icuSources/tools/genctd/genctd.cpp

   1 /*
   2 **********************************************************************
   3 *   Copyright (C) 2002-2006, International Business Machines
   4 *   Corporation and others.  All Rights Reserved.
   5 **********************************************************************
   6 *
   7 * File genctd.c
   8 */
   9
  10 //--------------------------------------------------------------------
  11 //
  12 //   Tool for generating CompactTrieDictionary data files (.ctd files).
  13 //
  14 //   Usage:  genctd [options] -o output-file.ctd input-file
  15 //
  16 //       options:   -v         verbose
  17 //                  -? or -h   help
  18 //
  19 //   The input  file is a plain text file containing words, one per line.
  20 //    Words end at the first whitespace; lines beginning with whitespace
  21 //    are ignored.
  22 //    The file can be encoded as utf-8, or utf-16 (either endian), or
  23 //    in the default code page (platform dependent.).  utf encoded
  24 //    files must include a BOM.
  25 //
  26 //--------------------------------------------------------------------
  27
  28 #include "unicode/utypes.h"
  29 #include "unicode/uchar.h"
  30 #include "unicode/ucnv.h"
  31 #include "unicode/uniset.h"
  32 #include "unicode/unistr.h"
  33 #include "unicode/uclean.h"
  34 #include "unicode/udata.h"
  35 #include "unicode/putil.h"
  36
  37 #include "uoptions.h"
  38 #include "unewdata.h"
  39 #include "ucmndata.h"
  40 #include "rbbidata.h"
  41 #include "triedict.h"
  42 #include "cmemory.h"
  43
  44 #include <stdio.h>
  45 #include <stdlib.h>
  46 #include <string.h>
  47
  48 U_NAMESPACE_USE
  49
  50 static char *progName;
  51 static UOption options[]={
  52     UOPTION_HELP_H,             /* 0 */
  53     UOPTION_HELP_QUESTION_MARK, /* 1 */
  54     UOPTION_VERBOSE,            /* 2 */
  55     { "out",   NULL, NULL, NULL, 'o', UOPT_REQUIRES_ARG, 0 },   /* 3 */
  56     UOPTION_ICUDATADIR,         /* 4 */
  57     UOPTION_DESTDIR,            /* 5 */
  58     UOPTION_COPYRIGHT,          /* 6 */
  59 };
  60
  61 void usageAndDie(int retCode) {
  62         printf("Usage: %s [-v] [-options] -o output-file dictionary-file\n", progName);
  63         printf("\tRead in word list and write out compact trie dictionary\n"
  64             "options:\n"
  65             "\t-h or -? or --help  this usage text\n"
  66             "\t-V or --version     show a version message\n"
  67             "\t-c or --copyright   include a copyright notice\n"
  68             "\t-v or --verbose     turn on verbose output\n"
  69             "\t-i or --icudatadir  directory for locating any needed intermediate data files,\n"
  70             "\t                    followed by path, defaults to %s\n"
  71             "\t-d or --destdir     destination directory, followed by the path\n",
  72             u_getDataDirectory());
  73         exit (retCode);
  74 }
  75
  76
  77 #if UCONFIG_NO_BREAK_ITERATION
  78
  79 /* dummy UDataInfo cf. udata.h */
  80 static UDataInfo dummyDataInfo = {
  81     sizeof(UDataInfo),
  82     0,
  83
  84     U_IS_BIG_ENDIAN,
  85     U_CHARSET_FAMILY,
  86     U_SIZEOF_UCHAR,
  87     0,
  88
  89     { 0, 0, 0, 0 },                 /* dummy dataFormat */
  90     { 0, 0, 0, 0 },                 /* dummy formatVersion */
  91     { 0, 0, 0, 0 }                  /* dummy dataVersion */
  92 };
  93
  94 #else
  95
  96 //
  97 //  Set up the ICU data header, defined in ucmndata.h
  98 //
  99 DataHeader dh ={
 100     {sizeof(DataHeader),           // Struct MappedData
 101         0xda,
 102         0x27},
 103
 104     {                               // struct UDataInfo
 105         sizeof(UDataInfo),          //     size
 106         0,                          //     reserved
 107         U_IS_BIG_ENDIAN,
 108         U_CHARSET_FAMILY,
 109         U_SIZEOF_UCHAR,
 110         0,                          //     reserved
 111
 112     { 0x54, 0x72, 0x44, 0x63 },     // "TrDc" Trie Dictionary
 113     { 1, 0, 0, 0 },                 // 1.0.0.0
 114     { 0, 0, 0, 0 },                 // Irrelevant for this data type
 115     }};
 116
 117 #endif
 118
 119 //----------------------------------------------------------------------------
 120 //
 121 //  main      for genctd
 122 //
 123 //----------------------------------------------------------------------------
 124 int  main(int argc, char **argv) {
 125     UErrorCode  status = U_ZERO_ERROR;
 126     const char *wordFileName;
 127     const char *outFileName;
 128     const char *outDir = NULL;
 129     const char *copyright = NULL;
 130
 131     //
 132     // Pick up and check the command line arguments,
 133     //    using the standard ICU tool utils option handling.
 134     //
 135     U_MAIN_INIT_ARGS(argc, argv);
 136     progName = argv[0];
 137     argc=u_parseArgs(argc, argv, sizeof(options)/sizeof(options[0]), options);
 138     if(argc<0) {
 139         // Unrecognized option
 140         fprintf(stderr, "error in command line argument \"%s\"\n", argv[-argc]);
 141         usageAndDie(U_ILLEGAL_ARGUMENT_ERROR);
 142     }
 143
 144     if(options[0].doesOccur || options[1].doesOccur) {
 145         //  -? or -h for help.
 146         usageAndDie(0);
 147     }
 148
 149     if (!options[3].doesOccur || argc < 2) {
 150         fprintf(stderr, "input and output file must both be specified.\n");
 151         usageAndDie(U_ILLEGAL_ARGUMENT_ERROR);
 152     }
 153     outFileName  = options[3].value;
 154     wordFileName = argv[1];
 155
 156     if (options[4].doesOccur) {
 157         u_setDataDirectory(options[4].value);
 158     }
 159
 160     /* Initialize ICU */
 161     u_init(&status);
 162     if (U_FAILURE(status)) {
 163         fprintf(stderr, "%s: can not initialize ICU.  status = %s\n",
 164             argv[0], u_errorName(status));
 165         exit(1);
 166     }
 167     status = U_ZERO_ERROR;
 168
 169     /* Combine the directory with the file name */
 170     if(options[5].doesOccur) {
 171         outDir = options[5].value;
 172     }
 173     if (options[6].doesOccur) {
 174         copyright = U_COPYRIGHT_STRING;
 175     }
 176
 177 #if UCONFIG_NO_BREAK_ITERATION
 178
 179     UNewDataMemory *pData;
 180     char msg[1024];
 181
 182     /* write message with just the name */
 183     sprintf(msg, "genctd writes dummy %s because of UCONFIG_NO_BREAK_ITERATION, see uconfig.h", outFileName);
 184     fprintf(stderr, "%s\n", msg);
 185
 186     /* write the dummy data file */
 187     pData = udata_create(outDir, NULL, outFileName, &dummyDataInfo, NULL, &status);
 188     udata_writeBlock(pData, msg, strlen(msg));
 189     udata_finish(pData, &status);
 190     return (int)status;
 191
 192 #else
 193
 194     //
 195     //  Read in the dictionary source file
 196     //
 197     long        result;
 198     long        wordFileSize;
 199     FILE        *file;
 200     char        *wordBufferC;
 201
 202     file = fopen(wordFileName, "rb");
 203     if( file == 0 ) {
 204         fprintf(stderr, "Could not open file \"%s\"\n", wordFileName);
 205         exit(-1);
 206     }
 207     fseek(file, 0, SEEK_END);
 208     wordFileSize = ftell(file);
 209     fseek(file, 0, SEEK_SET);
 210     wordBufferC = new char[wordFileSize+10];
 211
 212     result = (long)fread(wordBufferC, 1, wordFileSize, file);
 213     if (result != wordFileSize)  {
 214         fprintf(stderr, "Error reading file \"%s\"\n", wordFileName);
 215         exit (-1);
 216     }
 217     wordBufferC[wordFileSize]=0;
 218     fclose(file);
 219
 220     //
 221     // Look for a Unicode Signature (BOM) on the word file
 222     //
 223     int32_t        signatureLength;
 224     const char *   wordSourceC = wordBufferC;
 225     const char*    encoding = ucnv_detectUnicodeSignature(
 226                            wordSourceC, wordFileSize, &signatureLength, &status);
 227     if (U_FAILURE(status)) {
 228         exit(status);
 229     }
 230     if(encoding!=NULL ){
 231         wordSourceC  += signatureLength;
 232         wordFileSize -= signatureLength;
 233     }
 234
 235     //
 236     // Open a converter to take the rule file to UTF-16
 237     //
 238     UConverter* conv;
 239     conv = ucnv_open(encoding, &status);
 240     if (U_FAILURE(status)) {
 241         fprintf(stderr, "ucnv_open: ICU Error \"%s\"\n", u_errorName(status));
 242         exit(status);
 243     }
 244
 245     //
 246     // Convert the words to UChar.
 247     //  Preflight first to determine required buffer size.
 248     //
 249     uint32_t destCap = ucnv_toUChars(conv,
 250                        NULL,           //  dest,
 251                        0,              //  destCapacity,
 252                        wordSourceC,
 253                        wordFileSize,
 254                        &status);
 255     if (status != U_BUFFER_OVERFLOW_ERROR) {
 256         fprintf(stderr, "ucnv_toUChars: ICU Error \"%s\"\n", u_errorName(status));
 257         exit(status);
 258     };
 259
 260     status = U_ZERO_ERROR;
 261     UChar *wordSourceU = new UChar[destCap+1];
 262     ucnv_toUChars(conv,
 263                   wordSourceU,     //  dest,
 264                   destCap+1,
 265                   wordSourceC,
 266                   wordFileSize,
 267                   &status);
 268     if (U_FAILURE(status)) {
 269         fprintf(stderr, "ucnv_toUChars: ICU Error \"%s\"\n", u_errorName(status));
 270         exit(status);
 271     };
 272     ucnv_close(conv);
 273
 274     // Get rid of the original file buffer
 275     delete[] wordBufferC;
 276
 277     // Create a MutableTrieDictionary, and loop through all the lines, inserting
 278     // words.
 279
 280     // First, pick a median character.
 281     UChar *current = wordSourceU + (destCap/2);
 282     UChar uc = *current++;
 283     UnicodeSet breaks;
 284     breaks.add(0x000A);     // Line Feed
 285     breaks.add(0x000D);     // Carriage Return
 286     breaks.add(0x2028);     // Line Separator
 287     breaks.add(0x2029);     // Paragraph Separator
 288
 289     do {
 290         // Look for line break
 291         while (uc && !breaks.contains(uc)) {
 292             uc = *current++;
 293         }
 294         // Now skip to first non-line-break
 295         while (uc && breaks.contains(uc)) {
 296             uc = *current++;
 297         }
 298     }
 299     while (uc && (breaks.contains(uc) || u_isspace(uc)));
 300
 301     MutableTrieDictionary *mtd = new MutableTrieDictionary(uc, status);
 302
 303     if (U_FAILURE(status)) {
 304         fprintf(stderr, "new MutableTrieDictionary: ICU Error \"%s\"\n", u_errorName(status));
 305         exit(status);
 306     }
 307
 308     // Now add the words. Words are non-space characters at the beginning of
 309     // lines, and must be at least one UChar.
 310     current = wordSourceU;
 311     UChar *candidate = current;
 312     uc = *current++;
 313     int32_t length = 0;
 314
 315     while (uc) {
 316         while (uc && !u_isspace(uc)) {
 317             ++length;
 318             uc = *current++;
 319         }
 320         if (length > 0) {
 321             mtd->addWord(candidate, length, status);
 322             if (U_FAILURE(status)) {
 323                 fprintf(stderr, "MutableTrieDictionary::addWord: ICU Error \"%s\"\n",
 324                         u_errorName(status));
 325                 exit(status);
 326             }
 327         }
 328         // Find beginning of next line
 329         while (uc && !breaks.contains(uc)) {
 330             uc = *current++;
 331         }
 332         while (uc && breaks.contains(uc)) {
 333             uc = *current++;
 334         }
 335         candidate = current-1;
 336         length = 0;
 337     }
 338
 339     // Get rid of the Unicode text buffer
 340     delete[] wordSourceU;
 341
 342     // Now, create a CompactTrieDictionary from the mutable dictionary
 343     CompactTrieDictionary *ctd = new CompactTrieDictionary(*mtd, status);
 344     if (U_FAILURE(status)) {
 345         fprintf(stderr, "new CompactTrieDictionary: ICU Error \"%s\"\n", u_errorName(status));
 346         exit(status);
 347     }
 348
 349     // Get rid of the MutableTrieDictionary
 350     delete mtd;
 351
 352     //
 353     //  Get the binary data from the dictionary.
 354     //
 355     uint32_t        outDataSize = ctd->dataSize();
 356     const uint8_t  *outData = (const uint8_t *)ctd->data();
 357
 358     //
 359     //  Create the output file
 360     //
 361     size_t bytesWritten;
 362     UNewDataMemory *pData;
 363     pData = udata_create(outDir, NULL, outFileName, &(dh.info), copyright, &status);
 364     if(U_FAILURE(status)) {
 365         fprintf(stderr, "genctd: Could not open output file \"%s\", \"%s\"\n",
 366                          outFileName, u_errorName(status));
 367         exit(status);
 368     }
 369
 370
 371     //  Write the data itself.
 372     udata_writeBlock(pData, outData, outDataSize);
 373     // finish up
 374     bytesWritten = udata_finish(pData, &status);
 375     if(U_FAILURE(status)) {
 376         fprintf(stderr, "genctd: error \"%s\" writing the output file\n", u_errorName(status));
 377         exit(status);
 378     }
 379
 380     if (bytesWritten != outDataSize) {
 381         fprintf(stderr, "Error writing to output file \"%s\"\n", outFileName);
 382         exit(-1);
 383     }
 384
 385     // Get rid of the CompactTrieDictionary
 386     delete ctd;
 387
 388     u_cleanup();
 389
 390     printf("genctd: tool completed successfully.\n");
 391     return 0;
 392
 393 #endif /* #if !UCONFIG_NO_BREAK_ITERATION */
 394 }
 395