icuSources/test/cintltst/trie2test.c

   1 /*
   2 ******************************************************************************
   3 *
   4 *   Copyright (C) 2001-2008, International Business Machines
   5 *   Corporation and others.  All Rights Reserved.
   6 *
   7 ******************************************************************************
   8 *   file name:  trietest.c
   9 *   encoding:   US-ASCII
  10 *   tab size:   8 (not used)
  11 *   indentation:4
  12 *
  13 *   created on: 2008sep01 (starting from a copy of trietest.c)
  14 *   created by: Markus W. Scherer
  15 */
  16
  17 #include <stdio.h>
  18 #include "unicode/utypes.h"
  19 #include "utrie2.h"
  20 #include "utrie.h"
  21 #include "cstring.h"
  22 #include "cmemory.h"
  23 #include "udataswp.h"
  24 #include "cintltst.h"
  25
  26 #define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0]))
  27
  28 void addTrie2Test(TestNode** root);
  29
  30 /* Values for setting possibly overlapping, out-of-order ranges of values */
  31 typedef struct SetRange {
  32     UChar32 start, limit;
  33     uint32_t value;
  34     UBool overwrite;
  35 } SetRange;
  36
  37 /*
  38  * Values for testing:
  39  * value is set from the previous boundary's limit to before
  40  * this boundary's limit
  41  *
  42  * There must be an entry with limit 0 and the intialValue.
  43  * It may be preceded by an entry with negative limit and the errorValue.
  44  */
  45 typedef struct CheckRange {
  46     UChar32 limit;
  47     uint32_t value;
  48 } CheckRange;
  49
  50 static int32_t
  51 skipSpecialValues(const CheckRange checkRanges[], int32_t countCheckRanges) {
  52     int32_t i;
  53     for(i=0; i<countCheckRanges && checkRanges[i].limit<=0; ++i) {}
  54     return i;
  55 }
  56
  57 static int32_t
  58 getSpecialValues(const CheckRange checkRanges[], int32_t countCheckRanges,
  59                  uint32_t *pInitialValue, uint32_t *pErrorValue) {
  60     int32_t i=0;
  61     if(i<countCheckRanges && checkRanges[i].limit<0) {
  62         *pErrorValue=checkRanges[i++].value;
  63     } else {
  64         *pErrorValue=0xbad;
  65     }
  66     if(i<countCheckRanges && checkRanges[i].limit==0) {
  67         *pInitialValue=checkRanges[i++].value;
  68     } else {
  69         *pInitialValue=0;
  70     }
  71     return i;
  72 }
  73
  74 /* utrie2_enum() callback, modifies a value */
  75 static uint32_t U_CALLCONV
  76 testEnumValue(const void *context, uint32_t value) {
  77     return value^0x5555;
  78 }
  79
  80 /* utrie2_enum() callback, verifies a range */
  81 static UBool U_CALLCONV
  82 testEnumRange(const void *context, UChar32 start, UChar32 end, uint32_t value) {
  83     const CheckRange **pb=(const CheckRange **)context;
  84     const CheckRange *b=(*pb)++;
  85     UChar32 limit=end+1;
  86
  87     value^=0x5555;
  88     if(start!=(b-1)->limit || limit!=b->limit || value!=b->value) {
  89         log_err("error: utrie2_enum() delivers wrong range [U+%04lx..U+%04lx].0x%lx instead of [U+%04lx..U+%04lx].0x%lx\n",
  90             (long)start, (long)end, (long)value,
  91             (long)(b-1)->limit, (long)b->limit-1, (long)b->value);
  92     }
  93     return TRUE;
  94 }
  95
  96 static void
  97 testTrieEnum(const char *testName,
  98              const UTrie2 *trie,
  99              const CheckRange checkRanges[], int32_t countCheckRanges) {
 100     /* skip over special values */
 101     while(countCheckRanges>0 && checkRanges[0].limit<=0) {
 102         ++checkRanges;
 103         --countCheckRanges;
 104     }
 105     utrie2_enum(trie, testEnumValue, testEnumRange, &checkRanges);
 106 }
 107
 108 /* verify all expected values via UTRIE2_GETxx() */
 109 static void
 110 testTrieGetters(const char *testName,
 111                 const UTrie2 *trie, UTrie2ValueBits valueBits,
 112                 const CheckRange checkRanges[], int32_t countCheckRanges) {
 113     uint32_t initialValue, errorValue;
 114     uint32_t value, value2;
 115     UChar32 start, limit;
 116     int32_t i, countSpecials;
 117
 118     UBool isFrozen=utrie2_isFrozen(trie);
 119     const char *const typeName= isFrozen ? "frozen trie" : "newTrie";
 120
 121     countSpecials=getSpecialValues(checkRanges, countCheckRanges, &initialValue, &errorValue);
 122
 123     start=0;
 124     for(i=countSpecials; i<countCheckRanges; ++i) {
 125         limit=checkRanges[i].limit;
 126         value=checkRanges[i].value;
 127
 128         while(start<limit) {
 129             if(isFrozen) {
 130                 if(start<=0xffff) {
 131                     if(!U_IS_LEAD(start)) {
 132                         if(valueBits==UTRIE2_16_VALUE_BITS) {
 133                             value2=UTRIE2_GET16_FROM_U16_SINGLE_LEAD(trie, start);
 134                         } else {
 135                             value2=UTRIE2_GET32_FROM_U16_SINGLE_LEAD(trie, start);
 136                         }
 137                         if(value!=value2) {
 138                             log_err("error: %s(%s).fromBMP(U+%04lx)==0x%lx instead of 0x%lx\n",
 139                                     typeName, testName, (long)start, (long)value2, (long)value);
 140                         }
 141                     }
 142                 } else {
 143                     if(valueBits==UTRIE2_16_VALUE_BITS) {
 144                         value2=UTRIE2_GET16_FROM_SUPP(trie, start);
 145                     } else {
 146                         value2=UTRIE2_GET32_FROM_SUPP(trie, start);
 147                     }
 148                     if(value!=value2) {
 149                         log_err("error: %s(%s).fromSupp(U+%04lx)==0x%lx instead of 0x%lx\n",
 150                                 typeName, testName, (long)start, (long)value2, (long)value);
 151                     }
 152                 }
 153                 if(valueBits==UTRIE2_16_VALUE_BITS) {
 154                     value2=UTRIE2_GET16(trie, start);
 155                 } else {
 156                     value2=UTRIE2_GET32(trie, start);
 157                 }
 158                 if(value!=value2) {
 159                     log_err("error: %s(%s).get(U+%04lx)==0x%lx instead of 0x%lx\n",
 160                             typeName, testName, (long)start, (long)value2, (long)value);
 161                 }
 162             }
 163             value2=utrie2_get32(trie, start);
 164             if(value!=value2) {
 165                 log_err("error: %s(%s).get32(U+%04lx)==0x%lx instead of 0x%lx\n",
 166                         typeName, testName, (long)start, (long)value2, (long)value);
 167             }
 168             ++start;
 169         }
 170     }
 171
 172     if(isFrozen) {
 173         /* test linear ASCII range from the data array pointer (access to "internal" field) */
 174         start=0;
 175         for(i=countSpecials; i<countCheckRanges && start<=0x7f; ++i) {
 176             limit=checkRanges[i].limit;
 177             value=checkRanges[i].value;
 178
 179             while(start<limit && start<=0x7f) {
 180                 if(valueBits==UTRIE2_16_VALUE_BITS) {
 181                     value2=trie->data16[start];
 182                 } else {
 183                     value2=trie->data32[start];
 184                 }
 185                 if(value!=value2) {
 186                     log_err("error: %s(%s).asciiData[U+%04lx]==0x%lx instead of 0x%lx\n",
 187                             typeName, testName, (long)start, (long)value2, (long)value);
 188                 }
 189                 ++start;
 190             }
 191         }
 192         while(start<=0xbf) {
 193             if(valueBits==UTRIE2_16_VALUE_BITS) {
 194                 value2=trie->data16[start];
 195             } else {
 196                 value2=trie->data32[start];
 197             }
 198             if(errorValue!=value2) {
 199                 log_err("error: %s(%s).badData[U+%04lx]==0x%lx instead of 0x%lx\n",
 200                         typeName, testName, (long)start, (long)value2, (long)errorValue);
 201             }
 202             ++start;
 203         }
 204     }
 205
 206     if(0!=strncmp(testName, "dummy", 5) && 0!=strncmp(testName, "trie1", 5)) {
 207         /* test values for lead surrogate code units */
 208         for(start=0xd7ff; start<0xdc01; ++start) {
 209             switch(start) {
 210             case 0xd7ff:
 211             case 0xdc00:
 212                 value=errorValue;
 213                 break;
 214             case 0xd800:
 215                 value=90;
 216                 break;
 217             case 0xd999:
 218                 value=94;
 219                 break;
 220             case 0xdbff:
 221                 value=99;
 222                 break;
 223             default:
 224                 value=initialValue;
 225                 break;
 226             }
 227             if(isFrozen && U_IS_LEAD(start)) {
 228                 if(valueBits==UTRIE2_16_VALUE_BITS) {
 229                     value2=UTRIE2_GET16_FROM_U16_SINGLE_LEAD(trie, start);
 230                 } else {
 231                     value2=UTRIE2_GET32_FROM_U16_SINGLE_LEAD(trie, start);
 232                 }
 233                 if(value2!=value) {
 234                     log_err("error: %s(%s).LSCU(U+%04lx)==0x%lx instead of 0x%lx\n",
 235                             typeName, testName, (long)start, (long)value2, (long)value);
 236                 }
 237             }
 238             value2=utrie2_get32FromLeadSurrogateCodeUnit(trie, start);
 239             if(value2!=value) {
 240                 log_err("error: %s(%s).lscu(U+%04lx)==0x%lx instead of 0x%lx\n",
 241                         typeName, testName, (long)start, (long)value2, (long)value);
 242             }
 243         }
 244     }
 245
 246     /* test errorValue */
 247     if(isFrozen) {
 248         if(valueBits==UTRIE2_16_VALUE_BITS) {
 249             value=UTRIE2_GET16(trie, -1);
 250             value2=UTRIE2_GET16(trie, 0x110000);
 251         } else {
 252             value=UTRIE2_GET32(trie, -1);
 253             value2=UTRIE2_GET32(trie, 0x110000);
 254         }
 255         if(value!=errorValue || value2!=errorValue) {
 256             log_err("error: %s(%s).get(out of range) != errorValue\n",
 257                     typeName, testName);
 258         }
 259     }
 260     value=utrie2_get32(trie, -1);
 261     value2=utrie2_get32(trie, 0x110000);
 262     if(value!=errorValue || value2!=errorValue) {
 263         log_err("error: %s(%s).get32(out of range) != errorValue\n",
 264                 typeName, testName);
 265     }
 266 }
 267
 268 static void
 269 testTrieUTF16(const char *testName,
 270               const UTrie2 *trie, UTrie2ValueBits valueBits,
 271               const CheckRange checkRanges[], int32_t countCheckRanges) {
 272     UChar s[200];
 273     uint32_t values[100];
 274
 275     const UChar *p, *limit;
 276
 277     uint32_t value;
 278     UChar32 prevCP, c, c2;
 279     int32_t i, length, sIndex, countValues;
 280
 281     /* write a string */
 282     prevCP=0;
 283     length=countValues=0;
 284     for(i=skipSpecialValues(checkRanges, countCheckRanges); i<countCheckRanges; ++i) {
 285         value=checkRanges[i].value;
 286         /* write three code points */
 287         U16_APPEND_UNSAFE(s, length, prevCP);   /* start of the range */
 288         values[countValues++]=value;
 289         c=checkRanges[i].limit;
 290         prevCP=(prevCP+c)/2;                    /* middle of the range */
 291         U16_APPEND_UNSAFE(s, length, prevCP);
 292         values[countValues++]=value;
 293         prevCP=c;
 294         --c;                                    /* end of the range */
 295         U16_APPEND_UNSAFE(s, length, c);
 296         values[countValues++]=value;
 297     }
 298     limit=s+length;
 299
 300     /* try forward */
 301     p=s;
 302     i=0;
 303     while(p<limit) {
 304         sIndex=(int32_t)(p-s);
 305         U16_NEXT(s, sIndex, length, c2);
 306         c=0x33;
 307         if(valueBits==UTRIE2_16_VALUE_BITS) {
 308             UTRIE2_U16_NEXT16(trie, p, limit, c, value);
 309         } else {
 310             UTRIE2_U16_NEXT32(trie, p, limit, c, value);
 311         }
 312         if(value!=values[i]) {
 313             log_err("error: wrong value from UTRIE2_NEXT(%s)(U+%04lx): 0x%lx instead of 0x%lx\n",
 314                     testName, (long)c, (long)value, (long)values[i]);
 315         }
 316         if(c!=c2) {
 317             log_err("error: wrong code point from UTRIE2_NEXT(%s): U+%04lx != U+%04lx\n",
 318                     testName, (long)c, (long)c2);
 319             continue;
 320         }
 321         ++i;
 322     }
 323
 324     /* try backward */
 325     p=limit;
 326     i=countValues;
 327     while(s<p) {
 328         --i;
 329         sIndex=(int32_t)(p-s);
 330         U16_PREV(s, 0, sIndex, c2);
 331         c=0x33;
 332         if(valueBits==UTRIE2_16_VALUE_BITS) {
 333             UTRIE2_U16_PREV16(trie, s, p, c, value);
 334         } else {
 335             UTRIE2_U16_PREV32(trie, s, p, c, value);
 336         }
 337         if(value!=values[i]) {
 338             log_err("error: wrong value from UTRIE2_PREV(%s)(U+%04lx): 0x%lx instead of 0x%lx\n",
 339                     testName, (long)c, (long)value, (long)values[i]);
 340         }
 341         if(c!=c2) {
 342             log_err("error: wrong code point from UTRIE2_PREV(%s): U+%04lx != U+%04lx\n",
 343                     testName, c, c2);
 344         }
 345     }
 346 }
 347
 348 static void
 349 testTrieUTF8(const char *testName,
 350              const UTrie2 *trie, UTrie2ValueBits valueBits,
 351              const CheckRange checkRanges[], int32_t countCheckRanges) {
 352     static const uint8_t illegal[]={
 353         0xc0, 0x80,                         /* non-shortest U+0000 */
 354         0xc1, 0xbf,                         /* non-shortest U+007f */
 355         0xc2,                               /* truncated */
 356         0xe0, 0x90, 0x80,                   /* non-shortest U+0400 */
 357         0xe0, 0xa0,                         /* truncated */
 358         0xed, 0xa0, 0x80,                   /* lead surrogate U+d800 */
 359         0xed, 0xbf, 0xbf,                   /* trail surrogate U+dfff */
 360         0xf0, 0x8f, 0xbf, 0xbf,             /* non-shortest U+ffff */
 361         0xf0, 0x90, 0x80,                   /* truncated */
 362         0xf4, 0x90, 0x80, 0x80,             /* beyond-Unicode U+110000 */
 363         0xf8, 0x80, 0x80, 0x80,             /* truncated */
 364         0xf8, 0x80, 0x80, 0x80, 0x80,       /* 5-byte UTF-8 */
 365         0xfd, 0xbf, 0xbf, 0xbf, 0xbf,       /* truncated */
 366         0xfd, 0xbf, 0xbf, 0xbf, 0xbf, 0xbf, /* 6-byte UTF-8 */
 367         0xfe,
 368         0xff
 369     };
 370     uint8_t s[600];
 371     uint32_t values[200];
 372
 373     const uint8_t *p, *limit;
 374
 375     uint32_t initialValue, errorValue;
 376     uint32_t value, bytes;
 377     UChar32 prevCP, c;
 378     int32_t i, countSpecials, length, countValues;
 379     int32_t prev8, i8;
 380
 381     countSpecials=getSpecialValues(checkRanges, countCheckRanges, &initialValue, &errorValue);
 382
 383     /* write a string */
 384     prevCP=0;
 385     length=countValues=0;
 386     /* first a couple of trail bytes in lead position */
 387     s[length++]=0x80;
 388     values[countValues++]=errorValue;
 389     s[length++]=0xbf;
 390     values[countValues++]=errorValue;
 391     prev8=i8=0;
 392     for(i=countSpecials; i<countCheckRanges; ++i) {
 393         value=checkRanges[i].value;
 394         /* write three legal (or surrogate) code points */
 395         U8_APPEND_UNSAFE(s, length, prevCP);    /* start of the range */
 396         values[countValues++]=U_IS_SURROGATE(prevCP) ? errorValue : value;
 397         c=checkRanges[i].limit;
 398         prevCP=(prevCP+c)/2;                    /* middle of the range */
 399         U8_APPEND_UNSAFE(s, length, prevCP);
 400         values[countValues++]=U_IS_SURROGATE(prevCP) ? errorValue : value;
 401         prevCP=c;
 402         --c;                                    /* end of the range */
 403         U8_APPEND_UNSAFE(s, length, c);
 404         values[countValues++]=U_IS_SURROGATE(c) ? errorValue : value;
 405         /* write an illegal byte sequence */
 406         if(i8<sizeof(illegal)) {
 407             U8_FWD_1(illegal, i8, sizeof(illegal));
 408             while(prev8<i8) {
 409                 s[length++]=illegal[prev8++];
 410             }
 411             values[countValues++]=errorValue;
 412         }
 413     }
 414     /* write the remaining illegal byte sequences */
 415     while(i8<sizeof(illegal)) {
 416         U8_FWD_1(illegal, i8, sizeof(illegal));
 417         while(prev8<i8) {
 418             s[length++]=illegal[prev8++];
 419         }
 420         values[countValues++]=errorValue;
 421     }
 422     limit=s+length;
 423
 424     /* try forward */
 425     p=s;
 426     i=0;
 427     while(p<limit) {
 428         prev8=i8=(int32_t)(p-s);
 429         U8_NEXT(s, i8, length, c);
 430         if(valueBits==UTRIE2_16_VALUE_BITS) {
 431             UTRIE2_U8_NEXT16(trie, p, limit, value);
 432         } else {
 433             UTRIE2_U8_NEXT32(trie, p, limit, value);
 434         }
 435         bytes=0;
 436         if(value!=values[i] || i8!=(p-s)) {
 437             while(prev8<i8) {
 438                 bytes=(bytes<<8)|s[prev8++];
 439             }
 440         }
 441         if(value!=values[i]) {
 442             log_err("error: wrong value from UTRIE2_U8_NEXT(%s)(%lx->U+%04lx): 0x%lx instead of 0x%lx\n",
 443                     testName, (unsigned long)bytes, (long)c, (long)value, (long)values[i]);
 444         }
 445         if(i8!=(p-s)) {
 446             log_err("error: wrong end index from UTRIE2_U8_NEXT(%s)(%lx->U+%04lx): %ld != %ld\n",
 447                     testName, (unsigned long)bytes, (long)c, (long)(p-s), (long)i8);
 448             continue;
 449         }
 450         ++i;
 451     }
 452
 453     /* try backward */
 454     p=limit;
 455     i=countValues;
 456     while(s<p) {
 457         --i;
 458         prev8=i8=(int32_t)(p-s);
 459         U8_PREV(s, 0, i8, c);
 460         if(valueBits==UTRIE2_16_VALUE_BITS) {
 461             UTRIE2_U8_PREV16(trie, s, p, value);
 462         } else {
 463             UTRIE2_U8_PREV32(trie, s, p, value);
 464         }
 465         bytes=0;
 466         if(value!=values[i] || i8!=(p-s)) {
 467             int32_t k=i8;
 468             while(k<prev8) {
 469                 bytes=(bytes<<8)|s[k++];
 470             }
 471         }
 472         if(value!=values[i]) {
 473             log_err("error: wrong value from UTRIE2_U8_PREV(%s)(%lx->U+%04lx): 0x%lx instead of 0x%lx\n",
 474                     testName, (unsigned long)bytes, (long)c, (long)value, (long)values[i]);
 475         }
 476         if(i8!=(p-s)) {
 477             log_err("error: wrong end index from UTRIE2_U8_PREV(%s)(%lx->U+%04lx): %ld != %ld\n",
 478                     testName, (unsigned long)bytes, (long)c, (long)(p-s), (long)i8);
 479             continue;
 480         }
 481     }
 482 }
 483
 484 static void
 485 testFrozenTrie(const char *testName,
 486                UTrie2 *trie, UTrie2ValueBits valueBits,
 487                const CheckRange checkRanges[], int32_t countCheckRanges) {
 488     UErrorCode errorCode;
 489     uint32_t value, value2;
 490
 491     if(!utrie2_isFrozen(trie)) {
 492         log_err("error: utrie2_isFrozen(frozen %s) returned FALSE (not frozen)\n",
 493                 testName);
 494         return;
 495     }
 496
 497     testTrieGetters(testName, trie, valueBits, checkRanges, countCheckRanges);
 498     testTrieEnum(testName, trie, checkRanges, countCheckRanges);
 499     testTrieUTF16(testName, trie, valueBits, checkRanges, countCheckRanges);
 500     testTrieUTF8(testName, trie, valueBits, checkRanges, countCheckRanges);
 501
 502     errorCode=U_ZERO_ERROR;
 503     value=utrie2_get32(trie, 1);
 504     utrie2_set32(trie, 1, 234, &errorCode);
 505     value2=utrie2_get32(trie, 1);
 506     if(errorCode!=U_NO_WRITE_PERMISSION || value2!=value) {
 507         log_err("error: utrie2_set32(frozen %s) failed: it set %s != U_NO_WRITE_PERMISSION\n",
 508                 testName, u_errorName(errorCode));
 509         return;
 510     }
 511
 512     errorCode=U_ZERO_ERROR;
 513     utrie2_setRange32(trie, 1, 5, 234, TRUE, &errorCode);
 514     value2=utrie2_get32(trie, 1);
 515     if(errorCode!=U_NO_WRITE_PERMISSION || value2!=value) {
 516         log_err("error: utrie2_setRange32(frozen %s) failed: it set %s != U_NO_WRITE_PERMISSION\n",
 517                 testName, u_errorName(errorCode));
 518         return;
 519     }
 520
 521     errorCode=U_ZERO_ERROR;
 522     value=utrie2_get32FromLeadSurrogateCodeUnit(trie, 0xd801);
 523     utrie2_set32ForLeadSurrogateCodeUnit(trie, 0xd801, 234, &errorCode);
 524     value2=utrie2_get32FromLeadSurrogateCodeUnit(trie, 0xd801);
 525     if(errorCode!=U_NO_WRITE_PERMISSION || value2!=value) {
 526         log_err("error: utrie2_set32ForLeadSurrogateCodeUnit(frozen %s) failed: "
 527                 "it set %s != U_NO_WRITE_PERMISSION\n",
 528                 testName, u_errorName(errorCode));
 529         return;
 530     }
 531 }
 532
 533 static void
 534 testNewTrie(const char *testName, const UTrie2 *trie,
 535             const CheckRange checkRanges[], int32_t countCheckRanges) {
 536     /* The valueBits are ignored for an unfrozen trie. */
 537     testTrieGetters(testName, trie, UTRIE2_COUNT_VALUE_BITS, checkRanges, countCheckRanges);
 538     testTrieEnum(testName, trie, checkRanges, countCheckRanges);
 539 }
 540
 541 static void
 542 testTrieSerialize(const char *testName,
 543                   UTrie2 *trie, UTrie2ValueBits valueBits,
 544                   UBool withSwap,
 545                   const CheckRange checkRanges[], int32_t countCheckRanges) {
 546     uint32_t storage[10000];
 547     int32_t length1, length2, length3;
 548     UTrie2ValueBits otherValueBits;
 549     UErrorCode errorCode;
 550
 551     /* clone the trie so that the caller can reuse the original */
 552     errorCode=U_ZERO_ERROR;
 553     trie=utrie2_clone(trie, &errorCode);
 554     if(U_FAILURE(errorCode)) {
 555         log_err("error: utrie2_clone(unfrozen %s) failed - %s\n",
 556                 testName, u_errorName(errorCode));
 557         return;
 558     }
 559
 560     /*
 561      * This is not a loop, but simply a block that we can exit with "break"
 562      * when something goes wrong.
 563      */
 564     do {
 565         errorCode=U_ZERO_ERROR;
 566         utrie2_serialize(trie, storage, sizeof(storage), &errorCode);
 567         if(errorCode!=U_ILLEGAL_ARGUMENT_ERROR) {
 568             log_err("error: utrie2_serialize(unfrozen %s) set %s != U_ILLEGAL_ARGUMENT_ERROR\n",
 569                     testName, u_errorName(errorCode));
 570             break;
 571         }
 572         errorCode=U_ZERO_ERROR;
 573         utrie2_freeze(trie, valueBits, &errorCode);
 574         if(U_FAILURE(errorCode) || !utrie2_isFrozen(trie)) {
 575             log_err("error: utrie2_freeze(%s) failed: %s isFrozen: %d\n",
 576                     testName, u_errorName(errorCode), utrie2_isFrozen(trie));
 577             break;
 578         }
 579         otherValueBits= valueBits==UTRIE2_16_VALUE_BITS ? UTRIE2_32_VALUE_BITS : UTRIE2_16_VALUE_BITS;
 580         utrie2_freeze(trie, otherValueBits, &errorCode);
 581         if(errorCode!=U_ILLEGAL_ARGUMENT_ERROR) {
 582             log_err("error: utrie2_freeze(already-frozen with other valueBits %s) "
 583                     "set %s != U_ILLEGAL_ARGUMENT_ERROR\n",
 584                     testName, u_errorName(errorCode));
 585             break;
 586         }
 587         errorCode=U_ZERO_ERROR;
 588         if(withSwap) {
 589             /* clone a frozen trie */
 590             UTrie2 *clone=utrie2_clone(trie, &errorCode);
 591             if(U_FAILURE(errorCode)) {
 592                 log_err("error: cloning a frozen UTrie2 failed (%s) - %s\n",
 593                         testName, u_errorName(errorCode));
 594                 errorCode=U_ZERO_ERROR;  /* continue with the original */
 595             } else {
 596                 utrie2_close(trie);
 597                 trie=clone;
 598             }
 599         }
 600         length1=utrie2_serialize(trie, NULL, 0, &errorCode);
 601         if(errorCode!=U_BUFFER_OVERFLOW_ERROR) {
 602             log_err("error: utrie2_serialize(%s) preflighting set %s != U_BUFFER_OVERFLOW_ERROR\n",
 603                     testName, u_errorName(errorCode));
 604             break;
 605         }
 606         errorCode=U_ZERO_ERROR;
 607         length2=utrie2_serialize(trie, storage, sizeof(storage), &errorCode);
 608         if(errorCode==U_BUFFER_OVERFLOW_ERROR) {
 609             log_err("error: utrie2_serialize(%s) needs more memory\n", testName);
 610             break;
 611         }
 612         if(U_FAILURE(errorCode)) {
 613             log_err("error: utrie2_serialize(%s) failed: %s\n", testName, u_errorName(errorCode));
 614             break;
 615         }
 616         if(length1!=length2) {
 617             log_err("error: trie serialization (%s) lengths different: "
 618                     "preflight vs. serialize\n", testName);
 619             break;
 620         }
 621
 622         testFrozenTrie(testName, trie, valueBits, checkRanges, countCheckRanges);
 623         utrie2_close(trie);
 624         trie=NULL;
 625
 626         if(withSwap) {
 627             uint32_t swapped[10000];
 628             int32_t swappedLength;
 629
 630             UDataSwapper *ds;
 631
 632             /* swap to opposite-endian */
 633             uprv_memset(swapped, 0x55, length2);
 634             ds=udata_openSwapper(U_IS_BIG_ENDIAN, U_CHARSET_FAMILY,
 635                                  !U_IS_BIG_ENDIAN, U_CHARSET_FAMILY, &errorCode);
 636             swappedLength=utrie2_swap(ds, storage, -1, NULL, &errorCode);
 637             if(U_FAILURE(errorCode) || swappedLength!=length2) {
 638                 log_err("error: utrie2_swap(%s to OE preflighting) failed (%s) "
 639                         "or before/after lengths different\n",
 640                         testName, u_errorName(errorCode));
 641                 udata_closeSwapper(ds);
 642                 break;
 643             }
 644             swappedLength=utrie2_swap(ds, storage, length2, swapped, &errorCode);
 645             udata_closeSwapper(ds);
 646             if(U_FAILURE(errorCode) || swappedLength!=length2) {
 647                 log_err("error: utrie2_swap(%s to OE) failed (%s) or before/after lengths different\n",
 648                         testName, u_errorName(errorCode));
 649                 break;
 650             }
 651
 652             /* swap back to platform-endian */
 653             uprv_memset(storage, 0xaa, length2);
 654             ds=udata_openSwapper(!U_IS_BIG_ENDIAN, U_CHARSET_FAMILY,
 655                                  U_IS_BIG_ENDIAN, U_CHARSET_FAMILY, &errorCode);
 656             swappedLength=utrie2_swap(ds, swapped, -1, NULL, &errorCode);
 657             if(U_FAILURE(errorCode) || swappedLength!=length2) {
 658                 log_err("error: utrie2_swap(%s to PE preflighting) failed (%s) "
 659                         "or before/after lengths different\n",
 660                         testName, u_errorName(errorCode));
 661                 udata_closeSwapper(ds);
 662                 break;
 663             }
 664             swappedLength=utrie2_swap(ds, swapped, length2, storage, &errorCode);
 665             udata_closeSwapper(ds);
 666             if(U_FAILURE(errorCode) || swappedLength!=length2) {
 667                 log_err("error: utrie2_swap(%s to PE) failed (%s) or before/after lengths different\n",
 668                         testName, u_errorName(errorCode));
 669                 break;
 670             }
 671         }
 672
 673         trie=utrie2_openFromSerialized(valueBits, storage, length2, &length3, &errorCode);
 674         if(U_FAILURE(errorCode)) {
 675             log_err("error: utrie2_openFromSerialized(%s) failed, %s\n", testName, u_errorName(errorCode));
 676             break;
 677         }
 678         if((valueBits==UTRIE2_16_VALUE_BITS)!=(trie->data32==NULL)) {
 679             log_err("error: trie serialization (%s) did not preserve 32-bitness\n", testName);
 680             break;
 681         }
 682         if(length2!=length3) {
 683             log_err("error: trie serialization (%s) lengths different: "
 684                     "serialize vs. unserialize\n", testName);
 685             break;
 686         }
 687         /* overwrite the storage that is not supposed to be needed */
 688         uprv_memset((char *)storage+length3, 0xfa, (int32_t)(sizeof(storage)-length3));
 689
 690         utrie2_freeze(trie, valueBits, &errorCode);
 691         if(U_FAILURE(errorCode) || !utrie2_isFrozen(trie)) {
 692             log_err("error: utrie2_freeze(unserialized %s) failed: %s isFrozen: %d\n",
 693                     testName, u_errorName(errorCode), utrie2_isFrozen(trie));
 694             break;
 695         }
 696         utrie2_freeze(trie, otherValueBits, &errorCode);
 697         if(errorCode!=U_ILLEGAL_ARGUMENT_ERROR) {
 698             log_err("error: utrie2_freeze(unserialized with other valueBits %s) "
 699                     "set %s != U_ILLEGAL_ARGUMENT_ERROR\n",
 700                     testName, u_errorName(errorCode));
 701             break;
 702         }
 703         errorCode=U_ZERO_ERROR;
 704         if(withSwap) {
 705             /* clone an unserialized trie */
 706             UTrie2 *clone=utrie2_clone(trie, &errorCode);
 707             if(U_FAILURE(errorCode)) {
 708                 log_err("error: utrie2_clone(unserialized %s) failed - %s\n",
 709                         testName, u_errorName(errorCode));
 710                 errorCode=U_ZERO_ERROR;
 711                 /* no need to break: just test the original trie */
 712             } else {
 713                 utrie2_close(trie);
 714                 trie=clone;
 715                 uprv_memset(storage, 0, sizeof(storage));
 716             }
 717         }
 718         testFrozenTrie(testName, trie, valueBits, checkRanges, countCheckRanges);
 719         {
 720             /* clone-as-thawed an unserialized trie */
 721             UTrie2 *clone=utrie2_cloneAsThawed(trie, &errorCode);
 722             if(U_FAILURE(errorCode) || utrie2_isFrozen(clone)) {
 723                 log_err("error: utrie2_cloneAsThawed(unserialized %s) failed - "
 724                         "%s (isFrozen: %d)\n",
 725                         testName, u_errorName(errorCode), clone!=NULL && utrie2_isFrozen(trie));
 726                 break;
 727             } else {
 728                 utrie2_close(trie);
 729                 trie=clone;
 730             }
 731         }
 732         {
 733             uint32_t value, value2;
 734
 735             value=utrie2_get32(trie, 0xa1);
 736             utrie2_set32(trie, 0xa1, 789, &errorCode);
 737             value2=utrie2_get32(trie, 0xa1);
 738             utrie2_set32(trie, 0xa1, value, &errorCode);
 739             if(U_FAILURE(errorCode) || value2!=789) {
 740                 log_err("error: modifying a cloneAsThawed UTrie2 (%s) failed - %s\n",
 741                         testName, u_errorName(errorCode));
 742             }
 743         }
 744         testNewTrie(testName, trie, checkRanges, countCheckRanges);
 745     } while(0);
 746
 747     utrie2_close(trie);
 748 }
 749
 750 static UTrie2 *
 751 testTrieSerializeAllValueBits(const char *testName,
 752                               UTrie2 *trie, UBool withClone,
 753                               const CheckRange checkRanges[], int32_t countCheckRanges) {
 754     char name[40];
 755
 756     /* verify that all the expected values are in the unfrozen trie */
 757     testNewTrie(testName, trie, checkRanges, countCheckRanges);
 758
 759     /*
 760      * Test with both valueBits serializations,
 761      * and that utrie2_serialize() can be called multiple times.
 762      */
 763     uprv_strcpy(name, testName);
 764     uprv_strcat(name, ".16");
 765     testTrieSerialize(name, trie,
 766                       UTRIE2_16_VALUE_BITS, withClone,
 767                       checkRanges, countCheckRanges);
 768
 769     if(withClone) {
 770         /*
 771          * try cloning after the first serialization;
 772          * clone-as-thawed just to sometimes try it on an unfrozen trie
 773          */
 774         UErrorCode errorCode=U_ZERO_ERROR;
 775         UTrie2 *clone=utrie2_cloneAsThawed(trie, &errorCode);
 776         if(U_FAILURE(errorCode)) {
 777             log_err("error: utrie2_cloneAsThawed(%s) after serialization failed - %s\n",
 778                     testName, u_errorName(errorCode));
 779         } else {
 780             utrie2_close(trie);
 781             trie=clone;
 782
 783             testNewTrie(testName, trie, checkRanges, countCheckRanges);
 784         }
 785     }
 786
 787     uprv_strcpy(name, testName);
 788     uprv_strcat(name, ".32");
 789     testTrieSerialize(name, trie,
 790                       UTRIE2_32_VALUE_BITS, withClone,
 791                       checkRanges, countCheckRanges);
 792
 793     return trie; /* could be the clone */
 794 }
 795
 796 static UTrie2 *
 797 makeTrieWithRanges(const char *testName, UBool withClone,
 798                    const SetRange setRanges[], int32_t countSetRanges,
 799                    const CheckRange checkRanges[], int32_t countCheckRanges) {
 800     UTrie2 *trie;
 801     uint32_t initialValue, errorValue;
 802     uint32_t value;
 803     UChar32 start, limit;
 804     int32_t i;
 805     UErrorCode errorCode;
 806     UBool overwrite;
 807
 808     log_verbose("\ntesting Trie '%s'\n", testName);
 809     errorCode=U_ZERO_ERROR;
 810     getSpecialValues(checkRanges, countCheckRanges, &initialValue, &errorValue);
 811     trie=utrie2_open(initialValue, errorValue, &errorCode);
 812     if(U_FAILURE(errorCode)) {
 813         log_err("error: utrie2_open(%s) failed: %s\n", testName, u_errorName(errorCode));
 814         return NULL;
 815     }
 816
 817     /* set values from setRanges[] */
 818     for(i=0; i<countSetRanges; ++i) {
 819         if(withClone && i==countSetRanges/2) {
 820             /* switch to a clone in the middle of setting values */
 821             UTrie2 *clone=utrie2_clone(trie, &errorCode);
 822             if(U_FAILURE(errorCode)) {
 823                 log_err("error: utrie2_clone(%s) failed - %s\n",
 824                         testName, u_errorName(errorCode));
 825                 errorCode=U_ZERO_ERROR;  /* continue with the original */
 826             } else {
 827                 utrie2_close(trie);
 828                 trie=clone;
 829             }
 830         }
 831         start=setRanges[i].start;
 832         limit=setRanges[i].limit;
 833         value=setRanges[i].value;
 834         overwrite=setRanges[i].overwrite;
 835         if((limit-start)==1 && overwrite) {
 836             utrie2_set32(trie, start, value, &errorCode);
 837         } else {
 838             utrie2_setRange32(trie, start, limit-1, value, overwrite, &errorCode);
 839         }
 840     }
 841
 842     /* set some values for lead surrogate code units */
 843     utrie2_set32ForLeadSurrogateCodeUnit(trie, 0xd800, 90, &errorCode);
 844     utrie2_set32ForLeadSurrogateCodeUnit(trie, 0xd999, 94, &errorCode);
 845     utrie2_set32ForLeadSurrogateCodeUnit(trie, 0xdbff, 99, &errorCode);
 846     if(U_SUCCESS(errorCode)) {
 847         return trie;
 848     } else {
 849         log_err("error: setting values into a trie (%s) failed - %s\n",
 850                 testName, u_errorName(errorCode));
 851         utrie2_close(trie);
 852         return NULL;
 853     }
 854 }
 855
 856 static void
 857 testTrieRanges(const char *testName, UBool withClone,
 858                const SetRange setRanges[], int32_t countSetRanges,
 859                const CheckRange checkRanges[], int32_t countCheckRanges) {
 860     UTrie2 *trie=makeTrieWithRanges(testName, withClone,
 861                                     setRanges, countSetRanges,
 862                                     checkRanges, countCheckRanges);
 863     if(trie!=NULL) {
 864         trie=testTrieSerializeAllValueBits(testName, trie, withClone,
 865                                            checkRanges, countCheckRanges);
 866         utrie2_close(trie);
 867     }
 868 }
 869
 870 /* test data ----------------------------------------------------------------*/
 871
 872 /* set consecutive ranges, even with value 0 */
 873 static const SetRange
 874 setRanges1[]={
 875     { 0,        0x40,     0,      FALSE },
 876     { 0x40,     0xe7,     0x1234, FALSE },
 877     { 0xe7,     0x3400,   0,      FALSE },
 878     { 0x3400,   0x9fa6,   0x6162, FALSE },
 879     { 0x9fa6,   0xda9e,   0x3132, FALSE },
 880     { 0xdada,   0xeeee,   0x87ff, FALSE },
 881     { 0xeeee,   0x11111,  1,      FALSE },
 882     { 0x11111,  0x44444,  0x6162, FALSE },
 883     { 0x44444,  0x60003,  0,      FALSE },
 884     { 0xf0003,  0xf0004,  0xf,    FALSE },
 885     { 0xf0004,  0xf0006,  0x10,   FALSE },
 886     { 0xf0006,  0xf0007,  0x11,   FALSE },
 887     { 0xf0007,  0xf0040,  0x12,   FALSE },
 888     { 0xf0040,  0x110000, 0,      FALSE }
 889 };
 890
 891 static const CheckRange
 892 checkRanges1[]={
 893     { 0,        0 },
 894     { 0x40,     0 },
 895     { 0xe7,     0x1234 },
 896     { 0x3400,   0 },
 897     { 0x9fa6,   0x6162 },
 898     { 0xda9e,   0x3132 },
 899     { 0xdada,   0 },
 900     { 0xeeee,   0x87ff },
 901     { 0x11111,  1 },
 902     { 0x44444,  0x6162 },
 903     { 0xf0003,  0 },
 904     { 0xf0004,  0xf },
 905     { 0xf0006,  0x10 },
 906     { 0xf0007,  0x11 },
 907     { 0xf0040,  0x12 },
 908     { 0x110000, 0 }
 909 };
 910
 911 /* set some interesting overlapping ranges */
 912 static const SetRange
 913 setRanges2[]={
 914     { 0x21,     0x7f,     0x5555, TRUE },
 915     { 0x2f800,  0x2fedc,  0x7a,   TRUE },
 916     { 0x72,     0xdd,     3,      TRUE },
 917     { 0xdd,     0xde,     4,      FALSE },
 918     { 0x201,    0x240,    6,      TRUE },  /* 3 consecutive blocks with the same pattern but */
 919     { 0x241,    0x280,    6,      TRUE },  /* discontiguous value ranges, testing utrie2_enum() */
 920     { 0x281,    0x2c0,    6,      TRUE },
 921     { 0x2f987,  0x2fa98,  5,      TRUE },
 922     { 0x2f777,  0x2f883,  0,      TRUE },
 923     { 0x2f900,  0x2ffaa,  1,      FALSE },
 924     { 0x2ffaa,  0x2ffab,  2,      TRUE },
 925     { 0x2ffbb,  0x2ffc0,  7,      TRUE }
 926 };
 927
 928 static const CheckRange
 929 checkRanges2[]={
 930     { 0,        0 },
 931     { 0x21,     0 },
 932     { 0x72,     0x5555 },
 933     { 0xdd,     3 },
 934     { 0xde,     4 },
 935     { 0x201,    0 },
 936     { 0x240,    6 },
 937     { 0x241,    0 },
 938     { 0x280,    6 },
 939     { 0x281,    0 },
 940     { 0x2c0,    6 },
 941     { 0x2f883,  0 },
 942     { 0x2f987,  0x7a },
 943     { 0x2fa98,  5 },
 944     { 0x2fedc,  0x7a },
 945     { 0x2ffaa,  1 },
 946     { 0x2ffab,  2 },
 947     { 0x2ffbb,  0 },
 948     { 0x2ffc0,  7 },
 949     { 0x110000, 0 }
 950 };
 951
 952 static const CheckRange
 953 checkRanges2_d800[]={
 954     { 0x10000,  0 },
 955     { 0x10400,  0 }
 956 };
 957
 958 static const CheckRange
 959 checkRanges2_d87e[]={
 960     { 0x2f800,  6 },
 961     { 0x2f883,  0 },
 962     { 0x2f987,  0x7a },
 963     { 0x2fa98,  5 },
 964     { 0x2fc00,  0x7a }
 965 };
 966
 967 static const CheckRange
 968 checkRanges2_d87f[]={
 969     { 0x2fc00,  0 },
 970     { 0x2fedc,  0x7a },
 971     { 0x2ffaa,  1 },
 972     { 0x2ffab,  2 },
 973     { 0x2ffbb,  0 },
 974     { 0x2ffc0,  7 },
 975     { 0x30000,  0 }
 976 };
 977
 978 static const CheckRange
 979 checkRanges2_dbff[]={
 980     { 0x10fc00, 0 },
 981     { 0x110000, 0 }
 982 };
 983
 984 /* use a non-zero initial value */
 985 static const SetRange
 986 setRanges3[]={
 987     { 0x31,     0xa4,     1, FALSE },
 988     { 0x3400,   0x6789,   2, FALSE },
 989     { 0x8000,   0x89ab,   9, TRUE },
 990     { 0x9000,   0xa000,   4, TRUE },
 991     { 0xabcd,   0xbcde,   3, TRUE },
 992     { 0x55555,  0x110000, 6, TRUE },  /* highStart<U+ffff with non-initialValue */
 993     { 0xcccc,   0x55555,  6, TRUE }
 994 };
 995
 996 static const CheckRange
 997 checkRanges3[]={
 998     { 0,        9 },  /* non-zero initialValue */
 999     { 0x31,     9 },
1000     { 0xa4,     1 },
1001     { 0x3400,   9 },
1002     { 0x6789,   2 },
1003     { 0x9000,   9 },
1004     { 0xa000,   4 },
1005     { 0xabcd,   9 },
1006     { 0xbcde,   3 },
1007     { 0xcccc,   9 },
1008     { 0x110000, 6 }
1009 };
1010
1011 /* empty or single-value tries, testing highStart==0 */
1012 static const SetRange
1013 setRangesEmpty[]={
1014     { 0,        0,        0, FALSE },  /* need some values for it to compile */
1015 };
1016
1017 static const CheckRange
1018 checkRangesEmpty[]={
1019     { 0,        3 },
1020     { 0x110000, 3 }
1021 };
1022
1023 static const SetRange
1024 setRangesSingleValue[]={
1025     { 0,        0x110000, 5, TRUE },
1026 };
1027
1028 static const CheckRange
1029 checkRangesSingleValue[]={
1030     { 0,        3 },
1031     { 0x110000, 5 }
1032 };
1033
1034 static void
1035 TrieTest(void) {
1036     testTrieRanges("set1", FALSE,
1037         setRanges1, LENGTHOF(setRanges1),
1038         checkRanges1, LENGTHOF(checkRanges1));
1039     testTrieRanges("set2-overlap", FALSE,
1040         setRanges2, LENGTHOF(setRanges2),
1041         checkRanges2, LENGTHOF(checkRanges2));
1042     testTrieRanges("set3-initial-9", FALSE,
1043         setRanges3, LENGTHOF(setRanges3),
1044         checkRanges3, LENGTHOF(checkRanges3));
1045     testTrieRanges("set-empty", FALSE,
1046         setRangesEmpty, 0,
1047         checkRangesEmpty, LENGTHOF(checkRangesEmpty));
1048     testTrieRanges("set-single-value", FALSE,
1049         setRangesSingleValue, LENGTHOF(setRangesSingleValue),
1050         checkRangesSingleValue, LENGTHOF(checkRangesSingleValue));
1051
1052     testTrieRanges("set2-overlap.withClone", TRUE,
1053         setRanges2, LENGTHOF(setRanges2),
1054         checkRanges2, LENGTHOF(checkRanges2));
1055 }
1056
1057 static void
1058 EnumNewTrieForLeadSurrogateTest(void) {
1059     static const char *const testName="enum-for-lead";
1060     UTrie2 *trie=makeTrieWithRanges(testName, FALSE,
1061                                     setRanges2, LENGTHOF(setRanges2),
1062                                     checkRanges2, LENGTHOF(checkRanges2));
1063     while(trie!=NULL) {
1064         const CheckRange *checkRanges;
1065
1066         checkRanges=checkRanges2_d800+1;
1067         utrie2_enumForLeadSurrogate(trie, 0xd800,
1068                                     testEnumValue, testEnumRange,
1069                                     &checkRanges);
1070         checkRanges=checkRanges2_d87e+1;
1071         utrie2_enumForLeadSurrogate(trie, 0xd87e,
1072                                     testEnumValue, testEnumRange,
1073                                     &checkRanges);
1074         checkRanges=checkRanges2_d87f+1;
1075         utrie2_enumForLeadSurrogate(trie, 0xd87f,
1076                                     testEnumValue, testEnumRange,
1077                                     &checkRanges);
1078         checkRanges=checkRanges2_dbff+1;
1079         utrie2_enumForLeadSurrogate(trie, 0xdbff,
1080                                     testEnumValue, testEnumRange,
1081                                     &checkRanges);
1082         if(!utrie2_isFrozen(trie)) {
1083             UErrorCode errorCode=U_ZERO_ERROR;
1084             utrie2_freeze(trie, UTRIE2_16_VALUE_BITS, &errorCode);
1085             if(U_FAILURE(errorCode)) {
1086                 log_err("error: utrie2_freeze(%s) failed\n", testName);
1087                 utrie2_close(trie);
1088                 return;
1089             }
1090         } else {
1091             utrie2_close(trie);
1092             break;
1093         }
1094     }
1095 }
1096
1097 /* test utrie2_openDummy() -------------------------------------------------- */
1098
1099 static void
1100 dummyTest(UTrie2ValueBits valueBits) {
1101     CheckRange
1102     checkRanges[]={
1103         { -1,       0 },
1104         { 0,        0 },
1105         { 0x110000, 0 }
1106     };
1107
1108     UTrie2 *trie;
1109     UErrorCode errorCode;
1110
1111     const char *testName;
1112     uint32_t initialValue, errorValue;
1113
1114     if(valueBits==UTRIE2_16_VALUE_BITS) {
1115         testName="dummy.16";
1116         initialValue=0x313;
1117         errorValue=0xaffe;
1118     } else {
1119         testName="dummy.32";
1120         initialValue=0x01234567;
1121         errorValue=0x89abcdef;
1122     }
1123     checkRanges[0].value=errorValue;
1124     checkRanges[1].value=checkRanges[2].value=initialValue;
1125
1126     errorCode=U_ZERO_ERROR;
1127     trie=utrie2_openDummy(valueBits, initialValue, errorValue, &errorCode);
1128     if(U_FAILURE(errorCode)) {
1129         log_err("utrie2_openDummy(valueBits=%d) failed - %s\n", valueBits, u_errorName(errorCode));
1130         return;
1131     }
1132
1133     testFrozenTrie(testName, trie, valueBits, checkRanges, LENGTHOF(checkRanges));
1134     utrie2_close(trie);
1135 }
1136
1137 static void
1138 DummyTrieTest(void) {
1139     dummyTest(UTRIE2_16_VALUE_BITS);
1140     dummyTest(UTRIE2_32_VALUE_BITS);
1141 }
1142
1143 /* test builder memory management ------------------------------------------- */
1144
1145 static void
1146 FreeBlocksTest(void) {
1147     static const CheckRange
1148     checkRanges[]={
1149         { 0,        1 },
1150         { 0x740,    1 },
1151         { 0x780,    2 },
1152         { 0x880,    3 },
1153         { 0x110000, 1 }
1154     };
1155     static const char *const testName="free-blocks";
1156
1157     UTrie2 *trie;
1158     int32_t i;
1159     UErrorCode errorCode;
1160
1161     errorCode=U_ZERO_ERROR;
1162     trie=utrie2_open(1, 0xbad, &errorCode);
1163     if(U_FAILURE(errorCode)) {
1164         log_err("error: utrie2_open(%s) failed: %s\n", testName, u_errorName(errorCode));
1165         return;
1166     }
1167
1168     /*
1169      * Repeatedly set overlapping same-value ranges to stress the free-data-block management.
1170      * If it fails, it will overflow the data array.
1171      */
1172     for(i=0; i<(0x120000>>UTRIE2_SHIFT_2)/2; ++i) {
1173         utrie2_setRange32(trie, 0x740, 0x840-1, 1, TRUE, &errorCode);
1174         utrie2_setRange32(trie, 0x780, 0x880-1, 1, TRUE, &errorCode);
1175         utrie2_setRange32(trie, 0x740, 0x840-1, 2, TRUE, &errorCode);
1176         utrie2_setRange32(trie, 0x780, 0x880-1, 3, TRUE, &errorCode);
1177     }
1178     /* make blocks that will be free during compaction */
1179     utrie2_setRange32(trie, 0x1000, 0x3000-1, 2, TRUE, &errorCode);
1180     utrie2_setRange32(trie, 0x2000, 0x4000-1, 3, TRUE, &errorCode);
1181     utrie2_setRange32(trie, 0x1000, 0x4000-1, 1, TRUE, &errorCode);
1182     /* set some values for lead surrogate code units */
1183     utrie2_set32ForLeadSurrogateCodeUnit(trie, 0xd800, 90, &errorCode);
1184     utrie2_set32ForLeadSurrogateCodeUnit(trie, 0xd999, 94, &errorCode);
1185     utrie2_set32ForLeadSurrogateCodeUnit(trie, 0xdbff, 99, &errorCode);
1186     if(U_FAILURE(errorCode)) {
1187         log_err("error: setting lots of ranges into a trie (%s) failed - %s\n",
1188                 testName, u_errorName(errorCode));
1189         utrie2_close(trie);
1190         return;
1191     }
1192
1193     trie=testTrieSerializeAllValueBits(testName, trie, FALSE,
1194                                        checkRanges, LENGTHOF(checkRanges));
1195     utrie2_close(trie);
1196 }
1197
1198 static void
1199 GrowDataArrayTest(void) {
1200     static const CheckRange
1201     checkRanges[]={
1202         { 0,        1 },
1203         { 0x720,    2 },
1204         { 0x7a0,    3 },
1205         { 0x8a0,    4 },
1206         { 0x110000, 5 }
1207     };
1208     static const char *const testName="grow-data";
1209
1210     UTrie2 *trie;
1211     int32_t i;
1212     UErrorCode errorCode;
1213
1214     errorCode=U_ZERO_ERROR;
1215     trie=utrie2_open(1, 0xbad, &errorCode);
1216     if(U_FAILURE(errorCode)) {
1217         log_err("error: utrie2_open(%s) failed: %s\n", testName, u_errorName(errorCode));
1218         return;
1219     }
1220
1221     /*
1222      * Use utrie2_set32() not utrie2_setRange32() to write non-initialValue-data.
1223      * Should grow/reallocate the data array to a sufficient length.
1224      */
1225     for(i=0; i<0x1000; ++i) {
1226         utrie2_set32(trie, i, 2, &errorCode);
1227     }
1228     for(i=0x720; i<0x1100; ++i) { /* some overlap */
1229         utrie2_set32(trie, i, 3, &errorCode);
1230     }
1231     for(i=0x7a0; i<0x900; ++i) {
1232         utrie2_set32(trie, i, 4, &errorCode);
1233     }
1234     for(i=0x8a0; i<0x110000; ++i) {
1235         utrie2_set32(trie, i, 5, &errorCode);
1236     }
1237     for(i=0xd800; i<0xdc00; ++i) {
1238         utrie2_set32ForLeadSurrogateCodeUnit(trie, i, 1, &errorCode);
1239     }
1240     /* set some values for lead surrogate code units */
1241     utrie2_set32ForLeadSurrogateCodeUnit(trie, 0xd800, 90, &errorCode);
1242     utrie2_set32ForLeadSurrogateCodeUnit(trie, 0xd999, 94, &errorCode);
1243     utrie2_set32ForLeadSurrogateCodeUnit(trie, 0xdbff, 99, &errorCode);
1244     if(U_FAILURE(errorCode)) {
1245         log_err("error: setting lots of values into a trie (%s) failed - %s\n",
1246                 testName, u_errorName(errorCode));
1247         utrie2_close(trie);
1248         return;
1249     }
1250
1251     trie=testTrieSerializeAllValueBits(testName, trie, FALSE,
1252                                           checkRanges, LENGTHOF(checkRanges));
1253     utrie2_close(trie);
1254 }
1255
1256 /* versions 1 and 2 --------------------------------------------------------- */
1257
1258 static void
1259 GetVersionTest(void) {
1260     uint32_t data[4];
1261     if( /* version 1 */
1262         (data[0]=0x54726965, 1!=utrie2_getVersion(data, sizeof(data), FALSE)) ||
1263         (data[0]=0x54726965, 1!=utrie2_getVersion(data, sizeof(data), TRUE)) ||
1264         (data[0]=0x65697254, 0!=utrie2_getVersion(data, sizeof(data), FALSE)) ||
1265         (data[0]=0x65697254, 1!=utrie2_getVersion(data, sizeof(data), TRUE)) ||
1266         /* version 2 */
1267         (data[0]=0x54726932, 2!=utrie2_getVersion(data, sizeof(data), FALSE)) ||
1268         (data[0]=0x54726932, 2!=utrie2_getVersion(data, sizeof(data), TRUE)) ||
1269         (data[0]=0x32697254, 0!=utrie2_getVersion(data, sizeof(data), FALSE)) ||
1270         (data[0]=0x32697254, 2!=utrie2_getVersion(data, sizeof(data), TRUE)) ||
1271         /* illegal arguments */
1272         (data[0]=0x54726932, 0!=utrie2_getVersion(NULL, sizeof(data), FALSE)) ||
1273         (data[0]=0x54726932, 0!=utrie2_getVersion(data, 3, FALSE)) ||
1274         (data[0]=0x54726932, 0!=utrie2_getVersion((char *)data+1, sizeof(data), FALSE)) ||
1275         /* unknown signature values */
1276         (data[0]=0x11223344, 0!=utrie2_getVersion(data, sizeof(data), FALSE)) ||
1277         (data[0]=0x54726933, 0!=utrie2_getVersion(data, sizeof(data), FALSE))
1278     ) {
1279         log_err("error: utrie2_getVersion() is not working as expected\n");
1280     }
1281 }
1282
1283 static UNewTrie *
1284 makeNewTrie1WithRanges(const char *testName,
1285                        const SetRange setRanges[], int32_t countSetRanges,
1286                        const CheckRange checkRanges[], int32_t countCheckRanges) {
1287     UNewTrie *newTrie;
1288     uint32_t initialValue, errorValue;
1289     uint32_t value;
1290     UChar32 start, limit;
1291     int32_t i;
1292     UErrorCode errorCode;
1293     UBool overwrite, ok;
1294
1295     log_verbose("\ntesting Trie '%s'\n", testName);
1296     errorCode=U_ZERO_ERROR;
1297     getSpecialValues(checkRanges, countCheckRanges, &initialValue, &errorValue);
1298     newTrie=utrie_open(NULL, NULL, 2000,
1299                        initialValue, initialValue,
1300                        FALSE);
1301     if(U_FAILURE(errorCode)) {
1302         log_err("error: utrie_open(%s) failed: %s\n", testName, u_errorName(errorCode));
1303         return NULL;
1304     }
1305
1306     /* set values from setRanges[] */
1307     ok=TRUE;
1308     for(i=0; i<countSetRanges; ++i) {
1309         start=setRanges[i].start;
1310         limit=setRanges[i].limit;
1311         value=setRanges[i].value;
1312         overwrite=setRanges[i].overwrite;
1313         if((limit-start)==1 && overwrite) {
1314             ok&=utrie_set32(newTrie, start, value);
1315         } else {
1316             ok&=utrie_setRange32(newTrie, start, limit, value, overwrite);
1317         }
1318     }
1319     if(ok) {
1320         return newTrie;
1321     } else {
1322         log_err("error: setting values into a trie1 (%s) failed\n", testName);
1323         utrie_close(newTrie);
1324         return NULL;
1325     }
1326 }
1327
1328 static void
1329 testTrie2FromTrie1(const char *testName,
1330                    const SetRange setRanges[], int32_t countSetRanges,
1331                    const CheckRange checkRanges[], int32_t countCheckRanges) {
1332     uint32_t memory1_16[3000], memory1_32[3000];
1333     int32_t length16, length32;
1334     UChar lead;
1335
1336     char name[40];
1337
1338     UNewTrie *newTrie1_16, *newTrie1_32;
1339     UTrie trie1_16, trie1_32;
1340     UTrie2 *trie2;
1341     uint32_t initialValue, errorValue;
1342     UErrorCode errorCode;
1343
1344     newTrie1_16=makeNewTrie1WithRanges(testName,
1345                                        setRanges, countSetRanges,
1346                                        checkRanges, countCheckRanges);
1347     if(newTrie1_16==NULL) {
1348         return;
1349     }
1350     newTrie1_32=utrie_clone(NULL, newTrie1_16, NULL, 0);
1351     if(newTrie1_32==NULL) {
1352         utrie_close(newTrie1_16);
1353         return;
1354     }
1355     errorCode=U_ZERO_ERROR;
1356     length16=utrie_serialize(newTrie1_16, memory1_16, sizeof(memory1_16),
1357                              NULL, TRUE, &errorCode);
1358     length32=utrie_serialize(newTrie1_32, memory1_32, sizeof(memory1_32),
1359                              NULL, FALSE, &errorCode);
1360     utrie_unserialize(&trie1_16, memory1_16, length16, &errorCode);
1361     utrie_unserialize(&trie1_32, memory1_32, length32, &errorCode);
1362     utrie_close(newTrie1_16);
1363     utrie_close(newTrie1_32);
1364     if(U_FAILURE(errorCode)) {
1365         log_err("error: utrie_serialize or unserialize(%s) failed: %s\n",
1366                 testName, u_errorName(errorCode));
1367         return;
1368     }
1369
1370     getSpecialValues(checkRanges, countCheckRanges, &initialValue, &errorValue);
1371
1372     uprv_strcpy(name, testName);
1373     uprv_strcat(name, ".16");
1374     trie2=utrie2_fromUTrie(&trie1_16, errorValue, &errorCode);
1375     if(U_SUCCESS(errorCode)) {
1376         testFrozenTrie(name, trie2, UTRIE2_16_VALUE_BITS, checkRanges, countCheckRanges);
1377         for(lead=0xd800; lead<0xdc00; ++lead) {
1378             uint32_t value1, value2;
1379             value1=UTRIE_GET16_FROM_LEAD(&trie1_16, lead);
1380             value2=UTRIE2_GET16_FROM_U16_SINGLE_LEAD(trie2, lead);
1381             if(value1!=value2) {
1382                 log_err("error: utrie2_fromUTrie(%s) wrong value %ld!=%ld "
1383                         "from lead surrogate code unit U+%04lx\n",
1384                         name, (long)value2, (long)value1, (long)lead);
1385                 break;
1386             }
1387         }
1388     }
1389     utrie2_close(trie2);
1390
1391     uprv_strcpy(name, testName);
1392     uprv_strcat(name, ".32");
1393     trie2=utrie2_fromUTrie(&trie1_32, errorValue, &errorCode);
1394     if(U_SUCCESS(errorCode)) {
1395         testFrozenTrie(name, trie2, UTRIE2_32_VALUE_BITS, checkRanges, countCheckRanges);
1396         for(lead=0xd800; lead<0xdc00; ++lead) {
1397             uint32_t value1, value2;
1398             value1=UTRIE_GET32_FROM_LEAD(&trie1_32, lead);
1399             value2=UTRIE2_GET32_FROM_U16_SINGLE_LEAD(trie2, lead);
1400             if(value1!=value2) {
1401                 log_err("error: utrie2_fromUTrie(%s) wrong value %ld!=%ld "
1402                         "from lead surrogate code unit U+%04lx\n",
1403                         name, (long)value2, (long)value1, (long)lead);
1404                 break;
1405             }
1406         }
1407     }
1408     utrie2_close(trie2);
1409 }
1410
1411 static void
1412 Trie12ConversionTest(void) {
1413     testTrie2FromTrie1("trie1->trie2",
1414                        setRanges2, LENGTHOF(setRanges2),
1415                        checkRanges2, LENGTHOF(checkRanges2));
1416 }
1417
1418 void
1419 addTrie2Test(TestNode** root) {
1420     addTest(root, &TrieTest, "tsutil/trie2test/TrieTest");
1421     addTest(root, &EnumNewTrieForLeadSurrogateTest,
1422                   "tsutil/trie2test/EnumNewTrieForLeadSurrogateTest");
1423     addTest(root, &DummyTrieTest, "tsutil/trie2test/DummyTrieTest");
1424     addTest(root, &FreeBlocksTest, "tsutil/trie2test/FreeBlocksTest");
1425     addTest(root, &GrowDataArrayTest, "tsutil/trie2test/GrowDataArrayTest");
1426     addTest(root, &GetVersionTest, "tsutil/trie2test/GetVersionTest");
1427     addTest(root, &Trie12ConversionTest, "tsutil/trie2test/Trie12ConversionTest");
1428 }