1 | // © 2016 and later: Unicode, Inc. and others. |
2 | // License & terms of use: http://www.unicode.org/copyright.html |
3 | /* |
4 | ********************************************************************** |
5 | * Copyright (C) 2008-2016, International Business Machines |
6 | * Corporation and others. All Rights Reserved. |
7 | ********************************************************************** |
8 | */ |
9 | |
10 | #include "unicode/utypes.h" |
11 | #include "unicode/uspoof.h" |
12 | #include "unicode/uchar.h" |
13 | #include "unicode/uniset.h" |
14 | #include "unicode/utf16.h" |
15 | #include "utrie2.h" |
16 | #include "cmemory.h" |
17 | #include "cstring.h" |
18 | #include "scriptset.h" |
19 | #include "umutex.h" |
20 | #include "udataswp.h" |
21 | #include "uassert.h" |
22 | #include "ucln_in.h" |
23 | #include "uspoof_impl.h" |
24 | |
25 | #if !UCONFIG_NO_NORMALIZATION |
26 | |
27 | |
28 | U_NAMESPACE_BEGIN |
29 | |
30 | UOBJECT_DEFINE_RTTI_IMPLEMENTATION(SpoofImpl) |
31 | |
32 | SpoofImpl::SpoofImpl(SpoofData *data, UErrorCode& status) { |
33 | construct(status); |
34 | fSpoofData = data; |
35 | } |
36 | |
37 | SpoofImpl::SpoofImpl(UErrorCode& status) { |
38 | construct(status); |
39 | |
40 | // TODO: Call this method where it is actually needed, instead of in the |
41 | // constructor, to allow for lazy data loading. See #12696. |
42 | fSpoofData = SpoofData::getDefault(status); |
43 | } |
44 | |
45 | SpoofImpl::SpoofImpl() { |
46 | UErrorCode status = U_ZERO_ERROR; |
47 | construct(status); |
48 | |
49 | // TODO: Call this method where it is actually needed, instead of in the |
50 | // constructor, to allow for lazy data loading. See #12696. |
51 | fSpoofData = SpoofData::getDefault(status); |
52 | } |
53 | |
54 | void SpoofImpl::construct(UErrorCode& status) { |
55 | fChecks = USPOOF_ALL_CHECKS; |
56 | fSpoofData = nullptr; |
57 | fAllowedCharsSet = nullptr; |
58 | fAllowedLocales = nullptr; |
59 | fRestrictionLevel = USPOOF_HIGHLY_RESTRICTIVE; |
60 | |
61 | if (U_FAILURE(status)) { return; } |
62 | |
63 | UnicodeSet * = new UnicodeSet(0, 0x10ffff); |
64 | fAllowedCharsSet = allowedCharsSet; |
65 | fAllowedLocales = uprv_strdup("" ); |
66 | if (fAllowedCharsSet == nullptr || fAllowedLocales == nullptr) { |
67 | status = U_MEMORY_ALLOCATION_ERROR; |
68 | return; |
69 | } |
70 | allowedCharsSet->freeze(); |
71 | } |
72 | |
73 | |
74 | // Copy Constructor, used by the user level clone() function. |
75 | SpoofImpl::SpoofImpl(const SpoofImpl &src, UErrorCode &status) : |
76 | fChecks(USPOOF_ALL_CHECKS), fSpoofData(nullptr), fAllowedCharsSet(nullptr) , |
77 | fAllowedLocales(nullptr) { |
78 | if (U_FAILURE(status)) { |
79 | return; |
80 | } |
81 | fChecks = src.fChecks; |
82 | if (src.fSpoofData != nullptr) { |
83 | fSpoofData = src.fSpoofData->addReference(); |
84 | } |
85 | fAllowedCharsSet = src.fAllowedCharsSet->clone(); |
86 | fAllowedLocales = uprv_strdup(src.fAllowedLocales); |
87 | if (fAllowedCharsSet == nullptr || fAllowedLocales == nullptr) { |
88 | status = U_MEMORY_ALLOCATION_ERROR; |
89 | } |
90 | fRestrictionLevel = src.fRestrictionLevel; |
91 | } |
92 | |
93 | SpoofImpl::~SpoofImpl() { |
94 | if (fSpoofData != nullptr) { |
95 | fSpoofData->removeReference(); // Will delete if refCount goes to zero. |
96 | } |
97 | delete fAllowedCharsSet; |
98 | uprv_free((void *)fAllowedLocales); |
99 | } |
100 | |
101 | // Cast this instance as a USpoofChecker for the C API. |
102 | USpoofChecker *SpoofImpl::asUSpoofChecker() { |
103 | return exportForC(); |
104 | } |
105 | |
106 | // |
107 | // Incoming parameter check on Status and the SpoofChecker object |
108 | // received from the C API. |
109 | // |
110 | const SpoofImpl *SpoofImpl::validateThis(const USpoofChecker *sc, UErrorCode &status) { |
111 | auto* This = validate(sc, status); |
112 | if (U_FAILURE(status)) { |
113 | return nullptr; |
114 | } |
115 | if (This->fSpoofData != nullptr && !This->fSpoofData->validateDataVersion(status)) { |
116 | return nullptr; |
117 | } |
118 | return This; |
119 | } |
120 | |
121 | SpoofImpl *SpoofImpl::validateThis(USpoofChecker *sc, UErrorCode &status) { |
122 | return const_cast<SpoofImpl *> |
123 | (SpoofImpl::validateThis(const_cast<const USpoofChecker *>(sc), status)); |
124 | } |
125 | |
126 | |
127 | void SpoofImpl::setAllowedLocales(const char *localesList, UErrorCode &status) { |
128 | UnicodeSet allowedChars; |
129 | UnicodeSet *tmpSet = nullptr; |
130 | const char *locStart = localesList; |
131 | const char *locEnd = nullptr; |
132 | const char *localesListEnd = localesList + uprv_strlen(localesList); |
133 | int32_t localeListCount = 0; // Number of locales provided by caller. |
134 | |
135 | // Loop runs once per locale from the localesList, a comma separated list of locales. |
136 | do { |
137 | locEnd = uprv_strchr(locStart, ','); |
138 | if (locEnd == nullptr) { |
139 | locEnd = localesListEnd; |
140 | } |
141 | while (*locStart == ' ') { |
142 | locStart++; |
143 | } |
144 | const char *trimmedEnd = locEnd-1; |
145 | while (trimmedEnd > locStart && *trimmedEnd == ' ') { |
146 | trimmedEnd--; |
147 | } |
148 | if (trimmedEnd <= locStart) { |
149 | break; |
150 | } |
151 | const char *locale = uprv_strndup(locStart, (int32_t)(trimmedEnd + 1 - locStart)); |
152 | localeListCount++; |
153 | |
154 | // We have one locale from the locales list. |
155 | // Add the script chars for this locale to the accumulating set of allowed chars. |
156 | // If the locale is no good, we will be notified back via status. |
157 | addScriptChars(locale, &allowedChars, status); |
158 | uprv_free((void *)locale); |
159 | if (U_FAILURE(status)) { |
160 | break; |
161 | } |
162 | locStart = locEnd + 1; |
163 | } while (locStart < localesListEnd); |
164 | |
165 | // If our caller provided an empty list of locales, we disable the allowed characters checking |
166 | if (localeListCount == 0) { |
167 | uprv_free((void *)fAllowedLocales); |
168 | fAllowedLocales = uprv_strdup("" ); |
169 | tmpSet = new UnicodeSet(0, 0x10ffff); |
170 | if (fAllowedLocales == nullptr || tmpSet == nullptr) { |
171 | status = U_MEMORY_ALLOCATION_ERROR; |
172 | return; |
173 | } |
174 | tmpSet->freeze(); |
175 | delete fAllowedCharsSet; |
176 | fAllowedCharsSet = tmpSet; |
177 | fChecks &= ~USPOOF_CHAR_LIMIT; |
178 | return; |
179 | } |
180 | |
181 | |
182 | // Add all common and inherited characters to the set of allowed chars. |
183 | UnicodeSet tempSet; |
184 | tempSet.applyIntPropertyValue(UCHAR_SCRIPT, USCRIPT_COMMON, status); |
185 | allowedChars.addAll(tempSet); |
186 | tempSet.applyIntPropertyValue(UCHAR_SCRIPT, USCRIPT_INHERITED, status); |
187 | allowedChars.addAll(tempSet); |
188 | |
189 | // If anything went wrong, we bail out without changing |
190 | // the state of the spoof checker. |
191 | if (U_FAILURE(status)) { |
192 | return; |
193 | } |
194 | |
195 | // Store the updated spoof checker state. |
196 | tmpSet = allowedChars.clone(); |
197 | const char *tmpLocalesList = uprv_strdup(localesList); |
198 | if (tmpSet == nullptr || tmpLocalesList == nullptr) { |
199 | status = U_MEMORY_ALLOCATION_ERROR; |
200 | return; |
201 | } |
202 | uprv_free((void *)fAllowedLocales); |
203 | fAllowedLocales = tmpLocalesList; |
204 | tmpSet->freeze(); |
205 | delete fAllowedCharsSet; |
206 | fAllowedCharsSet = tmpSet; |
207 | fChecks |= USPOOF_CHAR_LIMIT; |
208 | } |
209 | |
210 | |
211 | const char * SpoofImpl::getAllowedLocales(UErrorCode &/*status*/) { |
212 | return fAllowedLocales; |
213 | } |
214 | |
215 | |
216 | // Given a locale (a language), add all the characters from all of the scripts used with that language |
217 | // to the allowedChars UnicodeSet |
218 | |
219 | void SpoofImpl::addScriptChars(const char *locale, UnicodeSet *allowedChars, UErrorCode &status) { |
220 | UScriptCode scripts[30]; |
221 | |
222 | int32_t numScripts = uscript_getCode(locale, scripts, UPRV_LENGTHOF(scripts), &status); |
223 | if (U_FAILURE(status)) { |
224 | return; |
225 | } |
226 | if (status == U_USING_DEFAULT_WARNING) { |
227 | status = U_ILLEGAL_ARGUMENT_ERROR; |
228 | return; |
229 | } |
230 | UnicodeSet tmpSet; |
231 | int32_t i; |
232 | for (i=0; i<numScripts; i++) { |
233 | tmpSet.applyIntPropertyValue(UCHAR_SCRIPT, scripts[i], status); |
234 | allowedChars->addAll(tmpSet); |
235 | } |
236 | } |
237 | |
238 | // Computes the augmented script set for a code point, according to UTS 39 section 5.1. |
239 | void SpoofImpl::getAugmentedScriptSet(UChar32 codePoint, ScriptSet& result, UErrorCode& status) { |
240 | result.resetAll(); |
241 | result.setScriptExtensions(codePoint, status); |
242 | if (U_FAILURE(status)) { return; } |
243 | |
244 | // Section 5.1 step 1 |
245 | if (result.test(USCRIPT_HAN, status)) { |
246 | result.set(USCRIPT_HAN_WITH_BOPOMOFO, status); |
247 | result.set(USCRIPT_JAPANESE, status); |
248 | result.set(USCRIPT_KOREAN, status); |
249 | } |
250 | if (result.test(USCRIPT_HIRAGANA, status)) { |
251 | result.set(USCRIPT_JAPANESE, status); |
252 | } |
253 | if (result.test(USCRIPT_KATAKANA, status)) { |
254 | result.set(USCRIPT_JAPANESE, status); |
255 | } |
256 | if (result.test(USCRIPT_HANGUL, status)) { |
257 | result.set(USCRIPT_KOREAN, status); |
258 | } |
259 | if (result.test(USCRIPT_BOPOMOFO, status)) { |
260 | result.set(USCRIPT_HAN_WITH_BOPOMOFO, status); |
261 | } |
262 | |
263 | // Section 5.1 step 2 |
264 | if (result.test(USCRIPT_COMMON, status) || result.test(USCRIPT_INHERITED, status)) { |
265 | result.setAll(); |
266 | } |
267 | } |
268 | |
269 | // Computes the resolved script set for a string, according to UTS 39 section 5.1. |
270 | void SpoofImpl::getResolvedScriptSet(const UnicodeString& input, ScriptSet& result, UErrorCode& status) const { |
271 | getResolvedScriptSetWithout(input, USCRIPT_CODE_LIMIT, result, status); |
272 | } |
273 | |
274 | // Computes the resolved script set for a string, omitting characters having the specified script. |
275 | // If USCRIPT_CODE_LIMIT is passed as the second argument, all characters are included. |
276 | void SpoofImpl::getResolvedScriptSetWithout(const UnicodeString& input, UScriptCode script, ScriptSet& result, UErrorCode& status) const { |
277 | result.setAll(); |
278 | |
279 | ScriptSet temp; |
280 | UChar32 codePoint; |
281 | for (int32_t i = 0; i < input.length(); i += U16_LENGTH(codePoint)) { |
282 | codePoint = input.char32At(i); |
283 | |
284 | // Compute the augmented script set for the character |
285 | getAugmentedScriptSet(codePoint, temp, status); |
286 | if (U_FAILURE(status)) { return; } |
287 | |
288 | // Intersect the augmented script set with the resolved script set, but only if the character doesn't |
289 | // have the script specified in the function call |
290 | if (script == USCRIPT_CODE_LIMIT || !temp.test(script, status)) { |
291 | result.intersect(temp); |
292 | } |
293 | } |
294 | } |
295 | |
296 | // Computes the set of numerics for a string, according to UTS 39 section 5.3. |
297 | void SpoofImpl::getNumerics(const UnicodeString& input, UnicodeSet& result, UErrorCode& /*status*/) const { |
298 | result.clear(); |
299 | |
300 | UChar32 codePoint; |
301 | for (int32_t i = 0; i < input.length(); i += U16_LENGTH(codePoint)) { |
302 | codePoint = input.char32At(i); |
303 | |
304 | // Store a representative character for each kind of decimal digit |
305 | if (u_charType(codePoint) == U_DECIMAL_DIGIT_NUMBER) { |
306 | // Store the zero character as a representative for comparison. |
307 | // Unicode guarantees it is codePoint - value |
308 | result.add(codePoint - (UChar32)u_getNumericValue(codePoint)); |
309 | } |
310 | } |
311 | } |
312 | |
313 | // Computes the restriction level of a string, according to UTS 39 section 5.2. |
314 | URestrictionLevel SpoofImpl::getRestrictionLevel(const UnicodeString& input, UErrorCode& status) const { |
315 | // Section 5.2 step 1: |
316 | if (!fAllowedCharsSet->containsAll(input)) { |
317 | return USPOOF_UNRESTRICTIVE; |
318 | } |
319 | |
320 | // Section 5.2 step 2 |
321 | // Java use a static UnicodeSet for this test. In C++, avoid the static variable |
322 | // and just do a simple for loop. |
323 | UBool allASCII = true; |
324 | for (int32_t i=0, length=input.length(); i<length; i++) { |
325 | if (input.charAt(i) > 0x7f) { |
326 | allASCII = false; |
327 | break; |
328 | } |
329 | } |
330 | if (allASCII) { |
331 | return USPOOF_ASCII; |
332 | } |
333 | |
334 | // Section 5.2 steps 3: |
335 | ScriptSet resolvedScriptSet; |
336 | getResolvedScriptSet(input, resolvedScriptSet, status); |
337 | if (U_FAILURE(status)) { return USPOOF_UNRESTRICTIVE; } |
338 | |
339 | // Section 5.2 step 4: |
340 | if (!resolvedScriptSet.isEmpty()) { |
341 | return USPOOF_SINGLE_SCRIPT_RESTRICTIVE; |
342 | } |
343 | |
344 | // Section 5.2 step 5: |
345 | ScriptSet resolvedNoLatn; |
346 | getResolvedScriptSetWithout(input, USCRIPT_LATIN, resolvedNoLatn, status); |
347 | if (U_FAILURE(status)) { return USPOOF_UNRESTRICTIVE; } |
348 | |
349 | // Section 5.2 step 6: |
350 | if (resolvedNoLatn.test(USCRIPT_HAN_WITH_BOPOMOFO, status) |
351 | || resolvedNoLatn.test(USCRIPT_JAPANESE, status) |
352 | || resolvedNoLatn.test(USCRIPT_KOREAN, status)) { |
353 | return USPOOF_HIGHLY_RESTRICTIVE; |
354 | } |
355 | |
356 | // Section 5.2 step 7: |
357 | if (!resolvedNoLatn.isEmpty() |
358 | && !resolvedNoLatn.test(USCRIPT_CYRILLIC, status) |
359 | && !resolvedNoLatn.test(USCRIPT_GREEK, status) |
360 | && !resolvedNoLatn.test(USCRIPT_CHEROKEE, status)) { |
361 | return USPOOF_MODERATELY_RESTRICTIVE; |
362 | } |
363 | |
364 | // Section 5.2 step 8: |
365 | return USPOOF_MINIMALLY_RESTRICTIVE; |
366 | } |
367 | |
368 | int32_t SpoofImpl::findHiddenOverlay(const UnicodeString& input, UErrorCode&) const { |
369 | bool sawLeadCharacter = false; |
370 | for (int32_t i=0; i<input.length();) { |
371 | UChar32 cp = input.char32At(i); |
372 | if (sawLeadCharacter && cp == 0x0307) { |
373 | return i; |
374 | } |
375 | uint8_t combiningClass = u_getCombiningClass(cp); |
376 | // Skip over characters except for those with combining class 0 (non-combining characters) or with |
377 | // combining class 230 (same class as U+0307) |
378 | U_ASSERT(u_getCombiningClass(0x0307) == 230); |
379 | if (combiningClass == 0 || combiningClass == 230) { |
380 | sawLeadCharacter = isIllegalCombiningDotLeadCharacter(cp); |
381 | } |
382 | i += U16_LENGTH(cp); |
383 | } |
384 | return -1; |
385 | } |
386 | |
387 | static inline bool isIllegalCombiningDotLeadCharacterNoLookup(UChar32 cp) { |
388 | return cp == u'i' || cp == u'j' || cp == u'ı' || cp == u'ȷ' || cp == u'l' || |
389 | u_hasBinaryProperty(cp, UCHAR_SOFT_DOTTED); |
390 | } |
391 | |
392 | bool SpoofImpl::isIllegalCombiningDotLeadCharacter(UChar32 cp) const { |
393 | if (isIllegalCombiningDotLeadCharacterNoLookup(cp)) { |
394 | return true; |
395 | } |
396 | UnicodeString skelStr; |
397 | fSpoofData->confusableLookup(cp, skelStr); |
398 | UChar32 finalCp = skelStr.char32At(skelStr.moveIndex32(skelStr.length(), -1)); |
399 | if (finalCp != cp && isIllegalCombiningDotLeadCharacterNoLookup(finalCp)) { |
400 | return true; |
401 | } |
402 | return false; |
403 | } |
404 | |
405 | |
406 | |
407 | // Convert a text format hex number. Utility function used by builder code. Static. |
408 | // Input: char16_t *string text. Output: a UChar32 |
409 | // Input has been pre-checked, and will have no non-hex chars. |
410 | // The number must fall in the code point range of 0..0x10ffff |
411 | // Static Function. |
412 | UChar32 SpoofImpl::ScanHex(const char16_t *s, int32_t start, int32_t limit, UErrorCode &status) { |
413 | if (U_FAILURE(status)) { |
414 | return 0; |
415 | } |
416 | U_ASSERT(limit-start > 0); |
417 | uint32_t val = 0; |
418 | int i; |
419 | for (i=start; i<limit; i++) { |
420 | int digitVal = s[i] - 0x30; |
421 | if (digitVal>9) { |
422 | digitVal = 0xa + (s[i] - 0x41); // Upper Case 'A' |
423 | } |
424 | if (digitVal>15) { |
425 | digitVal = 0xa + (s[i] - 0x61); // Lower Case 'a' |
426 | } |
427 | U_ASSERT(digitVal <= 0xf); |
428 | val <<= 4; |
429 | val += digitVal; |
430 | } |
431 | if (val > 0x10ffff) { |
432 | status = U_PARSE_ERROR; |
433 | val = 0; |
434 | } |
435 | return (UChar32)val; |
436 | } |
437 | |
438 | |
439 | //----------------------------------------- |
440 | // |
441 | // class CheckResult Implementation |
442 | // |
443 | //----------------------------------------- |
444 | |
445 | CheckResult::CheckResult() { |
446 | clear(); |
447 | } |
448 | |
449 | USpoofCheckResult* CheckResult::asUSpoofCheckResult() { |
450 | return exportForC(); |
451 | } |
452 | |
453 | // |
454 | // Incoming parameter check on Status and the CheckResult object |
455 | // received from the C API. |
456 | // |
457 | const CheckResult* CheckResult::validateThis(const USpoofCheckResult *ptr, UErrorCode &status) { |
458 | return validate(ptr, status); |
459 | } |
460 | |
461 | CheckResult* CheckResult::validateThis(USpoofCheckResult *ptr, UErrorCode &status) { |
462 | return validate(ptr, status); |
463 | } |
464 | |
465 | void CheckResult::clear() { |
466 | fChecks = 0; |
467 | fNumerics.clear(); |
468 | fRestrictionLevel = USPOOF_UNDEFINED_RESTRICTIVE; |
469 | } |
470 | |
471 | int32_t CheckResult::toCombinedBitmask(int32_t enabledChecks) { |
472 | if ((enabledChecks & USPOOF_AUX_INFO) != 0 && fRestrictionLevel != USPOOF_UNDEFINED_RESTRICTIVE) { |
473 | return fChecks | fRestrictionLevel; |
474 | } else { |
475 | return fChecks; |
476 | } |
477 | } |
478 | |
479 | CheckResult::~CheckResult() { |
480 | } |
481 | |
482 | //---------------------------------------------------------------------------------------------- |
483 | // |
484 | // class SpoofData Implementation |
485 | // |
486 | //---------------------------------------------------------------------------------------------- |
487 | |
488 | |
489 | UBool SpoofData::validateDataVersion(UErrorCode &status) const { |
490 | if (U_FAILURE(status) || |
491 | fRawData == nullptr || |
492 | fRawData->fMagic != USPOOF_MAGIC || |
493 | fRawData->fFormatVersion[0] != USPOOF_CONFUSABLE_DATA_FORMAT_VERSION || |
494 | fRawData->fFormatVersion[1] != 0 || |
495 | fRawData->fFormatVersion[2] != 0 || |
496 | fRawData->fFormatVersion[3] != 0) { |
497 | status = U_INVALID_FORMAT_ERROR; |
498 | return false; |
499 | } |
500 | return true; |
501 | } |
502 | |
503 | static UBool U_CALLCONV |
504 | spoofDataIsAcceptable(void *context, |
505 | const char * /* type */, const char * /*name*/, |
506 | const UDataInfo *pInfo) { |
507 | if( |
508 | pInfo->size >= 20 && |
509 | pInfo->isBigEndian == U_IS_BIG_ENDIAN && |
510 | pInfo->charsetFamily == U_CHARSET_FAMILY && |
511 | pInfo->dataFormat[0] == 0x43 && // dataFormat="Cfu " |
512 | pInfo->dataFormat[1] == 0x66 && |
513 | pInfo->dataFormat[2] == 0x75 && |
514 | pInfo->dataFormat[3] == 0x20 && |
515 | pInfo->formatVersion[0] == USPOOF_CONFUSABLE_DATA_FORMAT_VERSION |
516 | ) { |
517 | UVersionInfo *version = static_cast<UVersionInfo *>(context); |
518 | if(version != nullptr) { |
519 | uprv_memcpy(version, pInfo->dataVersion, 4); |
520 | } |
521 | return true; |
522 | } else { |
523 | return false; |
524 | } |
525 | } |
526 | |
527 | // Methods for the loading of the default confusables data file. The confusable |
528 | // data is loaded only when it is needed. |
529 | // |
530 | // SpoofData::getDefault() - Return the default confusables data, and call the |
531 | // initOnce() if it is not available. Adds a reference |
532 | // to the SpoofData that the caller is responsible for |
533 | // decrementing when they are done with the data. |
534 | // |
535 | // uspoof_loadDefaultData - Called once, from initOnce(). The resulting SpoofData |
536 | // is shared by all spoof checkers using the default data. |
537 | // |
538 | // uspoof_cleanupDefaultData - Called during cleanup. |
539 | // |
540 | |
541 | static UInitOnce gSpoofInitDefaultOnce {}; |
542 | static SpoofData* gDefaultSpoofData; |
543 | |
544 | static UBool U_CALLCONV |
545 | uspoof_cleanupDefaultData() { |
546 | if (gDefaultSpoofData) { |
547 | // Will delete, assuming all user-level spoof checkers were closed. |
548 | gDefaultSpoofData->removeReference(); |
549 | gDefaultSpoofData = nullptr; |
550 | gSpoofInitDefaultOnce.reset(); |
551 | } |
552 | return true; |
553 | } |
554 | |
555 | static void U_CALLCONV uspoof_loadDefaultData(UErrorCode& status) { |
556 | UDataMemory *udm = udata_openChoice(nullptr, "cfu" , "confusables" , |
557 | spoofDataIsAcceptable, |
558 | nullptr, // context, would receive dataVersion if supplied. |
559 | &status); |
560 | if (U_FAILURE(status)) { return; } |
561 | gDefaultSpoofData = new SpoofData(udm, status); |
562 | if (U_FAILURE(status)) { |
563 | delete gDefaultSpoofData; |
564 | gDefaultSpoofData = nullptr; |
565 | return; |
566 | } |
567 | if (gDefaultSpoofData == nullptr) { |
568 | status = U_MEMORY_ALLOCATION_ERROR; |
569 | return; |
570 | } |
571 | ucln_i18n_registerCleanup(UCLN_I18N_SPOOFDATA, uspoof_cleanupDefaultData); |
572 | } |
573 | |
574 | SpoofData* SpoofData::getDefault(UErrorCode& status) { |
575 | umtx_initOnce(gSpoofInitDefaultOnce, &uspoof_loadDefaultData, status); |
576 | if (U_FAILURE(status)) { return nullptr; } |
577 | gDefaultSpoofData->addReference(); |
578 | return gDefaultSpoofData; |
579 | } |
580 | |
581 | |
582 | |
583 | SpoofData::SpoofData(UDataMemory *udm, UErrorCode &status) |
584 | { |
585 | reset(); |
586 | if (U_FAILURE(status)) { |
587 | return; |
588 | } |
589 | fUDM = udm; |
590 | // fRawData is non-const because it may be constructed by the data builder. |
591 | fRawData = reinterpret_cast<SpoofDataHeader *>( |
592 | const_cast<void *>(udata_getMemory(udm))); |
593 | validateDataVersion(status); |
594 | initPtrs(status); |
595 | } |
596 | |
597 | |
598 | SpoofData::SpoofData(const void *data, int32_t length, UErrorCode &status) |
599 | { |
600 | reset(); |
601 | if (U_FAILURE(status)) { |
602 | return; |
603 | } |
604 | if ((size_t)length < sizeof(SpoofDataHeader)) { |
605 | status = U_INVALID_FORMAT_ERROR; |
606 | return; |
607 | } |
608 | if (data == nullptr) { |
609 | status = U_ILLEGAL_ARGUMENT_ERROR; |
610 | return; |
611 | } |
612 | void *ncData = const_cast<void *>(data); |
613 | fRawData = static_cast<SpoofDataHeader *>(ncData); |
614 | if (length < fRawData->fLength) { |
615 | status = U_INVALID_FORMAT_ERROR; |
616 | return; |
617 | } |
618 | validateDataVersion(status); |
619 | initPtrs(status); |
620 | } |
621 | |
622 | |
623 | // Spoof Data constructor for use from data builder. |
624 | // Initializes a new, empty data area that will be populated later. |
625 | SpoofData::SpoofData(UErrorCode &status) { |
626 | reset(); |
627 | if (U_FAILURE(status)) { |
628 | return; |
629 | } |
630 | fDataOwned = true; |
631 | |
632 | // The spoof header should already be sized to be a multiple of 16 bytes. |
633 | // Just in case it's not, round it up. |
634 | uint32_t initialSize = (sizeof(SpoofDataHeader) + 15) & ~15; |
635 | U_ASSERT(initialSize == sizeof(SpoofDataHeader)); |
636 | |
637 | fRawData = static_cast<SpoofDataHeader *>(uprv_malloc(initialSize)); |
638 | fMemLimit = initialSize; |
639 | if (fRawData == nullptr) { |
640 | status = U_MEMORY_ALLOCATION_ERROR; |
641 | return; |
642 | } |
643 | uprv_memset(fRawData, 0, initialSize); |
644 | |
645 | fRawData->fMagic = USPOOF_MAGIC; |
646 | fRawData->fFormatVersion[0] = USPOOF_CONFUSABLE_DATA_FORMAT_VERSION; |
647 | fRawData->fFormatVersion[1] = 0; |
648 | fRawData->fFormatVersion[2] = 0; |
649 | fRawData->fFormatVersion[3] = 0; |
650 | initPtrs(status); |
651 | } |
652 | |
653 | // reset() - initialize all fields. |
654 | // Should be updated if any new fields are added. |
655 | // Called by constructors to put things in a known initial state. |
656 | void SpoofData::reset() { |
657 | fRawData = nullptr; |
658 | fDataOwned = false; |
659 | fUDM = nullptr; |
660 | fMemLimit = 0; |
661 | fRefCount = 1; |
662 | fCFUKeys = nullptr; |
663 | fCFUValues = nullptr; |
664 | fCFUStrings = nullptr; |
665 | } |
666 | |
667 | |
668 | // SpoofData::initPtrs() |
669 | // Initialize the pointers to the various sections of the raw data. |
670 | // |
671 | // This function is used both during the Trie building process (multiple |
672 | // times, as the individual data sections are added), and |
673 | // during the opening of a Spoof Checker from prebuilt data. |
674 | // |
675 | // The pointers for non-existent data sections (identified by an offset of 0) |
676 | // are set to nullptr. |
677 | // |
678 | // Note: During building the data, adding each new data section |
679 | // reallocs the raw data area, which likely relocates it, which |
680 | // in turn requires reinitializing all of the pointers into it, hence |
681 | // multiple calls to this function during building. |
682 | // |
683 | void SpoofData::initPtrs(UErrorCode &status) { |
684 | fCFUKeys = nullptr; |
685 | fCFUValues = nullptr; |
686 | fCFUStrings = nullptr; |
687 | if (U_FAILURE(status)) { |
688 | return; |
689 | } |
690 | if (fRawData->fCFUKeys != 0) { |
691 | fCFUKeys = (int32_t *)((char *)fRawData + fRawData->fCFUKeys); |
692 | } |
693 | if (fRawData->fCFUStringIndex != 0) { |
694 | fCFUValues = (uint16_t *)((char *)fRawData + fRawData->fCFUStringIndex); |
695 | } |
696 | if (fRawData->fCFUStringTable != 0) { |
697 | fCFUStrings = (char16_t *)((char *)fRawData + fRawData->fCFUStringTable); |
698 | } |
699 | } |
700 | |
701 | |
702 | SpoofData::~SpoofData() { |
703 | if (fDataOwned) { |
704 | uprv_free(fRawData); |
705 | } |
706 | fRawData = nullptr; |
707 | if (fUDM != nullptr) { |
708 | udata_close(fUDM); |
709 | } |
710 | fUDM = nullptr; |
711 | } |
712 | |
713 | |
714 | void SpoofData::removeReference() { |
715 | if (umtx_atomic_dec(&fRefCount) == 0) { |
716 | delete this; |
717 | } |
718 | } |
719 | |
720 | |
721 | SpoofData *SpoofData::addReference() { |
722 | umtx_atomic_inc(&fRefCount); |
723 | return this; |
724 | } |
725 | |
726 | |
727 | void *SpoofData::reserveSpace(int32_t numBytes, UErrorCode &status) { |
728 | if (U_FAILURE(status)) { |
729 | return nullptr; |
730 | } |
731 | if (!fDataOwned) { |
732 | UPRV_UNREACHABLE_EXIT; |
733 | } |
734 | |
735 | numBytes = (numBytes + 15) & ~15; // Round up to a multiple of 16 |
736 | uint32_t returnOffset = fMemLimit; |
737 | fMemLimit += numBytes; |
738 | fRawData = static_cast<SpoofDataHeader *>(uprv_realloc(fRawData, fMemLimit)); |
739 | fRawData->fLength = fMemLimit; |
740 | uprv_memset((char *)fRawData + returnOffset, 0, numBytes); |
741 | initPtrs(status); |
742 | return (char *)fRawData + returnOffset; |
743 | } |
744 | |
745 | int32_t SpoofData::serialize(void *buf, int32_t capacity, UErrorCode &status) const { |
746 | int32_t dataSize = fRawData->fLength; |
747 | if (capacity < dataSize) { |
748 | status = U_BUFFER_OVERFLOW_ERROR; |
749 | return dataSize; |
750 | } |
751 | uprv_memcpy(buf, fRawData, dataSize); |
752 | return dataSize; |
753 | } |
754 | |
755 | int32_t SpoofData::size() const { |
756 | return fRawData->fLength; |
757 | } |
758 | |
759 | //------------------------------- |
760 | // |
761 | // Front-end APIs for SpoofData |
762 | // |
763 | //------------------------------- |
764 | |
765 | int32_t SpoofData::confusableLookup(UChar32 inChar, UnicodeString &dest) const { |
766 | // Perform a binary search. |
767 | // [lo, hi), i.e lo is inclusive, hi is exclusive. |
768 | // The result after the loop will be in lo. |
769 | int32_t lo = 0; |
770 | int32_t hi = length(); |
771 | do { |
772 | int32_t mid = (lo + hi) / 2; |
773 | if (codePointAt(mid) > inChar) { |
774 | hi = mid; |
775 | } else if (codePointAt(mid) < inChar) { |
776 | lo = mid; |
777 | } else { |
778 | // Found result. Break early. |
779 | lo = mid; |
780 | break; |
781 | } |
782 | } while (hi - lo > 1); |
783 | |
784 | // Did we find an entry? If not, the char maps to itself. |
785 | if (codePointAt(lo) != inChar) { |
786 | dest.append(inChar); |
787 | return 1; |
788 | } |
789 | |
790 | // Add the element to the string builder and return. |
791 | return appendValueTo(lo, dest); |
792 | } |
793 | |
794 | int32_t SpoofData::length() const { |
795 | return fRawData->fCFUKeysSize; |
796 | } |
797 | |
798 | UChar32 SpoofData::codePointAt(int32_t index) const { |
799 | return ConfusableDataUtils::keyToCodePoint(fCFUKeys[index]); |
800 | } |
801 | |
802 | int32_t SpoofData::appendValueTo(int32_t index, UnicodeString& dest) const { |
803 | int32_t stringLength = ConfusableDataUtils::keyToLength(fCFUKeys[index]); |
804 | |
805 | // Value is either a char (for strings of length 1) or |
806 | // an index into the string table (for longer strings) |
807 | uint16_t value = fCFUValues[index]; |
808 | if (stringLength == 1) { |
809 | dest.append((char16_t)value); |
810 | } else { |
811 | dest.append(fCFUStrings + value, stringLength); |
812 | } |
813 | |
814 | return stringLength; |
815 | } |
816 | |
817 | |
818 | U_NAMESPACE_END |
819 | |
820 | U_NAMESPACE_USE |
821 | |
822 | //----------------------------------------------------------------------------- |
823 | // |
824 | // uspoof_swap - byte swap and char encoding swap of spoof data |
825 | // |
826 | //----------------------------------------------------------------------------- |
827 | U_CAPI int32_t U_EXPORT2 |
828 | uspoof_swap(const UDataSwapper *ds, const void *inData, int32_t length, void *outData, |
829 | UErrorCode *status) { |
830 | |
831 | if (status == nullptr || U_FAILURE(*status)) { |
832 | return 0; |
833 | } |
834 | if(ds==nullptr || inData==nullptr || length<-1 || (length>0 && outData==nullptr)) { |
835 | *status=U_ILLEGAL_ARGUMENT_ERROR; |
836 | return 0; |
837 | } |
838 | |
839 | // |
840 | // Check that the data header is for spoof data. |
841 | // (Header contents are defined in gencfu.cpp) |
842 | // |
843 | const UDataInfo *pInfo = (const UDataInfo *)((const char *)inData+4); |
844 | if(!( pInfo->dataFormat[0]==0x43 && /* dataFormat="Cfu " */ |
845 | pInfo->dataFormat[1]==0x66 && |
846 | pInfo->dataFormat[2]==0x75 && |
847 | pInfo->dataFormat[3]==0x20 && |
848 | pInfo->formatVersion[0]==USPOOF_CONFUSABLE_DATA_FORMAT_VERSION && |
849 | pInfo->formatVersion[1]==0 && |
850 | pInfo->formatVersion[2]==0 && |
851 | pInfo->formatVersion[3]==0 )) { |
852 | udata_printError(ds, "uspoof_swap(): data format %02x.%02x.%02x.%02x " |
853 | "(format version %02x %02x %02x %02x) is not recognized\n" , |
854 | pInfo->dataFormat[0], pInfo->dataFormat[1], |
855 | pInfo->dataFormat[2], pInfo->dataFormat[3], |
856 | pInfo->formatVersion[0], pInfo->formatVersion[1], |
857 | pInfo->formatVersion[2], pInfo->formatVersion[3]); |
858 | *status=U_UNSUPPORTED_ERROR; |
859 | return 0; |
860 | } |
861 | |
862 | // |
863 | // Swap the data header. (This is the generic ICU Data Header, not the uspoof Specific |
864 | // header). This swap also conveniently gets us |
865 | // the size of the ICU d.h., which lets us locate the start |
866 | // of the uspoof specific data. |
867 | // |
868 | int32_t =udata_swapDataHeader(ds, inData, length, outData, status); |
869 | |
870 | |
871 | // |
872 | // Get the Spoof Data Header, and check that it appears to be OK. |
873 | // |
874 | // |
875 | const uint8_t *inBytes =(const uint8_t *)inData+headerSize; |
876 | SpoofDataHeader *spoofDH = (SpoofDataHeader *)inBytes; |
877 | if (ds->readUInt32(spoofDH->fMagic) != USPOOF_MAGIC || |
878 | ds->readUInt32(spoofDH->fLength) < sizeof(SpoofDataHeader)) |
879 | { |
880 | udata_printError(ds, "uspoof_swap(): Spoof Data header is invalid.\n" ); |
881 | *status=U_UNSUPPORTED_ERROR; |
882 | return 0; |
883 | } |
884 | |
885 | // |
886 | // Prefight operation? Just return the size |
887 | // |
888 | int32_t spoofDataLength = ds->readUInt32(spoofDH->fLength); |
889 | int32_t totalSize = headerSize + spoofDataLength; |
890 | if (length < 0) { |
891 | return totalSize; |
892 | } |
893 | |
894 | // |
895 | // Check that length passed in is consistent with length from Spoof data header. |
896 | // |
897 | if (length < totalSize) { |
898 | udata_printError(ds, "uspoof_swap(): too few bytes (%d after ICU Data header) for spoof data.\n" , |
899 | spoofDataLength); |
900 | *status=U_INDEX_OUTOFBOUNDS_ERROR; |
901 | return 0; |
902 | } |
903 | |
904 | |
905 | // |
906 | // Swap the Data. Do the data itself first, then the Spoof Data Header, because |
907 | // we need to reference the header to locate the data, and an |
908 | // inplace swap of the header leaves it unusable. |
909 | // |
910 | uint8_t *outBytes = (uint8_t *)outData + headerSize; |
911 | SpoofDataHeader *outputDH = (SpoofDataHeader *)outBytes; |
912 | |
913 | int32_t sectionStart; |
914 | int32_t sectionLength; |
915 | |
916 | // |
917 | // If not swapping in place, zero out the output buffer before starting. |
918 | // Gaps may exist between the individual sections, and these must be zeroed in |
919 | // the output buffer. The simplest way to do that is to just zero the whole thing. |
920 | // |
921 | if (inBytes != outBytes) { |
922 | uprv_memset(outBytes, 0, spoofDataLength); |
923 | } |
924 | |
925 | // Confusables Keys Section (fCFUKeys) |
926 | sectionStart = ds->readUInt32(spoofDH->fCFUKeys); |
927 | sectionLength = ds->readUInt32(spoofDH->fCFUKeysSize) * 4; |
928 | ds->swapArray32(ds, inBytes+sectionStart, sectionLength, outBytes+sectionStart, status); |
929 | |
930 | // String Index Section |
931 | sectionStart = ds->readUInt32(spoofDH->fCFUStringIndex); |
932 | sectionLength = ds->readUInt32(spoofDH->fCFUStringIndexSize) * 2; |
933 | ds->swapArray16(ds, inBytes+sectionStart, sectionLength, outBytes+sectionStart, status); |
934 | |
935 | // String Table Section |
936 | sectionStart = ds->readUInt32(spoofDH->fCFUStringTable); |
937 | sectionLength = ds->readUInt32(spoofDH->fCFUStringTableLen) * 2; |
938 | ds->swapArray16(ds, inBytes+sectionStart, sectionLength, outBytes+sectionStart, status); |
939 | |
940 | // And, last, swap the header itself. |
941 | // int32_t fMagic // swap this |
942 | // uint8_t fFormatVersion[4] // Do not swap this, just copy |
943 | // int32_t fLength and all the rest // Swap the rest, all is 32 bit stuff. |
944 | // |
945 | uint32_t magic = ds->readUInt32(spoofDH->fMagic); |
946 | ds->writeUInt32((uint32_t *)&outputDH->fMagic, magic); |
947 | |
948 | if (inBytes != outBytes) { |
949 | uprv_memcpy(outputDH->fFormatVersion, spoofDH->fFormatVersion, sizeof(spoofDH->fFormatVersion)); |
950 | } |
951 | // swap starting at fLength |
952 | ds->swapArray32(ds, &spoofDH->fLength, sizeof(SpoofDataHeader)-8 /* minus magic and fFormatVersion[4] */, &outputDH->fLength, status); |
953 | |
954 | return totalSize; |
955 | } |
956 | |
957 | #endif |
958 | |
959 | |
960 | |