source/tools/genrb/wrtxml.cpp

   1 // Copyright (C) 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*
   4 *******************************************************************************
   5 *
   6 *   Copyright (C) 2002-2015, International Business Machines
   7 *   Corporation and others.  All Rights Reserved.
   8 *
   9 *******************************************************************************
  10 *
  11 * File wrtxml.cpp
  12 *
  13 * Modification History:
  14 *
  15 *   Date        Name        Description
  16 *   10/01/02    Ram         Creation.
  17 *   02/07/08    Spieth      Correct XLIFF generation on EBCDIC platform
  18 *
  19 *******************************************************************************
  20 */
  21
  22 // Safer use of UnicodeString.
  23 #ifndef UNISTR_FROM_CHAR_EXPLICIT
  24 #   define UNISTR_FROM_CHAR_EXPLICIT explicit
  25 #endif
  26
  27 // Less important, but still a good idea.
  28 #ifndef UNISTR_FROM_STRING_EXPLICIT
  29 #   define UNISTR_FROM_STRING_EXPLICIT explicit
  30 #endif
  31
  32 #include "reslist.h"
  33 #include "unewdata.h"
  34 #include "unicode/ures.h"
  35 #include "errmsg.h"
  36 #include "filestrm.h"
  37 #include "cstring.h"
  38 #include "unicode/ucnv.h"
  39 #include "genrb.h"
  40 #include "rle.h"
  41 #include "uhash.h"
  42 #include "uresimp.h"
  43 #include "unicode/ustring.h"
  44 #include "unicode/uchar.h"
  45 #include "ustr.h"
  46 #include "prscmnts.h"
  47 #include "unicode/unistr.h"
  48 #include "unicode/utf8.h"
  49 #include "unicode/utf16.h"
  50 #include <time.h>
  51
  52 U_NAMESPACE_USE
  53
  54 static int tabCount = 0;
  55
  56 static FileStream* out=NULL;
  57 static struct SRBRoot* srBundle ;
  58 static const char* outDir = NULL;
  59 static const char* enc ="";
  60 static UConverter* conv = NULL;
  61
  62 const char* const* ISOLanguages;
  63 const char* const* ISOCountries;
  64 const char* textExt = ".txt";
  65 const char* xliffExt = ".xlf";
  66
  67 static int32_t write_utf8_file(FileStream* fileStream, UnicodeString outString)
  68 {
  69     UErrorCode status = U_ZERO_ERROR;
  70     int32_t len = 0;
  71
  72     // preflight to get the destination buffer size
  73     u_strToUTF8(NULL,
  74                 0,
  75                 &len,
  76                 outString.getBuffer(),
  77                 outString.length(),
  78                 &status);
  79
  80     // allocate the buffer
  81     char* dest = (char*)uprv_malloc(len);
  82     status = U_ZERO_ERROR;
  83
  84     // convert the data
  85     u_strToUTF8(dest,
  86                 len,
  87                 &len,
  88                 outString.getBuffer(),
  89                 outString.length(),
  90                 &status);
  91
  92     // write data to out file
  93     int32_t ret = T_FileStream_write(fileStream, dest, len);
  94     uprv_free(dest);
  95     return (ret);
  96 }
  97
  98 /*write indentation for formatting*/
  99 static void write_tabs(FileStream* os){
 100     int i=0;
 101     for(;i<=tabCount;i++){
 102         write_utf8_file(os,UnicodeString("    "));
 103     }
 104 }
 105
 106 /*get ID for each element. ID is globally unique.*/
 107 static char* getID(const char* id, const char* curKey, char* result) {
 108     if(curKey == NULL) {
 109         result = (char *)uprv_malloc(sizeof(char)*uprv_strlen(id) + 1);
 110         uprv_memset(result, 0, sizeof(char)*uprv_strlen(id) + 1);
 111         uprv_strcpy(result, id);
 112     } else {
 113         result = (char *)uprv_malloc(sizeof(char)*(uprv_strlen(id) + 1 + uprv_strlen(curKey)) + 1);
 114         uprv_memset(result, 0, sizeof(char)*(uprv_strlen(id) + 1 + uprv_strlen(curKey)) + 1);
 115         if(id[0]!='\0'){
 116             uprv_strcpy(result, id);
 117             uprv_strcat(result, "_");
 118         }
 119         uprv_strcat(result, curKey);
 120     }
 121     return result;
 122 }
 123
 124 /*compute CRC for binary code*/
 125 /* The code is from  http://www.theorem.com/java/CRC32.java
 126  * Calculates the CRC32 - 32 bit Cyclical Redundancy Check
 127  * <P> This check is used in numerous systems to verify the integrity
 128  * of information.  It's also used as a hashing function.  Unlike a regular
 129  * checksum, it's sensitive to the order of the characters.
 130  * It produces a 32 bit
 131  *
 132  * @author Michael Lecuyer (mjl@theorem.com)
 133  * @version 1.1 August 11, 1998
 134  */
 135
 136 /* ICU is not endian portable, because ICU data generated on big endian machines can be
 137  * ported to big endian machines but not to little endian machines and vice versa. The
 138  * conversion is not portable across platforms with different endianess.
 139  */
 140
 141 uint32_t computeCRC(const char *ptr, uint32_t len, uint32_t lastcrc){
 142     int32_t crc;
 143     uint32_t temp1;
 144     uint32_t temp2;
 145
 146     int32_t crc_ta[256];
 147     int i = 0;
 148     int j = 0;
 149     uint32_t crc2 = 0;
 150
 151 #define CRC32_POLYNOMIAL 0xEDB88320
 152
 153     /*build crc table*/
 154     for (i = 0; i <= 255; i++) {
 155         crc2 = i;
 156         for (j = 8; j > 0; j--) {
 157             if ((crc2 & 1) == 1) {
 158                 crc2 = (crc2 >> 1) ^ CRC32_POLYNOMIAL;
 159             } else {
 160                 crc2 >>= 1;
 161             }
 162         }
 163         crc_ta[i] = crc2;
 164     }
 165
 166     crc = lastcrc;
 167     while(len--!=0) {
 168         temp1 = (uint32_t)crc>>8;
 169         temp2 = crc_ta[(crc^*ptr) & 0xFF];
 170         crc = temp1^temp2;
 171         ptr++;
 172     }
 173     return(crc);
 174 }
 175
 176 static void strnrepchr(char* src, int32_t srcLen, char s, char r){
 177     int32_t i = 0;
 178     for(i=0;i<srcLen;i++){
 179         if(src[i]==s){
 180             src[i]=r;
 181         }
 182     }
 183 }
 184 /* Parse the filename, and get its language information.
 185  * If it fails to get the language information from the filename,
 186  * use "en" as the default value for language
 187  */
 188 static char* parseFilename(const char* id, char* /*lang*/) {
 189     int idLen = (int) uprv_strlen(id);
 190     char* localeID = (char*) uprv_malloc(idLen);
 191     int pos = 0;
 192     int canonCapacity = 0;
 193     char* canon = NULL;
 194     int canonLen = 0;
 195     /*int i;*/
 196     UErrorCode status = U_ZERO_ERROR;
 197     const char *ext = uprv_strchr(id, '.');
 198
 199     if(ext != NULL){
 200         pos = (int) (ext - id);
 201     } else {
 202         pos = idLen;
 203     }
 204     uprv_memcpy(localeID, id, pos);
 205     localeID[pos]=0; /* NUL terminate the string */
 206
 207     canonCapacity =pos*3;
 208     canon = (char*) uprv_malloc(canonCapacity);
 209     canonLen = uloc_canonicalize(localeID, canon, canonCapacity, &status);
 210
 211     if(U_FAILURE(status)){
 212         fprintf(stderr, "Could not canonicalize the locale ID: %s. Error: %s\n", localeID, u_errorName(status));
 213         exit(status);
 214     }
 215     strnrepchr(canon, canonLen, '_', '-');
 216     return canon;
 217 }
 218
 219 static const char* xmlHeader = "<?xml version=\"1.0\" encoding=\"utf-8\"?>\n";
 220 #if 0
 221 static const char* bundleStart = "<xliff version = \"1.2\" "
 222                                         "xmlns='urn:oasis:names:tc:xliff:document:1.2' "
 223                                         "xmlns:xsi='http://www.w3.org/2001/XMLSchema-instance' "
 224                                         "xsi:schemaLocation='urn:oasis:names:tc:xliff:document:1.2 xliff-core-1.2-transitional.xsd'>\n";
 225 #else
 226 static const char* bundleStart = "<xliff version = \"1.1\" "
 227                                         "xmlns='urn:oasis:names:tc:xliff:document:1.1' "
 228                                         "xmlns:xsi='http://www.w3.org/2001/XMLSchema-instance' "
 229                                         "xsi:schemaLocation='urn:oasis:names:tc:xliff:document:1.1 http://www.oasis-open.org/committees/xliff/documents/xliff-core-1.1.xsd'>\n";
 230 #endif
 231 static const char* bundleEnd   = "</xliff>\n";
 232
 233 void res_write_xml(struct SResource *res, const char* id, const char* language, UBool isTopLevel, UErrorCode *status);
 234
 235 static char* convertAndEscape(char** pDest, int32_t destCap, int32_t* destLength,
 236                               const UChar* src, int32_t srcLen, UErrorCode* status){
 237     int32_t srcIndex=0;
 238     char* dest=NULL;
 239     char* temp=NULL;
 240     int32_t destLen=0;
 241     UChar32 c = 0;
 242
 243     if(status==NULL || U_FAILURE(*status) || pDest==NULL  || srcLen==0 || src == NULL){
 244         return NULL;
 245     }
 246     dest =*pDest;
 247     if(dest==NULL || destCap <=0){
 248         destCap = srcLen * 8;
 249         dest = (char*) uprv_malloc(sizeof(char) * destCap);
 250         if(dest==NULL){
 251             *status=U_MEMORY_ALLOCATION_ERROR;
 252             return NULL;
 253         }
 254     }
 255
 256     dest[0]=0;
 257
 258     while(srcIndex<srcLen){
 259         U16_NEXT(src, srcIndex, srcLen, c);
 260
 261         if (U16_IS_LEAD(c) || U16_IS_TRAIL(c)) {
 262             *status = U_ILLEGAL_CHAR_FOUND;
 263             fprintf(stderr, "Illegal Surrogate! \n");
 264             uprv_free(dest);
 265             return NULL;
 266         }
 267
 268         if((destLen+U8_LENGTH(c)) < destCap){
 269
 270             /* ASCII Range */
 271             if(c <=0x007F){
 272                 switch(c) {
 273                 case '\x26':
 274                     uprv_strcpy(dest+( destLen),"\x26\x61\x6d\x70\x3b"); /* &amp;*/
 275                     destLen+=(int32_t)uprv_strlen("\x26\x61\x6d\x70\x3b");
 276                     break;
 277                 case '\x3c':
 278                     uprv_strcpy(dest+(destLen),"\x26\x6c\x74\x3b"); /* &lt;*/
 279                     destLen+=(int32_t)uprv_strlen("\x26\x6c\x74\x3b");
 280                     break;
 281                 case '\x3e':
 282                     uprv_strcpy(dest+(destLen),"\x26\x67\x74\x3b"); /* &gt;*/
 283                     destLen+=(int32_t)uprv_strlen("\x26\x67\x74\x3b");
 284                     break;
 285                 case '\x22':
 286                     uprv_strcpy(dest+(destLen),"\x26\x71\x75\x6f\x74\x3b"); /* &quot;*/
 287                     destLen+=(int32_t)uprv_strlen("\x26\x71\x75\x6f\x74\x3b");
 288                     break;
 289                 case '\x27':
 290                     uprv_strcpy(dest+(destLen),"\x26\x61\x70\x6f\x73\x3b"); /* &apos; */
 291                     destLen+=(int32_t)uprv_strlen("\x26\x61\x70\x6f\x73\x3b");
 292                     break;
 293
 294                  /* Disallow C0 controls except TAB, CR, LF*/
 295                 case 0x00:
 296                 case 0x01:
 297                 case 0x02:
 298                 case 0x03:
 299                 case 0x04:
 300                 case 0x05:
 301                 case 0x06:
 302                 case 0x07:
 303                 case 0x08:
 304                 /*case 0x09:*/
 305                 /*case 0x0A: */
 306                 case 0x0B:
 307                 case 0x0C:
 308                 /*case 0x0D:*/
 309                 case 0x0E:
 310                 case 0x0F:
 311                 case 0x10:
 312                 case 0x11:
 313                 case 0x12:
 314                 case 0x13:
 315                 case 0x14:
 316                 case 0x15:
 317                 case 0x16:
 318                 case 0x17:
 319                 case 0x18:
 320                 case 0x19:
 321                 case 0x1A:
 322                 case 0x1B:
 323                 case 0x1C:
 324                 case 0x1D:
 325                 case 0x1E:
 326                 case 0x1F:
 327                     *status = U_ILLEGAL_CHAR_FOUND;
 328                     fprintf(stderr, "Illegal Character \\u%04X!\n",(int)c);
 329                     uprv_free(dest);
 330                     return NULL;
 331                 default:
 332                     dest[destLen++]=(char)c;
 333                 }
 334             }else{
 335                 UBool isError = FALSE;
 336                 U8_APPEND((unsigned char*)dest,destLen,destCap,c,isError);
 337                 if(isError){
 338                     *status = U_ILLEGAL_CHAR_FOUND;
 339                     fprintf(stderr, "Illegal Character \\U%08X!\n",(int)c);
 340                     uprv_free(dest);
 341                     return NULL;
 342                 }
 343             }
 344         }else{
 345             destCap += destLen;
 346
 347             temp = (char*) uprv_malloc(sizeof(char)*destCap);
 348             if(temp==NULL){
 349                 *status=U_MEMORY_ALLOCATION_ERROR;
 350                 uprv_free(dest);
 351                 return NULL;
 352             }
 353             uprv_memmove(temp,dest,destLen);
 354             destLen=0;
 355             uprv_free(dest);
 356             dest=temp;
 357             temp=NULL;
 358         }
 359
 360     }
 361     *destLength = destLen;
 362     return dest;
 363 }
 364
 365 #define ASTERISK 0x002A
 366 #define SPACE    0x0020
 367 #define CR       0x000A
 368 #define LF       0x000D
 369 #define AT_SIGN  0x0040
 370
 371 static void
 372 trim(char **src, int32_t *len){
 373
 374     char *s = NULL;
 375     int32_t i = 0;
 376     if(src == NULL || *src == NULL){
 377         return;
 378     }
 379     s = *src;
 380     /* trim from the end */
 381     for( i=(*len-1); i>= 0; i--){
 382         switch(s[i]){
 383         case ASTERISK:
 384         case SPACE:
 385         case CR:
 386         case LF:
 387             s[i] = 0;
 388             continue;
 389         default:
 390             break;
 391         }
 392         break;
 393
 394     }
 395     *len = i+1;
 396 }
 397
 398 static void
 399 print(UChar* src, int32_t srcLen,const char *tagStart,const char *tagEnd,  UErrorCode *status){
 400     int32_t bufCapacity   = srcLen*4;
 401     char *buf       = NULL;
 402     int32_t bufLen = 0;
 403
 404     if(U_FAILURE(*status)){
 405         return;
 406     }
 407
 408     buf = (char*) (uprv_malloc(bufCapacity));
 409     if(buf==0){
 410         fprintf(stderr, "Could not allocate memory!!");
 411         exit(U_MEMORY_ALLOCATION_ERROR);
 412     }
 413     buf = convertAndEscape(&buf, bufCapacity, &bufLen, src, srcLen,status);
 414     if(U_SUCCESS(*status)){
 415         trim(&buf,&bufLen);
 416         write_utf8_file(out,UnicodeString(tagStart));
 417         write_utf8_file(out,UnicodeString(buf, bufLen, "UTF-8"));
 418         write_utf8_file(out,UnicodeString(tagEnd));
 419         write_utf8_file(out,UnicodeString("\n"));
 420
 421     }
 422 }
 423 static void
 424 printNoteElements(const UString *src, UErrorCode *status){
 425
 426 #if UCONFIG_NO_REGULAR_EXPRESSIONS==0 /* donot compile when no RegularExpressions are available */
 427
 428     int32_t capacity = 0;
 429     UChar* note = NULL;
 430     int32_t noteLen = 0;
 431     int32_t count = 0,i;
 432
 433     if(src == NULL){
 434         return;
 435     }
 436
 437     capacity = src->fLength;
 438     note  = (UChar*) uprv_malloc(U_SIZEOF_UCHAR * capacity);
 439
 440     count = getCount(src->fChars,src->fLength, UPC_NOTE, status);
 441     if(U_FAILURE(*status)){
 442         uprv_free(note);
 443         return;
 444     }
 445     for(i=0; i < count; i++){
 446         noteLen =  getAt(src->fChars,src->fLength, &note, capacity, i, UPC_NOTE, status);
 447         if(U_FAILURE(*status)){
 448             uprv_free(note);
 449             return;
 450         }
 451         if(noteLen > 0){
 452             write_tabs(out);
 453             print(note, noteLen,"<note>", "</note>", status);
 454         }
 455     }
 456     uprv_free(note);
 457 #else
 458
 459     fprintf(stderr, "Warning: Could not output comments to XLIFF file. ICU has been built without RegularExpression support.\n");
 460
 461 #endif /* UCONFIG_NO_REGULAR_EXPRESSIONS */
 462
 463 }
 464
 465 static void printAttribute(const char *name, const char *value, int32_t /*len*/)
 466 {
 467     write_utf8_file(out, UnicodeString(" "));
 468     write_utf8_file(out, UnicodeString(name));
 469     write_utf8_file(out, UnicodeString(" = \""));
 470     write_utf8_file(out, UnicodeString(value));
 471     write_utf8_file(out, UnicodeString("\""));
 472 }
 473
 474 static void printAttribute(const char *name, const UnicodeString value, int32_t /*len*/)
 475 {
 476     write_utf8_file(out, UnicodeString(" "));
 477     write_utf8_file(out, UnicodeString(name));
 478     write_utf8_file(out, UnicodeString(" = \""));
 479     write_utf8_file(out, value);
 480     write_utf8_file(out, UnicodeString("\""));
 481 }
 482
 483 static void
 484 printComments(struct UString *src, const char *resName, UBool printTranslate, UErrorCode *status){
 485
 486 #if UCONFIG_NO_REGULAR_EXPRESSIONS==0 /* donot compile when no RegularExpressions are available */
 487
 488     if(status==NULL || U_FAILURE(*status)){
 489         return;
 490     }
 491
 492     int32_t capacity = src->fLength + 1;
 493     char* buf = NULL;
 494     int32_t bufLen = 0;
 495     UChar* desc  = (UChar*) uprv_malloc(U_SIZEOF_UCHAR * capacity);
 496     UChar* trans = (UChar*) uprv_malloc(U_SIZEOF_UCHAR * capacity);
 497
 498     int32_t descLen = 0, transLen=0;
 499     if(desc==NULL || trans==NULL){
 500         *status = U_MEMORY_ALLOCATION_ERROR;
 501         uprv_free(desc);
 502         uprv_free(trans);
 503         return;
 504     }
 505     // TODO: make src const, stop modifying it in-place, make printContainer() take const resource, etc.
 506     src->fLength = removeCmtText(src->fChars, src->fLength, status);
 507     descLen  = getDescription(src->fChars,src->fLength, &desc, capacity, status);
 508     transLen = getTranslate(src->fChars,src->fLength, &trans, capacity, status);
 509
 510     /* first print translate attribute */
 511     if(transLen > 0){
 512         if(printTranslate){
 513             /* print translate attribute */
 514             buf = convertAndEscape(&buf, 0, &bufLen, trans, transLen, status);
 515             if(U_SUCCESS(*status)){
 516                 printAttribute("translate", UnicodeString(buf, bufLen, "UTF-8"), bufLen);
 517                 write_utf8_file(out,UnicodeString(">\n"));
 518             }
 519         }else if(getShowWarning()){
 520             fprintf(stderr, "Warning: Tranlate attribute for resource %s cannot be set. XLIFF prohibits it.\n", resName);
 521             /* no translate attribute .. just close the tag */
 522             write_utf8_file(out,UnicodeString(">\n"));
 523         }
 524     }else{
 525         /* no translate attribute .. just close the tag */
 526         write_utf8_file(out,UnicodeString(">\n"));
 527     }
 528
 529     if(descLen > 0){
 530         write_tabs(out);
 531         print(desc, descLen, "<!--", "-->", status);
 532     }
 533
 534     uprv_free(desc);
 535     uprv_free(trans);
 536 #else
 537
 538     fprintf(stderr, "Warning: Could not output comments to XLIFF file. ICU has been built without RegularExpression support.\n");
 539
 540 #endif /* UCONFIG_NO_REGULAR_EXPRESSIONS */
 541
 542 }
 543
 544 /*
 545  * Print out a containing element, like:
 546  * <trans-unit id = "blah" resname = "blah" restype = "x-id-alias" translate = "no">
 547  * <group id "calendar_gregorian" resname = "gregorian" restype = "x-icu-array">
 548  */
 549 static char *printContainer(SResource *res, const char *container, const char *restype, const char *mimetype, const char *id, UErrorCode *status)
 550 {
 551     const char *resname = NULL;
 552     char *sid = NULL;
 553
 554     write_tabs(out);
 555
 556     resname = res->getKeyString(srBundle);
 557     if (resname != NULL && *resname != 0) {
 558         sid = getID(id, resname, sid);
 559     } else {
 560         sid = getID(id, NULL, sid);
 561     }
 562
 563     write_utf8_file(out, UnicodeString("<"));
 564     write_utf8_file(out, UnicodeString(container));
 565     printAttribute("id", sid, (int32_t) uprv_strlen(sid));
 566
 567     if (resname != NULL) {
 568         printAttribute("resname", resname, (int32_t) uprv_strlen(resname));
 569     }
 570
 571     if (mimetype != NULL) {
 572         printAttribute("mime-type", mimetype, (int32_t) uprv_strlen(mimetype));
 573     }
 574
 575     if (restype != NULL) {
 576         printAttribute("restype", restype, (int32_t) uprv_strlen(restype));
 577     }
 578
 579     tabCount += 1;
 580     if (res->fComment.fLength > 0) {
 581         /* printComments will print the closing ">\n" */
 582         printComments(&res->fComment, resname, TRUE, status);
 583     } else {
 584         write_utf8_file(out, UnicodeString(">\n"));
 585     }
 586
 587     return sid;
 588 }
 589
 590 /* Writing Functions */
 591
 592 static const char *trans_unit = "trans-unit";
 593 static const char *close_trans_unit = "</trans-unit>\n";
 594 static const char *source = "<source>";
 595 static const char *close_source = "</source>\n";
 596 static const char *group = "group";
 597 static const char *close_group = "</group>\n";
 598
 599 static const char *bin_unit = "bin-unit";
 600 static const char *close_bin_unit = "</bin-unit>\n";
 601 static const char *bin_source = "<bin-source>\n";
 602 static const char *close_bin_source = "</bin-source>\n";
 603 static const char *external_file = "<external-file";
 604 /*static const char *close_external_file = "</external-file>\n";*/
 605 static const char *internal_file = "<internal-file";
 606 static const char *close_internal_file = "</internal-file>\n";
 607
 608 static const char *application_mimetype = "application"; /* add "/octet-stream"? */
 609
 610 static const char *alias_restype     = "x-icu-alias";
 611 static const char *array_restype     = "x-icu-array";
 612 static const char *binary_restype    = "x-icu-binary";
 613 static const char *integer_restype   = "x-icu-integer";
 614 static const char *intvector_restype = "x-icu-intvector";
 615 static const char *table_restype     = "x-icu-table";
 616
 617 static void
 618 string_write_xml(StringResource *res, const char* id, const char* /*language*/, UErrorCode *status) {
 619
 620     char *sid = NULL;
 621     char* buf = NULL;
 622     int32_t bufLen = 0;
 623
 624     if(status==NULL || U_FAILURE(*status)){
 625         return;
 626     }
 627
 628     sid = printContainer(res, trans_unit, NULL, NULL, id, status);
 629
 630     write_tabs(out);
 631
 632     write_utf8_file(out, UnicodeString(source));
 633
 634     buf = convertAndEscape(&buf, 0, &bufLen, res->getBuffer(), res->length(), status);
 635
 636     if (U_FAILURE(*status)) {
 637         return;
 638     }
 639
 640     write_utf8_file(out, UnicodeString(buf, bufLen, "UTF-8"));
 641     write_utf8_file(out, UnicodeString(close_source));
 642
 643     printNoteElements(&res->fComment, status);
 644
 645     tabCount -= 1;
 646     write_tabs(out);
 647
 648     write_utf8_file(out, UnicodeString(close_trans_unit));
 649
 650     uprv_free(buf);
 651     uprv_free(sid);
 652 }
 653
 654 static void
 655 alias_write_xml(AliasResource *res, const char* id, const char* /*language*/, UErrorCode *status) {
 656     char *sid = NULL;
 657     char* buf = NULL;
 658     int32_t bufLen=0;
 659
 660     sid = printContainer(res, trans_unit, alias_restype, NULL, id, status);
 661
 662     write_tabs(out);
 663
 664     write_utf8_file(out, UnicodeString(source));
 665
 666     buf = convertAndEscape(&buf, 0, &bufLen, res->getBuffer(), res->length(), status);
 667
 668     if(U_FAILURE(*status)){
 669         return;
 670     }
 671     write_utf8_file(out, UnicodeString(buf, bufLen, "UTF-8"));
 672     write_utf8_file(out, UnicodeString(close_source));
 673
 674     printNoteElements(&res->fComment, status);
 675
 676     tabCount -= 1;
 677     write_tabs(out);
 678
 679     write_utf8_file(out, UnicodeString(close_trans_unit));
 680
 681     uprv_free(buf);
 682     uprv_free(sid);
 683 }
 684
 685 static void
 686 array_write_xml(ArrayResource *res, const char* id, const char* language, UErrorCode *status) {
 687     char* sid = NULL;
 688     int index = 0;
 689
 690     struct SResource *current = NULL;
 691
 692     sid = printContainer(res, group, array_restype, NULL, id, status);
 693
 694     current = res->fFirst;
 695
 696     while (current != NULL) {
 697         char c[256] = {0};
 698         char* subId = NULL;
 699
 700         itostr(c, index, 10, 0);
 701         index += 1;
 702         subId = getID(sid, c, subId);
 703
 704         res_write_xml(current, subId, language, FALSE, status);
 705         uprv_free(subId);
 706         subId = NULL;
 707
 708         if(U_FAILURE(*status)){
 709             return;
 710         }
 711
 712         current = current->fNext;
 713     }
 714
 715     tabCount -= 1;
 716     write_tabs(out);
 717     write_utf8_file(out, UnicodeString(close_group));
 718
 719     uprv_free(sid);
 720 }
 721
 722 static void
 723 intvector_write_xml(IntVectorResource *res, const char* id, const char* /*language*/, UErrorCode *status) {
 724     char* sid = NULL;
 725     char* ivd = NULL;
 726     uint32_t i=0;
 727     uint32_t len=0;
 728     char buf[256] = {'0'};
 729
 730     sid = printContainer(res, group, intvector_restype, NULL, id, status);
 731
 732     for(i = 0; i < res->fCount; i += 1) {
 733         char c[256] = {0};
 734
 735         itostr(c, i, 10, 0);
 736         ivd = getID(sid, c, ivd);
 737         len = itostr(buf, res->fArray[i], 10, 0);
 738
 739         write_tabs(out);
 740         write_utf8_file(out, UnicodeString("<"));
 741         write_utf8_file(out, UnicodeString(trans_unit));
 742
 743         printAttribute("id", ivd, (int32_t)uprv_strlen(ivd));
 744         printAttribute("restype", integer_restype, (int32_t) strlen(integer_restype));
 745
 746         write_utf8_file(out, UnicodeString(">\n"));
 747
 748         tabCount += 1;
 749         write_tabs(out);
 750         write_utf8_file(out, UnicodeString(source));
 751
 752         write_utf8_file(out, UnicodeString(buf, len));
 753
 754         write_utf8_file(out, UnicodeString(close_source));
 755         tabCount -= 1;
 756         write_tabs(out);
 757         write_utf8_file(out, UnicodeString(close_trans_unit));
 758
 759         uprv_free(ivd);
 760         ivd = NULL;
 761     }
 762
 763     tabCount -= 1;
 764     write_tabs(out);
 765
 766     write_utf8_file(out, UnicodeString(close_group));
 767     uprv_free(sid);
 768     sid = NULL;
 769 }
 770
 771 static void
 772 int_write_xml(IntResource *res, const char* id, const char* /*language*/, UErrorCode *status) {
 773     char* sid = NULL;
 774     char buf[256] = {0};
 775     uint32_t len = 0;
 776
 777     sid = printContainer(res, trans_unit, integer_restype, NULL, id, status);
 778
 779     write_tabs(out);
 780
 781     write_utf8_file(out, UnicodeString(source));
 782
 783     len = itostr(buf, res->fValue, 10, 0);
 784     write_utf8_file(out, UnicodeString(buf, len));
 785
 786     write_utf8_file(out, UnicodeString(close_source));
 787
 788     printNoteElements(&res->fComment, status);
 789
 790     tabCount -= 1;
 791     write_tabs(out);
 792
 793     write_utf8_file(out, UnicodeString(close_trans_unit));
 794
 795     uprv_free(sid);
 796     sid = NULL;
 797 }
 798
 799 static void
 800 bin_write_xml(BinaryResource *res, const char* id, const char* /*language*/, UErrorCode *status) {
 801     const char* m_type = application_mimetype;
 802     char* sid = NULL;
 803     uint32_t crc = 0xFFFFFFFF;
 804
 805     char fileName[1024] ={0};
 806     int32_t tLen = ( outDir == NULL) ? 0 :(int32_t)uprv_strlen(outDir);
 807     char* fn =  (char*) uprv_malloc(sizeof(char) * (tLen+1024 +
 808                                                     (res->fFileName !=NULL ?
 809                                                     uprv_strlen(res->fFileName) :0)));
 810     const char* ext = NULL;
 811
 812     char* f = NULL;
 813
 814     fn[0]=0;
 815
 816     if(res->fFileName != NULL){
 817         uprv_strcpy(fileName, res->fFileName);
 818         f = uprv_strrchr(fileName, '\\');
 819
 820         if (f != NULL) {
 821             f++;
 822         } else {
 823             f = fileName;
 824         }
 825
 826         ext = uprv_strrchr(fileName, '.');
 827
 828         if (ext == NULL) {
 829             fprintf(stderr, "Error: %s is an unknown binary filename type.\n", fileName);
 830             exit(U_ILLEGAL_ARGUMENT_ERROR);
 831         }
 832
 833         if(uprv_strcmp(ext, ".jpg")==0 || uprv_strcmp(ext, ".jpeg")==0 || uprv_strcmp(ext, ".gif")==0 ){
 834             m_type = "image";
 835         } else if(uprv_strcmp(ext, ".wav")==0 || uprv_strcmp(ext, ".au")==0 ){
 836             m_type = "audio";
 837         } else if(uprv_strcmp(ext, ".avi")==0 || uprv_strcmp(ext, ".mpg")==0 || uprv_strcmp(ext, ".mpeg")==0){
 838             m_type = "video";
 839         } else if(uprv_strcmp(ext, ".txt")==0 || uprv_strcmp(ext, ".text")==0){
 840             m_type = "text";
 841         }
 842
 843         sid = printContainer(res, bin_unit, binary_restype, m_type, id, status);
 844
 845         write_tabs(out);
 846
 847         write_utf8_file(out, UnicodeString(bin_source));
 848
 849         tabCount+= 1;
 850         write_tabs(out);
 851
 852         write_utf8_file(out, UnicodeString(external_file));
 853         printAttribute("href", f, (int32_t)uprv_strlen(f));
 854         write_utf8_file(out, UnicodeString("/>\n"));
 855         tabCount -= 1;
 856         write_tabs(out);
 857
 858         write_utf8_file(out, UnicodeString(close_bin_source));
 859
 860         printNoteElements(&res->fComment, status);
 861         tabCount -= 1;
 862         write_tabs(out);
 863         write_utf8_file(out, UnicodeString(close_bin_unit));
 864     } else {
 865         char temp[256] = {0};
 866         uint32_t i = 0;
 867         int32_t len=0;
 868
 869         sid = printContainer(res, bin_unit, binary_restype, m_type, id, status);
 870
 871         write_tabs(out);
 872         write_utf8_file(out, UnicodeString(bin_source));
 873
 874         tabCount += 1;
 875         write_tabs(out);
 876
 877         write_utf8_file(out, UnicodeString(internal_file));
 878         printAttribute("form", application_mimetype, (int32_t) uprv_strlen(application_mimetype));
 879
 880         while(i <res->fLength){
 881             len = itostr(temp, res->fData[i], 16, 2);
 882             crc = computeCRC(temp, len, crc);
 883             i++;
 884         }
 885
 886         len = itostr(temp, crc, 10, 0);
 887         printAttribute("crc", temp, len);
 888
 889         write_utf8_file(out, UnicodeString(">"));
 890
 891         i = 0;
 892         while(i <res->fLength){
 893             len = itostr(temp, res->fData[i], 16, 2);
 894             write_utf8_file(out, UnicodeString(temp));
 895             i += 1;
 896         }
 897
 898         write_utf8_file(out, UnicodeString(close_internal_file));
 899
 900         tabCount -= 2;
 901         write_tabs(out);
 902
 903         write_utf8_file(out, UnicodeString(close_bin_source));
 904         printNoteElements(&res->fComment, status);
 905
 906         tabCount -= 1;
 907         write_tabs(out);
 908         write_utf8_file(out, UnicodeString(close_bin_unit));
 909
 910         uprv_free(sid);
 911         sid = NULL;
 912     }
 913
 914     uprv_free(fn);
 915 }
 916
 917
 918
 919 static void
 920 table_write_xml(TableResource *res, const char* id, const char* language, UBool isTopLevel, UErrorCode *status) {
 921
 922     uint32_t  i         = 0;
 923
 924     struct SResource *current = NULL;
 925     char* sid = NULL;
 926
 927     if (U_FAILURE(*status)) {
 928         return ;
 929     }
 930
 931     sid = printContainer(res, group, table_restype, NULL, id, status);
 932
 933     if(isTopLevel) {
 934         sid[0] = '\0';
 935     }
 936
 937     current = res->fFirst;
 938     i = 0;
 939
 940     while (current != NULL) {
 941         res_write_xml(current, sid, language, FALSE, status);
 942
 943         if(U_FAILURE(*status)){
 944             return;
 945         }
 946
 947         i += 1;
 948         current = current->fNext;
 949     }
 950
 951     tabCount -= 1;
 952     write_tabs(out);
 953
 954     write_utf8_file(out, UnicodeString(close_group));
 955
 956     uprv_free(sid);
 957     sid = NULL;
 958 }
 959
 960 void
 961 res_write_xml(struct SResource *res, const char* id,  const char* language, UBool isTopLevel, UErrorCode *status) {
 962
 963     if (U_FAILURE(*status)) {
 964         return ;
 965     }
 966
 967     if (res != NULL) {
 968         switch (res->fType) {
 969         case URES_STRING:
 970              string_write_xml    (static_cast<StringResource *>(res), id, language, status);
 971              return;
 972
 973         case URES_ALIAS:
 974              alias_write_xml     (static_cast<AliasResource *>(res), id, language, status);
 975              return;
 976
 977         case URES_INT_VECTOR:
 978              intvector_write_xml (static_cast<IntVectorResource *>(res), id, language, status);
 979              return;
 980
 981         case URES_BINARY:
 982              bin_write_xml       (static_cast<BinaryResource *>(res), id, language, status);
 983              return;
 984
 985         case URES_INT:
 986              int_write_xml       (static_cast<IntResource *>(res), id, language, status);
 987              return;
 988
 989         case URES_ARRAY:
 990              array_write_xml     (static_cast<ArrayResource *>(res), id, language, status);
 991              return;
 992
 993         case URES_TABLE:
 994              table_write_xml     (static_cast<TableResource *>(res), id, language, isTopLevel, status);
 995              return;
 996
 997         default:
 998             break;
 999         }
1000     }
1001
1002     *status = U_INTERNAL_PROGRAM_ERROR;
1003 }
1004
1005 void
1006 bundle_write_xml(struct SRBRoot *bundle, const char *outputDir,const char* outputEnc, const char* filename,
1007                   char *writtenFilename, int writtenFilenameLen,
1008                   const char* language, const char* outFileName, UErrorCode *status) {
1009
1010     char* xmlfileName = NULL;
1011     char* outputFileName = NULL;
1012     char* originalFileName = NULL;
1013     const char* fileStart = "<file xml:space = \"preserve\" source-language = \"";
1014     const char* file1 = "\" datatype = \"x-icu-resource-bundle\" ";
1015     const char* file2 = "original = \"";
1016     const char* file4 = "\" date = \"";
1017     const char* fileEnd = "</file>\n";
1018     const char* headerStart = "<header>\n";
1019     const char* headerEnd = "</header>\n";
1020     const char* bodyStart = "<body>\n";
1021     const char* bodyEnd = "</body>\n";
1022
1023     const char *tool_start = "<tool";
1024     const char *tool_id = "genrb-" GENRB_VERSION "-icu-" U_ICU_VERSION;
1025     const char *tool_name = "genrb";
1026
1027     char* temp = NULL;
1028     char* lang = NULL;
1029     const char* pos = NULL;
1030     int32_t first, index;
1031     time_t currTime;
1032     char timeBuf[128];
1033
1034     outDir = outputDir;
1035
1036     srBundle = bundle;
1037
1038     pos = uprv_strrchr(filename, '\\');
1039     if(pos != NULL) {
1040         first = (int32_t)(pos - filename + 1);
1041     } else {
1042         first = 0;
1043     }
1044     index = (int32_t)(uprv_strlen(filename) - uprv_strlen(textExt) - first);
1045     originalFileName = (char *)uprv_malloc(sizeof(char)*index+1);
1046     uprv_memset(originalFileName, 0, sizeof(char)*index+1);
1047     uprv_strncpy(originalFileName, filename + first, index);
1048
1049     if(uprv_strcmp(originalFileName, srBundle->fLocale) != 0) {
1050         fprintf(stdout, "Warning: The file name is not same as the resource name!\n");
1051     }
1052
1053     temp = originalFileName;
1054     originalFileName = (char *)uprv_malloc(sizeof(char)* (uprv_strlen(temp)+uprv_strlen(textExt)) + 1);
1055     uprv_memset(originalFileName, 0, sizeof(char)* (uprv_strlen(temp)+uprv_strlen(textExt)) + 1);
1056     uprv_strcat(originalFileName, temp);
1057     uprv_strcat(originalFileName, textExt);
1058     uprv_free(temp);
1059     temp = NULL;
1060
1061
1062     if (language == NULL) {
1063 /*        lang = parseFilename(filename, lang);
1064         if (lang == NULL) {*/
1065             /* now check if locale name is valid or not
1066              * this is to cater for situation where
1067              * pegasusServer.txt contains
1068              *
1069              * en{
1070              *      ..
1071              * }
1072              */
1073              lang = parseFilename(srBundle->fLocale, lang);
1074              /*
1075               * Neither  the file name nor the table name inside the
1076               * txt file contain a valid country and language codes
1077               * throw an error.
1078               * pegasusServer.txt contains
1079               *
1080               * testelements{
1081               *     ....
1082               * }
1083               */
1084              if(lang==NULL){
1085                  fprintf(stderr, "Error: The file name and table name do not contain a valid language code. Please use -l option to specify it.\n");
1086                  exit(U_ILLEGAL_ARGUMENT_ERROR);
1087              }
1088        /* }*/
1089     } else {
1090         lang = (char *)uprv_malloc(sizeof(char)*uprv_strlen(language) +1);
1091         uprv_memset(lang, 0, sizeof(char)*uprv_strlen(language) +1);
1092         uprv_strcpy(lang, language);
1093     }
1094
1095     if(outFileName) {
1096         outputFileName = (char *)uprv_malloc(sizeof(char)*uprv_strlen(outFileName) + 1);
1097         uprv_memset(outputFileName, 0, sizeof(char)*uprv_strlen(outFileName) + 1);
1098         uprv_strcpy(outputFileName,outFileName);
1099     } else {
1100         outputFileName = (char *)uprv_malloc(sizeof(char)*uprv_strlen(srBundle->fLocale) + 1);
1101         uprv_memset(outputFileName, 0, sizeof(char)*uprv_strlen(srBundle->fLocale) + 1);
1102         uprv_strcpy(outputFileName,srBundle->fLocale);
1103     }
1104
1105     if(outputDir) {
1106         xmlfileName = (char *)uprv_malloc(sizeof(char)*(uprv_strlen(outputDir) + uprv_strlen(outputFileName) + uprv_strlen(xliffExt) + 1) +1);
1107         uprv_memset(xmlfileName, 0, sizeof(char)*(uprv_strlen(outputDir)+ uprv_strlen(outputFileName) + uprv_strlen(xliffExt) + 1) +1);
1108     } else {
1109         xmlfileName = (char *)uprv_malloc(sizeof(char)*(uprv_strlen(outputFileName) + uprv_strlen(xliffExt)) +1);
1110         uprv_memset(xmlfileName, 0, sizeof(char)*(uprv_strlen(outputFileName) + uprv_strlen(xliffExt)) +1);
1111     }
1112
1113     if(outputDir){
1114         uprv_strcpy(xmlfileName, outputDir);
1115         if(outputDir[uprv_strlen(outputDir)-1] !=U_FILE_SEP_CHAR){
1116             uprv_strcat(xmlfileName,U_FILE_SEP_STRING);
1117         }
1118     }
1119     uprv_strcat(xmlfileName,outputFileName);
1120     uprv_strcat(xmlfileName,xliffExt);
1121
1122     if (writtenFilename) {
1123         uprv_strncpy(writtenFilename, xmlfileName, writtenFilenameLen);
1124     }
1125
1126     if (U_FAILURE(*status)) {
1127         goto cleanup_bundle_write_xml;
1128     }
1129
1130     out= T_FileStream_open(xmlfileName,"w");
1131
1132     if(out==NULL){
1133         *status = U_FILE_ACCESS_ERROR;
1134         goto cleanup_bundle_write_xml;
1135     }
1136     write_utf8_file(out, UnicodeString(xmlHeader));
1137
1138     if(outputEnc && *outputEnc!='\0'){
1139         /* store the output encoding */
1140         enc = outputEnc;
1141         conv=ucnv_open(enc,status);
1142         if(U_FAILURE(*status)){
1143             goto cleanup_bundle_write_xml;
1144         }
1145     }
1146     write_utf8_file(out, UnicodeString(bundleStart));
1147     write_tabs(out);
1148     write_utf8_file(out, UnicodeString(fileStart));
1149     /* check if lang and language are the same */
1150     if(language != NULL && uprv_strcmp(lang, srBundle->fLocale)!=0){
1151         fprintf(stderr,"Warning: The top level tag in the resource and language specified are not the same. Please check the input.\n");
1152     }
1153     write_utf8_file(out, UnicodeString(lang));
1154     write_utf8_file(out, UnicodeString(file1));
1155     write_utf8_file(out, UnicodeString(file2));
1156     write_utf8_file(out, UnicodeString(originalFileName));
1157     write_utf8_file(out, UnicodeString(file4));
1158
1159     time(&currTime);
1160     strftime(timeBuf, sizeof(timeBuf), "%Y-%m-%dT%H:%M:%SZ", gmtime(&currTime));
1161     write_utf8_file(out, UnicodeString(timeBuf));
1162     write_utf8_file(out, UnicodeString("\">\n"));
1163
1164     tabCount += 1;
1165     write_tabs(out);
1166     write_utf8_file(out, UnicodeString(headerStart));
1167
1168     tabCount += 1;
1169     write_tabs(out);
1170
1171     write_utf8_file(out, UnicodeString(tool_start));
1172     printAttribute("tool-id", tool_id, (int32_t) uprv_strlen(tool_id));
1173     printAttribute("tool-name", tool_name, (int32_t) uprv_strlen(tool_name));
1174     write_utf8_file(out, UnicodeString("/>\n"));
1175
1176     tabCount -= 1;
1177     write_tabs(out);
1178
1179     write_utf8_file(out, UnicodeString(headerEnd));
1180
1181     write_tabs(out);
1182     tabCount += 1;
1183
1184     write_utf8_file(out, UnicodeString(bodyStart));
1185
1186
1187     res_write_xml(bundle->fRoot, bundle->fLocale, lang, TRUE, status);
1188
1189     tabCount -= 1;
1190     write_tabs(out);
1191
1192     write_utf8_file(out, UnicodeString(bodyEnd));
1193     tabCount--;
1194     write_tabs(out);
1195     write_utf8_file(out, UnicodeString(fileEnd));
1196     tabCount--;
1197     write_tabs(out);
1198     write_utf8_file(out, UnicodeString(bundleEnd));
1199     T_FileStream_close(out);
1200
1201     ucnv_close(conv);
1202
1203 cleanup_bundle_write_xml:
1204     uprv_free(originalFileName);
1205     uprv_free(lang);
1206     if(xmlfileName != NULL) {
1207         uprv_free(xmlfileName);
1208     }
1209     if(outputFileName != NULL){
1210         uprv_free(outputFileName);
1211     }
1212 }