icuSources/tools/gencnval/gencnval.c

   1 /*
   2 *******************************************************************************
   3 *
   4 *   Copyright (C) 1999-2011, 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
 146 /**
 147  * path to convrtrs.txt
 148  */
 149 const char *path;
 150
 151 /* prototypes --------------------------------------------------------------- */
 152
 153 static void
 154 parseLine(const char *line);
 155
 156 static void
 157 parseFile(FileStream *in);
 158
 159 static int32_t
 160 chomp(char *line);
 161
 162 static void
 163 addOfficialTaggedStandards(char *line, int32_t lineLen);
 164
 165 static uint16_t
 166 addAlias(const char *alias, uint16_t standard, uint16_t converter, UBool defaultName);
 167
 168 static uint16_t
 169 addConverter(const char *converter);
 170
 171 static char *
 172 allocString(StringBlock *block, const char *s, int32_t length);
 173
 174 static uint16_t
 175 addToKnownAliases(const char *alias);
 176
 177 static int
 178 compareAliases(const void *alias1, const void *alias2);
 179
 180 static uint16_t
 181 getTagNumber(const char *tag, uint16_t tagLen);
 182
 183 /*static void
 184 addTaggedAlias(uint16_t tag, const char *alias, uint16_t converter);*/
 185
 186 static void
 187 writeAliasTable(UNewDataMemory *out);
 188
 189 /* -------------------------------------------------------------------------- */
 190
 191 /* Presumes that you used allocString() */
 192 #define GET_ALIAS_STR(index) (stringStore + ((size_t)(index) << 1))
 193 #define GET_TAG_STR(index) (tagStore + ((size_t)(index) << 1))
 194
 195 /* Presumes that you used allocString() */
 196 #define GET_ALIAS_NUM(str) ((uint16_t)((str - stringStore) >> 1))
 197 #define GET_TAG_NUM(str) ((uint16_t)((str - tagStore) >> 1))
 198
 199 enum
 200 {
 201     HELP1,
 202     HELP2,
 203     VERBOSE,
 204     COPYRIGHT,
 205     DESTDIR,
 206     SOURCEDIR
 207 };
 208
 209 static UOption options[]={
 210     UOPTION_HELP_H,
 211     UOPTION_HELP_QUESTION_MARK,
 212     UOPTION_VERBOSE,
 213     UOPTION_COPYRIGHT,
 214     UOPTION_DESTDIR,
 215     UOPTION_SOURCEDIR
 216 };
 217
 218 extern int
 219 main(int argc, char* argv[]) {
 220     char pathBuf[512];
 221     FileStream *in;
 222     UNewDataMemory *out;
 223     UErrorCode errorCode=U_ZERO_ERROR;
 224
 225     U_MAIN_INIT_ARGS(argc, argv);
 226
 227     /* preset then read command line options */
 228     options[DESTDIR].value=options[SOURCEDIR].value=u_getDataDirectory();
 229     argc=u_parseArgs(argc, argv, sizeof(options)/sizeof(options[0]), options);
 230
 231     /* error handling, printing usage message */
 232     if(argc<0) {
 233         fprintf(stderr,
 234             "error in command line argument \"%s\"\n",
 235             argv[-argc]);
 236     }
 237     if(argc<0 || options[HELP1].doesOccur || options[HELP2].doesOccur) {
 238         fprintf(stderr,
 239             "usage: %s [-options] [convrtrs.txt]\n"
 240             "\tread convrtrs.txt and create " U_ICUDATA_NAME "_" DATA_NAME "." DATA_TYPE "\n"
 241             "options:\n"
 242             "\t-h or -? or --help  this usage text\n"
 243             "\t-v or --verbose     prints out extra information about the alias table\n"
 244             "\t-c or --copyright   include a copyright notice\n"
 245             "\t-d or --destdir     destination directory, followed by the path\n"
 246             "\t-s or --sourcedir   source directory, followed by the path\n",
 247             argv[0]);
 248         return argc<0 ? U_ILLEGAL_ARGUMENT_ERROR : U_ZERO_ERROR;
 249     }
 250
 251     if(options[VERBOSE].doesOccur) {
 252         verbose = TRUE;
 253     }
 254
 255     if(argc>=2) {
 256         path=argv[1];
 257     } else {
 258         path=options[SOURCEDIR].value;
 259         if(path!=NULL && *path!=0) {
 260             char *end;
 261
 262             uprv_strcpy(pathBuf, path);
 263             end = uprv_strchr(pathBuf, 0);
 264             if(*(end-1)!=U_FILE_SEP_CHAR) {
 265                 *(end++)=U_FILE_SEP_CHAR;
 266             }
 267             uprv_strcpy(end, "convrtrs.txt");
 268             path=pathBuf;
 269         } else {
 270             path = "convrtrs.txt";
 271         }
 272     }
 273
 274     uprv_memset(stringStore, 0, sizeof(stringStore));
 275     uprv_memset(tagStore, 0, sizeof(tagStore));
 276     uprv_memset(converters, 0, sizeof(converters));
 277     uprv_memset(tags, 0, sizeof(tags));
 278     uprv_memset(aliasLists, 0, sizeof(aliasLists));
 279     uprv_memset(knownAliases, 0, sizeof(aliasLists));
 280
 281
 282     in=T_FileStream_open(path, "r");
 283     if(in==NULL) {
 284         fprintf(stderr, "gencnval: unable to open input file %s\n", path);
 285         exit(U_FILE_ACCESS_ERROR);
 286     }
 287     parseFile(in);
 288     T_FileStream_close(in);
 289
 290     /* create the output file */
 291     out=udata_create(options[DESTDIR].value, DATA_TYPE, DATA_NAME, &dataInfo,
 292                      options[COPYRIGHT].doesOccur ? U_COPYRIGHT_STRING : NULL, &errorCode);
 293     if(U_FAILURE(errorCode)) {
 294         fprintf(stderr, "gencnval: unable to open output file - error %s\n", u_errorName(errorCode));
 295         exit(errorCode);
 296     }
 297
 298     /* write the table of aliases based on a tag/converter name combination */
 299     writeAliasTable(out);
 300
 301     /* finish */
 302     udata_finish(out, &errorCode);
 303     if(U_FAILURE(errorCode)) {
 304         fprintf(stderr, "gencnval: error finishing output file - %s\n", u_errorName(errorCode));
 305         exit(errorCode);
 306     }
 307
 308     return 0;
 309 }
 310
 311 static void
 312 parseFile(FileStream *in) {
 313     char line[MAX_LINE_SIZE];
 314     char lastLine[MAX_LINE_SIZE];
 315     int32_t lineSize = 0;
 316     int32_t lastLineSize = 0;
 317     UBool validParse = TRUE;
 318
 319     lineNum = 0;
 320
 321     /* Add the empty tag, which is for untagged aliases */
 322     getTagNumber("", 0);
 323     getTagNumber(ALL_TAG_STR, 3);
 324     allocString(&stringBlock, "", 0);
 325
 326     /* read the list of aliases */
 327     while (validParse) {
 328         validParse = FALSE;
 329
 330         /* Read non-empty lines that don't start with a space character. */
 331         while (T_FileStream_readLine(in, lastLine, MAX_LINE_SIZE) != NULL) {
 332             lastLineSize = chomp(lastLine);
 333             if (lineSize == 0 || (lastLineSize > 0 && isspace((int)*lastLine))) {
 334                 uprv_strcpy(line + lineSize, lastLine);
 335                 lineSize += lastLineSize;
 336             } else if (lineSize > 0) {
 337                 validParse = TRUE;
 338                 break;
 339             }
 340             lineNum++;
 341         }
 342
 343         if (validParse || lineSize > 0) {
 344             if (isspace((int)*line)) {
 345                 fprintf(stderr, "%s:%d: error: cannot start an alias with a space\n", path, lineNum-1);
 346                 exit(U_PARSE_ERROR);
 347             } else if (line[0] == '{') {
 348                 if (!standardTagsUsed && line[lineSize - 1] != '}') {
 349                     fprintf(stderr, "%s:%d: error: alias needs to start with a converter name\n", path, lineNum);
 350                     exit(U_PARSE_ERROR);
 351                 }
 352                 addOfficialTaggedStandards(line, lineSize);
 353                 standardTagsUsed = TRUE;
 354             } else {
 355                 if (standardTagsUsed) {
 356                     parseLine(line);
 357                 }
 358                 else {
 359                     fprintf(stderr, "%s:%d: error: alias table needs to start a list of standard tags\n", path, lineNum);
 360                     exit(U_PARSE_ERROR);
 361                 }
 362             }
 363             /* Was the last line consumed */
 364             if (lastLineSize > 0) {
 365                 uprv_strcpy(line, lastLine);
 366                 lineSize = lastLineSize;
 367             }
 368             else {
 369                 lineSize = 0;
 370             }
 371         }
 372         lineNum++;
 373     }
 374 }
 375
 376 /* This works almost like the Perl chomp.
 377  It removes the newlines, comments and trailing whitespace (not preceding whitespace).
 378 */
 379 static int32_t
 380 chomp(char *line) {
 381     char *s = line;
 382     char *lastNonSpace = line;
 383     while(*s!=0) {
 384         /* truncate at a newline or a comment */
 385         if(*s == '\r' || *s == '\n' || *s == '#') {
 386             *s = 0;
 387             break;
 388         }
 389         if (!isspace((int)*s)) {
 390             lastNonSpace = s;
 391         }
 392         ++s;
 393     }
 394     if (lastNonSpace++ > line) {
 395         *lastNonSpace = 0;
 396         s = lastNonSpace;
 397     }
 398     return (int32_t)(s - line);
 399 }
 400
 401 static void
 402 parseLine(const char *line) {
 403     uint16_t pos=0, start, limit, length, cnv;
 404     char *converter, *alias;
 405
 406     /* skip leading white space */
 407     /* There is no whitespace at the beginning anymore */
 408 /*    while(line[pos]!=0 && isspace(line[pos])) {
 409         ++pos;
 410     }
 411 */
 412
 413     /* is there nothing on this line? */
 414     if(line[pos]==0) {
 415         return;
 416     }
 417
 418     /* get the converter name */
 419     start=pos;
 420     while(line[pos]!=0 && !isspace((int)line[pos])) {
 421         ++pos;
 422     }
 423     limit=pos;
 424
 425     /* store the converter name */
 426     length=(uint16_t)(limit-start);
 427     converter=allocString(&stringBlock, line+start, length);
 428
 429     /* add the converter to the converter table */
 430     cnv=addConverter(converter);
 431
 432     /* The name itself may be tagged, so let's added it to the aliases list properly */
 433     pos = start;
 434
 435     /* get all the real aliases */
 436     for(;;) {
 437
 438         /* skip white space */
 439         while(line[pos]!=0 && isspace((int)line[pos])) {
 440             ++pos;
 441         }
 442
 443         /* is there no more alias name on this line? */
 444         if(line[pos]==0) {
 445             break;
 446         }
 447
 448         /* get an alias name */
 449         start=pos;
 450         while(line[pos]!=0 && line[pos]!='{' && !isspace((int)line[pos])) {
 451             ++pos;
 452         }
 453         limit=pos;
 454
 455         /* store the alias name */
 456         length=(uint16_t)(limit-start);
 457         if (start == 0) {
 458             /* add the converter as its own alias to the alias table */
 459             alias = converter;
 460             addAlias(alias, ALL_TAG_NUM, cnv, TRUE);
 461         }
 462         else {
 463             alias=allocString(&stringBlock, line+start, length);
 464             addAlias(alias, ALL_TAG_NUM, cnv, FALSE);
 465         }
 466         addToKnownAliases(alias);
 467
 468         /* add the alias/converter pair to the alias table */
 469         /* addAlias(alias, 0, cnv, FALSE);*/
 470
 471         /* skip whitespace */
 472         while (line[pos] && isspace((int)line[pos])) {
 473             ++pos;
 474         }
 475
 476         /* handle tags if they are present */
 477         if (line[pos] == '{') {
 478             ++pos;
 479             do {
 480                 start = pos;
 481                 while (line[pos] && line[pos] != '}' && !isspace((int)line[pos])) {
 482                     ++pos;
 483                 }
 484                 limit = pos;
 485
 486                 if (start != limit) {
 487                     /* add the tag to the tag table */
 488                     uint16_t tag = getTagNumber(line + start, (uint16_t)(limit - start));
 489                     addAlias(alias, tag, cnv, (UBool)(line[limit-1] == '*'));
 490                 }
 491
 492                 while (line[pos] && isspace((int)line[pos])) {
 493                     ++pos;
 494                 }
 495             } while (line[pos] && line[pos] != '}');
 496
 497             if (line[pos] == '}') {
 498                 ++pos;
 499             } else {
 500                 fprintf(stderr, "%s:%d: Unterminated tag list\n", path, lineNum);
 501                 exit(U_UNMATCHED_BRACES);
 502             }
 503         } else {
 504             addAlias(alias, EMPTY_TAG_NUM, cnv, (UBool)(tags[0].aliasList[cnv].aliasCount == 0));
 505         }
 506     }
 507 }
 508
 509 static uint16_t
 510 getTagNumber(const char *tag, uint16_t tagLen) {
 511     char *atag;
 512     uint16_t t;
 513     UBool preferredName = ((tagLen > 0) ? (tag[tagLen - 1] == '*') : (FALSE));
 514
 515     if (tagCount >= MAX_TAG_COUNT) {
 516         fprintf(stderr, "%s:%d: too many tags\n", path, lineNum);
 517         exit(U_BUFFER_OVERFLOW_ERROR);
 518     }
 519
 520     if (preferredName) {
 521 /*        puts(tag);*/
 522         tagLen--;
 523     }
 524
 525     for (t = 0; t < tagCount; ++t) {
 526         const char *currTag = GET_TAG_STR(tags[t].tag);
 527         if (uprv_strlen(currTag) == tagLen && !uprv_strnicmp(currTag, tag, tagLen)) {
 528             return t;
 529         }
 530     }
 531
 532     /* we need to add this tag */
 533     if (tagCount >= MAX_TAG_COUNT) {
 534         fprintf(stderr, "%s:%d: error: too many tags\n", path, lineNum);
 535         exit(U_BUFFER_OVERFLOW_ERROR);
 536     }
 537
 538     /* allocate a new entry in the tag table */
 539     atag = allocString(&tagBlock, tag, tagLen);
 540
 541     if (standardTagsUsed) {
 542         fprintf(stderr, "%s:%d: error: Tag \"%s\" is not declared at the beginning of the alias table.\n",
 543             path, lineNum, atag);
 544         exit(1);
 545     }
 546     else if (tagLen > 0 && strcmp(tag, ALL_TAG_STR) != 0) {
 547         fprintf(stderr, "%s:%d: warning: Tag \"%s\" was added to the list of standards because it was not declared at beginning of the alias table.\n",
 548             path, lineNum, atag);
 549     }
 550
 551     /* add the tag to the tag table */
 552     tags[tagCount].tag = GET_TAG_NUM(atag);
 553     /* The aliasList should be set to 0's already */
 554
 555     return tagCount++;
 556 }
 557
 558 /*static void
 559 addTaggedAlias(uint16_t tag, const char *alias, uint16_t converter) {
 560     tags[tag].aliases[converter] = alias;
 561 }
 562 */
 563
 564 static void
 565 addOfficialTaggedStandards(char *line, int32_t lineLen) {
 566     char *atag;
 567     char *endTagExp;
 568     char *tag;
 569     static const char WHITESPACE[] = " \t";
 570
 571     if (tagCount > UCNV_NUM_RESERVED_TAGS) {
 572         fprintf(stderr, "%s:%d: error: official tags already added\n", path, lineNum);
 573         exit(U_BUFFER_OVERFLOW_ERROR);
 574     }
 575     tag = strchr(line, '{');
 576     if (tag == NULL) {
 577         /* Why were we called? */
 578         fprintf(stderr, "%s:%d: error: Missing start of tag group\n", path, lineNum);
 579         exit(U_PARSE_ERROR);
 580     }
 581     tag++;
 582     endTagExp = strchr(tag, '}');
 583     if (endTagExp == NULL) {
 584         fprintf(stderr, "%s:%d: error: Missing end of tag group\n", path, lineNum);
 585         exit(U_PARSE_ERROR);
 586     }
 587     endTagExp[0] = 0;
 588
 589     tag = strtok(tag, WHITESPACE);
 590     while (tag != NULL) {
 591 /*        printf("Adding original tag \"%s\"\n", tag);*/
 592
 593         /* allocate a new entry in the tag table */
 594         atag = allocString(&tagBlock, tag, -1);
 595
 596         /* add the tag to the tag table */
 597         tags[tagCount++].tag = (uint16_t)((atag - tagStore) >> 1);
 598
 599         /* The aliasList should already be set to 0's */
 600
 601         /* Get next tag */
 602         tag = strtok(NULL, WHITESPACE);
 603     }
 604 }
 605
 606 static uint16_t
 607 addToKnownAliases(const char *alias) {
 608 /*    uint32_t idx; */
 609     /* strict matching */
 610 /*    for (idx = 0; idx < knownAliasesCount; idx++) {
 611         uint16_t num = GET_ALIAS_NUM(alias);
 612         if (knownAliases[idx] != num
 613             && uprv_strcmp(alias, GET_ALIAS_STR(knownAliases[idx])) == 0)
 614         {
 615             fprintf(stderr, "%s:%d: warning: duplicate alias %s and %s found\n", path,
 616                 lineNum, alias, GET_ALIAS_STR(knownAliases[idx]));
 617             duplicateKnownAliasesCount++;
 618             break;
 619         }
 620         else if (knownAliases[idx] != num
 621             && ucnv_compareNames(alias, GET_ALIAS_STR(knownAliases[idx])) == 0)
 622         {
 623             if (verbose) {
 624                 fprintf(stderr, "%s:%d: information: duplicate alias %s and %s found\n", path,
 625                     lineNum, alias, GET_ALIAS_STR(knownAliases[idx]));
 626             }
 627             duplicateKnownAliasesCount++;
 628             break;
 629         }
 630     }
 631 */
 632     if (knownAliasesCount >= MAX_ALIAS_COUNT) {
 633         fprintf(stderr, "%s:%d: warning: Too many aliases defined for all converters\n",
 634             path, lineNum);
 635         exit(U_BUFFER_OVERFLOW_ERROR);
 636     }
 637     /* TODO: We could try to unlist exact duplicates. */
 638     return knownAliases[knownAliasesCount++] = GET_ALIAS_NUM(alias);
 639 }
 640
 641 /*
 642 @param standard When standard is 0, then it's the "empty" tag.
 643 */
 644 static uint16_t
 645 addAlias(const char *alias, uint16_t standard, uint16_t converter, UBool defaultName) {
 646     uint32_t idx, idx2;
 647     UBool dupFound = FALSE;
 648     UBool startEmptyWithoutDefault = FALSE;
 649     AliasList *aliasList;
 650
 651     if(standard>=MAX_TAG_COUNT) {
 652         fprintf(stderr, "%s:%d: error: too many standard tags\n", path, lineNum);
 653         exit(U_BUFFER_OVERFLOW_ERROR);
 654     }
 655     if(converter>=MAX_CONV_COUNT) {
 656         fprintf(stderr, "%s:%d: error: too many converter names\n", path, lineNum);
 657         exit(U_BUFFER_OVERFLOW_ERROR);
 658     }
 659     aliasList = &tags[standard].aliasList[converter];
 660
 661     if (strchr(alias, '}')) {
 662         fprintf(stderr, "%s:%d: error: unmatched } found\n", path,
 663             lineNum);
 664     }
 665
 666     if(aliasList->aliasCount + 1 >= MAX_TC_ALIAS_COUNT) {
 667         fprintf(stderr, "%s:%d: error: too many aliases for alias %s and converter %s\n", path,
 668             lineNum, alias, GET_ALIAS_STR(converters[converter].converter));
 669         exit(U_BUFFER_OVERFLOW_ERROR);
 670     }
 671
 672     /* Show this warning only once. All aliases are added to the "ALL" tag. */
 673     if (standard == ALL_TAG_NUM && GET_ALIAS_STR(converters[converter].converter) != alias) {
 674         /* Normally these option values are parsed at runtime, and they can
 675            be discarded when the alias is a default converter. Options should
 676            only be on a converter and not an alias. */
 677         if (uprv_strchr(alias, UCNV_OPTION_SEP_CHAR) != 0)
 678         {
 679             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",
 680                 lineNum, alias);
 681         }
 682         if (uprv_strchr(alias, UCNV_VALUE_SEP_CHAR) != 0)
 683         {
 684             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",
 685                 lineNum, alias);
 686         }
 687     }
 688
 689     if (standard != ALL_TAG_NUM) {
 690         /* Check for duplicate aliases for this tag on all converters */
 691         for (idx = 0; idx < converterCount; idx++) {
 692             for (idx2 = 0; idx2 < tags[standard].aliasList[idx].aliasCount; idx2++) {
 693                 uint16_t aliasNum = tags[standard].aliasList[idx].aliases[idx2];
 694                 if (aliasNum
 695                     && ucnv_compareNames(alias, GET_ALIAS_STR(aliasNum)) == 0)
 696                 {
 697                     if (idx == converter) {
 698                         /*
 699                          * (alias, standard) duplicates are harmless if they map to the same converter.
 700                          * Only print a warning in verbose mode, or if the alias is a precise duplicate,
 701                          * not just a lenient-match duplicate.
 702                          */
 703                         if (verbose || 0 == uprv_strcmp(alias, GET_ALIAS_STR(aliasNum))) {
 704                             fprintf(stderr, "%s:%d: warning: duplicate aliases %s and %s found for standard %s and converter %s\n", path,
 705                                 lineNum, alias, GET_ALIAS_STR(aliasNum),
 706                                 GET_TAG_STR(tags[standard].tag),
 707                                 GET_ALIAS_STR(converters[converter].converter));
 708                         }
 709                     } else {
 710                         fprintf(stderr, "%s:%d: warning: duplicate aliases %s and %s found for standard tag %s between converter %s and converter %s\n", path,
 711                             lineNum, alias, GET_ALIAS_STR(aliasNum),
 712                             GET_TAG_STR(tags[standard].tag),
 713                             GET_ALIAS_STR(converters[converter].converter),
 714                             GET_ALIAS_STR(converters[idx].converter));
 715                     }
 716                     dupFound = TRUE;
 717                     break;
 718                 }
 719             }
 720         }
 721
 722         /* Check for duplicate default aliases for this converter on all tags */
 723         /* It's okay to have multiple standards prefer the same name */
 724 /*        if (verbose && !dupFound) {
 725             for (idx = 0; idx < tagCount; idx++) {
 726                 if (tags[idx].aliasList[converter].aliases) {
 727                     uint16_t aliasNum = tags[idx].aliasList[converter].aliases[0];
 728                     if (aliasNum
 729                         && ucnv_compareNames(alias, GET_ALIAS_STR(aliasNum)) == 0)
 730                     {
 731                         fprintf(stderr, "%s:%d: warning: duplicate alias %s found for converter %s and standard tag %s\n", path,
 732                             lineNum, alias, GET_ALIAS_STR(converters[converter].converter), GET_TAG_STR(tags[standard].tag));
 733                         break;
 734                     }
 735                 }
 736             }
 737         }*/
 738     }
 739
 740     if (aliasList->aliasCount <= 0) {
 741         aliasList->aliasCount++;
 742         startEmptyWithoutDefault = TRUE;
 743     }
 744     aliasList->aliases = (uint16_t *)uprv_realloc(aliasList->aliases, (aliasList->aliasCount + 1) * sizeof(aliasList->aliases[0]));
 745     if (startEmptyWithoutDefault) {
 746         aliasList->aliases[0] = 0;
 747     }
 748     if (defaultName) {
 749         if (aliasList->aliases[0] != 0) {
 750             fprintf(stderr, "%s:%d: error: Alias %s and %s cannot both be the default alias for standard tag %s and converter %s\n", path,
 751                 lineNum,
 752                 alias,
 753                 GET_ALIAS_STR(aliasList->aliases[0]),
 754                 GET_TAG_STR(tags[standard].tag),
 755                 GET_ALIAS_STR(converters[converter].converter));
 756             exit(U_PARSE_ERROR);
 757         }
 758         aliasList->aliases[0] = GET_ALIAS_NUM(alias);
 759     } else {
 760         aliasList->aliases[aliasList->aliasCount++] = GET_ALIAS_NUM(alias);
 761     }
 762 /*    aliasList->converter = converter;*/
 763
 764     converters[converter].totalAliasCount++; /* One more to the column */
 765     tags[standard].totalAliasCount++; /* One more to the row */
 766
 767     return aliasList->aliasCount;
 768 }
 769
 770 static uint16_t
 771 addConverter(const char *converter) {
 772     uint32_t idx;
 773     if(converterCount>=MAX_CONV_COUNT) {
 774         fprintf(stderr, "%s:%d: error: too many converters\n", path, lineNum);
 775         exit(U_BUFFER_OVERFLOW_ERROR);
 776     }
 777
 778     for (idx = 0; idx < converterCount; idx++) {
 779         if (ucnv_compareNames(converter, GET_ALIAS_STR(converters[idx].converter)) == 0) {
 780             fprintf(stderr, "%s:%d: error: duplicate converter %s found!\n", path, lineNum, converter);
 781             exit(U_PARSE_ERROR);
 782             break;
 783         }
 784     }
 785
 786     converters[converterCount].converter = GET_ALIAS_NUM(converter);
 787     converters[converterCount].totalAliasCount = 0;
 788
 789     return converterCount++;
 790 }
 791
 792 /* resolve this alias based on the prioritization of the standard tags. */
 793 static void
 794 resolveAliasToConverter(uint16_t alias, uint16_t *tagNum, uint16_t *converterNum) {
 795     uint16_t idx, idx2, idx3;
 796
 797     for (idx = UCNV_NUM_RESERVED_TAGS; idx < tagCount; idx++) {
 798         for (idx2 = 0; idx2 < converterCount; idx2++) {
 799             for (idx3 = 0; idx3 < tags[idx].aliasList[idx2].aliasCount; idx3++) {
 800                 uint16_t aliasNum = tags[idx].aliasList[idx2].aliases[idx3];
 801                 if (aliasNum == alias) {
 802                     *tagNum = idx;
 803                     *converterNum = idx2;
 804                     return;
 805                 }
 806             }
 807         }
 808     }
 809     /* Do the leftovers last, just in case */
 810     /* There is no need to do the ALL tag */
 811     idx = 0;
 812     for (idx2 = 0; idx2 < converterCount; idx2++) {
 813         for (idx3 = 0; idx3 < tags[idx].aliasList[idx2].aliasCount; idx3++) {
 814             uint16_t aliasNum = tags[idx].aliasList[idx2].aliases[idx3];
 815             if (aliasNum == alias) {
 816                 *tagNum = idx;
 817                 *converterNum = idx2;
 818                 return;
 819             }
 820         }
 821     }
 822     *tagNum = UINT16_MAX;
 823     *converterNum = UINT16_MAX;
 824     fprintf(stderr, "%s: warning: alias %s not found\n",
 825         path,
 826         GET_ALIAS_STR(alias));
 827     return;
 828 }
 829
 830 /* The knownAliases should be sorted before calling this function */
 831 static uint32_t
 832 resolveAliases(uint16_t *uniqueAliasArr, uint16_t *uniqueAliasToConverterArr, uint16_t aliasOffset) {
 833     uint32_t uniqueAliasIdx = 0;
 834     uint32_t idx;
 835     uint16_t currTagNum, oldTagNum;
 836     uint16_t currConvNum, oldConvNum;
 837     const char *lastName;
 838
 839     resolveAliasToConverter(knownAliases[0], &oldTagNum, &currConvNum);
 840     uniqueAliasToConverterArr[uniqueAliasIdx] = currConvNum;
 841     oldConvNum = currConvNum;
 842     uniqueAliasArr[uniqueAliasIdx] = knownAliases[0] + aliasOffset;
 843     uniqueAliasIdx++;
 844     lastName = GET_ALIAS_STR(knownAliases[0]);
 845
 846     for (idx = 1; idx < knownAliasesCount; idx++) {
 847         resolveAliasToConverter(knownAliases[idx], &currTagNum, &currConvNum);
 848         if (ucnv_compareNames(lastName, GET_ALIAS_STR(knownAliases[idx])) == 0) {
 849             /* duplicate found */
 850             if ((currTagNum < oldTagNum && currTagNum >= UCNV_NUM_RESERVED_TAGS)
 851                 || oldTagNum == 0) {
 852                 oldTagNum = currTagNum;
 853                 uniqueAliasToConverterArr[uniqueAliasIdx - 1] = currConvNum;
 854                 uniqueAliasArr[uniqueAliasIdx - 1] = knownAliases[idx] + aliasOffset;
 855                 if (verbose) {
 856                     printf("using %s instead of %s -> %s",
 857                         GET_ALIAS_STR(knownAliases[idx]),
 858                         lastName,
 859                         GET_ALIAS_STR(converters[currConvNum].converter));
 860                     if (oldConvNum != currConvNum) {
 861                         printf(" (alias conflict)");
 862                     }
 863                     puts("");
 864                 }
 865             }
 866             else {
 867                 /* else ignore it */
 868                 if (verbose) {
 869                     printf("folding %s into %s -> %s",
 870                         GET_ALIAS_STR(knownAliases[idx]),
 871                         lastName,
 872                         GET_ALIAS_STR(converters[oldConvNum].converter));
 873                     if (oldConvNum != currConvNum) {
 874                         printf(" (alias conflict)");
 875                     }
 876                     puts("");
 877                 }
 878             }
 879             if (oldConvNum != currConvNum) {
 880                 uniqueAliasToConverterArr[uniqueAliasIdx - 1] |= UCNV_AMBIGUOUS_ALIAS_MAP_BIT;
 881             }
 882         }
 883         else {
 884             uniqueAliasToConverterArr[uniqueAliasIdx] = currConvNum;
 885             oldConvNum = currConvNum;
 886             uniqueAliasArr[uniqueAliasIdx] = knownAliases[idx] + aliasOffset;
 887             uniqueAliasIdx++;
 888             lastName = GET_ALIAS_STR(knownAliases[idx]);
 889             oldTagNum = currTagNum;
 890             /*printf("%s -> %s\n", GET_ALIAS_STR(knownAliases[idx]), GET_ALIAS_STR(converters[currConvNum].converter));*/
 891         }
 892         if (uprv_strchr(GET_ALIAS_STR(converters[currConvNum].converter), UCNV_OPTION_SEP_CHAR) != NULL) {
 893             uniqueAliasToConverterArr[uniqueAliasIdx-1] |= UCNV_CONTAINS_OPTION_BIT;
 894         }
 895     }
 896     return uniqueAliasIdx;
 897 }
 898
 899 static void
 900 createOneAliasList(uint16_t *aliasArrLists, uint32_t tag, uint32_t converter, uint16_t offset) {
 901     uint32_t aliasNum;
 902     AliasList *aliasList = &tags[tag].aliasList[converter];
 903
 904     if (aliasList->aliasCount == 0) {
 905         aliasArrLists[tag*converterCount + converter] = 0;
 906     }
 907     else {
 908         aliasLists[aliasListsSize++] = aliasList->aliasCount;
 909
 910         /* write into the array area a 1's based index. */
 911         aliasArrLists[tag*converterCount + converter] = aliasListsSize;
 912
 913 /*        printf("tag %s converter %s\n",
 914             GET_TAG_STR(tags[tag].tag),
 915             GET_ALIAS_STR(converters[converter].converter));*/
 916         for (aliasNum = 0; aliasNum < aliasList->aliasCount; aliasNum++) {
 917             uint16_t value;
 918 /*            printf("   %s\n",
 919                 GET_ALIAS_STR(aliasList->aliases[aliasNum]));*/
 920             if (aliasList->aliases[aliasNum]) {
 921                 value = aliasList->aliases[aliasNum] + offset;
 922             } else {
 923                 value = 0;
 924                 if (tag != 0) { /* Only show the warning when it's not the leftover tag. */
 925                     fprintf(stderr, "%s: warning: tag %s does not have a default alias for %s\n",
 926                             path,
 927                             GET_TAG_STR(tags[tag].tag),
 928                             GET_ALIAS_STR(converters[converter].converter));
 929                 }
 930             }
 931             aliasLists[aliasListsSize++] = value;
 932             if (aliasListsSize >= MAX_LIST_SIZE) {
 933                 fprintf(stderr, "%s: error: Too many alias lists\n", path);
 934                 exit(U_BUFFER_OVERFLOW_ERROR);
 935             }
 936
 937         }
 938     }
 939 }
 940
 941 static void
 942 createNormalizedAliasStrings(char *normalizedStrings, const char *origStringBlock, int32_t stringBlockLength) {
 943     int32_t currStrLen;
 944     uprv_memcpy(normalizedStrings, origStringBlock, stringBlockLength);
 945     while ((currStrLen = (int32_t)uprv_strlen(origStringBlock)) < stringBlockLength) {
 946         int32_t currStrSize = currStrLen + 1;
 947         if (currStrLen > 0) {
 948             int32_t normStrLen;
 949             ucnv_io_stripForCompare(normalizedStrings, origStringBlock);
 950             normStrLen = uprv_strlen(normalizedStrings);
 951             if (normStrLen > 0) {
 952                 uprv_memset(normalizedStrings + normStrLen, 0, currStrSize - normStrLen);
 953             }
 954         }
 955         stringBlockLength -= currStrSize;
 956         normalizedStrings += currStrSize;
 957         origStringBlock += currStrSize;
 958     }
 959 }
 960
 961 static void
 962 writeAliasTable(UNewDataMemory *out) {
 963     uint32_t i, j;
 964     uint32_t uniqueAliasesSize;
 965     uint16_t aliasOffset = (uint16_t)(tagBlock.top/sizeof(uint16_t));
 966     uint16_t *aliasArrLists = (uint16_t *)uprv_malloc(tagCount * converterCount * sizeof(uint16_t));
 967     uint16_t *uniqueAliases = (uint16_t *)uprv_malloc(knownAliasesCount * sizeof(uint16_t));
 968     uint16_t *uniqueAliasesToConverter = (uint16_t *)uprv_malloc(knownAliasesCount * sizeof(uint16_t));
 969
 970     qsort(knownAliases, knownAliasesCount, sizeof(knownAliases[0]), compareAliases);
 971     uniqueAliasesSize = resolveAliases(uniqueAliases, uniqueAliasesToConverter, aliasOffset);
 972
 973     /* Array index starts at 1. aliasLists[0] is the size of the lists section. */
 974     aliasListsSize = 0;
 975
 976     /* write the offsets of all the aliases lists in a 2D array, and create the lists. */
 977     for (i = 0; i < tagCount; ++i) {
 978         for (j = 0; j < converterCount; ++j) {
 979             createOneAliasList(aliasArrLists, i, j, aliasOffset);
 980         }
 981     }
 982
 983     /* Write the size of the TOC */
 984     if (tableOptions.stringNormalizationType == UCNV_IO_UNNORMALIZED) {
 985         udata_write32(out, 8);
 986     }
 987     else {
 988         udata_write32(out, 9);
 989     }
 990
 991     /* Write the sizes of each section */
 992     /* All sizes are the number of uint16_t units, not bytes */
 993     udata_write32(out, converterCount);
 994     udata_write32(out, tagCount);
 995     udata_write32(out, uniqueAliasesSize);  /* list of aliases */
 996     udata_write32(out, uniqueAliasesSize);  /* The preresolved form of mapping an untagged the alias to a converter */
 997     udata_write32(out, tagCount * converterCount);
 998     udata_write32(out, aliasListsSize + 1);
 999     udata_write32(out, sizeof(tableOptions) / sizeof(uint16_t));
1000     udata_write32(out, (tagBlock.top + stringBlock.top) / sizeof(uint16_t));
1001     if (tableOptions.stringNormalizationType != UCNV_IO_UNNORMALIZED) {
1002         udata_write32(out, (tagBlock.top + stringBlock.top) / sizeof(uint16_t));
1003     }
1004
1005     /* write the table of converters */
1006     /* Think of this as the column headers */
1007     for(i=0; i<converterCount; ++i) {
1008         udata_write16(out, (uint16_t)(converters[i].converter + aliasOffset));
1009     }
1010
1011     /* write the table of tags */
1012     /* Think of this as the row headers */
1013     for(i=UCNV_NUM_RESERVED_TAGS; i<tagCount; ++i) {
1014         udata_write16(out, tags[i].tag);
1015     }
1016     /* The empty tag is considered the leftover list, and put that at the end of the priority list. */
1017     udata_write16(out, tags[EMPTY_TAG_NUM].tag);
1018     udata_write16(out, tags[ALL_TAG_NUM].tag);
1019
1020     /* Write the unique list of aliases */
1021     udata_writeBlock(out, uniqueAliases, uniqueAliasesSize * sizeof(uint16_t));
1022
1023     /* Write the unique list of aliases */
1024     udata_writeBlock(out, uniqueAliasesToConverter, uniqueAliasesSize * sizeof(uint16_t));
1025
1026     /* Write the array to the lists */
1027     udata_writeBlock(out, (const void *)(aliasArrLists + (2*converterCount)), (((tagCount - 2) * converterCount) * sizeof(uint16_t)));
1028     /* Now write the leftover part of the array for the EMPTY and ALL lists */
1029     udata_writeBlock(out, (const void *)aliasArrLists, (2 * converterCount * sizeof(uint16_t)));
1030
1031     /* Offset the next array to make the index start at 1. */
1032     udata_write16(out, 0xDEAD);
1033
1034     /* Write the lists */
1035     udata_writeBlock(out, (const void *)aliasLists, aliasListsSize * sizeof(uint16_t));
1036
1037     /* Write any options for the alias table. */
1038     udata_writeBlock(out, (const void *)&tableOptions, sizeof(tableOptions));
1039
1040     /* write the tags strings */
1041     udata_writeString(out, tagBlock.store, tagBlock.top);
1042
1043     /* write the aliases strings */
1044     udata_writeString(out, stringBlock.store, stringBlock.top);
1045
1046     /* write the normalized aliases strings */
1047     if (tableOptions.stringNormalizationType != UCNV_IO_UNNORMALIZED) {
1048         char *normalizedStrings = (char *)uprv_malloc(tagBlock.top + stringBlock.top);
1049         createNormalizedAliasStrings(normalizedStrings, tagBlock.store, tagBlock.top);
1050         createNormalizedAliasStrings(normalizedStrings + tagBlock.top, stringBlock.store, stringBlock.top);
1051
1052         /* Write out the complete normalized array. */
1053         udata_writeString(out, normalizedStrings, tagBlock.top + stringBlock.top);
1054         uprv_free(normalizedStrings);
1055     }
1056
1057     uprv_free(aliasArrLists);
1058     uprv_free(uniqueAliases);
1059 }
1060
1061 static char *
1062 allocString(StringBlock *block, const char *s, int32_t length) {
1063     uint32_t top;
1064     char *p;
1065
1066     if(length<0) {
1067         length=(int32_t)uprv_strlen(s);
1068     }
1069
1070     /*
1071      * add 1 for the terminating NUL
1072      * and round up (+1 &~1)
1073      * to keep the addresses on a 16-bit boundary
1074      */
1075     top=block->top + (uint32_t)((length + 1 + 1) & ~1);
1076
1077     if(top >= block->max) {
1078         fprintf(stderr, "%s:%d: error: out of memory\n", path, lineNum);
1079         exit(U_MEMORY_ALLOCATION_ERROR);
1080     }
1081
1082     /* get the pointer and copy the string */
1083     p = block->store + block->top;
1084     uprv_memcpy(p, s, length);
1085     p[length] = 0; /* NUL-terminate it */
1086     if((length & 1) == 0) {
1087         p[length + 1] = 0; /* set the padding byte */
1088     }
1089
1090     /* check for invariant characters now that we have a NUL-terminated string for easy output */
1091     if(!uprv_isInvariantString(p, length)) {
1092         fprintf(stderr, "%s:%d: error: the name %s contains not just invariant characters\n", path, lineNum, p);
1093         exit(U_INVALID_TABLE_FORMAT);
1094     }
1095
1096     block->top = top;
1097     return p;
1098 }
1099
1100 static int
1101 compareAliases(const void *alias1, const void *alias2) {
1102     /* Names like IBM850 and ibm-850 need to be sorted together */
1103     int result = ucnv_compareNames(GET_ALIAS_STR(*(uint16_t*)alias1), GET_ALIAS_STR(*(uint16_t*)alias2));
1104     if (!result) {
1105         /* Sort the shortest first */
1106         return (int)uprv_strlen(GET_ALIAS_STR(*(uint16_t*)alias1)) - (int)uprv_strlen(GET_ALIAS_STR(*(uint16_t*)alias2));
1107     }
1108     return result;
1109 }
1110
1111 /*
1112  * Hey, Emacs, please set the following:
1113  *
1114  * Local Variables:
1115  * indent-tabs-mode: nil
1116  * End:
1117  *
1118  */
1119