]>
Commit | Line | Data |
---|---|---|
f3c0d7a5 A |
1 | // © 2016 and later: Unicode, Inc. and others. |
2 | // License & terms of use: http://www.unicode.org/copyright.html | |
729e4ab9 | 3 | /******************************************************************** |
2ca993e8 A |
4 | * COPYRIGHT: |
5 | * Copyright (c) 2009-2016, International Business Machines Corporation and | |
729e4ab9 A |
6 | * others. All Rights Reserved. |
7 | ********************************************************************/ | |
8 | /******************************************************************************** | |
9 | * | |
10 | * File spooftest.c | |
11 | * | |
12 | *********************************************************************************/ | |
13 | /*C API TEST for the uspoof Unicode Indentifier Spoofing and Security API */ | |
14 | /** | |
15 | * This is an API test for ICU spoof detection in plain C. It doesn't test very many cases, and doesn't | |
16 | * try to test the full functionality. It just calls each function and verifies that it | |
17 | * works on a basic level. | |
18 | * | |
19 | * More complete testing of spoof detection functionality is done with the C++ tests. | |
20 | **/ | |
21 | ||
22 | #include "unicode/utypes.h" | |
23 | #if !UCONFIG_NO_REGULAR_EXPRESSIONS && !UCONFIG_NO_NORMALIZATION | |
24 | ||
25 | #include <stdlib.h> | |
26 | #include <stdio.h> | |
27 | #include <string.h> | |
28 | #include "unicode/uspoof.h" | |
29 | #include "unicode/ustring.h" | |
30 | #include "unicode/uset.h" | |
31 | #include "cintltst.h" | |
2ca993e8 | 32 | #include "cmemory.h" |
729e4ab9 A |
33 | |
34 | #define TEST_ASSERT_SUCCESS(status) {if (U_FAILURE(status)) { \ | |
35 | log_err_status(status, "Failure at file %s, line %d, error = %s\n", __FILE__, __LINE__, u_errorName(status));}} | |
36 | ||
37 | #define TEST_ASSERT(expr) {if ((expr)==FALSE) { \ | |
38 | log_err("Test Failure at file %s, line %d: \"%s\" is false.\n", __FILE__, __LINE__, #expr);};} | |
39 | ||
40 | #define TEST_ASSERT_EQ(a, b) { if ((a) != (b)) { \ | |
41 | log_err("Test Failure at file %s, line %d: \"%s\" (%d) != \"%s\" (%d) \n", \ | |
42 | __FILE__, __LINE__, #a, (a), #b, (b)); }} | |
43 | ||
44 | #define TEST_ASSERT_NE(a, b) { if ((a) == (b)) { \ | |
45 | log_err("Test Failure at file %s, line %d: \"%s\" (%d) == \"%s\" (%d) \n", \ | |
46 | __FILE__, __LINE__, #a, (a), #b, (b)); }} | |
47 | ||
48 | ||
49 | /* | |
50 | * TEST_SETUP and TEST_TEARDOWN | |
51 | * macros to handle the boilerplate around setting up test case. | |
52 | * Put arbitrary test code between SETUP and TEARDOWN. | |
53 | * "sc" is the ready-to-go SpoofChecker for use in the tests. | |
54 | */ | |
55 | #define TEST_SETUP { \ | |
56 | UErrorCode status = U_ZERO_ERROR; \ | |
57 | USpoofChecker *sc; \ | |
58 | sc = uspoof_open(&status); \ | |
59 | TEST_ASSERT_SUCCESS(status); \ | |
60 | if (U_SUCCESS(status)){ | |
61 | ||
62 | #define TEST_TEARDOWN \ | |
63 | } \ | |
64 | TEST_ASSERT_SUCCESS(status); \ | |
65 | uspoof_close(sc); \ | |
66 | } | |
67 | ||
2ca993e8 | 68 | static void TestOpenFromSource(void); |
729e4ab9 A |
69 | static void TestUSpoofCAPI(void); |
70 | ||
71 | void addUSpoofTest(TestNode** root); | |
72 | ||
73 | void addUSpoofTest(TestNode** root) | |
74 | { | |
75 | #if !UCONFIG_NO_FILE_IO | |
2ca993e8 | 76 | addTest(root, &TestOpenFromSource, "uspoof/TestOpenFromSource"); |
729e4ab9 | 77 | #endif |
2ca993e8 | 78 | addTest(root, &TestUSpoofCAPI, "uspoof/TestUSpoofCAPI"); |
729e4ab9 A |
79 | } |
80 | ||
81 | /* | |
82 | * Identifiers for verifying that spoof checking is minimally alive and working. | |
83 | */ | |
84 | const UChar goodLatin[] = {(UChar)0x75, (UChar)0x7a, 0}; /* "uz", all ASCII */ | |
85 | /* (not confusable) */ | |
86 | const UChar scMixed[] = {(UChar)0x73, (UChar)0x0441, 0}; /* "sc", with Cyrillic 'c' */ | |
87 | /* (mixed script, confusable */ | |
88 | ||
89 | const UChar scLatin[] = {(UChar)0x73, (UChar)0x63, 0}; /* "sc", plain ascii. */ | |
90 | const UChar goodCyrl[] = {(UChar)0x438, (UChar)0x43B, 0}; /* Plain lower case Cyrillic letters, | |
91 | no latin confusables */ | |
92 | ||
93 | const UChar goodGreek[] = {(UChar)0x3c0, (UChar)0x3c6, 0}; /* Plain lower case Greek letters */ | |
94 | ||
95 | const UChar lll_Latin_a[] = {(UChar)0x6c, (UChar)0x49, (UChar)0x31, 0}; /* lI1, all ASCII */ | |
96 | ||
97 | /* Full-width I, Small Roman Numeral fifty, Latin Cap Letter IOTA*/ | |
98 | const UChar lll_Latin_b[] = {(UChar)0xff29, (UChar)0x217c, (UChar)0x196, 0}; | |
99 | ||
100 | const UChar lll_Cyrl[] = {(UChar)0x0406, (UChar)0x04C0, (UChar)0x31, 0}; | |
101 | ||
102 | /* The skeleton transform for all of thes 'lll' lookalikes is all lower case l. */ | |
103 | const UChar lll_Skel[] = {(UChar)0x6c, (UChar)0x6c, (UChar)0x6c, 0}; | |
104 | ||
4388f060 A |
105 | const UChar han_Hiragana[] = {(UChar)0x3086, (UChar)0x308A, (UChar)0x0020, (UChar)0x77F3, (UChar)0x7530, 0}; |
106 | ||
729e4ab9 A |
107 | /* Provide better code coverage */ |
108 | const char goodLatinUTF8[] = {0x75, 0x77, 0}; | |
729e4ab9 | 109 | |
2ca993e8 A |
110 | // Test open from source rules. |
111 | // Run this in isolation to verify initialization. | |
112 | static void TestOpenFromSource() { | |
113 | // No TEST_SETUP because that calls uspoof_open(). | |
114 | UErrorCode status = U_ZERO_ERROR; | |
729e4ab9 A |
115 | const char *dataSrcDir; |
116 | char *fileName; | |
117 | char *confusables; | |
57a6839d | 118 | int confusablesLength = 0; |
729e4ab9 | 119 | char *confusablesWholeScript; |
57a6839d | 120 | int confusablesWholeScriptLength = 0; |
729e4ab9 A |
121 | FILE *f; |
122 | UParseError pe; | |
123 | int32_t errType; | |
2ca993e8 | 124 | int32_t checkResults; |
729e4ab9 | 125 | USpoofChecker *rsc; |
2ca993e8 | 126 | |
729e4ab9 A |
127 | dataSrcDir = ctest_dataSrcDir(); |
128 | fileName = malloc(strlen(dataSrcDir) + 100); | |
129 | strcpy(fileName, dataSrcDir); | |
130 | strcat(fileName, U_FILE_SEP_STRING "unidata" U_FILE_SEP_STRING "confusables.txt"); | |
4388f060 | 131 | f = fopen(fileName, "rb"); |
729e4ab9 A |
132 | TEST_ASSERT_NE(f, NULL); |
133 | confusables = malloc(3000000); | |
51004dcb | 134 | if (f != NULL) { |
57a6839d A |
135 | confusablesLength = fread(confusables, 1, 3000000, f); |
136 | fclose(f); | |
51004dcb | 137 | } |
729e4ab9 | 138 | |
729e4ab9 A |
139 | strcpy(fileName, dataSrcDir); |
140 | strcat(fileName, U_FILE_SEP_STRING "unidata" U_FILE_SEP_STRING "confusablesWholeScript.txt"); | |
4388f060 | 141 | f = fopen(fileName, "rb"); |
729e4ab9 A |
142 | TEST_ASSERT_NE(f, NULL); |
143 | confusablesWholeScript = malloc(1000000); | |
51004dcb | 144 | if (f != NULL) { |
57a6839d A |
145 | confusablesWholeScriptLength = fread(confusablesWholeScript, 1, 1000000, f); |
146 | fclose(f); | |
51004dcb | 147 | } |
729e4ab9 A |
148 | |
149 | rsc = uspoof_openFromSource(confusables, confusablesLength, | |
2ca993e8 A |
150 | confusablesWholeScript, confusablesWholeScriptLength, |
151 | &errType, &pe, &status); | |
729e4ab9 A |
152 | TEST_ASSERT_SUCCESS(status); |
153 | ||
2ca993e8 A |
154 | // Ticket #11860: uspoof_openFromSource() did not initialize for use. |
155 | // Verify that the spoof checker does not crash. | |
156 | checkResults = uspoof_check(rsc, goodLatin, -1, NULL, &status); | |
157 | TEST_ASSERT_SUCCESS(status); | |
158 | TEST_ASSERT_EQ(0, checkResults); | |
159 | ||
729e4ab9 A |
160 | free(confusablesWholeScript); |
161 | free(confusables); | |
162 | free(fileName); | |
163 | uspoof_close(rsc); | |
164 | /* printf("ParseError Line is %d\n", pe.line); */ | |
2ca993e8 | 165 | } |
729e4ab9 | 166 | |
2ca993e8 A |
167 | /* |
168 | * Spoof Detection C API Tests | |
169 | */ | |
170 | static void TestUSpoofCAPI(void) { | |
171 | ||
172 | /* | |
173 | * basic uspoof_open(). | |
174 | */ | |
175 | { | |
176 | USpoofChecker *sc; | |
177 | UErrorCode status = U_ZERO_ERROR; | |
178 | sc = uspoof_open(&status); | |
179 | TEST_ASSERT_SUCCESS(status); | |
180 | if (U_FAILURE(status)) { | |
181 | /* If things are so broken that we can't even open a default spoof checker, */ | |
182 | /* don't even try the rest of the tests. They would all fail. */ | |
183 | return; | |
184 | } | |
185 | uspoof_close(sc); | |
186 | } | |
729e4ab9 A |
187 | |
188 | /* | |
189 | * openFromSerialized and serialize | |
190 | */ | |
191 | TEST_SETUP | |
192 | int32_t serializedSize = 0; | |
193 | int32_t actualLength = 0; | |
194 | char *buf; | |
195 | USpoofChecker *sc2; | |
196 | int32_t checkResults; | |
197 | ||
198 | ||
199 | serializedSize = uspoof_serialize(sc, NULL, 0, &status); | |
200 | TEST_ASSERT_EQ(status, U_BUFFER_OVERFLOW_ERROR); | |
201 | TEST_ASSERT(serializedSize > 0); | |
202 | ||
203 | /* Serialize the default spoof checker */ | |
204 | status = U_ZERO_ERROR; | |
205 | buf = (char *)malloc(serializedSize + 10); | |
206 | TEST_ASSERT(buf != NULL); | |
207 | buf[serializedSize] = 42; | |
208 | uspoof_serialize(sc, buf, serializedSize, &status); | |
209 | TEST_ASSERT_SUCCESS(status); | |
210 | TEST_ASSERT_EQ(42, buf[serializedSize]); | |
211 | ||
212 | /* Create a new spoof checker from the freshly serialized data */ | |
213 | sc2 = uspoof_openFromSerialized(buf, serializedSize+10, &actualLength, &status); | |
214 | TEST_ASSERT_SUCCESS(status); | |
215 | TEST_ASSERT_NE(NULL, sc2); | |
216 | TEST_ASSERT_EQ(serializedSize, actualLength); | |
217 | ||
218 | /* Verify that the new spoof checker at least wiggles */ | |
219 | checkResults = uspoof_check(sc2, goodLatin, -1, NULL, &status); | |
220 | TEST_ASSERT_SUCCESS(status); | |
221 | TEST_ASSERT_EQ(0, checkResults); | |
222 | ||
223 | checkResults = uspoof_check(sc2, scMixed, -1, NULL, &status); | |
224 | TEST_ASSERT_SUCCESS(status); | |
f3c0d7a5 | 225 | TEST_ASSERT_EQ(USPOOF_SINGLE_SCRIPT, checkResults); |
729e4ab9 A |
226 | |
227 | uspoof_close(sc2); | |
228 | free(buf); | |
229 | TEST_TEARDOWN; | |
230 | ||
231 | ||
232 | ||
233 | /* | |
234 | * Set & Get Check Flags | |
235 | */ | |
236 | TEST_SETUP | |
237 | int32_t t; | |
238 | uspoof_setChecks(sc, USPOOF_ALL_CHECKS, &status); | |
239 | TEST_ASSERT_SUCCESS(status); | |
240 | t = uspoof_getChecks(sc, &status); | |
241 | TEST_ASSERT_EQ(t, USPOOF_ALL_CHECKS); | |
242 | ||
243 | uspoof_setChecks(sc, 0, &status); | |
244 | TEST_ASSERT_SUCCESS(status); | |
245 | t = uspoof_getChecks(sc, &status); | |
246 | TEST_ASSERT_EQ(0, t); | |
247 | ||
248 | uspoof_setChecks(sc, | |
249 | USPOOF_WHOLE_SCRIPT_CONFUSABLE | USPOOF_MIXED_SCRIPT_CONFUSABLE | USPOOF_ANY_CASE, | |
250 | &status); | |
251 | TEST_ASSERT_SUCCESS(status); | |
252 | t = uspoof_getChecks(sc, &status); | |
253 | TEST_ASSERT_SUCCESS(status); | |
254 | TEST_ASSERT_EQ(USPOOF_WHOLE_SCRIPT_CONFUSABLE | USPOOF_MIXED_SCRIPT_CONFUSABLE | USPOOF_ANY_CASE, t); | |
255 | TEST_TEARDOWN; | |
256 | ||
257 | /* | |
258 | * get & setAllowedChars | |
259 | */ | |
260 | TEST_SETUP | |
261 | USet *us; | |
262 | const USet *uset; | |
263 | ||
264 | uset = uspoof_getAllowedChars(sc, &status); | |
265 | TEST_ASSERT_SUCCESS(status); | |
266 | TEST_ASSERT(uset_isFrozen(uset)); | |
267 | us = uset_open((UChar32)0x41, (UChar32)0x5A); /* [A-Z] */ | |
268 | uspoof_setAllowedChars(sc, us, &status); | |
269 | TEST_ASSERT_SUCCESS(status); | |
270 | TEST_ASSERT_NE(us, uspoof_getAllowedChars(sc, &status)); | |
271 | TEST_ASSERT(uset_equals(us, uspoof_getAllowedChars(sc, &status))); | |
272 | TEST_ASSERT_SUCCESS(status); | |
273 | uset_close(us); | |
274 | TEST_TEARDOWN; | |
275 | ||
276 | /* | |
277 | * clone() | |
278 | */ | |
279 | ||
280 | TEST_SETUP | |
281 | USpoofChecker *clone1 = NULL; | |
282 | USpoofChecker *clone2 = NULL; | |
283 | int32_t checkResults = 0; | |
284 | ||
285 | clone1 = uspoof_clone(sc, &status); | |
286 | TEST_ASSERT_SUCCESS(status); | |
287 | TEST_ASSERT_NE(clone1, sc); | |
288 | ||
289 | clone2 = uspoof_clone(clone1, &status); | |
290 | TEST_ASSERT_SUCCESS(status); | |
291 | TEST_ASSERT_NE(clone2, clone1); | |
292 | ||
293 | uspoof_close(clone1); | |
294 | ||
295 | /* Verify that the cloned spoof checker is alive */ | |
296 | checkResults = uspoof_check(clone2, goodLatin, -1, NULL, &status); | |
297 | TEST_ASSERT_SUCCESS(status); | |
298 | TEST_ASSERT_EQ(0, checkResults); | |
299 | ||
300 | checkResults = uspoof_check(clone2, scMixed, -1, NULL, &status); | |
301 | TEST_ASSERT_SUCCESS(status); | |
f3c0d7a5 | 302 | TEST_ASSERT_EQ(USPOOF_SINGLE_SCRIPT, checkResults); |
729e4ab9 A |
303 | uspoof_close(clone2); |
304 | TEST_TEARDOWN; | |
305 | ||
4388f060 A |
306 | /* |
307 | * basic uspoof_check() | |
308 | */ | |
309 | TEST_SETUP | |
310 | int32_t result; | |
311 | result = uspoof_check(sc, goodLatin, -1, NULL, &status); | |
312 | TEST_ASSERT_SUCCESS(status); | |
313 | TEST_ASSERT_EQ(0, result); | |
314 | ||
315 | result = uspoof_check(sc, han_Hiragana, -1, NULL, &status); | |
316 | TEST_ASSERT_SUCCESS(status); | |
317 | TEST_ASSERT_EQ(0, result); | |
318 | ||
319 | result = uspoof_check(sc, scMixed, -1, NULL, &status); | |
320 | TEST_ASSERT_SUCCESS(status); | |
f3c0d7a5 | 321 | TEST_ASSERT_EQ(USPOOF_SINGLE_SCRIPT, result); |
4388f060 A |
322 | TEST_TEARDOWN |
323 | ||
324 | ||
729e4ab9 A |
325 | /* |
326 | * get & set Checks | |
327 | */ | |
328 | TEST_SETUP | |
329 | int32_t checks; | |
330 | int32_t checks2; | |
331 | int32_t checkResults; | |
332 | ||
333 | checks = uspoof_getChecks(sc, &status); | |
334 | TEST_ASSERT_SUCCESS(status); | |
335 | TEST_ASSERT_EQ(USPOOF_ALL_CHECKS, checks); | |
336 | ||
337 | checks &= ~(USPOOF_SINGLE_SCRIPT | USPOOF_MIXED_SCRIPT_CONFUSABLE); | |
338 | uspoof_setChecks(sc, checks, &status); | |
339 | TEST_ASSERT_SUCCESS(status); | |
340 | checks2 = uspoof_getChecks(sc, &status); | |
341 | TEST_ASSERT_EQ(checks, checks2); | |
342 | ||
343 | /* The checks that were disabled just above are the same ones that the "scMixed" test fails. | |
344 | So with those tests gone checking that Identifier should now succeed */ | |
345 | checkResults = uspoof_check(sc, scMixed, -1, NULL, &status); | |
346 | TEST_ASSERT_SUCCESS(status); | |
347 | TEST_ASSERT_EQ(0, checkResults); | |
348 | TEST_TEARDOWN; | |
349 | ||
350 | /* | |
351 | * AllowedLoacles | |
352 | */ | |
353 | ||
354 | TEST_SETUP | |
355 | const char *allowedLocales; | |
356 | int32_t checkResults; | |
357 | ||
358 | /* Default allowed locales list should be empty */ | |
359 | allowedLocales = uspoof_getAllowedLocales(sc, &status); | |
360 | TEST_ASSERT_SUCCESS(status); | |
361 | TEST_ASSERT(strcmp("", allowedLocales) == 0) | |
362 | ||
363 | /* Allow en and ru, which should enable Latin and Cyrillic only to pass */ | |
364 | uspoof_setAllowedLocales(sc, "en, ru_RU", &status); | |
365 | TEST_ASSERT_SUCCESS(status); | |
366 | allowedLocales = uspoof_getAllowedLocales(sc, &status); | |
367 | TEST_ASSERT_SUCCESS(status); | |
368 | TEST_ASSERT(strstr(allowedLocales, "en") != NULL); | |
369 | TEST_ASSERT(strstr(allowedLocales, "ru") != NULL); | |
370 | ||
371 | /* Limit checks to USPOOF_CHAR_LIMIT. Some of the test data has whole script confusables also, | |
372 | * which we don't want to see in this test. */ | |
373 | uspoof_setChecks(sc, USPOOF_CHAR_LIMIT, &status); | |
374 | TEST_ASSERT_SUCCESS(status); | |
375 | ||
376 | checkResults = uspoof_check(sc, goodLatin, -1, NULL, &status); | |
377 | TEST_ASSERT_SUCCESS(status); | |
378 | TEST_ASSERT_EQ(0, checkResults); | |
379 | ||
380 | checkResults = uspoof_check(sc, goodGreek, -1, NULL, &status); | |
381 | TEST_ASSERT_SUCCESS(status); | |
382 | TEST_ASSERT_EQ(USPOOF_CHAR_LIMIT, checkResults); | |
383 | ||
384 | checkResults = uspoof_check(sc, goodCyrl, -1, NULL, &status); | |
385 | TEST_ASSERT_SUCCESS(status); | |
386 | TEST_ASSERT_EQ(0, checkResults); | |
387 | ||
388 | /* Reset with an empty locale list, which should allow all characters to pass */ | |
389 | uspoof_setAllowedLocales(sc, " ", &status); | |
390 | TEST_ASSERT_SUCCESS(status); | |
391 | ||
392 | checkResults = uspoof_check(sc, goodGreek, -1, NULL, &status); | |
393 | TEST_ASSERT_SUCCESS(status); | |
394 | TEST_ASSERT_EQ(0, checkResults); | |
395 | TEST_TEARDOWN; | |
396 | ||
397 | /* | |
398 | * AllowedChars set/get the USet of allowed characters. | |
399 | */ | |
400 | TEST_SETUP | |
401 | const USet *set; | |
402 | USet *tmpSet; | |
403 | int32_t checkResults; | |
404 | ||
405 | /* By default, we should see no restriction; the USet should allow all characters. */ | |
406 | set = uspoof_getAllowedChars(sc, &status); | |
407 | TEST_ASSERT_SUCCESS(status); | |
408 | tmpSet = uset_open(0, 0x10ffff); | |
409 | TEST_ASSERT(uset_equals(tmpSet, set)); | |
410 | ||
411 | /* Setting the allowed chars should enable the check. */ | |
412 | uspoof_setChecks(sc, USPOOF_ALL_CHECKS & ~USPOOF_CHAR_LIMIT, &status); | |
413 | TEST_ASSERT_SUCCESS(status); | |
414 | ||
415 | /* Remove a character that is in our good Latin test identifier from the allowed chars set. */ | |
416 | uset_remove(tmpSet, goodLatin[1]); | |
417 | uspoof_setAllowedChars(sc, tmpSet, &status); | |
418 | TEST_ASSERT_SUCCESS(status); | |
419 | uset_close(tmpSet); | |
420 | ||
51004dcb A |
421 | /* Latin Identifier should now fail; other non-latin test cases should still be OK |
422 | * Note: fail of CHAR_LIMIT also causes the restriction level to be USPOOF_UNRESTRICTIVE | |
423 | * which will give us a USPOOF_RESTRICTION_LEVEL failure. | |
424 | */ | |
729e4ab9 A |
425 | checkResults = uspoof_check(sc, goodLatin, -1, NULL, &status); |
426 | TEST_ASSERT_SUCCESS(status); | |
51004dcb | 427 | TEST_ASSERT_EQ(USPOOF_CHAR_LIMIT | USPOOF_RESTRICTION_LEVEL, checkResults); |
729e4ab9 A |
428 | |
429 | checkResults = uspoof_check(sc, goodGreek, -1, NULL, &status); | |
430 | TEST_ASSERT_SUCCESS(status); | |
f3c0d7a5 | 431 | TEST_ASSERT_EQ(0, checkResults); |
729e4ab9 A |
432 | TEST_TEARDOWN; |
433 | ||
434 | /* | |
435 | * check UTF-8 | |
436 | */ | |
437 | TEST_SETUP | |
438 | char utf8buf[200]; | |
f3c0d7a5 | 439 | int32_t checkResults, checkResults2; |
729e4ab9 A |
440 | int32_t position; |
441 | ||
442 | u_strToUTF8(utf8buf, sizeof(utf8buf), NULL, goodLatin, -1, &status); | |
443 | TEST_ASSERT_SUCCESS(status); | |
444 | position = 666; | |
445 | checkResults = uspoof_checkUTF8(sc, utf8buf, -1, &position, &status); | |
446 | TEST_ASSERT_SUCCESS(status); | |
447 | TEST_ASSERT_EQ(0, checkResults); | |
51004dcb | 448 | TEST_ASSERT_EQ(0, position); |
729e4ab9 A |
449 | |
450 | u_strToUTF8(utf8buf, sizeof(utf8buf), NULL, goodCyrl, -1, &status); | |
451 | TEST_ASSERT_SUCCESS(status); | |
452 | checkResults = uspoof_checkUTF8(sc, utf8buf, -1, &position, &status); | |
453 | TEST_ASSERT_SUCCESS(status); | |
454 | TEST_ASSERT_EQ(0, checkResults); | |
455 | ||
456 | u_strToUTF8(utf8buf, sizeof(utf8buf), NULL, scMixed, -1, &status); | |
457 | TEST_ASSERT_SUCCESS(status); | |
458 | position = 666; | |
459 | checkResults = uspoof_checkUTF8(sc, utf8buf, -1, &position, &status); | |
f3c0d7a5 | 460 | checkResults2 = uspoof_check(sc, scMixed, -1, NULL, &status); |
729e4ab9 | 461 | TEST_ASSERT_SUCCESS(status); |
f3c0d7a5 | 462 | TEST_ASSERT_EQ(USPOOF_SINGLE_SCRIPT , checkResults); |
51004dcb | 463 | TEST_ASSERT_EQ(0, position); |
f3c0d7a5 A |
464 | TEST_ASSERT_EQ(checkResults , checkResults2); |
465 | ||
466 | TEST_TEARDOWN; | |
467 | ||
468 | /* | |
469 | * uspoof_check2 variants | |
470 | */ | |
471 | TEST_SETUP | |
472 | int32_t result1, result2; | |
473 | char utf8buf[200]; | |
474 | uspoof_setChecks(sc, USPOOF_ALL_CHECKS | USPOOF_AUX_INFO, &status); | |
475 | USpoofCheckResult* checkResult = uspoof_openCheckResult(&status); | |
476 | TEST_ASSERT_SUCCESS(status); | |
477 | ||
478 | const UChar* tests[] = { goodLatin, scMixed, scLatin, | |
479 | goodCyrl, goodGreek, lll_Latin_a, lll_Latin_b, han_Hiragana }; | |
480 | ||
481 | for (int32_t i=0; i<UPRV_LENGTHOF(tests); i++) { | |
482 | const UChar* str = tests[i]; | |
483 | ||
484 | // Basic test | |
485 | result1 = uspoof_check(sc, str, -1, NULL, &status); | |
486 | result2 = uspoof_check2(sc, str, -1, NULL, &status); | |
487 | TEST_ASSERT_SUCCESS(status); | |
488 | TEST_ASSERT_EQ(result1, result2); | |
489 | ||
490 | // With check result parameter | |
491 | result1 = uspoof_check(sc, str, -1, NULL, &status); | |
492 | result2 = uspoof_check2(sc, str, -1, checkResult, &status); | |
493 | TEST_ASSERT_SUCCESS(status); | |
494 | TEST_ASSERT_EQ(result1, result2); | |
495 | ||
496 | // Checks from checkResult should be same as those from bitmask | |
497 | TEST_ASSERT_EQ(result1 & USPOOF_ALL_CHECKS, uspoof_getCheckResultChecks(checkResult, &status)); | |
498 | ||
499 | // Restriction level from checkResult should be same as that from bitmask | |
500 | URestrictionLevel restrictionLevel = uspoof_getCheckResultRestrictionLevel(checkResult, &status); | |
501 | TEST_ASSERT_EQ(result1 & restrictionLevel, restrictionLevel); | |
502 | ||
503 | // UTF8 endpoint | |
504 | u_strToUTF8(utf8buf, sizeof(utf8buf), NULL, goodLatin, -1, &status); | |
505 | TEST_ASSERT_SUCCESS(status); | |
506 | result1 = uspoof_checkUTF8(sc, utf8buf, -1, NULL, &status); | |
507 | result2 = uspoof_check2UTF8(sc, utf8buf, -1, NULL, &status); | |
508 | TEST_ASSERT_SUCCESS(status); | |
509 | TEST_ASSERT_EQ(result1, result2); | |
510 | } | |
729e4ab9 | 511 | |
f3c0d7a5 | 512 | uspoof_closeCheckResult(checkResult); |
729e4ab9 A |
513 | TEST_TEARDOWN; |
514 | ||
515 | /* | |
516 | * uspoof_areConfusable() | |
517 | */ | |
518 | TEST_SETUP | |
519 | int32_t checkResults; | |
520 | ||
521 | checkResults = uspoof_areConfusable(sc, scLatin, -1, scMixed, -1, &status); | |
522 | TEST_ASSERT_SUCCESS(status); | |
523 | TEST_ASSERT_EQ(USPOOF_MIXED_SCRIPT_CONFUSABLE, checkResults); | |
524 | ||
525 | checkResults = uspoof_areConfusable(sc, goodGreek, -1, scLatin, -1, &status); | |
526 | TEST_ASSERT_SUCCESS(status); | |
527 | TEST_ASSERT_EQ(0, checkResults); | |
528 | ||
529 | checkResults = uspoof_areConfusable(sc, lll_Latin_a, -1, lll_Latin_b, -1, &status); | |
530 | TEST_ASSERT_SUCCESS(status); | |
531 | TEST_ASSERT_EQ(USPOOF_SINGLE_SCRIPT_CONFUSABLE, checkResults); | |
532 | ||
533 | TEST_TEARDOWN; | |
534 | ||
535 | /* | |
536 | * areConfusableUTF8 | |
537 | */ | |
538 | TEST_SETUP | |
539 | int32_t checkResults; | |
540 | char s1[200]; | |
541 | char s2[200]; | |
542 | ||
543 | ||
544 | u_strToUTF8(s1, sizeof(s1), NULL, scLatin, -1, &status); | |
545 | u_strToUTF8(s2, sizeof(s2), NULL, scMixed, -1, &status); | |
546 | TEST_ASSERT_SUCCESS(status); | |
547 | checkResults = uspoof_areConfusableUTF8(sc, s1, -1, s2, -1, &status); | |
548 | TEST_ASSERT_SUCCESS(status); | |
549 | TEST_ASSERT_EQ(USPOOF_MIXED_SCRIPT_CONFUSABLE, checkResults); | |
550 | ||
551 | u_strToUTF8(s1, sizeof(s1), NULL, goodGreek, -1, &status); | |
552 | u_strToUTF8(s2, sizeof(s2), NULL, scLatin, -1, &status); | |
553 | TEST_ASSERT_SUCCESS(status); | |
554 | checkResults = uspoof_areConfusableUTF8(sc, s1, -1, s2, -1, &status); | |
555 | TEST_ASSERT_SUCCESS(status); | |
556 | TEST_ASSERT_EQ(0, checkResults); | |
557 | ||
558 | u_strToUTF8(s1, sizeof(s1), NULL, lll_Latin_a, -1, &status); | |
559 | u_strToUTF8(s2, sizeof(s2), NULL, lll_Latin_b, -1, &status); | |
560 | TEST_ASSERT_SUCCESS(status); | |
561 | checkResults = uspoof_areConfusableUTF8(sc, s1, -1, s2, -1, &status); | |
562 | TEST_ASSERT_SUCCESS(status); | |
563 | TEST_ASSERT_EQ(USPOOF_SINGLE_SCRIPT_CONFUSABLE, checkResults); | |
564 | ||
565 | TEST_TEARDOWN; | |
566 | ||
567 | ||
568 | /* | |
569 | * getSkeleton | |
570 | */ | |
571 | ||
572 | TEST_SETUP | |
573 | UChar dest[100]; | |
574 | int32_t skelLength; | |
575 | ||
2ca993e8 | 576 | skelLength = uspoof_getSkeleton(sc, USPOOF_ANY_CASE, lll_Latin_a, -1, dest, UPRV_LENGTHOF(dest), &status); |
729e4ab9 A |
577 | TEST_ASSERT_SUCCESS(status); |
578 | TEST_ASSERT_EQ(0, u_strcmp(lll_Skel, dest)); | |
579 | TEST_ASSERT_EQ(u_strlen(lll_Skel), skelLength); | |
580 | ||
581 | skelLength = uspoof_getSkeletonUTF8(sc, USPOOF_ANY_CASE, goodLatinUTF8, -1, (char*)dest, | |
2ca993e8 | 582 | UPRV_LENGTHOF(dest), &status); |
729e4ab9 A |
583 | TEST_ASSERT_SUCCESS(status); |
584 | ||
585 | skelLength = uspoof_getSkeleton(sc, USPOOF_ANY_CASE, lll_Latin_a, -1, NULL, 0, &status); | |
586 | TEST_ASSERT_EQ(U_BUFFER_OVERFLOW_ERROR, status); | |
587 | TEST_ASSERT_EQ(3, skelLength); | |
588 | status = U_ZERO_ERROR; | |
589 | ||
590 | TEST_TEARDOWN; | |
b331163b A |
591 | |
592 | /* | |
593 | * get Inclusion and Recommended sets | |
594 | */ | |
595 | TEST_SETUP | |
596 | const USet *inclusions = NULL; | |
597 | const USet *recommended = NULL; | |
598 | ||
599 | inclusions = uspoof_getInclusionSet(&status); | |
600 | TEST_ASSERT_SUCCESS(status); | |
601 | TEST_ASSERT_EQ(TRUE, uset_isFrozen(inclusions)); | |
602 | ||
603 | status = U_ZERO_ERROR; | |
604 | recommended = uspoof_getRecommendedSet(&status); | |
605 | TEST_ASSERT_SUCCESS(status); | |
606 | TEST_ASSERT_EQ(TRUE, uset_isFrozen(recommended)); | |
607 | TEST_TEARDOWN; | |
608 | ||
729e4ab9 A |
609 | } |
610 | ||
611 | #endif /* UCONFIG_NO_REGULAR_EXPRESSIONS */ |