icuSources/tools/gencnval/gencnval.c

   1 /*
   2 *******************************************************************************
   3 *
   4 *   Copyright (C) 1999-2006, International Business Machines
   5 *   Corporation and others.  All Rights Reserved.
   6 *
   7 *******************************************************************************
   8 *   file name:  gencnval.c
   9 *   encoding:   US-ASCII
  10 *   tab size:   8 (not used)
  11 *   indentation:4
  12 *
  13 *   created on: 1999nov05
  14 *   created by: Markus W. Scherer
  15 *
  16 *   This program reads convrtrs.txt and writes a memory-mappable
  17 *   converter name alias table to cnvalias.dat .
  18 *
  19 *   This program currently writes version 2.1 of the data format. See
  20 *   ucnv_io.c for more details on the format. Note that version 2.1
  21 *   is written in such a way that a 2.0 reader will be able to use it,
  22 *   and a 2.1 reader will be able to read 2.0.
  23 */
  24
  25 #include "unicode/utypes.h"
  26 #include "unicode/putil.h"
  27 #include "unicode/ucnv.h" /* ucnv_compareNames() */
  28 #include "ucnv_io.h"
  29 #include "cmemory.h"
  30 #include "cstring.h"
  31 #include "uinvchar.h"
  32 #include "filestrm.h"
  33 #include "unicode/uclean.h"
  34 #include "unewdata.h"
  35 #include "uoptions.h"
  36
  37 #include <stdio.h>
  38 #include <stdlib.h>
  39 #include <ctype.h>
  40
  41 /* TODO: Need to check alias name length is less than UCNV_MAX_CONVERTER_NAME_LENGTH */
  42
  43 /* STRING_STORE_SIZE + TAG_STORE_SIZE <= ((2^16 - 1) * 2)
  44  That is the maximum size for the string stores combined
  45  because the strings are index at 16-bit boundries by a
  46  16-bit index, and there is only one section for the
  47  strings.
  48  */
  49 #define STRING_STORE_SIZE 0x1FBFE   /* 130046 */
  50 #define TAG_STORE_SIZE      0x400   /* 1024 */
  51
  52 /* The combined tag and converter count can affect the number of lists
  53  created.  The size of all lists must be less than (2^17 - 1)
  54  because the lists are indexed as a 16-bit array with a 16-bit index.
  55  */
  56 #define MAX_TAG_COUNT 0x3F      /* 63 */
  57 #define MAX_CONV_COUNT UCNV_CONVERTER_INDEX_MASK
  58 #define MAX_ALIAS_COUNT 0xFFFF  /* 65535 */
  59
  60 /* The maximum number of aliases that a standard tag/converter combination can have.
  61  At this moment 6/18/2002, IANA has 12 names for ASCII. Don't go below 15 for
  62  this value. I don't recommend more than 31 for this value.
  63  */
  64 #define MAX_TC_ALIAS_COUNT 0x1F    /* 31 */
  65
  66 #define MAX_LINE_SIZE 0x7FFF    /* 32767 */
  67 #define MAX_LIST_SIZE 0xFFFF    /* 65535 */
  68
  69 #define DATA_NAME "cnvalias"
  70 #define DATA_TYPE "icu" /* ICU alias table */
  71
  72 #define ALL_TAG_STR "ALL"
  73 #define ALL_TAG_NUM 1
  74 #define EMPTY_TAG_NUM 0
  75
  76 /* UDataInfo cf. udata.h */
  77 static const UDataInfo dataInfo={
  78     sizeof(UDataInfo),
  79     0,
  80
  81     U_IS_BIG_ENDIAN,
  82     U_CHARSET_FAMILY,
  83     sizeof(UChar),
  84     0,
  85
  86     {0x43, 0x76, 0x41, 0x6c},     /* dataFormat="CvAl" */
  87     {3, 0, 1, 0},                 /* formatVersion */
  88     {1, 4, 2, 0}                  /* dataVersion */
  89 };
  90
  91 typedef struct {
  92     char *store;
  93     uint32_t top;
  94     uint32_t max;
  95 } StringBlock;
  96
  97 static char stringStore[STRING_STORE_SIZE];
  98 static StringBlock stringBlock = { stringStore, 0, STRING_STORE_SIZE };
  99
 100 typedef struct {
 101     uint16_t    aliasCount;
 102     uint16_t    *aliases;     /* Index into stringStore */
 103 } AliasList;
 104
 105 typedef struct {
 106     uint16_t converter;     /* Index into stringStore */
 107     uint16_t totalAliasCount;    /* Total aliases in this column */
 108 } Converter;
 109
 110 static Converter converters[MAX_CONV_COUNT];
 111 static uint16_t converterCount=0;
 112
 113 static char tagStore[TAG_STORE_SIZE];
 114 static StringBlock tagBlock = { tagStore, 0, TAG_STORE_SIZE };
 115
 116 typedef struct {
 117     uint16_t    tag;        /* Index into tagStore */
 118     uint16_t    totalAliasCount; /* Total aliases in this row */
 119     AliasList   aliasList[MAX_CONV_COUNT];
 120 } Tag;
 121
 122 /* Think of this as a 3D array. It's tagCount by converterCount by aliasCount */
 123 static Tag tags[MAX_TAG_COUNT];
 124 static uint16_t tagCount = 0;
 125
 126 /* Used for storing all aliases  */
 127 static uint16_t knownAliases[MAX_ALIAS_COUNT];
 128 static uint16_t knownAliasesCount = 0;
 129 /*static uint16_t duplicateKnownAliasesCount = 0;*/
 130
 131 /* Used for storing the lists section that point to aliases */
 132 static uint16_t aliasLists[MAX_LIST_SIZE];
 133 static uint16_t aliasListsSize = 0;
 134
 135 /* Were the standard tags declared before the aliases. */
 136 static UBool standardTagsUsed = FALSE;
 137 static UBool verbose = FALSE;
 138 static int lineNum = 1;
 139
 140 static UConverterAliasOptions tableOptions = {
 141     UCNV_IO_STD_NORMALIZED,
 142     1 /* containsCnvOptionInfo */
 143 };
 144
 145 /* prototypes --------------------------------------------------------------- */
 146
 147 static void
 148 parseLine(const char *line);
 149
 150 static void
 151 parseFile(FileStream *in);
 152
 153 static int32_t
 154 chomp(char *line);
 155
 156 static void
 157 addOfficialTaggedStandards(char *line, int32_t lineLen);
 158
 159 static uint16_t
 160 addAlias(const char *alias, uint16_t standard, uint16_t converter, UBool defaultName);
 161
 162 static uint16_t
 163 addConverter(const char *converter);
 164
 165 static char *
 166 allocString(StringBlock *block, const char *s, int32_t length);
 167
 168 static uint16_t
 169 addToKnownAliases(const char *alias);
 170
 171 static int
 172 compareAliases(const void *alias1, const void *alias2);
 173
 174 static uint16_t
 175 getTagNumber(const char *tag, uint16_t tagLen);
 176
 177 /*static void
 178 addTaggedAlias(uint16_t tag, const char *alias, uint16_t converter);*/
 179
 180 static void
 181 writeAliasTable(UNewDataMemory *out);
 182
 183 /* -------------------------------------------------------------------------- */
 184
 185 /* Presumes that you used allocString() */
 186 #define GET_ALIAS_STR(index) (stringStore + ((size_t)(index) << 1))
 187 #define GET_TAG_STR(index) (tagStore + ((size_t)(index) << 1))
 188
 189 /* Presumes that you used allocString() */
 190 #define GET_ALIAS_NUM(str) ((uint16_t)((str - stringStore) >> 1))
 191 #define GET_TAG_NUM(str) ((uint16_t)((str - tagStore) >> 1))
 192
 193 enum
 194 {
 195     HELP1,
 196     HELP2,
 197     VERBOSE,
 198     COPYRIGHT,
 199     DESTDIR,
 200     SOURCEDIR
 201 };
 202
 203 static UOption options[]={
 204     UOPTION_HELP_H,
 205     UOPTION_HELP_QUESTION_MARK,
 206     UOPTION_VERBOSE,
 207     UOPTION_COPYRIGHT,
 208     UOPTION_DESTDIR,
 209     UOPTION_SOURCEDIR
 210 };
 211
 212 extern int
 213 main(int argc, char* argv[]) {
 214     char pathBuf[512];
 215     const char *path;
 216     FileStream *in;
 217     UNewDataMemory *out;
 218     UErrorCode errorCode=U_ZERO_ERROR;
 219
 220     U_MAIN_INIT_ARGS(argc, argv);
 221
 222     /* preset then read command line options */
 223     options[DESTDIR].value=options[SOURCEDIR].value=u_getDataDirectory();
 224     argc=u_parseArgs(argc, argv, sizeof(options)/sizeof(options[0]), options);
 225
 226     /* error handling, printing usage message */
 227     if(argc<0) {
 228         fprintf(stderr,
 229             "error in command line argument \"%s\"\n",
 230             argv[-argc]);
 231     }
 232     if(argc<0 || options[HELP1].doesOccur || options[HELP2].doesOccur) {
 233         fprintf(stderr,
 234             "usage: %s [-options] [convrtrs.txt]\n"
 235             "\tread convrtrs.txt and create " U_ICUDATA_NAME "_" DATA_NAME "." DATA_TYPE "\n"
 236             "options:\n"
 237             "\t-h or -? or --help  this usage text\n"
 238             "\t-v or --verbose     prints out extra information about the alias table\n"
 239             "\t-c or --copyright   include a copyright notice\n"
 240             "\t-d or --destdir     destination directory, followed by the path\n"
 241             "\t-s or --sourcedir   source directory, followed by the path\n",
 242             argv[0]);
 243         return argc<0 ? U_ILLEGAL_ARGUMENT_ERROR : U_ZERO_ERROR;
 244     }
 245
 246     if(options[VERBOSE].doesOccur) {
 247         verbose = TRUE;
 248     }
 249
 250     if(argc>=2) {
 251         path=argv[1];
 252     } else {
 253         path=options[SOURCEDIR].value;
 254         if(path!=NULL && *path!=0) {
 255             char *end;
 256
 257             uprv_strcpy(pathBuf, path);
 258             end = uprv_strchr(pathBuf, 0);
 259             if(*(end-1)!=U_FILE_SEP_CHAR) {
 260                 *(end++)=U_FILE_SEP_CHAR;
 261             }
 262             uprv_strcpy(end, "convrtrs.txt");
 263             path=pathBuf;
 264         } else {
 265             path = "convrtrs.txt";
 266         }
 267     }
 268
 269     uprv_memset(stringStore, 0, sizeof(stringStore));
 270     uprv_memset(tagStore, 0, sizeof(tagStore));
 271     uprv_memset(converters, 0, sizeof(converters));
 272     uprv_memset(tags, 0, sizeof(tags));
 273     uprv_memset(aliasLists, 0, sizeof(aliasLists));
 274     uprv_memset(knownAliases, 0, sizeof(aliasLists));
 275
 276
 277     in=T_FileStream_open(path, "r");
 278     if(in==NULL) {
 279         fprintf(stderr, "gencnval: unable to open input file convrtrs.txt\n");
 280         exit(U_FILE_ACCESS_ERROR);
 281     }
 282     parseFile(in);
 283     T_FileStream_close(in);
 284
 285     /* create the output file */
 286     out=udata_create(options[DESTDIR].value, DATA_TYPE, DATA_NAME, &dataInfo,
 287                      options[COPYRIGHT].doesOccur ? U_COPYRIGHT_STRING : NULL, &errorCode);
 288     if(U_FAILURE(errorCode)) {
 289         fprintf(stderr, "gencnval: unable to open output file - error %s\n", u_errorName(errorCode));
 290         exit(errorCode);
 291     }
 292
 293     /* write the table of aliases based on a tag/converter name combination */
 294     writeAliasTable(out);
 295
 296     /* finish */
 297     udata_finish(out, &errorCode);
 298     if(U_FAILURE(errorCode)) {
 299         fprintf(stderr, "gencnval: error finishing output file - %s\n", u_errorName(errorCode));
 300         exit(errorCode);
 301     }
 302
 303     return 0;
 304 }
 305
 306 static void
 307 parseFile(FileStream *in) {
 308     char line[MAX_LINE_SIZE];
 309     char lastLine[MAX_LINE_SIZE];
 310     int32_t lineSize = 0;
 311     int32_t lastLineSize = 0;
 312     UBool validParse = TRUE;
 313
 314     lineNum = 0;
 315
 316     /* Add the empty tag, which is for untagged aliases */
 317     getTagNumber("", 0);
 318     getTagNumber(ALL_TAG_STR, 3);
 319     allocString(&stringBlock, "", 0);
 320
 321     /* read the list of aliases */
 322     while (validParse) {
 323         validParse = FALSE;
 324
 325         /* Read non-empty lines that don't start with a space character. */
 326         while (T_FileStream_readLine(in, lastLine, MAX_LINE_SIZE) != NULL) {
 327             lastLineSize = chomp(lastLine);
 328             if (lineSize == 0 || (lastLineSize > 0 && isspace(*lastLine))) {
 329                 uprv_strcpy(line + lineSize, lastLine);
 330                 lineSize += lastLineSize;
 331             } else if (lineSize > 0) {
 332                 validParse = TRUE;
 333                 break;
 334             }
 335             lineNum++;
 336         }
 337
 338         if (validParse || lineSize > 0) {
 339             if (isspace(*line)) {
 340                 fprintf(stderr, "error(line %d): cannot start an alias with a space\n", lineNum-1);
 341                 exit(U_PARSE_ERROR);
 342             } else if (line[0] == '{') {
 343                 if (!standardTagsUsed && line[lineSize - 1] != '}') {
 344                     fprintf(stderr, "error(line %d): alias needs to start with a converter name\n", lineNum);
 345                     exit(U_PARSE_ERROR);
 346                 }
 347                 addOfficialTaggedStandards(line, lineSize);
 348                 standardTagsUsed = TRUE;
 349             } else {
 350                 if (standardTagsUsed) {
 351                     parseLine(line);
 352                 }
 353                 else {
 354                     fprintf(stderr, "error(line %d): alias table needs to start a list of standard tags\n", lineNum);
 355                     exit(U_PARSE_ERROR);
 356                 }
 357             }
 358             /* Was the last line consumed */
 359             if (lastLineSize > 0) {
 360                 uprv_strcpy(line, lastLine);
 361                 lineSize = lastLineSize;
 362             }
 363             else {
 364                 lineSize = 0;
 365             }
 366         }
 367         lineNum++;
 368     }
 369 }
 370
 371 /* This works almost like the Perl chomp.
 372  It removes the newlines, comments and trailing whitespace (not preceding whitespace).
 373 */
 374 static int32_t
 375 chomp(char *line) {
 376     char *s = line;
 377     char *lastNonSpace = line;
 378     while(*s!=0) {
 379         /* truncate at a newline or a comment */
 380         if(*s == '\r' || *s == '\n' || *s == '#') {
 381             *s = 0;
 382             break;
 383         }
 384         if (!isspace(*s)) {
 385             lastNonSpace = s;
 386         }
 387         ++s;
 388     }
 389     if (lastNonSpace++ > line) {
 390         *lastNonSpace = 0;
 391         s = lastNonSpace;
 392     }
 393     return (int32_t)(s - line);
 394 }
 395
 396 static void
 397 parseLine(const char *line) {
 398     uint16_t pos=0, start, limit, length, cnv;
 399     char *converter, *alias;
 400
 401     /* skip leading white space */
 402     /* There is no whitespace at the beginning anymore */
 403 /*    while(line[pos]!=0 && isspace(line[pos])) {
 404         ++pos;
 405     }
 406 */
 407
 408     /* is there nothing on this line? */
 409     if(line[pos]==0) {
 410         return;
 411     }
 412
 413     /* get the converter name */
 414     start=pos;
 415     while(line[pos]!=0 && !isspace(line[pos])) {
 416         ++pos;
 417     }
 418     limit=pos;
 419
 420     /* store the converter name */
 421     length=(uint16_t)(limit-start);
 422     converter=allocString(&stringBlock, line+start, length);
 423
 424     /* add the converter to the converter table */
 425     cnv=addConverter(converter);
 426
 427     /* The name itself may be tagged, so let's added it to the aliases list properly */
 428     pos = start;
 429
 430     /* get all the real aliases */
 431     for(;;) {
 432
 433         /* skip white space */
 434         while(line[pos]!=0 && isspace(line[pos])) {
 435             ++pos;
 436         }
 437
 438         /* is there no more alias name on this line? */
 439         if(line[pos]==0) {
 440             break;
 441         }
 442
 443         /* get an alias name */
 444         start=pos;
 445         while(line[pos]!=0 && line[pos]!='{' && !isspace(line[pos])) {
 446             ++pos;
 447         }
 448         limit=pos;
 449
 450         /* store the alias name */
 451         length=(uint16_t)(limit-start);
 452         if (start == 0) {
 453             /* add the converter as its own alias to the alias table */
 454             alias = converter;
 455             addAlias(alias, ALL_TAG_NUM, cnv, TRUE);
 456         }
 457         else {
 458             alias=allocString(&stringBlock, line+start, length);
 459             addAlias(alias, ALL_TAG_NUM, cnv, FALSE);
 460         }
 461         addToKnownAliases(alias);
 462
 463         /* add the alias/converter pair to the alias table */
 464         /* addAlias(alias, 0, cnv, FALSE);*/
 465
 466         /* skip whitespace */
 467         while (line[pos] && isspace(line[pos])) {
 468             ++pos;
 469         }
 470
 471         /* handle tags if they are present */
 472         if (line[pos] == '{') {
 473             ++pos;
 474             do {
 475                 start = pos;
 476                 while (line[pos] && line[pos] != '}' && !isspace( line[pos])) {
 477                     ++pos;
 478                 }
 479                 limit = pos;
 480
 481                 if (start != limit) {
 482                     /* add the tag to the tag table */
 483                     uint16_t tag = getTagNumber(line + start, (uint16_t)(limit - start));
 484                     addAlias(alias, tag, cnv, (UBool)(line[limit-1] == '*'));
 485                 }
 486
 487                 while (line[pos] && isspace(line[pos])) {
 488                     ++pos;
 489                 }
 490             } while (line[pos] && line[pos] != '}');
 491
 492             if (line[pos] == '}') {
 493                 ++pos;
 494             } else {
 495                 fprintf(stderr, "error(line %d): Unterminated tag list\n", lineNum);
 496                 exit(U_UNMATCHED_BRACES);
 497             }
 498         } else {
 499             addAlias(alias, EMPTY_TAG_NUM, cnv, (UBool)(tags[0].aliasList[cnv].aliasCount == 0));
 500         }
 501     }
 502 }
 503
 504 static uint16_t
 505 getTagNumber(const char *tag, uint16_t tagLen) {
 506     char *atag;
 507     uint16_t t;
 508     UBool preferredName = ((tagLen > 0) ? (tag[tagLen - 1] == '*') : (FALSE));
 509
 510     if (tagCount >= MAX_TAG_COUNT) {
 511         fprintf(stderr, "error(line %d): too many tags\n", lineNum);
 512         exit(U_BUFFER_OVERFLOW_ERROR);
 513     }
 514
 515     if (preferredName) {
 516 /*        puts(tag);*/
 517         tagLen--;
 518     }
 519
 520     for (t = 0; t < tagCount; ++t) {
 521         const char *currTag = GET_TAG_STR(tags[t].tag);
 522         if (uprv_strlen(currTag) == tagLen && !uprv_strnicmp(currTag, tag, tagLen)) {
 523             return t;
 524         }
 525     }
 526
 527     /* we need to add this tag */
 528     if (tagCount >= MAX_TAG_COUNT) {
 529         fprintf(stderr, "error(line %d): too many tags\n", lineNum);
 530         exit(U_BUFFER_OVERFLOW_ERROR);
 531     }
 532
 533     /* allocate a new entry in the tag table */
 534     atag = allocString(&tagBlock, tag, tagLen);
 535
 536     if (standardTagsUsed) {
 537         fprintf(stderr, "error(line %d): Tag \"%s\" is not declared at the beginning of the alias table.\n",
 538             lineNum, atag);
 539         exit(1);
 540     }
 541     else if (tagLen > 0 && strcmp(tag, ALL_TAG_STR) != 0) {
 542         fprintf(stderr, "warning(line %d): Tag \"%s\" was added to the list of standards because it was not declared at beginning of the alias table.\n",
 543             lineNum, atag);
 544     }
 545
 546     /* add the tag to the tag table */
 547     tags[tagCount].tag = GET_TAG_NUM(atag);
 548     /* The aliasList should be set to 0's already */
 549
 550     return tagCount++;
 551 }
 552
 553 /*static void
 554 addTaggedAlias(uint16_t tag, const char *alias, uint16_t converter) {
 555     tags[tag].aliases[converter] = alias;
 556 }
 557 */
 558
 559 static void
 560 addOfficialTaggedStandards(char *line, int32_t lineLen) {
 561     char *atag;
 562     char *tag = strchr(line, '{') + 1;
 563     static const char WHITESPACE[] = " \t";
 564
 565     if (tagCount > UCNV_NUM_RESERVED_TAGS) {
 566         fprintf(stderr, "error(line %d): official tags already added\n", lineNum);
 567         exit(U_BUFFER_OVERFLOW_ERROR);
 568     }
 569     strchr(tag, '}')[0] = 0;
 570
 571     tag = strtok(tag, WHITESPACE);
 572     while (tag != NULL) {
 573 /*        printf("Adding original tag \"%s\"\n", tag);*/
 574
 575         /* allocate a new entry in the tag table */
 576         atag = allocString(&tagBlock, tag, -1);
 577
 578         /* add the tag to the tag table */
 579         tags[tagCount++].tag = (uint16_t)((atag - tagStore) >> 1);
 580
 581         /* The aliasList should already be set to 0's */
 582
 583         /* Get next tag */
 584         tag = strtok(NULL, WHITESPACE);
 585     }
 586 }
 587
 588 static uint16_t
 589 addToKnownAliases(const char *alias) {
 590 /*    uint32_t idx; */
 591     /* strict matching */
 592 /*    for (idx = 0; idx < knownAliasesCount; idx++) {
 593         uint16_t num = GET_ALIAS_NUM(alias);
 594         if (knownAliases[idx] != num
 595             && uprv_strcmp(alias, GET_ALIAS_STR(knownAliases[idx])) == 0)
 596         {
 597             fprintf(stderr, "warning(line %d): duplicate alias %s and %s found\n",
 598                 lineNum, alias, GET_ALIAS_STR(knownAliases[idx]));
 599             duplicateKnownAliasesCount++;
 600             break;
 601         }
 602         else if (knownAliases[idx] != num
 603             && ucnv_compareNames(alias, GET_ALIAS_STR(knownAliases[idx])) == 0)
 604         {
 605             if (verbose) {
 606                 fprintf(stderr, "information(line %d): duplicate alias %s and %s found\n",
 607                     lineNum, alias, GET_ALIAS_STR(knownAliases[idx]));
 608             }
 609             duplicateKnownAliasesCount++;
 610             break;
 611         }
 612     }
 613 */
 614     if (knownAliasesCount >= MAX_ALIAS_COUNT) {
 615         fprintf(stderr, "warning(line %d): Too many aliases defined for all converters\n",
 616             lineNum);
 617         exit(U_BUFFER_OVERFLOW_ERROR);
 618     }
 619     /* TODO: We could try to unlist exact duplicates. */
 620     return knownAliases[knownAliasesCount++] = GET_ALIAS_NUM(alias);
 621 }
 622
 623 /*
 624 @param standard When standard is 0, then it's the "empty" tag.
 625 */
 626 static uint16_t
 627 addAlias(const char *alias, uint16_t standard, uint16_t converter, UBool defaultName) {
 628     uint32_t idx, idx2;
 629     UBool dupFound = FALSE;
 630     UBool startEmptyWithoutDefault = FALSE;
 631     AliasList *aliasList;
 632
 633     if(standard>=MAX_TAG_COUNT) {
 634         fprintf(stderr, "error(line %d): too many standard tags\n", lineNum);
 635         exit(U_BUFFER_OVERFLOW_ERROR);
 636     }
 637     if(converter>=MAX_CONV_COUNT) {
 638         fprintf(stderr, "error(line %d): too many converter names\n", lineNum);
 639         exit(U_BUFFER_OVERFLOW_ERROR);
 640     }
 641     aliasList = &tags[standard].aliasList[converter];
 642
 643     if (strchr(alias, '}')) {
 644         fprintf(stderr, "error(line %d): unmatched } found\n",
 645             lineNum);
 646     }
 647
 648     if(aliasList->aliasCount + 1 >= MAX_TC_ALIAS_COUNT) {
 649         fprintf(stderr, "error(line %d): too many aliases for alias %s and converter %s\n",
 650             lineNum, alias, GET_ALIAS_STR(converters[converter].converter));
 651         exit(U_BUFFER_OVERFLOW_ERROR);
 652     }
 653
 654     /* Show this warning only once. All aliases are added to the "ALL" tag. */
 655     if (standard == ALL_TAG_NUM && GET_ALIAS_STR(converters[converter].converter) != alias) {
 656         /* Normally these option values are parsed at runtime, and they can
 657            be discarded when the alias is a default converter. Options should
 658            only be on a converter and not an alias. */
 659         if (uprv_strchr(alias, UCNV_OPTION_SEP_CHAR) != 0)
 660         {
 661             fprintf(stderr, "warning(line %d): alias %s contains a \""UCNV_OPTION_SEP_STRING"\". Options are parsed at run-time and do not need to be in the alias table.\n",
 662                 lineNum, alias);
 663         }
 664         if (uprv_strchr(alias, UCNV_VALUE_SEP_CHAR) != 0)
 665         {
 666             fprintf(stderr, "warning(line %d): alias %s contains an \""UCNV_VALUE_SEP_STRING"\". Options are parsed at run-time and do not need to be in the alias table.\n",
 667                 lineNum, alias);
 668         }
 669     }
 670
 671     if (standard != ALL_TAG_NUM) {
 672         /* Check for duplicate aliases for this tag on all converters */
 673         for (idx = 0; idx < converterCount; idx++) {
 674             for (idx2 = 0; idx2 < tags[standard].aliasList[idx].aliasCount; idx2++) {
 675                 uint16_t aliasNum = tags[standard].aliasList[idx].aliases[idx2];
 676                 if (aliasNum
 677                     && ucnv_compareNames(alias, GET_ALIAS_STR(aliasNum)) == 0)
 678                 {
 679                     if (idx == converter) {
 680                         /*
 681                          * (alias, standard) duplicates are harmless if they map to the same converter.
 682                          * Only print a warning in verbose mode, or if the alias is a precise duplicate,
 683                          * not just a lenient-match duplicate.
 684                          */
 685                         if (verbose || 0 == uprv_strcmp(alias, GET_ALIAS_STR(aliasNum))) {
 686                             fprintf(stderr, "warning(line %d): duplicate aliases %s and %s found for standard %s and converter %s\n",
 687                                 lineNum, alias, GET_ALIAS_STR(aliasNum),
 688                                 GET_TAG_STR(tags[standard].tag),
 689                                 GET_ALIAS_STR(converters[converter].converter));
 690                         }
 691                     } else {
 692                         fprintf(stderr, "warning(line %d): duplicate aliases %s and %s found for standard tag %s between converter %s and converter %s\n",
 693                             lineNum, alias, GET_ALIAS_STR(aliasNum),
 694                             GET_TAG_STR(tags[standard].tag),
 695                             GET_ALIAS_STR(converters[converter].converter),
 696                             GET_ALIAS_STR(converters[idx].converter));
 697                     }
 698                     dupFound = TRUE;
 699                     break;
 700                 }
 701             }
 702         }
 703
 704         /* Check for duplicate default aliases for this converter on all tags */
 705         /* It's okay to have multiple standards prefer the same name */
 706 /*        if (verbose && !dupFound) {
 707             for (idx = 0; idx < tagCount; idx++) {
 708                 if (tags[idx].aliasList[converter].aliases) {
 709                     uint16_t aliasNum = tags[idx].aliasList[converter].aliases[0];
 710                     if (aliasNum
 711                         && ucnv_compareNames(alias, GET_ALIAS_STR(aliasNum)) == 0)
 712                     {
 713                         fprintf(stderr, "warning(line %d): duplicate alias %s found for converter %s and standard tag %s\n",
 714                             lineNum, alias, GET_ALIAS_STR(converters[converter].converter), GET_TAG_STR(tags[standard].tag));
 715                         break;
 716                     }
 717                 }
 718             }
 719         }*/
 720     }
 721
 722     if (aliasList->aliasCount <= 0) {
 723         aliasList->aliasCount++;
 724         startEmptyWithoutDefault = TRUE;
 725     }
 726     aliasList->aliases = (uint16_t *)uprv_realloc(aliasList->aliases, (aliasList->aliasCount + 1) * sizeof(aliasList->aliases[0]));
 727     if (startEmptyWithoutDefault) {
 728         aliasList->aliases[0] = 0;
 729     }
 730     if (defaultName) {
 731         if (aliasList->aliases[0] != 0) {
 732             fprintf(stderr, "error(line %d): Alias %s and %s cannot both be the default alias for standard tag %s and converter %s\n",
 733                 lineNum,
 734                 alias,
 735                 GET_ALIAS_STR(aliasList->aliases[0]),
 736                 GET_TAG_STR(tags[standard].tag),
 737                 GET_ALIAS_STR(converters[converter].converter));
 738             exit(U_PARSE_ERROR);
 739         }
 740         aliasList->aliases[0] = GET_ALIAS_NUM(alias);
 741     } else {
 742         aliasList->aliases[aliasList->aliasCount++] = GET_ALIAS_NUM(alias);
 743     }
 744 /*    aliasList->converter = converter;*/
 745
 746     converters[converter].totalAliasCount++; /* One more to the column */
 747     tags[standard].totalAliasCount++; /* One more to the row */
 748
 749     return aliasList->aliasCount;
 750 }
 751
 752 static uint16_t
 753 addConverter(const char *converter) {
 754     uint32_t idx;
 755     if(converterCount>=MAX_CONV_COUNT) {
 756         fprintf(stderr, "error(line %d): too many converters\n", lineNum);
 757         exit(U_BUFFER_OVERFLOW_ERROR);
 758     }
 759
 760     for (idx = 0; idx < converterCount; idx++) {
 761         if (ucnv_compareNames(converter, GET_ALIAS_STR(converters[idx].converter)) == 0) {
 762             fprintf(stderr, "error(line %d): duplicate converter %s found!\n", lineNum, converter);
 763             exit(U_PARSE_ERROR);
 764             break;
 765         }
 766     }
 767
 768     converters[converterCount].converter = GET_ALIAS_NUM(converter);
 769     converters[converterCount].totalAliasCount = 0;
 770
 771     return converterCount++;
 772 }
 773
 774 /* resolve this alias based on the prioritization of the standard tags. */
 775 static void
 776 resolveAliasToConverter(uint16_t alias, uint16_t *tagNum, uint16_t *converterNum) {
 777     uint16_t idx, idx2, idx3;
 778
 779     for (idx = UCNV_NUM_RESERVED_TAGS; idx < tagCount; idx++) {
 780         for (idx2 = 0; idx2 < converterCount; idx2++) {
 781             for (idx3 = 0; idx3 < tags[idx].aliasList[idx2].aliasCount; idx3++) {
 782                 uint16_t aliasNum = tags[idx].aliasList[idx2].aliases[idx3];
 783                 if (aliasNum == alias) {
 784                     *tagNum = idx;
 785                     *converterNum = idx2;
 786                     return;
 787                 }
 788             }
 789         }
 790     }
 791     /* Do the leftovers last, just in case */
 792     /* There is no need to do the ALL tag */
 793     idx = 0;
 794     for (idx2 = 0; idx2 < converterCount; idx2++) {
 795         for (idx3 = 0; idx3 < tags[idx].aliasList[idx2].aliasCount; idx3++) {
 796             uint16_t aliasNum = tags[idx].aliasList[idx2].aliases[idx3];
 797             if (aliasNum == alias) {
 798                 *tagNum = idx;
 799                 *converterNum = idx2;
 800                 return;
 801             }
 802         }
 803     }
 804     *tagNum = UINT16_MAX;
 805     *converterNum = UINT16_MAX;
 806     fprintf(stderr, "warning: alias %s not found\n",
 807         GET_ALIAS_STR(alias));
 808     return;
 809 }
 810
 811 /* The knownAliases should be sorted before calling this function */
 812 static uint32_t
 813 resolveAliases(uint16_t *uniqueAliasArr, uint16_t *uniqueAliasToConverterArr, uint16_t aliasOffset) {
 814     uint32_t uniqueAliasIdx = 0;
 815     uint32_t idx;
 816     uint16_t currTagNum, oldTagNum;
 817     uint16_t currConvNum, oldConvNum;
 818     const char *lastName;
 819
 820     resolveAliasToConverter(knownAliases[0], &oldTagNum, &currConvNum);
 821     uniqueAliasToConverterArr[uniqueAliasIdx] = currConvNum;
 822     oldConvNum = currConvNum;
 823     uniqueAliasArr[uniqueAliasIdx] = knownAliases[0] + aliasOffset;
 824     uniqueAliasIdx++;
 825     lastName = GET_ALIAS_STR(knownAliases[0]);
 826
 827     for (idx = 1; idx < knownAliasesCount; idx++) {
 828         resolveAliasToConverter(knownAliases[idx], &currTagNum, &currConvNum);
 829         if (ucnv_compareNames(lastName, GET_ALIAS_STR(knownAliases[idx])) == 0) {
 830             /* duplicate found */
 831             if ((currTagNum < oldTagNum && currTagNum >= UCNV_NUM_RESERVED_TAGS)
 832                 || oldTagNum == 0) {
 833                 oldTagNum = currTagNum;
 834                 uniqueAliasToConverterArr[uniqueAliasIdx - 1] = currConvNum;
 835                 uniqueAliasArr[uniqueAliasIdx - 1] = knownAliases[idx] + aliasOffset;
 836                 if (verbose) {
 837                     printf("using %s instead of %s -> %s",
 838                         GET_ALIAS_STR(knownAliases[idx]),
 839                         lastName,
 840                         GET_ALIAS_STR(converters[currConvNum].converter));
 841                     if (oldConvNum != currConvNum) {
 842                         printf(" (alias conflict)");
 843                     }
 844                     puts("");
 845                 }
 846             }
 847             else {
 848                 /* else ignore it */
 849                 if (verbose) {
 850                     printf("folding %s into %s -> %s",
 851                         GET_ALIAS_STR(knownAliases[idx]),
 852                         lastName,
 853                         GET_ALIAS_STR(converters[oldConvNum].converter));
 854                     if (oldConvNum != currConvNum) {
 855                         printf(" (alias conflict)");
 856                     }
 857                     puts("");
 858                 }
 859             }
 860             if (oldConvNum != currConvNum) {
 861                 uniqueAliasToConverterArr[uniqueAliasIdx - 1] |= UCNV_AMBIGUOUS_ALIAS_MAP_BIT;
 862             }
 863         }
 864         else {
 865             uniqueAliasToConverterArr[uniqueAliasIdx] = currConvNum;
 866             oldConvNum = currConvNum;
 867             uniqueAliasArr[uniqueAliasIdx] = knownAliases[idx] + aliasOffset;
 868             uniqueAliasIdx++;
 869             lastName = GET_ALIAS_STR(knownAliases[idx]);
 870             oldTagNum = currTagNum;
 871             /*printf("%s -> %s\n", GET_ALIAS_STR(knownAliases[idx]), GET_ALIAS_STR(converters[currConvNum].converter));*/
 872         }
 873         if (uprv_strchr(GET_ALIAS_STR(converters[currConvNum].converter), UCNV_OPTION_SEP_CHAR) != NULL) {
 874             uniqueAliasToConverterArr[uniqueAliasIdx-1] |= UCNV_CONTAINS_OPTION_BIT;
 875         }
 876     }
 877     return uniqueAliasIdx;
 878 }
 879
 880 static void
 881 createOneAliasList(uint16_t *aliasArrLists, uint32_t tag, uint32_t converter, uint16_t offset) {
 882     uint32_t aliasNum;
 883     AliasList *aliasList = &tags[tag].aliasList[converter];
 884
 885     if (aliasList->aliasCount == 0) {
 886         aliasArrLists[tag*converterCount + converter] = 0;
 887     }
 888     else {
 889         aliasLists[aliasListsSize++] = aliasList->aliasCount;
 890
 891         /* write into the array area a 1's based index. */
 892         aliasArrLists[tag*converterCount + converter] = aliasListsSize;
 893
 894 /*        printf("tag %s converter %s\n",
 895             GET_TAG_STR(tags[tag].tag),
 896             GET_ALIAS_STR(converters[converter].converter));*/
 897         for (aliasNum = 0; aliasNum < aliasList->aliasCount; aliasNum++) {
 898             uint16_t value;
 899 /*            printf("   %s\n",
 900                 GET_ALIAS_STR(aliasList->aliases[aliasNum]));*/
 901             if (aliasList->aliases[aliasNum]) {
 902                 value = aliasList->aliases[aliasNum] + offset;
 903             } else {
 904                 value = 0;
 905                 if (tag != 0) { /* Only show the warning when it's not the leftover tag. */
 906                     printf("warning: tag %s does not have a default alias for %s\n",
 907                             GET_TAG_STR(tags[tag].tag),
 908                             GET_ALIAS_STR(converters[converter].converter));
 909                 }
 910             }
 911             aliasLists[aliasListsSize++] = value;
 912             if (aliasListsSize >= MAX_LIST_SIZE) {
 913                 fprintf(stderr, "error: Too many alias lists\n");
 914                 exit(U_BUFFER_OVERFLOW_ERROR);
 915             }
 916
 917         }
 918     }
 919 }
 920
 921 static void
 922 createNormalizedAliasStrings(char *normalizedStrings, const char *origStringBlock, int32_t stringBlockLength) {
 923     int32_t currStrLen;
 924     uprv_memcpy(normalizedStrings, origStringBlock, stringBlockLength);
 925     while ((currStrLen = (int32_t)uprv_strlen(origStringBlock)) < stringBlockLength) {
 926         int32_t currStrSize = currStrLen + 1;
 927         if (currStrLen > 0) {
 928             int32_t normStrLen;
 929             ucnv_io_stripForCompare(normalizedStrings, origStringBlock);
 930             normStrLen = uprv_strlen(normalizedStrings);
 931             if (normStrLen > 0) {
 932                 uprv_memset(normalizedStrings + normStrLen, 0, currStrSize - normStrLen);
 933             }
 934         }
 935         stringBlockLength -= currStrSize;
 936         normalizedStrings += currStrSize;
 937         origStringBlock += currStrSize;
 938     }
 939 }
 940
 941 static void
 942 writeAliasTable(UNewDataMemory *out) {
 943     uint32_t i, j;
 944     uint32_t uniqueAliasesSize;
 945     uint16_t aliasOffset = (uint16_t)(tagBlock.top/sizeof(uint16_t));
 946     uint16_t *aliasArrLists = (uint16_t *)uprv_malloc(tagCount * converterCount * sizeof(uint16_t));
 947     uint16_t *uniqueAliases = (uint16_t *)uprv_malloc(knownAliasesCount * sizeof(uint16_t));
 948     uint16_t *uniqueAliasesToConverter = (uint16_t *)uprv_malloc(knownAliasesCount * sizeof(uint16_t));
 949
 950     qsort(knownAliases, knownAliasesCount, sizeof(knownAliases[0]), compareAliases);
 951     uniqueAliasesSize = resolveAliases(uniqueAliases, uniqueAliasesToConverter, aliasOffset);
 952
 953     /* Array index starts at 1. aliasLists[0] is the size of the lists section. */
 954     aliasListsSize = 0;
 955
 956     /* write the offsets of all the aliases lists in a 2D array, and create the lists. */
 957     for (i = 0; i < tagCount; ++i) {
 958         for (j = 0; j < converterCount; ++j) {
 959             createOneAliasList(aliasArrLists, i, j, aliasOffset);
 960         }
 961     }
 962
 963     /* Write the size of the TOC */
 964     if (tableOptions.stringNormalizationType == UCNV_IO_UNNORMALIZED) {
 965         udata_write32(out, 8);
 966     }
 967     else {
 968         udata_write32(out, 9);
 969     }
 970
 971     /* Write the sizes of each section */
 972     /* All sizes are the number of uint16_t units, not bytes */
 973     udata_write32(out, converterCount);
 974     udata_write32(out, tagCount);
 975     udata_write32(out, uniqueAliasesSize);  /* list of aliases */
 976     udata_write32(out, uniqueAliasesSize);  /* The preresolved form of mapping an untagged the alias to a converter */
 977     udata_write32(out, tagCount * converterCount);
 978     udata_write32(out, aliasListsSize + 1);
 979     udata_write32(out, sizeof(tableOptions) / sizeof(uint16_t));
 980     udata_write32(out, (tagBlock.top + stringBlock.top) / sizeof(uint16_t));
 981     if (tableOptions.stringNormalizationType != UCNV_IO_UNNORMALIZED) {
 982         udata_write32(out, (tagBlock.top + stringBlock.top) / sizeof(uint16_t));
 983     }
 984
 985     /* write the table of converters */
 986     /* Think of this as the column headers */
 987     for(i=0; i<converterCount; ++i) {
 988         udata_write16(out, (uint16_t)(converters[i].converter + aliasOffset));
 989     }
 990
 991     /* write the table of tags */
 992     /* Think of this as the row headers */
 993     for(i=UCNV_NUM_RESERVED_TAGS; i<tagCount; ++i) {
 994         udata_write16(out, tags[i].tag);
 995     }
 996     /* The empty tag is considered the leftover list, and put that at the end of the priority list. */
 997     udata_write16(out, tags[EMPTY_TAG_NUM].tag);
 998     udata_write16(out, tags[ALL_TAG_NUM].tag);
 999
1000     /* Write the unique list of aliases */
1001     udata_writeBlock(out, uniqueAliases, uniqueAliasesSize * sizeof(uint16_t));
1002
1003     /* Write the unique list of aliases */
1004     udata_writeBlock(out, uniqueAliasesToConverter, uniqueAliasesSize * sizeof(uint16_t));
1005
1006     /* Write the array to the lists */
1007     udata_writeBlock(out, (const void *)(aliasArrLists + (2*converterCount)), (((tagCount - 2) * converterCount) * sizeof(uint16_t)));
1008     /* Now write the leftover part of the array for the EMPTY and ALL lists */
1009     udata_writeBlock(out, (const void *)aliasArrLists, (2 * converterCount * sizeof(uint16_t)));
1010
1011     /* Offset the next array to make the index start at 1. */
1012     udata_write16(out, 0xDEAD);
1013
1014     /* Write the lists */
1015     udata_writeBlock(out, (const void *)aliasLists, aliasListsSize * sizeof(uint16_t));
1016
1017     /* Write any options for the alias table. */
1018     udata_writeBlock(out, (const void *)&tableOptions, sizeof(tableOptions));
1019
1020     /* write the tags strings */
1021     udata_writeString(out, tagBlock.store, tagBlock.top);
1022
1023     /* write the aliases strings */
1024     udata_writeString(out, stringBlock.store, stringBlock.top);
1025
1026     /* write the normalized aliases strings */
1027     if (tableOptions.stringNormalizationType != UCNV_IO_UNNORMALIZED) {
1028         char *normalizedStrings = (char *)uprv_malloc(tagBlock.top + stringBlock.top);
1029         createNormalizedAliasStrings(normalizedStrings, tagBlock.store, tagBlock.top);
1030         createNormalizedAliasStrings(normalizedStrings + tagBlock.top, stringBlock.store, stringBlock.top);
1031
1032         /* Write out the complete normalized array. */
1033         udata_writeString(out, normalizedStrings, tagBlock.top + stringBlock.top);
1034         uprv_free(normalizedStrings);
1035     }
1036
1037     uprv_free(aliasArrLists);
1038     uprv_free(uniqueAliases);
1039 }
1040
1041 static char *
1042 allocString(StringBlock *block, const char *s, int32_t length) {
1043     uint32_t top;
1044     char *p;
1045
1046     if(length<0) {
1047         length=(int32_t)uprv_strlen(s);
1048     }
1049
1050     /*
1051      * add 1 for the terminating NUL
1052      * and round up (+1 &~1)
1053      * to keep the addresses on a 16-bit boundary
1054      */
1055     top=block->top + (uint32_t)((length + 1 + 1) & ~1);
1056
1057     if(top >= block->max) {
1058         fprintf(stderr, "error(line %d): out of memory\n", lineNum);
1059         exit(U_MEMORY_ALLOCATION_ERROR);
1060     }
1061
1062     /* get the pointer and copy the string */
1063     p = block->store + block->top;
1064     uprv_memcpy(p, s, length);
1065     p[length] = 0; /* NUL-terminate it */
1066     if((length & 1) == 0) {
1067         p[length + 1] = 0; /* set the padding byte */
1068     }
1069
1070     /* check for invariant characters now that we have a NUL-terminated string for easy output */
1071     if(!uprv_isInvariantString(p, length)) {
1072         fprintf(stderr, "error(line %d): the name %s contains not just invariant characters\n", lineNum, p);
1073         exit(U_INVALID_TABLE_FORMAT);
1074     }
1075
1076     block->top = top;
1077     return p;
1078 }
1079
1080 static int
1081 compareAliases(const void *alias1, const void *alias2) {
1082     /* Names like IBM850 and ibm-850 need to be sorted together */
1083     int result = ucnv_compareNames(GET_ALIAS_STR(*(uint16_t*)alias1), GET_ALIAS_STR(*(uint16_t*)alias2));
1084     if (!result) {
1085         /* Sort the shortest first */
1086         return (int)uprv_strlen(GET_ALIAS_STR(*(uint16_t*)alias1)) - (int)uprv_strlen(GET_ALIAS_STR(*(uint16_t*)alias2));
1087     }
1088     return result;
1089 }
1090
1091 /*
1092  * Hey, Emacs, please set the following:
1093  *
1094  * Local Variables:
1095  * indent-tabs-mode: nil
1096  * End:
1097  *
1098  */
1099