icuSources/test/intltest/testidn.cpp

   1 /*
   2 *******************************************************************************
   3 *
   4 *   Copyright (C) 2003-2009, 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                                     "rfc3491.txt"
  65                                  };
  66 static const UTrie *idnTrie              = NULL;
  67 static const int32_t *indexes            = NULL;
  68 static const uint16_t *mappingData       = NULL;
  69 /* -------------------------------------------------------------------------- */
  70
  71 /* file definitions */
  72 #define DATA_TYPE "icu"
  73
  74 #define SPREP_DIR "sprep"
  75
  76 extern int
  77 testData(TestIDNA& test) {
  78     char *basename=NULL;
  79     UErrorCode errorCode=U_ZERO_ERROR;
  80     char *saveBasename =NULL;
  81
  82     LocalUStringPrepProfilePointer profile(usprep_openByType(USPREP_RFC3491_NAMEPREP, &errorCode));
  83     if(U_FAILURE(errorCode)){
  84         test.errcheckln(errorCode, "Failed to load IDNA data file. " + UnicodeString(u_errorName(errorCode)));
  85         return errorCode;
  86     }
  87
  88     char* filename = (char*) malloc(strlen(IntlTest::pathToDataDirectory())*1024);
  89     //TODO get the srcDir dynamically
  90     const char *srcDir=IntlTest::pathToDataDirectory();
  91
  92     idnTrie     = &profile->sprepTrie;
  93     indexes     = profile->indexes;
  94     mappingData = profile->mappingData;
  95
  96     //initialize
  97     pTestIDNA = &test;
  98
  99     /* prepare the filename beginning with the source dir */
 100     if(uprv_strchr(srcDir,U_FILE_SEP_CHAR) == NULL){
 101         filename[0] = 0x2E;
 102         filename[1] = U_FILE_SEP_CHAR;
 103         uprv_strcpy(filename+2,srcDir);
 104     }else{
 105         uprv_strcpy(filename, srcDir);
 106     }
 107     basename=filename+uprv_strlen(filename);
 108     if(basename>filename && *(basename-1)!=U_FILE_SEP_CHAR) {
 109         *basename++=U_FILE_SEP_CHAR;
 110     }
 111
 112     /* process unassigned */
 113     basename=filename+uprv_strlen(filename);
 114     if(basename>filename && *(basename-1)!=U_FILE_SEP_CHAR) {
 115         *basename++=U_FILE_SEP_CHAR;
 116     }
 117
 118     /* first copy misc directory */
 119     saveBasename = basename;
 120     uprv_strcpy(basename,SPREP_DIR);
 121     basename = basename + uprv_strlen(SPREP_DIR);
 122     *basename++=U_FILE_SEP_CHAR;
 123
 124     /* process unassigned */
 125     uprv_strcpy(basename,fileNames[0]);
 126     parseMappings(filename,TRUE, test,&errorCode);
 127     if(U_FAILURE(errorCode)) {
 128         test.errln( "Could not open file %s for reading \n", filename);
 129         return errorCode;
 130     }
 131
 132     testAllCodepoints(test);
 133
 134     pTestIDNA = NULL;
 135     free(filename);
 136     return errorCode;
 137 }
 138 U_CDECL_BEGIN
 139
 140 static void U_CALLCONV
 141 strprepProfileLineFn(void * /*context*/,
 142               char *fields[][2], int32_t fieldCount,
 143               UErrorCode *pErrorCode) {
 144     uint32_t mapping[40];
 145     char *end, *map;
 146     uint32_t code;
 147     int32_t length;
 148    /*UBool* mapWithNorm = (UBool*) context;*/
 149     const char* typeName;
 150     uint32_t rangeStart=0,rangeEnd =0;
 151     const char *s;
 152
 153     s = u_skipWhitespace(fields[0][0]);
 154     if (*s == '@') {
 155         /* a special directive introduced in 4.2 */
 156         return;
 157     }
 158
 159     if(fieldCount != 3){
 160         *pErrorCode = U_INVALID_FORMAT_ERROR;
 161         return;
 162     }
 163
 164     typeName = fields[2][0];
 165     map = fields[1][0];
 166
 167     if(uprv_strstr(typeName, usprepTypeNames[USPREP_UNASSIGNED])!=NULL){
 168
 169         u_parseCodePointRange(s, &rangeStart,&rangeEnd, pErrorCode);
 170
 171         /* store the range */
 172         compareFlagsForRange(rangeStart,rangeEnd,USPREP_UNASSIGNED);
 173
 174     }else if(uprv_strstr(typeName, usprepTypeNames[USPREP_PROHIBITED])!=NULL){
 175
 176         u_parseCodePointRange(s, &rangeStart,&rangeEnd, pErrorCode);
 177
 178         /* store the range */
 179         compareFlagsForRange(rangeStart,rangeEnd,USPREP_PROHIBITED);
 180
 181     }else if(uprv_strstr(typeName, usprepTypeNames[USPREP_MAP])!=NULL){
 182         /* get the character code, field 0 */
 183         code=(uint32_t)uprv_strtoul(s, &end, 16);
 184
 185         /* parse the mapping string */
 186         length=u_parseCodePoints(map, mapping, sizeof(mapping)/4, pErrorCode);
 187
 188         /* store the mapping */
 189         compareMapping(code,mapping, length,USPREP_MAP);
 190
 191     }else{
 192         *pErrorCode = U_INVALID_FORMAT_ERROR;
 193     }
 194
 195 }
 196
 197 U_CDECL_END
 198
 199 static void
 200 parseMappings(const char *filename,UBool reportError, TestIDNA& test, UErrorCode *pErrorCode) {
 201     char *fields[3][2];
 202
 203     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
 204         return;
 205     }
 206
 207     u_parseDelimitedFile(filename, ';', fields, 3, strprepProfileLineFn, (void*)filename, pErrorCode);
 208
 209     //fprintf(stdout,"Number of code points that have mappings with length >1 : %i\n",len);
 210
 211     if(U_FAILURE(*pErrorCode) && (reportError || *pErrorCode!=U_FILE_ACCESS_ERROR)) {
 212         test.errln( "testidn error: u_parseDelimitedFile(\"%s\") failed - %s\n", filename, u_errorName(*pErrorCode));
 213     }
 214 }
 215
 216
 217 static inline UStringPrepType
 218 getValues(uint32_t result, int32_t& value, UBool& isIndex){
 219
 220     UStringPrepType type;
 221
 222     if(result == 0){
 223         /*
 224          * Initial value stored in the mapping table
 225          * just return USPREP_TYPE_LIMIT .. so that
 226          * the source codepoint is copied to the destination
 227          */
 228         type = USPREP_TYPE_LIMIT;
 229         isIndex =FALSE;
 230         value = 0;
 231     }else if(result >= _SPREP_TYPE_THRESHOLD){
 232         type = (UStringPrepType) (result - _SPREP_TYPE_THRESHOLD);
 233         isIndex =FALSE;
 234         value = 0;
 235     }else{
 236         /* get the state */
 237         type = USPREP_MAP;
 238         /* ascertain if the value is index or delta */
 239         if(result & 0x02){
 240             isIndex = TRUE;
 241             value = result  >> 2; //mask off the lower 2 bits and shift
 242
 243         }else{
 244             isIndex = FALSE;
 245             value = (int16_t)result;
 246             value =  (value >> 2);
 247
 248         }
 249         if((result>>2) == _SPREP_MAX_INDEX_VALUE){
 250             type = USPREP_DELETE;
 251             isIndex =FALSE;
 252             value = 0;
 253         }
 254     }
 255     return type;
 256 }
 257
 258
 259
 260 static void
 261 testAllCodepoints(TestIDNA& test){
 262     /*
 263     {
 264         UChar str[19] = {
 265                             0xC138, 0xACC4, 0xC758, 0xBAA8, 0xB4E0, 0xC0AC, 0xB78C, 0xB4E4, 0xC774,
 266                             0x070F,//prohibited
 267                             0xD55C, 0xAD6D, 0xC5B4, 0xB97C, 0xC774, 0xD574, 0xD55C, 0xB2E4, 0xBA74
 268                         };
 269         uint32_t in[19] = {0};
 270         UErrorCode status = U_ZERO_ERROR;
 271         int32_t inLength=0, outLength=100;
 272         char output[100] = {0};
 273         punycode_status error;
 274         u_strToUTF32((UChar32*)in,19,&inLength,str,19,&status);
 275
 276         error= punycode_encode(inLength, in, NULL, (uint32_t*)&outLength, output);
 277         printf(output);
 278
 279     }
 280     */
 281
 282     uint32_t i = 0;
 283     int32_t unassigned      = 0;
 284     int32_t prohibited      = 0;
 285     int32_t mappedWithNorm  = 0;
 286     int32_t mapped          = 0;
 287     int32_t noValueInTrie   = 0;
 288
 289     UStringPrepType type;
 290     int32_t value;
 291     UBool isIndex = FALSE;
 292
 293     for(i=0;i<=0x10FFFF;i++){
 294         uint32_t result = 0;
 295         UTRIE_GET16(idnTrie,i, result);
 296         type = getValues(result,value, isIndex);
 297         if(type != USPREP_TYPE_LIMIT ){
 298             if(type == USPREP_UNASSIGNED){
 299                 unassigned++;
 300             }
 301             if(type == USPREP_PROHIBITED){
 302                 prohibited++;
 303             }
 304             if(type == USPREP_MAP){
 305                 mapped++;
 306             }
 307         }else{
 308             noValueInTrie++;
 309             if(result > 0){
 310                 test.errln("The return value for 0x%06X is wrong. %i\n",i,result);
 311             }
 312         }
 313     }
 314
 315     test.logln("Number of Unassinged code points : %i \n",unassigned);
 316     test.logln("Number of Prohibited code points : %i \n",prohibited);
 317     test.logln("Number of Mapped code points : %i \n",mapped);
 318     test.logln("Number of Mapped with NFKC code points : %i \n",mappedWithNorm);
 319     test.logln("Number of code points that have no value in Trie: %i \n",noValueInTrie);
 320
 321
 322 }
 323
 324 static void
 325 compareMapping(uint32_t codepoint, uint32_t* mapping,int32_t mapLength,
 326                UStringPrepType type){
 327     uint32_t result = 0;
 328     UTRIE_GET16(idnTrie,codepoint, result);
 329
 330     int32_t length=0;
 331     UBool isIndex;
 332     UStringPrepType retType;
 333     int32_t value, index=0, delta=0;
 334
 335     retType = getValues(result,value,isIndex);
 336
 337
 338     if(type != retType && retType != USPREP_DELETE){
 339
 340         pTestIDNA->errln( "Did not get the assigned type for codepoint 0x%08X. Expected: %i Got: %i\n",codepoint, USPREP_MAP, type);
 341
 342     }
 343
 344     if(isIndex){
 345         index = value;
 346         if(index >= indexes[_SPREP_ONE_UCHAR_MAPPING_INDEX_START] &&
 347                  index < indexes[_SPREP_TWO_UCHARS_MAPPING_INDEX_START]){
 348             length = 1;
 349         }else if(index >= indexes[_SPREP_TWO_UCHARS_MAPPING_INDEX_START] &&
 350                  index < indexes[_SPREP_THREE_UCHARS_MAPPING_INDEX_START]){
 351             length = 2;
 352         }else if(index >= indexes[_SPREP_THREE_UCHARS_MAPPING_INDEX_START] &&
 353                  index < indexes[_SPREP_FOUR_UCHARS_MAPPING_INDEX_START]){
 354             length = 3;
 355         }else{
 356             length = mappingData[index++];
 357         }
 358     }else{
 359         delta = value;
 360         length = (retType == USPREP_DELETE)? 0 :  1;
 361     }
 362
 363     int32_t realLength =0;
 364     /* figure out the real length */
 365     for(int32_t j=0; j<mapLength; j++){
 366         if(mapping[j] > 0xFFFF){
 367             realLength +=2;
 368         }else{
 369             realLength++;
 370         }
 371     }
 372
 373     if(realLength != length){
 374         pTestIDNA->errln( "Did not get the expected length. Expected: %i Got: %i\n", mapLength, length);
 375     }
 376
 377     if(isIndex){
 378         for(int8_t i =0; i< mapLength; i++){
 379             if(mapping[i] <= 0xFFFF){
 380                 if(mappingData[index+i] != (uint16_t)mapping[i]){
 381                     pTestIDNA->errln("Did not get the expected result. Expected: 0x%04X Got: 0x%04X \n", mapping[i], mappingData[index+i]);
 382                 }
 383             }else{
 384                 UChar lead  = UTF16_LEAD(mapping[i]);
 385                 UChar trail = UTF16_TRAIL(mapping[i]);
 386                 if(mappingData[index+i] != lead ||
 387                     mappingData[index+i+1] != trail){
 388                     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]);
 389                 }
 390             }
 391         }
 392     }else{
 393         if(retType!=USPREP_DELETE && (codepoint-delta) != (uint16_t)mapping[0]){
 394             pTestIDNA->errln("Did not get the expected result. Expected: 0x%04X Got: 0x%04X \n", mapping[0],(codepoint-delta));
 395         }
 396     }
 397
 398 }
 399
 400 static void
 401 compareFlagsForRange(uint32_t start, uint32_t end,
 402                      UStringPrepType type){
 403
 404     uint32_t result =0 ;
 405     UStringPrepType retType;
 406     UBool isIndex=FALSE;
 407     int32_t value=0;
 408 /*
 409     // supplementary code point
 410     UChar __lead16=UTF16_LEAD(0x2323E);
 411     int32_t __offset;
 412
 413     // get data for lead surrogate
 414     (result)=_UTRIE_GET_RAW((&idnTrie), index, 0, (__lead16));
 415     __offset=(&idnTrie)->getFoldingOffset(result);
 416
 417     // get the real data from the folded lead/trail units
 418     if(__offset>0) {
 419         (result)=_UTRIE_GET_RAW((&idnTrie), index, __offset, (0x2323E)&0x3ff);
 420     } else {
 421         (result)=(uint32_t)((&idnTrie)->initialValue);
 422     }
 423
 424     UTRIE_GET16(&idnTrie,0x2323E, result);
 425 */
 426     while(start < end+1){
 427         UTRIE_GET16(idnTrie,start, result);
 428         retType = getValues(result,value,isIndex);
 429         if(result > _SPREP_TYPE_THRESHOLD){
 430             if(retType != type){
 431                 pTestIDNA->errln( "FAIL: Did not get the expected type for 0x%06X. Expected: %s Got: %s\n",start,usprepTypeNames[type], usprepTypeNames[retType]);
 432             }
 433         }else{
 434             if(type == USPREP_PROHIBITED && ((result & 0x01) != 0x01)){
 435                 pTestIDNA->errln( "FAIL: Did not get the expected type for 0x%06X. Expected: %s Got: %s\n",start,usprepTypeNames[type], usprepTypeNames[retType]);
 436             }
 437         }
 438
 439         start++;
 440     }
 441
 442 }
 443
 444
 445 #endif /* #if !UCONFIG_NO_IDNA */
 446
 447 /*
 448  * Hey, Emacs, please set the following:
 449  *
 450  * Local Variables:
 451  * indent-tabs-mode: nil
 452  * End:
 453  *
 454  */