1 /********************************************************************
2 * COPYRIGHT:
3 * Copyright (c) 1997-2014, International Business Machines Corporation and
4 * others. All Rights Reserved.
5 ********************************************************************/
6
7 #include "unicode/utypes.h"
8
9 #if !UCONFIG_NO_COLLATION
10
11 #include "unicode/coll.h"
12 #include "unicode/tblcoll.h"
13 #include "unicode/unistr.h"
14 #include "unicode/sortkey.h"
15 #include "g7coll.h"
16 #include "sfwdchit.h"
17 #include "cmemory.h"
18
19 static const UChar testCases[][G7CollationTest::MAX_TOKEN_LEN] = {
20 { 0x0062 /*'b'*/, 0x006c /*'l'*/, 0x0061 /*'a'*/, 0x0062 /*'c'*/, 0x006b /*'k'*/,
21 0x0062 /*'b'*/, 0x0069 /*'i'*/, 0x0072 /*'r'*/, 0x0064 /*'d'*/, 0x0073 /*'s'*/, 0x0000}, /* 9 */
22 { 0x0050 /*'P'*/, 0x0061 /*'a'*/, 0x0074/*'t'*/, 0x0000}, /* 1 */
23 { 0x0070 /*'p'*/, 0x00E9, 0x0063 /*'c'*/, 0x0068 /*'h'*/, 0x00E9, 0x0000}, /* 2 */
24 { 0x0070 /*'p'*/, 0x00EA, 0x0063 /*'c'*/, 0x0068 /*'h'*/, 0x0065 /*'e'*/, 0x0000}, /* 3 */
25 { 0x0070 /*'p'*/, 0x00E9, 0x0063 /*'c'*/, 0x0068 /*'h'*/, 0x0065 /*'e'*/, 0x0072 /*'r'*/, 0x0000}, /* 4 */
26 { 0x0070 /*'p'*/, 0x00EA, 0x0063 /*'c'*/, 0x0068 /*'h'*/, 0x0065 /*'e'*/, 0x0072 /*'r'*/, 0x0000}, /* 5 */
27 { 0x0054 /*'T'*/, 0x006f /*'o'*/, 0x0064 /*'d'*/, 0x0000}, /* 6 */
28 { 0x0054 /*'T'*/, 0x00F6, 0x006e /*'n'*/, 0x0065 /*'e'*/, 0x0000}, /* 7 */
29 { 0x0054 /*'T'*/, 0x006f /*'o'*/, 0x0066 /*'f'*/, 0x0075 /*'u'*/, 0x0000}, /* 8 */
30 { 0x0062 /*'b'*/, 0x006c /*'l'*/, 0x0061 /*'a'*/, 0x0062 /*'c'*/, 0x006b /*'k'*/,
31 0x0062 /*'b'*/, 0x0069 /*'i'*/, 0x0072 /*'r'*/, 0x0064 /*'d'*/, 0x0000}, /* 12 */
32 { 0x0054 /*'T'*/, 0x006f /*'o'*/, 0x006e /*'n'*/, 0x0000}, /* 10 */
33 { 0x0050 /*'P'*/, 0x0041 /*'A'*/, 0x0054 /*'T'*/, 0x0000}, /* 11 */
34 { 0x0062 /*'b'*/, 0x006c /*'l'*/, 0x0061 /*'a'*/, 0x0062 /*'c'*/, 0x006b /*'k'*/,
35 0x002d /*'-'*/, 0x0062 /*'b'*/, 0x0069 /*'i'*/, 0x0072 /*'r'*/, 0x0064 /*'d'*/, 0x0000}, /* 13 */
36 { 0x0062 /*'b'*/, 0x006c /*'l'*/, 0x0061 /*'a'*/, 0x0062 /*'c'*/, 0x006b /*'k'*/,
37 0x002d /*'-'*/, 0x0062 /*'b'*/, 0x0069 /*'i'*/, 0x0072 /*'r'*/, 0x0064 /*'d'*/, 0x0073/*'s'*/, 0x0000}, /* 0 */
38 {0x0070 /*'p'*/, 0x0061 /*'a'*/, 0x0074 /*'t'*/, 0x0000}, /* 14 */
39 /* Additional tests */
40 { 0x0063 /*'c'*/, 0x007a /*'z'*/, 0x0061 /*'a'*/, 0x0072 /*'r'*/, 0x0000 }, /* 15 */
41 { 0x0063 /*'c'*/, 0x0068 /*'h'*/, 0x0075 /*'u'*/, 0x0072 /*'r'*/, 0x006f /*'o'*/, 0x0000 }, /* 16 */
42 { 0x0063 /*'c'*/, 0x0061 /*'a'*/, 0x0074 /*'t'*/, 0x000 }, /* 17 */
43 { 0x0064 /*'d'*/, 0x0061 /*'a'*/, 0x0072 /*'r'*/, 0x006e /*'n'*/, 0x0000 }, /* 18 */
44 { 0x003f /*'?'*/, 0x0000 }, /* 19 */
45 { 0x0071 /*'q'*/, 0x0075 /*'u'*/, 0x0069 /*'i'*/, 0x0063 /*'c'*/, 0x006b /*'k'*/, 0x0000 }, /* 20 */
46 { 0x0023 /*'#'*/, 0x0000 }, /* 21 */
47 { 0x0026 /*'&'*/, 0x0000 }, /* 22 */
48 { 0x0061 /*'a'*/, 0x002d /*'-'*/, 0x0072 /*'r'*/, 0x0064 /*'d'*/, 0x0076 /*'v'*/, 0x0061 /*'a'*/,
49 0x0072/*'r'*/, 0x006b/*'k'*/, 0x0000}, /* 24 */
50 { 0x0061 /*'a'*/, 0x0061 /*'a'*/, 0x0072 /*'r'*/, 0x0064 /*'d'*/, 0x0076 /*'v'*/, 0x0061 /*'a'*/,
51 0x0072/*'r'*/, 0x006b/*'k'*/, 0x0000}, /* 23 */
52 { 0x0061 /*'a'*/, 0x0062 /*'b'*/, 0x0062 /*'b'*/, 0x006f /*'o'*/, 0x0074 /*'t'*/, 0x0000}, /* 25 */
53 { 0x0063 /*'c'*/, 0x006f /*'o'*/, 0x002d /*'-'*/, 0x0070 /*'p'*/, 0x0000}, /* 27 */
54 { 0x0063 /*'c'*/, 0x006f /*'o'*/, 0x0070 /*'p'*/, 0x0000}, /* 28 */
55 { 0x0063 /*'c'*/, 0x006f /*'o'*/, 0x006f /*'o'*/, 0x0070 /*'p'*/, 0x0000}, /* 26 */
56 { 0x007a /*'z'*/, 0x0065 /*'e'*/, 0x0062 /*'b'*/, 0x0072 /*'r'*/, 0x0061 /*'a'*/, 0x0000} /* 29 */
57 };
58
59 static const int32_t results[G7CollationTest::TESTLOCALES][G7CollationTest::TOTALTESTSET] = {
60 { 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 */
61 { 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 */
62 { 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 */
63 { 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 */
64 { 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 */
65 { 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 */
66 { 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 */
67 { 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 */
68 /* new table collation with rules "& Z < p, P" loop to FIXEDTESTSET */
69 { 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 },
70 /* new table collation with rules "& C < ch , cH, Ch, CH " loop to TOTALTESTSET */
71 { 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 },
72 /* new table collation with rules "& Question-mark ; ? & Hash-mark ; # & Ampersand ; '&' " loop to TOTALTESTSET */
73 { 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 },
74 /* analogous to Japanese rules " & aa ; a- & ee ; e- & ii ; i- & oo ; o- & uu ; u- " */ /* loop to TOTALTESTSET */
75 { 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 }
76 };
77
~G7CollationTest()78 G7CollationTest::~G7CollationTest() {}
79
TestG7Locales()80 void G7CollationTest::TestG7Locales(/* char* par */)
81 {
82 int32_t i;
83 const Locale locales[8] = {
84 Locale("en", "US", ""),
85 Locale("en", "GB", ""),
86 Locale("en", "CA", ""),
87 Locale("fr", "FR", ""),
88 Locale("fr", "CA", ""),
89 Locale("de", "DE", ""),
90 Locale("it", "IT", ""),
91 Locale("ja", "JP", "")
92 };
93
94 for (i = 0; i < UPRV_LENGTHOF(locales); i++)
95 {
96 UnicodeString dispName;
97 UErrorCode status = U_ZERO_ERROR;
98
99 const Locale &locale = locales[i];
100 LocalPointer<Collator> myCollation(Collator::createInstance(locale, status));
101 if(U_FAILURE(status)) {
102 errcheckln(status, "Couldn't instantiate collator. Error: %s", u_errorName(status));
103 return;
104 }
105 myCollation->setStrength(Collator::QUATERNARY);
106 myCollation->setAttribute(UCOL_ALTERNATE_HANDLING, UCOL_SHIFTED, status);
107 if (U_FAILURE(status)) {
108 errln("Locale %s creation failed - %s", locale.getName(), u_errorName(status));
109 continue;
110 }
111
112 const UnicodeString &rules = ((RuleBasedCollator*)myCollation.getAlias())->getRules();
113 if (rules.isEmpty() &&
114 (locale == Locale::getCanadaFrench() || locale == Locale::getJapanese())) {
115 dataerrln("%s Collator missing rule string", locale.getName());
116 if (logKnownIssue("10671", "TestG7Locales does not test ignore-punctuation")) {
117 continue;
118 }
119 } else {
120 status = U_ZERO_ERROR;
121 RuleBasedCollator *tblColl1 = new RuleBasedCollator(rules, status);
122 if (U_FAILURE(status)) {
123 errln("Recreate %s collation failed - %s", locale.getName(), u_errorName(status));
124 continue;
125 }
126 myCollation.adoptInstead(tblColl1);
127 }
128
129 UnicodeString msg;
130
131 msg += "Locale ";
132 msg += locales[i].getDisplayName(dispName);
133 msg += "tests start :";
134 logln(msg);
135
136 int32_t j, n;
137 for (j = 0; j < FIXEDTESTSET; j++)
138 {
139 for (n = j+1; n < FIXEDTESTSET; n++)
140 {
141 doTest(myCollation.getAlias(), testCases[results[i][j]], testCases[results[i][n]], Collator::LESS);
142 }
143 }
144 }
145 }
146
TestDemo1()147 void G7CollationTest::TestDemo1(/* char* par */)
148 {
149 logln("Demo Test 1 : Create a new table collation with rules \"& Z < p, P\"");
150 UErrorCode status = U_ZERO_ERROR;
151 Collator *col = Collator::createInstance("en_US", status);
152 if(U_FAILURE(status)) {
153 delete col;
154 errcheckln(status, "Couldn't instantiate collator. Error: %s", u_errorName(status));
155 return;
156 }
157 const UnicodeString baseRules = ((RuleBasedCollator*)col)->getRules();
158 UnicodeString newRules(" & Z < p, P");
159 newRules.insert(0, baseRules);
160 RuleBasedCollator *myCollation = new RuleBasedCollator(newRules, status);
161
162 if (U_FAILURE(status))
163 {
164 errln( "Demo Test 1 Table Collation object creation failed.");
165 return;
166 }
167
168 int32_t j, n;
169 for (j = 0; j < FIXEDTESTSET; j++)
170 {
171 for (n = j+1; n < FIXEDTESTSET; n++)
172 {
173 doTest(myCollation, testCases[results[8][j]], testCases[results[8][n]], Collator::LESS);
174 }
175 }
176
177 delete myCollation;
178 delete col;
179 }
180
TestDemo2()181 void G7CollationTest::TestDemo2(/* char* par */)
182 {
183 logln("Demo Test 2 : Create a new table collation with rules \"& C < ch , cH, Ch, CH\"");
184 UErrorCode status = U_ZERO_ERROR;
185 Collator *col = Collator::createInstance("en_US", status);
186 if(U_FAILURE(status)) {
187 delete col;
188 errcheckln(status, "Couldn't instantiate collator. Error: %s", u_errorName(status));
189 return;
190 }
191 const UnicodeString baseRules = ((RuleBasedCollator*)col)->getRules();
192 UnicodeString newRules("& C < ch , cH, Ch, CH");
193 newRules.insert(0, baseRules);
194 RuleBasedCollator *myCollation = new RuleBasedCollator(newRules, status);
195
196 if (U_FAILURE(status))
197 {
198 errln("Demo Test 2 Table Collation object creation failed.");
199 return;
200 }
201
202 int32_t j, n;
203 for (j = 0; j < TOTALTESTSET; j++)
204 {
205 for (n = j+1; n < TOTALTESTSET; n++)
206 {
207 doTest(myCollation, testCases[results[9][j]], testCases[results[9][n]], Collator::LESS);
208 }
209 }
210
211 delete myCollation;
212 delete col;
213 }
214
TestDemo3()215 void G7CollationTest::TestDemo3(/* char* par */)
216 {
217 logln("Demo Test 3 : Create a new table collation with rules \"& Question'-'mark ; '?' & Hash'-'mark ; '#' & Ampersand ; '&'\"");
218 UErrorCode status = U_ZERO_ERROR;
219 Collator *col = Collator::createInstance("en_US", status);
220 if(U_FAILURE(status)) {
221 errcheckln(status, "Couldn't instantiate collator. Error: %s", u_errorName(status));
222 delete col;
223 return;
224 }
225 const UnicodeString baseRules = ((RuleBasedCollator*)col)->getRules();
226 UnicodeString newRules = "& Question'-'mark ; '?' & Hash'-'mark ; '#' & Ampersand ; '&'";
227 newRules.insert(0, baseRules);
228 RuleBasedCollator *myCollation = new RuleBasedCollator(newRules, status);
229
230 if (U_FAILURE(status))
231 {
232 errln("Demo Test 3 Table Collation object creation failed.");
233 return;
234 }
235
236 int32_t j, n;
237 for (j = 0; j < TOTALTESTSET; j++)
238 {
239 for (n = j+1; n < TOTALTESTSET; n++)
240 {
241 doTest(myCollation, testCases[results[10][j]], testCases[results[10][n]], Collator::LESS);
242 }
243 }
244
245 delete myCollation;
246 delete col;
247 }
248
TestDemo4()249 void G7CollationTest::TestDemo4(/* char* par */)
250 {
251 logln("Demo Test 4 : Create a new table collation with rules \" & aa ; a'-' & ee ; e'-' & ii ; i'-' & oo ; o'-' & uu ; u'-' \"");
252 UErrorCode status = U_ZERO_ERROR;
253 Collator *col = Collator::createInstance("en_US", status);
254 if(U_FAILURE(status)) {
255 delete col;
256 errcheckln(status, "Couldn't instantiate collator. Error: %s", u_errorName(status));
257 return;
258 }
259
260 const UnicodeString baseRules = ((RuleBasedCollator*)col)->getRules();
261 UnicodeString newRules = " & aa ; a'-' & ee ; e'-' & ii ; i'-' & oo ; o'-' & uu ; u'-' ";
262 newRules.insert(0, baseRules);
263 RuleBasedCollator *myCollation = new RuleBasedCollator(newRules, status);
264
265 int32_t j, n;
266 for (j = 0; j < TOTALTESTSET; j++)
267 {
268 for (n = j+1; n < TOTALTESTSET; n++)
269 {
270 doTest(myCollation, testCases[results[11][j]], testCases[results[11][n]], Collator::LESS);
271 }
272 }
273
274 delete myCollation;
275 delete col;
276 }
277
runIndexedTest(int32_t index,UBool exec,const char * & name,char *)278 void G7CollationTest::runIndexedTest( int32_t index, UBool exec, const char* &name, char* /*par*/ )
279 {
280 if (exec) logln("TestSuite G7CollationTest: ");
281 switch (index) {
282 case 0: name = "TestG7Locales"; if (exec) TestG7Locales(/* par */); break;
283 case 1: name = "TestDemo1"; if (exec) TestDemo1(/* par */); break;
284 case 2: name = "TestDemo2"; if (exec) TestDemo2(/* par */); break;
285 case 3: name = "TestDemo3"; if (exec) TestDemo3(/* par */); break;
286 case 4: name = "TestDemo4"; if (exec) TestDemo4(/* par */); break;
287 default: name = ""; break;
288 }
289 }
290
291 #endif /* #if !UCONFIG_NO_COLLATION */
292