libs/xslt/libxslt/xsltlocale.c

   1 /*
   2  * xsltlocale.c: locale handling
   3  *
   4  * Reference:
   5  * RFC 3066: Tags for the Identification of Languages
   6  * http://www.ietf.org/rfc/rfc3066.txt
   7  * ISO 639-1, ISO 3166-1
   8  *
   9  * Author: Nick Wellnhofer
  10  * winapi port: Roumen Petrov
  11  */
  12
  13 #define IN_LIBXSLT
  14 #include "libxslt.h"
  15
  16 #include <string.h>
  17 #include <libxml/xmlmemory.h>
  18
  19 #include "xsltlocale.h"
  20 #include "xsltutils.h"
  21
  22 #define TOUPPER(c) (c & ~0x20)
  23 #define TOLOWER(c) (c | 0x20)
  24 #define ISALPHA(c) ((unsigned)(TOUPPER(c) - 'A') < 26)
  25
  26 /*without terminating null character*/
  27 #define XSLTMAX_ISO639LANGLEN           8
  28 #define XSLTMAX_ISO3166CNTRYLEN         8
  29                                         /* <lang>-<cntry> */
  30 #define XSLTMAX_LANGTAGLEN              (XSLTMAX_ISO639LANGLEN+1+XSLTMAX_ISO3166CNTRYLEN)
  31
  32 static const xmlChar* xsltDefaultRegion(const xmlChar *localeName);
  33
  34 #ifdef XSLT_LOCALE_WINAPI
  35 xmlRMutexPtr xsltLocaleMutex = NULL;
  36
  37 struct xsltRFC1766Info_s {
  38       /*note typedef unsigned char xmlChar !*/
  39     xmlChar    tag[XSLTMAX_LANGTAGLEN+1];
  40       /*note typedef LCID xsltLocale !*/
  41     xsltLocale lcid;
  42 };
  43 typedef struct xsltRFC1766Info_s xsltRFC1766Info;
  44
  45 static int xsltLocaleListSize = 0;
  46 static xsltRFC1766Info *xsltLocaleList = NULL;
  47
  48
  49 static xsltLocale
  50 xslt_locale_WINAPI(const xmlChar *languageTag) {
  51     int k;
  52     xsltRFC1766Info *p = xsltLocaleList;
  53
  54     for (k=0; k<xsltLocaleListSize; k++, p++)
  55         if (xmlStrcmp(p->tag, languageTag) == 0) return p->lcid;
  56     return((xsltLocale)0);
  57 }
  58
  59 static void xsltEnumSupportedLocales(void);
  60 #endif
  61
  62 /**
  63  * xsltFreeLocales:
  64  *
  65  * Cleanup function for the locale support on shutdown
  66  */
  67 void
  68 xsltFreeLocales(void) {
  69 #ifdef XSLT_LOCALE_WINAPI
  70     xmlRMutexLock(xsltLocaleMutex);
  71     xmlFree(xsltLocaleList);
  72     xsltLocaleList = NULL;
  73     xmlRMutexUnlock(xsltLocaleMutex);
  74 #endif
  75 }
  76
  77 /**
  78  * xsltNewLocale:
  79  * @languageTag: RFC 3066 language tag
  80  *
  81  * Creates a new locale of an opaque system dependent type based on the
  82  * language tag.
  83  *
  84  * Returns the locale or NULL on error or if no matching locale was found
  85  */
  86 xsltLocale
  87 xsltNewLocale(const xmlChar *languageTag) {
  88 #ifdef XSLT_LOCALE_POSIX
  89     xsltLocale locale;
  90     char localeName[XSLTMAX_LANGTAGLEN+6]; /* 6 chars for ".utf8\0" */
  91     const xmlChar *p = languageTag;
  92     const char *region = NULL;
  93     char *q = localeName;
  94     int i, llen;
  95
  96     /* Convert something like "pt-br" to "pt_BR.utf8" */
  97
  98     if (languageTag == NULL)
  99         return(NULL);
 100
 101     for (i=0; i<XSLTMAX_ISO639LANGLEN && ISALPHA(*p); ++i)
 102         *q++ = TOLOWER(*p++);
 103
 104     if (i == 0)
 105         return(NULL);
 106
 107     llen = i;
 108
 109     if (*p) {
 110         if (*p++ != '-')
 111             return(NULL);
 112         *q++ = '_';
 113
 114         for (i=0; i<XSLTMAX_ISO3166CNTRYLEN && ISALPHA(*p); ++i)
 115             *q++ = TOUPPER(*p++);
 116
 117         if (i == 0 || *p)
 118             return(NULL);
 119
 120         memcpy(q, ".utf8", 6);
 121         locale = newlocale(LC_COLLATE_MASK, localeName, NULL);
 122         if (locale != NULL)
 123             return(locale);
 124
 125         /* Continue without using country code */
 126
 127         q = localeName + llen;
 128     }
 129
 130     /* Try locale without territory, e.g. for Esperanto (eo) */
 131
 132     memcpy(q, ".utf8", 6);
 133     locale = newlocale(LC_COLLATE_MASK, localeName, NULL);
 134     if (locale != NULL)
 135         return(locale);
 136
 137     /* Try to find most common country for language */
 138
 139     if (llen != 2)
 140         return(NULL);
 141
 142     region = (char *)xsltDefaultRegion((xmlChar *)localeName);
 143     if (region == NULL)
 144         return(NULL);
 145
 146     q = localeName + llen;
 147     *q++ = '_';
 148     *q++ = region[0];
 149     *q++ = region[1];
 150     memcpy(q, ".utf8", 6);
 151     locale = newlocale(LC_COLLATE_MASK, localeName, NULL);
 152
 153     return(locale);
 154 #endif
 155
 156 #ifdef XSLT_LOCALE_WINAPI
 157 {
 158     xsltLocale    locale = (xsltLocale)0;
 159     xmlChar       localeName[XSLTMAX_LANGTAGLEN+1];
 160     xmlChar       *q = localeName;
 161     const xmlChar *p = languageTag;
 162     int           i, llen;
 163     const xmlChar *region = NULL;
 164
 165     if (languageTag == NULL) goto end;
 166
 167     xsltEnumSupportedLocales();
 168
 169     for (i=0; i<XSLTMAX_ISO639LANGLEN && ISALPHA(*p); ++i)
 170         *q++ = TOLOWER(*p++);
 171     if (i == 0) goto end;
 172
 173     llen = i;
 174     *q++ = '-';
 175     if (*p) { /*if country tag is given*/
 176         if (*p++ != '-') goto end;
 177
 178         for (i=0; i<XSLTMAX_ISO3166CNTRYLEN && ISALPHA(*p); ++i)
 179             *q++ = TOUPPER(*p++);
 180         if (i == 0 || *p) goto end;
 181
 182         *q = '\0';
 183         locale = xslt_locale_WINAPI(localeName);
 184         if (locale != (xsltLocale)0) goto end;
 185     }
 186     /* Try to find most common country for language */
 187     region = xsltDefaultRegion(localeName);
 188     if (region == NULL) goto end;
 189
 190     strcpy((char *) localeName + llen + 1, (char *) region);
 191     locale = xslt_locale_WINAPI(localeName);
 192 end:
 193     return(locale);
 194 }
 195 #endif
 196
 197 #ifdef XSLT_LOCALE_NONE
 198     return(NULL);
 199 #endif
 200 }
 201
 202 static const xmlChar*
 203 xsltDefaultRegion(const xmlChar *localeName) {
 204     xmlChar c;
 205     /* region should be xmlChar, but gcc warns on all string assignments */
 206     const char *region = NULL;
 207
 208     c = localeName[1];
 209     /* This is based on the locales from glibc 2.3.3 */
 210
 211     switch (localeName[0]) {
 212         case 'a':
 213             if (c == 'a' || c == 'm') region = "ET";
 214             else if (c == 'f') region = "ZA";
 215             else if (c == 'n') region = "ES";
 216             else if (c == 'r') region = "AE";
 217             else if (c == 'z') region = "AZ";
 218             break;
 219         case 'b':
 220             if (c == 'e') region = "BY";
 221             else if (c == 'g') region = "BG";
 222             else if (c == 'n') region = "BD";
 223             else if (c == 'r') region = "FR";
 224             else if (c == 's') region = "BA";
 225             break;
 226         case 'c':
 227             if (c == 'a') region = "ES";
 228             else if (c == 's') region = "CZ";
 229             else if (c == 'y') region = "GB";
 230             break;
 231         case 'd':
 232             if (c == 'a') region = "DK";
 233             else if (c == 'e') region = "DE";
 234             break;
 235         case 'e':
 236             if (c == 'l') region = "GR";
 237             else if (c == 'n' || c == 'o') region = "US";
 238             else if (c == 's' || c == 'u') region = "ES";
 239             else if (c == 't') region = "EE";
 240             break;
 241         case 'f':
 242             if (c == 'a') region = "IR";
 243             else if (c == 'i') region = "FI";
 244             else if (c == 'o') region = "FO";
 245             else if (c == 'r') region = "FR";
 246             break;
 247         case 'g':
 248             if (c == 'a') region = "IE";
 249             else if (c == 'l') region = "ES";
 250             else if (c == 'v') region = "GB";
 251             break;
 252         case 'h':
 253             if (c == 'e') region = "IL";
 254             else if (c == 'i') region = "IN";
 255             else if (c == 'r') region = "HT";
 256             else if (c == 'u') region = "HU";
 257             break;
 258         case 'i':
 259             if (c == 'd') region = "ID";
 260             else if (c == 's') region = "IS";
 261             else if (c == 't') region = "IT";
 262             else if (c == 'w') region = "IL";
 263             break;
 264         case 'j':
 265             if (c == 'a') region = "JP";
 266             break;
 267         case 'k':
 268             if (c == 'l') region = "GL";
 269             else if (c == 'o') region = "KR";
 270             else if (c == 'w') region = "GB";
 271             break;
 272         case 'l':
 273             if (c == 't') region = "LT";
 274             else if (c == 'v') region = "LV";
 275             break;
 276         case 'm':
 277             if (c == 'k') region = "MK";
 278             else if (c == 'l' || c == 'r') region = "IN";
 279             else if (c == 'n') region = "MN";
 280             else if (c == 's') region = "MY";
 281             else if (c == 't') region = "MT";
 282             break;
 283         case 'n':
 284             if (c == 'b' || c == 'n' || c == 'o') region = "NO";
 285             else if (c == 'e') region = "NP";
 286             else if (c == 'l') region = "NL";
 287             break;
 288         case 'o':
 289             if (c == 'm') region = "ET";
 290             break;
 291         case 'p':
 292             if (c == 'a') region = "IN";
 293             else if (c == 'l') region = "PL";
 294             else if (c == 't') region = "PT";
 295             break;
 296         case 'r':
 297             if (c == 'o') region = "RO";
 298             else if (c == 'u') region = "RU";
 299             break;
 300         case 's':
 301             switch (c) {
 302                 case 'e': region = "NO"; break;
 303                 case 'h': region = "YU"; break;
 304                 case 'k': region = "SK"; break;
 305                 case 'l': region = "SI"; break;
 306                 case 'o': region = "ET"; break;
 307                 case 'q': region = "AL"; break;
 308                 case 't': region = "ZA"; break;
 309                 case 'v': region = "SE"; break;
 310             }
 311             break;
 312         case 't':
 313             if (c == 'a' || c == 'e') region = "IN";
 314             else if (c == 'h') region = "TH";
 315             else if (c == 'i') region = "ER";
 316             else if (c == 'r') region = "TR";
 317             else if (c == 't') region = "RU";
 318             break;
 319         case 'u':
 320             if (c == 'k') region = "UA";
 321             else if (c == 'r') region = "PK";
 322             break;
 323         case 'v':
 324             if (c == 'i') region = "VN";
 325             break;
 326         case 'w':
 327             if (c == 'a') region = "BE";
 328             break;
 329         case 'x':
 330             if (c == 'h') region = "ZA";
 331             break;
 332         case 'z':
 333             if (c == 'h') region = "CN";
 334             else if (c == 'u') region = "ZA";
 335             break;
 336     }
 337     return((xmlChar *)region);
 338 }
 339
 340 /**
 341  * xsltFreeLocale:
 342  * @locale: the locale to free
 343  *
 344  * Frees a locale created with xsltNewLocale
 345  */
 346 void
 347 xsltFreeLocale(xsltLocale locale) {
 348 #ifdef XSLT_LOCALE_POSIX
 349     freelocale(locale);
 350 #endif
 351 }
 352
 353 /**
 354  * xsltStrxfrm:
 355  * @locale: locale created with xsltNewLocale
 356  * @string: UTF-8 string to transform
 357  *
 358  * Transforms a string according to locale. The transformed string must then be
 359  * compared with xsltLocaleStrcmp and freed with xmlFree.
 360  *
 361  * Returns the transformed string or NULL on error
 362  */
 363 xsltLocaleChar *
 364 xsltStrxfrm(xsltLocale locale, const xmlChar *string)
 365 {
 366 #ifdef XSLT_LOCALE_NONE
 367     return(NULL);
 368 #else
 369     size_t xstrlen, r;
 370     xsltLocaleChar *xstr;
 371
 372 #ifdef XSLT_LOCALE_POSIX
 373     xstrlen = strxfrm_l(NULL, (const char *)string, 0, locale) + 1;
 374     xstr = (xsltLocaleChar *) xmlMalloc(xstrlen);
 375     if (xstr == NULL) {
 376         xsltTransformError(NULL, NULL, NULL,
 377             "xsltStrxfrm : out of memory error\n");
 378         return(NULL);
 379     }
 380
 381     r = strxfrm_l((char *)xstr, (const char *)string, xstrlen, locale);
 382 #endif
 383
 384 #ifdef XSLT_LOCALE_WINAPI
 385     xstrlen = MultiByteToWideChar(CP_UTF8, 0, (char *) string, -1, NULL, 0);
 386     if (xstrlen == 0) {
 387         xsltTransformError(NULL, NULL, NULL, "xsltStrxfrm : MultiByteToWideChar check failed\n");
 388         return(NULL);
 389     }
 390     xstr = (xsltLocaleChar*) xmlMalloc(xstrlen * sizeof(xsltLocaleChar));
 391     if (xstr == NULL) {
 392         xsltTransformError(NULL, NULL, NULL, "xsltStrxfrm : out of memory\n");
 393         return(NULL);
 394     }
 395     r = MultiByteToWideChar(CP_UTF8, 0, (char *) string, -1, xstr, xstrlen);
 396     if (r == 0) {
 397         xsltTransformError(NULL, NULL, NULL, "xsltStrxfrm : MultiByteToWideChar failed\n");
 398         xmlFree(xstr);
 399         return(NULL);
 400     }
 401     return(xstr);
 402 #endif /* XSLT_LOCALE_WINAPI */
 403
 404     if (r >= xstrlen) {
 405         xsltTransformError(NULL, NULL, NULL, "xsltStrxfrm : strxfrm failed\n");
 406         xmlFree(xstr);
 407         return(NULL);
 408     }
 409
 410     return(xstr);
 411 #endif /* XSLT_LOCALE_NONE */
 412 }
 413
 414 /**
 415  * xsltLocaleStrcmp:
 416  * @locale: a locale identifier
 417  * @str1: a string transformed with xsltStrxfrm
 418  * @str2: a string transformed with xsltStrxfrm
 419  *
 420  * Compares two strings transformed with xsltStrxfrm
 421  *
 422  * Returns a value < 0 if str1 sorts before str2,
 423  *         a value > 0 if str1 sorts after str2,
 424  *         0 if str1 and str2 are equal wrt sorting
 425  */
 426 int
 427 xsltLocaleStrcmp(xsltLocale locale, const xsltLocaleChar *str1, const xsltLocaleChar *str2) {
 428     (void)locale;
 429 #ifdef XSLT_LOCALE_WINAPI
 430 {
 431     int ret;
 432     if (str1 == str2) return(0);
 433     if (str1 == NULL) return(-1);
 434     if (str2 == NULL) return(1);
 435     ret = CompareStringW(locale, 0, str1, -1, str2, -1);
 436     if (ret == 0) {
 437         xsltTransformError(NULL, NULL, NULL, "xsltLocaleStrcmp : CompareStringW fail\n");
 438         return(0);
 439     }
 440     return(ret - 2);
 441 }
 442 #else
 443     return(xmlStrcmp(str1, str2));
 444 #endif
 445 }
 446
 447 #ifdef XSLT_LOCALE_WINAPI
 448 /**
 449  * xsltCountSupportedLocales:
 450  * @lcid: not used
 451  *
 452  * callback used to count locales
 453  *
 454  * Returns TRUE
 455  */
 456 BOOL CALLBACK
 457 xsltCountSupportedLocales(LPSTR lcid) {
 458     (void) lcid;
 459     ++xsltLocaleListSize;
 460     return(TRUE);
 461 }
 462
 463 /**
 464  * xsltIterateSupportedLocales:
 465  * @lcid: not used
 466  *
 467  * callback used to track locales
 468  *
 469  * Returns TRUE if not at the end of the array
 470  */
 471 BOOL CALLBACK
 472 xsltIterateSupportedLocales(LPSTR lcid) {
 473     static int count = 0;
 474     xmlChar    iso639lang [XSLTMAX_ISO639LANGLEN  +1];
 475     xmlChar    iso3136ctry[XSLTMAX_ISO3166CNTRYLEN+1];
 476     int        k, l;
 477     xsltRFC1766Info *p = xsltLocaleList + count;
 478
 479     k = sscanf(lcid, "%lx", (long*)&p->lcid);
 480     if (k < 1) goto end;
 481     /*don't count terminating null character*/
 482     k = GetLocaleInfoA(p->lcid, LOCALE_SISO639LANGNAME,
 483                        (char *) iso639lang, sizeof(iso639lang));
 484     if (--k < 1) goto end;
 485     l = GetLocaleInfoA(p->lcid, LOCALE_SISO3166CTRYNAME,
 486                        (char *) iso3136ctry, sizeof(iso3136ctry));
 487     if (--l < 1) goto end;
 488
 489     {  /*fill results*/
 490         xmlChar    *q = p->tag;
 491         memcpy(q, iso639lang, k);
 492         q += k;
 493         *q++ = '-';
 494         memcpy(q, iso3136ctry, l);
 495         q += l;
 496         *q = '\0';
 497     }
 498     ++count;
 499 end:
 500     return((count < xsltLocaleListSize) ? TRUE : FALSE);
 501 }
 502
 503
 504 static void
 505 xsltEnumSupportedLocales(void) {
 506     xmlRMutexLock(xsltLocaleMutex);
 507     if (xsltLocaleListSize <= 0) {
 508         size_t len;
 509
 510         EnumSystemLocalesA(xsltCountSupportedLocales, LCID_SUPPORTED);
 511
 512         len = xsltLocaleListSize * sizeof(xsltRFC1766Info);
 513         xsltLocaleList = xmlMalloc(len);
 514         memset(xsltLocaleList, 0, len);
 515         EnumSystemLocalesA(xsltIterateSupportedLocales, LCID_SUPPORTED);
 516     }
 517     xmlRMutexUnlock(xsltLocaleMutex);
 518 }
 519
 520 #endif /*def XSLT_LOCALE_WINAPI*/