icuSources/test/cintltst/cg7coll.c

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /********************************************************************
   4  * COPYRIGHT:
   5  * Copyright (c) 1997-2014, International Business Machines Corporation and
   6  * others. All Rights Reserved.
   7  ********************************************************************/
   8 /********************************************************************************
   9 *
  10 * File CG7COLL.C
  11 *
  12 * Modification History:
  13 *        Name                     Description
  14 *     Madhu Katragadda            Ported for C API
  15 *********************************************************************************/
  16 /**
  17  * G7CollationTest is a third level test class.  This test performs the examples
  18  * mentioned on the IBM Java international demos web site.
  19  * Sample Rules: & Z < p , P
  20  * Effect :  Making P sort after Z.
  21  *
  22  * Sample Rules: & c < ch , cH, Ch, CH
  23  * Effect : As well as adding sequences of characters that act as a single character (this is
  24  * known as contraction), you can also add characters that act like a sequence of
  25  * characters (this is known as expansion).
  26  *
  27  * Sample Rules: & Question'-'mark ; '?' & Hash'-'mark ; '#' & Ampersand ; '&'
  28  * Effect : Expansion and contraction can actually be combined.
  29  *
  30  * Sample Rules: & aa ; a'-' & ee ; e'-' & ii ; i'-' & oo ; o'-' & uu ; u'-'
  31  * Effect : sorted sequence as the following,
  32  * aardvark
  33  * a-rdvark
  34  * abbot
  35  * coop
  36  * co-p
  37  * cop
  38  */
  39 #include <stdlib.h>
  40 #include <string.h>
  41 #include <stdio.h>
  42
  43 #include "unicode/utypes.h"
  44
  45 #if !UCONFIG_NO_COLLATION
  46
  47 #include "unicode/ucol.h"
  48 #include "unicode/uloc.h"
  49 #include "cintltst.h"
  50 #include "cg7coll.h"
  51 #include "ccolltst.h"
  52 #include "callcoll.h"
  53 #include "unicode/ustring.h"
  54
  55 const char* locales[8] = {
  56         "en_US",
  57         "en_GB",
  58         "en_CA",
  59         "fr_FR",
  60         "fr_CA",
  61         "de_DE",
  62         "it_IT",
  63         "ja_JP"
  64 };
  65
  66
  67
  68 const static UChar testCases[][MAX_TOKEN_LEN] = {
  69     {  0x0062 /*'b'*/, 0x006c /*'l'*/, 0x0061 /*'a'*/, 0x0062 /*'c'*/, 0x006b /*'k'*/,
  70         0x0062 /*'b'*/, 0x0069 /*'i'*/, 0x0072 /*'r'*/, 0x0064 /*'d'*/, 0x0073 /*'s'*/, 0x0000},                    /* 9 */
  71     { 0x0050 /*'P'*/, 0x0061 /*'a'*/, 0x0074/*'t'*/, 0x0000},                                                    /* 1 */
  72     { 0x0070 /*'p'*/, 0x00E9, 0x0063 /*'c'*/, 0x0068 /*'h'*/, 0x00E9, 0x0000},                                    /* 2 */
  73     { 0x0070 /*'p'*/, 0x00EA, 0x0063 /*'c'*/, 0x0068 /*'h'*/, 0x0065 /*'e'*/, 0x0000},                           /* 3 */
  74     { 0x0070 /*'p'*/, 0x00E9, 0x0063 /*'c'*/, 0x0068 /*'h'*/, 0x0065 /*'e'*/, 0x0072 /*'r'*/, 0x0000},            /* 4 */
  75     { 0x0070 /*'p'*/, 0x00EA, 0x0063 /*'c'*/, 0x0068 /*'h'*/, 0x0065 /*'e'*/, 0x0072 /*'r'*/, 0x0000},            /* 5 */
  76     { 0x0054 /*'T'*/, 0x006f /*'o'*/, 0x0064 /*'d'*/, 0x0000},                                                    /* 6 */
  77     { 0x0054 /*'T'*/, 0x00F6, 0x006e /*'n'*/, 0x0065 /*'e'*/, 0x0000},                                            /* 7 */
  78     { 0x0054 /*'T'*/, 0x006f /*'o'*/, 0x0066 /*'f'*/, 0x0075 /*'u'*/, 0x0000},                                   /* 8 */
  79     { 0x0062 /*'b'*/, 0x006c /*'l'*/, 0x0061 /*'a'*/, 0x0062 /*'c'*/, 0x006b /*'k'*/,
  80       0x0062  /*'b'*/, 0x0069 /*'i'*/, 0x0072 /*'r'*/, 0x0064 /*'d'*/, 0x0000},                                    /* 12 */
  81     { 0x0054 /*'T'*/, 0x006f /*'o'*/, 0x006e /*'n'*/, 0x0000},                                                    /* 10 */
  82     { 0x0050  /*'P'*/, 0x0041 /*'A'*/, 0x0054 /*'T'*/, 0x0000},                                                    /* 11 */
  83     { 0x0062 /*'b'*/, 0x006c /*'l'*/, 0x0061 /*'a'*/, 0x0062 /*'c'*/, 0x006b /*'k'*/,
  84         0x002d /*'-'*/,  0x0062 /*'b'*/, 0x0069 /*'i'*/, 0x0072 /*'r'*/, 0x0064 /*'d'*/, 0x0000},                /* 13 */
  85     { 0x0062 /*'b'*/, 0x006c /*'l'*/, 0x0061 /*'a'*/, 0x0062 /*'c'*/, 0x006b /*'k'*/,
  86         0x002d /*'-'*/,  0x0062 /*'b'*/, 0x0069 /*'i'*/, 0x0072 /*'r'*/, 0x0064 /*'d'*/, 0x0073/*'s'*/, 0x0000},  /* 0 */
  87     {0x0070 /*'p'*/, 0x0061 /*'a'*/, 0x0074 /*'t'*/, 0x0000},                                                    /* 14 */
  88     /* Additional tests */
  89     { 0x0063 /*'c'*/, 0x007a /*'z'*/, 0x0061 /*'a'*/, 0x0072 /*'r'*/, 0x0000 },                                 /* 15 */
  90     { 0x0063 /*'c'*/, 0x0068 /*'h'*/, 0x0075 /*'u'*/, 0x0072 /*'r'*/, 0x006f /*'o'*/, 0x0000 },                  /* 16 */
  91     { 0x0063 /*'c'*/, 0x0061 /*'a'*/, 0x0074 /*'t'*/, 0x000 },                                                    /* 17 */
  92     { 0x0064 /*'d'*/, 0x0061 /*'a'*/, 0x0072 /*'r'*/, 0x006e /*'n'*/, 0x0000 },                                 /* 18 */
  93     { 0x003f /*'?'*/, 0x0000 },                                                                                /* 19 */
  94     { 0x0071 /*'q'*/, 0x0075 /*'u'*/, 0x0069 /*'i'*/, 0x0063 /*'c'*/, 0x006b /*'k'*/, 0x0000 },                  /* 20 */
  95     { 0x0023 /*'#'*/, 0x0000 },                                                                                /* 21 */
  96     { 0x0026 /*'&'*/, 0x0000 },                                                                                /* 22 */
  97     {  0x0061 /*'a'*/, 0x002d /*'-'*/, 0x0072 /*'r'*/, 0x0064 /*'d'*/, 0x0076 /*'v'*/, 0x0061 /*'a'*/,
  98                 0x0072/*'r'*/, 0x006b/*'k'*/, 0x0000},                                                        /* 24 */
  99     { 0x0061 /*'a'*/, 0x0061 /*'a'*/, 0x0072 /*'r'*/, 0x0064 /*'d'*/, 0x0076 /*'v'*/, 0x0061 /*'a'*/,
 100                 0x0072/*'r'*/, 0x006b/*'k'*/, 0x0000},                                                        /* 23 */
 101     { 0x0061 /*'a'*/, 0x0062 /*'b'*/, 0x0062 /*'b'*/, 0x006f /*'o'*/, 0x0074 /*'t'*/, 0x0000},                   /* 25 */
 102     { 0x0063 /*'c'*/, 0x006f /*'o'*/, 0x002d /*'-'*/, 0x0070 /*'p'*/, 0x0000},                                 /* 27 */
 103     { 0x0063 /*'c'*/, 0x006f  /*'o'*/, 0x0070 /*'p'*/, 0x0000},                                                /* 28 */
 104     { 0x0063 /*'c'*/, 0x006f /*'o'*/, 0x006f /*'o'*/, 0x0070 /*'p'*/, 0x0000},                                 /* 26 */
 105     { 0x007a /*'z'*/, 0x0065  /*'e'*/, 0x0062 /*'b'*/, 0x0072 /*'r'*/, 0x0061 /*'a'*/, 0x0000}                    /* 29 */
 106 };
 107
 108 const static int32_t results[TESTLOCALES][TOTALTESTSET] = {
 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_US */
 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 }, /* en_GB */
 111     { 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 */
 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 }, /* fr_FR */
 113     { 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 */
 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 }, /* de_DE */
 115     { 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 */
 116     { 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 */
 117     /* new table collation with rules "& Z < p, P"  loop to FIXEDTESTSET */
 118     { 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 },
 119     /* new table collation with rules "& C < ch , cH, Ch, CH " loop to TOTALTESTSET */
 120     { 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 },
 121     /* new table collation with rules "& Question-mark ; ? & Hash-mark ; # & Ampersand ; '&'  " loop to TOTALTESTSET */
 122     { 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 },
 123     /* analogous to Japanese rules " & aa ; a- & ee ; e- & ii ; i- & oo ; o- & uu ; u- " */  /* loop to TOTALTESTSET */
 124     { 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 }
 125 };
 126
 127 void addRuleBasedCollTest(TestNode** root)
 128 {
 129     addTest(root, &TestG7Locales, "tscoll/cg7coll/TestG7Locales");
 130     addTest(root, &TestDemo1, "tscoll/cg7coll/TestDemo1");
 131     addTest(root, &TestDemo2, "tscoll/cg7coll/TestDemo2");
 132     addTest(root, &TestDemo3, "tscoll/cg7coll/TestDemo3");
 133     addTest(root, &TestDemo4, "tscoll/cg7coll/TestDemo4");
 134
 135
 136 }
 137
 138 static void TestG7Locales()
 139 {
 140     UCollator *myCollation;
 141     UErrorCode status = U_ZERO_ERROR;
 142     const UChar *defRules;
 143     int32_t i, rlen, j, n;
 144     log_verbose("Testing  ucol_openRules for all the locales\n");
 145     for (i = 0; i < UPRV_LENGTHOF(locales); i++)
 146     {
 147         const char *locale = locales[i];
 148         status = U_ZERO_ERROR;
 149         myCollation = ucol_open(locale, &status);
 150         ucol_setAttribute(myCollation, UCOL_STRENGTH, UCOL_QUATERNARY, &status);
 151         ucol_setAttribute(myCollation, UCOL_ALTERNATE_HANDLING, UCOL_SHIFTED, &status);
 152
 153         if (U_FAILURE(status))
 154         {
 155             log_err_status(status, "Error in creating collator in %s:  %s\n", locale, myErrorName(status));
 156             ucol_close(myCollation);
 157             continue;
 158         }
 159
 160         defRules = ucol_getRules(myCollation, &rlen);
 161         if (rlen == 0 && (strcmp(locale, "fr_CA") == 0 || strcmp(locale, "ja_JP") == 0)) {
 162             log_data_err("%s UCollator missing rule string\n", locale);
 163             if (log_knownIssue("10671", "TestG7Locales does not test ignore-punctuation")) {
 164                 ucol_close(myCollation);
 165                 continue;
 166             }
 167         } else {
 168             UCollator *tblColl1;
 169             status = U_ZERO_ERROR;
 170             tblColl1 = ucol_openRules(defRules, rlen, UCOL_OFF,
 171                     UCOL_DEFAULT_STRENGTH,NULL, &status);
 172             ucol_close(myCollation);
 173             if (U_FAILURE(status))
 174             {
 175                 log_err_status(status, "Error in creating collator in %s:  %s\n", locale, myErrorName(status));
 176                 continue;
 177             }
 178             myCollation = tblColl1;
 179         }
 180
 181         log_verbose("Locale  %s\n", locales[i]);
 182         log_verbose("  tests start...\n");
 183
 184         j = 0;
 185         n = 0;
 186         for (j = 0; j < FIXEDTESTSET; j++)
 187         {
 188             for (n = j+1; n < FIXEDTESTSET; n++)
 189             {
 190                 doTest(myCollation, testCases[results[i][j]], testCases[results[i][n]], UCOL_LESS);
 191             }
 192         }
 193
 194         ucol_close(myCollation);
 195     }
 196 }
 197
 198 static void TestDemo1()
 199 {
 200     UCollator *myCollation;
 201     int32_t j, n;
 202     static const char rules[] = "& Z < p, P";
 203     int32_t len=(int32_t)strlen(rules);
 204     UChar temp[sizeof(rules)];
 205     UErrorCode status = U_ZERO_ERROR;
 206     u_uastrcpy(temp, rules);
 207
 208     log_verbose("Demo Test 1 : Create a new table collation with rules \" & Z < p, P \" \n");
 209
 210     myCollation = ucol_openRules(temp, len, UCOL_OFF, UCOL_DEFAULT_STRENGTH,NULL, &status);
 211
 212     if (U_FAILURE(status))
 213     {
 214         log_err_status(status, "Demo Test 1 Rule collation object creation failed. : %s\n", myErrorName(status));
 215         return;
 216     }
 217
 218     for (j = 0; j < FIXEDTESTSET; j++)
 219     {
 220         for (n = j+1; n < FIXEDTESTSET; n++)
 221         {
 222             doTest(myCollation, testCases[results[8][j]], testCases[results[8][n]], UCOL_LESS);
 223         }
 224     }
 225
 226     ucol_close(myCollation);
 227 }
 228
 229 static void TestDemo2()
 230 {
 231     UCollator *myCollation;
 232     int32_t j, n;
 233     static const char rules[] = "& C < ch , cH, Ch, CH";
 234     int32_t len=(int32_t)strlen(rules);
 235     UChar temp[sizeof(rules)];
 236     UErrorCode status = U_ZERO_ERROR;
 237     u_uastrcpy(temp, rules);
 238
 239     log_verbose("Demo Test 2 : Create a new table collation with rules \"& C < ch , cH, Ch, CH\"");
 240
 241     myCollation = ucol_openRules(temp, len, UCOL_OFF, UCOL_DEFAULT_STRENGTH, NULL, &status);
 242
 243     if (U_FAILURE(status))
 244     {
 245         log_err_status(status, "Demo Test 2 Rule collation object creation failed.: %s\n", myErrorName(status));
 246         return;
 247     }
 248     for (j = 0; j < TOTALTESTSET; j++)
 249     {
 250         for (n = j+1; n < TOTALTESTSET; n++)
 251         {
 252             doTest(myCollation, testCases[results[9][j]], testCases[results[9][n]], UCOL_LESS);
 253         }
 254     }
 255     ucol_close(myCollation);
 256 }
 257
 258 static void TestDemo3()
 259 {
 260     UCollator *myCollation;
 261     int32_t j, n;
 262     static const char rules[] = "& Question'-'mark ; '?' & Hash'-'mark ; '#' & Ampersand ; '&'";
 263     int32_t len=(int32_t)strlen(rules);
 264     UChar temp[sizeof(rules)];
 265     UErrorCode status = U_ZERO_ERROR;
 266     u_uastrcpy(temp, rules);
 267
 268     log_verbose("Demo Test 3 : Create a new table collation with rules \"& Question'-'mark ; '?' & Hash'-'mark ; '#' & Ampersand ; '&'\" \n");
 269
 270     myCollation = ucol_openRules(temp, len, UCOL_OFF, UCOL_DEFAULT_STRENGTH, NULL, &status);
 271
 272     if (U_FAILURE(status))
 273     {
 274         log_err_status(status, "Demo Test 3 Rule collation object creation failed.: %s\n", myErrorName(status));
 275         return;
 276     }
 277
 278     for (j = 0; j < TOTALTESTSET; j++)
 279     {
 280         for (n = j+1; n < TOTALTESTSET; n++)
 281         {
 282             doTest(myCollation, testCases[results[10][j]], testCases[results[10][n]], UCOL_LESS);
 283         }
 284     }
 285     ucol_close(myCollation);
 286 }
 287
 288 static void TestDemo4()
 289 {
 290     UCollator *myCollation;
 291     int32_t j, n;
 292     static const char rules[] = " & aa ; a'-' & ee ; e'-' & ii ; i'-' & oo ; o'-' & uu ; u'-' ";
 293     int32_t len=(int32_t)strlen(rules);
 294     UChar temp[sizeof(rules)];
 295     UErrorCode status = U_ZERO_ERROR;
 296     u_uastrcpy(temp, rules);
 297
 298     log_verbose("Demo Test 4 : Create a new table collation with rules \" & aa ; a'-' & ee ; e'-' & ii ; i'-' & oo ; o'-' & uu ; u'-' \"\n");
 299
 300     myCollation = ucol_openRules(temp, len, UCOL_OFF, UCOL_DEFAULT_STRENGTH, NULL, &status);
 301
 302     if (U_FAILURE(status))
 303     {
 304         log_err_status(status, "Demo Test 4 Rule collation object creation failed.: %s\n", myErrorName(status));
 305         return;
 306     }
 307     for (j = 0; j < TOTALTESTSET; j++)
 308     {
 309         for (n = j+1; n < TOTALTESTSET; n++)
 310         {
 311             doTest(myCollation, testCases[results[11][j]], testCases[results[11][n]], UCOL_LESS);
 312         }
 313     }
 314     ucol_close(myCollation);
 315 }
 316
 317 #endif /* #if !UCONFIG_NO_COLLATION */