icuSources/tools/gencnval/gencnval.c

   1 /*
   2 *******************************************************************************
   3 *
   4 *   Copyright (C) 1999-2012, 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     int i, n;
 221     char pathBuf[512];
 222     FileStream *in;
 223     UNewDataMemory *out;
 224     UErrorCode errorCode=U_ZERO_ERROR;
 225
 226     U_MAIN_INIT_ARGS(argc, argv);
 227
 228     /* preset then read command line options */
 229     options[DESTDIR].value=options[SOURCEDIR].value=u_getDataDirectory();
 230     argc=u_parseArgs(argc, argv, sizeof(options)/sizeof(options[0]), options);
 231
 232     /* error handling, printing usage message */
 233     if(argc<0) {
 234         fprintf(stderr,
 235             "error in command line argument \"%s\"\n",
 236             argv[-argc]);
 237     }
 238     if(argc<0 || options[HELP1].doesOccur || options[HELP2].doesOccur) {
 239         fprintf(stderr,
 240             "usage: %s [-options] [convrtrs.txt]\n"
 241             "\tread convrtrs.txt and create " U_ICUDATA_NAME "_" DATA_NAME "." DATA_TYPE "\n"
 242             "options:\n"
 243             "\t-h or -? or --help  this usage text\n"
 244             "\t-v or --verbose     prints out extra information about the alias table\n"
 245             "\t-c or --copyright   include a copyright notice\n"
 246             "\t-d or --destdir     destination directory, followed by the path\n"
 247             "\t-s or --sourcedir   source directory, followed by the path\n",
 248             argv[0]);
 249         return argc<0 ? U_ILLEGAL_ARGUMENT_ERROR : U_ZERO_ERROR;
 250     }
 251
 252     if(options[VERBOSE].doesOccur) {
 253         verbose = TRUE;
 254     }
 255
 256     if(argc>=2) {
 257         path=argv[1];
 258     } else {
 259         path=options[SOURCEDIR].value;
 260         if(path!=NULL && *path!=0) {
 261             char *end;
 262
 263             uprv_strcpy(pathBuf, path);
 264             end = uprv_strchr(pathBuf, 0);
 265             if(*(end-1)!=U_FILE_SEP_CHAR) {
 266                 *(end++)=U_FILE_SEP_CHAR;
 267             }
 268             uprv_strcpy(end, "convrtrs.txt");
 269             path=pathBuf;
 270         } else {
 271             path = "convrtrs.txt";
 272         }
 273     }
 274
 275     uprv_memset(stringStore, 0, sizeof(stringStore));
 276     uprv_memset(tagStore, 0, sizeof(tagStore));
 277     uprv_memset(converters, 0, sizeof(converters));
 278     uprv_memset(tags, 0, sizeof(tags));
 279     uprv_memset(aliasLists, 0, sizeof(aliasLists));
 280     uprv_memset(knownAliases, 0, sizeof(aliasLists));
 281
 282
 283     in=T_FileStream_open(path, "r");
 284     if(in==NULL) {
 285         fprintf(stderr, "gencnval: unable to open input file %s\n", path);
 286         exit(U_FILE_ACCESS_ERROR);
 287     }
 288     parseFile(in);
 289     T_FileStream_close(in);
 290
 291     /* create the output file */
 292     out=udata_create(options[DESTDIR].value, DATA_TYPE, DATA_NAME, &dataInfo,
 293                      options[COPYRIGHT].doesOccur ? U_COPYRIGHT_STRING : NULL, &errorCode);
 294     if(U_FAILURE(errorCode)) {
 295         fprintf(stderr, "gencnval: unable to open output file - error %s\n", u_errorName(errorCode));
 296         exit(errorCode);
 297     }
 298
 299     /* write the table of aliases based on a tag/converter name combination */
 300     writeAliasTable(out);
 301
 302     /* finish */
 303     udata_finish(out, &errorCode);
 304     if(U_FAILURE(errorCode)) {
 305         fprintf(stderr, "gencnval: error finishing output file - %s\n", u_errorName(errorCode));
 306         exit(errorCode);
 307     }
 308
 309     /* clean up tags */
 310     for (i = 0; i < MAX_TAG_COUNT; i++) {
 311         for (n = 0; n < MAX_CONV_COUNT; n++) {
 312             if (tags[i].aliasList[n].aliases!=NULL) {
 313                 uprv_free(tags[i].aliasList[n].aliases);
 314             }
 315         }
 316     }
 317
 318     return 0;
 319 }
 320
 321 static void
 322 parseFile(FileStream *in) {
 323     char line[MAX_LINE_SIZE];
 324     char lastLine[MAX_LINE_SIZE];
 325     int32_t lineSize = 0;
 326     int32_t lastLineSize = 0;
 327     UBool validParse = TRUE;
 328
 329     lineNum = 0;
 330
 331     /* Add the empty tag, which is for untagged aliases */
 332     getTagNumber("", 0);
 333     getTagNumber(ALL_TAG_STR, 3);
 334     allocString(&stringBlock, "", 0);
 335
 336     /* read the list of aliases */
 337     while (validParse) {
 338         validParse = FALSE;
 339
 340         /* Read non-empty lines that don't start with a space character. */
 341         while (T_FileStream_readLine(in, lastLine, MAX_LINE_SIZE) != NULL) {
 342             lastLineSize = chomp(lastLine);
 343             if (lineSize == 0 || (lastLineSize > 0 && isspace((int)*lastLine))) {
 344                 uprv_strcpy(line + lineSize, lastLine);
 345                 lineSize += lastLineSize;
 346             } else if (lineSize > 0) {
 347                 validParse = TRUE;
 348                 break;
 349             }
 350             lineNum++;
 351         }
 352
 353         if (validParse || lineSize > 0) {
 354             if (isspace((int)*line)) {
 355                 fprintf(stderr, "%s:%d: error: cannot start an alias with a space\n", path, lineNum-1);
 356                 exit(U_PARSE_ERROR);
 357             } else if (line[0] == '{') {
 358                 if (!standardTagsUsed && line[lineSize - 1] != '}') {
 359                     fprintf(stderr, "%s:%d: error: alias needs to start with a converter name\n", path, lineNum);
 360                     exit(U_PARSE_ERROR);
 361                 }
 362                 addOfficialTaggedStandards(line, lineSize);
 363                 standardTagsUsed = TRUE;
 364             } else {
 365                 if (standardTagsUsed) {
 366                     parseLine(line);
 367                 }
 368                 else {
 369                     fprintf(stderr, "%s:%d: error: alias table needs to start a list of standard tags\n", path, lineNum);
 370                     exit(U_PARSE_ERROR);
 371                 }
 372             }
 373             /* Was the last line consumed */
 374             if (lastLineSize > 0) {
 375                 uprv_strcpy(line, lastLine);
 376                 lineSize = lastLineSize;
 377             }
 378             else {
 379                 lineSize = 0;
 380             }
 381         }
 382         lineNum++;
 383     }
 384 }
 385
 386 /* This works almost like the Perl chomp.
 387  It removes the newlines, comments and trailing whitespace (not preceding whitespace).
 388 */
 389 static int32_t
 390 chomp(char *line) {
 391     char *s = line;
 392     char *lastNonSpace = line;
 393     while(*s!=0) {
 394         /* truncate at a newline or a comment */
 395         if(*s == '\r' || *s == '\n' || *s == '#') {
 396             *s = 0;
 397             break;
 398         }
 399         if (!isspace((int)*s)) {
 400             lastNonSpace = s;
 401         }
 402         ++s;
 403     }
 404     if (lastNonSpace++ > line) {
 405         *lastNonSpace = 0;
 406         s = lastNonSpace;
 407     }
 408     return (int32_t)(s - line);
 409 }
 410
 411 static void
 412 parseLine(const char *line) {
 413     uint16_t pos=0, start, limit, length, cnv;
 414     char *converter, *alias;
 415
 416     /* skip leading white space */
 417     /* There is no whitespace at the beginning anymore */
 418 /*    while(line[pos]!=0 && isspace(line[pos])) {
 419         ++pos;
 420     }
 421 */
 422
 423     /* is there nothing on this line? */
 424     if(line[pos]==0) {
 425         return;
 426     }
 427
 428     /* get the converter name */
 429     start=pos;
 430     while(line[pos]!=0 && !isspace((int)line[pos])) {
 431         ++pos;
 432     }
 433     limit=pos;
 434
 435     /* store the converter name */
 436     length=(uint16_t)(limit-start);
 437     converter=allocString(&stringBlock, line+start, length);
 438
 439     /* add the converter to the converter table */
 440     cnv=addConverter(converter);
 441
 442     /* The name itself may be tagged, so let's added it to the aliases list properly */
 443     pos = start;
 444
 445     /* get all the real aliases */
 446     for(;;) {
 447
 448         /* skip white space */
 449         while(line[pos]!=0 && isspace((int)line[pos])) {
 450             ++pos;
 451         }
 452
 453         /* is there no more alias name on this line? */
 454         if(line[pos]==0) {
 455             break;
 456         }
 457
 458         /* get an alias name */
 459         start=pos;
 460         while(line[pos]!=0 && line[pos]!='{' && !isspace((int)line[pos])) {
 461             ++pos;
 462         }
 463         limit=pos;
 464
 465         /* store the alias name */
 466         length=(uint16_t)(limit-start);
 467         if (start == 0) {
 468             /* add the converter as its own alias to the alias table */
 469             alias = converter;
 470             addAlias(alias, ALL_TAG_NUM, cnv, TRUE);
 471         }
 472         else {
 473             alias=allocString(&stringBlock, line+start, length);
 474             addAlias(alias, ALL_TAG_NUM, cnv, FALSE);
 475         }
 476         addToKnownAliases(alias);
 477
 478         /* add the alias/converter pair to the alias table */
 479         /* addAlias(alias, 0, cnv, FALSE);*/
 480
 481         /* skip whitespace */
 482         while (line[pos] && isspace((int)line[pos])) {
 483             ++pos;
 484         }
 485
 486         /* handle tags if they are present */
 487         if (line[pos] == '{') {
 488             ++pos;
 489             do {
 490                 start = pos;
 491                 while (line[pos] && line[pos] != '}' && !isspace((int)line[pos])) {
 492                     ++pos;
 493                 }
 494                 limit = pos;
 495
 496                 if (start != limit) {
 497                     /* add the tag to the tag table */
 498                     uint16_t tag = getTagNumber(line + start, (uint16_t)(limit - start));
 499                     addAlias(alias, tag, cnv, (UBool)(line[limit-1] == '*'));
 500                 }
 501
 502                 while (line[pos] && isspace((int)line[pos])) {
 503                     ++pos;
 504                 }
 505             } while (line[pos] && line[pos] != '}');
 506
 507             if (line[pos] == '}') {
 508                 ++pos;
 509             } else {
 510                 fprintf(stderr, "%s:%d: Unterminated tag list\n", path, lineNum);
 511                 exit(U_UNMATCHED_BRACES);
 512             }
 513         } else {
 514             addAlias(alias, EMPTY_TAG_NUM, cnv, (UBool)(tags[0].aliasList[cnv].aliasCount == 0));
 515         }
 516     }
 517 }
 518
 519 static uint16_t
 520 getTagNumber(const char *tag, uint16_t tagLen) {
 521     char *atag;
 522     uint16_t t;
 523     UBool preferredName = ((tagLen > 0) ? (tag[tagLen - 1] == '*') : (FALSE));
 524
 525     if (tagCount >= MAX_TAG_COUNT) {
 526         fprintf(stderr, "%s:%d: too many tags\n", path, lineNum);
 527         exit(U_BUFFER_OVERFLOW_ERROR);
 528     }
 529
 530     if (preferredName) {
 531 /*        puts(tag);*/
 532         tagLen--;
 533     }
 534
 535     for (t = 0; t < tagCount; ++t) {
 536         const char *currTag = GET_TAG_STR(tags[t].tag);
 537         if (uprv_strlen(currTag) == tagLen && !uprv_strnicmp(currTag, tag, tagLen)) {
 538             return t;
 539         }
 540     }
 541
 542     /* we need to add this tag */
 543     if (tagCount >= MAX_TAG_COUNT) {
 544         fprintf(stderr, "%s:%d: error: too many tags\n", path, lineNum);
 545         exit(U_BUFFER_OVERFLOW_ERROR);
 546     }
 547
 548     /* allocate a new entry in the tag table */
 549     atag = allocString(&tagBlock, tag, tagLen);
 550
 551     if (standardTagsUsed) {
 552         fprintf(stderr, "%s:%d: error: Tag \"%s\" is not declared at the beginning of the alias table.\n",
 553             path, lineNum, atag);
 554         exit(1);
 555     }
 556     else if (tagLen > 0 && strcmp(tag, ALL_TAG_STR) != 0) {
 557         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",
 558             path, lineNum, atag);
 559     }
 560
 561     /* add the tag to the tag table */
 562     tags[tagCount].tag = GET_TAG_NUM(atag);
 563     /* The aliasList should be set to 0's already */
 564
 565     return tagCount++;
 566 }
 567
 568 /*static void
 569 addTaggedAlias(uint16_t tag, const char *alias, uint16_t converter) {
 570     tags[tag].aliases[converter] = alias;
 571 }
 572 */
 573
 574 static void
 575 addOfficialTaggedStandards(char *line, int32_t lineLen) {
 576     char *atag;
 577     char *endTagExp;
 578     char *tag;
 579     static const char WHITESPACE[] = " \t";
 580
 581     if (tagCount > UCNV_NUM_RESERVED_TAGS) {
 582         fprintf(stderr, "%s:%d: error: official tags already added\n", path, lineNum);
 583         exit(U_BUFFER_OVERFLOW_ERROR);
 584     }
 585     tag = strchr(line, '{');
 586     if (tag == NULL) {
 587         /* Why were we called? */
 588         fprintf(stderr, "%s:%d: error: Missing start of tag group\n", path, lineNum);
 589         exit(U_PARSE_ERROR);
 590     }
 591     tag++;
 592     endTagExp = strchr(tag, '}');
 593     if (endTagExp == NULL) {
 594         fprintf(stderr, "%s:%d: error: Missing end of tag group\n", path, lineNum);
 595         exit(U_PARSE_ERROR);
 596     }
 597     endTagExp[0] = 0;
 598
 599     tag = strtok(tag, WHITESPACE);
 600     while (tag != NULL) {
 601 /*        printf("Adding original tag \"%s\"\n", tag);*/
 602
 603         /* allocate a new entry in the tag table */
 604         atag = allocString(&tagBlock, tag, -1);
 605
 606         /* add the tag to the tag table */
 607         tags[tagCount++].tag = (uint16_t)((atag - tagStore) >> 1);
 608
 609         /* The aliasList should already be set to 0's */
 610
 611         /* Get next tag */
 612         tag = strtok(NULL, WHITESPACE);
 613     }
 614 }
 615
 616 static uint16_t
 617 addToKnownAliases(const char *alias) {
 618 /*    uint32_t idx; */
 619     /* strict matching */
 620 /*    for (idx = 0; idx < knownAliasesCount; idx++) {
 621         uint16_t num = GET_ALIAS_NUM(alias);
 622         if (knownAliases[idx] != num
 623             && uprv_strcmp(alias, GET_ALIAS_STR(knownAliases[idx])) == 0)
 624         {
 625             fprintf(stderr, "%s:%d: warning: duplicate alias %s and %s found\n", path,
 626                 lineNum, alias, GET_ALIAS_STR(knownAliases[idx]));
 627             duplicateKnownAliasesCount++;
 628             break;
 629         }
 630         else if (knownAliases[idx] != num
 631             && ucnv_compareNames(alias, GET_ALIAS_STR(knownAliases[idx])) == 0)
 632         {
 633             if (verbose) {
 634                 fprintf(stderr, "%s:%d: information: duplicate alias %s and %s found\n", path,
 635                     lineNum, alias, GET_ALIAS_STR(knownAliases[idx]));
 636             }
 637             duplicateKnownAliasesCount++;
 638             break;
 639         }
 640     }
 641 */
 642     if (knownAliasesCount >= MAX_ALIAS_COUNT) {
 643         fprintf(stderr, "%s:%d: warning: Too many aliases defined for all converters\n",
 644             path, lineNum);
 645         exit(U_BUFFER_OVERFLOW_ERROR);
 646     }
 647     /* TODO: We could try to unlist exact duplicates. */
 648     return knownAliases[knownAliasesCount++] = GET_ALIAS_NUM(alias);
 649 }
 650
 651 /*
 652 @param standard When standard is 0, then it's the "empty" tag.
 653 */
 654 static uint16_t
 655 addAlias(const char *alias, uint16_t standard, uint16_t converter, UBool defaultName) {
 656     uint32_t idx, idx2;
 657     UBool startEmptyWithoutDefault = FALSE;
 658     AliasList *aliasList;
 659
 660     if(standard>=MAX_TAG_COUNT) {
 661         fprintf(stderr, "%s:%d: error: too many standard tags\n", path, lineNum);
 662         exit(U_BUFFER_OVERFLOW_ERROR);
 663     }
 664     if(converter>=MAX_CONV_COUNT) {
 665         fprintf(stderr, "%s:%d: error: too many converter names\n", path, lineNum);
 666         exit(U_BUFFER_OVERFLOW_ERROR);
 667     }
 668     aliasList = &tags[standard].aliasList[converter];
 669
 670     if (strchr(alias, '}')) {
 671         fprintf(stderr, "%s:%d: error: unmatched } found\n", path,
 672             lineNum);
 673     }
 674
 675     if(aliasList->aliasCount + 1 >= MAX_TC_ALIAS_COUNT) {
 676         fprintf(stderr, "%s:%d: error: too many aliases for alias %s and converter %s\n", path,
 677             lineNum, alias, GET_ALIAS_STR(converters[converter].converter));
 678         exit(U_BUFFER_OVERFLOW_ERROR);
 679     }
 680
 681     /* Show this warning only once. All aliases are added to the "ALL" tag. */
 682     if (standard == ALL_TAG_NUM && GET_ALIAS_STR(converters[converter].converter) != alias) {
 683         /* Normally these option values are parsed at runtime, and they can
 684            be discarded when the alias is a default converter. Options should
 685            only be on a converter and not an alias. */
 686         if (uprv_strchr(alias, UCNV_OPTION_SEP_CHAR) != 0)
 687         {
 688             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",
 689                 lineNum, alias);
 690         }
 691         if (uprv_strchr(alias, UCNV_VALUE_SEP_CHAR) != 0)
 692         {
 693             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",
 694                 lineNum, alias);
 695         }
 696     }
 697
 698     if (standard != ALL_TAG_NUM) {
 699         /* Check for duplicate aliases for this tag on all converters */
 700         for (idx = 0; idx < converterCount; idx++) {
 701             for (idx2 = 0; idx2 < tags[standard].aliasList[idx].aliasCount; idx2++) {
 702                 uint16_t aliasNum = tags[standard].aliasList[idx].aliases[idx2];
 703                 if (aliasNum
 704                     && ucnv_compareNames(alias, GET_ALIAS_STR(aliasNum)) == 0)
 705                 {
 706                     if (idx == converter) {
 707                         /*
 708                          * (alias, standard) duplicates are harmless if they map to the same converter.
 709                          * Only print a warning in verbose mode, or if the alias is a precise duplicate,
 710                          * not just a lenient-match duplicate.
 711                          */
 712                         if (verbose || 0 == uprv_strcmp(alias, GET_ALIAS_STR(aliasNum))) {
 713                             fprintf(stderr, "%s:%d: warning: duplicate aliases %s and %s found for standard %s and converter %s\n", path,
 714                                 lineNum, alias, GET_ALIAS_STR(aliasNum),
 715                                 GET_TAG_STR(tags[standard].tag),
 716                                 GET_ALIAS_STR(converters[converter].converter));
 717                         }
 718                     } else {
 719                         fprintf(stderr, "%s:%d: warning: duplicate aliases %s and %s found for standard tag %s between converter %s and converter %s\n", path,
 720                             lineNum, alias, GET_ALIAS_STR(aliasNum),
 721                             GET_TAG_STR(tags[standard].tag),
 722                             GET_ALIAS_STR(converters[converter].converter),
 723                             GET_ALIAS_STR(converters[idx].converter));
 724                     }
 725                     break;
 726                 }
 727             }
 728         }
 729
 730         /* Check for duplicate default aliases for this converter on all tags */
 731         /* It's okay to have multiple standards prefer the same name */
 732 /*        if (verbose && !dupFound) {
 733             for (idx = 0; idx < tagCount; idx++) {
 734                 if (tags[idx].aliasList[converter].aliases) {
 735                     uint16_t aliasNum = tags[idx].aliasList[converter].aliases[0];
 736                     if (aliasNum
 737                         && ucnv_compareNames(alias, GET_ALIAS_STR(aliasNum)) == 0)
 738                     {
 739                         fprintf(stderr, "%s:%d: warning: duplicate alias %s found for converter %s and standard tag %s\n", path,
 740                             lineNum, alias, GET_ALIAS_STR(converters[converter].converter), GET_TAG_STR(tags[standard].tag));
 741                         break;
 742                     }
 743                 }
 744             }
 745         }*/
 746     }
 747
 748     if (aliasList->aliasCount <= 0) {
 749         aliasList->aliasCount++;
 750         startEmptyWithoutDefault = TRUE;
 751     }
 752     aliasList->aliases = (uint16_t *)uprv_realloc(aliasList->aliases, (aliasList->aliasCount + 1) * sizeof(aliasList->aliases[0]));
 753     if (startEmptyWithoutDefault) {
 754         aliasList->aliases[0] = 0;
 755     }
 756     if (defaultName) {
 757         if (aliasList->aliases[0] != 0) {
 758             fprintf(stderr, "%s:%d: error: Alias %s and %s cannot both be the default alias for standard tag %s and converter %s\n", path,
 759                 lineNum,
 760                 alias,
 761                 GET_ALIAS_STR(aliasList->aliases[0]),
 762                 GET_TAG_STR(tags[standard].tag),
 763                 GET_ALIAS_STR(converters[converter].converter));
 764             exit(U_PARSE_ERROR);
 765         }
 766         aliasList->aliases[0] = GET_ALIAS_NUM(alias);
 767     } else {
 768         aliasList->aliases[aliasList->aliasCount++] = GET_ALIAS_NUM(alias);
 769     }
 770 /*    aliasList->converter = converter;*/
 771
 772     converters[converter].totalAliasCount++; /* One more to the column */
 773     tags[standard].totalAliasCount++; /* One more to the row */
 774
 775     return aliasList->aliasCount;
 776 }
 777
 778 static uint16_t
 779 addConverter(const char *converter) {
 780     uint32_t idx;
 781     if(converterCount>=MAX_CONV_COUNT) {
 782         fprintf(stderr, "%s:%d: error: too many converters\n", path, lineNum);
 783         exit(U_BUFFER_OVERFLOW_ERROR);
 784     }
 785
 786     for (idx = 0; idx < converterCount; idx++) {
 787         if (ucnv_compareNames(converter, GET_ALIAS_STR(converters[idx].converter)) == 0) {
 788             fprintf(stderr, "%s:%d: error: duplicate converter %s found!\n", path, lineNum, converter);
 789             exit(U_PARSE_ERROR);
 790             break;
 791         }
 792     }
 793
 794     converters[converterCount].converter = GET_ALIAS_NUM(converter);
 795     converters[converterCount].totalAliasCount = 0;
 796
 797     return converterCount++;
 798 }
 799
 800 /* resolve this alias based on the prioritization of the standard tags. */
 801 static void
 802 resolveAliasToConverter(uint16_t alias, uint16_t *tagNum, uint16_t *converterNum) {
 803     uint16_t idx, idx2, idx3;
 804
 805     for (idx = UCNV_NUM_RESERVED_TAGS; idx < tagCount; idx++) {
 806         for (idx2 = 0; idx2 < converterCount; idx2++) {
 807             for (idx3 = 0; idx3 < tags[idx].aliasList[idx2].aliasCount; idx3++) {
 808                 uint16_t aliasNum = tags[idx].aliasList[idx2].aliases[idx3];
 809                 if (aliasNum == alias) {
 810                     *tagNum = idx;
 811                     *converterNum = idx2;
 812                     return;
 813                 }
 814             }
 815         }
 816     }
 817     /* Do the leftovers last, just in case */
 818     /* There is no need to do the ALL tag */
 819     idx = 0;
 820     for (idx2 = 0; idx2 < converterCount; idx2++) {
 821         for (idx3 = 0; idx3 < tags[idx].aliasList[idx2].aliasCount; idx3++) {
 822             uint16_t aliasNum = tags[idx].aliasList[idx2].aliases[idx3];
 823             if (aliasNum == alias) {
 824                 *tagNum = idx;
 825                 *converterNum = idx2;
 826                 return;
 827             }
 828         }
 829     }
 830     *tagNum = UINT16_MAX;
 831     *converterNum = UINT16_MAX;
 832     fprintf(stderr, "%s: warning: alias %s not found\n",
 833         path,
 834         GET_ALIAS_STR(alias));
 835     return;
 836 }
 837
 838 /* The knownAliases should be sorted before calling this function */
 839 static uint32_t
 840 resolveAliases(uint16_t *uniqueAliasArr, uint16_t *uniqueAliasToConverterArr, uint16_t aliasOffset) {
 841     uint32_t uniqueAliasIdx = 0;
 842     uint32_t idx;
 843     uint16_t currTagNum, oldTagNum;
 844     uint16_t currConvNum, oldConvNum;
 845     const char *lastName;
 846
 847     resolveAliasToConverter(knownAliases[0], &oldTagNum, &currConvNum);
 848     uniqueAliasToConverterArr[uniqueAliasIdx] = currConvNum;
 849     oldConvNum = currConvNum;
 850     uniqueAliasArr[uniqueAliasIdx] = knownAliases[0] + aliasOffset;
 851     uniqueAliasIdx++;
 852     lastName = GET_ALIAS_STR(knownAliases[0]);
 853
 854     for (idx = 1; idx < knownAliasesCount; idx++) {
 855         resolveAliasToConverter(knownAliases[idx], &currTagNum, &currConvNum);
 856         if (ucnv_compareNames(lastName, GET_ALIAS_STR(knownAliases[idx])) == 0) {
 857             /* duplicate found */
 858             if ((currTagNum < oldTagNum && currTagNum >= UCNV_NUM_RESERVED_TAGS)
 859                 || oldTagNum == 0) {
 860                 oldTagNum = currTagNum;
 861                 uniqueAliasToConverterArr[uniqueAliasIdx - 1] = currConvNum;
 862                 uniqueAliasArr[uniqueAliasIdx - 1] = knownAliases[idx] + aliasOffset;
 863                 if (verbose) {
 864                     printf("using %s instead of %s -> %s",
 865                         GET_ALIAS_STR(knownAliases[idx]),
 866                         lastName,
 867                         GET_ALIAS_STR(converters[currConvNum].converter));
 868                     if (oldConvNum != currConvNum) {
 869                         printf(" (alias conflict)");
 870                     }
 871                     puts("");
 872                 }
 873             }
 874             else {
 875                 /* else ignore it */
 876                 if (verbose) {
 877                     printf("folding %s into %s -> %s",
 878                         GET_ALIAS_STR(knownAliases[idx]),
 879                         lastName,
 880                         GET_ALIAS_STR(converters[oldConvNum].converter));
 881                     if (oldConvNum != currConvNum) {
 882                         printf(" (alias conflict)");
 883                     }
 884                     puts("");
 885                 }
 886             }
 887             if (oldConvNum != currConvNum) {
 888                 uniqueAliasToConverterArr[uniqueAliasIdx - 1] |= UCNV_AMBIGUOUS_ALIAS_MAP_BIT;
 889             }
 890         }
 891         else {
 892             uniqueAliasToConverterArr[uniqueAliasIdx] = currConvNum;
 893             oldConvNum = currConvNum;
 894             uniqueAliasArr[uniqueAliasIdx] = knownAliases[idx] + aliasOffset;
 895             uniqueAliasIdx++;
 896             lastName = GET_ALIAS_STR(knownAliases[idx]);
 897             oldTagNum = currTagNum;
 898             /*printf("%s -> %s\n", GET_ALIAS_STR(knownAliases[idx]), GET_ALIAS_STR(converters[currConvNum].converter));*/
 899         }
 900         if (uprv_strchr(GET_ALIAS_STR(converters[currConvNum].converter), UCNV_OPTION_SEP_CHAR) != NULL) {
 901             uniqueAliasToConverterArr[uniqueAliasIdx-1] |= UCNV_CONTAINS_OPTION_BIT;
 902         }
 903     }
 904     return uniqueAliasIdx;
 905 }
 906
 907 static void
 908 createOneAliasList(uint16_t *aliasArrLists, uint32_t tag, uint32_t converter, uint16_t offset) {
 909     uint32_t aliasNum;
 910     AliasList *aliasList = &tags[tag].aliasList[converter];
 911
 912     if (aliasList->aliasCount == 0) {
 913         aliasArrLists[tag*converterCount + converter] = 0;
 914     }
 915     else {
 916         aliasLists[aliasListsSize++] = aliasList->aliasCount;
 917
 918         /* write into the array area a 1's based index. */
 919         aliasArrLists[tag*converterCount + converter] = aliasListsSize;
 920
 921 /*        printf("tag %s converter %s\n",
 922             GET_TAG_STR(tags[tag].tag),
 923             GET_ALIAS_STR(converters[converter].converter));*/
 924         for (aliasNum = 0; aliasNum < aliasList->aliasCount; aliasNum++) {
 925             uint16_t value;
 926 /*            printf("   %s\n",
 927                 GET_ALIAS_STR(aliasList->aliases[aliasNum]));*/
 928             if (aliasList->aliases[aliasNum]) {
 929                 value = aliasList->aliases[aliasNum] + offset;
 930             } else {
 931                 value = 0;
 932                 if (tag != 0) { /* Only show the warning when it's not the leftover tag. */
 933                     fprintf(stderr, "%s: warning: tag %s does not have a default alias for %s\n",
 934                             path,
 935                             GET_TAG_STR(tags[tag].tag),
 936                             GET_ALIAS_STR(converters[converter].converter));
 937                 }
 938             }
 939             aliasLists[aliasListsSize++] = value;
 940             if (aliasListsSize >= MAX_LIST_SIZE) {
 941                 fprintf(stderr, "%s: error: Too many alias lists\n", path);
 942                 exit(U_BUFFER_OVERFLOW_ERROR);
 943             }
 944
 945         }
 946     }
 947 }
 948
 949 static void
 950 createNormalizedAliasStrings(char *normalizedStrings, const char *origStringBlock, int32_t stringBlockLength) {
 951     int32_t currStrLen;
 952     uprv_memcpy(normalizedStrings, origStringBlock, stringBlockLength);
 953     while ((currStrLen = (int32_t)uprv_strlen(origStringBlock)) < stringBlockLength) {
 954         int32_t currStrSize = currStrLen + 1;
 955         if (currStrLen > 0) {
 956             int32_t normStrLen;
 957             ucnv_io_stripForCompare(normalizedStrings, origStringBlock);
 958             normStrLen = uprv_strlen(normalizedStrings);
 959             if (normStrLen > 0) {
 960                 uprv_memset(normalizedStrings + normStrLen, 0, currStrSize - normStrLen);
 961             }
 962         }
 963         stringBlockLength -= currStrSize;
 964         normalizedStrings += currStrSize;
 965         origStringBlock += currStrSize;
 966     }
 967 }
 968
 969 static void
 970 writeAliasTable(UNewDataMemory *out) {
 971     uint32_t i, j;
 972     uint32_t uniqueAliasesSize;
 973     uint16_t aliasOffset = (uint16_t)(tagBlock.top/sizeof(uint16_t));
 974     uint16_t *aliasArrLists = (uint16_t *)uprv_malloc(tagCount * converterCount * sizeof(uint16_t));
 975     uint16_t *uniqueAliases = (uint16_t *)uprv_malloc(knownAliasesCount * sizeof(uint16_t));
 976     uint16_t *uniqueAliasesToConverter = (uint16_t *)uprv_malloc(knownAliasesCount * sizeof(uint16_t));
 977
 978     qsort(knownAliases, knownAliasesCount, sizeof(knownAliases[0]), compareAliases);
 979     uniqueAliasesSize = resolveAliases(uniqueAliases, uniqueAliasesToConverter, aliasOffset);
 980
 981     /* Array index starts at 1. aliasLists[0] is the size of the lists section. */
 982     aliasListsSize = 0;
 983
 984     /* write the offsets of all the aliases lists in a 2D array, and create the lists. */
 985     for (i = 0; i < tagCount; ++i) {
 986         for (j = 0; j < converterCount; ++j) {
 987             createOneAliasList(aliasArrLists, i, j, aliasOffset);
 988         }
 989     }
 990
 991     /* Write the size of the TOC */
 992     if (tableOptions.stringNormalizationType == UCNV_IO_UNNORMALIZED) {
 993         udata_write32(out, 8);
 994     }
 995     else {
 996         udata_write32(out, 9);
 997     }
 998
 999     /* Write the sizes of each section */
1000     /* All sizes are the number of uint16_t units, not bytes */
1001     udata_write32(out, converterCount);
1002     udata_write32(out, tagCount);
1003     udata_write32(out, uniqueAliasesSize);  /* list of aliases */
1004     udata_write32(out, uniqueAliasesSize);  /* The preresolved form of mapping an untagged the alias to a converter */
1005     udata_write32(out, tagCount * converterCount);
1006     udata_write32(out, aliasListsSize + 1);
1007     udata_write32(out, sizeof(tableOptions) / sizeof(uint16_t));
1008     udata_write32(out, (tagBlock.top + stringBlock.top) / sizeof(uint16_t));
1009     if (tableOptions.stringNormalizationType != UCNV_IO_UNNORMALIZED) {
1010         udata_write32(out, (tagBlock.top + stringBlock.top) / sizeof(uint16_t));
1011     }
1012
1013     /* write the table of converters */
1014     /* Think of this as the column headers */
1015     for(i=0; i<converterCount; ++i) {
1016         udata_write16(out, (uint16_t)(converters[i].converter + aliasOffset));
1017     }
1018
1019     /* write the table of tags */
1020     /* Think of this as the row headers */
1021     for(i=UCNV_NUM_RESERVED_TAGS; i<tagCount; ++i) {
1022         udata_write16(out, tags[i].tag);
1023     }
1024     /* The empty tag is considered the leftover list, and put that at the end of the priority list. */
1025     udata_write16(out, tags[EMPTY_TAG_NUM].tag);
1026     udata_write16(out, tags[ALL_TAG_NUM].tag);
1027
1028     /* Write the unique list of aliases */
1029     udata_writeBlock(out, uniqueAliases, uniqueAliasesSize * sizeof(uint16_t));
1030
1031     /* Write the unique list of aliases */
1032     udata_writeBlock(out, uniqueAliasesToConverter, uniqueAliasesSize * sizeof(uint16_t));
1033
1034     /* Write the array to the lists */
1035     udata_writeBlock(out, (const void *)(aliasArrLists + (2*converterCount)), (((tagCount - 2) * converterCount) * sizeof(uint16_t)));
1036     /* Now write the leftover part of the array for the EMPTY and ALL lists */
1037     udata_writeBlock(out, (const void *)aliasArrLists, (2 * converterCount * sizeof(uint16_t)));
1038
1039     /* Offset the next array to make the index start at 1. */
1040     udata_write16(out, 0xDEAD);
1041
1042     /* Write the lists */
1043     udata_writeBlock(out, (const void *)aliasLists, aliasListsSize * sizeof(uint16_t));
1044
1045     /* Write any options for the alias table. */
1046     udata_writeBlock(out, (const void *)&tableOptions, sizeof(tableOptions));
1047
1048     /* write the tags strings */
1049     udata_writeString(out, tagBlock.store, tagBlock.top);
1050
1051     /* write the aliases strings */
1052     udata_writeString(out, stringBlock.store, stringBlock.top);
1053
1054     /* write the normalized aliases strings */
1055     if (tableOptions.stringNormalizationType != UCNV_IO_UNNORMALIZED) {
1056         char *normalizedStrings = (char *)uprv_malloc(tagBlock.top + stringBlock.top);
1057         createNormalizedAliasStrings(normalizedStrings, tagBlock.store, tagBlock.top);
1058         createNormalizedAliasStrings(normalizedStrings + tagBlock.top, stringBlock.store, stringBlock.top);
1059
1060         /* Write out the complete normalized array. */
1061         udata_writeString(out, normalizedStrings, tagBlock.top + stringBlock.top);
1062         uprv_free(normalizedStrings);
1063     }
1064
1065     uprv_free(uniqueAliasesToConverter);
1066     uprv_free(uniqueAliases);
1067     uprv_free(aliasArrLists);
1068 }
1069
1070 static char *
1071 allocString(StringBlock *block, const char *s, int32_t length) {
1072     uint32_t top;
1073     char *p;
1074
1075     if(length<0) {
1076         length=(int32_t)uprv_strlen(s);
1077     }
1078
1079     /*
1080      * add 1 for the terminating NUL
1081      * and round up (+1 &~1)
1082      * to keep the addresses on a 16-bit boundary
1083      */
1084     top=block->top + (uint32_t)((length + 1 + 1) & ~1);
1085
1086     if(top >= block->max) {
1087         fprintf(stderr, "%s:%d: error: out of memory\n", path, lineNum);
1088         exit(U_MEMORY_ALLOCATION_ERROR);
1089     }
1090
1091     /* get the pointer and copy the string */
1092     p = block->store + block->top;
1093     uprv_memcpy(p, s, length);
1094     p[length] = 0; /* NUL-terminate it */
1095     if((length & 1) == 0) {
1096         p[length + 1] = 0; /* set the padding byte */
1097     }
1098
1099     /* check for invariant characters now that we have a NUL-terminated string for easy output */
1100     if(!uprv_isInvariantString(p, length)) {
1101         fprintf(stderr, "%s:%d: error: the name %s contains not just invariant characters\n", path, lineNum, p);
1102         exit(U_INVALID_TABLE_FORMAT);
1103     }
1104
1105     block->top = top;
1106     return p;
1107 }
1108
1109 static int
1110 compareAliases(const void *alias1, const void *alias2) {
1111     /* Names like IBM850 and ibm-850 need to be sorted together */
1112     int result = ucnv_compareNames(GET_ALIAS_STR(*(uint16_t*)alias1), GET_ALIAS_STR(*(uint16_t*)alias2));
1113     if (!result) {
1114         /* Sort the shortest first */
1115         return (int)uprv_strlen(GET_ALIAS_STR(*(uint16_t*)alias1)) - (int)uprv_strlen(GET_ALIAS_STR(*(uint16_t*)alias2));
1116     }
1117     return result;
1118 }
1119
1120 /*
1121  * Hey, Emacs, please set the following:
1122  *
1123  * Local Variables:
1124  * indent-tabs-mode: nil
1125  * End:
1126  *
1127  */
1128