icuSources/test/intltest/testidn.cpp

   1 /*
   2 *******************************************************************************
   3 *
   4 *   Copyright (C) 2003-2006, International Business Machines
   5 *   Corporation and others.  All Rights Reserved.
   6 *
   7 *******************************************************************************
   8 *   file name:  testidn.cpp
   9 *   encoding:   US-ASCII
  10 *   tab size:   8 (not used)
  11 *   indentation:4
  12 *
  13 *   created on: 2003-02-06
  14 *   created by: Ram Viswanadha
  15 *
  16 *   This program reads the rfc3454_*.txt files,
  17 *   parses them, and extracts the data for Nameprep conformance.
  18 *   It then preprocesses it and writes a binary file for efficient use
  19 *   in various IDNA conversion processes.
  20 */
  21
  22 #include "unicode/utypes.h"
  23
  24 #if !UCONFIG_NO_IDNA && !UCONFIG_NO_TRANSLITERATION
  25
  26 #define USPREP_TYPE_NAMES_ARRAY
  27
  28 #include "unicode/uchar.h"
  29 #include "unicode/putil.h"
  30 #include "cmemory.h"
  31 #include "cstring.h"
  32 #include "unicode/udata.h"
  33 #include "unewdata.h"
  34 #include "uoptions.h"
  35 #include "uparse.h"
  36 #include "utrie.h"
  37 #include "umutex.h"
  38 #include "sprpimpl.h"
  39 #include "testidna.h"
  40 #include "punyref.h"
  41 #include <stdlib.h>
  42
  43 UBool beVerbose=FALSE, haveCopyright=TRUE;
  44
  45 /* prototypes --------------------------------------------------------------- */
  46
  47
  48 static void
  49 parseMappings(const char *filename, UBool reportError,TestIDNA& test, UErrorCode *pErrorCode);
  50
  51 static void
  52 compareMapping(uint32_t codepoint, uint32_t* mapping, int32_t mapLength,
  53                UStringPrepType option);
  54
  55 static void
  56 compareFlagsForRange(uint32_t start, uint32_t end,UStringPrepType option);
  57
  58 static void
  59 testAllCodepoints(TestIDNA& test);
  60
  61 static TestIDNA* pTestIDNA =NULL;
  62
  63 static const char* fileNames[] = {
  64                                     "NamePrepProfile.txt"
  65                                  };
  66 static       UStringPrepProfile *profile = NULL;
  67 static const UTrie *idnTrie              = NULL;
  68 static const int32_t *indexes            = NULL;
  69 static const uint16_t *mappingData       = NULL;
  70 /* -------------------------------------------------------------------------- */
  71
  72 /* file definitions */
  73 #define DATA_NAME "uidna"
  74 #define DATA_TYPE "icu"
  75
  76 #define MISC_DIR "misc"
  77
  78 extern int
  79 testData(TestIDNA& test) {
  80     char* filename = (char*) malloc(strlen(IntlTest::pathToDataDirectory())*5555);
  81     //TODO get the srcDir dynamically
  82     const char *srcDir=IntlTest::pathToDataDirectory();
  83     char *basename=NULL;
  84     UErrorCode errorCode=U_ZERO_ERROR;
  85     char *saveBasename =NULL;
  86
  87     profile = usprep_open(NULL, DATA_NAME, &errorCode);
  88     if(U_FAILURE(errorCode)){
  89         test.errln("Failed to load IDNA data file. " + UnicodeString(u_errorName(errorCode)));
  90         return errorCode;
  91     }
  92
  93     idnTrie     = &profile->sprepTrie;
  94     indexes     = profile->indexes;
  95     mappingData = profile->mappingData;
  96
  97     //initialize
  98     pTestIDNA = &test;
  99
 100     /* prepare the filename beginning with the source dir */
 101     if(uprv_strchr(srcDir,U_FILE_SEP_CHAR) == NULL){
 102         filename[0] = 0x2E;
 103         filename[1] = U_FILE_SEP_CHAR;
 104         uprv_strcpy(filename+2,srcDir);
 105     }else{
 106         uprv_strcpy(filename, srcDir);
 107     }
 108     basename=filename+uprv_strlen(filename);
 109     if(basename>filename && *(basename-1)!=U_FILE_SEP_CHAR) {
 110         *basename++=U_FILE_SEP_CHAR;
 111     }
 112
 113     /* process unassigned */
 114     basename=filename+uprv_strlen(filename);
 115     if(basename>filename && *(basename-1)!=U_FILE_SEP_CHAR) {
 116         *basename++=U_FILE_SEP_CHAR;
 117     }
 118
 119     /* first copy misc directory */
 120     saveBasename = basename;
 121     uprv_strcpy(basename,MISC_DIR);
 122     basename = basename + uprv_strlen(MISC_DIR);
 123     *basename++=U_FILE_SEP_CHAR;
 124
 125     /* process unassigned */
 126     uprv_strcpy(basename,fileNames[0]);
 127     parseMappings(filename,TRUE, test,&errorCode);
 128     if(U_FAILURE(errorCode)) {
 129         test.errln( "Could not open file %s for reading \n", filename);
 130         return errorCode;
 131     }
 132
 133     testAllCodepoints(test);
 134
 135     usprep_close(profile);
 136     pTestIDNA = NULL;
 137     free(filename);
 138     return errorCode;
 139 }
 140 U_CDECL_BEGIN
 141
 142 static void U_CALLCONV
 143 strprepProfileLineFn(void * /*context*/,
 144               char *fields[][2], int32_t fieldCount,
 145               UErrorCode *pErrorCode) {
 146     uint32_t mapping[40];
 147     char *end, *map;
 148     uint32_t code;
 149     int32_t length;
 150    /*UBool* mapWithNorm = (UBool*) context;*/
 151     const char* typeName;
 152     uint32_t rangeStart=0,rangeEnd =0;
 153
 154     if(fieldCount != 3){
 155         *pErrorCode = U_INVALID_FORMAT_ERROR;
 156         return;
 157     }
 158
 159     typeName = fields[2][0];
 160     map = fields[1][0];
 161
 162     if(uprv_strstr(typeName, usprepTypeNames[USPREP_UNASSIGNED])!=NULL){
 163
 164         u_parseCodePointRange(fields[0][0], &rangeStart,&rangeEnd, pErrorCode);
 165
 166         /* store the range */
 167         compareFlagsForRange(rangeStart,rangeEnd,USPREP_UNASSIGNED);
 168
 169     }else if(uprv_strstr(typeName, usprepTypeNames[USPREP_PROHIBITED])!=NULL){
 170
 171         u_parseCodePointRange(fields[0][0], &rangeStart,&rangeEnd, pErrorCode);
 172
 173         /* store the range */
 174         compareFlagsForRange(rangeStart,rangeEnd,USPREP_PROHIBITED);
 175
 176     }else if(uprv_strstr(typeName, usprepTypeNames[USPREP_MAP])!=NULL){
 177         /* get the character code, field 0 */
 178         code=(uint32_t)uprv_strtoul(fields[0][0], &end, 16);
 179
 180         /* parse the mapping string */
 181         length=u_parseCodePoints(map, mapping, sizeof(mapping)/4, pErrorCode);
 182
 183         /* store the mapping */
 184         compareMapping(code,mapping, length,USPREP_MAP);
 185
 186     }else{
 187         *pErrorCode = U_INVALID_FORMAT_ERROR;
 188     }
 189
 190 }
 191
 192 U_CDECL_END
 193
 194 static void
 195 parseMappings(const char *filename,UBool reportError, TestIDNA& test, UErrorCode *pErrorCode) {
 196     char *fields[3][2];
 197
 198     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
 199         return;
 200     }
 201
 202     u_parseDelimitedFile(filename, ';', fields, 3, strprepProfileLineFn, (void*)filename, pErrorCode);
 203
 204     //fprintf(stdout,"Number of code points that have mappings with length >1 : %i\n",len);
 205
 206     if(U_FAILURE(*pErrorCode) && (reportError || *pErrorCode!=U_FILE_ACCESS_ERROR)) {
 207         test.errln( "testidn error: u_parseDelimitedFile(\"%s\") failed - %s\n", filename, u_errorName(*pErrorCode));
 208     }
 209 }
 210
 211
 212 static inline UStringPrepType
 213 getValues(uint32_t result, int32_t& value, UBool& isIndex){
 214
 215     UStringPrepType type;
 216
 217     if(result == 0){
 218         /*
 219          * Initial value stored in the mapping table
 220          * just return USPREP_TYPE_LIMIT .. so that
 221          * the source codepoint is copied to the destination
 222          */
 223         type = USPREP_TYPE_LIMIT;
 224         isIndex =FALSE;
 225         value = 0;
 226     }else if(result >= _SPREP_TYPE_THRESHOLD){
 227         type = (UStringPrepType) (result - _SPREP_TYPE_THRESHOLD);
 228         isIndex =FALSE;
 229         value = 0;
 230     }else{
 231         /* get the state */
 232         type = USPREP_MAP;
 233         /* ascertain if the value is index or delta */
 234         if(result & 0x02){
 235             isIndex = TRUE;
 236             value = result  >> 2; //mask off the lower 2 bits and shift
 237
 238         }else{
 239             isIndex = FALSE;
 240             value = (int16_t)result;
 241             value =  (value >> 2);
 242
 243         }
 244         if((result>>2) == _SPREP_MAX_INDEX_VALUE){
 245             type = USPREP_DELETE;
 246             isIndex =FALSE;
 247             value = 0;
 248         }
 249     }
 250     return type;
 251 }
 252
 253
 254
 255 static void
 256 testAllCodepoints(TestIDNA& test){
 257     /*
 258     {
 259         UChar str[19] = {
 260                             0xC138, 0xACC4, 0xC758, 0xBAA8, 0xB4E0, 0xC0AC, 0xB78C, 0xB4E4, 0xC774,
 261                             0x070F,//prohibited
 262                             0xD55C, 0xAD6D, 0xC5B4, 0xB97C, 0xC774, 0xD574, 0xD55C, 0xB2E4, 0xBA74
 263                         };
 264         uint32_t in[19] = {0};
 265         UErrorCode status = U_ZERO_ERROR;
 266         int32_t inLength=0, outLength=100;
 267         char output[100] = {0};
 268         punycode_status error;
 269         u_strToUTF32((UChar32*)in,19,&inLength,str,19,&status);
 270
 271         error= punycode_encode(inLength, in, NULL, (uint32_t*)&outLength, output);
 272         printf(output);
 273
 274     }
 275     */
 276
 277     uint32_t i = 0;
 278     int32_t unassigned      = 0;
 279     int32_t prohibited      = 0;
 280     int32_t mappedWithNorm  = 0;
 281     int32_t mapped          = 0;
 282     int32_t noValueInTrie   = 0;
 283
 284     UStringPrepType type;
 285     int32_t value;
 286     UBool isIndex = FALSE;
 287
 288     for(i=0;i<=0x10FFFF;i++){
 289         uint32_t result = 0;
 290         UTRIE_GET16(idnTrie,i, result);
 291         type = getValues(result,value, isIndex);
 292         if(type != USPREP_TYPE_LIMIT ){
 293             if(type == USPREP_UNASSIGNED){
 294                 unassigned++;
 295             }
 296             if(type == USPREP_PROHIBITED){
 297                 prohibited++;
 298             }
 299             if(type == USPREP_MAP){
 300                 mapped++;
 301             }
 302         }else{
 303             noValueInTrie++;
 304             if(result > 0){
 305                 test.errln("The return value for 0x%06X is wrong. %i\n",i,result);
 306             }
 307         }
 308     }
 309
 310     test.logln("Number of Unassinged code points : %i \n",unassigned);
 311     test.logln("Number of Prohibited code points : %i \n",prohibited);
 312     test.logln("Number of Mapped code points : %i \n",mapped);
 313     test.logln("Number of Mapped with NFKC code points : %i \n",mappedWithNorm);
 314     test.logln("Number of code points that have no value in Trie: %i \n",noValueInTrie);
 315
 316
 317 }
 318
 319 static void
 320 compareMapping(uint32_t codepoint, uint32_t* mapping,int32_t mapLength,
 321                UStringPrepType type){
 322     uint32_t result = 0;
 323     UTRIE_GET16(idnTrie,codepoint, result);
 324
 325     int32_t length=0;
 326     UBool isIndex;
 327     UStringPrepType retType;
 328     int32_t value, index=0, delta=0;
 329
 330     retType = getValues(result,value,isIndex);
 331
 332
 333     if(type != retType && retType != USPREP_DELETE){
 334
 335         pTestIDNA->errln( "Did not get the assigned type for codepoint 0x%08X. Expected: %i Got: %i\n",codepoint, USPREP_MAP, type);
 336
 337     }
 338
 339     if(isIndex){
 340         index = value;
 341         if(index >= indexes[_SPREP_ONE_UCHAR_MAPPING_INDEX_START] &&
 342                  index < indexes[_SPREP_TWO_UCHARS_MAPPING_INDEX_START]){
 343             length = 1;
 344         }else if(index >= indexes[_SPREP_TWO_UCHARS_MAPPING_INDEX_START] &&
 345                  index < indexes[_SPREP_THREE_UCHARS_MAPPING_INDEX_START]){
 346             length = 2;
 347         }else if(index >= indexes[_SPREP_THREE_UCHARS_MAPPING_INDEX_START] &&
 348                  index < indexes[_SPREP_FOUR_UCHARS_MAPPING_INDEX_START]){
 349             length = 3;
 350         }else{
 351             length = mappingData[index++];
 352         }
 353     }else{
 354         delta = value;
 355         length = (retType == USPREP_DELETE)? 0 :  1;
 356     }
 357
 358     int32_t realLength =0;
 359     /* figure out the real length */
 360     for(int32_t j=0; j<mapLength; j++){
 361         if(mapping[j] > 0xFFFF){
 362             realLength +=2;
 363         }else{
 364             realLength++;
 365         }
 366     }
 367
 368     if(realLength != length){
 369         pTestIDNA->errln( "Did not get the expected length. Expected: %i Got: %i\n", mapLength, length);
 370     }
 371
 372     if(isIndex){
 373         for(int8_t i =0; i< mapLength; i++){
 374             if(mapping[i] <= 0xFFFF){
 375                 if(mappingData[index+i] != (uint16_t)mapping[i]){
 376                     pTestIDNA->errln("Did not get the expected result. Expected: 0x%04X Got: 0x%04X \n", mapping[i], mappingData[index+i]);
 377                 }
 378             }else{
 379                 UChar lead  = UTF16_LEAD(mapping[i]);
 380                 UChar trail = UTF16_TRAIL(mapping[i]);
 381                 if(mappingData[index+i] != lead ||
 382                     mappingData[index+i+1] != trail){
 383                     pTestIDNA->errln( "Did not get the expected result. Expected: 0x%04X 0x%04X  Got: 0x%04X 0x%04X", lead, trail, mappingData[index+i], mappingData[index+i+1]);
 384                 }
 385             }
 386         }
 387     }else{
 388         if(retType!=USPREP_DELETE && (codepoint-delta) != (uint16_t)mapping[0]){
 389             pTestIDNA->errln("Did not get the expected result. Expected: 0x%04X Got: 0x%04X \n", mapping[0],(codepoint-delta));
 390         }
 391     }
 392
 393 }
 394
 395 static void
 396 compareFlagsForRange(uint32_t start, uint32_t end,
 397                      UStringPrepType type){
 398
 399     uint32_t result =0 ;
 400     UStringPrepType retType;
 401     UBool isIndex=FALSE;
 402     int32_t value=0;
 403 /*
 404     // supplementary code point
 405     UChar __lead16=UTF16_LEAD(0x2323E);
 406     int32_t __offset;
 407
 408     // get data for lead surrogate
 409     (result)=_UTRIE_GET_RAW((&idnTrie), index, 0, (__lead16));
 410     __offset=(&idnTrie)->getFoldingOffset(result);
 411
 412     // get the real data from the folded lead/trail units
 413     if(__offset>0) {
 414         (result)=_UTRIE_GET_RAW((&idnTrie), index, __offset, (0x2323E)&0x3ff);
 415     } else {
 416         (result)=(uint32_t)((&idnTrie)->initialValue);
 417     }
 418
 419     UTRIE_GET16(&idnTrie,0x2323E, result);
 420 */
 421     while(start < end+1){
 422         UTRIE_GET16(idnTrie,start, result);
 423         retType = getValues(result,value,isIndex);
 424         if(result > _SPREP_TYPE_THRESHOLD){
 425             if(retType != type){
 426                 pTestIDNA->errln( "FAIL: Did not get the expected type for 0x%06X. Expected: %s Got: %s\n",start,usprepTypeNames[type], usprepTypeNames[retType]);
 427             }
 428         }else{
 429             if(type == USPREP_PROHIBITED && ((result & 0x01) != 0x01)){
 430                 pTestIDNA->errln( "FAIL: Did not get the expected type for 0x%06X. Expected: %s Got: %s\n",start,usprepTypeNames[type], usprepTypeNames[retType]);
 431             }
 432         }
 433
 434         start++;
 435     }
 436
 437 }
 438
 439
 440 #endif /* #if !UCONFIG_NO_IDNA */
 441
 442 /*
 443  * Hey, Emacs, please set the following:
 444  *
 445  * Local Variables:
 446  * indent-tabs-mode: nil
 447  * End:
 448  *
 449  */