icuSources/test/cintltst/cg7coll.c

   1 /********************************************************************
   2  * COPYRIGHT:
   3  * Copyright (c) 1997-2014, International Business Machines Corporation and
   4  * others. All Rights Reserved.
   5  ********************************************************************/
   6 /********************************************************************************
   7 *
   8 * File CG7COLL.C
   9 *
  10 * Modification History:
  11 *        Name                     Description
  12 *     Madhu Katragadda            Ported for C API
  13 *********************************************************************************/
  14 /**
  15  * G7CollationTest is a third level test class.  This test performs the examples
  16  * mentioned on the IBM Java international demos web site.
  17  * Sample Rules: & Z < p , P
  18  * Effect :  Making P sort after Z.
  19  *
  20  * Sample Rules: & c < ch , cH, Ch, CH
  21  * Effect : As well as adding sequences of characters that act as a single character (this is
  22  * known as contraction), you can also add characters that act like a sequence of
  23  * characters (this is known as expansion).
  24  *
  25  * Sample Rules: & Question'-'mark ; '?' & Hash'-'mark ; '#' & Ampersand ; '&'
  26  * Effect : Expansion and contraction can actually be combined.
  27  *
  28  * Sample Rules: & aa ; a'-' & ee ; e'-' & ii ; i'-' & oo ; o'-' & uu ; u'-'
  29  * Effect : sorted sequence as the following,
  30  * aardvark
  31  * a-rdvark
  32  * abbot
  33  * coop
  34  * co-p
  35  * cop
  36  */
  37 #include <stdlib.h>
  38 #include <string.h>
  39 #include <stdio.h>
  40
  41 #include "unicode/utypes.h"
  42
  43 #if !UCONFIG_NO_COLLATION
  44
  45 #include "unicode/ucol.h"
  46 #include "unicode/uloc.h"
  47 #include "cintltst.h"
  48 #include "cg7coll.h"
  49 #include "ccolltst.h"
  50 #include "callcoll.h"
  51 #include "unicode/ustring.h"
  52
  53 const char* locales[8] = {
  54         "en_US",
  55         "en_GB",
  56         "en_CA",
  57         "fr_FR",
  58         "fr_CA",
  59         "de_DE",
  60         "it_IT",
  61         "ja_JP"
  62 };
  63
  64
  65
  66 const static UChar testCases[][MAX_TOKEN_LEN] = {
  67     {  0x0062 /*'b'*/, 0x006c /*'l'*/, 0x0061 /*'a'*/, 0x0062 /*'c'*/, 0x006b /*'k'*/,
  68         0x0062 /*'b'*/, 0x0069 /*'i'*/, 0x0072 /*'r'*/, 0x0064 /*'d'*/, 0x0073 /*'s'*/, 0x0000},                    /* 9 */
  69     { 0x0050 /*'P'*/, 0x0061 /*'a'*/, 0x0074/*'t'*/, 0x0000},                                                    /* 1 */
  70     { 0x0070 /*'p'*/, 0x00E9, 0x0063 /*'c'*/, 0x0068 /*'h'*/, 0x00E9, 0x0000},                                    /* 2 */
  71     { 0x0070 /*'p'*/, 0x00EA, 0x0063 /*'c'*/, 0x0068 /*'h'*/, 0x0065 /*'e'*/, 0x0000},                           /* 3 */
  72     { 0x0070 /*'p'*/, 0x00E9, 0x0063 /*'c'*/, 0x0068 /*'h'*/, 0x0065 /*'e'*/, 0x0072 /*'r'*/, 0x0000},            /* 4 */
  73     { 0x0070 /*'p'*/, 0x00EA, 0x0063 /*'c'*/, 0x0068 /*'h'*/, 0x0065 /*'e'*/, 0x0072 /*'r'*/, 0x0000},            /* 5 */
  74     { 0x0054 /*'T'*/, 0x006f /*'o'*/, 0x0064 /*'d'*/, 0x0000},                                                    /* 6 */
  75     { 0x0054 /*'T'*/, 0x00F6, 0x006e /*'n'*/, 0x0065 /*'e'*/, 0x0000},                                            /* 7 */
  76     { 0x0054 /*'T'*/, 0x006f /*'o'*/, 0x0066 /*'f'*/, 0x0075 /*'u'*/, 0x0000},                                   /* 8 */
  77     { 0x0062 /*'b'*/, 0x006c /*'l'*/, 0x0061 /*'a'*/, 0x0062 /*'c'*/, 0x006b /*'k'*/,
  78       0x0062  /*'b'*/, 0x0069 /*'i'*/, 0x0072 /*'r'*/, 0x0064 /*'d'*/, 0x0000},                                    /* 12 */
  79     { 0x0054 /*'T'*/, 0x006f /*'o'*/, 0x006e /*'n'*/, 0x0000},                                                    /* 10 */
  80     { 0x0050  /*'P'*/, 0x0041 /*'A'*/, 0x0054 /*'T'*/, 0x0000},                                                    /* 11 */
  81     { 0x0062 /*'b'*/, 0x006c /*'l'*/, 0x0061 /*'a'*/, 0x0062 /*'c'*/, 0x006b /*'k'*/,
  82         0x002d /*'-'*/,  0x0062 /*'b'*/, 0x0069 /*'i'*/, 0x0072 /*'r'*/, 0x0064 /*'d'*/, 0x0000},                /* 13 */
  83     { 0x0062 /*'b'*/, 0x006c /*'l'*/, 0x0061 /*'a'*/, 0x0062 /*'c'*/, 0x006b /*'k'*/,
  84         0x002d /*'-'*/,  0x0062 /*'b'*/, 0x0069 /*'i'*/, 0x0072 /*'r'*/, 0x0064 /*'d'*/, 0x0073/*'s'*/, 0x0000},  /* 0 */
  85     {0x0070 /*'p'*/, 0x0061 /*'a'*/, 0x0074 /*'t'*/, 0x0000},                                                    /* 14 */
  86     /* Additional tests */
  87     { 0x0063 /*'c'*/, 0x007a /*'z'*/, 0x0061 /*'a'*/, 0x0072 /*'r'*/, 0x0000 },                                 /* 15 */
  88     { 0x0063 /*'c'*/, 0x0068 /*'h'*/, 0x0075 /*'u'*/, 0x0072 /*'r'*/, 0x006f /*'o'*/, 0x0000 },                  /* 16 */
  89     { 0x0063 /*'c'*/, 0x0061 /*'a'*/, 0x0074 /*'t'*/, 0x000 },                                                    /* 17 */
  90     { 0x0064 /*'d'*/, 0x0061 /*'a'*/, 0x0072 /*'r'*/, 0x006e /*'n'*/, 0x0000 },                                 /* 18 */
  91     { 0x003f /*'?'*/, 0x0000 },                                                                                /* 19 */
  92     { 0x0071 /*'q'*/, 0x0075 /*'u'*/, 0x0069 /*'i'*/, 0x0063 /*'c'*/, 0x006b /*'k'*/, 0x0000 },                  /* 20 */
  93     { 0x0023 /*'#'*/, 0x0000 },                                                                                /* 21 */
  94     { 0x0026 /*'&'*/, 0x0000 },                                                                                /* 22 */
  95     {  0x0061 /*'a'*/, 0x002d /*'-'*/, 0x0072 /*'r'*/, 0x0064 /*'d'*/, 0x0076 /*'v'*/, 0x0061 /*'a'*/,
  96                 0x0072/*'r'*/, 0x006b/*'k'*/, 0x0000},                                                        /* 24 */
  97     { 0x0061 /*'a'*/, 0x0061 /*'a'*/, 0x0072 /*'r'*/, 0x0064 /*'d'*/, 0x0076 /*'v'*/, 0x0061 /*'a'*/,
  98                 0x0072/*'r'*/, 0x006b/*'k'*/, 0x0000},                                                        /* 23 */
  99     { 0x0061 /*'a'*/, 0x0062 /*'b'*/, 0x0062 /*'b'*/, 0x006f /*'o'*/, 0x0074 /*'t'*/, 0x0000},                   /* 25 */
 100     { 0x0063 /*'c'*/, 0x006f /*'o'*/, 0x002d /*'-'*/, 0x0070 /*'p'*/, 0x0000},                                 /* 27 */
 101     { 0x0063 /*'c'*/, 0x006f  /*'o'*/, 0x0070 /*'p'*/, 0x0000},                                                /* 28 */
 102     { 0x0063 /*'c'*/, 0x006f /*'o'*/, 0x006f /*'o'*/, 0x0070 /*'p'*/, 0x0000},                                 /* 26 */
 103     { 0x007a /*'z'*/, 0x0065  /*'e'*/, 0x0062 /*'b'*/, 0x0072 /*'r'*/, 0x0061 /*'a'*/, 0x0000}                    /* 29 */
 104 };
 105
 106 const static int32_t results[TESTLOCALES][TOTALTESTSET] = {
 107     { 12, 13, 9, 0, 14, 1, 11, 2, 3, 4, 5, 6, 8, 10, 7, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31 }, /* en_US */
 108     { 12, 13, 9, 0, 14, 1, 11, 2, 3, 4, 5, 6, 8, 10, 7, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31 }, /* en_GB */
 109     { 12, 13, 9, 0, 14, 1, 11, 2, 3, 4, 5, 6, 8, 10, 7, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31 }, /* en_CA */
 110     { 12, 13, 9, 0, 14, 1, 11, 2, 3, 4, 5, 6, 8, 10, 7, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31 }, /* fr_FR */
 111     { 12, 13, 9, 0, 14, 1, 11, 3, 2, 4, 5, 6, 8, 10, 7, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31 }, /* fr_CA */
 112     { 12, 13, 9, 0, 14, 1, 11, 2, 3, 4, 5, 6, 8, 10, 7, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31 }, /* de_DE */
 113     { 12, 13, 9, 0, 14, 1, 11, 2, 3, 4, 5, 6, 8, 10, 7, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31 }, /* it_IT */
 114     { 12, 13, 9, 0, 14, 1, 11, 2, 3, 4, 5, 6, 8, 10, 7, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31 }, /* ja_JP */
 115     /* new table collation with rules "& Z < p, P"  loop to FIXEDTESTSET */
 116     { 12, 13, 9, 0, 6, 8, 10, 7, 14, 1, 11, 2, 3, 4, 5, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31 },
 117     /* new table collation with rules "& C < ch , cH, Ch, CH " loop to TOTALTESTSET */
 118     { 19, 22, 21, 23, 24, 25, 12, 13, 9, 0, 17, 26, 28, 27, 15, 16, 18, 14, 1, 11, 2, 3, 4, 5, 20, 6, 8, 10, 7, 29 },
 119     /* new table collation with rules "& Question-mark ; ? & Hash-mark ; # & Ampersand ; '&'  " loop to TOTALTESTSET */
 120     { 23, 24, 25, 22, 12, 13, 9, 0, 17, 16, 26, 28, 27, 15, 18, 21, 14, 1, 11, 2, 3, 4, 5, 19, 20, 6, 8, 10, 7, 29 },
 121     /* analogous to Japanese rules " & aa ; a- & ee ; e- & ii ; i- & oo ; o- & uu ; u- " */  /* loop to TOTALTESTSET */
 122     { 19, 22, 21, 24, 23, 25, 12, 13, 9, 0, 17, 16, 28, 26, 27, 15, 18, 14, 1, 11, 2, 3, 4, 5, 20, 6, 8, 10, 7, 29 }
 123 };
 124
 125 void addRuleBasedCollTest(TestNode** root)
 126 {
 127     addTest(root, &TestG7Locales, "tscoll/cg7coll/TestG7Locales");
 128     addTest(root, &TestDemo1, "tscoll/cg7coll/TestDemo1");
 129     addTest(root, &TestDemo2, "tscoll/cg7coll/TestDemo2");
 130     addTest(root, &TestDemo3, "tscoll/cg7coll/TestDemo3");
 131     addTest(root, &TestDemo4, "tscoll/cg7coll/TestDemo4");
 132
 133
 134 }
 135
 136 static void TestG7Locales()
 137 {
 138     UCollator *myCollation;
 139     UErrorCode status = U_ZERO_ERROR;
 140     const UChar *defRules;
 141     int32_t i, rlen, j, n;
 142     log_verbose("Testing  ucol_openRules for all the locales\n");
 143     for (i = 0; i < UPRV_LENGTHOF(locales); i++)
 144     {
 145         const char *locale = locales[i];
 146         status = U_ZERO_ERROR;
 147         myCollation = ucol_open(locale, &status);
 148         ucol_setAttribute(myCollation, UCOL_STRENGTH, UCOL_QUATERNARY, &status);
 149         ucol_setAttribute(myCollation, UCOL_ALTERNATE_HANDLING, UCOL_SHIFTED, &status);
 150
 151         if (U_FAILURE(status))
 152         {
 153             log_err_status(status, "Error in creating collator in %s:  %s\n", locale, myErrorName(status));
 154             ucol_close(myCollation);
 155             continue;
 156         }
 157
 158         defRules = ucol_getRules(myCollation, &rlen);
 159         if (rlen == 0 && (strcmp(locale, "fr_CA") == 0 || strcmp(locale, "ja_JP") == 0)) {
 160             log_data_err("%s UCollator missing rule string\n", locale);
 161             if (log_knownIssue("10671", "TestG7Locales does not test ignore-punctuation")) {
 162                 ucol_close(myCollation);
 163                 continue;
 164             }
 165         } else {
 166             UCollator *tblColl1;
 167             status = U_ZERO_ERROR;
 168             tblColl1 = ucol_openRules(defRules, rlen, UCOL_OFF,
 169                     UCOL_DEFAULT_STRENGTH,NULL, &status);
 170             ucol_close(myCollation);
 171             if (U_FAILURE(status))
 172             {
 173                 log_err_status(status, "Error in creating collator in %s:  %s\n", locale, myErrorName(status));
 174                 continue;
 175             }
 176             myCollation = tblColl1;
 177         }
 178
 179         log_verbose("Locale  %s\n", locales[i]);
 180         log_verbose("  tests start...\n");
 181
 182         j = 0;
 183         n = 0;
 184         for (j = 0; j < FIXEDTESTSET; j++)
 185         {
 186             for (n = j+1; n < FIXEDTESTSET; n++)
 187             {
 188                 doTest(myCollation, testCases[results[i][j]], testCases[results[i][n]], UCOL_LESS);
 189             }
 190         }
 191
 192         ucol_close(myCollation);
 193     }
 194 }
 195
 196 static void TestDemo1()
 197 {
 198     UCollator *myCollation;
 199     int32_t j, n;
 200     static const char rules[] = "& Z < p, P";
 201     int32_t len=(int32_t)strlen(rules);
 202     UChar temp[sizeof(rules)];
 203     UErrorCode status = U_ZERO_ERROR;
 204     u_uastrcpy(temp, rules);
 205
 206     log_verbose("Demo Test 1 : Create a new table collation with rules \" & Z < p, P \" \n");
 207
 208     myCollation = ucol_openRules(temp, len, UCOL_OFF, UCOL_DEFAULT_STRENGTH,NULL, &status);
 209
 210     if (U_FAILURE(status))
 211     {
 212         log_err_status(status, "Demo Test 1 Rule collation object creation failed. : %s\n", myErrorName(status));
 213         return;
 214     }
 215
 216     for (j = 0; j < FIXEDTESTSET; j++)
 217     {
 218         for (n = j+1; n < FIXEDTESTSET; n++)
 219         {
 220             doTest(myCollation, testCases[results[8][j]], testCases[results[8][n]], UCOL_LESS);
 221         }
 222     }
 223
 224     ucol_close(myCollation);
 225 }
 226
 227 static void TestDemo2()
 228 {
 229     UCollator *myCollation;
 230     int32_t j, n;
 231     static const char rules[] = "& C < ch , cH, Ch, CH";
 232     int32_t len=(int32_t)strlen(rules);
 233     UChar temp[sizeof(rules)];
 234     UErrorCode status = U_ZERO_ERROR;
 235     u_uastrcpy(temp, rules);
 236
 237     log_verbose("Demo Test 2 : Create a new table collation with rules \"& C < ch , cH, Ch, CH\"");
 238
 239     myCollation = ucol_openRules(temp, len, UCOL_OFF, UCOL_DEFAULT_STRENGTH, NULL, &status);
 240
 241     if (U_FAILURE(status))
 242     {
 243         log_err_status(status, "Demo Test 2 Rule collation object creation failed.: %s\n", myErrorName(status));
 244         return;
 245     }
 246     for (j = 0; j < TOTALTESTSET; j++)
 247     {
 248         for (n = j+1; n < TOTALTESTSET; n++)
 249         {
 250             doTest(myCollation, testCases[results[9][j]], testCases[results[9][n]], UCOL_LESS);
 251         }
 252     }
 253     ucol_close(myCollation);
 254 }
 255
 256 static void TestDemo3()
 257 {
 258     UCollator *myCollation;
 259     int32_t j, n;
 260     static const char rules[] = "& Question'-'mark ; '?' & Hash'-'mark ; '#' & Ampersand ; '&'";
 261     int32_t len=(int32_t)strlen(rules);
 262     UChar temp[sizeof(rules)];
 263     UErrorCode status = U_ZERO_ERROR;
 264     u_uastrcpy(temp, rules);
 265
 266     log_verbose("Demo Test 3 : Create a new table collation with rules \"& Question'-'mark ; '?' & Hash'-'mark ; '#' & Ampersand ; '&'\" \n");
 267
 268     myCollation = ucol_openRules(temp, len, UCOL_OFF, UCOL_DEFAULT_STRENGTH, NULL, &status);
 269
 270     if (U_FAILURE(status))
 271     {
 272         log_err_status(status, "Demo Test 3 Rule collation object creation failed.: %s\n", myErrorName(status));
 273         return;
 274     }
 275
 276     for (j = 0; j < TOTALTESTSET; j++)
 277     {
 278         for (n = j+1; n < TOTALTESTSET; n++)
 279         {
 280             doTest(myCollation, testCases[results[10][j]], testCases[results[10][n]], UCOL_LESS);
 281         }
 282     }
 283     ucol_close(myCollation);
 284 }
 285
 286 static void TestDemo4()
 287 {
 288     UCollator *myCollation;
 289     int32_t j, n;
 290     static const char rules[] = " & aa ; a'-' & ee ; e'-' & ii ; i'-' & oo ; o'-' & uu ; u'-' ";
 291     int32_t len=(int32_t)strlen(rules);
 292     UChar temp[sizeof(rules)];
 293     UErrorCode status = U_ZERO_ERROR;
 294     u_uastrcpy(temp, rules);
 295
 296     log_verbose("Demo Test 4 : Create a new table collation with rules \" & aa ; a'-' & ee ; e'-' & ii ; i'-' & oo ; o'-' & uu ; u'-' \"\n");
 297
 298     myCollation = ucol_openRules(temp, len, UCOL_OFF, UCOL_DEFAULT_STRENGTH, NULL, &status);
 299
 300     if (U_FAILURE(status))
 301     {
 302         log_err_status(status, "Demo Test 4 Rule collation object creation failed.: %s\n", myErrorName(status));
 303         return;
 304     }
 305     for (j = 0; j < TOTALTESTSET; j++)
 306     {
 307         for (n = j+1; n < TOTALTESTSET; n++)
 308         {
 309             doTest(myCollation, testCases[results[11][j]], testCases[results[11][n]], UCOL_LESS);
 310         }
 311     }
 312     ucol_close(myCollation);
 313 }
 314
 315 #endif /* #if !UCONFIG_NO_COLLATION */