icuSources/tools/gencfu/gencfu.cpp

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*
   4 **********************************************************************
   5 *   Copyright (C) 2009-2016, International Business Machines
   6 *   Corporation and others.  All Rights Reserved.
   7 **********************************************************************
   8 *
   9 * File gencfu.c
  10 */
  11
  12 //--------------------------------------------------------------------
  13 //
  14 //   Tool for generating Unicode Confusable data files (.cfu files).
  15 //   .cfu files contain the compiled of the confusable data
  16 //   derived from the Unicode Consortium data described in
  17 //   Unicode UAX 39.
  18 //
  19 //   Usage:  gencfu [options] -r confusables-file.txt -o output-file.cfu
  20 //
  21 //       options:   -v         verbose
  22 //                  -? or -h   help
  23 //
  24 //   The input rule filew is are plain text files containing confusable character
  25 //    definitions in the input format defined by Unicode UAX39 for the files
  26 //    confusables.txt.  This source (.txt) format
  27 //    is also accepted direaccepted by ICU spoof detedtors.  The
  28 //    files must be encoded in utf-8 format, with or without a BOM.
  29 //
  30 //   The script used to compile confusablesWholeScript.txt into the CFU file
  31 //    until the Unicode consortium deprecated it.
  32 //
  33 //--------------------------------------------------------------------
  34
  35 #include "unicode/utypes.h"
  36 #include "unicode/unistr.h"
  37 #include "unicode/uclean.h"
  38 #include "unicode/udata.h"
  39 #include "unicode/putil.h"
  40
  41 #include "uoptions.h"
  42 #include "unewdata.h"
  43 #include "ucmndata.h"
  44 #include "uspoof_impl.h"
  45 #include "cmemory.h"
  46
  47 #include <stdio.h>
  48 #include <stdlib.h>
  49 #include <string.h>
  50
  51 U_NAMESPACE_USE
  52
  53 static char *progName;
  54 static UOption options[]={
  55     UOPTION_HELP_H,             /* 0 */
  56     UOPTION_HELP_QUESTION_MARK, /* 1 */
  57     UOPTION_VERBOSE,            /* 2 */
  58     { "rules", NULL, NULL, NULL, 'r', UOPT_REQUIRES_ARG, 0 },   /* 3 */
  59     { "wsrules", NULL, NULL, NULL, 'w', UOPT_REQUIRES_ARG, 0},  /* 4 */  // deprecated
  60     { "out",   NULL, NULL, NULL, 'o', UOPT_REQUIRES_ARG, 0 },   /* 5 */
  61     UOPTION_ICUDATADIR,         /* 6 */
  62     UOPTION_DESTDIR,            /* 7 */
  63     UOPTION_COPYRIGHT,          /* 8 */
  64     UOPTION_QUIET,              /* 9 */
  65 };
  66
  67 void usageAndDie(int retCode) {
  68         printf("Usage: %s [-v] [-options] -r confusablesRules.txt -o output-file\n", progName);
  69         printf("\tRead in Unicode confusable character definitions and write out the binary data\n"
  70             "options:\n"
  71             "\t-h or -? or --help  this usage text\n"
  72             "\t-V or --version     show a version message\n"
  73             "\t-c or --copyright   include a copyright notice\n"
  74             "\t-v or --verbose     turn on verbose output\n"
  75             "\t-q or --quiet       do not display warnings and progress\n"
  76             "\t-i or --icudatadir  directory for locating any needed intermediate data files,\n"
  77             "\t                    followed by path, defaults to %s\n"
  78             "\t-d or --destdir     destination directory, followed by the path\n",
  79             u_getDataDirectory());
  80         exit (retCode);
  81 }
  82
  83
  84 #if UCONFIG_NO_REGULAR_EXPRESSIONS || UCONFIG_NO_NORMALIZATION || UCONFIG_NO_FILE_IO
  85
  86 /* dummy UDataInfo cf. udata.h */
  87 static UDataInfo dummyDataInfo = {
  88     sizeof(UDataInfo),
  89     0,
  90
  91     U_IS_BIG_ENDIAN,
  92     U_CHARSET_FAMILY,
  93     U_SIZEOF_UCHAR,
  94     0,
  95
  96     { 0, 0, 0, 0 },                 /* dummy dataFormat */
  97     { 0, 0, 0, 0 },                 /* dummy formatVersion */
  98     { 0, 0, 0, 0 }                  /* dummy dataVersion */
  99 };
 100
 101 #else
 102
 103 //
 104 //  Set up the ICU data header, defined in ucmndata.h
 105 //
 106 DataHeader dh ={
 107     {sizeof(DataHeader),           // Struct MappedData
 108         0xda,
 109         0x27},
 110
 111     {                               // struct UDataInfo
 112         sizeof(UDataInfo),          //     size
 113         0,                          //     reserved
 114         U_IS_BIG_ENDIAN,
 115         U_CHARSET_FAMILY,
 116         U_SIZEOF_UCHAR,
 117         0,                          //     reserved
 118
 119     { 0x43, 0x66, 0x75, 0x20 },     //     dataFormat="Cfu "
 120     { 0xff, 0, 0, 0 },              //     formatVersion.  Filled in later with values
 121                                     //      from the  builder.  The  values declared
 122                                     //      here should never appear in any real data.
 123         { 5, 1, 0, 0 }              //   dataVersion (Unicode version)
 124     }};
 125
 126 #endif
 127
 128 // Forward declaration for function for reading source files.
 129 static const char *readFile(const char *fileName, int32_t *len);
 130
 131 //----------------------------------------------------------------------------
 132 //
 133 //  main      for gencfu
 134 //
 135 //----------------------------------------------------------------------------
 136 int  main(int argc, char **argv) {
 137     UErrorCode  status = U_ZERO_ERROR;
 138     const char *confFileName;
 139     const char *outFileName;
 140     const char *outDir = NULL;
 141     const char *copyright = NULL;
 142
 143     //
 144     // Pick up and check the command line arguments,
 145     //    using the standard ICU tool utils option handling.
 146     //
 147     U_MAIN_INIT_ARGS(argc, argv);
 148     progName = argv[0];
 149     argc=u_parseArgs(argc, argv, UPRV_LENGTHOF(options), options);
 150     if(argc<0) {
 151         // Unrecognized option
 152         fprintf(stderr, "error in command line argument \"%s\"\n", argv[-argc]);
 153         usageAndDie(U_ILLEGAL_ARGUMENT_ERROR);
 154     }
 155
 156     if(options[0].doesOccur || options[1].doesOccur) {
 157         //  -? or -h for help.
 158         usageAndDie(0);
 159     }
 160
 161     if (!(options[3].doesOccur && options[5].doesOccur)) {
 162         fprintf(stderr, "confusables file and output file must all be specified.\n");
 163         usageAndDie(U_ILLEGAL_ARGUMENT_ERROR);
 164     }
 165     confFileName   = options[3].value;
 166     outFileName    = options[5].value;
 167
 168     if (options[6].doesOccur) {
 169         u_setDataDirectory(options[6].value);
 170     }
 171
 172     status = U_ZERO_ERROR;
 173
 174     /* Combine the directory with the file name */
 175     if(options[7].doesOccur) {
 176         outDir = options[7].value;
 177     }
 178     if (options[8].doesOccur) {
 179         copyright = U_COPYRIGHT_STRING;
 180     }
 181
 182     UBool quiet = FALSE;
 183     if (options[9].doesOccur) {
 184       quiet = TRUE;
 185     }
 186
 187 #if UCONFIG_NO_REGULAR_EXPRESSIONS || UCONFIG_NO_NORMALIZATION || UCONFIG_NO_FILE_IO
 188     // spoof detection data file parsing is dependent on regular expressions.
 189     // TODO: have the tool return an error status.  Requires fixing the ICU data build
 190     //       so that it doesn't abort entirely on that error.
 191
 192     UNewDataMemory *pData;
 193     char msg[1024];
 194
 195     /* write message with just the name */
 196     sprintf(msg, "gencfu writes dummy %s because of UCONFIG_NO_REGULAR_EXPRESSIONS and/or UCONFIG_NO_NORMALIZATION and/or UCONFIG_NO_FILE_IO, see uconfig.h", outFileName);
 197     fprintf(stderr, "%s\n", msg);
 198
 199     /* write the dummy data file */
 200     pData = udata_create(outDir, NULL, outFileName, &dummyDataInfo, NULL, &status);
 201     udata_writeBlock(pData, msg, strlen(msg));
 202     udata_finish(pData, &status);
 203     return (int)status;
 204
 205 #else
 206     /* Initialize ICU */
 207     u_init(&status);
 208     if (U_FAILURE(status)) {
 209         fprintf(stderr, "%s: can not initialize ICU.  status = %s\n",
 210             argv[0], u_errorName(status));
 211         exit(1);
 212     }
 213     status = U_ZERO_ERROR;
 214
 215     //  Read in the confusables source file
 216
 217     int32_t      confusablesLen = 0;
 218     const char  *confusables = readFile(confFileName, &confusablesLen);
 219     if (confusables == NULL) {
 220         printf("gencfu: error reading file  \"%s\"\n", confFileName);
 221         exit(-1);
 222     }
 223
 224     //
 225     //  Create the Spoof Detector from the source confusables files.
 226     //     This will compile the data.
 227     //
 228     UParseError parseError;
 229     parseError.line = 0;
 230     parseError.offset = 0;
 231     int32_t errType;
 232     USpoofChecker *sc = uspoof_openFromSource(confusables, confusablesLen,
 233                                               NULL, 0,
 234                                               &errType, &parseError, &status);
 235     if (U_FAILURE(status)) {
 236         fprintf(stderr, "gencfu: uspoof_openFromSource error \"%s\"  at file %s, line %d, column %d\n",
 237                 u_errorName(status), confFileName, (int)parseError.line, (int)parseError.offset);
 238         exit(status);
 239     };
 240
 241
 242     //
 243     //  Get the compiled rule data from the USpoofChecker.
 244     //
 245     uint32_t        outDataSize;
 246     uint8_t        *outData;
 247     outDataSize = uspoof_serialize(sc, NULL, 0, &status);
 248     if (status != U_BUFFER_OVERFLOW_ERROR) {
 249         fprintf(stderr, "gencfu: uspoof_serialize() returned %s\n", u_errorName(status));
 250         exit(status);
 251     }
 252     status = U_ZERO_ERROR;
 253     outData = new uint8_t[outDataSize];
 254     uspoof_serialize(sc, outData, outDataSize, &status);
 255
 256     // Copy the data format version numbers from the spoof data header into the UDataMemory header.
 257
 258     uprv_memcpy(dh.info.formatVersion,
 259                 reinterpret_cast<SpoofDataHeader *>(outData)->fFormatVersion,
 260                 sizeof(dh.info.formatVersion));
 261
 262     //
 263     //  Create the output file
 264     //
 265     size_t bytesWritten;
 266     UNewDataMemory *pData;
 267     pData = udata_create(outDir, NULL, outFileName, &(dh.info), copyright, &status);
 268     if(U_FAILURE(status)) {
 269         fprintf(stderr, "gencfu: Could not open output file \"%s\", \"%s\"\n",
 270                          outFileName, u_errorName(status));
 271         exit(status);
 272     }
 273
 274
 275     //  Write the data itself.
 276     udata_writeBlock(pData, outData, outDataSize);
 277     // finish up
 278     bytesWritten = udata_finish(pData, &status);
 279     if(U_FAILURE(status)) {
 280         fprintf(stderr, "gencfu: Error %d writing the output file\n", status);
 281         exit(status);
 282     }
 283
 284     if (bytesWritten != outDataSize) {
 285         fprintf(stderr, "gencfu: Error writing to output file \"%s\"\n", outFileName);
 286         exit(-1);
 287     }
 288
 289     uspoof_close(sc);
 290     delete [] outData;
 291     delete [] confusables;
 292     u_cleanup();
 293     if (!quiet) {
 294         printf("gencfu: tool completed successfully.\n");
 295     }
 296     return 0;
 297 #endif   // UCONFIG_NO_REGULAR_EXPRESSIONS
 298 }
 299
 300
 301  //
 302  //  Read in a confusables source file
 303  //
 304  static const char *readFile(const char *fileName, int32_t *len) {
 305     char       *result;
 306     long        fileSize;
 307     FILE        *file;
 308
 309     file = fopen(fileName, "rb");
 310     if( file == 0 ) {
 311         return NULL;
 312     }
 313     fseek(file, 0, SEEK_END);
 314     fileSize = ftell(file);
 315     fseek(file, 0, SEEK_SET);
 316     result = new char[fileSize+10];
 317     if (result==NULL) {
 318         fclose(file);
 319         return NULL;
 320     }
 321
 322     long t = fread(result, 1, fileSize, file);
 323     if (t != fileSize)  {
 324         delete [] result;
 325         fclose(file);
 326         return NULL;
 327     }
 328     result[fileSize]=0;
 329     *len = static_cast<int32_t>(fileSize);
 330     fclose(file);
 331     return result;
 332  }