName;
340 if(nameChoice!=U_UNICODE_CHAR_NAME && nameChoice!=U_EXTENDED_CHAR_NAME) {
341 /*
342 * skip the modern name if it is not requested _and_
343 * if the semicolon byte value is a character, not a token number
344 */
345 if((uint8_t)';'>=tokenCount || tokens[(uint8_t)';']==(uint16_t)(-1)) {
346 int fieldIndex= nameChoice==U_ISO_COMMENT ? 2 : nameChoice;
347 do {
348 while(nameLength>0) {
349 --nameLength;
350 if(*name++==';') {
351 break;
352 }
353 }
354 } while(--fieldIndex>0);
355 } else {
356 /*
357 * the semicolon byte value is a token number, therefore
358 * only modern names are stored in unames.dat and there is no
359 * such requested alternate name here
360 */
361 nameLength=0;
362 }
363 }
365 /* compare each letter directly, and compare a token word per token */
366 while(nameLength>0) {
367 --nameLength;
368 c=*name++;
370 if(c>=tokenCount) {
371 if(c!=';') {
372 /* implicit letter */
373 if((char)c!=*otherName++) {
374 return FALSE;
375 }
376 } else {
377 /* finished */
378 break;
379 }
380 } else {
381 token=tokens[c];
382 if(token==(uint16_t)(-2)) {
383 /* this is a lead byte for a double-byte token */
384 token=tokens[c<<8|*name++];
385 --nameLength;
386 }
387 if(token==(uint16_t)(-1)) {
388 if(c!=';') {
389 /* explicit letter */
390 if((char)c!=*otherName++) {
391 return FALSE;
392 }
393 } else {
394 /* stop, but skip the semicolon if we are seeking
395 extended names and there was no 2.0 name but there
396 is a 1.0 name. */
397 if(otherName == origOtherName && nameChoice == U_EXTENDED_CHAR_NAME) {
398 if ((uint8_t)';'>=tokenCount || tokens[(uint8_t)';']==(uint16_t)(-1)) {
399 continue;
400 }
401 }
402 /* finished */
403 break;
404 }
405 } else {
406 /* write token word */
407 uint8_t *tokenString=tokenStrings+token;
408 while((c=*tokenString++)!=0) {
409 if((char)c!=*otherName++) {
410 return FALSE;
411 }
412 }
413 }
414 }
415 }
417 /* complete match? */
418 return (UBool)(*otherName==0);
419 }
421 static uint8_t getCharCat(UChar32 cp) {
422 uint8_t cat;
424 if (U_IS_UNICODE_NONCHAR(cp)) {
425 return U_NONCHARACTER_CODE_POINT;
426 }
428 if ((cat = u_charType(cp)) == U_SURROGATE) {
429 cat = U_IS_LEAD(cp) ? U_LEAD_SURROGATE : U_TRAIL_SURROGATE;
430 }
432 return cat;
433 }
435 static const char *getCharCatName(UChar32 cp) {
436 uint8_t cat = getCharCat(cp);
438 /* Return unknown if the table of names above is not up to
439 date. */
441 if (cat >= LENGTHOF(charCatNames)) {
442 return "unknown";
443 } else {
444 return charCatNames[cat];
445 }
446 }
448 static uint16_t getExtName(uint32_t code, char *buffer, uint16_t bufferLength) {
449 const char *catname = getCharCatName(code);
450 uint16_t length = 0;
452 UChar32 cp;
453 int ndigits, i;
455 WRITE_CHAR(buffer, bufferLength, length, '<');
456 while (catname[length - 1]) {
457 WRITE_CHAR(buffer, bufferLength, length, catname[length - 1]);
458 }
459 WRITE_CHAR(buffer, bufferLength, length, '-');
460 for (cp = code, ndigits = 0; cp; ++ndigits, cp >>= 4)
461 ;
462 if (ndigits < 4)
463 ndigits = 4;
464 for (cp = code, i = ndigits; (cp || i > 0) && bufferLength; cp >>= 4, bufferLength--) {
465 uint8_t v = (uint8_t)(cp & 0xf);
466 buffer[--i] = (v < 10 ? '0' + v : 'A' + v - 10);
467 }
468 buffer += ndigits;
469 length += ndigits;
470 WRITE_CHAR(buffer, bufferLength, length, '>');
472 return length;
473 }
475 /*
476 * getGroup() does a binary search for the group that contains the
477 * Unicode code point "code".
478 * The return value is always a valid Group* that may contain "code"
479 * or else is the highest group before "code".
480 * If the lowest group is after "code", then that one is returned.
481 */
482 static const uint16_t *
483 getGroup(UCharNames *names, uint32_t code) {
484 const uint16_t *groups=GET_GROUPS(names);
485 uint16_t groupMSB=(uint16_t)(code>>GROUP_SHIFT),
486 start=0,
487 limit=*groups++,
488 number;
490 /* binary search for the group of names that contains the one for code */
491 while(start<limit-1) {
492 number=(uint16_t)((start+limit)/2);
493 if(groupMSB<groups[number*GROUP_LENGTH+GROUP_MSB]) {
494 limit=number;
495 } else {
496 start=number;
497 }
498 }
500 /* return this regardless of whether it is an exact match */
501 return groups+start*GROUP_LENGTH;
502 }
504 /*
505 * expandGroupLengths() reads a block of compressed lengths of 32 strings and
506 * expands them into offsets and lengths for each string.
507 * Lengths are stored with a variable-width encoding in consecutive nibbles:
508 * If a nibble<0xc, then it is the length itself (0=empty string).
509 * If a nibble>=0xc, then it forms a length value with the following nibble.
510 * Calculation see below.
511 * The offsets and lengths arrays must be at least 33 (one more) long because
512 * there is no check here at the end if the last nibble is still used.
513 */
514 static const uint8_t *
515 expandGroupLengths(const uint8_t *s,
516 uint16_t offsets[LINES_PER_GROUP+1], uint16_t lengths[LINES_PER_GROUP+1]) {
517 /* read the lengths of the 32 strings in this group and get each string's offset */
518 uint16_t i=0, offset=0, length=0;
519 uint8_t lengthByte;
521 /* all 32 lengths must be read to get the offset of the first group string */
522 while(i<LINES_PER_GROUP) {
523 lengthByte=*s++;
525 /* read even nibble - MSBs of lengthByte */
526 if(length>=12) {
527 /* double-nibble length spread across two bytes */
528 length=(uint16_t)(((length&0x3)<<4|lengthByte>>4)+12);
529 lengthByte&=0xf;
530 } else if((lengthByte /* &0xf0 */)>=0xc0) {
531 /* double-nibble length spread across this one byte */
532 length=(uint16_t)((lengthByte&0x3f)+12);
533 } else {
534 /* single-nibble length in MSBs */
535 length=(uint16_t)(lengthByte>>4);
536 lengthByte&=0xf;
537 }
539 *offsets++=offset;
540 *lengths++=length;
542 offset+=length;
543 ++i;
545 /* read odd nibble - LSBs of lengthByte */
546 if((lengthByte&0xf0)==0) {
547 /* this nibble was not consumed for a double-nibble length above */
548 length=lengthByte;
549 if(length<12) {
550 /* single-nibble length in LSBs */
551 *offsets++=offset;
552 *lengths++=length;
554 offset+=length;
555 ++i;
556 }
557 } else {
558 length=0; /* prevent double-nibble detection in the next iteration */
559 }
560 }
562 /* now, s is at the first group string */
563 return s;
564 }
566 static uint16_t
567 expandGroupName(UCharNames *names, const uint16_t *group,
568 uint16_t lineNumber, UCharNameChoice nameChoice,
569 char *buffer, uint16_t bufferLength) {
570 uint16_t offsets[LINES_PER_GROUP+2], lengths[LINES_PER_GROUP+2];
571 const uint8_t *s=(uint8_t *)names+names->groupStringOffset+GET_GROUP_OFFSET(group);
572 s=expandGroupLengths(s, offsets, lengths);
573 return expandName(names, s+offsets[lineNumber], lengths[lineNumber], nameChoice,
574 buffer, bufferLength);
575 }
577 static uint16_t
578 getName(UCharNames *names, uint32_t code, UCharNameChoice nameChoice,
579 char *buffer, uint16_t bufferLength) {
580 const uint16_t *group=getGroup(names, code);
581 if((uint16_t)(code>>GROUP_SHIFT)==group[GROUP_MSB]) {
582 return expandGroupName(names, group, (uint16_t)(code&GROUP_MASK), nameChoice,
583 buffer, bufferLength);
584 } else {
585 /* group not found */
586 /* zero-terminate */
587 if(bufferLength>0) {
588 *buffer=0;
589 }
590 return 0;
591 }
592 }
594 /*
595 * enumGroupNames() enumerates all the names in a 32-group
596 * and either calls the enumerator function or finds a given input name.
597 */
598 static UBool
599 enumGroupNames(UCharNames *names, const uint16_t *group,
600 UChar32 start, UChar32 end,
601 UEnumCharNamesFn *fn, void *context,
602 UCharNameChoice nameChoice) {
603 uint16_t offsets[LINES_PER_GROUP+2], lengths[LINES_PER_GROUP+2];
604 const uint8_t *s=(uint8_t *)names+names->groupStringOffset+GET_GROUP_OFFSET(group);
606 s=expandGroupLengths(s, offsets, lengths);
607 if(fn!=DO_FIND_NAME) {
608 char buffer[200];
609 uint16_t length;
611 while(start<=end) {
612 length=expandName(names, s+offsets[start&GROUP_MASK], lengths[start&GROUP_MASK], nameChoice, buffer, sizeof(buffer));
613 if (!length && nameChoice == U_EXTENDED_CHAR_NAME) {
614 buffer[length = getExtName(start, buffer, sizeof(buffer))] = 0;
615 }
616 /* here, we assume that the buffer is large enough */
617 if(length>0) {
618 if(!fn(context, start, nameChoice, buffer, length)) {
619 return FALSE;
620 }
621 }
622 ++start;
623 }
624 } else {
625 const char *otherName=((FindName *)context)->otherName;
626 while(start<=end) {
627 if(compareName(names, s+offsets[start&GROUP_MASK], lengths[start&GROUP_MASK], nameChoice, otherName)) {
628 ((FindName *)context)->code=start;
629 return FALSE;
630 }
631 ++start;
632 }
633 }
634 return TRUE;
635 }
637 /*
638 * enumExtNames enumerate extended names.
639 * It only needs to do it if it is called with a real function and not
640 * with the dummy DO_FIND_NAME, because u_charFromName() does a check
641 * for extended names by itself.
642 */
643 static UBool
644 enumExtNames(UChar32 start, UChar32 end,
645 UEnumCharNamesFn *fn, void *context)
646 {
647 if(fn!=DO_FIND_NAME) {
648 char buffer[200];
649 uint16_t length;
651 while(start<=end) {
652 buffer[length = getExtName(start, buffer, sizeof(buffer))] = 0;
653 /* here, we assume that the buffer is large enough */
654 if(length>0) {
655 if(!fn(context, start, U_EXTENDED_CHAR_NAME, buffer, length)) {
656 return FALSE;
657 }
658 }
659 ++start;
660 }
661 }
663 return TRUE;
664 }
666 static UBool
667 enumNames(UCharNames *names,
668 UChar32 start, UChar32 limit,
669 UEnumCharNamesFn *fn, void *context,
670 UCharNameChoice nameChoice) {
671 uint16_t startGroupMSB, endGroupMSB, groupCount;
672 const uint16_t *group, *groupLimit;
674 startGroupMSB=(uint16_t)(start>>GROUP_SHIFT);
675 endGroupMSB=(uint16_t)((limit-1)>>GROUP_SHIFT);
677 /* find the group that contains start, or the highest before it */
678 group=getGroup(names, start);
680 if(startGroupMSB<group[GROUP_MSB] && nameChoice==U_EXTENDED_CHAR_NAME) {
681 /* enumerate synthetic names between start and the group start */
682 UChar32 extLimit=((UChar32)group[GROUP_MSB]<<GROUP_SHIFT);
683 if(extLimit>limit) {
684 extLimit=limit;
685 }
686 if(!enumExtNames(start, extLimit-1, fn, context)) {
687 return FALSE;
688 }
689 start=extLimit;
690 }
692 if(startGroupMSB==endGroupMSB) {
693 if(startGroupMSB==group[GROUP_MSB]) {
694 /* if start and limit-1 are in the same group, then enumerate only in that one */
695 return enumGroupNames(names, group, start, limit-1, fn, context, nameChoice);
696 }
697 } else {
698 const uint16_t *groups=GET_GROUPS(names);
699 groupCount=*groups++;
700 groupLimit=groups+groupCount*GROUP_LENGTH;
702 if(startGroupMSB==group[GROUP_MSB]) {
703 /* enumerate characters in the partial start group */
704 if((start&GROUP_MASK)!=0) {
705 if(!enumGroupNames(names, group,
706 start, ((UChar32)startGroupMSB<<GROUP_SHIFT)+LINES_PER_GROUP-1,
707 fn, context, nameChoice)) {
708 return FALSE;
709 }
710 group=NEXT_GROUP(group); /* continue with the next group */
711 }
712 } else if(startGroupMSB>group[GROUP_MSB]) {
713 /* make sure that we start enumerating with the first group after start */
714 const uint16_t *nextGroup=NEXT_GROUP(group);
715 if (nextGroup < groupLimit && nextGroup[GROUP_MSB] > startGroupMSB && nameChoice == U_EXTENDED_CHAR_NAME) {
716 UChar32 end = nextGroup[GROUP_MSB] << GROUP_SHIFT;
717 if (end > limit) {
718 end = limit;
719 }
720 if (!enumExtNames(start, end - 1, fn, context)) {
721 return FALSE;
722 }
723 }
724 group=nextGroup;
725 }
727 /* enumerate entire groups between the start- and end-groups */
728 while(group<groupLimit && group[GROUP_MSB]<endGroupMSB) {
729 const uint16_t *nextGroup;
730 start=(UChar32)group[GROUP_MSB]<<GROUP_SHIFT;
731 if(!enumGroupNames(names, group, start, start+LINES_PER_GROUP-1, fn, context, nameChoice)) {
732 return FALSE;
733 }
734 nextGroup=NEXT_GROUP(group);
735 if (nextGroup < groupLimit && nextGroup[GROUP_MSB] > group[GROUP_MSB] + 1 && nameChoice == U_EXTENDED_CHAR_NAME) {
736 UChar32 end = nextGroup[GROUP_MSB] << GROUP_SHIFT;
737 if (end > limit) {
738 end = limit;
739 }
740 if (!enumExtNames((group[GROUP_MSB] + 1) << GROUP_SHIFT, end - 1, fn, context)) {
741 return FALSE;
742 }
743 }
744 group=nextGroup;
745 }
747 /* enumerate within the end group (group[GROUP_MSB]==endGroupMSB) */
748 if(group<groupLimit && group[GROUP_MSB]==endGroupMSB) {
749 return enumGroupNames(names, group, (limit-1)&~GROUP_MASK, limit-1, fn, context, nameChoice);
750 } else if (nameChoice == U_EXTENDED_CHAR_NAME && group == groupLimit) {
751 UChar32 next = (PREV_GROUP(group)[GROUP_MSB] + 1) << GROUP_SHIFT;
752 if (next > start) {
753 start = next;
754 }
755 } else {
756 return TRUE;
757 }
758 }
760 /* we have not found a group, which means everything is made of
761 extended names. */
762 if (nameChoice == U_EXTENDED_CHAR_NAME) {
763 if (limit > UCHAR_MAX_VALUE + 1) {
764 limit = UCHAR_MAX_VALUE + 1;
765 }
766 return enumExtNames(start, limit - 1, fn, context);
767 }
769 return TRUE;
770 }
772 static uint16_t
773 writeFactorSuffix(const uint16_t *factors, uint16_t count,
774 const char *s, /* suffix elements */
775 uint32_t code,
776 uint16_t indexes[8], /* output fields from here */
777 const char *elementBases[8], const char *elements[8],
778 char *buffer, uint16_t bufferLength) {
779 uint16_t i, factor, bufferPos=0;
780 char c;
782 /* write elements according to the factors */
784 /*
785 * the factorized elements are determined by modulo arithmetic
786 * with the factors of this algorithm
787 *
788 * note that for fewer operations, count is decremented here
789 */
790 --count;
791 for(i=count; i>0; --i) {
792 factor=factors[i];
793 indexes[i]=(uint16_t)(code%factor);
794 code/=factor;
795 }
796 /*
797 * we don't need to calculate the last modulus because start<=code<=end
798 * guarantees here that code<=factors[0]
799 */
800 indexes[0]=(uint16_t)code;
802 /* write each element */
803 for(;;) {
804 if(elementBases!=NULL) {
805 *elementBases++=s;
806 }
808 /* skip indexes[i] strings */
809 factor=indexes[i];
810 while(factor>0) {
811 while(*s++!=0) {}
812 --factor;
813 }
814 if(elements!=NULL) {
815 *elements++=s;
816 }
818 /* write element */
819 while((c=*s++)!=0) {
820 WRITE_CHAR(buffer, bufferLength, bufferPos, c);
821 }
823 /* we do not need to perform the rest of this loop for i==count - break here */
824 if(i>=count) {
825 break;
826 }
828 /* skip the rest of the strings for this factors[i] */
829 factor=(uint16_t)(factors[i]-indexes[i]-1);
830 while(factor>0) {
831 while(*s++!=0) {}
832 --factor;
833 }
835 ++i;
836 }
838 /* zero-terminate */
839 if(bufferLength>0) {
840 *buffer=0;
841 }
843 return bufferPos;
844 }
846 /*
847 * Important:
848 * Parts of findAlgName() are almost the same as some of getAlgName().
849 * Fixes must be applied to both.
850 */
851 static uint16_t
852 getAlgName(AlgorithmicRange *range, uint32_t code, UCharNameChoice nameChoice,
853 char *buffer, uint16_t bufferLength) {
854 uint16_t bufferPos=0;
856 /* Only the normative character name can be algorithmic. */
857 if(nameChoice!=U_UNICODE_CHAR_NAME && nameChoice!=U_EXTENDED_CHAR_NAME) {
858 /* zero-terminate */
859 if(bufferLength>0) {
860 *buffer=0;
861 }
862 return 0;
863 }
865 switch(range->type) {
866 case 0: {
867 /* name = prefix hex-digits */
868 const char *s=(const char *)(range+1);
869 char c;
871 uint16_t i, count;
873 /* copy prefix */
874 while((c=*s++)!=0) {
875 WRITE_CHAR(buffer, bufferLength, bufferPos, c);
876 }
878 /* write hexadecimal code point value */
879 count=range->variant;
881 /* zero-terminate */
882 if(count<bufferLength) {
883 buffer[count]=0;
884 }
886 for(i=count; i>0;) {
887 if(--i<bufferLength) {
888 c=(char)(code&0xf);
889 if(c<10) {
890 c+='0';
891 } else {
892 c+='A'-10;
893 }
894 buffer[i]=c;
895 }
896 code>>=4;
897 }
899 bufferPos+=count;
900 break;
901 }
902 case 1: {
903 /* name = prefix factorized-elements */
904 uint16_t indexes[8];
905 const uint16_t *factors=(const uint16_t *)(range+1);
906 uint16_t count=range->variant;
907 const char *s=(const char *)(factors+count);
908 char c;
910 /* copy prefix */
911 while((c=*s++)!=0) {
912 WRITE_CHAR(buffer, bufferLength, bufferPos, c);
913 }
915 bufferPos+=writeFactorSuffix(factors, count,
916 s, code-range->start, indexes, NULL, NULL, buffer, bufferLength);
917 break;
918 }
919 default:
920 /* undefined type */
921 /* zero-terminate */
922 if(bufferLength>0) {
923 *buffer=0;
924 }
925 break;
926 }
928 return bufferPos;
929 }
931 /*
932 * Important: enumAlgNames() and findAlgName() are almost the same.
933 * Any fix must be applied to both.
934 */
935 static UBool
936 enumAlgNames(AlgorithmicRange *range,
937 UChar32 start, UChar32 limit,
938 UEnumCharNamesFn *fn, void *context,
939 UCharNameChoice nameChoice) {
940 char buffer[200];
941 uint16_t length;
943 if(nameChoice!=U_UNICODE_CHAR_NAME && nameChoice!=U_EXTENDED_CHAR_NAME) {
944 return TRUE;
945 }
947 switch(range->type) {
948 case 0: {
949 char *s, *end;
950 char c;
952 /* get the full name of the start character */
953 length=getAlgName(range, (uint32_t)start, nameChoice, buffer, sizeof(buffer));
954 if(length<=0) {
955 return TRUE;
956 }
958 /* call the enumerator function with this first character */
959 if(!fn(context, start, nameChoice, buffer, length)) {
960 return FALSE;
961 }
963 /* go to the end of the name; all these names have the same length */
964 end=buffer;
965 while(*end!=0) {
966 ++end;
967 }
969 /* enumerate the rest of the names */
970 while(++start<limit) {
971 /* increment the hexadecimal number on a character-basis */
972 s=end;
973 for (;;) {
974 c=*--s;
975 if(('0'<=c && c<'9') || ('A'<=c && c<'F')) {
976 *s=(char)(c+1);
977 break;
978 } else if(c=='9') {
979 *s='A';
980 break;
981 } else if(c=='F') {
982 *s='0';
983 }
984 }
986 if(!fn(context, start, nameChoice, buffer, length)) {
987 return FALSE;
988 }
989 }
990 break;
991 }
992 case 1: {
993 uint16_t indexes[8];
994 const char *elementBases[8], *elements[8];
995 const uint16_t *factors=(const uint16_t *)(range+1);
996 uint16_t count=range->variant;
997 const char *s=(const char *)(factors+count);
998 char *suffix, *t;
999 uint16_t prefixLength, i, idx;
1001 char c;
1003 /* name = prefix factorized-elements */
1005 /* copy prefix */
1006 suffix=buffer;
1007 prefixLength=0;
1008 while((c=*s++)!=0) {
1009 *suffix++=c;
1010 ++prefixLength;
1011 }
1013 /* append the suffix of the start character */
1014 length=(uint16_t)(prefixLength+writeFactorSuffix(factors, count,
1015 s, (uint32_t)start-range->start,
1016 indexes, elementBases, elements,
1017 suffix, (uint16_t)(sizeof(buffer)-prefixLength)));
1019 /* call the enumerator function with this first character */
1020 if(!fn(context, start, nameChoice, buffer, length)) {
1021 return FALSE;
1022 }
1024 /* enumerate the rest of the names */
1025 while(++start<limit) {
1026 /* increment the indexes in lexical order bound by the factors */
1027 i=count;
1028 for (;;) {
1029 idx=(uint16_t)(indexes[--i]+1);
1030 if(idx<factors[i]) {
1031 /* skip one index and its element string */
1032 indexes[i]=idx;
1033 s=elements[i];
1034 while(*s++!=0) {
1035 }
1036 elements[i]=s;
1037 break;
1038 } else {
1039 /* reset this index to 0 and its element string to the first one */
1040 indexes[i]=0;
1041 elements[i]=elementBases[i];
1042 }
1043 }
1045 /* to make matters a little easier, just append all elements to the suffix */
1046 t=suffix;
1047 length=prefixLength;
1048 for(i=0; i<count; ++i) {
1049 s=elements[i];
1050 while((c=*s++)!=0) {
1051 *t++=c;
1052 ++length;
1053 }
1054 }
1055 /* zero-terminate */
1056 *t=0;
1058 if(!fn(context, start, nameChoice, buffer, length)) {
1059 return FALSE;
1060 }
1061 }
1062 break;
1063 }
1064 default:
1065 /* undefined type */
1066 break;
1067 }
1069 return TRUE;
1070 }
1072 /*
1073 * findAlgName() is almost the same as enumAlgNames() except that it
1074 * returns the code point for a name if it fits into the range.
1075 * It returns 0xffff otherwise.
1076 */
1077 static UChar32
1078 findAlgName(AlgorithmicRange *range, UCharNameChoice nameChoice, const char *otherName) {
1079 UChar32 code;
1081 if(nameChoice!=U_UNICODE_CHAR_NAME && nameChoice!=U_EXTENDED_CHAR_NAME) {
1082 return 0xffff;
1083 }
1085 switch(range->type) {
1086 case 0: {
1087 /* name = prefix hex-digits */
1088 const char *s=(const char *)(range+1);
1089 char c;
1091 uint16_t i, count;
1093 /* compare prefix */
1094 while((c=*s++)!=0) {
1095 if((char)c!=*otherName++) {
1096 return 0xffff;
1097 }
1098 }
1100 /* read hexadecimal code point value */
1101 count=range->variant;
1102 code=0;
1103 for(i=0; i<count; ++i) {
1104 c=*otherName++;
1105 if('0'<=c && c<='9') {
1106 code=(code<<4)|(c-'0');
1107 } else if('A'<=c && c<='F') {
1108 code=(code<<4)|(c-'A'+10);
1109 } else {
1110 return 0xffff;
1111 }
1112 }
1114 /* does it fit into the range? */
1115 if(*otherName==0 && range->start<=(uint32_t)code && (uint32_t)code<=range->end) {
1116 return code;
1117 }
1118 break;
1119 }
1120 case 1: {
1121 char buffer[64];
1122 uint16_t indexes[8];
1123 const char *elementBases[8], *elements[8];
1124 const uint16_t *factors=(const uint16_t *)(range+1);
1125 uint16_t count=range->variant;
1126 const char *s=(const char *)(factors+count), *t;
1127 UChar32 start, limit;
1128 uint16_t i, idx;
1130 char c;
1132 /* name = prefix factorized-elements */
1134 /* compare prefix */
1135 while((c=*s++)!=0) {
1136 if((char)c!=*otherName++) {
1137 return 0xffff;
1138 }
1139 }
1141 start=(UChar32)range->start;
1142 limit=(UChar32)(range->end+1);
1144 /* initialize the suffix elements for enumeration; indexes should all be set to 0 */
1145 writeFactorSuffix(factors, count, s, 0,
1146 indexes, elementBases, elements, buffer, sizeof(buffer));
1148 /* compare the first suffix */
1149 if(0==uprv_strcmp(otherName, buffer)) {
1150 return start;
1151 }
1153 /* enumerate and compare the rest of the suffixes */
1154 while(++start<limit) {
1155 /* increment the indexes in lexical order bound by the factors */
1156 i=count;
1157 for (;;) {
1158 idx=(uint16_t)(indexes[--i]+1);
1159 if(idx<factors[i]) {
1160 /* skip one index and its element string */
1161 indexes[i]=idx;
1162 s=elements[i];
1163 while(*s++!=0) {}
1164 elements[i]=s;
1165 break;
1166 } else {
1167 /* reset this index to 0 and its element string to the first one */
1168 indexes[i]=0;
1169 elements[i]=elementBases[i];
1170 }
1171 }
1173 /* to make matters a little easier, just compare all elements of the suffix */
1174 t=otherName;
1175 for(i=0; i<count; ++i) {
1176 s=elements[i];
1177 while((c=*s++)!=0) {
1178 if(c!=*t++) {
1179 s=""; /* does not match */
1180 i=99;
1181 }
1182 }
1183 }
1184 if(i<99 && *t==0) {
1185 return start;
1186 }
1187 }
1188 break;
1189 }
1190 default:
1191 /* undefined type */
1192 break;
1193 }
1195 return 0xffff;
1196 }
1198 /* sets of name characters, maximum name lengths ---------------------------- */
1200 #define SET_ADD(set, c) ((set)[(uint8_t)c>>5]|=((uint32_t)1<<((uint8_t)c&0x1f)))
1201 #define SET_CONTAINS(set, c) (((set)[(uint8_t)c>>5]&((uint32_t)1<<((uint8_t)c&0x1f)))!=0)
1203 static int32_t
1204 calcStringSetLength(uint32_t set[8], const char *s) {
1205 int32_t length=0;
1206 char c;
1208 while((c=*s++)!=0) {
1209 SET_ADD(set, c);
1210 ++length;
1211 }
1212 return length;
1213 }
1215 static int32_t
1216 calcAlgNameSetsLengths(int32_t maxNameLength) {
1217 AlgorithmicRange *range;
1218 uint32_t *p;
1219 uint32_t rangeCount;
1220 int32_t length;
1222 /* enumerate algorithmic ranges */
1223 p=(uint32_t *)((uint8_t *)uCharNames+uCharNames->algNamesOffset);
1224 rangeCount=*p;
1225 range=(AlgorithmicRange *)(p+1);
1226 while(rangeCount>0) {
1227 switch(range->type) {
1228 case 0:
1229 /* name = prefix + (range->variant times) hex-digits */
1230 /* prefix */
1231 length=calcStringSetLength(gNameSet, (const char *)(range+1))+range->variant;
1232 if(length>maxNameLength) {
1233 maxNameLength=length;
1234 }
1235 break;
1236 case 1: {
1237 /* name = prefix factorized-elements */
1238 const uint16_t *factors=(const uint16_t *)(range+1);
1239 const char *s;
1240 int32_t i, count=range->variant, factor, factorLength, maxFactorLength;
1242 /* prefix length */
1243 s=(const char *)(factors+count);
1244 length=calcStringSetLength(gNameSet, s);
1245 s+=length+1; /* start of factor suffixes */
1247 /* get the set and maximum factor suffix length for each factor */
1248 for(i=0; i<count; ++i) {
1249 maxFactorLength=0;
1250 for(factor=factors[i]; factor>0; --factor) {
1251 factorLength=calcStringSetLength(gNameSet, s);
1252 s+=factorLength+1;
1253 if(factorLength>maxFactorLength) {
1254 maxFactorLength=factorLength;
1255 }
1256 }
1257 length+=maxFactorLength;
1258 }
1260 if(length>maxNameLength) {
1261 maxNameLength=length;
1262 }
1263 break;
1264 }
1265 default:
1266 /* unknown type */
1267 break;
1268 }
1270 range=(AlgorithmicRange *)((uint8_t *)range+range->size);
1271 --rangeCount;
1272 }
1273 return maxNameLength;
1274 }
1276 static int32_t
1277 calcExtNameSetsLengths(int32_t maxNameLength) {
1278 int32_t i, length;
1280 for(i=0; i<LENGTHOF(charCatNames); ++i) {
1281 /*
1282 * for each category, count the length of the category name
1283 * plus 9=
1284 * 2 for <>
1285 * 1 for -
1286 * 6 for most hex digits per code point
1287 */
1288 length=9+calcStringSetLength(gNameSet, charCatNames[i]);
1289 if(length>maxNameLength) {
1290 maxNameLength=length;
1291 }
1292 }
1293 return maxNameLength;
1294 }
1296 static int32_t
1297 calcNameSetLength(const uint16_t *tokens, uint16_t tokenCount, const uint8_t *tokenStrings, int8_t *tokenLengths,
1298 uint32_t set[8],
1299 const uint8_t **pLine, const uint8_t *lineLimit) {
1300 const uint8_t *line=*pLine;
1301 int32_t length=0, tokenLength;
1302 uint16_t c, token;
1304 while(line!=lineLimit && (c=*line++)!=(uint8_t)';') {
1305 if(c>=tokenCount) {
1306 /* implicit letter */
1307 SET_ADD(set, c);
1308 ++length;
1309 } else {
1310 token=tokens[c];
1311 if(token==(uint16_t)(-2)) {
1312 /* this is a lead byte for a double-byte token */
1313 c=c<<8|*line++;
1314 token=tokens[c];
1315 }
1316 if(token==(uint16_t)(-1)) {
1317 /* explicit letter */
1318 SET_ADD(set, c);
1319 ++length;
1320 } else {
1321 /* count token word */
1322 if(tokenLengths!=NULL) {
1323 /* use cached token length */
1324 tokenLength=tokenLengths[c];
1325 if(tokenLength==0) {
1326 tokenLength=calcStringSetLength(set, (const char *)tokenStrings+token);
1327 tokenLengths[c]=(int8_t)tokenLength;
1328 }
1329 } else {
1330 tokenLength=calcStringSetLength(set, (const char *)tokenStrings+token);
1331 }
1332 length+=tokenLength;
1333 }
1334 }
1335 }
1337 *pLine=line;
1338 return length;
1339 }
1341 static void
1342 calcGroupNameSetsLengths(int32_t maxNameLength) {
1343 uint16_t offsets[LINES_PER_GROUP+2], lengths[LINES_PER_GROUP+2];
1345 uint16_t *tokens=(uint16_t *)uCharNames+8;
1346 uint16_t tokenCount=*tokens++;
1347 uint8_t *tokenStrings=(uint8_t *)uCharNames+uCharNames->tokenStringOffset;
1349 int8_t *tokenLengths;
1351 const uint16_t *group;
1352 const uint8_t *s, *line, *lineLimit;
1354 int32_t groupCount, lineNumber, length;
1356 tokenLengths=(int8_t *)uprv_malloc(tokenCount);
1357 if(tokenLengths!=NULL) {
1358 uprv_memset(tokenLengths, 0, tokenCount);
1359 }
1361 group=GET_GROUPS(uCharNames);
1362 groupCount=*group++;
1364 /* enumerate all groups */
1365 while(groupCount>0) {
1366 s=(uint8_t *)uCharNames+uCharNames->groupStringOffset+GET_GROUP_OFFSET(group);
1367 s=expandGroupLengths(s, offsets, lengths);
1369 /* enumerate all lines in each group */
1370 for(lineNumber=0; lineNumber<LINES_PER_GROUP; ++lineNumber) {
1371 line=s+offsets[lineNumber];
1372 length=lengths[lineNumber];
1373 if(length==0) {
1374 continue;
1375 }
1377 lineLimit=line+length;
1379 /* read regular name */
1380 length=calcNameSetLength(tokens, tokenCount, tokenStrings, tokenLengths, gNameSet, &line, lineLimit);
1381 if(length>maxNameLength) {
1382 maxNameLength=length;
1383 }
1384 if(line==lineLimit) {
1385 continue;
1386 }
1388 /* read Unicode 1.0 name */
1389 length=calcNameSetLength(tokens, tokenCount, tokenStrings, tokenLengths, gNameSet, &line, lineLimit);
1390 if(length>maxNameLength) {
1391 maxNameLength=length;
1392 }
1393 if(line==lineLimit) {
1394 continue;
1395 }
1397 /* read ISO comment */
1398 /*length=calcNameSetLength(tokens, tokenCount, tokenStrings, tokenLengths, gISOCommentSet, &line, lineLimit);*/
1399 }
1401 group=NEXT_GROUP(group);
1402 --groupCount;
1403 }
1405 if(tokenLengths!=NULL) {
1406 uprv_free(tokenLengths);
1407 }
1409 /* set gMax... - name length last for threading */
1410 gMaxNameLength=maxNameLength;
1411 }
1413 static UBool
1414 calcNameSetsLengths(UErrorCode *pErrorCode) {
1415 static const char extChars[]="0123456789ABCDEF<>-";
1416 int32_t i, maxNameLength;
1418 if(gMaxNameLength!=0) {
1419 return TRUE;
1420 }
1422 if(!isDataLoaded(pErrorCode)) {
1423 return FALSE;
1424 }
1426 /* set hex digits, used in various names, and <>-, used in extended names */
1427 for(i=0; i<(int32_t)sizeof(extChars)-1; ++i) {
1428 SET_ADD(gNameSet, extChars[i]);
1429 }
1431 /* set sets and lengths from algorithmic names */
1432 maxNameLength=calcAlgNameSetsLengths(0);
1434 /* set sets and lengths from extended names */
1435 maxNameLength=calcExtNameSetsLengths(maxNameLength);
1437 /* set sets and lengths from group names, set global maximum values */
1438 calcGroupNameSetsLengths(maxNameLength);
1440 return TRUE;
1441 }
1443 /* public API --------------------------------------------------------------- */
1445 U_CAPI int32_t U_EXPORT2
1446 u_charName(UChar32 code, UCharNameChoice nameChoice,
1447 char *buffer, int32_t bufferLength,
1448 UErrorCode *pErrorCode) {
1449 AlgorithmicRange *algRange;
1450 uint32_t *p;
1451 uint32_t i;
1452 int32_t length;
1454 /* check the argument values */
1455 if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
1456 return 0;
1457 } else if(nameChoice>=U_CHAR_NAME_CHOICE_COUNT ||
1458 bufferLength<0 || (bufferLength>0 && buffer==NULL)
1459 ) {
1460 *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
1461 return 0;
1462 }
1464 if((uint32_t)code>UCHAR_MAX_VALUE || !isDataLoaded(pErrorCode)) {
1465 return u_terminateChars(buffer, bufferLength, 0, pErrorCode);
1466 }
1468 length=0;
1470 /* try algorithmic names first */
1471 p=(uint32_t *)((uint8_t *)uCharNames+uCharNames->algNamesOffset);
1472 i=*p;
1473 algRange=(AlgorithmicRange *)(p+1);
1474 while(i>0) {
1475 if(algRange->start<=(uint32_t)code && (uint32_t)code<=algRange->end) {
1476 length=getAlgName(algRange, (uint32_t)code, nameChoice, buffer, (uint16_t)bufferLength);
1477 break;
1478 }
1479 algRange=(AlgorithmicRange *)((uint8_t *)algRange+algRange->size);
1480 --i;
1481 }
1483 if(i==0) {
1484 if (nameChoice == U_EXTENDED_CHAR_NAME) {
1485 length = getName(uCharNames, (uint32_t )code, U_EXTENDED_CHAR_NAME, buffer, (uint16_t) bufferLength);
1486 if (!length) {
1487 /* extended character name */
1488 length = getExtName((uint32_t) code, buffer, (uint16_t) bufferLength);
1489 }
1490 } else {
1491 /* normal character name */
1492 length=getName(uCharNames, (uint32_t)code, nameChoice, buffer, (uint16_t)bufferLength);
1493 }
1494 }
1496 return u_terminateChars(buffer, bufferLength, length, pErrorCode);
1497 }
1499 U_CAPI int32_t U_EXPORT2
1500 u_getISOComment(UChar32 /*c*/,
1501 char *dest, int32_t destCapacity,
1502 UErrorCode *pErrorCode) {
1503 /* check the argument values */
1504 if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
1505 return 0;
1506 } else if(destCapacity<0 || (destCapacity>0 && dest==NULL)) {
1507 *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
1508 return 0;
1509 }
1511 return u_terminateChars(dest, destCapacity, 0, pErrorCode);
1512 }
1514 U_CAPI UChar32 U_EXPORT2
1515 u_charFromName(UCharNameChoice nameChoice,
1516 const char *name,
1517 UErrorCode *pErrorCode) {
1518 char upper[120], lower[120];
1519 FindName findName;
1520 AlgorithmicRange *algRange;
1521 uint32_t *p;
1522 uint32_t i;
1523 UChar32 cp = 0;
1524 char c0;
1525 UChar32 error = 0xffff; /* Undefined, but use this for backwards compatibility. */
1527 if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
1528 return error;
1529 }
1531 if(nameChoice>=U_CHAR_NAME_CHOICE_COUNT || name==NULL || *name==0) {
1532 *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
1533 return error;
1534 }
1536 if(!isDataLoaded(pErrorCode)) {
1537 return error;
1538 }
1540 /* construct the uppercase and lowercase of the name first */
1541 for(i=0; i<sizeof(upper); ++i) {
1542 if((c0=*name++)!=0) {
1543 upper[i]=uprv_toupper(c0);
1544 lower[i]=uprv_tolower(c0);
1545 } else {
1546 upper[i]=lower[i]=0;
1547 break;
1548 }
1549 }
1550 if(i==sizeof(upper)) {
1551 /* name too long, there is no such character */
1552 *pErrorCode = U_ILLEGAL_CHAR_FOUND;
1553 return error;
1554 }
1556 /* try extended names first */
1557 if (lower[0] == '<') {
1558 if (nameChoice == U_EXTENDED_CHAR_NAME) {
1559 if (lower[--i] == '>') {
1560 for (--i; lower[i] && lower[i] != '-'; --i) {
1561 }
1563 if (lower[i] == '-') { /* We've got a category. */
1564 uint32_t cIdx;
1566 lower[i] = 0;
1568 for (++i; lower[i] != '>'; ++i) {
1569 if (lower[i] >= '0' && lower[i] <= '9') {
1570 cp = (cp << 4) + lower[i] - '0';
1571 } else if (lower[i] >= 'a' && lower[i] <= 'f') {
1572 cp = (cp << 4) + lower[i] - 'a' + 10;
1573 } else {
1574 *pErrorCode = U_ILLEGAL_CHAR_FOUND;
1575 return error;
1576 }
1577 }
1579 /* Now validate the category name.
1580 We could use a binary search, or a trie, if
1581 we really wanted to. */
1583 for (lower[i] = 0, cIdx = 0; cIdx < LENGTHOF(charCatNames); ++cIdx) {
1585 if (!uprv_strcmp(lower + 1, charCatNames[cIdx])) {
1586 if (getCharCat(cp) == cIdx) {
1587 return cp;
1588 }
1589 break;
1590 }
1591 }
1592 }
1593 }
1594 }
1596 *pErrorCode = U_ILLEGAL_CHAR_FOUND;
1597 return error;
1598 }
1600 /* try algorithmic names now */
1601 p=(uint32_t *)((uint8_t *)uCharNames+uCharNames->algNamesOffset);
1602 i=*p;
1603 algRange=(AlgorithmicRange *)(p+1);
1604 while(i>0) {
1605 if((cp=findAlgName(algRange, nameChoice, upper))!=0xffff) {
1606 return cp;
1607 }
1608 algRange=(AlgorithmicRange *)((uint8_t *)algRange+algRange->size);
1609 --i;
1610 }
1612 /* normal character name */
1613 findName.otherName=upper;
1614 findName.code=error;
1615 enumNames(uCharNames, 0, UCHAR_MAX_VALUE + 1, DO_FIND_NAME, &findName, nameChoice);
1616 if (findName.code == error) {
1617 *pErrorCode = U_ILLEGAL_CHAR_FOUND;
1618 }
1619 return findName.code;
1620 }
1622 U_CAPI void U_EXPORT2
1623 u_enumCharNames(UChar32 start, UChar32 limit,
1624 UEnumCharNamesFn *fn,
1625 void *context,
1626 UCharNameChoice nameChoice,
1627 UErrorCode *pErrorCode) {
1628 AlgorithmicRange *algRange;
1629 uint32_t *p;
1630 uint32_t i;
1632 if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
1633 return;
1634 }
1636 if(nameChoice>=U_CHAR_NAME_CHOICE_COUNT || fn==NULL) {
1637 *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
1638 return;
1639 }
1641 if((uint32_t) limit > UCHAR_MAX_VALUE + 1) {
1642 limit = UCHAR_MAX_VALUE + 1;
1643 }
1644 if((uint32_t)start>=(uint32_t)limit) {
1645 return;
1646 }
1648 if(!isDataLoaded(pErrorCode)) {
1649 return;
1650 }
1652 /* interleave the data-driven ones with the algorithmic ones */
1653 /* iterate over all algorithmic ranges; assume that they are in ascending order */
1654 p=(uint32_t *)((uint8_t *)uCharNames+uCharNames->algNamesOffset);
1655 i=*p;
1656 algRange=(AlgorithmicRange *)(p+1);
1657 while(i>0) {
1658 /* enumerate the character names before the current algorithmic range */
1659 /* here: start<limit */
1660 if((uint32_t)start<algRange->start) {
1661 if((uint32_t)limit<=algRange->start) {
1662 enumNames(uCharNames, start, limit, fn, context, nameChoice);
1663 return;
1664 }
1665 if(!enumNames(uCharNames, start, (UChar32)algRange->start, fn, context, nameChoice)) {
1666 return;
1667 }
1668 start=(UChar32)algRange->start;
1669 }
1670 /* enumerate the character names in the current algorithmic range */
1671 /* here: algRange->start<=start<limit */
1672 if((uint32_t)start<=algRange->end) {
1673 if((uint32_t)limit<=(algRange->end+1)) {
1674 enumAlgNames(algRange, start, limit, fn, context, nameChoice);
1675 return;
1676 }
1677 if(!enumAlgNames(algRange, start, (UChar32)algRange->end+1, fn, context, nameChoice)) {
1678 return;
1679 }
1680 start=(UChar32)algRange->end+1;
1681 }
1682 /* continue to the next algorithmic range (here: start<limit) */
1683 algRange=(AlgorithmicRange *)((uint8_t *)algRange+algRange->size);
1684 --i;
1685 }
1686 /* enumerate the character names after the last algorithmic range */
1687 enumNames(uCharNames, start, limit, fn, context, nameChoice);
1688 }
1690 U_CAPI int32_t U_EXPORT2
1691 uprv_getMaxCharNameLength() {
1692 UErrorCode errorCode=U_ZERO_ERROR;
1693 if(calcNameSetsLengths(&errorCode)) {
1694 return gMaxNameLength;
1695 } else {
1696 return 0;
1697 }
1698 }
1700 /**
1701 * Converts the char set cset into a Unicode set uset.
1702 * @param cset Set of 256 bit flags corresponding to a set of chars.
1703 * @param uset USet to receive characters. Existing contents are deleted.
1704 */
1705 static void
1706 charSetToUSet(uint32_t cset[8], const USetAdder *sa) {
1707 UChar us[256];
1708 char cs[256];
1710 int32_t i, length;
1711 UErrorCode errorCode;
1713 errorCode=U_ZERO_ERROR;
1715 if(!calcNameSetsLengths(&errorCode)) {
1716 return;
1717 }
1719 /* build a char string with all chars that are used in character names */
1720 length=0;
1721 for(i=0; i<256; ++i) {
1722 if(SET_CONTAINS(cset, i)) {
1723 cs[length++]=(char)i;
1724 }
1725 }
1727 /* convert the char string to a UChar string */
1728 u_charsToUChars(cs, us, length);
1730 /* add each UChar to the USet */
1731 for(i=0; i<length; ++i) {
1732 if(us[i]!=0 || cs[i]==0) { /* non-invariant chars become (UChar)0 */
1733 sa->add(sa->set, us[i]);
1734 }
1735 }
1736 }
1738 /**
1739 * Fills set with characters that are used in Unicode character names.
1740 * @param set USet to receive characters.
1741 */
1742 U_CAPI void U_EXPORT2
1743 uprv_getCharNameCharacters(const USetAdder *sa) {
1744 charSetToUSet(gNameSet, sa);
1745 }
1747 /* data swapping ------------------------------------------------------------ */
1749 /*
1750 * The token table contains non-negative entries for token bytes,
1751 * and -1 for bytes that represent themselves in the data file's charset.
1752 * -2 entries are used for lead bytes.
1753 *
1754 * Direct bytes (-1 entries) must be translated from the input charset family
1755 * to the output charset family.
1756 * makeTokenMap() writes a permutation mapping for this.
1757 * Use it once for single-/lead-byte tokens and once more for all trail byte
1758 * tokens. (';' is an unused trail byte marked with -1.)
1759 */
1760 static void
1761 makeTokenMap(const UDataSwapper *ds,
1762 int16_t tokens[], uint16_t tokenCount,
1763 uint8_t map[256],
1764 UErrorCode *pErrorCode) {
1765 UBool usedOutChar[256];
1766 uint16_t i, j;
1767 uint8_t c1, c2;
1769 if(U_FAILURE(*pErrorCode)) {
1770 return;
1771 }
1773 if(ds->inCharset==ds->outCharset) {
1774 /* Same charset family: identity permutation */
1775 for(i=0; i<256; ++i) {
1776 map[i]=(uint8_t)i;
1777 }
1778 } else {
1779 uprv_memset(map, 0, 256);
1780 uprv_memset(usedOutChar, 0, 256);
1782 if(tokenCount>256) {
1783 tokenCount=256;
1784 }
1786 /* set the direct bytes (byte 0 always maps to itself) */
1787 for(i=1; i<tokenCount; ++i) {
1788 if(tokens[i]==-1) {
1789 /* convert the direct byte character */
1790 c1=(uint8_t)i;
1791 ds->swapInvChars(ds, &c1, 1, &c2, pErrorCode);
1792 if(U_FAILURE(*pErrorCode)) {
1793 udata_printError(ds, "unames/makeTokenMap() finds variant character 0x%02x used (input charset family %d)\n",
1794 i, ds->inCharset);
1795 return;
1796 }
1798 /* enter the converted character into the map and mark it used */
1799 map[c1]=c2;
1800 usedOutChar[c2]=TRUE;
1801 }
1802 }
1804 /* set the mappings for the rest of the permutation */
1805 for(i=j=1; i<tokenCount; ++i) {
1806 /* set mappings that were not set for direct bytes */
1807 if(map[i]==0) {
1808 /* set an output byte value that was not used as an output byte above */
1809 while(usedOutChar[j]) {
1810 ++j;
1811 }
1812 map[i]=(uint8_t)j++;
1813 }
1814 }
1816 /*
1817 * leave mappings at tokenCount and above unset if tokenCount<256
1818 * because they won't be used
1819 */
1820 }
1821 }
1823 U_CAPI int32_t U_EXPORT2
1824 uchar_swapNames(const UDataSwapper *ds,
1825 const void *inData, int32_t length, void *outData,
1826 UErrorCode *pErrorCode) {
1827 const UDataInfo *pInfo;
1828 int32_t headerSize;
1830 const uint8_t *inBytes;
1831 uint8_t *outBytes;
1833 uint32_t tokenStringOffset, groupsOffset, groupStringOffset, algNamesOffset,
1834 offset, i, count, stringsCount;
1836 const AlgorithmicRange *inRange;
1837 AlgorithmicRange *outRange;
1839 /* udata_swapDataHeader checks the arguments */
1840 headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode);
1841 if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
1842 return 0;
1843 }
1845 /* check data format and format version */
1846 pInfo=(const UDataInfo *)((const char *)inData+4);
1847 if(!(
1848 pInfo->dataFormat[0]==0x75 && /* dataFormat="unam" */
1849 pInfo->dataFormat[1]==0x6e &&
1850 pInfo->dataFormat[2]==0x61 &&
1851 pInfo->dataFormat[3]==0x6d &&
1852 pInfo->formatVersion[0]==1
1853 )) {
1854 udata_printError(ds, "uchar_swapNames(): data format %02x.%02x.%02x.%02x (format version %02x) is not recognized as unames.icu\n",
1855 pInfo->dataFormat[0], pInfo->dataFormat[1],
1856 pInfo->dataFormat[2], pInfo->dataFormat[3],
1857 pInfo->formatVersion[0]);
1858 *pErrorCode=U_UNSUPPORTED_ERROR;
1859 return 0;
1860 }
1862 inBytes=(const uint8_t *)inData+headerSize;
1863 outBytes=(uint8_t *)outData+headerSize;
1864 if(length<0) {
1865 algNamesOffset=ds->readUInt32(((const uint32_t *)inBytes)[3]);
1866 } else {
1867 length-=headerSize;
1868 if( length<20 ||
1869 (uint32_t)length<(algNamesOffset=ds->readUInt32(((const uint32_t *)inBytes)[3]))
1870 ) {
1871 udata_printError(ds, "uchar_swapNames(): too few bytes (%d after header) for unames.icu\n",
1872 length);
1873 *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
1874 return 0;
1875 }
1876 }
1878 if(length<0) {
1879 /* preflighting: iterate through algorithmic ranges */
1880 offset=algNamesOffset;
1881 count=ds->readUInt32(*((const uint32_t *)(inBytes+offset)));
1882 offset+=4;
1884 for(i=0; i<count; ++i) {
1885 inRange=(const AlgorithmicRange *)(inBytes+offset);
1886 offset+=ds->readUInt16(inRange->size);
1887 }
1888 } else {
1889 /* swap data */
1890 const uint16_t *p;
1891 uint16_t *q, *temp;
1893 int16_t tokens[512];
1894 uint16_t tokenCount;
1896 uint8_t map[256], trailMap[256];
1898 /* copy the data for inaccessible bytes */
1899 if(inBytes!=outBytes) {
1900 uprv_memcpy(outBytes, inBytes, length);
1901 }
1903 /* the initial 4 offsets first */
1904 tokenStringOffset=ds->readUInt32(((const uint32_t *)inBytes)[0]);
1905 groupsOffset=ds->readUInt32(((const uint32_t *)inBytes)[1]);
1906 groupStringOffset=ds->readUInt32(((const uint32_t *)inBytes)[2]);
1907 ds->swapArray32(ds, inBytes, 16, outBytes, pErrorCode);
1909 /*
1910 * now the tokens table
1911 * it needs to be permutated along with the compressed name strings
1912 */
1913 p=(const uint16_t *)(inBytes+16);
1914 q=(uint16_t *)(outBytes+16);
1916 /* read and swap the tokenCount */
1917 tokenCount=ds->readUInt16(*p);
1918 ds->swapArray16(ds, p, 2, q, pErrorCode);
1919 ++p;
1920 ++q;
1922 /* read the first 512 tokens and make the token maps */
1923 if(tokenCount<=512) {
1924 count=tokenCount;
1925 } else {
1926 count=512;
1927 }
1928 for(i=0; i<count; ++i) {
1929 tokens[i]=udata_readInt16(ds, p[i]);
1930 }
1931 for(; i<512; ++i) {
1932 tokens[i]=0; /* fill the rest of the tokens array if tokenCount<512 */
1933 }
1934 makeTokenMap(ds, tokens, tokenCount, map, pErrorCode);
1935 makeTokenMap(ds, tokens+256, (uint16_t)(tokenCount>256 ? tokenCount-256 : 0), trailMap, pErrorCode);
1936 if(U_FAILURE(*pErrorCode)) {
1937 return 0;
1938 }
1940 /*
1941 * swap and permutate the tokens
1942 * go through a temporary array to support in-place swapping
1943 */
1944 temp=(uint16_t *)uprv_malloc(tokenCount*2);
1945 if(temp==NULL) {
1946 udata_printError(ds, "out of memory swapping %u unames.icu tokens\n",
1947 tokenCount);
1948 *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
1949 return 0;
1950 }
1952 /* swap and permutate single-/lead-byte tokens */
1953 for(i=0; i<tokenCount && i<256; ++i) {
1954 ds->swapArray16(ds, p+i, 2, temp+map[i], pErrorCode);
1955 }
1957 /* swap and permutate trail-byte tokens */
1958 for(; i<tokenCount; ++i) {
1959 ds->swapArray16(ds, p+i, 2, temp+(i&0xffffff00)+trailMap[i&0xff], pErrorCode);
1960 }
1962 /* copy the result into the output and free the temporary array */
1963 uprv_memcpy(q, temp, tokenCount*2);
1964 uprv_free(temp);
1966 /*
1967 * swap the token strings but not a possible padding byte after
1968 * the terminating NUL of the last string
1969 */
1970 udata_swapInvStringBlock(ds, inBytes+tokenStringOffset, (int32_t)(groupsOffset-tokenStringOffset),
1971 outBytes+tokenStringOffset, pErrorCode);
1972 if(U_FAILURE(*pErrorCode)) {
1973 udata_printError(ds, "uchar_swapNames(token strings) failed\n");
1974 return 0;
1975 }
1977 /* swap the group table */
1978 count=ds->readUInt16(*((const uint16_t *)(inBytes+groupsOffset)));
1979 ds->swapArray16(ds, inBytes+groupsOffset, (int32_t)((1+count*3)*2),
1980 outBytes+groupsOffset, pErrorCode);
1982 /*
1983 * swap the group strings
1984 * swap the string bytes but not the nibble-encoded string lengths
1985 */
1986 if(ds->inCharset!=ds->outCharset) {
1987 uint16_t offsets[LINES_PER_GROUP+1], lengths[LINES_PER_GROUP+1];
1989 const uint8_t *inStrings, *nextInStrings;
1990 uint8_t *outStrings;
1992 uint8_t c;
1994 inStrings=inBytes+groupStringOffset;
1995 outStrings=outBytes+groupStringOffset;
1997 stringsCount=algNamesOffset-groupStringOffset;
1999 /* iterate through string groups until only a few padding bytes are left */
2000 while(stringsCount>32) {
2001 nextInStrings=expandGroupLengths(inStrings, offsets, lengths);
2003 /* move past the length bytes */
2004 stringsCount-=(uint32_t)(nextInStrings-inStrings);
2005 outStrings+=nextInStrings-inStrings;
2006 inStrings=nextInStrings;
2008 count=offsets[31]+lengths[31]; /* total number of string bytes in this group */
2009 stringsCount-=count;
2011 /* swap the string bytes using map[] and trailMap[] */
2012 while(count>0) {
2013 c=*inStrings++;
2014 *outStrings++=map[c];
2015 if(tokens[c]!=-2) {
2016 --count;
2017 } else {
2018 /* token lead byte: swap the trail byte, too */
2019 *outStrings++=trailMap[*inStrings++];
2020 count-=2;
2021 }
2022 }
2023 }
2024 }
2026 /* swap the algorithmic ranges */
2027 offset=algNamesOffset;
2028 count=ds->readUInt32(*((const uint32_t *)(inBytes+offset)));
2029 ds->swapArray32(ds, inBytes+offset, 4, outBytes+offset, pErrorCode);
2030 offset+=4;
2032 for(i=0; i<count; ++i) {
2033 if(offset>(uint32_t)length) {
2034 udata_printError(ds, "uchar_swapNames(): too few bytes (%d after header) for unames.icu algorithmic range %u\n",
2035 length, i);
2036 *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
2037 return 0;
2038 }
2040 inRange=(const AlgorithmicRange *)(inBytes+offset);
2041 outRange=(AlgorithmicRange *)(outBytes+offset);
2042 offset+=ds->readUInt16(inRange->size);
2044 ds->swapArray32(ds, inRange, 8, outRange, pErrorCode);
2045 ds->swapArray16(ds, &inRange->size, 2, &outRange->size, pErrorCode);
2046 switch(inRange->type) {
2047 case 0:
2048 /* swap prefix string */
2049 ds->swapInvChars(ds, inRange+1, (int32_t)uprv_strlen((const char *)(inRange+1)),
2050 outRange+1, pErrorCode);
2051 if(U_FAILURE(*pErrorCode)) {
2052 udata_printError(ds, "uchar_swapNames(prefix string of algorithmic range %u) failed\n",
2053 i);
2054 return 0;
2055 }
2056 break;
2057 case 1:
2058 {
2059 /* swap factors and the prefix and factor strings */
2060 uint32_t factorsCount;
2062 factorsCount=inRange->variant;
2063 p=(const uint16_t *)(inRange+1);
2064 q=(uint16_t *)(outRange+1);
2065 ds->swapArray16(ds, p, (int32_t)(factorsCount*2), q, pErrorCode);
2067 /* swap the strings, up to the last terminating NUL */
2068 p+=factorsCount;
2069 q+=factorsCount;
2070 stringsCount=(uint32_t)((inBytes+offset)-(const uint8_t *)p);
2071 while(stringsCount>0 && ((const uint8_t *)p)[stringsCount-1]!=0) {
2072 --stringsCount;
2073 }
2074 ds->swapInvChars(ds, p, (int32_t)stringsCount, q, pErrorCode);
2075 }
2076 break;
2077 default:
2078 udata_printError(ds, "uchar_swapNames(): unknown type %u of algorithmic range %u\n",
2079 inRange->type, i);
2080 *pErrorCode=U_UNSUPPORTED_ERROR;
2081 return 0;
2082 }
2083 }
2084 }
2086 return headerSize+(int32_t)offset;
2087 }
2089 U_NAMESPACE_END
2091 /*
2092 * Hey, Emacs, please set the following:
2093 *
2094 * Local Variables:
2095 * indent-tabs-mode: nil
2096 * End:
2097 *
2098 */
