src/third_party/icu/source/tools/genrb/wrtxml.cpp

   1 /*
   2 *******************************************************************************
   3 *
   4 *   Copyright (C) 2002-2012, International Business Machines
   5 *   Corporation and others.  All Rights Reserved.
   6 *
   7 *******************************************************************************
   8 *
   9 * File wrtxml.cpp
  10 *
  11 * Modification History:
  12 *
  13 *   Date        Name        Description
  14 *   10/01/02    Ram         Creation.
  15 *   02/07/08    Spieth      Correct XLIFF generation on EBCDIC platform
  16 *
  17 *******************************************************************************
  18 */
  19 #include "reslist.h"
  20 #include "unewdata.h"
  21 #include "unicode/ures.h"
  22 #include "errmsg.h"
  23 #include "filestrm.h"
  24 #include "cstring.h"
  25 #include "unicode/ucnv.h"
  26 #include "genrb.h"
  27 #include "rle.h"
  28 #include "ucol_tok.h"
  29 #include "uhash.h"
  30 #include "uresimp.h"
  31 #include "unicode/ustring.h"
  32 #include "unicode/uchar.h"
  33 #include "ustr.h"
  34 #include "prscmnts.h"
  35 #include "unicode/unistr.h"
  36 #include "unicode/utf8.h"
  37 #include "unicode/utf16.h"
  38 #include <time.h>
  39
  40 U_NAMESPACE_USE
  41
  42 static int tabCount = 0;
  43
  44 static FileStream* out=NULL;
  45 static struct SRBRoot* srBundle ;
  46 static const char* outDir = NULL;
  47 static const char* enc ="";
  48 static UConverter* conv = NULL;
  49
  50 const char* const* ISOLanguages;
  51 const char* const* ISOCountries;
  52 const char* textExt = ".txt";
  53 const char* xliffExt = ".xlf";
  54
  55 static int32_t write_utf8_file(FileStream* fileStream, UnicodeString outString)
  56 {
  57     UErrorCode status = U_ZERO_ERROR;
  58     int32_t len = 0;
  59
  60     // preflight to get the destination buffer size
  61     u_strToUTF8(NULL,
  62                 0,
  63                 &len,
  64                 outString.getBuffer(),
  65                 outString.length(),
  66                 &status);
  67
  68     // allocate the buffer
  69     char* dest = (char*)uprv_malloc(len);
  70     status = U_ZERO_ERROR;
  71
  72     // convert the data
  73     u_strToUTF8(dest,
  74                 len,
  75                 &len,
  76                 outString.getBuffer(),
  77                 outString.length(),
  78                 &status);
  79
  80     // write data to out file
  81     int32_t ret = T_FileStream_write(fileStream, dest, len);
  82     uprv_free(dest);
  83     return (ret);
  84 }
  85
  86 /*write indentation for formatting*/
  87 static void write_tabs(FileStream* os){
  88     int i=0;
  89     for(;i<=tabCount;i++){
  90         write_utf8_file(os,UnicodeString("    "));
  91     }
  92 }
  93
  94 /*get ID for each element. ID is globally unique.*/
  95 static char* getID(const char* id, const char* curKey, char* result) {
  96     if(curKey == NULL) {
  97         result = (char *)uprv_malloc(sizeof(char)*uprv_strlen(id) + 1);
  98         uprv_memset(result, 0, sizeof(char)*uprv_strlen(id) + 1);
  99         uprv_strcpy(result, id);
 100     } else {
 101         result = (char *)uprv_malloc(sizeof(char)*(uprv_strlen(id) + 1 + uprv_strlen(curKey)) + 1);
 102         uprv_memset(result, 0, sizeof(char)*(uprv_strlen(id) + 1 + uprv_strlen(curKey)) + 1);
 103         if(id[0]!='\0'){
 104             uprv_strcpy(result, id);
 105             uprv_strcat(result, "_");
 106         }
 107         uprv_strcat(result, curKey);
 108     }
 109     return result;
 110 }
 111
 112 /*compute CRC for binary code*/
 113 /* The code is from  http://www.theorem.com/java/CRC32.java
 114  * Calculates the CRC32 - 32 bit Cyclical Redundancy Check
 115  * <P> This check is used in numerous systems to verify the integrity
 116  * of information.  It's also used as a hashing function.  Unlike a regular
 117  * checksum, it's sensitive to the order of the characters.
 118  * It produces a 32 bit
 119  *
 120  * @author Michael Lecuyer (mjl@theorem.com)
 121  * @version 1.1 August 11, 1998
 122  */
 123
 124 /* ICU is not endian portable, because ICU data generated on big endian machines can be
 125  * ported to big endian machines but not to little endian machines and vice versa. The
 126  * conversion is not portable across platforms with different endianess.
 127  */
 128
 129 uint32_t computeCRC(char *ptr, uint32_t len, uint32_t lastcrc){
 130     int32_t crc;
 131     uint32_t temp1;
 132     uint32_t temp2;
 133
 134     int32_t crc_ta[256];
 135     int i = 0;
 136     int j = 0;
 137     uint32_t crc2 = 0;
 138
 139 #define CRC32_POLYNOMIAL 0xEDB88320
 140
 141     /*build crc table*/
 142     for (i = 0; i <= 255; i++) {
 143         crc2 = i;
 144         for (j = 8; j > 0; j--) {
 145             if ((crc2 & 1) == 1) {
 146                 crc2 = (crc2 >> 1) ^ CRC32_POLYNOMIAL;
 147             } else {
 148                 crc2 >>= 1;
 149             }
 150         }
 151         crc_ta[i] = crc2;
 152     }
 153
 154     crc = lastcrc;
 155     while(len--!=0) {
 156         temp1 = (uint32_t)crc>>8;
 157         temp2 = crc_ta[(crc^*ptr) & 0xFF];
 158         crc = temp1^temp2;
 159         ptr++;
 160     }
 161     return(crc);
 162 }
 163
 164 static void strnrepchr(char* src, int32_t srcLen, char s, char r){
 165     int32_t i = 0;
 166     for(i=0;i<srcLen;i++){
 167         if(src[i]==s){
 168             src[i]=r;
 169         }
 170     }
 171 }
 172 /* Parse the filename, and get its language information.
 173  * If it fails to get the language information from the filename,
 174  * use "en" as the default value for language
 175  */
 176 static char* parseFilename(const char* id, char* /*lang*/) {
 177     int idLen = (int) uprv_strlen(id);
 178     char* localeID = (char*) uprv_malloc(idLen);
 179     int pos = 0;
 180     int canonCapacity = 0;
 181     char* canon = NULL;
 182     int canonLen = 0;
 183     /*int i;*/
 184     UErrorCode status = U_ZERO_ERROR;
 185     const char *ext = uprv_strchr(id, '.');
 186
 187     if(ext != NULL){
 188         pos = (int) (ext - id);
 189     } else {
 190         pos = idLen;
 191     }
 192     uprv_memcpy(localeID, id, pos);
 193     localeID[pos]=0; /* NUL terminate the string */
 194
 195     canonCapacity =pos*3;
 196     canon = (char*) uprv_malloc(canonCapacity);
 197     canonLen = uloc_canonicalize(localeID, canon, canonCapacity, &status);
 198
 199     if(U_FAILURE(status)){
 200         fprintf(stderr, "Could not canonicalize the locale ID: %s. Error: %s\n", localeID, u_errorName(status));
 201         exit(status);
 202     }
 203     strnrepchr(canon, canonLen, '_', '-');
 204     return canon;
 205 }
 206
 207 static const char* xmlHeader = "<?xml version=\"1.0\" encoding=\"utf-8\"?>\n";
 208 #if 0
 209 static const char* bundleStart = "<xliff version = \"1.2\" "
 210                                         "xmlns='urn:oasis:names:tc:xliff:document:1.2' "
 211                                         "xmlns:xsi='http://www.w3.org/2001/XMLSchema-instance' "
 212                                         "xsi:schemaLocation='urn:oasis:names:tc:xliff:document:1.2 xliff-core-1.2-transitional.xsd'>\n";
 213 #else
 214 static const char* bundleStart = "<xliff version = \"1.1\" "
 215                                         "xmlns='urn:oasis:names:tc:xliff:document:1.1' "
 216                                         "xmlns:xsi='http://www.w3.org/2001/XMLSchema-instance' "
 217                                         "xsi:schemaLocation='urn:oasis:names:tc:xliff:document:1.1 http://www.oasis-open.org/committees/xliff/documents/xliff-core-1.1.xsd'>\n";
 218 #endif
 219 static const char* bundleEnd   = "</xliff>\n";
 220
 221 void res_write_xml(struct SResource *res, const char* id, const char* language, UBool isTopLevel, UErrorCode *status);
 222
 223 static char* convertAndEscape(char** pDest, int32_t destCap, int32_t* destLength,
 224                               const UChar* src, int32_t srcLen, UErrorCode* status){
 225     int32_t srcIndex=0;
 226     char* dest=NULL;
 227     char* temp=NULL;
 228     int32_t destLen=0;
 229     UChar32 c = 0;
 230
 231     if(status==NULL || U_FAILURE(*status) || pDest==NULL  || srcLen==0 || src == NULL){
 232         return NULL;
 233     }
 234     dest =*pDest;
 235     if(dest==NULL || destCap <=0){
 236         destCap = srcLen * 8;
 237         dest = (char*) uprv_malloc(sizeof(char) * destCap);
 238         if(dest==NULL){
 239             *status=U_MEMORY_ALLOCATION_ERROR;
 240             return NULL;
 241         }
 242     }
 243
 244     dest[0]=0;
 245
 246     while(srcIndex<srcLen){
 247         U16_NEXT(src, srcIndex, srcLen, c);
 248
 249         if (U16_IS_LEAD(c) || U16_IS_TRAIL(c)) {
 250             *status = U_ILLEGAL_CHAR_FOUND;
 251             fprintf(stderr, "Illegal Surrogate! \n");
 252             uprv_free(dest);
 253             return NULL;
 254         }
 255
 256         if((destLen+U8_LENGTH(c)) < destCap){
 257
 258             /* ASCII Range */
 259             if(c <=0x007F){
 260                 switch(c) {
 261                 case '\x26':
 262                     uprv_strcpy(dest+( destLen),"\x26\x61\x6d\x70\x3b"); /* &amp;*/
 263                     destLen+=(int32_t)uprv_strlen("\x26\x61\x6d\x70\x3b");
 264                     break;
 265                 case '\x3c':
 266                     uprv_strcpy(dest+(destLen),"\x26\x6c\x74\x3b"); /* &lt;*/
 267                     destLen+=(int32_t)uprv_strlen("\x26\x6c\x74\x3b");
 268                     break;
 269                 case '\x3e':
 270                     uprv_strcpy(dest+(destLen),"\x26\x67\x74\x3b"); /* &gt;*/
 271                     destLen+=(int32_t)uprv_strlen("\x26\x67\x74\x3b");
 272                     break;
 273                 case '\x22':
 274                     uprv_strcpy(dest+(destLen),"\x26\x71\x75\x6f\x74\x3b"); /* &quot;*/
 275                     destLen+=(int32_t)uprv_strlen("\x26\x71\x75\x6f\x74\x3b");
 276                     break;
 277                 case '\x27':
 278                     uprv_strcpy(dest+(destLen),"\x26\x61\x70\x6f\x73\x3b"); /* &apos; */
 279                     destLen+=(int32_t)uprv_strlen("\x26\x61\x70\x6f\x73\x3b");
 280                     break;
 281
 282                  /* Disallow C0 controls except TAB, CR, LF*/
 283                 case 0x00:
 284                 case 0x01:
 285                 case 0x02:
 286                 case 0x03:
 287                 case 0x04:
 288                 case 0x05:
 289                 case 0x06:
 290                 case 0x07:
 291                 case 0x08:
 292                 /*case 0x09:*/
 293                 /*case 0x0A: */
 294                 case 0x0B:
 295                 case 0x0C:
 296                 /*case 0x0D:*/
 297                 case 0x0E:
 298                 case 0x0F:
 299                 case 0x10:
 300                 case 0x11:
 301                 case 0x12:
 302                 case 0x13:
 303                 case 0x14:
 304                 case 0x15:
 305                 case 0x16:
 306                 case 0x17:
 307                 case 0x18:
 308                 case 0x19:
 309                 case 0x1A:
 310                 case 0x1B:
 311                 case 0x1C:
 312                 case 0x1D:
 313                 case 0x1E:
 314                 case 0x1F:
 315                     *status = U_ILLEGAL_CHAR_FOUND;
 316                     fprintf(stderr, "Illegal Character \\u%04X!\n",(int)c);
 317                     uprv_free(dest);
 318                     return NULL;
 319                 default:
 320                     dest[destLen++]=(char)c;
 321                 }
 322             }else{
 323                 UBool isError = FALSE;
 324                 U8_APPEND((unsigned char*)dest,destLen,destCap,c,isError);
 325                 if(isError){
 326                     *status = U_ILLEGAL_CHAR_FOUND;
 327                     fprintf(stderr, "Illegal Character \\U%08X!\n",(int)c);
 328                     uprv_free(dest);
 329                     return NULL;
 330                 }
 331             }
 332         }else{
 333             destCap += destLen;
 334
 335             temp = (char*) uprv_malloc(sizeof(char)*destCap);
 336             if(temp==NULL){
 337                 *status=U_MEMORY_ALLOCATION_ERROR;
 338                 uprv_free(dest);
 339                 return NULL;
 340             }
 341             uprv_memmove(temp,dest,destLen);
 342             destLen=0;
 343             uprv_free(dest);
 344             dest=temp;
 345             temp=NULL;
 346         }
 347
 348     }
 349     *destLength = destLen;
 350     return dest;
 351 }
 352
 353 #define ASTERISK 0x002A
 354 #define SPACE    0x0020
 355 #define CR       0x000A
 356 #define LF       0x000D
 357 #define AT_SIGN  0x0040
 358
 359 static void
 360 trim(char **src, int32_t *len){
 361
 362     char *s = NULL;
 363     int32_t i = 0;
 364     if(src == NULL || *src == NULL){
 365         return;
 366     }
 367     s = *src;
 368     /* trim from the end */
 369     for( i=(*len-1); i>= 0; i--){
 370         switch(s[i]){
 371         case ASTERISK:
 372         case SPACE:
 373         case CR:
 374         case LF:
 375             s[i] = 0;
 376             continue;
 377         default:
 378             break;
 379         }
 380         break;
 381
 382     }
 383     *len = i+1;
 384 }
 385
 386 static void
 387 print(UChar* src, int32_t srcLen,const char *tagStart,const char *tagEnd,  UErrorCode *status){
 388     int32_t bufCapacity   = srcLen*4;
 389     char *buf       = NULL;
 390     int32_t bufLen = 0;
 391
 392     if(U_FAILURE(*status)){
 393         return;
 394     }
 395
 396     buf = (char*) (uprv_malloc(bufCapacity));
 397     if(buf==0){
 398         fprintf(stderr, "Could not allocate memory!!");
 399         exit(U_MEMORY_ALLOCATION_ERROR);
 400     }
 401     buf = convertAndEscape(&buf, bufCapacity, &bufLen, src, srcLen,status);
 402     if(U_SUCCESS(*status)){
 403         trim(&buf,&bufLen);
 404         write_utf8_file(out,UnicodeString(tagStart));
 405         write_utf8_file(out,UnicodeString(buf, bufLen, "UTF-8"));
 406         write_utf8_file(out,UnicodeString(tagEnd));
 407         write_utf8_file(out,UnicodeString("\n"));
 408
 409     }
 410 }
 411 static void
 412 printNoteElements(struct UString *src, UErrorCode *status){
 413
 414 #if UCONFIG_NO_REGULAR_EXPRESSIONS==0 /* donot compile when no RegularExpressions are available */
 415
 416     int32_t capacity = 0;
 417     UChar* note = NULL;
 418     int32_t noteLen = 0;
 419     int32_t count = 0,i;
 420
 421     if(src == NULL){
 422         return;
 423     }
 424
 425     capacity = src->fLength;
 426     note  = (UChar*) uprv_malloc(U_SIZEOF_UCHAR * capacity);
 427
 428     count = getCount(src->fChars,src->fLength, UPC_NOTE, status);
 429     if(U_FAILURE(*status)){
 430         uprv_free(note);
 431         return;
 432     }
 433     for(i=0; i < count; i++){
 434         noteLen =  getAt(src->fChars,src->fLength, &note, capacity, i, UPC_NOTE, status);
 435         if(U_FAILURE(*status)){
 436             uprv_free(note);
 437             return;
 438         }
 439         if(noteLen > 0){
 440             write_tabs(out);
 441             print(note, noteLen,"<note>", "</note>", status);
 442         }
 443     }
 444     uprv_free(note);
 445 #else
 446
 447     fprintf(stderr, "Warning: Could not output comments to XLIFF file. ICU has been built without RegularExpression support.\n");
 448
 449 #endif /* UCONFIG_NO_REGULAR_EXPRESSIONS */
 450
 451 }
 452
 453 static void printAttribute(const char *name, const char *value, int32_t /*len*/)
 454 {
 455     write_utf8_file(out, UnicodeString(" "));
 456     write_utf8_file(out, UnicodeString(name));
 457     write_utf8_file(out, UnicodeString(" = \""));
 458     write_utf8_file(out, UnicodeString(value));
 459     write_utf8_file(out, UnicodeString("\""));
 460 }
 461
 462 static void printAttribute(const char *name, const UnicodeString value, int32_t /*len*/)
 463 {
 464     write_utf8_file(out, UnicodeString(" "));
 465     write_utf8_file(out, UnicodeString(name));
 466     write_utf8_file(out, UnicodeString(" = \""));
 467     write_utf8_file(out, value);
 468     write_utf8_file(out, UnicodeString("\""));
 469 }
 470
 471 static void
 472 printComments(struct UString *src, const char *resName, UBool printTranslate, UErrorCode *status){
 473
 474 #if UCONFIG_NO_REGULAR_EXPRESSIONS==0 /* donot compile when no RegularExpressions are available */
 475
 476     if(status==NULL || U_FAILURE(*status)){
 477         return;
 478     }
 479
 480     int32_t capacity = src->fLength + 1;
 481     char* buf = NULL;
 482     int32_t bufLen = 0;
 483     UChar* desc  = (UChar*) uprv_malloc(U_SIZEOF_UCHAR * capacity);
 484     UChar* trans = (UChar*) uprv_malloc(U_SIZEOF_UCHAR * capacity);
 485
 486     int32_t descLen = 0, transLen=0;
 487     if(desc==NULL || trans==NULL){
 488         *status = U_MEMORY_ALLOCATION_ERROR;
 489         uprv_free(desc);
 490         uprv_free(trans);
 491         return;
 492     }
 493     src->fLength = removeCmtText(src->fChars, src->fLength, status);
 494     descLen  = getDescription(src->fChars,src->fLength, &desc, capacity, status);
 495     transLen = getTranslate(src->fChars,src->fLength, &trans, capacity, status);
 496
 497     /* first print translate attribute */
 498     if(transLen > 0){
 499         if(printTranslate){
 500             /* print translate attribute */
 501             buf = convertAndEscape(&buf, 0, &bufLen, trans, transLen, status);
 502             if(U_SUCCESS(*status)){
 503                 printAttribute("translate", UnicodeString(buf, bufLen, "UTF-8"), bufLen);
 504                 write_utf8_file(out,UnicodeString(">\n"));
 505             }
 506         }else if(getShowWarning()){
 507             fprintf(stderr, "Warning: Tranlate attribute for resource %s cannot be set. XLIFF prohibits it.\n", resName);
 508             /* no translate attribute .. just close the tag */
 509             write_utf8_file(out,UnicodeString(">\n"));
 510         }
 511     }else{
 512         /* no translate attribute .. just close the tag */
 513         write_utf8_file(out,UnicodeString(">\n"));
 514     }
 515
 516     if(descLen > 0){
 517         write_tabs(out);
 518         print(desc, descLen, "<!--", "-->", status);
 519     }
 520
 521     uprv_free(desc);
 522     uprv_free(trans);
 523 #else
 524
 525     fprintf(stderr, "Warning: Could not output comments to XLIFF file. ICU has been built without RegularExpression support.\n");
 526
 527 #endif /* UCONFIG_NO_REGULAR_EXPRESSIONS */
 528
 529 }
 530
 531 /*
 532  * Print out a containing element, like:
 533  * <trans-unit id = "blah" resname = "blah" restype = "x-id-alias" translate = "no">
 534  * <group id "calendar_gregorian" resname = "gregorian" restype = "x-icu-array">
 535  */
 536 static char *printContainer(struct SResource *res, const char *container, const char *restype, const char *mimetype, const char *id, UErrorCode *status)
 537 {
 538     char resKeyBuffer[8];
 539     const char *resname = NULL;
 540     char *sid = NULL;
 541
 542     write_tabs(out);
 543
 544     resname = res_getKeyString(srBundle, res, resKeyBuffer);
 545     if (resname != NULL && *resname != 0) {
 546         sid = getID(id, resname, sid);
 547     } else {
 548         sid = getID(id, NULL, sid);
 549     }
 550
 551     write_utf8_file(out, UnicodeString("<"));
 552     write_utf8_file(out, UnicodeString(container));
 553     printAttribute("id", sid, (int32_t) uprv_strlen(sid));
 554
 555     if (resname != NULL) {
 556         printAttribute("resname", resname, (int32_t) uprv_strlen(resname));
 557     }
 558
 559     if (mimetype != NULL) {
 560         printAttribute("mime-type", mimetype, (int32_t) uprv_strlen(mimetype));
 561     }
 562
 563     if (restype != NULL) {
 564         printAttribute("restype", restype, (int32_t) uprv_strlen(restype));
 565     }
 566
 567     tabCount += 1;
 568     if (res->fComment.fLength > 0) {
 569         /* printComments will print the closing ">\n" */
 570         printComments(&res->fComment, resname, TRUE, status);
 571     } else {
 572         write_utf8_file(out, UnicodeString(">\n"));
 573     }
 574
 575     return sid;
 576 }
 577
 578 /* Writing Functions */
 579
 580 static const char *trans_unit = "trans-unit";
 581 static const char *close_trans_unit = "</trans-unit>\n";
 582 static const char *source = "<source>";
 583 static const char *close_source = "</source>\n";
 584 static const char *group = "group";
 585 static const char *close_group = "</group>\n";
 586
 587 static const char *bin_unit = "bin-unit";
 588 static const char *close_bin_unit = "</bin-unit>\n";
 589 static const char *bin_source = "<bin-source>\n";
 590 static const char *close_bin_source = "</bin-source>\n";
 591 static const char *external_file = "<external-file";
 592 /*static const char *close_external_file = "</external-file>\n";*/
 593 static const char *internal_file = "<internal-file";
 594 static const char *close_internal_file = "</internal-file>\n";
 595
 596 static const char *application_mimetype = "application"; /* add "/octet-stream"? */
 597
 598 static const char *alias_restype     = "x-icu-alias";
 599 static const char *array_restype     = "x-icu-array";
 600 static const char *binary_restype    = "x-icu-binary";
 601 static const char *integer_restype   = "x-icu-integer";
 602 static const char *intvector_restype = "x-icu-intvector";
 603 static const char *table_restype     = "x-icu-table";
 604
 605 static void
 606 string_write_xml(struct SResource *res, const char* id, const char* /*language*/, UErrorCode *status) {
 607
 608     char *sid = NULL;
 609     char* buf = NULL;
 610     int32_t bufLen = 0;
 611
 612     if(status==NULL || U_FAILURE(*status)){
 613         return;
 614     }
 615
 616     sid = printContainer(res, trans_unit, NULL, NULL, id, status);
 617
 618     write_tabs(out);
 619
 620     write_utf8_file(out, UnicodeString(source));
 621
 622     buf = convertAndEscape(&buf, 0, &bufLen, res->u.fString.fChars, res->u.fString.fLength, status);
 623
 624     if (U_FAILURE(*status)) {
 625         return;
 626     }
 627
 628     write_utf8_file(out, UnicodeString(buf, bufLen, "UTF-8"));
 629     write_utf8_file(out, UnicodeString(close_source));
 630
 631     printNoteElements(&res->fComment, status);
 632
 633     tabCount -= 1;
 634     write_tabs(out);
 635
 636     write_utf8_file(out, UnicodeString(close_trans_unit));
 637
 638     uprv_free(buf);
 639     uprv_free(sid);
 640 }
 641
 642 static void
 643 alias_write_xml(struct SResource *res, const char* id, const char* /*language*/, UErrorCode *status) {
 644     char *sid = NULL;
 645     char* buf = NULL;
 646     int32_t bufLen=0;
 647
 648     sid = printContainer(res, trans_unit, alias_restype, NULL, id, status);
 649
 650     write_tabs(out);
 651
 652     write_utf8_file(out, UnicodeString(source));
 653
 654     buf = convertAndEscape(&buf, 0, &bufLen, res->u.fString.fChars, res->u.fString.fLength, status);
 655
 656     if(U_FAILURE(*status)){
 657         return;
 658     }
 659     write_utf8_file(out, UnicodeString(buf, bufLen, "UTF-8"));
 660     write_utf8_file(out, UnicodeString(close_source));
 661
 662     printNoteElements(&res->fComment, status);
 663
 664     tabCount -= 1;
 665     write_tabs(out);
 666
 667     write_utf8_file(out, UnicodeString(close_trans_unit));
 668
 669     uprv_free(buf);
 670     uprv_free(sid);
 671 }
 672
 673 static void
 674 array_write_xml(struct SResource *res, const char* id, const char* language, UErrorCode *status) {
 675     char* sid = NULL;
 676     int index = 0;
 677
 678     struct SResource *current = NULL;
 679
 680     sid = printContainer(res, group, array_restype, NULL, id, status);
 681
 682     current = res->u.fArray.fFirst;
 683
 684     while (current != NULL) {
 685         char c[256] = {0};
 686         char* subId = NULL;
 687
 688         itostr(c, index, 10, 0);
 689         index += 1;
 690         subId = getID(sid, c, subId);
 691
 692         res_write_xml(current, subId, language, FALSE, status);
 693         uprv_free(subId);
 694         subId = NULL;
 695
 696         if(U_FAILURE(*status)){
 697             return;
 698         }
 699
 700         current = current->fNext;
 701     }
 702
 703     tabCount -= 1;
 704     write_tabs(out);
 705     write_utf8_file(out, UnicodeString(close_group));
 706
 707     uprv_free(sid);
 708 }
 709
 710 static void
 711 intvector_write_xml(struct SResource *res, const char* id, const char* /*language*/, UErrorCode *status) {
 712     char* sid = NULL;
 713     char* ivd = NULL;
 714     uint32_t i=0;
 715     uint32_t len=0;
 716     char buf[256] = {'0'};
 717
 718     sid = printContainer(res, group, intvector_restype, NULL, id, status);
 719
 720     for(i = 0; i < res->u.fIntVector.fCount; i += 1) {
 721         char c[256] = {0};
 722
 723         itostr(c, i, 10, 0);
 724         ivd = getID(sid, c, ivd);
 725         len = itostr(buf, res->u.fIntVector.fArray[i], 10, 0);
 726
 727         write_tabs(out);
 728         write_utf8_file(out, UnicodeString("<"));
 729         write_utf8_file(out, UnicodeString(trans_unit));
 730
 731         printAttribute("id", ivd, (int32_t)uprv_strlen(ivd));
 732         printAttribute("restype", integer_restype, (int32_t) strlen(integer_restype));
 733
 734         write_utf8_file(out, UnicodeString(">\n"));
 735
 736         tabCount += 1;
 737         write_tabs(out);
 738         write_utf8_file(out, UnicodeString(source));
 739
 740         write_utf8_file(out, UnicodeString(buf, len));
 741
 742         write_utf8_file(out, UnicodeString(close_source));
 743         tabCount -= 1;
 744         write_tabs(out);
 745         write_utf8_file(out, UnicodeString(close_trans_unit));
 746
 747         uprv_free(ivd);
 748         ivd = NULL;
 749     }
 750
 751     tabCount -= 1;
 752     write_tabs(out);
 753
 754     write_utf8_file(out, UnicodeString(close_group));
 755     uprv_free(sid);
 756     sid = NULL;
 757 }
 758
 759 static void
 760 int_write_xml(struct SResource *res, const char* id, const char* /*language*/, UErrorCode *status) {
 761     char* sid = NULL;
 762     char buf[256] = {0};
 763     uint32_t len = 0;
 764
 765     sid = printContainer(res, trans_unit, integer_restype, NULL, id, status);
 766
 767     write_tabs(out);
 768
 769     write_utf8_file(out, UnicodeString(source));
 770
 771     len = itostr(buf, res->u.fIntValue.fValue, 10, 0);
 772     write_utf8_file(out, UnicodeString(buf, len));
 773
 774     write_utf8_file(out, UnicodeString(close_source));
 775
 776     printNoteElements(&res->fComment, status);
 777
 778     tabCount -= 1;
 779     write_tabs(out);
 780
 781     write_utf8_file(out, UnicodeString(close_trans_unit));
 782
 783     uprv_free(sid);
 784     sid = NULL;
 785 }
 786
 787 static void
 788 bin_write_xml(struct SResource *res, const char* id, const char* /*language*/, UErrorCode *status) {
 789     const char* m_type = application_mimetype;
 790     char* sid = NULL;
 791     uint32_t crc = 0xFFFFFFFF;
 792
 793     char fileName[1024] ={0};
 794     int32_t tLen = ( outDir == NULL) ? 0 :(int32_t)uprv_strlen(outDir);
 795     char* fn =  (char*) uprv_malloc(sizeof(char) * (tLen+1024 +
 796                                                     (res->u.fBinaryValue.fFileName !=NULL ?
 797                                                     uprv_strlen(res->u.fBinaryValue.fFileName) :0)));
 798     const char* ext = NULL;
 799
 800     char* f = NULL;
 801
 802     fn[0]=0;
 803
 804     if(res->u.fBinaryValue.fFileName != NULL){
 805         uprv_strcpy(fileName, res->u.fBinaryValue.fFileName);
 806         f = uprv_strrchr(fileName, '\\');
 807
 808         if (f != NULL) {
 809             f++;
 810         } else {
 811             f = fileName;
 812         }
 813
 814         ext = uprv_strrchr(fileName, '.');
 815
 816         if (ext == NULL) {
 817             fprintf(stderr, "Error: %s is an unknown binary filename type.\n", fileName);
 818             exit(U_ILLEGAL_ARGUMENT_ERROR);
 819         }
 820
 821         if(uprv_strcmp(ext, ".jpg")==0 || uprv_strcmp(ext, ".jpeg")==0 || uprv_strcmp(ext, ".gif")==0 ){
 822             m_type = "image";
 823         } else if(uprv_strcmp(ext, ".wav")==0 || uprv_strcmp(ext, ".au")==0 ){
 824             m_type = "audio";
 825         } else if(uprv_strcmp(ext, ".avi")==0 || uprv_strcmp(ext, ".mpg")==0 || uprv_strcmp(ext, ".mpeg")==0){
 826             m_type = "video";
 827         } else if(uprv_strcmp(ext, ".txt")==0 || uprv_strcmp(ext, ".text")==0){
 828             m_type = "text";
 829         }
 830
 831         sid = printContainer(res, bin_unit, binary_restype, m_type, id, status);
 832
 833         write_tabs(out);
 834
 835         write_utf8_file(out, UnicodeString(bin_source));
 836
 837         tabCount+= 1;
 838         write_tabs(out);
 839
 840         write_utf8_file(out, UnicodeString(external_file));
 841         printAttribute("href", f, (int32_t)uprv_strlen(f));
 842         write_utf8_file(out, UnicodeString("/>\n"));
 843         tabCount -= 1;
 844         write_tabs(out);
 845
 846         write_utf8_file(out, UnicodeString(close_bin_source));
 847
 848         printNoteElements(&res->fComment, status);
 849         tabCount -= 1;
 850         write_tabs(out);
 851         write_utf8_file(out, UnicodeString(close_bin_unit));
 852     } else {
 853         char temp[256] = {0};
 854         uint32_t i = 0;
 855         int32_t len=0;
 856
 857         sid = printContainer(res, bin_unit, binary_restype, m_type, id, status);
 858
 859         write_tabs(out);
 860         write_utf8_file(out, UnicodeString(bin_source));
 861
 862         tabCount += 1;
 863         write_tabs(out);
 864
 865         write_utf8_file(out, UnicodeString(internal_file));
 866         printAttribute("form", application_mimetype, (int32_t) uprv_strlen(application_mimetype));
 867
 868         while(i <res->u.fBinaryValue.fLength){
 869             len = itostr(temp, res->u.fBinaryValue.fData[i], 16, 2);
 870             crc = computeCRC(temp, len, crc);
 871             i++;
 872         }
 873
 874         len = itostr(temp, crc, 10, 0);
 875         printAttribute("crc", temp, len);
 876
 877         write_utf8_file(out, UnicodeString(">"));
 878
 879         i = 0;
 880         while(i <res->u.fBinaryValue.fLength){
 881             len = itostr(temp, res->u.fBinaryValue.fData[i], 16, 2);
 882             write_utf8_file(out, UnicodeString(temp));
 883             i += 1;
 884         }
 885
 886         write_utf8_file(out, UnicodeString(close_internal_file));
 887
 888         tabCount -= 2;
 889         write_tabs(out);
 890
 891         write_utf8_file(out, UnicodeString(close_bin_source));
 892         printNoteElements(&res->fComment, status);
 893
 894         tabCount -= 1;
 895         write_tabs(out);
 896         write_utf8_file(out, UnicodeString(close_bin_unit));
 897
 898         uprv_free(sid);
 899         sid = NULL;
 900     }
 901
 902     uprv_free(fn);
 903 }
 904
 905
 906
 907 static void
 908 table_write_xml(struct SResource *res, const char* id, const char* language, UBool isTopLevel, UErrorCode *status) {
 909
 910     uint32_t  i         = 0;
 911
 912     struct SResource *current = NULL;
 913     char* sid = NULL;
 914
 915     if (U_FAILURE(*status)) {
 916         return ;
 917     }
 918
 919     sid = printContainer(res, group, table_restype, NULL, id, status);
 920
 921     if(isTopLevel) {
 922         sid[0] = '\0';
 923     }
 924
 925     current = res->u.fTable.fFirst;
 926     i = 0;
 927
 928     while (current != NULL) {
 929         res_write_xml(current, sid, language, FALSE, status);
 930
 931         if(U_FAILURE(*status)){
 932             return;
 933         }
 934
 935         i += 1;
 936         current = current->fNext;
 937     }
 938
 939     tabCount -= 1;
 940     write_tabs(out);
 941
 942     write_utf8_file(out, UnicodeString(close_group));
 943
 944     uprv_free(sid);
 945     sid = NULL;
 946 }
 947
 948 void
 949 res_write_xml(struct SResource *res, const char* id,  const char* language, UBool isTopLevel, UErrorCode *status) {
 950
 951     if (U_FAILURE(*status)) {
 952         return ;
 953     }
 954
 955     if (res != NULL) {
 956         switch (res->fType) {
 957         case URES_STRING:
 958              string_write_xml    (res, id, language, status);
 959              return;
 960
 961         case URES_ALIAS:
 962              alias_write_xml     (res, id, language, status);
 963              return;
 964
 965         case URES_INT_VECTOR:
 966              intvector_write_xml (res, id, language, status);
 967              return;
 968
 969         case URES_BINARY:
 970              bin_write_xml       (res, id, language, status);
 971              return;
 972
 973         case URES_INT:
 974              int_write_xml       (res, id, language, status);
 975              return;
 976
 977         case URES_ARRAY:
 978              array_write_xml     (res, id, language, status);
 979              return;
 980
 981         case URES_TABLE:
 982              table_write_xml     (res, id, language, isTopLevel, status);
 983              return;
 984
 985         default:
 986             break;
 987         }
 988     }
 989
 990     *status = U_INTERNAL_PROGRAM_ERROR;
 991 }
 992
 993 void
 994 bundle_write_xml(struct SRBRoot *bundle, const char *outputDir,const char* outputEnc, const char* filename,
 995                   char *writtenFilename, int writtenFilenameLen,
 996                   const char* language, const char* outFileName, UErrorCode *status) {
 997
 998     char* xmlfileName = NULL;
 999     char* outputFileName = NULL;
1000     char* originalFileName = NULL;
1001     const char* fileStart = "<file xml:space = \"preserve\" source-language = \"";
1002     const char* file1 = "\" datatype = \"x-icu-resource-bundle\" ";
1003     const char* file2 = "original = \"";
1004     const char* file4 = "\" date = \"";
1005     const char* fileEnd = "</file>\n";
1006     const char* headerStart = "<header>\n";
1007     const char* headerEnd = "</header>\n";
1008     const char* bodyStart = "<body>\n";
1009     const char* bodyEnd = "</body>\n";
1010
1011     const char *tool_start = "<tool";
1012     const char *tool_id = "genrb-" GENRB_VERSION "-icu-" U_ICU_VERSION;
1013     const char *tool_name = "genrb";
1014
1015     char* temp = NULL;
1016     char* lang = NULL;
1017     const char* pos = NULL;
1018     int32_t first, index;
1019     time_t currTime;
1020     char timeBuf[128];
1021
1022     outDir = outputDir;
1023
1024     srBundle = bundle;
1025
1026     pos = uprv_strrchr(filename, '\\');
1027     if(pos != NULL) {
1028         first = (int32_t)(pos - filename + 1);
1029     } else {
1030         first = 0;
1031     }
1032     index = (int32_t)(uprv_strlen(filename) - uprv_strlen(textExt) - first);
1033     originalFileName = (char *)uprv_malloc(sizeof(char)*index+1);
1034     uprv_memset(originalFileName, 0, sizeof(char)*index+1);
1035     uprv_strncpy(originalFileName, filename + first, index);
1036
1037     if(uprv_strcmp(originalFileName, srBundle->fLocale) != 0) {
1038         fprintf(stdout, "Warning: The file name is not same as the resource name!\n");
1039     }
1040
1041     temp = originalFileName;
1042     originalFileName = (char *)uprv_malloc(sizeof(char)* (uprv_strlen(temp)+uprv_strlen(textExt)) + 1);
1043     uprv_memset(originalFileName, 0, sizeof(char)* (uprv_strlen(temp)+uprv_strlen(textExt)) + 1);
1044     uprv_strcat(originalFileName, temp);
1045     uprv_strcat(originalFileName, textExt);
1046     uprv_free(temp);
1047     temp = NULL;
1048
1049
1050     if (language == NULL) {
1051 /*        lang = parseFilename(filename, lang);
1052         if (lang == NULL) {*/
1053             /* now check if locale name is valid or not
1054              * this is to cater for situation where
1055              * pegasusServer.txt contains
1056              *
1057              * en{
1058              *      ..
1059              * }
1060              */
1061              lang = parseFilename(srBundle->fLocale, lang);
1062              /*
1063               * Neither  the file name nor the table name inside the
1064               * txt file contain a valid country and language codes
1065               * throw an error.
1066               * pegasusServer.txt contains
1067               *
1068               * testelements{
1069               *     ....
1070               * }
1071               */
1072              if(lang==NULL){
1073                  fprintf(stderr, "Error: The file name and table name do not contain a valid language code. Please use -l option to specify it.\n");
1074                  exit(U_ILLEGAL_ARGUMENT_ERROR);
1075              }
1076        /* }*/
1077     } else {
1078         lang = (char *)uprv_malloc(sizeof(char)*uprv_strlen(language) +1);
1079         uprv_memset(lang, 0, sizeof(char)*uprv_strlen(language) +1);
1080         uprv_strcpy(lang, language);
1081     }
1082
1083     if(outFileName) {
1084         outputFileName = (char *)uprv_malloc(sizeof(char)*uprv_strlen(outFileName) + 1);
1085         uprv_memset(outputFileName, 0, sizeof(char)*uprv_strlen(outFileName) + 1);
1086         uprv_strcpy(outputFileName,outFileName);
1087     } else {
1088         outputFileName = (char *)uprv_malloc(sizeof(char)*uprv_strlen(srBundle->fLocale) + 1);
1089         uprv_memset(outputFileName, 0, sizeof(char)*uprv_strlen(srBundle->fLocale) + 1);
1090         uprv_strcpy(outputFileName,srBundle->fLocale);
1091     }
1092
1093     if(outputDir) {
1094         xmlfileName = (char *)uprv_malloc(sizeof(char)*(uprv_strlen(outputDir) + uprv_strlen(outputFileName) + uprv_strlen(xliffExt) + 1) +1);
1095         uprv_memset(xmlfileName, 0, sizeof(char)*(uprv_strlen(outputDir)+ uprv_strlen(outputFileName) + uprv_strlen(xliffExt) + 1) +1);
1096     } else {
1097         xmlfileName = (char *)uprv_malloc(sizeof(char)*(uprv_strlen(outputFileName) + uprv_strlen(xliffExt)) +1);
1098         uprv_memset(xmlfileName, 0, sizeof(char)*(uprv_strlen(outputFileName) + uprv_strlen(xliffExt)) +1);
1099     }
1100
1101     if(outputDir){
1102         uprv_strcpy(xmlfileName, outputDir);
1103         if(outputDir[uprv_strlen(outputDir)-1] !=U_FILE_SEP_CHAR){
1104             uprv_strcat(xmlfileName,U_FILE_SEP_STRING);
1105         }
1106     }
1107     uprv_strcat(xmlfileName,outputFileName);
1108     uprv_strcat(xmlfileName,xliffExt);
1109
1110     if (writtenFilename) {
1111         uprv_strncpy(writtenFilename, xmlfileName, writtenFilenameLen);
1112     }
1113
1114     if (U_FAILURE(*status)) {
1115         goto cleanup_bundle_write_xml;
1116     }
1117
1118     out= T_FileStream_open(xmlfileName,"w");
1119
1120     if(out==NULL){
1121         *status = U_FILE_ACCESS_ERROR;
1122         goto cleanup_bundle_write_xml;
1123     }
1124     write_utf8_file(out, xmlHeader);
1125
1126     if(outputEnc && *outputEnc!='\0'){
1127         /* store the output encoding */
1128         enc = outputEnc;
1129         conv=ucnv_open(enc,status);
1130         if(U_FAILURE(*status)){
1131             goto cleanup_bundle_write_xml;
1132         }
1133     }
1134     write_utf8_file(out, bundleStart);
1135     write_tabs(out);
1136     write_utf8_file(out, fileStart);
1137     /* check if lang and language are the same */
1138     if(language != NULL && uprv_strcmp(lang, srBundle->fLocale)!=0){
1139         fprintf(stderr,"Warning: The top level tag in the resource and language specified are not the same. Please check the input.\n");
1140     }
1141     write_utf8_file(out, UnicodeString(lang));
1142     write_utf8_file(out, UnicodeString(file1));
1143     write_utf8_file(out, UnicodeString(file2));
1144     write_utf8_file(out, UnicodeString(originalFileName));
1145     write_utf8_file(out, UnicodeString(file4));
1146
1147     time(&currTime);
1148     strftime(timeBuf, sizeof(timeBuf), "%Y-%m-%dT%H:%M:%SZ", gmtime(&currTime));
1149     write_utf8_file(out, UnicodeString(timeBuf));
1150     write_utf8_file(out, UnicodeString("\">\n"));
1151
1152     tabCount += 1;
1153     write_tabs(out);
1154     write_utf8_file(out, headerStart);
1155
1156     tabCount += 1;
1157     write_tabs(out);
1158
1159     write_utf8_file(out, tool_start);
1160     printAttribute("tool-id", tool_id, (int32_t) uprv_strlen(tool_id));
1161     printAttribute("tool-name", tool_name, (int32_t) uprv_strlen(tool_name));
1162     write_utf8_file(out, UnicodeString("/>\n"));
1163
1164     tabCount -= 1;
1165     write_tabs(out);
1166
1167     write_utf8_file(out, UnicodeString(headerEnd));
1168
1169     write_tabs(out);
1170     tabCount += 1;
1171
1172     write_utf8_file(out, UnicodeString(bodyStart));
1173
1174
1175     res_write_xml(bundle->fRoot, bundle->fLocale, lang, TRUE, status);
1176
1177     tabCount -= 1;
1178     write_tabs(out);
1179
1180     write_utf8_file(out, UnicodeString(bodyEnd));
1181     tabCount--;
1182     write_tabs(out);
1183     write_utf8_file(out, UnicodeString(fileEnd));
1184     tabCount--;
1185     write_tabs(out);
1186     write_utf8_file(out, UnicodeString(bundleEnd));
1187     T_FileStream_close(out);
1188
1189     ucnv_close(conv);
1190
1191 cleanup_bundle_write_xml:
1192     uprv_free(originalFileName);
1193     uprv_free(lang);
1194     if(xmlfileName != NULL) {
1195         uprv_free(xmlfileName);
1196     }
1197     if(outputFileName != NULL){
1198         uprv_free(outputFileName);
1199     }
1200 }