icuSources/test/cintltst/cfrtst.c

   1 /********************************************************************
   2  * COPYRIGHT:
   3  * Copyright (c) 1997-2010, International Business Machines Corporation and
   4  * others. All Rights Reserved.
   5  ********************************************************************/
   6 /********************************************************************************
   7 *
   8 * File CFRTST.C
   9 *
  10 * Modification History:
  11 *        Name                     Description
  12 *     Madhu Katragadda            Ported for C API
  13 *********************************************************************************/
  14 /**
  15  * CollationFrenchTest is a third level test class.  This tests the locale
  16  * specific primary, secondary and tertiary rules.  For example, the ignorable
  17  * character '-' in string "black-bird".  The en_US locale uses the default
  18  * collation rules as its sorting sequence.
  19  */
  20
  21 #include <stdlib.h>
  22
  23 #include "unicode/utypes.h"
  24
  25 #if !UCONFIG_NO_COLLATION
  26
  27 #include "unicode/ucol.h"
  28 #include "unicode/uloc.h"
  29 #include "cintltst.h"
  30 #include "ccolltst.h"
  31 #include "callcoll.h"
  32 #include "cfrtst.h"
  33 #include "unicode/ustring.h"
  34 #include "string.h"
  35
  36 static  UCollator *myCollation;
  37 const static UChar testSourceCases[][MAX_TOKEN_LEN] =
  38 {
  39     {0x0061/*'a'*/, 0x0062/*'b'*/, 0x0063/*'c'*/, 0x0000},
  40     {0x0043/*'C'*/, 0x004f/*'O'*/, 0x0054/*'T'*/, 0x0045/*'E'*/, 0x0000},
  41     {0x0063/*'c'*/, 0x006f/*'o'*/, 0x002d/*'-'*/, 0x006f/*'o'*/, 0x0070/*'p'*/, 0x0000},
  42     {0x0070/*'p'*/, 0x00EA, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x0065/*'e'*/, 0x0000},
  43     {0x0070/*'p'*/, 0x00EA, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x0065/*'e'*/, 0x0072/*'r'*/, 0x0000},
  44     {0x0070/*'p'*/, 0x00E9, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x0065/*'e'*/, 0x0072/*'r'*/, 0x0000},
  45     {0x0070/*'p'*/, 0x00E9, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x0065/*'e'*/, 0x0072/*'r'*/, 0x0000},
  46     {0x0048/*'H'*/, 0x0065/*'e'*/, 0x006c/*'l'*/, 0x006c/*'l'*/, 0x006f/*'o'*/, 0x0000},
  47     {0x01f1, 0x0000},
  48     {0xfb00, 0x0000},
  49     {0x01fa, 0x0000},
  50     {0x0101, 0x0000}
  51 };
  52
  53 const static UChar testTargetCases[][MAX_TOKEN_LEN] =
  54 {
  55     {0x0041/*'A'*/, 0x0042/*'B'*/, 0x0043/*'C'*/, 0x0000},
  56     {0x0063/*'c'*/, 0x00f4, 0x0074/*'t'*/, 0x0065/*'e'*/, 0x0000},
  57     {0x0043/*'C'*/, 0x004f/*'O'*/, 0x004f/*'O'*/, 0x0050/*'P'*/, 0x0000},
  58     {0x0070/*'p'*/, 0x00E9, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x00E9, 0x0000},
  59     {0x0070/*'p'*/,  0x00E9, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x00E9, 0x0000},
  60     {0x0070/*'p'*/, 0x00EA, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x0065/*'e'*/, 0x0000},
  61     {0x0070/*'p'*/, 0x00EA, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x0065/*'e'*/, 0x0072/*'r'*/, 0x0000},
  62     {0x0068/*'h'*/, 0x0065/*'e'*/, 0x006c/*'l'*/, 0x006c/*'l'*/, 0x004f/*'O'*/, 0x0000},
  63     {0x01ee, 0x0000},
  64     {0x25ca, 0x0000},
  65     {0x00e0, 0x0000},
  66     {0x01df, 0x0000}
  67 };
  68
  69 const static UCollationResult results[] =
  70 {
  71     UCOL_LESS,
  72     UCOL_LESS,
  73     UCOL_LESS, /*UCOL_GREATER,*/
  74     UCOL_LESS,
  75     UCOL_GREATER,
  76     UCOL_GREATER,
  77     UCOL_LESS,
  78     UCOL_GREATER,
  79     UCOL_LESS, /*UCOL_GREATER,*/
  80     UCOL_GREATER,
  81     UCOL_LESS,
  82     UCOL_LESS
  83 };
  84
  85 /* 0x0300 is grave, 0x0301 is acute*/
  86 /* the order of elements in this array must be different than the order in CollationEnglishTest*/
  87 const static UChar testAcute[][MAX_TOKEN_LEN] =
  88 {
  89 /*00*/    {0x0065/*'e'*/, 0x0065/*'e'*/,  0x0000},
  90 /*01*/    {0x0065/*'e'*/, 0x0301, 0x0065/*'e'*/,  0x0000},
  91 /*02*/    {0x0065/*'e'*/, 0x0300, 0x0301, 0x0065/*'e'*/,  0x0000},
  92 /*03*/    {0x0065/*'e'*/, 0x0300, 0x0065/*'e'*/,  0x0000},
  93 /*04*/    {0x0065/*'e'*/, 0x0301, 0x0300, 0x0065/*'e'*/,  0x0000},
  94 /*05*/    {0x0065/*'e'*/, 0x0065/*'e'*/, 0x0301, 0x0000},
  95 /*06*/    {0x0065/*'e'*/, 0x0301, 0x0065/*'e'*/, 0x0301, 0x0000},
  96 /*07*/    {0x0065/*'e'*/, 0x0300, 0x0301, 0x0065/*'e'*/, 0x0301, 0x0000},
  97 /*08*/    {0x0065/*'e'*/, 0x0300, 0x0065/*'e'*/, 0x0301, 0x0000},
  98 /*09*/    {0x0065/*'e'*/, 0x0301, 0x0300, 0x0065/*'e'*/, 0x0301, 0x0000},
  99 /*0a*/    {0x0065/*'e'*/, 0x0065/*'e'*/, 0x0300, 0x0301, 0x0000},
 100 /*0b*/    {0x0065/*'e'*/, 0x0301, 0x0065/*'e'*/, 0x0300, 0x0301, 0x0000},
 101 /*0c*/    {0x0065/*'e'*/, 0x0300, 0x0301, 0x0065/*'e'*/, 0x0300, 0x0301, 0x0000},
 102 /*0d*/    {0x0065/*'e'*/, 0x0300, 0x0065/*'e'*/, 0x0300, 0x0301, 0x0000},
 103 /*0e*/    {0x0065/*'e'*/, 0x0301, 0x0300, 0x0065/*'e'*/, 0x0300, 0x0301, 0x0000},
 104 /*0f*/    {0x0065/*'e'*/, 0x0065/*'e'*/, 0x0300, 0x0000},
 105 /*10*/    {0x0065/*'e'*/, 0x0301, 0x0065/*'e'*/, 0x0300, 0x0000},
 106 /*11*/    {0x0065/*'e'*/, 0x0300, 0x0301, 0x0065/*'e'*/, 0x0300, 0x0000},
 107 /*12*/    {0x0065/*'e'*/, 0x0300, 0x0065/*'e'*/, 0x0300, 0x0000},
 108 /*13*/    {0x0065/*'e'*/, 0x0301, 0x0300, 0x0065/*'e'*/, 0x0300, 0x0000},
 109 /*14*/    {0x0065/*'e'*/, 0x0065/*'e'*/, 0x0301, 0x0300, 0x0000},
 110 /*15*/    {0x0065/*'e'*/, 0x0301, 0x0065/*'e'*/, 0x0301, 0x0300, 0x0000},
 111 /*16*/    {0x0065/*'e'*/, 0x0300, 0x0301, 0x0065/*'e'*/, 0x0301, 0x0300, 0x0000},
 112 /*17*/    {0x0065/*'e'*/, 0x0300, 0x0065/*'e'*/, 0x0301, 0x0300, 0x0000},
 113 /*18*/    {0x0065/*'e'*/, 0x0301, 0x0300, 0x0065/*'e'*/, 0x0301, 0x0300, 0x0000}
 114 };
 115
 116 const static UChar testBugs[][MAX_TOKEN_LEN] =
 117 {
 118     {0x0061/*'a'*/, 0x000},
 119     {0x0041/*'A'*/, 0x000},
 120     {0x0065/*'e'*/, 0x000},
 121     {0x0045/*'E'*/, 0x000},
 122     {0x00e9, 0x000},
 123     {0x00e8, 0x000},
 124     {0x00ea, 0x000},
 125     {0x00eb, 0x000},
 126     {0x0065/*'e'*/, 0x0061/*'a'*/, 0x000},
 127     {0x0078/*'x'*/, 0x000}
 128 };
 129
 130
 131 static void TestGetSortKey(void);
 132
 133
 134 void addFrenchCollTest(TestNode** root)
 135 {
 136     addTest(root, &TestSecondary, "tscoll/cfrtst/TestSecondary");
 137     addTest(root, &TestTertiary, "tscoll/cfrtst/TestTertiary");
 138     addTest(root, &TestExtra, "tscoll/cfrtst/TestExtra");
 139     addTest(root, &TestGetSortKey, "tscoll/cfrtst/TestGetSortKey");
 140 }
 141
 142
 143 static void TestTertiary( )
 144 {
 145
 146     int32_t i;
 147     UErrorCode status = U_ZERO_ERROR;
 148     myCollation = ucol_open("fr_CA", &status);
 149     if(U_FAILURE(status) || !myCollation){
 150         log_err_status(status, "ERROR: in creation of rule based collator: %s\n", myErrorName(status));
 151         return;
 152     }
 153
 154     ucol_setAttribute(myCollation, UCOL_ALTERNATE_HANDLING, UCOL_SHIFTED, &status);
 155     if(U_FAILURE(status)){
 156         log_err("ERROR: in creation of rule based collator: %s\n", myErrorName(status));
 157         return;
 158     }
 159     log_verbose("Testing fr_CA Collation with Tertiary strength\n");
 160     ucol_setStrength(myCollation, UCOL_QUATERNARY);
 161     for (i = 0; i < 12 ; i++)
 162     {
 163         doTest(myCollation, testSourceCases[i], testTargetCases[i], results[i]);
 164     }
 165     ucol_close(myCollation);
 166 }
 167
 168 static void TestSecondary()
 169 {
 170     int32_t i,j, testAcuteSize;
 171     UCollationResult expected=UCOL_EQUAL;
 172     UErrorCode status = U_ZERO_ERROR;
 173     myCollation = ucol_open("fr_CA", &status);
 174     if(U_FAILURE(status)){
 175         log_err_status(status, "ERROR: in creation of rule based collator: %s\n", myErrorName(status));
 176         return;
 177     }
 178     ucol_setAttribute(myCollation, UCOL_STRENGTH, UCOL_SECONDARY, &status);
 179     if(U_FAILURE(status)){
 180         log_err("ERROR: in creation of rule based collator: %s\n", myErrorName(status));
 181         return;
 182     }
 183     log_verbose("Testing fr_CA Collation with Secondary strength\n");
 184     /*test acute and grave ordering (compare to french collation)*/
 185     testAcuteSize = sizeof(testAcute) / sizeof(testAcute[0]);
 186     for (i = 0; i < testAcuteSize; i++)
 187     {
 188         for (j = 0; j < testAcuteSize; j++)
 189         {
 190             if (i <  j) expected = UCOL_LESS;
 191             if (i == j) expected = UCOL_EQUAL;
 192             if (i >  j) expected = UCOL_GREATER;
 193             doTest(myCollation, testAcute[i], testAcute[j], expected );
 194         }
 195     }
 196     ucol_close(myCollation);
 197 }
 198
 199 static void TestExtra()
 200 {
 201     int32_t i, j;
 202     UErrorCode status = U_ZERO_ERROR;
 203     myCollation = ucol_open("fr_CA", &status);
 204     if(U_FAILURE(status)){
 205         log_err_status(status, "ERROR: in creation of rule based collator: %s\n", myErrorName(status));
 206         return;
 207     }
 208     log_verbose("Testing fr_CA Collation extra with secondary strength\n");
 209     ucol_setStrength(myCollation, UCOL_TERTIARY);
 210     for (i = 0; i < 9 ; i++)
 211     {
 212         for (j = i + 1; j < 10; j += 1)
 213         {
 214             doTest(myCollation, testBugs[i], testBugs[j], UCOL_LESS);
 215         }
 216     }
 217     ucol_close(myCollation);
 218 }
 219
 220 static void TestGetSortKey() {
 221     /* This is meant to test a buffer reallocation crash while using
 222     French secondary sorting with a large buffer.
 223     The fact that Japanese characters are used is irrelevant. */
 224     static const UChar pucUTF16[] = {
 225         0x3049,0x30b9,0x3088,0xfffd,0xfffd,0x308f,0xfffd,0x3042,
 226         0xfffd,0xfffd,0x305e,0xfffd,0x30b6,0x30bb,0x305b,0x30b1,
 227         0x3050,0x30af,0x304e,0x30bd,0xfffd,0x30c6,0xfffd,0xfffd,
 228         0x30e1,0xfffd,0xfffd,0x30d9,0xfffd,0x3092,0x3075,0x304a,
 229         0x3074,0x3070,0x30f5,0x30c4,0x306e,0x30df,0x3053,0xfffd,
 230         0x30a6,0x30b6,0x30e0,0xfffd,0x30bc,0x30ef,0x3087,0x30cc,
 231         0x305f,0x30de,0xfffd,0x3090,0x3063,0x30dc,0x30b6,0x30b9,
 232         0x30d2,0x3072,0x3061,0xfffd,0xfffd,0xfffd,0x307b,0x3092,
 233         0x30a5,0x30a9,0x30b1,0x30e7,0xfffd,0xfffd,0xfffd,0xfffd,
 234         0xfffd,0x305e,0xfffd,0x30c7,0x30ae,0x305b,0x308b,0x30c0,
 235         0x30f5,0xfffd,0xfffd,0xfffd,0x307d,0x304e,0xfffd,0xfffd,
 236         0x30c0,0x30c8,0x306f,0x307a,0x30dd,0x30e4,0x3084,0xfffd,
 237         0x308c,0x30f1,0xfffd,0x30c6,0xfffd,0x307a,0xfffd,0x3052,
 238         0x3056,0x305d,0x30b7,0xfffd,0x305b,0x30b0,0x30b9,0xfffd,
 239         0x30b2,0x306d,0x3044,0xfffd,0x3073,0xfffd,0x30be,0x30cf,
 240         0x3080,0xfffd,0x30a8,0x30f5,0x30a5,0x30c7,0x307c,0xfffd,
 241         0x30d1,0x305f,0x30b2,0xfffd,0x3053,0x30ca,0xfffd,0x30dd,
 242         0x3058,0x30c0,0x305d,0x30e1,0xfffd,0x30bb,0x305f,0x30d1,
 243         0x30f2,0x3058,0x3086,0x30ce,0x30db,0x30cb,0x30e9,0xfffd,
 244         0x308c,0xfffd,0xfffd,0x30af,0x30c4,0x3076,0x304c,0x30f5,
 245         0x30e8,0x308c,0xfffd,0x30e2,0x3073,0x30a3,0x304e,0x30ea,
 246         0xfffd,0x304f,0xfffd,0x306c,0x3044,0xfffd,0xfffd,0x30c9,
 247         0xfffd,0x30f5,0xfffd,0xfffd,0xfffd,0x30eb,0x30a8,0xfffd,
 248         0x306d,0x307d,0x30d8,0x3069,0xfffd,0xfffd,0x3086,0x30a9,
 249         0xfffd,0x3076,0x30e9,0x30cc,0x3074,0x30e0,0xfffd,0xfffd,
 250         0xfffd,0x30f0,0x3086,0x30ac,0x3076,0x3068,0x30c7,0xfffd,
 251         0x30b7,0x30d2,0x3048,0x308e,0x30e8,0x30d9,0x30ce,0x30d0,
 252         0x308b,0x30ee,0x30e6,0x3079,0x30f3,0x30af,0xfffd,0x3079,
 253         0xfffd,0xfffd,0x30ca,0x30bf,0xfffd,0x30b5,0xfffd,0xfffd,
 254         0x3093,0xfffd,0x30ba,0xfffd,0x3076,0x3047,0x304a,0xfffd,
 255         0xfffd,0x3086,0xfffd,0x3081,0xfffd,0x30f6,0x3066,0xfffd,
 256         0xfffd,0x30b6,0x30ef,0x30e2,0x30bf,0xfffd,0x3053,0x304a,
 257         0xfffd,0xfffd,0x304a,0x30e8,0xfffd,0x30e2,0xfffd,0xfffd,
 258         0x305c,0x3081,0x30c6,0xfffd,0x3091,0x3046,0x306a,0x3059,
 259         0xfffd,0xfffd,0x30dd,0x30d1,0x308a,0x30ee,0xfffd,0xfffd,
 260         0x308a,0x3042,0x30da,0xfffd,0x3064,0x30ef,0x305c,0x306b,
 261         0xfffd,0x30ca,0x3085,0x3067,0x30ea,0x30c2,0x30c8,0xfffd,
 262         0x30f5,0xfffd,0xfffd,0xfffd,0x30ca,0xfffd,0x3050,0x30f1,
 263         0x3050,0x3053,0x3072,0xfffd,0xfffd,0xfffd,0x3074,0xfffd,
 264         0x304b,0x30dd,0x306d,0xfffd,0x3049,0x30a1,0x30cc,0x30de,
 265         0x30ae,0x307b,0x308a,0xfffd,0x3065,0xfffd,0xfffd,0x30c0,
 266         0xfffd,0x3048,0x30dc,0x304f,0x3085,0x3059,0x304b,0x30d3,
 267         0x30eb,0x30a4,0x3073,0xfffd,0x30ba,0x308f,0x30a7,0x30c3,
 268         0x3074,0x30cf,0x306c,0x3053,0x30c0,0xfffd,0x3066,0xfffd,
 269         0x308f,0xfffd,0x30b5,0xfffd,0x3092,0x30c4,0xfffd,0x30d6,
 270         0x3056,0x30ad,0x30d2,0x30ba,0xfffd,0x30e6,0x304c,0x3088,
 271         0x30b6,0x3048,0x3077,0x30d1,0xfffd,0x3050,0xfffd,0x3042,
 272         0xfffd,0xfffd,0x308f,0xfffd,0x30c1,0xfffd,0x3074,0x3061,
 273         0x3056,0x30e5,0xfffd,0xfffd,0x3057,0xfffd,0xfffd,0xfffd,
 274         0xfffd,0x30bd,0x30b3,0x30ee,0xfffd,0x30f2,0x3084,0x3050,
 275         0xfffd,0x30e7,0xfffd,0xfffd,0x3060,0x3049,0x30f2,0x30ad,
 276         0x30bf,0x30f1,0x30a2,0xfffd,0x30af,0xfffd,0x3060,0x30a1,
 277         0x30e9,0x30c3,0xfffd,0x3072,0x3093,0x3070,0xfffd,0x308f,
 278         0x3060,0xfffd,0x3067,0x306f,0x3082,0x308b,0x3051,0xfffd,
 279         0x3058,0xfffd,0xfffd,0x30a8,0x3051,0x3054,0x30ad,0x30f0,
 280         0x3053,0xfffd,0x30e1,0x30d7,0x308d,0x307f,0x30be,0x30b0,
 281         0xfffd,0x30db,0xfffd,0x30d1,0xfffd,0x3054,0x30a5,0xfffd,
 282         0x306a,0xfffd,0x305c,0xfffd,0x3052,0x3088,0xfffd,0x306e,
 283         0xfffd,0x30a9,0x30a1,0x30b4,0x3083,0x30bd,0xfffd,0xfffd,
 284         0x306a,0x3070,0x30cd,0xfffd,0x3072,0x30ed,0x30c6,0x30be,
 285         0x30c4,0x305e,0x30b3,0x30e1,0x308a,0xfffd,0x305b,0xfffd,
 286         0x3042,0x3088,0xfffd,0x304c,0xfffd,0x3089,0x3071,0xfffd,
 287         0xfffd,0x30c6,0x3062,0x3079,0xfffd,0x304b,0x304a,0xfffd,
 288         0x30ad,0x3045,0x3045,0x3087,0xfffd,0x306a,0x308b,0x0000,
 289         0x30bd,0x3065,0x30b8,0x3086,0x30d3,0x3076,0xfffd,0xfffd,
 290         0x308f,0x3053,0x307c,0x3053,0x3084,0x30ae,0x30c4,0x3045,
 291         0x30a8,0x30d0,0x30e1,0x308c,0x30e6,0x30b7,0xfffd,0xfffd,
 292         0xfffd,0x3046,0x305f,0xfffd,0x3086,0x30ab,0xfffd,0xfffd,
 293         0x30c8,0xfffd,0x30a1,0x3052,0x3059,0xfffd,0x30a4,0xfffd,
 294         0xfffd,0x308c,0x3085,0x30ab,0x30b5,0x3091,0x30bf,0x30e3,
 295         0xfffd,0xfffd,0x3087,0xfffd,0x30f6,0x3051,0x30bd,0x3092,
 296         0x3063,0xfffd,0x30a9,0x3063,0x306e,0xfffd,0xfffd,0xfffd,
 297         0x306c,0xfffd,0x307e,0x30ad,0x3077,0x30c2,0x30e9,0x30d5,
 298         0xfffd,0xfffd,0x30c6,0x305c,0xfffd,0xfffd,0x3089,0xfffd,
 299         0x3048,0x30cb,0x308c,0xfffd,0xfffd,0x3044,0xfffd,0x3080,
 300         0x3063,0x3079,0xfffd,0x308a,0x30cb,0x3042,0x3057,0xfffd,
 301         0x307c,0x30c1,0x30a8,0x30cf,0xfffd,0x3083,0xfffd,0xfffd,
 302         0x306c,0xfffd,0x305e,0x3092,0xfffd,0x30dc,0x30b0,0x3081,
 303         0x30e3,0x30f0,0x304e,0x30cc,0x308e,0x30c4,0x30ad
 304     };
 305
 306     UErrorCode status = U_ZERO_ERROR;
 307     UCollator *pCollator;
 308     int32_t lenActualSortKey;
 309     uint8_t pucSortKey[4096];
 310     static const int32_t LENSORTKEY = (int32_t)sizeof(pucSortKey);
 311
 312     ucol_prepareShortStringOpen("LFR_AN_CX_EX_FO_HX_NX_S3", 0, NULL, &status);
 313
 314     pCollator = ucol_openFromShortString("LFR_AN_CX_EX_FO_HX_NX_S3", 0, NULL, &status);
 315
 316     if (U_FAILURE(status)) {
 317         log_data_err("error opening collator -> %s. (Are you missing data?)\n", u_errorName(status));
 318         return;
 319     }
 320
 321     lenActualSortKey = ucol_getSortKey(pCollator,
 322         (const UChar *)pucUTF16,
 323         sizeof(pucUTF16) / sizeof(pucUTF16[0]),
 324         pucSortKey,
 325         LENSORTKEY);
 326
 327     if (lenActualSortKey > LENSORTKEY) {
 328         log_err("sort key too big for original buffer. Got: %d Expected: %d\n", lenActualSortKey, LENSORTKEY);
 329         return;
 330     }
 331     /* If the test didn't crash, then the test succeeded. */
 332     ucol_close(pCollator);
 333 }
 334
 335 #endif /* #if !UCONFIG_NO_COLLATION */