ucaconf.cpp revision c69afcec261fc345fda8daf46f0ea6b4351dc777 
1/********************************************************************
2 * COPYRIGHT:
3 * Copyright (c) 2002-2008, International Business Machines Corporation and
4 * others. All Rights Reserved.
5 ********************************************************************/
6
7/**
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.
11 */
12
13#include "unicode/utypes.h"
14
15#if !UCONFIG_NO_COLLATION
16
17#include "ucaconf.h"
18#include "unicode/ustring.h"
19#include "cstring.h"
20#include "uparse.h"
21
22UCAConformanceTest::UCAConformanceTest() :
23rbUCA(NULL),
24testFile(NULL),
25status(U_ZERO_ERROR)
26{
27    UCA = ucol_open("root", &status);
28    if(U_FAILURE(status)) {
29        errln("ERROR - UCAConformanceTest: Unable to open UCA collator!");
30    }
31
32    const char *srcDir = IntlTest::getSourceTestData(status);
33    if (U_FAILURE(status)) {
34        dataerrln("[DATA] Could not open test data %s", u_errorName(status));
35        return;
36    }
37    uprv_strcpy(testDataPath, srcDir);
38    uprv_strcat(testDataPath, "CollationTest_");
39}
40
41UCAConformanceTest::~UCAConformanceTest()
42{
43    ucol_close(UCA);
44    if(rbUCA) {
45        ucol_close(rbUCA);
46    }
47    if(testFile) {
48        fclose(testFile);
49    }
50}
51
52void UCAConformanceTest::runIndexedTest( int32_t index, UBool exec, const char* &name, char* /*par */)
53{
54    if (exec) logln("TestSuite UCAConformanceTest: ");
55    if(U_SUCCESS(status)) {
56      switch (index) {
57          case 0: name = "TestTableNonIgnorable"; if (exec)   TestTableNonIgnorable(/* par */); break;
58          case 1: name = "TestTableShifted";      if (exec)   TestTableShifted(/* par */);      break;
59          case 2: name = "TestRulesNonIgnorable"; if (exec)   TestRulesNonIgnorable(/* par */); break;
60          case 3: name = "TestRulesShifted";      if (exec)   TestRulesShifted(/* par */);      break;
61          default: name = ""; break;
62      }
63    } else {
64      name = "";
65    }
66}
67
68void UCAConformanceTest::initRbUCA()
69{
70    if(!rbUCA) {
71        UParseError parseError;
72        UChar      *ucarules;
73        // preflight rules
74        int32_t size = ucol_getRulesEx(UCA, UCOL_FULL_RULES, NULL, 0);
75        ucarules = (UChar *)malloc(size * sizeof(UChar));
76
77        size = ucol_getRulesEx(UCA, UCOL_FULL_RULES, ucarules, size);
78        rbUCA = ucol_openRules(ucarules, size, UCOL_DEFAULT, UCOL_TERTIARY,
79            &parseError, &status);
80        free(ucarules);
81        if (U_FAILURE(status)) {
82            errln("Failure creating UCA rule-based collator: %s", u_errorName(status));
83            return;
84        }
85    }
86}
87
88void UCAConformanceTest::setCollNonIgnorable(UCollator *coll)
89{
90  ucol_setAttribute(coll, UCOL_NORMALIZATION_MODE, UCOL_ON, &status);
91  ucol_setAttribute(coll, UCOL_CASE_FIRST, UCOL_OFF, &status);
92  ucol_setAttribute(coll, UCOL_CASE_LEVEL, UCOL_OFF, &status);
93  ucol_setAttribute(coll, UCOL_STRENGTH, UCOL_TERTIARY, &status);
94  ucol_setAttribute(coll, UCOL_ALTERNATE_HANDLING, UCOL_NON_IGNORABLE, &status);
95}
96
97void UCAConformanceTest::setCollShifted(UCollator *coll)
98{
99    ucol_setAttribute(coll, UCOL_NORMALIZATION_MODE, UCOL_ON, &status);
100    ucol_setAttribute(coll, UCOL_CASE_FIRST, UCOL_OFF, &status);
101    ucol_setAttribute(coll, UCOL_CASE_LEVEL, UCOL_OFF, &status);
102    ucol_setAttribute(coll, UCOL_STRENGTH, UCOL_QUATERNARY, &status);
103    ucol_setAttribute(coll, UCOL_ALTERNATE_HANDLING, UCOL_SHIFTED, &status);
104}
105
106void UCAConformanceTest::openTestFile(const char *type)
107{
108    const char *ext = ".txt";
109    if(testFile) {
110        fclose(testFile);
111    }
112    char buffer[1024];
113    uprv_strcpy(buffer, testDataPath);
114    uprv_strcat(buffer, type);
115    int32_t bufLen = (int32_t)uprv_strlen(buffer);
116
117    // we try to open 3 files:
118    // path/CollationTest_type.txt
119    // path/CollationTest_type_SHORT.txt
120    // path/CollationTest_type_STUB.txt
121    // we are going to test with the first one that we manage to open.
122
123    uprv_strcpy(buffer+bufLen, ext);
124
125    testFile = fopen(buffer, "rb");
126
127    if(testFile == 0) {
128        uprv_strcpy(buffer+bufLen, "_SHORT");
129        uprv_strcat(buffer, ext);
130        testFile = fopen(buffer, "rb");
131
132        if(testFile == 0) {
133            uprv_strcpy(buffer+bufLen, "_STUB");
134            uprv_strcat(buffer, ext);
135            testFile = fopen(buffer, "rb");
136
137            if (testFile == 0) {
138                *(buffer+bufLen) = 0;
139                dataerrln("[DATA] Could not open any of the conformance test files, tried opening base %s\n", buffer);
140                return;
141            } else {
142                infoln(
143                    "INFO: Working with the stub file.\n"
144                    "If you need the full conformance test, please\n"
145                    "download the appropriate data files from:\n"
146                    "http://source.icu-project.org/repos/icu/tools/trunk/unicodetools/com/ibm/text/data/");
147            }
148        }
149    }
150}
151
152void UCAConformanceTest::testConformance(UCollator *coll)
153{
154    if(testFile == 0) {
155        return;
156    }
157
158    int32_t line = 0;
159
160    UChar b1[1024], b2[1024];
161    char lineB[1024];
162    UChar *buffer = b1, *oldB = NULL;
163    uint8_t sk1[1024], sk2[1024];
164    uint8_t *oldSk = NULL, *newSk = sk1;
165    int32_t resLen = 0, oldLen = 0;
166    int32_t buflen = 0, oldBlen = 0;
167    uint32_t first = 0;
168    uint32_t offset = 0;
169    UnicodeString oldS, newS;
170
171
172    while (fgets(lineB, 1024, testFile) != NULL) {
173        offset = 0;
174
175        line++;
176        if(*lineB == 0 || strlen(lineB) < 3 || lineB[0] == '#') {
177            continue;
178        }
179        offset = u_parseString(lineB, buffer, 1024, &first, &status);
180        buflen = offset;
181        buffer[offset++] = 0;
182
183        resLen = ucol_getSortKey(coll, buffer, buflen, newSk, 1024);
184
185        int32_t res = 0, cmpres = 0, cmpres2 = 0;
186
187        if(oldSk != NULL) {
188            res = strcmp((char *)oldSk, (char *)newSk);
189            cmpres = ucol_strcoll(coll, oldB, oldBlen, buffer, buflen);
190            cmpres2 = ucol_strcoll(coll, buffer, buflen, oldB, oldBlen);
191
192            if(cmpres != -cmpres2) {
193                errln("Compare result not symmetrical on line %i", line);
194            }
195
196            if(((res&0x80000000) != (cmpres&0x80000000)) || (res == 0 && cmpres != 0) || (res != 0 && cmpres == 0)) {
197                errln("Difference between ucol_strcoll and sortkey compare on line %i", line);
198                logln("Data line %s", lineB);
199            }
200
201            if(res > 0) {
202                errln("Line %i is not greater or equal than previous line", line);
203                logln("Data line %s", lineB);
204                prettify(CollationKey(oldSk, oldLen), oldS);
205                prettify(CollationKey(newSk, resLen), newS);
206                logln("Keys: "+oldS+" and "+newS);
207            } else if(res == 0) { /* equal */
208                res = u_strcmpCodePointOrder(oldB, buffer);
209                if (res == 0) {
210                    errln("Probable error in test file on line %i (comparing identical strings)", line);
211                    logln("Data line %s", lineB);
212                } else if (res > 0) {
213                    errln("Sortkeys are identical, but code point comapare gives >0 on line %i", line);
214                    logln("Data line %s", lineB);
215                }
216            }
217        }
218
219        oldSk = newSk;
220        oldLen = resLen;
221
222        newSk = (newSk == sk1)?sk2:sk1;
223        oldB = buffer;
224        oldBlen = buflen;
225        buffer = (buffer == b1)?b2:b1;
226    }
227}
228
229void UCAConformanceTest::TestTableNonIgnorable(/* par */) {
230    setCollNonIgnorable(UCA);
231    openTestFile("NON_IGNORABLE");
232    testConformance(UCA);
233}
234
235void UCAConformanceTest::TestTableShifted(/* par */) {
236    setCollShifted(UCA);
237    openTestFile("SHIFTED");
238    testConformance(UCA);
239}
240
241void UCAConformanceTest::TestRulesNonIgnorable(/* par */) {
242    initRbUCA();
243
244    if(U_SUCCESS(status)) {
245        setCollNonIgnorable(rbUCA);
246        openTestFile("NON_IGNORABLE");
247        testConformance(rbUCA);
248    }
249}
250
251void UCAConformanceTest::TestRulesShifted(/* par */) {
252    logln("This test is currently disabled, as it is impossible to "
253        "wholly represent fractional UCA using tailoring rules.");
254    return;
255
256    initRbUCA();
257
258    if(U_SUCCESS(status)) {
259        setCollShifted(rbUCA);
260        openTestFile("SHIFTED");
261        testConformance(rbUCA);
262    }
263}
264
265#endif /* #if !UCONFIG_NO_COLLATION */
266