gencfu.cpp (10659B)
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 file 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 by ICU spoof detectors. 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", nullptr, nullptr, nullptr, 'r', UOPT_REQUIRES_ARG, 0 }, /* 3 */ 59 { "wsrules", nullptr, nullptr, nullptr, 'w', UOPT_REQUIRES_ARG, 0}, /* 4 */ // deprecated 60 { "out", nullptr, nullptr, nullptr, '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 = nullptr; 141 const char *copyright = nullptr; 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 snprintf(msg, sizeof(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, nullptr, outFileName, &dummyDataInfo, nullptr, &status); 201 udata_writeBlock(pData, msg, strlen(msg)); 202 udata_finish(pData, &status); 203 return (int)status; 204 205 #else 206 207 // Read in the confusables source file 208 209 int32_t confusablesLen = 0; 210 const char *confusables = readFile(confFileName, &confusablesLen); 211 if (confusables == nullptr) { 212 printf("gencfu: error reading file \"%s\"\n", confFileName); 213 exit(-1); 214 } 215 216 // 217 // Create the Spoof Detector from the source confusables files. 218 // This will compile the data. 219 // 220 UParseError parseError; 221 parseError.line = 0; 222 parseError.offset = 0; 223 int32_t errType; 224 USpoofChecker *sc = uspoof_openFromSource(confusables, confusablesLen, 225 nullptr, 0, 226 &errType, &parseError, &status); 227 if (U_FAILURE(status)) { 228 fprintf(stderr, "gencfu: uspoof_openFromSource error \"%s\" at file %s, line %d, column %d\n", 229 u_errorName(status), confFileName, static_cast<int>(parseError.line), static_cast<int>(parseError.offset)); 230 exit(status); 231 } 232 233 234 // 235 // Get the compiled rule data from the USpoofChecker. 236 // 237 uint32_t outDataSize; 238 uint8_t *outData; 239 outDataSize = uspoof_serialize(sc, nullptr, 0, &status); 240 if (status != U_BUFFER_OVERFLOW_ERROR) { 241 fprintf(stderr, "gencfu: uspoof_serialize() returned %s\n", u_errorName(status)); 242 exit(status); 243 } 244 status = U_ZERO_ERROR; 245 outData = new uint8_t[outDataSize]; 246 uspoof_serialize(sc, outData, outDataSize, &status); 247 248 // Copy the data format version numbers from the spoof data header into the UDataMemory header. 249 250 uprv_memcpy(dh.info.formatVersion, 251 reinterpret_cast<SpoofDataHeader *>(outData)->fFormatVersion, 252 sizeof(dh.info.formatVersion)); 253 254 // 255 // Create the output file 256 // 257 size_t bytesWritten; 258 UNewDataMemory *pData; 259 pData = udata_create(outDir, nullptr, outFileName, &(dh.info), copyright, &status); 260 if(U_FAILURE(status)) { 261 fprintf(stderr, "gencfu: Could not open output file \"%s\", \"%s\"\n", 262 outFileName, u_errorName(status)); 263 exit(status); 264 } 265 266 267 // Write the data itself. 268 udata_writeBlock(pData, outData, outDataSize); 269 // finish up 270 bytesWritten = udata_finish(pData, &status); 271 if(U_FAILURE(status)) { 272 fprintf(stderr, "gencfu: Error %d writing the output file\n", status); 273 exit(status); 274 } 275 276 if (bytesWritten != outDataSize) { 277 fprintf(stderr, "gencfu: Error writing to output file \"%s\"\n", outFileName); 278 exit(-1); 279 } 280 281 uspoof_close(sc); 282 delete [] outData; 283 delete [] confusables; 284 u_cleanup(); 285 if (!quiet) { 286 printf("gencfu: tool completed successfully.\n"); 287 } 288 return 0; 289 #endif // UCONFIG_NO_REGULAR_EXPRESSIONS 290 } 291 292 293 // 294 // Read in a confusables source file 295 // 296 static const char *readFile(const char *fileName, int32_t *len) { 297 char *result; 298 long fileSize; 299 FILE *file; 300 301 file = fopen(fileName, "rb"); 302 if (file == nullptr) { 303 return nullptr; 304 } 305 fseek(file, 0, SEEK_END); 306 fileSize = ftell(file); 307 fseek(file, 0, SEEK_SET); 308 result = new char[fileSize+10]; 309 if (result==nullptr) { 310 fclose(file); 311 return nullptr; 312 } 313 314 long t = static_cast<long>(fread(result, 1, fileSize, file)); 315 if (t != fileSize) { 316 delete [] result; 317 fclose(file); 318 return nullptr; 319 } 320 result[fileSize]=0; 321 *len = static_cast<int32_t>(fileSize); 322 fclose(file); 323 return result; 324 }