]>
git.saurik.com Git - apple/icu.git/blob - icuSources/test/intltest/ucaconf.cpp
1 /********************************************************************
3 * Copyright (c) 2002-2012, International Business Machines Corporation and
4 * others. All Rights Reserved.
5 ********************************************************************/
8 * UCAConformanceTest performs conformance tests defined in the data
9 * files. ICU ships with stub data files, as the whole test are too
10 * long. To do the whole test, download the test files.
13 #include "unicode/utypes.h"
15 #if !UCONFIG_NO_COLLATION
18 #include "unicode/ustring.h"
23 UCAConformanceTest::UCAConformanceTest() :
28 UCA
= (RuleBasedCollator
*)Collator::createInstance(Locale::getRoot(), status
);
29 if(U_FAILURE(status
)) {
30 dataerrln("Error - UCAConformanceTest: Unable to open UCA collator! - %s", u_errorName(status
));
33 const char *srcDir
= IntlTest::getSourceTestData(status
);
34 if (U_FAILURE(status
)) {
35 dataerrln("Could not open test data %s", u_errorName(status
));
38 uprv_strcpy(testDataPath
, srcDir
);
39 uprv_strcat(testDataPath
, "CollationTest_");
41 UVersionInfo uniVersion
;
42 static const UVersionInfo v62
= { 6, 2, 0, 0 };
43 u_getUnicodeVersion(uniVersion
);
44 isAtLeastUCA62
= uprv_memcmp(uniVersion
, v62
, 4) >= 0;
47 UCAConformanceTest::~UCAConformanceTest()
56 void UCAConformanceTest::runIndexedTest( int32_t index
, UBool exec
, const char* &name
, char* /*par */)
59 logln("TestSuite UCAConformanceTest: ");
62 TESTCASE_AUTO(TestTableNonIgnorable
);
63 TESTCASE_AUTO(TestTableShifted
);
64 TESTCASE_AUTO(TestRulesNonIgnorable
);
65 TESTCASE_AUTO(TestRulesShifted
);
69 void UCAConformanceTest::initRbUCA()
72 UnicodeString ucarules
;
74 UCA
->getRules(UCOL_FULL_RULES
, ucarules
);
75 rbUCA
= new RuleBasedCollator(ucarules
, status
);
76 if (U_FAILURE(status
)) {
77 dataerrln("Failure creating UCA rule-based collator: %s", u_errorName(status
));
81 dataerrln("Failure creating UCA rule-based collator: %s", u_errorName(status
));
87 void UCAConformanceTest::setCollNonIgnorable(Collator
*coll
)
89 coll
->setAttribute(UCOL_NORMALIZATION_MODE
, UCOL_ON
, status
);
90 coll
->setAttribute(UCOL_CASE_FIRST
, UCOL_OFF
, status
);
91 coll
->setAttribute(UCOL_CASE_LEVEL
, UCOL_OFF
, status
);
92 coll
->setAttribute(UCOL_STRENGTH
, isAtLeastUCA62
? UCOL_IDENTICAL
: UCOL_TERTIARY
, status
);
93 coll
->setAttribute(UCOL_ALTERNATE_HANDLING
, UCOL_NON_IGNORABLE
, status
);
96 void UCAConformanceTest::setCollShifted(Collator
*coll
)
98 coll
->setAttribute(UCOL_NORMALIZATION_MODE
, UCOL_ON
, status
);
99 coll
->setAttribute(UCOL_CASE_FIRST
, UCOL_OFF
, status
);
100 coll
->setAttribute(UCOL_CASE_LEVEL
, UCOL_OFF
, status
);
101 coll
->setAttribute(UCOL_STRENGTH
, isAtLeastUCA62
? UCOL_IDENTICAL
: UCOL_QUATERNARY
, status
);
102 coll
->setAttribute(UCOL_ALTERNATE_HANDLING
, UCOL_SHIFTED
, status
);
105 void UCAConformanceTest::openTestFile(const char *type
)
107 const char *ext
= ".txt";
112 uprv_strcpy(buffer
, testDataPath
);
113 uprv_strcat(buffer
, type
);
114 int32_t bufLen
= (int32_t)uprv_strlen(buffer
);
116 // we try to open 3 files:
117 // path/CollationTest_type.txt
118 // path/CollationTest_type_SHORT.txt
119 // path/CollationTest_type_STUB.txt
120 // we are going to test with the first one that we manage to open.
122 uprv_strcpy(buffer
+bufLen
, ext
);
124 testFile
= fopen(buffer
, "rb");
127 uprv_strcpy(buffer
+bufLen
, "_SHORT");
128 uprv_strcat(buffer
, ext
);
129 testFile
= fopen(buffer
, "rb");
132 uprv_strcpy(buffer
+bufLen
, "_STUB");
133 uprv_strcat(buffer
, ext
);
134 testFile
= fopen(buffer
, "rb");
137 *(buffer
+bufLen
) = 0;
138 dataerrln("Could not open any of the conformance test files, tried opening base %s\n", buffer
);
142 "INFO: Working with the stub file.\n"
143 "If you need the full conformance test, please\n"
144 "download the appropriate data files from:\n"
145 "http://source.icu-project.org/repos/icu/tools/trunk/unicodetools/com/ibm/text/data/");
151 static const uint32_t IS_SHIFTED
= 1;
152 static const uint32_t FROM_RULES
= 2;
155 skipLineBecauseOfBug(const UChar
*s
, int32_t length
, uint32_t flags
) {
156 // TODO: Fix ICU ticket #8052
158 (s
[0] == 0xfb2 || s
[0] == 0xfb3) &&
160 (s
[2] == 0xf73 || s
[2] == 0xf75 || s
[2] == 0xf81)) {
163 // TODO: Fix ICU ticket #9361
164 if((flags
& IS_SHIFTED
) != 0 && length
>= 2 && s
[0] == 0xfffe) {
167 // TODO: Fix tailoring builder, ICU ticket #9593.
169 if((flags
& FROM_RULES
) != 0 && length
>= 2 && ((c
= s
[1]) == 0xedc || c
== 0xedd)) {
175 static UCollationResult
176 normalizeResult(int32_t result
) {
177 return result
<0 ? UCOL_LESS
: result
==0 ? UCOL_EQUAL
: UCOL_GREATER
;
180 void UCAConformanceTest::testConformance(const Collator
*coll
)
185 uint32_t skipFlags
= 0;
186 if(coll
->getAttribute(UCOL_ALTERNATE_HANDLING
, status
) == UCOL_SHIFTED
) {
187 skipFlags
|= IS_SHIFTED
;
190 skipFlags
|= FROM_RULES
;
195 UChar b1
[1024], b2
[1024];
196 UChar
*buffer
= b1
, *oldB
= NULL
;
198 char lineB1
[1024], lineB2
[1024];
199 char *lineB
= lineB1
, *oldLineB
= lineB2
;
201 uint8_t sk1
[1024], sk2
[1024];
202 uint8_t *oldSk
= NULL
, *newSk
= sk1
;
208 while (fgets(lineB
, 1024, testFile
) != NULL
) {
209 // remove trailing whitespace
213 if(*lineB
== 0 || lineB
[0] == '#') {
216 int32_t buflen
= u_parseString(lineB
, buffer
, 1024, &first
, &status
);
217 if(U_FAILURE(status
)) {
218 errln("Error parsing line %ld (%s): %s\n",
219 (long)line
, u_errorName(status
), lineB
);
220 status
= U_ZERO_ERROR
;
224 if(skipLineBecauseOfBug(buffer
, buflen
, skipFlags
)) {
225 logln("Skipping line %i because of a known bug", line
);
229 int32_t resLen
= coll
->getSortKey(buffer
, buflen
, newSk
, 1024);
232 int32_t skres
= strcmp((char *)oldSk
, (char *)newSk
);
233 int32_t cmpres
= coll
->compare(oldB
, oldBlen
, buffer
, buflen
, status
);
234 int32_t cmpres2
= coll
->compare(buffer
, buflen
, oldB
, oldBlen
, status
);
236 if(cmpres
!= -cmpres2
) {
237 errln("Compare result not symmetrical on line %i", line
);
240 if(cmpres
!= normalizeResult(skres
)) {
241 errln("Difference between coll->compare (%d) and sortkey compare (%d) on line %i",
242 cmpres
, skres
, line
);
243 errln(" Previous data line %s", oldLineB
);
244 errln(" Current data line %s", lineB
);
247 int32_t res
= cmpres
;
248 if(res
== 0 && !isAtLeastUCA62
) {
249 // Up to UCA 6.1, the collation test files use a custom tie-breaker,
250 // comparing the raw input strings.
251 res
= u_strcmpCodePointOrder(oldB
, buffer
);
252 // Starting with UCA 6.2, the collation test files use the standard UCA tie-breaker,
253 // comparing the NFD versions of the input strings,
254 // which we do via setting strength=identical.
257 errln("Line %i is not greater or equal than previous line", line
);
258 errln(" Previous data line %s", oldLineB
);
259 errln(" Current data line %s", lineB
);
260 UnicodeString oldS
, newS
;
261 prettify(CollationKey(oldSk
, oldLen
), oldS
);
262 prettify(CollationKey(newSk
, resLen
), newS
);
263 errln(" Previous key: "+oldS
);
264 errln(" Current key: "+newS
);
272 if(lineB
== lineB1
) {
286 void UCAConformanceTest::TestTableNonIgnorable(/* par */) {
287 if (U_FAILURE(status
)) {
288 dataerrln("Error running UCA Conformance Test: %s", u_errorName(status
));
291 setCollNonIgnorable(UCA
);
292 openTestFile("NON_IGNORABLE");
293 testConformance(UCA
);
296 void UCAConformanceTest::TestTableShifted(/* par */) {
297 if (U_FAILURE(status
)) {
298 dataerrln("Error running UCA Conformance Test: %s", u_errorName(status
));
302 openTestFile("SHIFTED");
303 testConformance(UCA
);
306 void UCAConformanceTest::TestRulesNonIgnorable(/* par */) {
309 if(U_SUCCESS(status
)) {
310 setCollNonIgnorable(rbUCA
);
311 openTestFile("NON_IGNORABLE");
312 testConformance(rbUCA
);
316 void UCAConformanceTest::TestRulesShifted(/* par */) {
317 logln("This test is currently disabled, as it is impossible to "
318 "wholly represent fractional UCA using tailoring rules.");
323 if(U_SUCCESS(status
)) {
324 setCollShifted(rbUCA
);
325 openTestFile("SHIFTED");
326 testConformance(rbUCA
);
330 #endif /* #if !UCONFIG_NO_COLLATION */