source/test/cintltst/cfrtst.c

   1 // Copyright (C) 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /********************************************************************
   4  * COPYRIGHT:
   5  * Copyright (c) 1997-2016, International Business Machines Corporation and
   6  * others. All Rights Reserved.
   7  ********************************************************************/
   8 /********************************************************************************
   9 *
  10 * File CFRTST.C
  11 *
  12 * Modification History:
  13 *        Name                     Description
  14 *     Madhu Katragadda            Ported for C API
  15 *********************************************************************************/
  16 /**
  17  * CollationFrenchTest is a third level test class.  This tests the locale
  18  * specific primary, secondary and tertiary rules.  For example, the ignorable
  19  * character '-' in string "black-bird".  The en_US locale uses the default
  20  * collation rules as its sorting sequence.
  21  */
  22
  23 #include <stdlib.h>
  24
  25 #include "unicode/utypes.h"
  26
  27 #if !UCONFIG_NO_COLLATION
  28
  29 #include "unicode/ucol.h"
  30 #include "unicode/uloc.h"
  31 #include "cintltst.h"
  32 #include "ccolltst.h"
  33 #include "callcoll.h"
  34 #include "cfrtst.h"
  35 #include "cmemory.h"
  36 #include "unicode/ustring.h"
  37 #include "string.h"
  38
  39 static  UCollator *myCollation;
  40 const static UChar testSourceCases[][MAX_TOKEN_LEN] =
  41 {
  42     {0x0061/*'a'*/, 0x0062/*'b'*/, 0x0063/*'c'*/, 0x0000},
  43     {0x0043/*'C'*/, 0x004f/*'O'*/, 0x0054/*'T'*/, 0x0045/*'E'*/, 0x0000},
  44     {0x0063/*'c'*/, 0x006f/*'o'*/, 0x002d/*'-'*/, 0x006f/*'o'*/, 0x0070/*'p'*/, 0x0000},
  45     {0x0070/*'p'*/, 0x00EA, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x0065/*'e'*/, 0x0000},
  46     {0x0070/*'p'*/, 0x00EA, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x0065/*'e'*/, 0x0072/*'r'*/, 0x0000},
  47     {0x0070/*'p'*/, 0x00E9, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x0065/*'e'*/, 0x0072/*'r'*/, 0x0000},
  48     {0x0070/*'p'*/, 0x00E9, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x0065/*'e'*/, 0x0072/*'r'*/, 0x0000},
  49     {0x0048/*'H'*/, 0x0065/*'e'*/, 0x006c/*'l'*/, 0x006c/*'l'*/, 0x006f/*'o'*/, 0x0000},
  50     {0x01f1, 0x0000},
  51     {0xfb00, 0x0000},
  52     {0x01fa, 0x0000},
  53     {0x0101, 0x0000}
  54 };
  55
  56 const static UChar testTargetCases[][MAX_TOKEN_LEN] =
  57 {
  58     {0x0041/*'A'*/, 0x0042/*'B'*/, 0x0043/*'C'*/, 0x0000},
  59     {0x0063/*'c'*/, 0x00f4, 0x0074/*'t'*/, 0x0065/*'e'*/, 0x0000},
  60     {0x0043/*'C'*/, 0x004f/*'O'*/, 0x004f/*'O'*/, 0x0050/*'P'*/, 0x0000},
  61     {0x0070/*'p'*/, 0x00E9, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x00E9, 0x0000},
  62     {0x0070/*'p'*/,  0x00E9, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x00E9, 0x0000},
  63     {0x0070/*'p'*/, 0x00EA, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x0065/*'e'*/, 0x0000},
  64     {0x0070/*'p'*/, 0x00EA, 0x0063/*'c'*/, 0x0068/*'h'*/, 0x0065/*'e'*/, 0x0072/*'r'*/, 0x0000},
  65     {0x0068/*'h'*/, 0x0065/*'e'*/, 0x006c/*'l'*/, 0x006c/*'l'*/, 0x004f/*'O'*/, 0x0000},
  66     {0x01ee, 0x0000},
  67     {0x25ca, 0x0000},
  68     {0x00e0, 0x0000},
  69     {0x01df, 0x0000}
  70 };
  71
  72 const static UCollationResult results[] =
  73 {
  74     UCOL_LESS,
  75     UCOL_LESS,
  76     UCOL_LESS, /*UCOL_GREATER,*/
  77     UCOL_LESS,
  78     UCOL_GREATER,
  79     UCOL_GREATER,
  80     UCOL_LESS,
  81     UCOL_GREATER,
  82     UCOL_LESS, /*UCOL_GREATER,*/
  83     UCOL_GREATER,
  84     UCOL_LESS,
  85     UCOL_LESS
  86 };
  87
  88 /* 0x0300 is grave, 0x0301 is acute*/
  89 /* the order of elements in this array must be different than the order in CollationEnglishTest*/
  90 const static UChar testAcute[][MAX_TOKEN_LEN] =
  91 {
  92 /*00*/    {0x0065/*'e'*/, 0x0065/*'e'*/,  0x0000},
  93 /*01*/    {0x0065/*'e'*/, 0x0301, 0x0065/*'e'*/,  0x0000},
  94 /*02*/    {0x0065/*'e'*/, 0x0300, 0x0301, 0x0065/*'e'*/,  0x0000},
  95 /*03*/    {0x0065/*'e'*/, 0x0300, 0x0065/*'e'*/,  0x0000},
  96 /*04*/    {0x0065/*'e'*/, 0x0301, 0x0300, 0x0065/*'e'*/,  0x0000},
  97 /*05*/    {0x0065/*'e'*/, 0x0065/*'e'*/, 0x0301, 0x0000},
  98 /*06*/    {0x0065/*'e'*/, 0x0301, 0x0065/*'e'*/, 0x0301, 0x0000},
  99 /*07*/    {0x0065/*'e'*/, 0x0300, 0x0301, 0x0065/*'e'*/, 0x0301, 0x0000},
 100 /*08*/    {0x0065/*'e'*/, 0x0300, 0x0065/*'e'*/, 0x0301, 0x0000},
 101 /*09*/    {0x0065/*'e'*/, 0x0301, 0x0300, 0x0065/*'e'*/, 0x0301, 0x0000},
 102 /*0a*/    {0x0065/*'e'*/, 0x0065/*'e'*/, 0x0300, 0x0301, 0x0000},
 103 /*0b*/    {0x0065/*'e'*/, 0x0301, 0x0065/*'e'*/, 0x0300, 0x0301, 0x0000},
 104 /*0c*/    {0x0065/*'e'*/, 0x0300, 0x0301, 0x0065/*'e'*/, 0x0300, 0x0301, 0x0000},
 105 /*0d*/    {0x0065/*'e'*/, 0x0300, 0x0065/*'e'*/, 0x0300, 0x0301, 0x0000},
 106 /*0e*/    {0x0065/*'e'*/, 0x0301, 0x0300, 0x0065/*'e'*/, 0x0300, 0x0301, 0x0000},
 107 /*0f*/    {0x0065/*'e'*/, 0x0065/*'e'*/, 0x0300, 0x0000},
 108 /*10*/    {0x0065/*'e'*/, 0x0301, 0x0065/*'e'*/, 0x0300, 0x0000},
 109 /*11*/    {0x0065/*'e'*/, 0x0300, 0x0301, 0x0065/*'e'*/, 0x0300, 0x0000},
 110 /*12*/    {0x0065/*'e'*/, 0x0300, 0x0065/*'e'*/, 0x0300, 0x0000},
 111 /*13*/    {0x0065/*'e'*/, 0x0301, 0x0300, 0x0065/*'e'*/, 0x0300, 0x0000},
 112 /*14*/    {0x0065/*'e'*/, 0x0065/*'e'*/, 0x0301, 0x0300, 0x0000},
 113 /*15*/    {0x0065/*'e'*/, 0x0301, 0x0065/*'e'*/, 0x0301, 0x0300, 0x0000},
 114 /*16*/    {0x0065/*'e'*/, 0x0300, 0x0301, 0x0065/*'e'*/, 0x0301, 0x0300, 0x0000},
 115 /*17*/    {0x0065/*'e'*/, 0x0300, 0x0065/*'e'*/, 0x0301, 0x0300, 0x0000},
 116 /*18*/    {0x0065/*'e'*/, 0x0301, 0x0300, 0x0065/*'e'*/, 0x0301, 0x0300, 0x0000}
 117 };
 118
 119 const static UChar testBugs[][MAX_TOKEN_LEN] =
 120 {
 121     {0x0061/*'a'*/, 0x000},
 122     {0x0041/*'A'*/, 0x000},
 123     {0x0065/*'e'*/, 0x000},
 124     {0x0045/*'E'*/, 0x000},
 125     {0x00e9, 0x000},
 126     {0x00e8, 0x000},
 127     {0x00ea, 0x000},
 128     {0x00eb, 0x000},
 129     {0x0065/*'e'*/, 0x0061/*'a'*/, 0x000},
 130     {0x0078/*'x'*/, 0x000}
 131 };
 132
 133
 134 static void TestGetSortKey(void);
 135
 136
 137 void addFrenchCollTest(TestNode** root)
 138 {
 139     addTest(root, &TestSecondary, "tscoll/cfrtst/TestSecondary");
 140     addTest(root, &TestTertiary, "tscoll/cfrtst/TestTertiary");
 141     addTest(root, &TestExtra, "tscoll/cfrtst/TestExtra");
 142     addTest(root, &TestGetSortKey, "tscoll/cfrtst/TestGetSortKey");
 143 }
 144
 145
 146 static void TestTertiary( )
 147 {
 148
 149     int32_t i;
 150     UErrorCode status = U_ZERO_ERROR;
 151     myCollation = ucol_open("fr_CA", &status);
 152     if(U_FAILURE(status) || !myCollation){
 153         log_err_status(status, "ERROR: in creation of rule based collator: %s\n", myErrorName(status));
 154         return;
 155     }
 156
 157     ucol_setAttribute(myCollation, UCOL_ALTERNATE_HANDLING, UCOL_SHIFTED, &status);
 158     if(U_FAILURE(status)){
 159         log_err("ERROR: in creation of rule based collator: %s\n", myErrorName(status));
 160         return;
 161     }
 162     log_verbose("Testing fr_CA Collation with Tertiary strength\n");
 163     ucol_setStrength(myCollation, UCOL_QUATERNARY);
 164     for (i = 0; i < 12 ; i++)
 165     {
 166         doTest(myCollation, testSourceCases[i], testTargetCases[i], results[i]);
 167     }
 168     ucol_close(myCollation);
 169 }
 170
 171 static void TestSecondary()
 172 {
 173     int32_t i,j, testAcuteSize;
 174     UCollationResult expected=UCOL_EQUAL;
 175     UErrorCode status = U_ZERO_ERROR;
 176     myCollation = ucol_open("fr_CA", &status);
 177     if(U_FAILURE(status)){
 178         log_err_status(status, "ERROR: in creation of rule based collator: %s\n", myErrorName(status));
 179         return;
 180     }
 181     ucol_setAttribute(myCollation, UCOL_STRENGTH, UCOL_SECONDARY, &status);
 182     if(U_FAILURE(status)){
 183         log_err("ERROR: in creation of rule based collator: %s\n", myErrorName(status));
 184         return;
 185     }
 186     log_verbose("Testing fr_CA Collation with Secondary strength\n");
 187     /*test acute and grave ordering (compare to french collation)*/
 188     testAcuteSize = UPRV_LENGTHOF(testAcute);
 189     for (i = 0; i < testAcuteSize; i++)
 190     {
 191         for (j = 0; j < testAcuteSize; j++)
 192         {
 193             if (i <  j) expected = UCOL_LESS;
 194             if (i == j) expected = UCOL_EQUAL;
 195             if (i >  j) expected = UCOL_GREATER;
 196             doTest(myCollation, testAcute[i], testAcute[j], expected );
 197         }
 198     }
 199     ucol_close(myCollation);
 200 }
 201
 202 static void TestExtra()
 203 {
 204     int32_t i, j;
 205     UErrorCode status = U_ZERO_ERROR;
 206     myCollation = ucol_open("fr_CA", &status);
 207     if(U_FAILURE(status)){
 208         log_err_status(status, "ERROR: in creation of rule based collator: %s\n", myErrorName(status));
 209         return;
 210     }
 211     log_verbose("Testing fr_CA Collation extra with secondary strength\n");
 212     ucol_setStrength(myCollation, UCOL_TERTIARY);
 213     for (i = 0; i < 9 ; i++)
 214     {
 215         for (j = i + 1; j < 10; j += 1)
 216         {
 217             doTest(myCollation, testBugs[i], testBugs[j], UCOL_LESS);
 218         }
 219     }
 220     ucol_close(myCollation);
 221 }
 222
 223 static void TestGetSortKey() {
 224     /* This is meant to test a buffer reallocation crash while using
 225     French secondary sorting with a large buffer.
 226     The fact that Japanese characters are used is irrelevant. */
 227     static const UChar pucUTF16[] = {
 228         0x3049,0x30b9,0x3088,0xfffd,0xfffd,0x308f,0xfffd,0x3042,
 229         0xfffd,0xfffd,0x305e,0xfffd,0x30b6,0x30bb,0x305b,0x30b1,
 230         0x3050,0x30af,0x304e,0x30bd,0xfffd,0x30c6,0xfffd,0xfffd,
 231         0x30e1,0xfffd,0xfffd,0x30d9,0xfffd,0x3092,0x3075,0x304a,
 232         0x3074,0x3070,0x30f5,0x30c4,0x306e,0x30df,0x3053,0xfffd,
 233         0x30a6,0x30b6,0x30e0,0xfffd,0x30bc,0x30ef,0x3087,0x30cc,
 234         0x305f,0x30de,0xfffd,0x3090,0x3063,0x30dc,0x30b6,0x30b9,
 235         0x30d2,0x3072,0x3061,0xfffd,0xfffd,0xfffd,0x307b,0x3092,
 236         0x30a5,0x30a9,0x30b1,0x30e7,0xfffd,0xfffd,0xfffd,0xfffd,
 237         0xfffd,0x305e,0xfffd,0x30c7,0x30ae,0x305b,0x308b,0x30c0,
 238         0x30f5,0xfffd,0xfffd,0xfffd,0x307d,0x304e,0xfffd,0xfffd,
 239         0x30c0,0x30c8,0x306f,0x307a,0x30dd,0x30e4,0x3084,0xfffd,
 240         0x308c,0x30f1,0xfffd,0x30c6,0xfffd,0x307a,0xfffd,0x3052,
 241         0x3056,0x305d,0x30b7,0xfffd,0x305b,0x30b0,0x30b9,0xfffd,
 242         0x30b2,0x306d,0x3044,0xfffd,0x3073,0xfffd,0x30be,0x30cf,
 243         0x3080,0xfffd,0x30a8,0x30f5,0x30a5,0x30c7,0x307c,0xfffd,
 244         0x30d1,0x305f,0x30b2,0xfffd,0x3053,0x30ca,0xfffd,0x30dd,
 245         0x3058,0x30c0,0x305d,0x30e1,0xfffd,0x30bb,0x305f,0x30d1,
 246         0x30f2,0x3058,0x3086,0x30ce,0x30db,0x30cb,0x30e9,0xfffd,
 247         0x308c,0xfffd,0xfffd,0x30af,0x30c4,0x3076,0x304c,0x30f5,
 248         0x30e8,0x308c,0xfffd,0x30e2,0x3073,0x30a3,0x304e,0x30ea,
 249         0xfffd,0x304f,0xfffd,0x306c,0x3044,0xfffd,0xfffd,0x30c9,
 250         0xfffd,0x30f5,0xfffd,0xfffd,0xfffd,0x30eb,0x30a8,0xfffd,
 251         0x306d,0x307d,0x30d8,0x3069,0xfffd,0xfffd,0x3086,0x30a9,
 252         0xfffd,0x3076,0x30e9,0x30cc,0x3074,0x30e0,0xfffd,0xfffd,
 253         0xfffd,0x30f0,0x3086,0x30ac,0x3076,0x3068,0x30c7,0xfffd,
 254         0x30b7,0x30d2,0x3048,0x308e,0x30e8,0x30d9,0x30ce,0x30d0,
 255         0x308b,0x30ee,0x30e6,0x3079,0x30f3,0x30af,0xfffd,0x3079,
 256         0xfffd,0xfffd,0x30ca,0x30bf,0xfffd,0x30b5,0xfffd,0xfffd,
 257         0x3093,0xfffd,0x30ba,0xfffd,0x3076,0x3047,0x304a,0xfffd,
 258         0xfffd,0x3086,0xfffd,0x3081,0xfffd,0x30f6,0x3066,0xfffd,
 259         0xfffd,0x30b6,0x30ef,0x30e2,0x30bf,0xfffd,0x3053,0x304a,
 260         0xfffd,0xfffd,0x304a,0x30e8,0xfffd,0x30e2,0xfffd,0xfffd,
 261         0x305c,0x3081,0x30c6,0xfffd,0x3091,0x3046,0x306a,0x3059,
 262         0xfffd,0xfffd,0x30dd,0x30d1,0x308a,0x30ee,0xfffd,0xfffd,
 263         0x308a,0x3042,0x30da,0xfffd,0x3064,0x30ef,0x305c,0x306b,
 264         0xfffd,0x30ca,0x3085,0x3067,0x30ea,0x30c2,0x30c8,0xfffd,
 265         0x30f5,0xfffd,0xfffd,0xfffd,0x30ca,0xfffd,0x3050,0x30f1,
 266         0x3050,0x3053,0x3072,0xfffd,0xfffd,0xfffd,0x3074,0xfffd,
 267         0x304b,0x30dd,0x306d,0xfffd,0x3049,0x30a1,0x30cc,0x30de,
 268         0x30ae,0x307b,0x308a,0xfffd,0x3065,0xfffd,0xfffd,0x30c0,
 269         0xfffd,0x3048,0x30dc,0x304f,0x3085,0x3059,0x304b,0x30d3,
 270         0x30eb,0x30a4,0x3073,0xfffd,0x30ba,0x308f,0x30a7,0x30c3,
 271         0x3074,0x30cf,0x306c,0x3053,0x30c0,0xfffd,0x3066,0xfffd,
 272         0x308f,0xfffd,0x30b5,0xfffd,0x3092,0x30c4,0xfffd,0x30d6,
 273         0x3056,0x30ad,0x30d2,0x30ba,0xfffd,0x30e6,0x304c,0x3088,
 274         0x30b6,0x3048,0x3077,0x30d1,0xfffd,0x3050,0xfffd,0x3042,
 275         0xfffd,0xfffd,0x308f,0xfffd,0x30c1,0xfffd,0x3074,0x3061,
 276         0x3056,0x30e5,0xfffd,0xfffd,0x3057,0xfffd,0xfffd,0xfffd,
 277         0xfffd,0x30bd,0x30b3,0x30ee,0xfffd,0x30f2,0x3084,0x3050,
 278         0xfffd,0x30e7,0xfffd,0xfffd,0x3060,0x3049,0x30f2,0x30ad,
 279         0x30bf,0x30f1,0x30a2,0xfffd,0x30af,0xfffd,0x3060,0x30a1,
 280         0x30e9,0x30c3,0xfffd,0x3072,0x3093,0x3070,0xfffd,0x308f,
 281         0x3060,0xfffd,0x3067,0x306f,0x3082,0x308b,0x3051,0xfffd,
 282         0x3058,0xfffd,0xfffd,0x30a8,0x3051,0x3054,0x30ad,0x30f0,
 283         0x3053,0xfffd,0x30e1,0x30d7,0x308d,0x307f,0x30be,0x30b0,
 284         0xfffd,0x30db,0xfffd,0x30d1,0xfffd,0x3054,0x30a5,0xfffd,
 285         0x306a,0xfffd,0x305c,0xfffd,0x3052,0x3088,0xfffd,0x306e,
 286         0xfffd,0x30a9,0x30a1,0x30b4,0x3083,0x30bd,0xfffd,0xfffd,
 287         0x306a,0x3070,0x30cd,0xfffd,0x3072,0x30ed,0x30c6,0x30be,
 288         0x30c4,0x305e,0x30b3,0x30e1,0x308a,0xfffd,0x305b,0xfffd,
 289         0x3042,0x3088,0xfffd,0x304c,0xfffd,0x3089,0x3071,0xfffd,
 290         0xfffd,0x30c6,0x3062,0x3079,0xfffd,0x304b,0x304a,0xfffd,
 291         0x30ad,0x3045,0x3045,0x3087,0xfffd,0x306a,0x308b,0x0000,
 292         0x30bd,0x3065,0x30b8,0x3086,0x30d3,0x3076,0xfffd,0xfffd,
 293         0x308f,0x3053,0x307c,0x3053,0x3084,0x30ae,0x30c4,0x3045,
 294         0x30a8,0x30d0,0x30e1,0x308c,0x30e6,0x30b7,0xfffd,0xfffd,
 295         0xfffd,0x3046,0x305f,0xfffd,0x3086,0x30ab,0xfffd,0xfffd,
 296         0x30c8,0xfffd,0x30a1,0x3052,0x3059,0xfffd,0x30a4,0xfffd,
 297         0xfffd,0x308c,0x3085,0x30ab,0x30b5,0x3091,0x30bf,0x30e3,
 298         0xfffd,0xfffd,0x3087,0xfffd,0x30f6,0x3051,0x30bd,0x3092,
 299         0x3063,0xfffd,0x30a9,0x3063,0x306e,0xfffd,0xfffd,0xfffd,
 300         0x306c,0xfffd,0x307e,0x30ad,0x3077,0x30c2,0x30e9,0x30d5,
 301         0xfffd,0xfffd,0x30c6,0x305c,0xfffd,0xfffd,0x3089,0xfffd,
 302         0x3048,0x30cb,0x308c,0xfffd,0xfffd,0x3044,0xfffd,0x3080,
 303         0x3063,0x3079,0xfffd,0x308a,0x30cb,0x3042,0x3057,0xfffd,
 304         0x307c,0x30c1,0x30a8,0x30cf,0xfffd,0x3083,0xfffd,0xfffd,
 305         0x306c,0xfffd,0x305e,0x3092,0xfffd,0x30dc,0x30b0,0x3081,
 306         0x30e3,0x30f0,0x304e,0x30cc,0x308e,0x30c4,0x30ad
 307     };
 308
 309     UErrorCode status = U_ZERO_ERROR;
 310     UCollator *pCollator;
 311     int32_t lenActualSortKey;
 312     uint8_t pucSortKey[4096];
 313     static const int32_t LENSORTKEY = (int32_t)sizeof(pucSortKey);
 314
 315     ucol_prepareShortStringOpen("LFR_AN_CX_EX_FO_HX_NX_S3", 0, NULL, &status);
 316
 317     pCollator = ucol_openFromShortString("LFR_AN_CX_EX_FO_HX_NX_S3", 0, NULL, &status);
 318
 319     if (U_FAILURE(status)) {
 320         log_data_err("error opening collator -> %s. (Are you missing data?)\n", u_errorName(status));
 321         return;
 322     }
 323
 324     lenActualSortKey = ucol_getSortKey(pCollator,
 325         (const UChar *)pucUTF16,
 326         UPRV_LENGTHOF(pucUTF16),
 327         pucSortKey,
 328         LENSORTKEY);
 329
 330     if (lenActualSortKey > LENSORTKEY) {
 331         log_err("sort key too big for original buffer. Got: %d Expected: %d\n", lenActualSortKey, LENSORTKEY);
 332         return;
 333     }
 334     /* If the test didn't crash, then the test succeeded. */
 335     ucol_close(pCollator);
 336 }
 337
 338 #endif /* #if !UCONFIG_NO_COLLATION */