icuSources/test/cintltst/cfrtst.c

   1 /********************************************************************
   2  * COPYRIGHT:
   3  * Copyright (c) 1997-2016, 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 "cmemory.h"
  34 #include "unicode/ustring.h"
  35 #include "string.h"
  36
  37 static  UCollator *myCollation;
  38 const static UChar testSourceCases[][MAX_TOKEN_LEN] =
  39 {
  40     {0x0061/*'a'*/, 0x0062/*'b'*/, 0x0063/*'c'*/, 0x0000},
  41     {0x0043/*'C'*/, 0x004f/*'O'*/, 0x0054/*'T'*/, 0x0045/*'E'*/, 0x0000},
  42     {0x0063/*'c'*/, 0x006f/*'o'*/, 0x002d/*'-'*/, 0x006f/*'o'*/, 0x0070/*'p'*/, 0x0000},
  43     {0x0070/*'p'*/, 0x00EA, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x0065/*'e'*/, 0x0000},
  44     {0x0070/*'p'*/, 0x00EA, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x0065/*'e'*/, 0x0072/*'r'*/, 0x0000},
  45     {0x0070/*'p'*/, 0x00E9, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x0065/*'e'*/, 0x0072/*'r'*/, 0x0000},
  46     {0x0070/*'p'*/, 0x00E9, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x0065/*'e'*/, 0x0072/*'r'*/, 0x0000},
  47     {0x0048/*'H'*/, 0x0065/*'e'*/, 0x006c/*'l'*/, 0x006c/*'l'*/, 0x006f/*'o'*/, 0x0000},
  48     {0x01f1, 0x0000},
  49     {0xfb00, 0x0000},
  50     {0x01fa, 0x0000},
  51     {0x0101, 0x0000}
  52 };
  53
  54 const static UChar testTargetCases[][MAX_TOKEN_LEN] =
  55 {
  56     {0x0041/*'A'*/, 0x0042/*'B'*/, 0x0043/*'C'*/, 0x0000},
  57     {0x0063/*'c'*/, 0x00f4, 0x0074/*'t'*/, 0x0065/*'e'*/, 0x0000},
  58     {0x0043/*'C'*/, 0x004f/*'O'*/, 0x004f/*'O'*/, 0x0050/*'P'*/, 0x0000},
  59     {0x0070/*'p'*/, 0x00E9, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x00E9, 0x0000},
  60     {0x0070/*'p'*/,  0x00E9, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x00E9, 0x0000},
  61     {0x0070/*'p'*/, 0x00EA, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x0065/*'e'*/, 0x0000},
  62     {0x0070/*'p'*/, 0x00EA, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x0065/*'e'*/, 0x0072/*'r'*/, 0x0000},
  63     {0x0068/*'h'*/, 0x0065/*'e'*/, 0x006c/*'l'*/, 0x006c/*'l'*/, 0x004f/*'O'*/, 0x0000},
  64     {0x01ee, 0x0000},
  65     {0x25ca, 0x0000},
  66     {0x00e0, 0x0000},
  67     {0x01df, 0x0000}
  68 };
  69
  70 const static UCollationResult results[] =
  71 {
  72     UCOL_LESS,
  73     UCOL_LESS,
  74     UCOL_LESS, /*UCOL_GREATER,*/
  75     UCOL_LESS,
  76     UCOL_GREATER,
  77     UCOL_GREATER,
  78     UCOL_LESS,
  79     UCOL_GREATER,
  80     UCOL_LESS, /*UCOL_GREATER,*/
  81     UCOL_GREATER,
  82     UCOL_LESS,
  83     UCOL_LESS
  84 };
  85
  86 /* 0x0300 is grave, 0x0301 is acute*/
  87 /* the order of elements in this array must be different than the order in CollationEnglishTest*/
  88 const static UChar testAcute[][MAX_TOKEN_LEN] =
  89 {
  90 /*00*/    {0x0065/*'e'*/, 0x0065/*'e'*/,  0x0000},
  91 /*01*/    {0x0065/*'e'*/, 0x0301, 0x0065/*'e'*/,  0x0000},
  92 /*02*/    {0x0065/*'e'*/, 0x0300, 0x0301, 0x0065/*'e'*/,  0x0000},
  93 /*03*/    {0x0065/*'e'*/, 0x0300, 0x0065/*'e'*/,  0x0000},
  94 /*04*/    {0x0065/*'e'*/, 0x0301, 0x0300, 0x0065/*'e'*/,  0x0000},
  95 /*05*/    {0x0065/*'e'*/, 0x0065/*'e'*/, 0x0301, 0x0000},
  96 /*06*/    {0x0065/*'e'*/, 0x0301, 0x0065/*'e'*/, 0x0301, 0x0000},
  97 /*07*/    {0x0065/*'e'*/, 0x0300, 0x0301, 0x0065/*'e'*/, 0x0301, 0x0000},
  98 /*08*/    {0x0065/*'e'*/, 0x0300, 0x0065/*'e'*/, 0x0301, 0x0000},
  99 /*09*/    {0x0065/*'e'*/, 0x0301, 0x0300, 0x0065/*'e'*/, 0x0301, 0x0000},
 100 /*0a*/    {0x0065/*'e'*/, 0x0065/*'e'*/, 0x0300, 0x0301, 0x0000},
 101 /*0b*/    {0x0065/*'e'*/, 0x0301, 0x0065/*'e'*/, 0x0300, 0x0301, 0x0000},
 102 /*0c*/    {0x0065/*'e'*/, 0x0300, 0x0301, 0x0065/*'e'*/, 0x0300, 0x0301, 0x0000},
 103 /*0d*/    {0x0065/*'e'*/, 0x0300, 0x0065/*'e'*/, 0x0300, 0x0301, 0x0000},
 104 /*0e*/    {0x0065/*'e'*/, 0x0301, 0x0300, 0x0065/*'e'*/, 0x0300, 0x0301, 0x0000},
 105 /*0f*/    {0x0065/*'e'*/, 0x0065/*'e'*/, 0x0300, 0x0000},
 106 /*10*/    {0x0065/*'e'*/, 0x0301, 0x0065/*'e'*/, 0x0300, 0x0000},
 107 /*11*/    {0x0065/*'e'*/, 0x0300, 0x0301, 0x0065/*'e'*/, 0x0300, 0x0000},
 108 /*12*/    {0x0065/*'e'*/, 0x0300, 0x0065/*'e'*/, 0x0300, 0x0000},
 109 /*13*/    {0x0065/*'e'*/, 0x0301, 0x0300, 0x0065/*'e'*/, 0x0300, 0x0000},
 110 /*14*/    {0x0065/*'e'*/, 0x0065/*'e'*/, 0x0301, 0x0300, 0x0000},
 111 /*15*/    {0x0065/*'e'*/, 0x0301, 0x0065/*'e'*/, 0x0301, 0x0300, 0x0000},
 112 /*16*/    {0x0065/*'e'*/, 0x0300, 0x0301, 0x0065/*'e'*/, 0x0301, 0x0300, 0x0000},
 113 /*17*/    {0x0065/*'e'*/, 0x0300, 0x0065/*'e'*/, 0x0301, 0x0300, 0x0000},
 114 /*18*/    {0x0065/*'e'*/, 0x0301, 0x0300, 0x0065/*'e'*/, 0x0301, 0x0300, 0x0000}
 115 };
 116
 117 const static UChar testBugs[][MAX_TOKEN_LEN] =
 118 {
 119     {0x0061/*'a'*/, 0x000},
 120     {0x0041/*'A'*/, 0x000},
 121     {0x0065/*'e'*/, 0x000},
 122     {0x0045/*'E'*/, 0x000},
 123     {0x00e9, 0x000},
 124     {0x00e8, 0x000},
 125     {0x00ea, 0x000},
 126     {0x00eb, 0x000},
 127     {0x0065/*'e'*/, 0x0061/*'a'*/, 0x000},
 128     {0x0078/*'x'*/, 0x000}
 129 };
 130
 131
 132 static void TestGetSortKey(void);
 133
 134
 135 void addFrenchCollTest(TestNode** root)
 136 {
 137     addTest(root, &TestSecondary, "tscoll/cfrtst/TestSecondary");
 138     addTest(root, &TestTertiary, "tscoll/cfrtst/TestTertiary");
 139     addTest(root, &TestExtra, "tscoll/cfrtst/TestExtra");
 140     addTest(root, &TestGetSortKey, "tscoll/cfrtst/TestGetSortKey");
 141 }
 142
 143
 144 static void TestTertiary( )
 145 {
 146
 147     int32_t i;
 148     UErrorCode status = U_ZERO_ERROR;
 149     myCollation = ucol_open("fr_CA", &status);
 150     if(U_FAILURE(status) || !myCollation){
 151         log_err_status(status, "ERROR: in creation of rule based collator: %s\n", myErrorName(status));
 152         return;
 153     }
 154
 155     ucol_setAttribute(myCollation, UCOL_ALTERNATE_HANDLING, UCOL_SHIFTED, &status);
 156     if(U_FAILURE(status)){
 157         log_err("ERROR: in creation of rule based collator: %s\n", myErrorName(status));
 158         return;
 159     }
 160     log_verbose("Testing fr_CA Collation with Tertiary strength\n");
 161     ucol_setStrength(myCollation, UCOL_QUATERNARY);
 162     for (i = 0; i < 12 ; i++)
 163     {
 164         doTest(myCollation, testSourceCases[i], testTargetCases[i], results[i]);
 165     }
 166     ucol_close(myCollation);
 167 }
 168
 169 static void TestSecondary()
 170 {
 171     int32_t i,j, testAcuteSize;
 172     UCollationResult expected=UCOL_EQUAL;
 173     UErrorCode status = U_ZERO_ERROR;
 174     myCollation = ucol_open("fr_CA", &status);
 175     if(U_FAILURE(status)){
 176         log_err_status(status, "ERROR: in creation of rule based collator: %s\n", myErrorName(status));
 177         return;
 178     }
 179     ucol_setAttribute(myCollation, UCOL_STRENGTH, UCOL_SECONDARY, &status);
 180     if(U_FAILURE(status)){
 181         log_err("ERROR: in creation of rule based collator: %s\n", myErrorName(status));
 182         return;
 183     }
 184     log_verbose("Testing fr_CA Collation with Secondary strength\n");
 185     /*test acute and grave ordering (compare to french collation)*/
 186     testAcuteSize = UPRV_LENGTHOF(testAcute);
 187     for (i = 0; i < testAcuteSize; i++)
 188     {
 189         for (j = 0; j < testAcuteSize; j++)
 190         {
 191             if (i <  j) expected = UCOL_LESS;
 192             if (i == j) expected = UCOL_EQUAL;
 193             if (i >  j) expected = UCOL_GREATER;
 194             doTest(myCollation, testAcute[i], testAcute[j], expected );
 195         }
 196     }
 197     ucol_close(myCollation);
 198 }
 199
 200 static void TestExtra()
 201 {
 202     int32_t i, j;
 203     UErrorCode status = U_ZERO_ERROR;
 204     myCollation = ucol_open("fr_CA", &status);
 205     if(U_FAILURE(status)){
 206         log_err_status(status, "ERROR: in creation of rule based collator: %s\n", myErrorName(status));
 207         return;
 208     }
 209     log_verbose("Testing fr_CA Collation extra with secondary strength\n");
 210     ucol_setStrength(myCollation, UCOL_TERTIARY);
 211     for (i = 0; i < 9 ; i++)
 212     {
 213         for (j = i + 1; j < 10; j += 1)
 214         {
 215             doTest(myCollation, testBugs[i], testBugs[j], UCOL_LESS);
 216         }
 217     }
 218     ucol_close(myCollation);
 219 }
 220
 221 static void TestGetSortKey() {
 222     /* This is meant to test a buffer reallocation crash while using
 223     French secondary sorting with a large buffer.
 224     The fact that Japanese characters are used is irrelevant. */
 225     static const UChar pucUTF16[] = {
 226         0x3049,0x30b9,0x3088,0xfffd,0xfffd,0x308f,0xfffd,0x3042,
 227         0xfffd,0xfffd,0x305e,0xfffd,0x30b6,0x30bb,0x305b,0x30b1,
 228         0x3050,0x30af,0x304e,0x30bd,0xfffd,0x30c6,0xfffd,0xfffd,
 229         0x30e1,0xfffd,0xfffd,0x30d9,0xfffd,0x3092,0x3075,0x304a,
 230         0x3074,0x3070,0x30f5,0x30c4,0x306e,0x30df,0x3053,0xfffd,
 231         0x30a6,0x30b6,0x30e0,0xfffd,0x30bc,0x30ef,0x3087,0x30cc,
 232         0x305f,0x30de,0xfffd,0x3090,0x3063,0x30dc,0x30b6,0x30b9,
 233         0x30d2,0x3072,0x3061,0xfffd,0xfffd,0xfffd,0x307b,0x3092,
 234         0x30a5,0x30a9,0x30b1,0x30e7,0xfffd,0xfffd,0xfffd,0xfffd,
 235         0xfffd,0x305e,0xfffd,0x30c7,0x30ae,0x305b,0x308b,0x30c0,
 236         0x30f5,0xfffd,0xfffd,0xfffd,0x307d,0x304e,0xfffd,0xfffd,
 237         0x30c0,0x30c8,0x306f,0x307a,0x30dd,0x30e4,0x3084,0xfffd,
 238         0x308c,0x30f1,0xfffd,0x30c6,0xfffd,0x307a,0xfffd,0x3052,
 239         0x3056,0x305d,0x30b7,0xfffd,0x305b,0x30b0,0x30b9,0xfffd,
 240         0x30b2,0x306d,0x3044,0xfffd,0x3073,0xfffd,0x30be,0x30cf,
 241         0x3080,0xfffd,0x30a8,0x30f5,0x30a5,0x30c7,0x307c,0xfffd,
 242         0x30d1,0x305f,0x30b2,0xfffd,0x3053,0x30ca,0xfffd,0x30dd,
 243         0x3058,0x30c0,0x305d,0x30e1,0xfffd,0x30bb,0x305f,0x30d1,
 244         0x30f2,0x3058,0x3086,0x30ce,0x30db,0x30cb,0x30e9,0xfffd,
 245         0x308c,0xfffd,0xfffd,0x30af,0x30c4,0x3076,0x304c,0x30f5,
 246         0x30e8,0x308c,0xfffd,0x30e2,0x3073,0x30a3,0x304e,0x30ea,
 247         0xfffd,0x304f,0xfffd,0x306c,0x3044,0xfffd,0xfffd,0x30c9,
 248         0xfffd,0x30f5,0xfffd,0xfffd,0xfffd,0x30eb,0x30a8,0xfffd,
 249         0x306d,0x307d,0x30d8,0x3069,0xfffd,0xfffd,0x3086,0x30a9,
 250         0xfffd,0x3076,0x30e9,0x30cc,0x3074,0x30e0,0xfffd,0xfffd,
 251         0xfffd,0x30f0,0x3086,0x30ac,0x3076,0x3068,0x30c7,0xfffd,
 252         0x30b7,0x30d2,0x3048,0x308e,0x30e8,0x30d9,0x30ce,0x30d0,
 253         0x308b,0x30ee,0x30e6,0x3079,0x30f3,0x30af,0xfffd,0x3079,
 254         0xfffd,0xfffd,0x30ca,0x30bf,0xfffd,0x30b5,0xfffd,0xfffd,
 255         0x3093,0xfffd,0x30ba,0xfffd,0x3076,0x3047,0x304a,0xfffd,
 256         0xfffd,0x3086,0xfffd,0x3081,0xfffd,0x30f6,0x3066,0xfffd,
 257         0xfffd,0x30b6,0x30ef,0x30e2,0x30bf,0xfffd,0x3053,0x304a,
 258         0xfffd,0xfffd,0x304a,0x30e8,0xfffd,0x30e2,0xfffd,0xfffd,
 259         0x305c,0x3081,0x30c6,0xfffd,0x3091,0x3046,0x306a,0x3059,
 260         0xfffd,0xfffd,0x30dd,0x30d1,0x308a,0x30ee,0xfffd,0xfffd,
 261         0x308a,0x3042,0x30da,0xfffd,0x3064,0x30ef,0x305c,0x306b,
 262         0xfffd,0x30ca,0x3085,0x3067,0x30ea,0x30c2,0x30c8,0xfffd,
 263         0x30f5,0xfffd,0xfffd,0xfffd,0x30ca,0xfffd,0x3050,0x30f1,
 264         0x3050,0x3053,0x3072,0xfffd,0xfffd,0xfffd,0x3074,0xfffd,
 265         0x304b,0x30dd,0x306d,0xfffd,0x3049,0x30a1,0x30cc,0x30de,
 266         0x30ae,0x307b,0x308a,0xfffd,0x3065,0xfffd,0xfffd,0x30c0,
 267         0xfffd,0x3048,0x30dc,0x304f,0x3085,0x3059,0x304b,0x30d3,
 268         0x30eb,0x30a4,0x3073,0xfffd,0x30ba,0x308f,0x30a7,0x30c3,
 269         0x3074,0x30cf,0x306c,0x3053,0x30c0,0xfffd,0x3066,0xfffd,
 270         0x308f,0xfffd,0x30b5,0xfffd,0x3092,0x30c4,0xfffd,0x30d6,
 271         0x3056,0x30ad,0x30d2,0x30ba,0xfffd,0x30e6,0x304c,0x3088,
 272         0x30b6,0x3048,0x3077,0x30d1,0xfffd,0x3050,0xfffd,0x3042,
 273         0xfffd,0xfffd,0x308f,0xfffd,0x30c1,0xfffd,0x3074,0x3061,
 274         0x3056,0x30e5,0xfffd,0xfffd,0x3057,0xfffd,0xfffd,0xfffd,
 275         0xfffd,0x30bd,0x30b3,0x30ee,0xfffd,0x30f2,0x3084,0x3050,
 276         0xfffd,0x30e7,0xfffd,0xfffd,0x3060,0x3049,0x30f2,0x30ad,
 277         0x30bf,0x30f1,0x30a2,0xfffd,0x30af,0xfffd,0x3060,0x30a1,
 278         0x30e9,0x30c3,0xfffd,0x3072,0x3093,0x3070,0xfffd,0x308f,
 279         0x3060,0xfffd,0x3067,0x306f,0x3082,0x308b,0x3051,0xfffd,
 280         0x3058,0xfffd,0xfffd,0x30a8,0x3051,0x3054,0x30ad,0x30f0,
 281         0x3053,0xfffd,0x30e1,0x30d7,0x308d,0x307f,0x30be,0x30b0,
 282         0xfffd,0x30db,0xfffd,0x30d1,0xfffd,0x3054,0x30a5,0xfffd,
 283         0x306a,0xfffd,0x305c,0xfffd,0x3052,0x3088,0xfffd,0x306e,
 284         0xfffd,0x30a9,0x30a1,0x30b4,0x3083,0x30bd,0xfffd,0xfffd,
 285         0x306a,0x3070,0x30cd,0xfffd,0x3072,0x30ed,0x30c6,0x30be,
 286         0x30c4,0x305e,0x30b3,0x30e1,0x308a,0xfffd,0x305b,0xfffd,
 287         0x3042,0x3088,0xfffd,0x304c,0xfffd,0x3089,0x3071,0xfffd,
 288         0xfffd,0x30c6,0x3062,0x3079,0xfffd,0x304b,0x304a,0xfffd,
 289         0x30ad,0x3045,0x3045,0x3087,0xfffd,0x306a,0x308b,0x0000,
 290         0x30bd,0x3065,0x30b8,0x3086,0x30d3,0x3076,0xfffd,0xfffd,
 291         0x308f,0x3053,0x307c,0x3053,0x3084,0x30ae,0x30c4,0x3045,
 292         0x30a8,0x30d0,0x30e1,0x308c,0x30e6,0x30b7,0xfffd,0xfffd,
 293         0xfffd,0x3046,0x305f,0xfffd,0x3086,0x30ab,0xfffd,0xfffd,
 294         0x30c8,0xfffd,0x30a1,0x3052,0x3059,0xfffd,0x30a4,0xfffd,
 295         0xfffd,0x308c,0x3085,0x30ab,0x30b5,0x3091,0x30bf,0x30e3,
 296         0xfffd,0xfffd,0x3087,0xfffd,0x30f6,0x3051,0x30bd,0x3092,
 297         0x3063,0xfffd,0x30a9,0x3063,0x306e,0xfffd,0xfffd,0xfffd,
 298         0x306c,0xfffd,0x307e,0x30ad,0x3077,0x30c2,0x30e9,0x30d5,
 299         0xfffd,0xfffd,0x30c6,0x305c,0xfffd,0xfffd,0x3089,0xfffd,
 300         0x3048,0x30cb,0x308c,0xfffd,0xfffd,0x3044,0xfffd,0x3080,
 301         0x3063,0x3079,0xfffd,0x308a,0x30cb,0x3042,0x3057,0xfffd,
 302         0x307c,0x30c1,0x30a8,0x30cf,0xfffd,0x3083,0xfffd,0xfffd,
 303         0x306c,0xfffd,0x305e,0x3092,0xfffd,0x30dc,0x30b0,0x3081,
 304         0x30e3,0x30f0,0x304e,0x30cc,0x308e,0x30c4,0x30ad
 305     };
 306
 307     UErrorCode status = U_ZERO_ERROR;
 308     UCollator *pCollator;
 309     int32_t lenActualSortKey;
 310     uint8_t pucSortKey[4096];
 311     static const int32_t LENSORTKEY = (int32_t)sizeof(pucSortKey);
 312
 313     ucol_prepareShortStringOpen("LFR_AN_CX_EX_FO_HX_NX_S3", 0, NULL, &status);
 314
 315     pCollator = ucol_openFromShortString("LFR_AN_CX_EX_FO_HX_NX_S3", 0, NULL, &status);
 316
 317     if (U_FAILURE(status)) {
 318         log_data_err("error opening collator -> %s. (Are you missing data?)\n", u_errorName(status));
 319         return;
 320     }
 321
 322     lenActualSortKey = ucol_getSortKey(pCollator,
 323         (const UChar *)pucUTF16,
 324         UPRV_LENGTHOF(pucUTF16),
 325         pucSortKey,
 326         LENSORTKEY);
 327
 328     if (lenActualSortKey > LENSORTKEY) {
 329         log_err("sort key too big for original buffer. Got: %d Expected: %d\n", lenActualSortKey, LENSORTKEY);
 330         return;
 331     }
 332     /* If the test didn't crash, then the test succeeded. */
 333     ucol_close(pCollator);
 334 }
 335
 336 #endif /* #if !UCONFIG_NO_COLLATION */