2 * Copyright (C) 2002 Manuel Novoa III
3 * Copyright (C) 2000-2005 Erik Andersen <andersen@uclibc.org>
5 * Licensed under the LGPL v2.1, see the file COPYING.LIB in this tarball.
9 * Initial test implementation of strcoll, strxfrm, wcscoll, and wcsxfrm.
10 * The code needs to be cleaned up a good bit, but I'd like to see people
22 #ifdef __UCLIBC_HAS_LOCALE__
23 #if defined(L_strxfrm) || defined(L_strxfrm_l) || defined(L_wcsxfrm) || defined(L_wcsxfrm_l)
27 #error WANT_WIDE should be defined for L_strxfrm
30 #error L_wcsxfrm already defined for L_strxfrm
32 #endif /* L_strxfrm */
34 #if defined(L_strxfrm) || defined(L_strxfrm_l)
36 #define wcscoll strcoll
37 #define wcscoll_l strcoll_l
38 #define wcsxfrm strxfrm
39 #define wcsxfrm_l strxfrm_l
49 #endif /* defined(L_strxfrm) || defined(L_strxfrm_l) */
51 #if defined(__UCLIBC_HAS_XLOCALE__) && !defined(__UCLIBC_DO_XLOCALE)
54 int wcscoll (const Wchar *s0, const Wchar *s1)
56 return wcscoll_l(s0, s1, __UCLIBC_CURLOCALE );
58 libc_hidden_def(wcscoll)
61 size_t wcsxfrm(Wchar *__restrict ws1, const Wchar *__restrict ws2, size_t n)
63 return wcsxfrm_l(ws1, ws2, n, __UCLIBC_CURLOCALE );
66 #else /* defined(__UCLIBC_HAS_XLOCALE__) && !defined(__UCLIBC_DO_XLOCALE) */
70 #define CUR_COLLATE (&__UCLIBC_CURLOCALE->collate)
72 #define CUR_COLLATE (& __LOCALE_PTR->collate)
79 const Wchar *eob; /* end of backward */
82 __uwchar_t ui_weight; /* undefined or invalid */
87 /* should be wchar_t. if wchar < 0 do EILSEQ? */
89 __uwchar_t ci_pending[MAX_PENDING]; /* nul-terminated */
92 char *bbe; /* end of back_buf (actual last... not 1 past end) */
93 char *bp; /* ptr into backbuf, NULL if not in backward mode */
101 #define WEIGHT_MASK 0x3fffU
102 #define RULE_MASK 0xc000U
104 #define RULE_FORWARD (1 << 14)
105 #define RULE_POSITION (1 << 15)
107 #define UI_IDX (WEIGHT_MASK-6)
108 #define POSIT_IDX (WEIGHT_MASK-5)
109 #define RANGE_IDX (WEIGHT_MASK-4)
110 #define UNDEF_IDX (WEIGHT_MASK-3)
111 #define INVAL_IDX (WEIGHT_MASK-2)
112 #define DITTO_IDX (WEIGHT_MASK-1)
117 #define TRACE(X) printf X
119 #define TRACE(X) ((void)0)
122 static int lookup(wchar_t wc __LOCALE_PARAM )
124 unsigned int sc, n, i0, i1;
126 if (((__uwchar_t) wc) > 0xffffU) {
130 sc = wc & CUR_COLLATE->ti_mask;
131 wc >>= CUR_COLLATE->ti_shift;
132 n = wc & CUR_COLLATE->ii_mask;
133 wc >>= CUR_COLLATE->ii_shift;
135 i0 = CUR_COLLATE->wcs2colidt_tbl[wc];
136 i0 <<= CUR_COLLATE->ii_shift;
137 i1 = CUR_COLLATE->wcs2colidt_tbl[CUR_COLLATE->ii_len + i0 + n];
138 i1 <<= CUR_COLLATE->ti_shift;
139 return CUR_COLLATE->wcs2colidt_tbl[CUR_COLLATE->ii_len + CUR_COLLATE->ti_len + i1 + sc];
143 static void init_col_state(col_state_t *cs, const Wchar *wcs)
145 memset(cs, 0, sizeof(col_state_t));
147 cs->bp = cs->back_buf = cs->ibb;
149 cs->bbe = cs->back_buf + (cs->bb_size -1);
152 static void next_weight(col_state_t *cs, int pass __LOCALE_PARAM )
154 int r, w, ru, ri, popping_backup_stack;
160 #else /* WANT_WIDE */
165 #endif /* WANT_WIDE */
171 TRACE(("ru_pushed = %d\n", ru));
176 #ifdef __UCLIBC_MJN3_ONLY__
177 #warning should we walk pendings backwards?
179 if (cs->cip) { /* possible pending weight */
180 if ((r = *(cs->cip++)) == 0) {
184 cs->weightidx = r & WEIGHT_MASK;
185 assert(cs->weightidx);
186 /* assert(cs->weightidx != WEIGHT_MASK); */
187 } else { /* get the next collation item from the string */
188 TRACE(("clearing popping flag\n"));
189 popping_backup_stack = 0;
192 /* keep first pos as 0 for a sentinal */
193 if (*cs->bp) { /* pending backward chars */
195 popping_backup_stack = 1;
196 TRACE(("setting popping flag\n"));
198 if (*cs->bp > 0) { /* singles pending */
200 if ((*cs->bp -= 1) == 0) {
203 } else { /* last was a multi */
207 } else if (!*cs->s) { /* not in backward mode and end of string */
217 cs->colitem = r = lookup(*cs->s __LOCALE_ARG );
218 #else /* WANT_WIDE */
219 n = n0 = __locale_mbrtowc_l(&WC, cs->s, __LOCALE_PTR);
225 cs->colitem = r = lookup(WC __LOCALE_ARG );
226 #endif /* WANT_WIDE */
228 TRACE((" r=%d WC=%#lx\n", r, (unsigned long)(WC)));
230 if (r > CUR_COLLATE->max_col_index) { /* starting char for one or more sequences */
231 p = CUR_COLLATE->multistart_tbl;
232 p += p[r-CUR_COLLATE->max_col_index -1];
237 if (!*p) { /* found it */
239 TRACE((" found multi %d\n", n));
243 /* the lookup check here is safe since we're assured that *p is a valid colidx */
244 if (!cs->s[n] || (lookup(cs->s[n] __LOCALE_ARG ) != *p)) {
250 #else /* WANT_WIDE */
252 nx = __locale_mbrtowc_l(&WC, cs->s + n, __LOCALE_PTR);
259 if (!cs->s[n] || (lookup(WC __LOCALE_ARG ) != *p)) {
264 n += nx; /* Only gets here if cs->s[n] != 0, so nx is set. */
265 #endif /* WANT_WIDE */
268 } else if (r == 0) { /* illegal, undefined, or part of a range */
269 if ((CUR_COLLATE->range_count)
270 #ifdef __UCLIBC_MJN3_ONLY__
271 #warning .. need to introduce range as a collating item?
273 && (((__uwchar_t)(WC - CUR_COLLATE->range_low)) <= CUR_COLLATE->range_count)
274 ) { /* part of a range */
275 /* Note: cs->colitem = 0 already. */
276 TRACE((" found range\n"));
277 ru = CUR_COLLATE->ruletable[CUR_COLLATE->range_rule_offset*CUR_COLLATE->MAX_WEIGHTS + pass];
278 assert((ru & WEIGHT_MASK) != DITTO_IDX);
279 if ((ru & WEIGHT_MASK) == WEIGHT_MASK) {
280 ru = (ru & RULE_MASK) | RANGE_IDX;
281 cs->weight = CUR_COLLATE->range_base_weight + (WC - CUR_COLLATE->range_low);
284 } else if (((__uwchar_t)(WC)) <= 0x7fffffffUL) { /* legal but undefined */
286 /* Note: cs->colitem = 0 already. */
287 ri = CUR_COLLATE->undefined_idx;
288 assert(ri != 0); /* implicit undefined isn't supported */
290 TRACE((" found explicit UNDEFINED\n"));
291 #ifdef __UCLIBC_MJN3_ONLY__
292 #warning right now single weight locales do not support ..
294 if (CUR_COLLATE->num_weights == 1) {
295 TRACE((" single weight UNDEFINED\n"));
296 cs->weightidx = RANGE_IDX;
302 ri = CUR_COLLATE->index2ruleidx[ri - 1];
303 ru = CUR_COLLATE->ruletable[ri * CUR_COLLATE->MAX_WEIGHTS + pass];
304 assert((ru & WEIGHT_MASK) != WEIGHT_MASK); /* TODO: handle ".." */
305 if ((ru & WEIGHT_MASK) == DITTO_IDX) {
306 cs->colitem = CUR_COLLATE->undefined_idx;
309 } else { /* illegal */
310 TRACE((" found illegal\n"));
312 /* We put all illegals in the same equiv class with maximal weight,
313 * and ignore them after the first pass. */
318 ru = (RULE_FORWARD | RANGE_IDX);
319 cs->weight = 0xffffU;
322 } else if (CUR_COLLATE->num_weights == 1) {
323 TRACE((" single weight\n"));
324 cs->weightidx = RANGE_IDX;
325 cs->weight = cs->colitem;
329 TRACE((" normal\n"));
332 /* if we get here, it is a normal char either singlely weighted, undefined, or in a range */
334 ri = CUR_COLLATE->index2ruleidx[cs->colitem - 1];
335 TRACE((" ri=%d ", ri));
336 #ifdef __UCLIBC_MJN3_ONLY__
337 #warning make sure this is correct
340 TRACE(("NOT IN THIS LOCALE\n"));
343 ru = CUR_COLLATE->ruletable[ri * CUR_COLLATE->MAX_WEIGHTS + pass];
347 #ifdef __UCLIBC_MJN3_ONLY__
348 #warning ignoreables probably should not interrupt backwards processing, but this is wrong
350 /* if (!(ru & WEIGHT_MASK)) { */
351 /* TRACE(("IGNORE\n")); */
357 TRACE((" rule = %#x weight = %#x popping = %d s = %p eob = %p\n",
358 ru & RULE_MASK, ru & WEIGHT_MASK, popping_backup_stack,
360 /* now we need to check if we're going backwards... */
362 if (!popping_backup_stack) {
363 if (!(ru & RULE_MASK)) { /* backward */
364 TRACE(("backwards\n"));
365 assert(cs->bp <= cs->bbe);
366 if (cs->bp == cs->bbe) {
367 if (cs->back_buf == cs->ibb) { /* was using internal buffer */
368 cs->bp = malloc(cs->bb_size + 128);
371 #ifdef __UCLIBC_MJN3_ONLY__
372 #warning what to do here?
377 memcpy(cs->bp, cs->back_buf, cs->bb_size);
380 cs->bp = realloc(cs->back_buf, cs->bb_size + 128);
383 #ifdef __UCLIBC_MJN3_ONLY__
384 #warning what to do here?
391 cs->bbe = cs->bp + (cs->bbe - cs->back_buf);
392 cs->back_buf = cs->bp;
396 if (n==1) { /* single char */
397 if (*cs->bp && (((unsigned char)(*cs->bp)) < CHAR_MAX)) {
398 *cs->bp += 1; /* increment last single's count */
399 } else { /* last was a multi, or just starting */
401 cs->bp = cs->back_buf;
403 assert(cs->bp < cs->bbe);
408 } else { /* multichar */
410 assert(cs->bp < cs->bbe);
417 /* end-of-string so start popping */
419 TRACE(("popping\n"));
421 } else if (*cs->bp) { /* was going backward but this element isn't */
422 /* discard current and use previous backward element */
425 TRACE(("popping\n"));
427 } else { /* was and still going forward */
428 TRACE(("forwards\n"));
429 if ((ru & (RULE_POSITION|WEIGHT_MASK)) > RULE_POSITION) {
430 assert(ru & WEIGHT_MASK);
432 cs->weight = cs->position;
433 #ifdef __UCLIBC_MJN3_ONLY__
436 cs->position = 0; /* reset to reduce size for strcoll? */
438 cs->weightidx = RANGE_IDX;
442 } else { /* popping backwards stack */
443 TRACE(("popping (continued)\n"));
452 cs->weightidx = ru & WEIGHT_MASK;
453 cs->rule = ru & RULE_MASK;
456 #ifdef __UCLIBC_MJN3_ONLY__
457 #warning for pending we only want the weight... _not_ the rule
459 if (!cs->weightidx) { /* ignore */
464 assert(cs->weightidx);
467 if (((unsigned int)(cs->weightidx - UI_IDX)) <= (INVAL_IDX-UI_IDX)) {
468 if (cs->weightidx == UI_IDX) {
469 cs->weight = cs->ui_weight;
474 assert(cs->weightidx != WEIGHT_MASK);
475 if (cs->weightidx == DITTO_IDX) { /* want the weight of the current collating item */
476 TRACE(("doing ditto\n"));
477 w = CUR_COLLATE->index2weight[cs->colitem -1];
478 } else if (cs->weightidx <= CUR_COLLATE->max_col_index) { /* normal */
479 TRACE(("doing normal\n"));
480 w = CUR_COLLATE->index2weight[cs->weightidx -1];
481 } else { /* a string */
482 TRACE(("doing string\n"));
483 assert(!(cs->weightidx & RULE_MASK));
484 /* note: iso14561 allows null string here */
485 p = CUR_COLLATE->weightstr + (cs->weightidx - (CUR_COLLATE->max_col_index + 2));
486 if (*p & WEIGHT_MASK) {
489 assert(r < MAX_PENDING);
490 cs->ci_pending[r++] = *p++;
491 } while (*p & WEIGHT_MASK);
492 cs->cip = cs->ci_pending;
502 int __XL_NPP(wcscoll) (const Wchar *s0, const Wchar *s1 __LOCALE_PARAM )
507 if (!CUR_COLLATE->num_weights) { /* C locale */
509 return wcscmp(s0, s1);
511 return strcmp(s0, s1);
516 do { /* loop through the weights levels */
517 init_col_state(ws, s0);
518 init_col_state(ws+1, s1);
519 do { /* loop through the strings */
520 /* for each string, get the next weight */
521 next_weight(ws, pass __LOCALE_ARG );
522 next_weight(ws+1, pass __LOCALE_ARG );
523 TRACE(("w0=%lu w1=%lu\n",
524 (unsigned long) ws[0].weight,
525 (unsigned long) ws[1].weight));
527 if (ws[0].weight != ws[1].weight) {
528 return ws[0].weight - ws[1].weight;
530 } while (ws[0].weight);
531 } while (++pass < CUR_COLLATE->num_weights);
535 libc_hidden_def(__XL_NPP(wcscoll))
539 size_t __XL_NPP(wcsxfrm)(wchar_t *__restrict ws1, const wchar_t *__restrict ws2,
540 size_t n __LOCALE_PARAM )
546 if (!CUR_COLLATE->num_weights) { /* C locale */
547 return __wcslcpy(ws1, ws2, n);
550 #ifdef __UCLIBC_MJN3_ONLY__
551 #warning handle empty string as a special case
555 do { /* loop through the weights levels */
556 init_col_state(&cs, ws2);
557 do { /* loop through the string */
558 next_weight(&cs, pass __LOCALE_ARG );
559 TRACE(("weight=%lu (%#lx)\n", (unsigned long) cs.weight, (unsigned long) cs.weight));
561 ws1[count] = cs.weight +1;
564 TRACE(("--------------------------------------------\n"));
566 if (count <= n) { /* overwrite the trailing 0 end-of-pass marker */
569 TRACE(("-------------------- pass %d --------------------\n", pass));
570 } while (++pass < CUR_COLLATE->num_weights);
571 if (count <= n) { /* oops... change it back */
576 #if defined L_strxfrm_l || defined L_wcsxfrm_l
577 libc_hidden_def(__XL_NPP(wcsxfrm))
580 #else /* WANT_WIDE */
582 static const unsigned long bound[] = {
590 static unsigned char first[] = {
591 0x0, 0xc0, 0xe0, 0xf0, 0xf8, 0xfc
594 /* Use an extension of UTF-8 to store a 32 bit val in max 6 bytes. */
596 static size_t store(unsigned char *s, size_t count, size_t n, __uwchar_t weight)
602 if (weight < bound[i]) {
605 } while (++i < sizeof(bound)/sizeof(bound[0]));
612 s[i] = 0x80 | (weight & 0x3f);
622 size_t __XL_NPP(strxfrm)(char *__restrict ws1, const char *__restrict ws2, size_t n
629 if (!CUR_COLLATE->num_weights) { /* C locale */
630 return strlcpy(ws1, ws2, n);
633 #ifdef __UCLIBC_MJN3_ONLY__
634 #warning handle empty string as a special case
637 inc = count = pass = 0;
638 do { /* loop through the weights levels */
639 init_col_state(&cs, ws2);
640 do { /* loop through the string */
641 next_weight(&cs, pass __LOCALE_ARG );
642 TRACE(("weight=%lu (%#lx)\n", (unsigned long) cs.weight, (unsigned long) cs.weight));
643 inc = store((unsigned char *)ws1, count, n, cs.weight + 1);
645 TRACE(("--------------------------------------------\n"));
647 /* overwrite the trailing 0 end-of-pass marker */
652 TRACE(("-------------------- pass %d --------------------\n", pass));
653 } while (++pass < CUR_COLLATE->num_weights);
654 if (count <= n) { /* oops... change it back */
660 libc_hidden_def(__XL_NPP(strxfrm))
663 #endif /* WANT_WIDE */
665 #endif /* defined(__UCLIBC_HAS_XLOCALE__) && !defined(__UCLIBC_DO_XLOCALE) */
667 #endif /* defined(L_strxfrm) || defined(L_strxfrm_l) || defined(L_wcsxfrm) || defined(L_wcsxfrm_l) */
669 #endif /* __UCLIBC_HAS_LOCALE__ */
670 /**********************************************************************/