1 /*******************************************************************
3 The early netatalk 2.x was based on UCS-2.
4 UCS-2 don't support chars above U+10000.
5 Recent netatalk is based on UTF-16.
6 UTF-16 can support chars above U+10000, using Surrogate Pair.
7 However, Surrogate Pair is complex, dirty, filthy and disagreeable.
8 There might still be latent bugs...
9 ********************************************************************/
13 #endif /* HAVE_CONFIG_H */
18 #include <sys/param.h>
20 #include <atalk/logger.h>
22 #include <arpa/inet.h>
24 #include <atalk/unicode.h>
25 #include "precompose.h"
26 #include "byteorder.h"
28 /*******************************************************************
29 Convert a string to lower case.
30 return True if any char is converted
31 ********************************************************************/
32 /* surrogate pair support */
34 int strlower_w(ucs2_t *s)
39 if ((0xD800 <= *s) && (*s < 0xDC00)) {
40 if ((0xDC00 <= s[1]) && (s[1] < 0xE000)) {
41 u_int32_t s_sp = (u_int32_t)*s << 16 | (u_int32_t)s[1];
42 u_int32_t v_sp = tolower_sp(s_sp);
51 ucs2_t v = tolower_w(*s);
62 /*******************************************************************
63 Convert a string to upper case.
64 return True if any char is converted
65 ********************************************************************/
66 /* surrogate pair support */
68 int strupper_w(ucs2_t *s)
73 if ((0xD800 <= *s) && (*s < 0xDC00)) {
74 if ((0xDC00 <= s[1]) && (s[1] < 0xE000)) {
75 u_int32_t s_sp = (u_int32_t)*s << 16 | (u_int32_t)s[1];
76 u_int32_t v_sp = toupper_sp(s_sp);
85 ucs2_t v = toupper_w(*s);
96 /*******************************************************************
98 determine if a character is lowercase
99 ********************************************************************/
100 /* These functions are not used. */
102 int islower_w(ucs2_t c)
104 return ( c == tolower_w(c));
107 int islower_sp(u_int32_t c_sp)
109 return ( c_sp == tolower_sp(c_sp));
112 /*******************************************************************
114 determine if a character is uppercase
115 ********************************************************************/
116 /* These functions are not used. */
118 int isupper_w(ucs2_t c)
120 return ( c == toupper_w(c));
123 int isupper_sp(u_int32_t c_sp)
125 return ( c_sp == toupper_sp(c_sp));
128 /*******************************************************************
130 Count the number of characters in a UTF-16 string.
131 ********************************************************************/
132 /* NOTE: one surrogate pair is two characters. */
134 size_t strlen_w(const ucs2_t *src)
138 for(len = 0; *src++; len++) ;
143 /*******************************************************************
145 Count up to max number of characters in a UTF-16 string.
146 ********************************************************************/
147 /* NOTE: one surrogate pair is two characters. */
149 size_t strnlen_w(const ucs2_t *src, size_t max)
153 for(len = 0; *src++ && (len < max); len++) ;
158 /*******************************************************************
160 ********************************************************************/
161 /* NOTE: hi and lo of surrogate pair are separately processed. */
163 ucs2_t *strchr_w(const ucs2_t *s, ucs2_t c)
166 if (c == *s) return (ucs2_t *)s;
169 if (c == *s) return (ucs2_t *)s;
174 /*******************************************************************
175 wide & sp strcasechr()
176 ********************************************************************/
177 /* NOTE: separately process BMP and surrogate pair */
179 ucs2_t *strcasechr_w(const ucs2_t *s, ucs2_t c)
182 /* LOG(log_debug, logtype_default, "Comparing %X to %X (%X - %X)", c, *s, toupper_w(c), toupper_w(*s));*/
183 if (toupper_w(c) == toupper_w(*s)) return (ucs2_t *)s;
186 if (c == *s) return (ucs2_t *)s;
191 ucs2_t *strcasechr_sp(const ucs2_t *s, u_int32_t c_sp)
193 if (*s == 0) return NULL;
195 if (toupper_sp(c_sp) == toupper_sp((u_int32_t)*s << 16 | (u_int32_t)s[1])) return (ucs2_t *)s;
202 /*******************************************************************
204 ********************************************************************/
205 /* no problem of surrogate pair */
207 int strcmp_w(const ucs2_t *a, const ucs2_t *b)
209 while (*b && *a == *b) { a++; b++; }
211 /* warning: if *a != *b and both are not 0 we retrun a random
212 greater or lesser than 0 number not realted to which
216 /*******************************************************************
218 ********************************************************************/
219 /* no problem of surrogate pair */
221 int strncmp_w(const ucs2_t *a, const ucs2_t *b, size_t len)
224 while ((n < len) && *b && *a == *b) { a++; b++; n++;}
225 return (len - n)?(*a - *b):0;
228 /*******************************************************************
230 ********************************************************************/
231 /* no problem of surrogate pair */
233 ucs2_t *strstr_w(const ucs2_t *s, const ucs2_t *ins)
238 if (!s || !*s || !ins || !*ins) return NULL;
240 inslen = strlen_w(ins);
242 while ((r = strchr_w(r, *ins))) {
243 if (strncmp_w(r, ins, inslen) == 0) return r;
249 /*******************************************************************
251 ********************************************************************/
254 ucs2_t *strcasestr_w(const ucs2_t *s, const ucs2_t *ins)
259 if (!s || !*s || !ins || !*ins) return NULL;
261 inslen = strlen_w(ins);
263 while ((r = strcasechr_w(r, *ins))) {
264 if (strncasecmp_w(r, ins, inslen) == 0) return r;
270 /*******************************************************************
272 case insensitive string comparison
273 ********************************************************************/
274 /* surrogate pair support */
276 int strcasecmp_w(const ucs2_t *a, const ucs2_t *b)
281 if ((0xD800 <= *a) && (*a < 0xDC00)) {
282 if (ret = tolower_sp((u_int32_t)*a << 16 | (u_int32_t)a[1]) - tolower_sp((u_int32_t)*b << 16 | (u_int32_t)b[1])) return ret;
285 if (!(*a && *b)) return (tolower_w(*a) - tolower_w(*b)); /* avoid buffer over run */
287 if (ret = tolower_w(*a) - tolower_w(*b)) return ret;
292 return (tolower_w(*a) - tolower_w(*b));
295 /*******************************************************************
297 case insensitive string comparison, length limited
298 ********************************************************************/
299 /* NOTE: compare up to 'len+1' if 'len' isolate surrogate pair */
301 int strncasecmp_w(const ucs2_t *a, const ucs2_t *b, size_t len)
306 while ((n < len) && *a && *b) {
307 if ((0xD800 <= *a) && (*a < 0xDC00)) {
308 if (ret = tolower_sp((u_int32_t)*a << 16 | (u_int32_t)a[1]) - tolower_sp((u_int32_t)*b << 16 | (u_int32_t)b[1])) return ret;
312 if (!((n < len) && *a && *b)) return (tolower_w(*a) - tolower_w(*b));
314 if (ret = tolower_w(*a) - tolower_w(*b)) return ret;
320 return (len - n)?(tolower_w(*a) - tolower_w(*b)):0;
323 /*******************************************************************
326 ********************************************************************/
327 /* NOTE: not check isolation of surrogate pair */
328 /* if len == 0 then duplicate the whole string */
330 ucs2_t *strndup_w(const ucs2_t *src, size_t len)
334 if (!len) len = strlen_w(src);
335 dest = (ucs2_t *)malloc((len + 1) * sizeof(ucs2_t));
337 LOG (log_error, logtype_default, "strdup_w: out of memory!");
341 memcpy(dest, src, len * sizeof(ucs2_t));
347 /*******************************************************************
350 ********************************************************************/
351 /* no problem of surrogate pair */
353 ucs2_t *strdup_w(const ucs2_t *src)
355 return strndup_w(src, 0);
358 /*******************************************************************
359 copy a string with max len
360 ********************************************************************/
362 ucs2_t *strncpy_w(ucs2_t *dest, const ucs2_t *src, const size_t max)
366 if (!dest || !src) return NULL;
368 for (len = 0; (src[len] != 0) && (len < max); len++)
369 dest[len] = src[len];
377 /*******************************************************************
378 append a string of len bytes and add a terminator
379 ********************************************************************/
381 ucs2_t *strncat_w(ucs2_t *dest, const ucs2_t *src, const size_t max)
386 if (!dest || !src) return NULL;
388 start = strlen_w(dest);
389 len = strnlen_w(src, max);
391 memcpy(&dest[start], src, len*sizeof(ucs2_t));
398 ucs2_t *strcat_w(ucs2_t *dest, const ucs2_t *src)
403 if (!dest || !src) return NULL;
405 start = strlen_w(dest);
408 memcpy(&dest[start], src, len*sizeof(ucs2_t));
415 /*******************************************************************
416 binary search for pre|decomposition
417 ********************************************************************/
419 static ucs2_t do_precomposition(unsigned int base, unsigned int comb)
422 int max = PRECOMP_COUNT - 1;
424 uint32_t sought = (base << 16) | comb, that;
428 mid = (min + max) / 2;
429 that = (precompositions[mid].base << 16) | (precompositions[mid].comb);
432 } else if (that > sought) {
435 return precompositions[mid].replacement;
442 /* ------------------------ */
443 static uint32_t do_precomposition_sp(unsigned int base_sp, unsigned int comb_sp)
446 int max = PRECOMP_SP_COUNT - 1;
448 uint64_t sought_sp = ((uint64_t)base_sp << 32) | (uint64_t)comb_sp, that_sp;
452 mid = (min + max) / 2;
453 that_sp = ((uint64_t)precompositions_sp[mid].base_sp << 32) | ((uint64_t)precompositions_sp[mid].comb_sp);
454 if (that_sp < sought_sp) {
456 } else if (that_sp > sought_sp) {
459 return precompositions_sp[mid].replacement_sp;
466 /* -------------------------- */
467 static uint32_t do_decomposition(ucs2_t base)
470 int max = DECOMP_COUNT - 1;
472 uint32_t sought = base;
473 uint32_t result, that;
477 mid = (min + max) / 2;
478 that = decompositions[mid].replacement;
481 } else if (that > sought) {
484 result = (decompositions[mid].base << 16) | (decompositions[mid].comb);
492 /* -------------------------- */
493 static uint64_t do_decomposition_sp(unsigned int base_sp)
496 int max = DECOMP_SP_COUNT - 1;
498 uint32_t sought_sp = base_sp;
504 mid = (min + max) / 2;
505 that_sp = decompositions_sp[mid].replacement_sp;
506 if (that_sp < sought_sp) {
508 } else if (that_sp > sought_sp) {
511 result_sp = ((uint64_t)decompositions_sp[mid].base_sp << 32) | ((uint64_t)decompositions_sp[mid].comb_sp);
519 /*******************************************************************
522 we can't use static, this stuff needs to be reentrant
523 static char comp[MAXPATHLEN +1];
525 We don't implement Singleton and Canonical Ordering.
526 We ignore CompositionExclusions.txt.
527 because they cause the problem of the roundtrip
528 such as Dancing Icon.
530 exclude U2000-U2FFF, UFE30-UFE4F and U2F800-U2FA1F ranges
531 in precompose.h from composition according to AFP 3.x spec
532 ********************************************************************/
534 size_t precompose_w (ucs2_t *name, size_t inplen, ucs2_t *comp, size_t *outlen)
538 uint32_t base_sp, comb_sp;
540 ucs2_t lindex, vindex;
543 size_t o_len = *outlen;
545 if (!inplen || (inplen & 1) || inplen > o_len)
553 while (*outlen > 2) {
562 return o_len - *outlen;
568 /* Non-Combination Character */
571 /* Unicode Standard Annex #15 A10.3 Hangul Composition */
573 else if ((VBASE <= comb) && (comb <= VBASE + VCOUNT)) {
574 if ((LBASE <= base) && (base < LBASE + LCOUNT)) {
576 lindex = base - LBASE;
577 vindex = comb - VBASE;
578 base = SBASE + (lindex * VCOUNT + vindex) * TCOUNT;
583 else if ((TBASE < comb) && (comb < TBASE + TCOUNT)) {
584 if ((SBASE <= base) && (base < SBASE + SCOUNT) && (((base - SBASE) % TCOUNT) == 0)) {
586 base += comb - TBASE;
590 /* Binary Search for Surrogate Pair */
591 else if ((0xD800 <= base) && (base < 0xDC00)) {
592 if ((0xDC00 <= comb) && (comb < 0xE000) && (i + 4 <= inplen)) {
593 base_sp = ((uint32_t)base << 16) | (uint32_t)comb;
595 comb_sp = ((uint32_t)in[1] << 16) | (uint32_t)in[2];
596 if (result_sp = do_precomposition_sp(base_sp, comb_sp)) {
601 } while ((i + 4 <= inplen) && result_sp) ;
603 *out = base_sp >> 16;
612 *out = base_sp & 0xFFFF;
624 /* Binary Search for BMP */
625 else if (result = do_precomposition(base, comb)) {
641 /* --------------- */
642 size_t decompose_w (ucs2_t *name, size_t inplen, ucs2_t *comp, size_t *outlen)
646 ucs2_t base, comb[COMBBUFLEN];
648 ucs2_t sindex, tjamo;
652 size_t o_len = *outlen;
654 if (!inplen || (inplen & 1))
664 /* check ASCII first. this is frequent. */
665 if (base <= 0x007f) ;
667 /* Unicode Standard Annex #15 A10.2 Hangul Decomposition */
668 else if ((SBASE <= base) && (base < SBASE + SCOUNT)) {
669 sindex = base - SBASE;
670 base = LBASE + sindex / NCOUNT;
671 comb[COMBBUFLEN-2] = VBASE + (sindex % NCOUNT) / TCOUNT;
674 if ((tjamo = TBASE + sindex % TCOUNT) == TBASE) {
675 comb[COMBBUFLEN-1] = comb[COMBBUFLEN-2];
681 comb[COMBBUFLEN-1] = tjamo;
686 /* Binary Search for Surrogate Pair */
687 else if ((0xD800 <= base) && (base < 0xDC00)) {
688 if (i + 2 < inplen) {
689 base_sp = ((uint32_t)base << 16) | (uint32_t)in[1];
691 if ( !(result_sp = do_decomposition_sp(base_sp))) break;
693 base_sp = result_sp >> 32;
694 comb[COMBBUFLEN-comblen] = (result_sp >> 16) & 0xFFFF; /* hi */
695 comb[COMBBUFLEN-comblen+1] = result_sp & 0xFFFF; /* lo */
696 } while (comblen < MAXCOMBSPLEN);
698 if (*outlen < (comblen + 1) << 1) {
703 *out = base_sp >> 16; /* hi */
707 base = base_sp & 0xFFFF; /* lo */
714 /* Binary Search for BMP */
717 if ( !(result = do_decomposition(base))) break;
720 comb[COMBBUFLEN-comblen] = result & 0xFFFF;
721 } while ((0x007f < base) && (comblen < MAXCOMBLEN));
724 if (*outlen < (comblen + 1) << 1) {
733 while ( comblen > 0 ) {
734 *out = comb[COMBBUFLEN-comblen];
745 return o_len-*outlen;
748 /*******************************************************************
749 length of UTF-8 character and string
750 ********************************************************************/
752 size_t utf8_charlen ( char* utf8 )
756 p = (unsigned char*) utf8;
760 else if ( *p > 0xC1 && *p < 0xe0 && *(p+1) > 0x7f && *(p+1) < 0xC0)
762 else if ( *p == 0xe0 && *(p+1) > 0x9f && *(p+1) < 0xc0 && *(p+2) > 0x7f && *(p+2) < 0xc0)
764 else if ( *p > 0xe0 && *p < 0xf0 && *(p+1) > 0x7f && *(p+1) < 0xc0 && *(p+2) > 0x7f && *(p+2) < 0xc0)
766 else if ( *p == 0xf0 && *(p+1) > 0x8f && *(p+1) < 0xc0 && *(p+2) > 0x7f && *(p+2) < 0xc0 && *(p+3) > 0x7f && *(p+3) < 0xc0 )
768 else if ( *p > 0xf0 && *p < 0xf4 && *(p+1) > 0x7f && *(p+1) < 0xc0 && *(p+2) > 0x7f && *(p+2) < 0xc0 && *(p+3) > 0x7f && *(p+3) < 0xc0 )
770 else if ( *p == 0xf4 && *(p+1) > 0x7f && *(p+1) < 0x90 && *(p+2) > 0x7f && *(p+2) < 0xc0 && *(p+3) > 0x7f && *(p+3) < 0xc0 )
773 return ((size_t) -1);
777 size_t utf8_strlen_validate ( char * utf8 )
782 p = (unsigned char*) utf8;
785 /* see http://www.unicode.org/unicode/reports/tr27/ for an explanation */
792 else if ( *p > 0xC1 && *p < 0xe0 && *(p+1) > 0x7f && *(p+1) < 0xC0)
795 else if ( *p == 0xe0 && *(p+1) > 0x9f && *(p+1) < 0xc0 && *(p+2) > 0x7f && *(p+2) < 0xc0)
798 else if ( *p > 0xe0 && *p < 0xf0 && *(p+1) > 0x7f && *(p+1) < 0xc0 && *(p+2) > 0x7f && *(p+2) < 0xc0)
801 else if ( *p == 0xf0 && *(p+1) > 0x8f && *(p+1) < 0xc0 && *(p+2) > 0x7f && *(p+2) < 0xc0 && *(p+3) > 0x7f && *(p+3) < 0xc0 )
804 else if ( *p > 0xf0 && *p < 0xf4 && *(p+1) > 0x7f && *(p+1) < 0xc0 && *(p+2) > 0x7f && *(p+2) < 0xc0 && *(p+3) > 0x7f && *(p+3) < 0xc0 )
807 else if ( *p == 0xf4 && *(p+1) > 0x7f && *(p+1) < 0x90 && *(p+2) > 0x7f && *(p+2) < 0xc0 && *(p+3) > 0x7f && *(p+3) < 0xc0 )
811 return ((size_t) -1);