/src/icu/icu4c/source/i18n/rematch.cpp
Line | Count | Source (jump to first uncovered line) |
1 | | // © 2016 and later: Unicode, Inc. and others. |
2 | | // License & terms of use: http://www.unicode.org/copyright.html |
3 | | /* |
4 | | ************************************************************************** |
5 | | * Copyright (C) 2002-2016 International Business Machines Corporation |
6 | | * and others. All rights reserved. |
7 | | ************************************************************************** |
8 | | */ |
9 | | // |
10 | | // file: rematch.cpp |
11 | | // |
12 | | // Contains the implementation of class RegexMatcher, |
13 | | // which is one of the main API classes for the ICU regular expression package. |
14 | | // |
15 | | |
16 | | #include "unicode/utypes.h" |
17 | | #if !UCONFIG_NO_REGULAR_EXPRESSIONS |
18 | | |
19 | | #include "unicode/regex.h" |
20 | | #include "unicode/uniset.h" |
21 | | #include "unicode/uchar.h" |
22 | | #include "unicode/ustring.h" |
23 | | #include "unicode/rbbi.h" |
24 | | #include "unicode/utf.h" |
25 | | #include "unicode/utf16.h" |
26 | | #include "uassert.h" |
27 | | #include "cmemory.h" |
28 | | #include "cstr.h" |
29 | | #include "uvector.h" |
30 | | #include "uvectr32.h" |
31 | | #include "uvectr64.h" |
32 | | #include "regeximp.h" |
33 | | #include "regexst.h" |
34 | | #include "regextxt.h" |
35 | | #include "ucase.h" |
36 | | |
37 | | // #include <malloc.h> // Needed for heapcheck testing |
38 | | |
39 | | |
40 | | U_NAMESPACE_BEGIN |
41 | | |
42 | | // Default limit for the size of the back track stack, to avoid system |
43 | | // failures causedby heap exhaustion. Units are in 32 bit words, not bytes. |
44 | | // This value puts ICU's limits higher than most other regexp implementations, |
45 | | // which use recursion rather than the heap, and take more storage per |
46 | | // backtrack point. |
47 | | // |
48 | | static const int32_t DEFAULT_BACKTRACK_STACK_CAPACITY = 8000000; |
49 | | |
50 | | // Time limit counter constant. |
51 | | // Time limits for expression evaluation are in terms of quanta of work by |
52 | | // the engine, each of which is 10,000 state saves. |
53 | | // This constant determines that state saves per tick number. |
54 | | static const int32_t TIMER_INITIAL_VALUE = 10000; |
55 | | |
56 | | |
57 | | // Test for any of the Unicode line terminating characters. |
58 | 0 | static inline UBool isLineTerminator(UChar32 c) { |
59 | 0 | if (c & ~(0x0a | 0x0b | 0x0c | 0x0d | 0x85 | 0x2028 | 0x2029)) { |
60 | 0 | return false; |
61 | 0 | } |
62 | 0 | return (c<=0x0d && c>=0x0a) || c==0x85 || c==0x2028 || c==0x2029; |
63 | 0 | } |
64 | | |
65 | | //----------------------------------------------------------------------------- |
66 | | // |
67 | | // Constructor and Destructor |
68 | | // |
69 | | //----------------------------------------------------------------------------- |
70 | 3.94k | RegexMatcher::RegexMatcher(const RegexPattern *pat) { |
71 | 3.94k | fDeferredStatus = U_ZERO_ERROR; |
72 | 3.94k | init(fDeferredStatus); |
73 | 3.94k | if (U_FAILURE(fDeferredStatus)) { |
74 | 0 | return; |
75 | 0 | } |
76 | 3.94k | if (pat==NULL) { |
77 | 0 | fDeferredStatus = U_ILLEGAL_ARGUMENT_ERROR; |
78 | 0 | return; |
79 | 0 | } |
80 | 3.94k | fPattern = pat; |
81 | 3.94k | init2(RegexStaticSets::gStaticSets->fEmptyText, fDeferredStatus); |
82 | 3.94k | } |
83 | | |
84 | | |
85 | | |
86 | | RegexMatcher::RegexMatcher(const UnicodeString ®exp, const UnicodeString &input, |
87 | 0 | uint32_t flags, UErrorCode &status) { |
88 | 0 | init(status); |
89 | 0 | if (U_FAILURE(status)) { |
90 | 0 | return; |
91 | 0 | } |
92 | 0 | UParseError pe; |
93 | 0 | fPatternOwned = RegexPattern::compile(regexp, flags, pe, status); |
94 | 0 | fPattern = fPatternOwned; |
95 | |
|
96 | 0 | UText inputText = UTEXT_INITIALIZER; |
97 | 0 | utext_openConstUnicodeString(&inputText, &input, &status); |
98 | 0 | init2(&inputText, status); |
99 | 0 | utext_close(&inputText); |
100 | |
|
101 | 0 | fInputUniStrMaybeMutable = true; |
102 | 0 | } |
103 | | |
104 | | |
105 | | RegexMatcher::RegexMatcher(UText *regexp, UText *input, |
106 | 0 | uint32_t flags, UErrorCode &status) { |
107 | 0 | init(status); |
108 | 0 | if (U_FAILURE(status)) { |
109 | 0 | return; |
110 | 0 | } |
111 | 0 | UParseError pe; |
112 | 0 | fPatternOwned = RegexPattern::compile(regexp, flags, pe, status); |
113 | 0 | if (U_FAILURE(status)) { |
114 | 0 | return; |
115 | 0 | } |
116 | | |
117 | 0 | fPattern = fPatternOwned; |
118 | 0 | init2(input, status); |
119 | 0 | } |
120 | | |
121 | | |
122 | | RegexMatcher::RegexMatcher(const UnicodeString ®exp, |
123 | 0 | uint32_t flags, UErrorCode &status) { |
124 | 0 | init(status); |
125 | 0 | if (U_FAILURE(status)) { |
126 | 0 | return; |
127 | 0 | } |
128 | 0 | UParseError pe; |
129 | 0 | fPatternOwned = RegexPattern::compile(regexp, flags, pe, status); |
130 | 0 | if (U_FAILURE(status)) { |
131 | 0 | return; |
132 | 0 | } |
133 | 0 | fPattern = fPatternOwned; |
134 | 0 | init2(RegexStaticSets::gStaticSets->fEmptyText, status); |
135 | 0 | } |
136 | | |
137 | | RegexMatcher::RegexMatcher(UText *regexp, |
138 | 0 | uint32_t flags, UErrorCode &status) { |
139 | 0 | init(status); |
140 | 0 | if (U_FAILURE(status)) { |
141 | 0 | return; |
142 | 0 | } |
143 | 0 | UParseError pe; |
144 | 0 | fPatternOwned = RegexPattern::compile(regexp, flags, pe, status); |
145 | 0 | if (U_FAILURE(status)) { |
146 | 0 | return; |
147 | 0 | } |
148 | | |
149 | 0 | fPattern = fPatternOwned; |
150 | 0 | init2(RegexStaticSets::gStaticSets->fEmptyText, status); |
151 | 0 | } |
152 | | |
153 | | |
154 | | |
155 | | |
156 | 3.94k | RegexMatcher::~RegexMatcher() { |
157 | 3.94k | delete fStack; |
158 | 3.94k | if (fData != fSmallData) { |
159 | 144 | uprv_free(fData); |
160 | 144 | fData = NULL; |
161 | 144 | } |
162 | 3.94k | if (fPatternOwned) { |
163 | 0 | delete fPatternOwned; |
164 | 0 | fPatternOwned = NULL; |
165 | 0 | fPattern = NULL; |
166 | 0 | } |
167 | | |
168 | 3.94k | if (fInput) { |
169 | 0 | delete fInput; |
170 | 0 | } |
171 | 3.94k | if (fInputText) { |
172 | 3.94k | utext_close(fInputText); |
173 | 3.94k | } |
174 | 3.94k | if (fAltInputText) { |
175 | 51 | utext_close(fAltInputText); |
176 | 51 | } |
177 | | |
178 | 3.94k | #if UCONFIG_NO_BREAK_ITERATION==0 |
179 | 3.94k | delete fWordBreakItr; |
180 | 3.94k | delete fGCBreakItr; |
181 | 3.94k | #endif |
182 | 3.94k | } |
183 | | |
184 | | // |
185 | | // init() common initialization for use by all constructors. |
186 | | // Initialize all fields, get the object into a consistent state. |
187 | | // This must be done even when the initial status shows an error, |
188 | | // so that the object is initialized sufficiently well for the destructor |
189 | | // to run safely. |
190 | | // |
191 | 3.94k | void RegexMatcher::init(UErrorCode &status) { |
192 | 3.94k | fPattern = NULL; |
193 | 3.94k | fPatternOwned = NULL; |
194 | 3.94k | fFrameSize = 0; |
195 | 3.94k | fRegionStart = 0; |
196 | 3.94k | fRegionLimit = 0; |
197 | 3.94k | fAnchorStart = 0; |
198 | 3.94k | fAnchorLimit = 0; |
199 | 3.94k | fLookStart = 0; |
200 | 3.94k | fLookLimit = 0; |
201 | 3.94k | fActiveStart = 0; |
202 | 3.94k | fActiveLimit = 0; |
203 | 3.94k | fTransparentBounds = false; |
204 | 3.94k | fAnchoringBounds = true; |
205 | 3.94k | fMatch = false; |
206 | 3.94k | fMatchStart = 0; |
207 | 3.94k | fMatchEnd = 0; |
208 | 3.94k | fLastMatchEnd = -1; |
209 | 3.94k | fAppendPosition = 0; |
210 | 3.94k | fHitEnd = false; |
211 | 3.94k | fRequireEnd = false; |
212 | 3.94k | fStack = NULL; |
213 | 3.94k | fFrame = NULL; |
214 | 3.94k | fTimeLimit = 0; |
215 | 3.94k | fTime = 0; |
216 | 3.94k | fTickCounter = 0; |
217 | 3.94k | fStackLimit = DEFAULT_BACKTRACK_STACK_CAPACITY; |
218 | 3.94k | fCallbackFn = NULL; |
219 | 3.94k | fCallbackContext = NULL; |
220 | 3.94k | fFindProgressCallbackFn = NULL; |
221 | 3.94k | fFindProgressCallbackContext = NULL; |
222 | 3.94k | fTraceDebug = false; |
223 | 3.94k | fDeferredStatus = status; |
224 | 3.94k | fData = fSmallData; |
225 | 3.94k | fWordBreakItr = NULL; |
226 | 3.94k | fGCBreakItr = NULL; |
227 | | |
228 | 3.94k | fStack = NULL; |
229 | 3.94k | fInputText = NULL; |
230 | 3.94k | fAltInputText = NULL; |
231 | 3.94k | fInput = NULL; |
232 | 3.94k | fInputLength = 0; |
233 | 3.94k | fInputUniStrMaybeMutable = false; |
234 | 3.94k | } |
235 | | |
236 | | // |
237 | | // init2() Common initialization for use by RegexMatcher constructors, part 2. |
238 | | // This handles the common setup to be done after the Pattern is available. |
239 | | // |
240 | 3.94k | void RegexMatcher::init2(UText *input, UErrorCode &status) { |
241 | 3.94k | if (U_FAILURE(status)) { |
242 | 0 | fDeferredStatus = status; |
243 | 0 | return; |
244 | 0 | } |
245 | | |
246 | 3.94k | if (fPattern->fDataSize > UPRV_LENGTHOF(fSmallData)) { |
247 | 144 | fData = (int64_t *)uprv_malloc(fPattern->fDataSize * sizeof(int64_t)); |
248 | 144 | if (fData == NULL) { |
249 | 0 | status = fDeferredStatus = U_MEMORY_ALLOCATION_ERROR; |
250 | 0 | return; |
251 | 0 | } |
252 | 144 | } |
253 | | |
254 | 3.94k | fStack = new UVector64(status); |
255 | 3.94k | if (fStack == NULL) { |
256 | 0 | status = fDeferredStatus = U_MEMORY_ALLOCATION_ERROR; |
257 | 0 | return; |
258 | 0 | } |
259 | | |
260 | 3.94k | reset(input); |
261 | 3.94k | setStackLimit(DEFAULT_BACKTRACK_STACK_CAPACITY, status); |
262 | 3.94k | if (U_FAILURE(status)) { |
263 | 0 | fDeferredStatus = status; |
264 | 0 | return; |
265 | 0 | } |
266 | 3.94k | } |
267 | | |
268 | | |
269 | | static const UChar BACKSLASH = 0x5c; |
270 | | static const UChar DOLLARSIGN = 0x24; |
271 | | static const UChar LEFTBRACKET = 0x7b; |
272 | | static const UChar RIGHTBRACKET = 0x7d; |
273 | | |
274 | | //-------------------------------------------------------------------------------- |
275 | | // |
276 | | // appendReplacement |
277 | | // |
278 | | //-------------------------------------------------------------------------------- |
279 | | RegexMatcher &RegexMatcher::appendReplacement(UnicodeString &dest, |
280 | | const UnicodeString &replacement, |
281 | 0 | UErrorCode &status) { |
282 | 0 | UText replacementText = UTEXT_INITIALIZER; |
283 | |
|
284 | 0 | utext_openConstUnicodeString(&replacementText, &replacement, &status); |
285 | 0 | if (U_SUCCESS(status)) { |
286 | 0 | UText resultText = UTEXT_INITIALIZER; |
287 | 0 | utext_openUnicodeString(&resultText, &dest, &status); |
288 | |
|
289 | 0 | if (U_SUCCESS(status)) { |
290 | 0 | appendReplacement(&resultText, &replacementText, status); |
291 | 0 | utext_close(&resultText); |
292 | 0 | } |
293 | 0 | utext_close(&replacementText); |
294 | 0 | } |
295 | |
|
296 | 0 | return *this; |
297 | 0 | } |
298 | | |
299 | | // |
300 | | // appendReplacement, UText mode |
301 | | // |
302 | | RegexMatcher &RegexMatcher::appendReplacement(UText *dest, |
303 | | UText *replacement, |
304 | 0 | UErrorCode &status) { |
305 | 0 | if (U_FAILURE(status)) { |
306 | 0 | return *this; |
307 | 0 | } |
308 | 0 | if (U_FAILURE(fDeferredStatus)) { |
309 | 0 | status = fDeferredStatus; |
310 | 0 | return *this; |
311 | 0 | } |
312 | 0 | if (fMatch == false) { |
313 | 0 | status = U_REGEX_INVALID_STATE; |
314 | 0 | return *this; |
315 | 0 | } |
316 | | |
317 | | // Copy input string from the end of previous match to start of current match |
318 | 0 | int64_t destLen = utext_nativeLength(dest); |
319 | 0 | if (fMatchStart > fAppendPosition) { |
320 | 0 | if (UTEXT_FULL_TEXT_IN_CHUNK(fInputText, fInputLength)) { |
321 | 0 | destLen += utext_replace(dest, destLen, destLen, fInputText->chunkContents+fAppendPosition, |
322 | 0 | (int32_t)(fMatchStart-fAppendPosition), &status); |
323 | 0 | } else { |
324 | 0 | int32_t len16; |
325 | 0 | if (UTEXT_USES_U16(fInputText)) { |
326 | 0 | len16 = (int32_t)(fMatchStart-fAppendPosition); |
327 | 0 | } else { |
328 | 0 | UErrorCode lengthStatus = U_ZERO_ERROR; |
329 | 0 | len16 = utext_extract(fInputText, fAppendPosition, fMatchStart, NULL, 0, &lengthStatus); |
330 | 0 | } |
331 | 0 | UChar *inputChars = (UChar *)uprv_malloc(sizeof(UChar)*(len16+1)); |
332 | 0 | if (inputChars == NULL) { |
333 | 0 | status = U_MEMORY_ALLOCATION_ERROR; |
334 | 0 | return *this; |
335 | 0 | } |
336 | 0 | utext_extract(fInputText, fAppendPosition, fMatchStart, inputChars, len16+1, &status); |
337 | 0 | destLen += utext_replace(dest, destLen, destLen, inputChars, len16, &status); |
338 | 0 | uprv_free(inputChars); |
339 | 0 | } |
340 | 0 | } |
341 | 0 | fAppendPosition = fMatchEnd; |
342 | | |
343 | | |
344 | | // scan the replacement text, looking for substitutions ($n) and \escapes. |
345 | | // TODO: optimize this loop by efficiently scanning for '$' or '\', |
346 | | // move entire ranges not containing substitutions. |
347 | 0 | UTEXT_SETNATIVEINDEX(replacement, 0); |
348 | 0 | for (UChar32 c = UTEXT_NEXT32(replacement); U_SUCCESS(status) && c != U_SENTINEL; c = UTEXT_NEXT32(replacement)) { |
349 | 0 | if (c == BACKSLASH) { |
350 | | // Backslash Escape. Copy the following char out without further checks. |
351 | | // Note: Surrogate pairs don't need any special handling |
352 | | // The second half wont be a '$' or a '\', and |
353 | | // will move to the dest normally on the next |
354 | | // loop iteration. |
355 | 0 | c = UTEXT_CURRENT32(replacement); |
356 | 0 | if (c == U_SENTINEL) { |
357 | 0 | break; |
358 | 0 | } |
359 | | |
360 | 0 | if (c==0x55/*U*/ || c==0x75/*u*/) { |
361 | | // We have a \udddd or \Udddddddd escape sequence. |
362 | 0 | int32_t offset = 0; |
363 | 0 | struct URegexUTextUnescapeCharContext context = U_REGEX_UTEXT_UNESCAPE_CONTEXT(replacement); |
364 | 0 | UChar32 escapedChar = u_unescapeAt(uregex_utext_unescape_charAt, &offset, INT32_MAX, &context); |
365 | 0 | if (escapedChar != (UChar32)0xFFFFFFFF) { |
366 | 0 | if (U_IS_BMP(escapedChar)) { |
367 | 0 | UChar c16 = (UChar)escapedChar; |
368 | 0 | destLen += utext_replace(dest, destLen, destLen, &c16, 1, &status); |
369 | 0 | } else { |
370 | 0 | UChar surrogate[2]; |
371 | 0 | surrogate[0] = U16_LEAD(escapedChar); |
372 | 0 | surrogate[1] = U16_TRAIL(escapedChar); |
373 | 0 | if (U_SUCCESS(status)) { |
374 | 0 | destLen += utext_replace(dest, destLen, destLen, surrogate, 2, &status); |
375 | 0 | } |
376 | 0 | } |
377 | | // TODO: Report errors for mal-formed \u escapes? |
378 | | // As this is, the original sequence is output, which may be OK. |
379 | 0 | if (context.lastOffset == offset) { |
380 | 0 | (void)UTEXT_PREVIOUS32(replacement); |
381 | 0 | } else if (context.lastOffset != offset-1) { |
382 | 0 | utext_moveIndex32(replacement, offset - context.lastOffset - 1); |
383 | 0 | } |
384 | 0 | } |
385 | 0 | } else { |
386 | 0 | (void)UTEXT_NEXT32(replacement); |
387 | | // Plain backslash escape. Just put out the escaped character. |
388 | 0 | if (U_IS_BMP(c)) { |
389 | 0 | UChar c16 = (UChar)c; |
390 | 0 | destLen += utext_replace(dest, destLen, destLen, &c16, 1, &status); |
391 | 0 | } else { |
392 | 0 | UChar surrogate[2]; |
393 | 0 | surrogate[0] = U16_LEAD(c); |
394 | 0 | surrogate[1] = U16_TRAIL(c); |
395 | 0 | if (U_SUCCESS(status)) { |
396 | 0 | destLen += utext_replace(dest, destLen, destLen, surrogate, 2, &status); |
397 | 0 | } |
398 | 0 | } |
399 | 0 | } |
400 | 0 | } else if (c != DOLLARSIGN) { |
401 | | // Normal char, not a $. Copy it out without further checks. |
402 | 0 | if (U_IS_BMP(c)) { |
403 | 0 | UChar c16 = (UChar)c; |
404 | 0 | destLen += utext_replace(dest, destLen, destLen, &c16, 1, &status); |
405 | 0 | } else { |
406 | 0 | UChar surrogate[2]; |
407 | 0 | surrogate[0] = U16_LEAD(c); |
408 | 0 | surrogate[1] = U16_TRAIL(c); |
409 | 0 | if (U_SUCCESS(status)) { |
410 | 0 | destLen += utext_replace(dest, destLen, destLen, surrogate, 2, &status); |
411 | 0 | } |
412 | 0 | } |
413 | 0 | } else { |
414 | | // We've got a $. Pick up a capture group name or number if one follows. |
415 | | // Consume digits so long as the resulting group number <= the number of |
416 | | // number of capture groups in the pattern. |
417 | |
|
418 | 0 | int32_t groupNum = 0; |
419 | 0 | int32_t numDigits = 0; |
420 | 0 | UChar32 nextChar = utext_current32(replacement); |
421 | 0 | if (nextChar == LEFTBRACKET) { |
422 | | // Scan for a Named Capture Group, ${name}. |
423 | 0 | UnicodeString groupName; |
424 | 0 | utext_next32(replacement); |
425 | 0 | while(U_SUCCESS(status) && nextChar != RIGHTBRACKET) { |
426 | 0 | nextChar = utext_next32(replacement); |
427 | 0 | if (nextChar == U_SENTINEL) { |
428 | 0 | status = U_REGEX_INVALID_CAPTURE_GROUP_NAME; |
429 | 0 | } else if ((nextChar >= 0x41 && nextChar <= 0x5a) || // A..Z |
430 | 0 | (nextChar >= 0x61 && nextChar <= 0x7a) || // a..z |
431 | 0 | (nextChar >= 0x31 && nextChar <= 0x39)) { // 0..9 |
432 | 0 | groupName.append(nextChar); |
433 | 0 | } else if (nextChar == RIGHTBRACKET) { |
434 | 0 | groupNum = fPattern->fNamedCaptureMap ? uhash_geti(fPattern->fNamedCaptureMap, &groupName) : 0; |
435 | 0 | if (groupNum == 0) { |
436 | 0 | status = U_REGEX_INVALID_CAPTURE_GROUP_NAME; |
437 | 0 | } |
438 | 0 | } else { |
439 | | // Character was something other than a name char or a closing '}' |
440 | 0 | status = U_REGEX_INVALID_CAPTURE_GROUP_NAME; |
441 | 0 | } |
442 | 0 | } |
443 | |
|
444 | 0 | } else if (u_isdigit(nextChar)) { |
445 | | // $n Scan for a capture group number |
446 | 0 | int32_t numCaptureGroups = fPattern->fGroupMap->size(); |
447 | 0 | for (;;) { |
448 | 0 | nextChar = UTEXT_CURRENT32(replacement); |
449 | 0 | if (nextChar == U_SENTINEL) { |
450 | 0 | break; |
451 | 0 | } |
452 | 0 | if (u_isdigit(nextChar) == false) { |
453 | 0 | break; |
454 | 0 | } |
455 | 0 | int32_t nextDigitVal = u_charDigitValue(nextChar); |
456 | 0 | if (groupNum*10 + nextDigitVal > numCaptureGroups) { |
457 | | // Don't consume the next digit if it makes the capture group number too big. |
458 | 0 | if (numDigits == 0) { |
459 | 0 | status = U_INDEX_OUTOFBOUNDS_ERROR; |
460 | 0 | } |
461 | 0 | break; |
462 | 0 | } |
463 | 0 | (void)UTEXT_NEXT32(replacement); |
464 | 0 | groupNum=groupNum*10 + nextDigitVal; |
465 | 0 | ++numDigits; |
466 | 0 | } |
467 | 0 | } else { |
468 | | // $ not followed by capture group name or number. |
469 | 0 | status = U_REGEX_INVALID_CAPTURE_GROUP_NAME; |
470 | 0 | } |
471 | |
|
472 | 0 | if (U_SUCCESS(status)) { |
473 | 0 | destLen += appendGroup(groupNum, dest, status); |
474 | 0 | } |
475 | 0 | } // End of $ capture group handling |
476 | 0 | } // End of per-character loop through the replacement string. |
477 | |
|
478 | 0 | return *this; |
479 | 0 | } |
480 | | |
481 | | |
482 | | |
483 | | //-------------------------------------------------------------------------------- |
484 | | // |
485 | | // appendTail Intended to be used in conjunction with appendReplacement() |
486 | | // To the destination string, append everything following |
487 | | // the last match position from the input string. |
488 | | // |
489 | | // Note: Match ranges do not affect appendTail or appendReplacement |
490 | | // |
491 | | //-------------------------------------------------------------------------------- |
492 | 0 | UnicodeString &RegexMatcher::appendTail(UnicodeString &dest) { |
493 | 0 | UErrorCode status = U_ZERO_ERROR; |
494 | 0 | UText resultText = UTEXT_INITIALIZER; |
495 | 0 | utext_openUnicodeString(&resultText, &dest, &status); |
496 | |
|
497 | 0 | if (U_SUCCESS(status)) { |
498 | 0 | appendTail(&resultText, status); |
499 | 0 | utext_close(&resultText); |
500 | 0 | } |
501 | |
|
502 | 0 | return dest; |
503 | 0 | } |
504 | | |
505 | | // |
506 | | // appendTail, UText mode |
507 | | // |
508 | 0 | UText *RegexMatcher::appendTail(UText *dest, UErrorCode &status) { |
509 | 0 | if (U_FAILURE(status)) { |
510 | 0 | return dest; |
511 | 0 | } |
512 | 0 | if (U_FAILURE(fDeferredStatus)) { |
513 | 0 | status = fDeferredStatus; |
514 | 0 | return dest; |
515 | 0 | } |
516 | | |
517 | 0 | if (fInputLength > fAppendPosition) { |
518 | 0 | if (UTEXT_FULL_TEXT_IN_CHUNK(fInputText, fInputLength)) { |
519 | 0 | int64_t destLen = utext_nativeLength(dest); |
520 | 0 | utext_replace(dest, destLen, destLen, fInputText->chunkContents+fAppendPosition, |
521 | 0 | (int32_t)(fInputLength-fAppendPosition), &status); |
522 | 0 | } else { |
523 | 0 | int32_t len16; |
524 | 0 | if (UTEXT_USES_U16(fInputText)) { |
525 | 0 | len16 = (int32_t)(fInputLength-fAppendPosition); |
526 | 0 | } else { |
527 | 0 | len16 = utext_extract(fInputText, fAppendPosition, fInputLength, NULL, 0, &status); |
528 | 0 | status = U_ZERO_ERROR; // buffer overflow |
529 | 0 | } |
530 | |
|
531 | 0 | UChar *inputChars = (UChar *)uprv_malloc(sizeof(UChar)*(len16)); |
532 | 0 | if (inputChars == NULL) { |
533 | 0 | fDeferredStatus = U_MEMORY_ALLOCATION_ERROR; |
534 | 0 | } else { |
535 | 0 | utext_extract(fInputText, fAppendPosition, fInputLength, inputChars, len16, &status); // unterminated |
536 | 0 | int64_t destLen = utext_nativeLength(dest); |
537 | 0 | utext_replace(dest, destLen, destLen, inputChars, len16, &status); |
538 | 0 | uprv_free(inputChars); |
539 | 0 | } |
540 | 0 | } |
541 | 0 | } |
542 | 0 | return dest; |
543 | 0 | } |
544 | | |
545 | | |
546 | | |
547 | | //-------------------------------------------------------------------------------- |
548 | | // |
549 | | // end |
550 | | // |
551 | | //-------------------------------------------------------------------------------- |
552 | 0 | int32_t RegexMatcher::end(UErrorCode &err) const { |
553 | 0 | return end(0, err); |
554 | 0 | } |
555 | | |
556 | 0 | int64_t RegexMatcher::end64(UErrorCode &err) const { |
557 | 0 | return end64(0, err); |
558 | 0 | } |
559 | | |
560 | 0 | int64_t RegexMatcher::end64(int32_t group, UErrorCode &err) const { |
561 | 0 | if (U_FAILURE(err)) { |
562 | 0 | return -1; |
563 | 0 | } |
564 | 0 | if (fMatch == false) { |
565 | 0 | err = U_REGEX_INVALID_STATE; |
566 | 0 | return -1; |
567 | 0 | } |
568 | 0 | if (group < 0 || group > fPattern->fGroupMap->size()) { |
569 | 0 | err = U_INDEX_OUTOFBOUNDS_ERROR; |
570 | 0 | return -1; |
571 | 0 | } |
572 | 0 | int64_t e = -1; |
573 | 0 | if (group == 0) { |
574 | 0 | e = fMatchEnd; |
575 | 0 | } else { |
576 | | // Get the position within the stack frame of the variables for |
577 | | // this capture group. |
578 | 0 | int32_t groupOffset = fPattern->fGroupMap->elementAti(group-1); |
579 | 0 | U_ASSERT(groupOffset < fPattern->fFrameSize); |
580 | 0 | U_ASSERT(groupOffset >= 0); |
581 | 0 | e = fFrame->fExtra[groupOffset + 1]; |
582 | 0 | } |
583 | |
|
584 | 0 | return e; |
585 | 0 | } |
586 | | |
587 | 0 | int32_t RegexMatcher::end(int32_t group, UErrorCode &err) const { |
588 | 0 | return (int32_t)end64(group, err); |
589 | 0 | } |
590 | | |
591 | | //-------------------------------------------------------------------------------- |
592 | | // |
593 | | // findProgressInterrupt This function is called once for each advance in the target |
594 | | // string from the find() function, and calls the user progress callback |
595 | | // function if there is one installed. |
596 | | // |
597 | | // Return: true if the find operation is to be terminated. |
598 | | // false if the find operation is to continue running. |
599 | | // |
600 | | //-------------------------------------------------------------------------------- |
601 | 0 | UBool RegexMatcher::findProgressInterrupt(int64_t pos, UErrorCode &status) { |
602 | 0 | if (fFindProgressCallbackFn && !(*fFindProgressCallbackFn)(fFindProgressCallbackContext, pos)) { |
603 | 0 | status = U_REGEX_STOPPED_BY_CALLER; |
604 | 0 | return true; |
605 | 0 | } |
606 | 0 | return false; |
607 | 0 | } |
608 | | |
609 | | //-------------------------------------------------------------------------------- |
610 | | // |
611 | | // find() |
612 | | // |
613 | | //-------------------------------------------------------------------------------- |
614 | 0 | UBool RegexMatcher::find() { |
615 | 0 | if (U_FAILURE(fDeferredStatus)) { |
616 | 0 | return false; |
617 | 0 | } |
618 | 0 | UErrorCode status = U_ZERO_ERROR; |
619 | 0 | UBool result = find(status); |
620 | 0 | return result; |
621 | 0 | } |
622 | | |
623 | | //-------------------------------------------------------------------------------- |
624 | | // |
625 | | // find() |
626 | | // |
627 | | //-------------------------------------------------------------------------------- |
628 | 0 | UBool RegexMatcher::find(UErrorCode &status) { |
629 | | // Start at the position of the last match end. (Will be zero if the |
630 | | // matcher has been reset.) |
631 | | // |
632 | 0 | if (U_FAILURE(status)) { |
633 | 0 | return false; |
634 | 0 | } |
635 | 0 | if (U_FAILURE(fDeferredStatus)) { |
636 | 0 | status = fDeferredStatus; |
637 | 0 | return false; |
638 | 0 | } |
639 | | |
640 | 0 | if (UTEXT_FULL_TEXT_IN_CHUNK(fInputText, fInputLength)) { |
641 | 0 | return findUsingChunk(status); |
642 | 0 | } |
643 | | |
644 | 0 | int64_t startPos = fMatchEnd; |
645 | 0 | if (startPos==0) { |
646 | 0 | startPos = fActiveStart; |
647 | 0 | } |
648 | |
|
649 | 0 | if (fMatch) { |
650 | | // Save the position of any previous successful match. |
651 | 0 | fLastMatchEnd = fMatchEnd; |
652 | |
|
653 | 0 | if (fMatchStart == fMatchEnd) { |
654 | | // Previous match had zero length. Move start position up one position |
655 | | // to avoid sending find() into a loop on zero-length matches. |
656 | 0 | if (startPos >= fActiveLimit) { |
657 | 0 | fMatch = false; |
658 | 0 | fHitEnd = true; |
659 | 0 | return false; |
660 | 0 | } |
661 | 0 | UTEXT_SETNATIVEINDEX(fInputText, startPos); |
662 | 0 | (void)UTEXT_NEXT32(fInputText); |
663 | 0 | startPos = UTEXT_GETNATIVEINDEX(fInputText); |
664 | 0 | } |
665 | 0 | } else { |
666 | 0 | if (fLastMatchEnd >= 0) { |
667 | | // A previous find() failed to match. Don't try again. |
668 | | // (without this test, a pattern with a zero-length match |
669 | | // could match again at the end of an input string.) |
670 | 0 | fHitEnd = true; |
671 | 0 | return false; |
672 | 0 | } |
673 | 0 | } |
674 | | |
675 | | |
676 | | // Compute the position in the input string beyond which a match can not begin, because |
677 | | // the minimum length match would extend past the end of the input. |
678 | | // Note: some patterns that cannot match anything will have fMinMatchLength==Max Int. |
679 | | // Be aware of possible overflows if making changes here. |
680 | 0 | int64_t testStartLimit; |
681 | 0 | if (UTEXT_USES_U16(fInputText)) { |
682 | 0 | testStartLimit = fActiveLimit - fPattern->fMinMatchLen; |
683 | 0 | if (startPos > testStartLimit) { |
684 | 0 | fMatch = false; |
685 | 0 | fHitEnd = true; |
686 | 0 | return false; |
687 | 0 | } |
688 | 0 | } else { |
689 | | // We don't know exactly how long the minimum match length is in native characters. |
690 | | // Treat anything > 0 as 1. |
691 | 0 | testStartLimit = fActiveLimit - (fPattern->fMinMatchLen > 0 ? 1 : 0); |
692 | 0 | } |
693 | | |
694 | 0 | UChar32 c; |
695 | 0 | U_ASSERT(startPos >= 0); |
696 | |
|
697 | 0 | switch (fPattern->fStartType) { |
698 | 0 | case START_NO_INFO: |
699 | | // No optimization was found. |
700 | | // Try a match at each input position. |
701 | 0 | for (;;) { |
702 | 0 | MatchAt(startPos, false, status); |
703 | 0 | if (U_FAILURE(status)) { |
704 | 0 | return false; |
705 | 0 | } |
706 | 0 | if (fMatch) { |
707 | 0 | return true; |
708 | 0 | } |
709 | 0 | if (startPos >= testStartLimit) { |
710 | 0 | fHitEnd = true; |
711 | 0 | return false; |
712 | 0 | } |
713 | 0 | UTEXT_SETNATIVEINDEX(fInputText, startPos); |
714 | 0 | (void)UTEXT_NEXT32(fInputText); |
715 | 0 | startPos = UTEXT_GETNATIVEINDEX(fInputText); |
716 | | // Note that it's perfectly OK for a pattern to have a zero-length |
717 | | // match at the end of a string, so we must make sure that the loop |
718 | | // runs with startPos == testStartLimit the last time through. |
719 | 0 | if (findProgressInterrupt(startPos, status)) |
720 | 0 | return false; |
721 | 0 | } |
722 | 0 | UPRV_UNREACHABLE_EXIT; |
723 | | |
724 | 0 | case START_START: |
725 | | // Matches are only possible at the start of the input string |
726 | | // (pattern begins with ^ or \A) |
727 | 0 | if (startPos > fActiveStart) { |
728 | 0 | fMatch = false; |
729 | 0 | return false; |
730 | 0 | } |
731 | 0 | MatchAt(startPos, false, status); |
732 | 0 | if (U_FAILURE(status)) { |
733 | 0 | return false; |
734 | 0 | } |
735 | 0 | return fMatch; |
736 | | |
737 | | |
738 | 0 | case START_SET: |
739 | 0 | { |
740 | | // Match may start on any char from a pre-computed set. |
741 | 0 | U_ASSERT(fPattern->fMinMatchLen > 0); |
742 | 0 | UTEXT_SETNATIVEINDEX(fInputText, startPos); |
743 | 0 | for (;;) { |
744 | 0 | int64_t pos = startPos; |
745 | 0 | c = UTEXT_NEXT32(fInputText); |
746 | 0 | startPos = UTEXT_GETNATIVEINDEX(fInputText); |
747 | | // c will be -1 (U_SENTINEL) at end of text, in which case we |
748 | | // skip this next block (so we don't have a negative array index) |
749 | | // and handle end of text in the following block. |
750 | 0 | if (c >= 0 && ((c<256 && fPattern->fInitialChars8->contains(c)) || |
751 | 0 | (c>=256 && fPattern->fInitialChars->contains(c)))) { |
752 | 0 | MatchAt(pos, false, status); |
753 | 0 | if (U_FAILURE(status)) { |
754 | 0 | return false; |
755 | 0 | } |
756 | 0 | if (fMatch) { |
757 | 0 | return true; |
758 | 0 | } |
759 | 0 | UTEXT_SETNATIVEINDEX(fInputText, pos); |
760 | 0 | } |
761 | 0 | if (startPos > testStartLimit) { |
762 | 0 | fMatch = false; |
763 | 0 | fHitEnd = true; |
764 | 0 | return false; |
765 | 0 | } |
766 | 0 | if (findProgressInterrupt(startPos, status)) |
767 | 0 | return false; |
768 | 0 | } |
769 | 0 | } |
770 | 0 | UPRV_UNREACHABLE_EXIT; |
771 | | |
772 | 0 | case START_STRING: |
773 | 0 | case START_CHAR: |
774 | 0 | { |
775 | | // Match starts on exactly one char. |
776 | 0 | U_ASSERT(fPattern->fMinMatchLen > 0); |
777 | 0 | UChar32 theChar = fPattern->fInitialChar; |
778 | 0 | UTEXT_SETNATIVEINDEX(fInputText, startPos); |
779 | 0 | for (;;) { |
780 | 0 | int64_t pos = startPos; |
781 | 0 | c = UTEXT_NEXT32(fInputText); |
782 | 0 | startPos = UTEXT_GETNATIVEINDEX(fInputText); |
783 | 0 | if (c == theChar) { |
784 | 0 | MatchAt(pos, false, status); |
785 | 0 | if (U_FAILURE(status)) { |
786 | 0 | return false; |
787 | 0 | } |
788 | 0 | if (fMatch) { |
789 | 0 | return true; |
790 | 0 | } |
791 | 0 | UTEXT_SETNATIVEINDEX(fInputText, startPos); |
792 | 0 | } |
793 | 0 | if (startPos > testStartLimit) { |
794 | 0 | fMatch = false; |
795 | 0 | fHitEnd = true; |
796 | 0 | return false; |
797 | 0 | } |
798 | 0 | if (findProgressInterrupt(startPos, status)) |
799 | 0 | return false; |
800 | 0 | } |
801 | 0 | } |
802 | 0 | UPRV_UNREACHABLE_EXIT; |
803 | | |
804 | 0 | case START_LINE: |
805 | 0 | { |
806 | 0 | UChar32 ch; |
807 | 0 | if (startPos == fAnchorStart) { |
808 | 0 | MatchAt(startPos, false, status); |
809 | 0 | if (U_FAILURE(status)) { |
810 | 0 | return false; |
811 | 0 | } |
812 | 0 | if (fMatch) { |
813 | 0 | return true; |
814 | 0 | } |
815 | 0 | UTEXT_SETNATIVEINDEX(fInputText, startPos); |
816 | 0 | ch = UTEXT_NEXT32(fInputText); |
817 | 0 | startPos = UTEXT_GETNATIVEINDEX(fInputText); |
818 | 0 | } else { |
819 | 0 | UTEXT_SETNATIVEINDEX(fInputText, startPos); |
820 | 0 | ch = UTEXT_PREVIOUS32(fInputText); |
821 | 0 | UTEXT_SETNATIVEINDEX(fInputText, startPos); |
822 | 0 | } |
823 | | |
824 | 0 | if (fPattern->fFlags & UREGEX_UNIX_LINES) { |
825 | 0 | for (;;) { |
826 | 0 | if (ch == 0x0a) { |
827 | 0 | MatchAt(startPos, false, status); |
828 | 0 | if (U_FAILURE(status)) { |
829 | 0 | return false; |
830 | 0 | } |
831 | 0 | if (fMatch) { |
832 | 0 | return true; |
833 | 0 | } |
834 | 0 | UTEXT_SETNATIVEINDEX(fInputText, startPos); |
835 | 0 | } |
836 | 0 | if (startPos >= testStartLimit) { |
837 | 0 | fMatch = false; |
838 | 0 | fHitEnd = true; |
839 | 0 | return false; |
840 | 0 | } |
841 | 0 | ch = UTEXT_NEXT32(fInputText); |
842 | 0 | startPos = UTEXT_GETNATIVEINDEX(fInputText); |
843 | | // Note that it's perfectly OK for a pattern to have a zero-length |
844 | | // match at the end of a string, so we must make sure that the loop |
845 | | // runs with startPos == testStartLimit the last time through. |
846 | 0 | if (findProgressInterrupt(startPos, status)) |
847 | 0 | return false; |
848 | 0 | } |
849 | 0 | } else { |
850 | 0 | for (;;) { |
851 | 0 | if (isLineTerminator(ch)) { |
852 | 0 | if (ch == 0x0d && startPos < fActiveLimit && UTEXT_CURRENT32(fInputText) == 0x0a) { |
853 | 0 | (void)UTEXT_NEXT32(fInputText); |
854 | 0 | startPos = UTEXT_GETNATIVEINDEX(fInputText); |
855 | 0 | } |
856 | 0 | MatchAt(startPos, false, status); |
857 | 0 | if (U_FAILURE(status)) { |
858 | 0 | return false; |
859 | 0 | } |
860 | 0 | if (fMatch) { |
861 | 0 | return true; |
862 | 0 | } |
863 | 0 | UTEXT_SETNATIVEINDEX(fInputText, startPos); |
864 | 0 | } |
865 | 0 | if (startPos >= testStartLimit) { |
866 | 0 | fMatch = false; |
867 | 0 | fHitEnd = true; |
868 | 0 | return false; |
869 | 0 | } |
870 | 0 | ch = UTEXT_NEXT32(fInputText); |
871 | 0 | startPos = UTEXT_GETNATIVEINDEX(fInputText); |
872 | | // Note that it's perfectly OK for a pattern to have a zero-length |
873 | | // match at the end of a string, so we must make sure that the loop |
874 | | // runs with startPos == testStartLimit the last time through. |
875 | 0 | if (findProgressInterrupt(startPos, status)) |
876 | 0 | return false; |
877 | 0 | } |
878 | 0 | } |
879 | 0 | } |
880 | | |
881 | 0 | default: |
882 | 0 | UPRV_UNREACHABLE_ASSERT; |
883 | | // Unknown value in fPattern->fStartType, should be from StartOfMatch enum. But |
884 | | // we have reports of this in production code, don't use UPRV_UNREACHABLE_EXIT. |
885 | | // See ICU-21669. |
886 | 0 | status = U_INTERNAL_PROGRAM_ERROR; |
887 | 0 | return false; |
888 | 0 | } |
889 | | |
890 | 0 | UPRV_UNREACHABLE_EXIT; |
891 | 0 | } |
892 | | |
893 | | |
894 | | |
895 | 0 | UBool RegexMatcher::find(int64_t start, UErrorCode &status) { |
896 | 0 | if (U_FAILURE(status)) { |
897 | 0 | return false; |
898 | 0 | } |
899 | 0 | if (U_FAILURE(fDeferredStatus)) { |
900 | 0 | status = fDeferredStatus; |
901 | 0 | return false; |
902 | 0 | } |
903 | 0 | this->reset(); // Note: Reset() is specified by Java Matcher documentation. |
904 | | // This will reset the region to be the full input length. |
905 | 0 | if (start < 0) { |
906 | 0 | status = U_INDEX_OUTOFBOUNDS_ERROR; |
907 | 0 | return false; |
908 | 0 | } |
909 | | |
910 | 0 | int64_t nativeStart = start; |
911 | 0 | if (nativeStart < fActiveStart || nativeStart > fActiveLimit) { |
912 | 0 | status = U_INDEX_OUTOFBOUNDS_ERROR; |
913 | 0 | return false; |
914 | 0 | } |
915 | 0 | fMatchEnd = nativeStart; |
916 | 0 | return find(status); |
917 | 0 | } |
918 | | |
919 | | |
920 | | //-------------------------------------------------------------------------------- |
921 | | // |
922 | | // findUsingChunk() -- like find(), but with the advance knowledge that the |
923 | | // entire string is available in the UText's chunk buffer. |
924 | | // |
925 | | //-------------------------------------------------------------------------------- |
926 | 0 | UBool RegexMatcher::findUsingChunk(UErrorCode &status) { |
927 | | // Start at the position of the last match end. (Will be zero if the |
928 | | // matcher has been reset. |
929 | | // |
930 | |
|
931 | 0 | int32_t startPos = (int32_t)fMatchEnd; |
932 | 0 | if (startPos==0) { |
933 | 0 | startPos = (int32_t)fActiveStart; |
934 | 0 | } |
935 | |
|
936 | 0 | const UChar *inputBuf = fInputText->chunkContents; |
937 | |
|
938 | 0 | if (fMatch) { |
939 | | // Save the position of any previous successful match. |
940 | 0 | fLastMatchEnd = fMatchEnd; |
941 | |
|
942 | 0 | if (fMatchStart == fMatchEnd) { |
943 | | // Previous match had zero length. Move start position up one position |
944 | | // to avoid sending find() into a loop on zero-length matches. |
945 | 0 | if (startPos >= fActiveLimit) { |
946 | 0 | fMatch = false; |
947 | 0 | fHitEnd = true; |
948 | 0 | return false; |
949 | 0 | } |
950 | 0 | U16_FWD_1(inputBuf, startPos, fInputLength); |
951 | 0 | } |
952 | 0 | } else { |
953 | 0 | if (fLastMatchEnd >= 0) { |
954 | | // A previous find() failed to match. Don't try again. |
955 | | // (without this test, a pattern with a zero-length match |
956 | | // could match again at the end of an input string.) |
957 | 0 | fHitEnd = true; |
958 | 0 | return false; |
959 | 0 | } |
960 | 0 | } |
961 | | |
962 | | |
963 | | // Compute the position in the input string beyond which a match can not begin, because |
964 | | // the minimum length match would extend past the end of the input. |
965 | | // Note: some patterns that cannot match anything will have fMinMatchLength==Max Int. |
966 | | // Be aware of possible overflows if making changes here. |
967 | | // Note: a match can begin at inputBuf + testLen; it is an inclusive limit. |
968 | 0 | int32_t testLen = (int32_t)(fActiveLimit - fPattern->fMinMatchLen); |
969 | 0 | if (startPos > testLen) { |
970 | 0 | fMatch = false; |
971 | 0 | fHitEnd = true; |
972 | 0 | return false; |
973 | 0 | } |
974 | | |
975 | 0 | UChar32 c; |
976 | 0 | U_ASSERT(startPos >= 0); |
977 | |
|
978 | 0 | switch (fPattern->fStartType) { |
979 | 0 | case START_NO_INFO: |
980 | | // No optimization was found. |
981 | | // Try a match at each input position. |
982 | 0 | for (;;) { |
983 | 0 | MatchChunkAt(startPos, false, status); |
984 | 0 | if (U_FAILURE(status)) { |
985 | 0 | return false; |
986 | 0 | } |
987 | 0 | if (fMatch) { |
988 | 0 | return true; |
989 | 0 | } |
990 | 0 | if (startPos >= testLen) { |
991 | 0 | fHitEnd = true; |
992 | 0 | return false; |
993 | 0 | } |
994 | 0 | U16_FWD_1(inputBuf, startPos, fActiveLimit); |
995 | | // Note that it's perfectly OK for a pattern to have a zero-length |
996 | | // match at the end of a string, so we must make sure that the loop |
997 | | // runs with startPos == testLen the last time through. |
998 | 0 | if (findProgressInterrupt(startPos, status)) |
999 | 0 | return false; |
1000 | 0 | } |
1001 | 0 | UPRV_UNREACHABLE_EXIT; |
1002 | | |
1003 | 0 | case START_START: |
1004 | | // Matches are only possible at the start of the input string |
1005 | | // (pattern begins with ^ or \A) |
1006 | 0 | if (startPos > fActiveStart) { |
1007 | 0 | fMatch = false; |
1008 | 0 | return false; |
1009 | 0 | } |
1010 | 0 | MatchChunkAt(startPos, false, status); |
1011 | 0 | if (U_FAILURE(status)) { |
1012 | 0 | return false; |
1013 | 0 | } |
1014 | 0 | return fMatch; |
1015 | | |
1016 | | |
1017 | 0 | case START_SET: |
1018 | 0 | { |
1019 | | // Match may start on any char from a pre-computed set. |
1020 | 0 | U_ASSERT(fPattern->fMinMatchLen > 0); |
1021 | 0 | for (;;) { |
1022 | 0 | int32_t pos = startPos; |
1023 | 0 | U16_NEXT(inputBuf, startPos, fActiveLimit, c); // like c = inputBuf[startPos++]; |
1024 | 0 | if ((c<256 && fPattern->fInitialChars8->contains(c)) || |
1025 | 0 | (c>=256 && fPattern->fInitialChars->contains(c))) { |
1026 | 0 | MatchChunkAt(pos, false, status); |
1027 | 0 | if (U_FAILURE(status)) { |
1028 | 0 | return false; |
1029 | 0 | } |
1030 | 0 | if (fMatch) { |
1031 | 0 | return true; |
1032 | 0 | } |
1033 | 0 | } |
1034 | 0 | if (startPos > testLen) { |
1035 | 0 | fMatch = false; |
1036 | 0 | fHitEnd = true; |
1037 | 0 | return false; |
1038 | 0 | } |
1039 | 0 | if (findProgressInterrupt(startPos, status)) |
1040 | 0 | return false; |
1041 | 0 | } |
1042 | 0 | } |
1043 | 0 | UPRV_UNREACHABLE_EXIT; |
1044 | | |
1045 | 0 | case START_STRING: |
1046 | 0 | case START_CHAR: |
1047 | 0 | { |
1048 | | // Match starts on exactly one char. |
1049 | 0 | U_ASSERT(fPattern->fMinMatchLen > 0); |
1050 | 0 | UChar32 theChar = fPattern->fInitialChar; |
1051 | 0 | for (;;) { |
1052 | 0 | int32_t pos = startPos; |
1053 | 0 | U16_NEXT(inputBuf, startPos, fActiveLimit, c); // like c = inputBuf[startPos++]; |
1054 | 0 | if (c == theChar) { |
1055 | 0 | MatchChunkAt(pos, false, status); |
1056 | 0 | if (U_FAILURE(status)) { |
1057 | 0 | return false; |
1058 | 0 | } |
1059 | 0 | if (fMatch) { |
1060 | 0 | return true; |
1061 | 0 | } |
1062 | 0 | } |
1063 | 0 | if (startPos > testLen) { |
1064 | 0 | fMatch = false; |
1065 | 0 | fHitEnd = true; |
1066 | 0 | return false; |
1067 | 0 | } |
1068 | 0 | if (findProgressInterrupt(startPos, status)) |
1069 | 0 | return false; |
1070 | 0 | } |
1071 | 0 | } |
1072 | 0 | UPRV_UNREACHABLE_EXIT; |
1073 | | |
1074 | 0 | case START_LINE: |
1075 | 0 | { |
1076 | 0 | UChar32 ch; |
1077 | 0 | if (startPos == fAnchorStart) { |
1078 | 0 | MatchChunkAt(startPos, false, status); |
1079 | 0 | if (U_FAILURE(status)) { |
1080 | 0 | return false; |
1081 | 0 | } |
1082 | 0 | if (fMatch) { |
1083 | 0 | return true; |
1084 | 0 | } |
1085 | 0 | U16_FWD_1(inputBuf, startPos, fActiveLimit); |
1086 | 0 | } |
1087 | | |
1088 | 0 | if (fPattern->fFlags & UREGEX_UNIX_LINES) { |
1089 | 0 | for (;;) { |
1090 | 0 | ch = inputBuf[startPos-1]; |
1091 | 0 | if (ch == 0x0a) { |
1092 | 0 | MatchChunkAt(startPos, false, status); |
1093 | 0 | if (U_FAILURE(status)) { |
1094 | 0 | return false; |
1095 | 0 | } |
1096 | 0 | if (fMatch) { |
1097 | 0 | return true; |
1098 | 0 | } |
1099 | 0 | } |
1100 | 0 | if (startPos >= testLen) { |
1101 | 0 | fMatch = false; |
1102 | 0 | fHitEnd = true; |
1103 | 0 | return false; |
1104 | 0 | } |
1105 | 0 | U16_FWD_1(inputBuf, startPos, fActiveLimit); |
1106 | | // Note that it's perfectly OK for a pattern to have a zero-length |
1107 | | // match at the end of a string, so we must make sure that the loop |
1108 | | // runs with startPos == testLen the last time through. |
1109 | 0 | if (findProgressInterrupt(startPos, status)) |
1110 | 0 | return false; |
1111 | 0 | } |
1112 | 0 | } else { |
1113 | 0 | for (;;) { |
1114 | 0 | ch = inputBuf[startPos-1]; |
1115 | 0 | if (isLineTerminator(ch)) { |
1116 | 0 | if (ch == 0x0d && startPos < fActiveLimit && inputBuf[startPos] == 0x0a) { |
1117 | 0 | startPos++; |
1118 | 0 | } |
1119 | 0 | MatchChunkAt(startPos, false, status); |
1120 | 0 | if (U_FAILURE(status)) { |
1121 | 0 | return false; |
1122 | 0 | } |
1123 | 0 | if (fMatch) { |
1124 | 0 | return true; |
1125 | 0 | } |
1126 | 0 | } |
1127 | 0 | if (startPos >= testLen) { |
1128 | 0 | fMatch = false; |
1129 | 0 | fHitEnd = true; |
1130 | 0 | return false; |
1131 | 0 | } |
1132 | 0 | U16_FWD_1(inputBuf, startPos, fActiveLimit); |
1133 | | // Note that it's perfectly OK for a pattern to have a zero-length |
1134 | | // match at the end of a string, so we must make sure that the loop |
1135 | | // runs with startPos == testLen the last time through. |
1136 | 0 | if (findProgressInterrupt(startPos, status)) |
1137 | 0 | return false; |
1138 | 0 | } |
1139 | 0 | } |
1140 | 0 | } |
1141 | | |
1142 | 0 | default: |
1143 | 0 | UPRV_UNREACHABLE_ASSERT; |
1144 | | // Unknown value in fPattern->fStartType, should be from StartOfMatch enum. But |
1145 | | // we have reports of this in production code, don't use UPRV_UNREACHABLE_EXIT. |
1146 | | // See ICU-21669. |
1147 | 0 | status = U_INTERNAL_PROGRAM_ERROR; |
1148 | 0 | return false; |
1149 | 0 | } |
1150 | | |
1151 | 0 | UPRV_UNREACHABLE_EXIT; |
1152 | 0 | } |
1153 | | |
1154 | | |
1155 | | |
1156 | | //-------------------------------------------------------------------------------- |
1157 | | // |
1158 | | // group() |
1159 | | // |
1160 | | //-------------------------------------------------------------------------------- |
1161 | 0 | UnicodeString RegexMatcher::group(UErrorCode &status) const { |
1162 | 0 | return group(0, status); |
1163 | 0 | } |
1164 | | |
1165 | | // Return immutable shallow clone |
1166 | 0 | UText *RegexMatcher::group(UText *dest, int64_t &group_len, UErrorCode &status) const { |
1167 | 0 | return group(0, dest, group_len, status); |
1168 | 0 | } |
1169 | | |
1170 | | // Return immutable shallow clone |
1171 | 0 | UText *RegexMatcher::group(int32_t groupNum, UText *dest, int64_t &group_len, UErrorCode &status) const { |
1172 | 0 | group_len = 0; |
1173 | 0 | if (U_FAILURE(status)) { |
1174 | 0 | return dest; |
1175 | 0 | } |
1176 | 0 | if (U_FAILURE(fDeferredStatus)) { |
1177 | 0 | status = fDeferredStatus; |
1178 | 0 | } else if (fMatch == false) { |
1179 | 0 | status = U_REGEX_INVALID_STATE; |
1180 | 0 | } else if (groupNum < 0 || groupNum > fPattern->fGroupMap->size()) { |
1181 | 0 | status = U_INDEX_OUTOFBOUNDS_ERROR; |
1182 | 0 | } |
1183 | |
|
1184 | 0 | if (U_FAILURE(status)) { |
1185 | 0 | return dest; |
1186 | 0 | } |
1187 | | |
1188 | 0 | int64_t s, e; |
1189 | 0 | if (groupNum == 0) { |
1190 | 0 | s = fMatchStart; |
1191 | 0 | e = fMatchEnd; |
1192 | 0 | } else { |
1193 | 0 | int32_t groupOffset = fPattern->fGroupMap->elementAti(groupNum-1); |
1194 | 0 | U_ASSERT(groupOffset < fPattern->fFrameSize); |
1195 | 0 | U_ASSERT(groupOffset >= 0); |
1196 | 0 | s = fFrame->fExtra[groupOffset]; |
1197 | 0 | e = fFrame->fExtra[groupOffset+1]; |
1198 | 0 | } |
1199 | |
|
1200 | 0 | if (s < 0) { |
1201 | | // A capture group wasn't part of the match |
1202 | 0 | return utext_clone(dest, fInputText, false, true, &status); |
1203 | 0 | } |
1204 | 0 | U_ASSERT(s <= e); |
1205 | 0 | group_len = e - s; |
1206 | |
|
1207 | 0 | dest = utext_clone(dest, fInputText, false, true, &status); |
1208 | 0 | if (dest) |
1209 | 0 | UTEXT_SETNATIVEINDEX(dest, s); |
1210 | 0 | return dest; |
1211 | 0 | } |
1212 | | |
1213 | 0 | UnicodeString RegexMatcher::group(int32_t groupNum, UErrorCode &status) const { |
1214 | 0 | UnicodeString result; |
1215 | 0 | int64_t groupStart = start64(groupNum, status); |
1216 | 0 | int64_t groupEnd = end64(groupNum, status); |
1217 | 0 | if (U_FAILURE(status) || groupStart == -1 || groupStart == groupEnd) { |
1218 | 0 | return result; |
1219 | 0 | } |
1220 | | |
1221 | | // Get the group length using a utext_extract preflight. |
1222 | | // UText is actually pretty efficient at this when underlying encoding is UTF-16. |
1223 | 0 | int32_t length = utext_extract(fInputText, groupStart, groupEnd, NULL, 0, &status); |
1224 | 0 | if (status != U_BUFFER_OVERFLOW_ERROR) { |
1225 | 0 | return result; |
1226 | 0 | } |
1227 | | |
1228 | 0 | status = U_ZERO_ERROR; |
1229 | 0 | UChar *buf = result.getBuffer(length); |
1230 | 0 | if (buf == NULL) { |
1231 | 0 | status = U_MEMORY_ALLOCATION_ERROR; |
1232 | 0 | } else { |
1233 | 0 | int32_t extractLength = utext_extract(fInputText, groupStart, groupEnd, buf, length, &status); |
1234 | 0 | result.releaseBuffer(extractLength); |
1235 | 0 | U_ASSERT(length == extractLength); |
1236 | 0 | } |
1237 | 0 | return result; |
1238 | 0 | } |
1239 | | |
1240 | | |
1241 | | //-------------------------------------------------------------------------------- |
1242 | | // |
1243 | | // appendGroup() -- currently internal only, appends a group to a UText rather |
1244 | | // than replacing its contents |
1245 | | // |
1246 | | //-------------------------------------------------------------------------------- |
1247 | | |
1248 | 0 | int64_t RegexMatcher::appendGroup(int32_t groupNum, UText *dest, UErrorCode &status) const { |
1249 | 0 | if (U_FAILURE(status)) { |
1250 | 0 | return 0; |
1251 | 0 | } |
1252 | 0 | if (U_FAILURE(fDeferredStatus)) { |
1253 | 0 | status = fDeferredStatus; |
1254 | 0 | return 0; |
1255 | 0 | } |
1256 | 0 | int64_t destLen = utext_nativeLength(dest); |
1257 | |
|
1258 | 0 | if (fMatch == false) { |
1259 | 0 | status = U_REGEX_INVALID_STATE; |
1260 | 0 | return utext_replace(dest, destLen, destLen, NULL, 0, &status); |
1261 | 0 | } |
1262 | 0 | if (groupNum < 0 || groupNum > fPattern->fGroupMap->size()) { |
1263 | 0 | status = U_INDEX_OUTOFBOUNDS_ERROR; |
1264 | 0 | return utext_replace(dest, destLen, destLen, NULL, 0, &status); |
1265 | 0 | } |
1266 | | |
1267 | 0 | int64_t s, e; |
1268 | 0 | if (groupNum == 0) { |
1269 | 0 | s = fMatchStart; |
1270 | 0 | e = fMatchEnd; |
1271 | 0 | } else { |
1272 | 0 | int32_t groupOffset = fPattern->fGroupMap->elementAti(groupNum-1); |
1273 | 0 | U_ASSERT(groupOffset < fPattern->fFrameSize); |
1274 | 0 | U_ASSERT(groupOffset >= 0); |
1275 | 0 | s = fFrame->fExtra[groupOffset]; |
1276 | 0 | e = fFrame->fExtra[groupOffset+1]; |
1277 | 0 | } |
1278 | |
|
1279 | 0 | if (s < 0) { |
1280 | | // A capture group wasn't part of the match |
1281 | 0 | return utext_replace(dest, destLen, destLen, NULL, 0, &status); |
1282 | 0 | } |
1283 | 0 | U_ASSERT(s <= e); |
1284 | |
|
1285 | 0 | int64_t deltaLen; |
1286 | 0 | if (UTEXT_FULL_TEXT_IN_CHUNK(fInputText, fInputLength)) { |
1287 | 0 | U_ASSERT(e <= fInputLength); |
1288 | 0 | deltaLen = utext_replace(dest, destLen, destLen, fInputText->chunkContents+s, (int32_t)(e-s), &status); |
1289 | 0 | } else { |
1290 | 0 | int32_t len16; |
1291 | 0 | if (UTEXT_USES_U16(fInputText)) { |
1292 | 0 | len16 = (int32_t)(e-s); |
1293 | 0 | } else { |
1294 | 0 | UErrorCode lengthStatus = U_ZERO_ERROR; |
1295 | 0 | len16 = utext_extract(fInputText, s, e, NULL, 0, &lengthStatus); |
1296 | 0 | } |
1297 | 0 | UChar *groupChars = (UChar *)uprv_malloc(sizeof(UChar)*(len16+1)); |
1298 | 0 | if (groupChars == NULL) { |
1299 | 0 | status = U_MEMORY_ALLOCATION_ERROR; |
1300 | 0 | return 0; |
1301 | 0 | } |
1302 | 0 | utext_extract(fInputText, s, e, groupChars, len16+1, &status); |
1303 | |
|
1304 | 0 | deltaLen = utext_replace(dest, destLen, destLen, groupChars, len16, &status); |
1305 | 0 | uprv_free(groupChars); |
1306 | 0 | } |
1307 | 0 | return deltaLen; |
1308 | 0 | } |
1309 | | |
1310 | | |
1311 | | |
1312 | | //-------------------------------------------------------------------------------- |
1313 | | // |
1314 | | // groupCount() |
1315 | | // |
1316 | | //-------------------------------------------------------------------------------- |
1317 | 0 | int32_t RegexMatcher::groupCount() const { |
1318 | 0 | return fPattern->fGroupMap->size(); |
1319 | 0 | } |
1320 | | |
1321 | | //-------------------------------------------------------------------------------- |
1322 | | // |
1323 | | // hasAnchoringBounds() |
1324 | | // |
1325 | | //-------------------------------------------------------------------------------- |
1326 | 0 | UBool RegexMatcher::hasAnchoringBounds() const { |
1327 | 0 | return fAnchoringBounds; |
1328 | 0 | } |
1329 | | |
1330 | | |
1331 | | //-------------------------------------------------------------------------------- |
1332 | | // |
1333 | | // hasTransparentBounds() |
1334 | | // |
1335 | | //-------------------------------------------------------------------------------- |
1336 | 0 | UBool RegexMatcher::hasTransparentBounds() const { |
1337 | 0 | return fTransparentBounds; |
1338 | 0 | } |
1339 | | |
1340 | | |
1341 | | |
1342 | | //-------------------------------------------------------------------------------- |
1343 | | // |
1344 | | // hitEnd() |
1345 | | // |
1346 | | //-------------------------------------------------------------------------------- |
1347 | 0 | UBool RegexMatcher::hitEnd() const { |
1348 | 0 | return fHitEnd; |
1349 | 0 | } |
1350 | | |
1351 | | |
1352 | | //-------------------------------------------------------------------------------- |
1353 | | // |
1354 | | // input() |
1355 | | // |
1356 | | //-------------------------------------------------------------------------------- |
1357 | 0 | const UnicodeString &RegexMatcher::input() const { |
1358 | 0 | if (!fInput) { |
1359 | 0 | UErrorCode status = U_ZERO_ERROR; |
1360 | 0 | int32_t len16; |
1361 | 0 | if (UTEXT_USES_U16(fInputText)) { |
1362 | 0 | len16 = (int32_t)fInputLength; |
1363 | 0 | } else { |
1364 | 0 | len16 = utext_extract(fInputText, 0, fInputLength, NULL, 0, &status); |
1365 | 0 | status = U_ZERO_ERROR; // overflow, length status |
1366 | 0 | } |
1367 | 0 | UnicodeString *result = new UnicodeString(len16, 0, 0); |
1368 | |
|
1369 | 0 | UChar *inputChars = result->getBuffer(len16); |
1370 | 0 | utext_extract(fInputText, 0, fInputLength, inputChars, len16, &status); // unterminated warning |
1371 | 0 | result->releaseBuffer(len16); |
1372 | |
|
1373 | 0 | (*(const UnicodeString **)&fInput) = result; // pointer assignment, rather than operator= |
1374 | 0 | } |
1375 | |
|
1376 | 0 | return *fInput; |
1377 | 0 | } |
1378 | | |
1379 | | //-------------------------------------------------------------------------------- |
1380 | | // |
1381 | | // inputText() |
1382 | | // |
1383 | | //-------------------------------------------------------------------------------- |
1384 | 0 | UText *RegexMatcher::inputText() const { |
1385 | 0 | return fInputText; |
1386 | 0 | } |
1387 | | |
1388 | | |
1389 | | //-------------------------------------------------------------------------------- |
1390 | | // |
1391 | | // getInput() -- like inputText(), but makes a clone or copies into another UText |
1392 | | // |
1393 | | //-------------------------------------------------------------------------------- |
1394 | 0 | UText *RegexMatcher::getInput (UText *dest, UErrorCode &status) const { |
1395 | 0 | if (U_FAILURE(status)) { |
1396 | 0 | return dest; |
1397 | 0 | } |
1398 | 0 | if (U_FAILURE(fDeferredStatus)) { |
1399 | 0 | status = fDeferredStatus; |
1400 | 0 | return dest; |
1401 | 0 | } |
1402 | | |
1403 | 0 | if (dest) { |
1404 | 0 | if (UTEXT_FULL_TEXT_IN_CHUNK(fInputText, fInputLength)) { |
1405 | 0 | utext_replace(dest, 0, utext_nativeLength(dest), fInputText->chunkContents, (int32_t)fInputLength, &status); |
1406 | 0 | } else { |
1407 | 0 | int32_t input16Len; |
1408 | 0 | if (UTEXT_USES_U16(fInputText)) { |
1409 | 0 | input16Len = (int32_t)fInputLength; |
1410 | 0 | } else { |
1411 | 0 | UErrorCode lengthStatus = U_ZERO_ERROR; |
1412 | 0 | input16Len = utext_extract(fInputText, 0, fInputLength, NULL, 0, &lengthStatus); // buffer overflow error |
1413 | 0 | } |
1414 | 0 | UChar *inputChars = (UChar *)uprv_malloc(sizeof(UChar)*(input16Len)); |
1415 | 0 | if (inputChars == NULL) { |
1416 | 0 | return dest; |
1417 | 0 | } |
1418 | | |
1419 | 0 | status = U_ZERO_ERROR; |
1420 | 0 | utext_extract(fInputText, 0, fInputLength, inputChars, input16Len, &status); // not terminated warning |
1421 | 0 | status = U_ZERO_ERROR; |
1422 | 0 | utext_replace(dest, 0, utext_nativeLength(dest), inputChars, input16Len, &status); |
1423 | |
|
1424 | 0 | uprv_free(inputChars); |
1425 | 0 | } |
1426 | 0 | return dest; |
1427 | 0 | } else { |
1428 | 0 | return utext_clone(NULL, fInputText, false, true, &status); |
1429 | 0 | } |
1430 | 0 | } |
1431 | | |
1432 | | |
1433 | | static UBool compat_SyncMutableUTextContents(UText *ut); |
1434 | 0 | static UBool compat_SyncMutableUTextContents(UText *ut) { |
1435 | 0 | UBool retVal = false; |
1436 | | |
1437 | | // In the following test, we're really only interested in whether the UText should switch |
1438 | | // between heap and stack allocation. If length hasn't changed, we won't, so the chunkContents |
1439 | | // will still point to the correct data. |
1440 | 0 | if (utext_nativeLength(ut) != ut->nativeIndexingLimit) { |
1441 | 0 | UnicodeString *us=(UnicodeString *)ut->context; |
1442 | | |
1443 | | // Update to the latest length. |
1444 | | // For example, (utext_nativeLength(ut) != ut->nativeIndexingLimit). |
1445 | 0 | int32_t newLength = us->length(); |
1446 | | |
1447 | | // Update the chunk description. |
1448 | | // The buffer may have switched between stack- and heap-based. |
1449 | 0 | ut->chunkContents = us->getBuffer(); |
1450 | 0 | ut->chunkLength = newLength; |
1451 | 0 | ut->chunkNativeLimit = newLength; |
1452 | 0 | ut->nativeIndexingLimit = newLength; |
1453 | 0 | retVal = true; |
1454 | 0 | } |
1455 | |
|
1456 | 0 | return retVal; |
1457 | 0 | } |
1458 | | |
1459 | | //-------------------------------------------------------------------------------- |
1460 | | // |
1461 | | // lookingAt() |
1462 | | // |
1463 | | //-------------------------------------------------------------------------------- |
1464 | 0 | UBool RegexMatcher::lookingAt(UErrorCode &status) { |
1465 | 0 | if (U_FAILURE(status)) { |
1466 | 0 | return false; |
1467 | 0 | } |
1468 | 0 | if (U_FAILURE(fDeferredStatus)) { |
1469 | 0 | status = fDeferredStatus; |
1470 | 0 | return false; |
1471 | 0 | } |
1472 | | |
1473 | 0 | if (fInputUniStrMaybeMutable) { |
1474 | 0 | if (compat_SyncMutableUTextContents(fInputText)) { |
1475 | 0 | fInputLength = utext_nativeLength(fInputText); |
1476 | 0 | reset(); |
1477 | 0 | } |
1478 | 0 | } |
1479 | 0 | else { |
1480 | 0 | resetPreserveRegion(); |
1481 | 0 | } |
1482 | 0 | if (UTEXT_FULL_TEXT_IN_CHUNK(fInputText, fInputLength)) { |
1483 | 0 | MatchChunkAt((int32_t)fActiveStart, false, status); |
1484 | 0 | } else { |
1485 | 0 | MatchAt(fActiveStart, false, status); |
1486 | 0 | } |
1487 | 0 | return fMatch; |
1488 | 0 | } |
1489 | | |
1490 | | |
1491 | 0 | UBool RegexMatcher::lookingAt(int64_t start, UErrorCode &status) { |
1492 | 0 | if (U_FAILURE(status)) { |
1493 | 0 | return false; |
1494 | 0 | } |
1495 | 0 | if (U_FAILURE(fDeferredStatus)) { |
1496 | 0 | status = fDeferredStatus; |
1497 | 0 | return false; |
1498 | 0 | } |
1499 | 0 | reset(); |
1500 | |
|
1501 | 0 | if (start < 0) { |
1502 | 0 | status = U_INDEX_OUTOFBOUNDS_ERROR; |
1503 | 0 | return false; |
1504 | 0 | } |
1505 | | |
1506 | 0 | if (fInputUniStrMaybeMutable) { |
1507 | 0 | if (compat_SyncMutableUTextContents(fInputText)) { |
1508 | 0 | fInputLength = utext_nativeLength(fInputText); |
1509 | 0 | reset(); |
1510 | 0 | } |
1511 | 0 | } |
1512 | |
|
1513 | 0 | int64_t nativeStart; |
1514 | 0 | nativeStart = start; |
1515 | 0 | if (nativeStart < fActiveStart || nativeStart > fActiveLimit) { |
1516 | 0 | status = U_INDEX_OUTOFBOUNDS_ERROR; |
1517 | 0 | return false; |
1518 | 0 | } |
1519 | | |
1520 | 0 | if (UTEXT_FULL_TEXT_IN_CHUNK(fInputText, fInputLength)) { |
1521 | 0 | MatchChunkAt((int32_t)nativeStart, false, status); |
1522 | 0 | } else { |
1523 | 0 | MatchAt(nativeStart, false, status); |
1524 | 0 | } |
1525 | 0 | return fMatch; |
1526 | 0 | } |
1527 | | |
1528 | | |
1529 | | |
1530 | | //-------------------------------------------------------------------------------- |
1531 | | // |
1532 | | // matches() |
1533 | | // |
1534 | | //-------------------------------------------------------------------------------- |
1535 | 0 | UBool RegexMatcher::matches(UErrorCode &status) { |
1536 | 0 | if (U_FAILURE(status)) { |
1537 | 0 | return false; |
1538 | 0 | } |
1539 | 0 | if (U_FAILURE(fDeferredStatus)) { |
1540 | 0 | status = fDeferredStatus; |
1541 | 0 | return false; |
1542 | 0 | } |
1543 | | |
1544 | 0 | if (fInputUniStrMaybeMutable) { |
1545 | 0 | if (compat_SyncMutableUTextContents(fInputText)) { |
1546 | 0 | fInputLength = utext_nativeLength(fInputText); |
1547 | 0 | reset(); |
1548 | 0 | } |
1549 | 0 | } |
1550 | 0 | else { |
1551 | 0 | resetPreserveRegion(); |
1552 | 0 | } |
1553 | |
|
1554 | 0 | if (UTEXT_FULL_TEXT_IN_CHUNK(fInputText, fInputLength)) { |
1555 | 0 | MatchChunkAt((int32_t)fActiveStart, true, status); |
1556 | 0 | } else { |
1557 | 0 | MatchAt(fActiveStart, true, status); |
1558 | 0 | } |
1559 | 0 | return fMatch; |
1560 | 0 | } |
1561 | | |
1562 | | |
1563 | 0 | UBool RegexMatcher::matches(int64_t start, UErrorCode &status) { |
1564 | 0 | if (U_FAILURE(status)) { |
1565 | 0 | return false; |
1566 | 0 | } |
1567 | 0 | if (U_FAILURE(fDeferredStatus)) { |
1568 | 0 | status = fDeferredStatus; |
1569 | 0 | return false; |
1570 | 0 | } |
1571 | 0 | reset(); |
1572 | |
|
1573 | 0 | if (start < 0) { |
1574 | 0 | status = U_INDEX_OUTOFBOUNDS_ERROR; |
1575 | 0 | return false; |
1576 | 0 | } |
1577 | | |
1578 | 0 | if (fInputUniStrMaybeMutable) { |
1579 | 0 | if (compat_SyncMutableUTextContents(fInputText)) { |
1580 | 0 | fInputLength = utext_nativeLength(fInputText); |
1581 | 0 | reset(); |
1582 | 0 | } |
1583 | 0 | } |
1584 | |
|
1585 | 0 | int64_t nativeStart; |
1586 | 0 | nativeStart = start; |
1587 | 0 | if (nativeStart < fActiveStart || nativeStart > fActiveLimit) { |
1588 | 0 | status = U_INDEX_OUTOFBOUNDS_ERROR; |
1589 | 0 | return false; |
1590 | 0 | } |
1591 | | |
1592 | 0 | if (UTEXT_FULL_TEXT_IN_CHUNK(fInputText, fInputLength)) { |
1593 | 0 | MatchChunkAt((int32_t)nativeStart, true, status); |
1594 | 0 | } else { |
1595 | 0 | MatchAt(nativeStart, true, status); |
1596 | 0 | } |
1597 | 0 | return fMatch; |
1598 | 0 | } |
1599 | | |
1600 | | |
1601 | | |
1602 | | //-------------------------------------------------------------------------------- |
1603 | | // |
1604 | | // pattern |
1605 | | // |
1606 | | //-------------------------------------------------------------------------------- |
1607 | 0 | const RegexPattern &RegexMatcher::pattern() const { |
1608 | 0 | return *fPattern; |
1609 | 0 | } |
1610 | | |
1611 | | |
1612 | | |
1613 | | //-------------------------------------------------------------------------------- |
1614 | | // |
1615 | | // region |
1616 | | // |
1617 | | //-------------------------------------------------------------------------------- |
1618 | 0 | RegexMatcher &RegexMatcher::region(int64_t regionStart, int64_t regionLimit, int64_t startIndex, UErrorCode &status) { |
1619 | 0 | if (U_FAILURE(status)) { |
1620 | 0 | return *this; |
1621 | 0 | } |
1622 | | |
1623 | 0 | if (regionStart>regionLimit || regionStart<0 || regionLimit<0) { |
1624 | 0 | status = U_ILLEGAL_ARGUMENT_ERROR; |
1625 | 0 | } |
1626 | |
|
1627 | 0 | int64_t nativeStart = regionStart; |
1628 | 0 | int64_t nativeLimit = regionLimit; |
1629 | 0 | if (nativeStart > fInputLength || nativeLimit > fInputLength) { |
1630 | 0 | status = U_ILLEGAL_ARGUMENT_ERROR; |
1631 | 0 | } |
1632 | |
|
1633 | 0 | if (startIndex == -1) |
1634 | 0 | this->reset(); |
1635 | 0 | else |
1636 | 0 | resetPreserveRegion(); |
1637 | |
|
1638 | 0 | fRegionStart = nativeStart; |
1639 | 0 | fRegionLimit = nativeLimit; |
1640 | 0 | fActiveStart = nativeStart; |
1641 | 0 | fActiveLimit = nativeLimit; |
1642 | |
|
1643 | 0 | if (startIndex != -1) { |
1644 | 0 | if (startIndex < fActiveStart || startIndex > fActiveLimit) { |
1645 | 0 | status = U_INDEX_OUTOFBOUNDS_ERROR; |
1646 | 0 | } |
1647 | 0 | fMatchEnd = startIndex; |
1648 | 0 | } |
1649 | |
|
1650 | 0 | if (!fTransparentBounds) { |
1651 | 0 | fLookStart = nativeStart; |
1652 | 0 | fLookLimit = nativeLimit; |
1653 | 0 | } |
1654 | 0 | if (fAnchoringBounds) { |
1655 | 0 | fAnchorStart = nativeStart; |
1656 | 0 | fAnchorLimit = nativeLimit; |
1657 | 0 | } |
1658 | 0 | return *this; |
1659 | 0 | } |
1660 | | |
1661 | 0 | RegexMatcher &RegexMatcher::region(int64_t start, int64_t limit, UErrorCode &status) { |
1662 | 0 | return region(start, limit, -1, status); |
1663 | 0 | } |
1664 | | |
1665 | | //-------------------------------------------------------------------------------- |
1666 | | // |
1667 | | // regionEnd |
1668 | | // |
1669 | | //-------------------------------------------------------------------------------- |
1670 | 0 | int32_t RegexMatcher::regionEnd() const { |
1671 | 0 | return (int32_t)fRegionLimit; |
1672 | 0 | } |
1673 | | |
1674 | 0 | int64_t RegexMatcher::regionEnd64() const { |
1675 | 0 | return fRegionLimit; |
1676 | 0 | } |
1677 | | |
1678 | | //-------------------------------------------------------------------------------- |
1679 | | // |
1680 | | // regionStart |
1681 | | // |
1682 | | //-------------------------------------------------------------------------------- |
1683 | 0 | int32_t RegexMatcher::regionStart() const { |
1684 | 0 | return (int32_t)fRegionStart; |
1685 | 0 | } |
1686 | | |
1687 | 0 | int64_t RegexMatcher::regionStart64() const { |
1688 | 0 | return fRegionStart; |
1689 | 0 | } |
1690 | | |
1691 | | |
1692 | | //-------------------------------------------------------------------------------- |
1693 | | // |
1694 | | // replaceAll |
1695 | | // |
1696 | | //-------------------------------------------------------------------------------- |
1697 | 0 | UnicodeString RegexMatcher::replaceAll(const UnicodeString &replacement, UErrorCode &status) { |
1698 | 0 | UText replacementText = UTEXT_INITIALIZER; |
1699 | 0 | UText resultText = UTEXT_INITIALIZER; |
1700 | 0 | UnicodeString resultString; |
1701 | 0 | if (U_FAILURE(status)) { |
1702 | 0 | return resultString; |
1703 | 0 | } |
1704 | | |
1705 | 0 | utext_openConstUnicodeString(&replacementText, &replacement, &status); |
1706 | 0 | utext_openUnicodeString(&resultText, &resultString, &status); |
1707 | |
|
1708 | 0 | replaceAll(&replacementText, &resultText, status); |
1709 | |
|
1710 | 0 | utext_close(&resultText); |
1711 | 0 | utext_close(&replacementText); |
1712 | |
|
1713 | 0 | return resultString; |
1714 | 0 | } |
1715 | | |
1716 | | |
1717 | | // |
1718 | | // replaceAll, UText mode |
1719 | | // |
1720 | 0 | UText *RegexMatcher::replaceAll(UText *replacement, UText *dest, UErrorCode &status) { |
1721 | 0 | if (U_FAILURE(status)) { |
1722 | 0 | return dest; |
1723 | 0 | } |
1724 | 0 | if (U_FAILURE(fDeferredStatus)) { |
1725 | 0 | status = fDeferredStatus; |
1726 | 0 | return dest; |
1727 | 0 | } |
1728 | | |
1729 | 0 | if (dest == NULL) { |
1730 | 0 | UnicodeString emptyString; |
1731 | 0 | UText empty = UTEXT_INITIALIZER; |
1732 | |
|
1733 | 0 | utext_openUnicodeString(&empty, &emptyString, &status); |
1734 | 0 | dest = utext_clone(NULL, &empty, true, false, &status); |
1735 | 0 | utext_close(&empty); |
1736 | 0 | } |
1737 | |
|
1738 | 0 | if (U_SUCCESS(status)) { |
1739 | 0 | reset(); |
1740 | 0 | while (find()) { |
1741 | 0 | appendReplacement(dest, replacement, status); |
1742 | 0 | if (U_FAILURE(status)) { |
1743 | 0 | break; |
1744 | 0 | } |
1745 | 0 | } |
1746 | 0 | appendTail(dest, status); |
1747 | 0 | } |
1748 | |
|
1749 | 0 | return dest; |
1750 | 0 | } |
1751 | | |
1752 | | |
1753 | | //-------------------------------------------------------------------------------- |
1754 | | // |
1755 | | // replaceFirst |
1756 | | // |
1757 | | //-------------------------------------------------------------------------------- |
1758 | 0 | UnicodeString RegexMatcher::replaceFirst(const UnicodeString &replacement, UErrorCode &status) { |
1759 | 0 | UText replacementText = UTEXT_INITIALIZER; |
1760 | 0 | UText resultText = UTEXT_INITIALIZER; |
1761 | 0 | UnicodeString resultString; |
1762 | |
|
1763 | 0 | utext_openConstUnicodeString(&replacementText, &replacement, &status); |
1764 | 0 | utext_openUnicodeString(&resultText, &resultString, &status); |
1765 | |
|
1766 | 0 | replaceFirst(&replacementText, &resultText, status); |
1767 | |
|
1768 | 0 | utext_close(&resultText); |
1769 | 0 | utext_close(&replacementText); |
1770 | |
|
1771 | 0 | return resultString; |
1772 | 0 | } |
1773 | | |
1774 | | // |
1775 | | // replaceFirst, UText mode |
1776 | | // |
1777 | 0 | UText *RegexMatcher::replaceFirst(UText *replacement, UText *dest, UErrorCode &status) { |
1778 | 0 | if (U_FAILURE(status)) { |
1779 | 0 | return dest; |
1780 | 0 | } |
1781 | 0 | if (U_FAILURE(fDeferredStatus)) { |
1782 | 0 | status = fDeferredStatus; |
1783 | 0 | return dest; |
1784 | 0 | } |
1785 | | |
1786 | 0 | reset(); |
1787 | 0 | if (!find()) { |
1788 | 0 | return getInput(dest, status); |
1789 | 0 | } |
1790 | | |
1791 | 0 | if (dest == NULL) { |
1792 | 0 | UnicodeString emptyString; |
1793 | 0 | UText empty = UTEXT_INITIALIZER; |
1794 | |
|
1795 | 0 | utext_openUnicodeString(&empty, &emptyString, &status); |
1796 | 0 | dest = utext_clone(NULL, &empty, true, false, &status); |
1797 | 0 | utext_close(&empty); |
1798 | 0 | } |
1799 | |
|
1800 | 0 | appendReplacement(dest, replacement, status); |
1801 | 0 | appendTail(dest, status); |
1802 | |
|
1803 | 0 | return dest; |
1804 | 0 | } |
1805 | | |
1806 | | |
1807 | | //-------------------------------------------------------------------------------- |
1808 | | // |
1809 | | // requireEnd |
1810 | | // |
1811 | | //-------------------------------------------------------------------------------- |
1812 | 0 | UBool RegexMatcher::requireEnd() const { |
1813 | 0 | return fRequireEnd; |
1814 | 0 | } |
1815 | | |
1816 | | |
1817 | | //-------------------------------------------------------------------------------- |
1818 | | // |
1819 | | // reset |
1820 | | // |
1821 | | //-------------------------------------------------------------------------------- |
1822 | 7.89k | RegexMatcher &RegexMatcher::reset() { |
1823 | 7.89k | fRegionStart = 0; |
1824 | 7.89k | fRegionLimit = fInputLength; |
1825 | 7.89k | fActiveStart = 0; |
1826 | 7.89k | fActiveLimit = fInputLength; |
1827 | 7.89k | fAnchorStart = 0; |
1828 | 7.89k | fAnchorLimit = fInputLength; |
1829 | 7.89k | fLookStart = 0; |
1830 | 7.89k | fLookLimit = fInputLength; |
1831 | 7.89k | resetPreserveRegion(); |
1832 | 7.89k | return *this; |
1833 | 7.89k | } |
1834 | | |
1835 | | |
1836 | | |
1837 | 7.89k | void RegexMatcher::resetPreserveRegion() { |
1838 | 7.89k | fMatchStart = 0; |
1839 | 7.89k | fMatchEnd = 0; |
1840 | 7.89k | fLastMatchEnd = -1; |
1841 | 7.89k | fAppendPosition = 0; |
1842 | 7.89k | fMatch = false; |
1843 | 7.89k | fHitEnd = false; |
1844 | 7.89k | fRequireEnd = false; |
1845 | 7.89k | fTime = 0; |
1846 | 7.89k | fTickCounter = TIMER_INITIAL_VALUE; |
1847 | | //resetStack(); // more expensive than it looks... |
1848 | 7.89k | } |
1849 | | |
1850 | | |
1851 | 0 | RegexMatcher &RegexMatcher::reset(const UnicodeString &input) { |
1852 | 0 | fInputText = utext_openConstUnicodeString(fInputText, &input, &fDeferredStatus); |
1853 | 0 | if (fPattern->fNeedsAltInput) { |
1854 | 0 | fAltInputText = utext_clone(fAltInputText, fInputText, false, true, &fDeferredStatus); |
1855 | 0 | } |
1856 | 0 | if (U_FAILURE(fDeferredStatus)) { |
1857 | 0 | return *this; |
1858 | 0 | } |
1859 | 0 | fInputLength = utext_nativeLength(fInputText); |
1860 | |
|
1861 | 0 | reset(); |
1862 | 0 | delete fInput; |
1863 | 0 | fInput = NULL; |
1864 | | |
1865 | | // Do the following for any UnicodeString. |
1866 | | // This is for compatibility for those clients who modify the input string "live" during regex operations. |
1867 | 0 | fInputUniStrMaybeMutable = true; |
1868 | |
|
1869 | 0 | #if UCONFIG_NO_BREAK_ITERATION==0 |
1870 | 0 | if (fWordBreakItr) { |
1871 | 0 | fWordBreakItr->setText(fInputText, fDeferredStatus); |
1872 | 0 | } |
1873 | 0 | if (fGCBreakItr) { |
1874 | 0 | fGCBreakItr->setText(fInputText, fDeferredStatus); |
1875 | 0 | } |
1876 | 0 | #endif |
1877 | |
|
1878 | 0 | return *this; |
1879 | 0 | } |
1880 | | |
1881 | | |
1882 | 3.94k | RegexMatcher &RegexMatcher::reset(UText *input) { |
1883 | 3.94k | if (fInputText != input) { |
1884 | 3.94k | fInputText = utext_clone(fInputText, input, false, true, &fDeferredStatus); |
1885 | 3.94k | if (fPattern->fNeedsAltInput) fAltInputText = utext_clone(fAltInputText, fInputText, false, true, &fDeferredStatus); |
1886 | 3.94k | if (U_FAILURE(fDeferredStatus)) { |
1887 | 0 | return *this; |
1888 | 0 | } |
1889 | 3.94k | fInputLength = utext_nativeLength(fInputText); |
1890 | | |
1891 | 3.94k | delete fInput; |
1892 | 3.94k | fInput = NULL; |
1893 | | |
1894 | 3.94k | #if UCONFIG_NO_BREAK_ITERATION==0 |
1895 | 3.94k | if (fWordBreakItr) { |
1896 | 0 | fWordBreakItr->setText(input, fDeferredStatus); |
1897 | 0 | } |
1898 | 3.94k | if (fGCBreakItr) { |
1899 | 0 | fGCBreakItr->setText(fInputText, fDeferredStatus); |
1900 | 0 | } |
1901 | 3.94k | #endif |
1902 | 3.94k | } |
1903 | 3.94k | reset(); |
1904 | 3.94k | fInputUniStrMaybeMutable = false; |
1905 | | |
1906 | 3.94k | return *this; |
1907 | 3.94k | } |
1908 | | |
1909 | | /*RegexMatcher &RegexMatcher::reset(const UChar *) { |
1910 | | fDeferredStatus = U_INTERNAL_PROGRAM_ERROR; |
1911 | | return *this; |
1912 | | }*/ |
1913 | | |
1914 | 0 | RegexMatcher &RegexMatcher::reset(int64_t position, UErrorCode &status) { |
1915 | 0 | if (U_FAILURE(status)) { |
1916 | 0 | return *this; |
1917 | 0 | } |
1918 | 0 | reset(); // Reset also resets the region to be the entire string. |
1919 | |
|
1920 | 0 | if (position < 0 || position > fActiveLimit) { |
1921 | 0 | status = U_INDEX_OUTOFBOUNDS_ERROR; |
1922 | 0 | return *this; |
1923 | 0 | } |
1924 | 0 | fMatchEnd = position; |
1925 | 0 | return *this; |
1926 | 0 | } |
1927 | | |
1928 | | |
1929 | | //-------------------------------------------------------------------------------- |
1930 | | // |
1931 | | // refresh |
1932 | | // |
1933 | | //-------------------------------------------------------------------------------- |
1934 | 0 | RegexMatcher &RegexMatcher::refreshInputText(UText *input, UErrorCode &status) { |
1935 | 0 | if (U_FAILURE(status)) { |
1936 | 0 | return *this; |
1937 | 0 | } |
1938 | 0 | if (input == NULL) { |
1939 | 0 | status = U_ILLEGAL_ARGUMENT_ERROR; |
1940 | 0 | return *this; |
1941 | 0 | } |
1942 | 0 | if (utext_nativeLength(fInputText) != utext_nativeLength(input)) { |
1943 | 0 | status = U_ILLEGAL_ARGUMENT_ERROR; |
1944 | 0 | return *this; |
1945 | 0 | } |
1946 | 0 | int64_t pos = utext_getNativeIndex(fInputText); |
1947 | | // Shallow read-only clone of the new UText into the existing input UText |
1948 | 0 | fInputText = utext_clone(fInputText, input, false, true, &status); |
1949 | 0 | if (U_FAILURE(status)) { |
1950 | 0 | return *this; |
1951 | 0 | } |
1952 | 0 | utext_setNativeIndex(fInputText, pos); |
1953 | |
|
1954 | 0 | if (fAltInputText != NULL) { |
1955 | 0 | pos = utext_getNativeIndex(fAltInputText); |
1956 | 0 | fAltInputText = utext_clone(fAltInputText, input, false, true, &status); |
1957 | 0 | if (U_FAILURE(status)) { |
1958 | 0 | return *this; |
1959 | 0 | } |
1960 | 0 | utext_setNativeIndex(fAltInputText, pos); |
1961 | 0 | } |
1962 | 0 | return *this; |
1963 | 0 | } |
1964 | | |
1965 | | |
1966 | | |
1967 | | //-------------------------------------------------------------------------------- |
1968 | | // |
1969 | | // setTrace |
1970 | | // |
1971 | | //-------------------------------------------------------------------------------- |
1972 | 0 | void RegexMatcher::setTrace(UBool state) { |
1973 | 0 | fTraceDebug = state; |
1974 | 0 | } |
1975 | | |
1976 | | |
1977 | | |
1978 | | /** |
1979 | | * UText, replace entire contents of the destination UText with a substring of the source UText. |
1980 | | * |
1981 | | * @param src The source UText |
1982 | | * @param dest The destination UText. Must be writable. |
1983 | | * May be NULL, in which case a new UText will be allocated. |
1984 | | * @param start Start index of source substring. |
1985 | | * @param limit Limit index of source substring. |
1986 | | * @param status An error code. |
1987 | | */ |
1988 | 0 | static UText *utext_extract_replace(UText *src, UText *dest, int64_t start, int64_t limit, UErrorCode *status) { |
1989 | 0 | if (U_FAILURE(*status)) { |
1990 | 0 | return dest; |
1991 | 0 | } |
1992 | 0 | if (start == limit) { |
1993 | 0 | if (dest) { |
1994 | 0 | utext_replace(dest, 0, utext_nativeLength(dest), NULL, 0, status); |
1995 | 0 | return dest; |
1996 | 0 | } else { |
1997 | 0 | return utext_openUChars(NULL, NULL, 0, status); |
1998 | 0 | } |
1999 | 0 | } |
2000 | 0 | int32_t length = utext_extract(src, start, limit, NULL, 0, status); |
2001 | 0 | if (*status != U_BUFFER_OVERFLOW_ERROR && U_FAILURE(*status)) { |
2002 | 0 | return dest; |
2003 | 0 | } |
2004 | 0 | *status = U_ZERO_ERROR; |
2005 | 0 | MaybeStackArray<UChar, 40> buffer; |
2006 | 0 | if (length >= buffer.getCapacity()) { |
2007 | 0 | UChar *newBuf = buffer.resize(length+1); // Leave space for terminating Nul. |
2008 | 0 | if (newBuf == NULL) { |
2009 | 0 | *status = U_MEMORY_ALLOCATION_ERROR; |
2010 | 0 | } |
2011 | 0 | } |
2012 | 0 | utext_extract(src, start, limit, buffer.getAlias(), length+1, status); |
2013 | 0 | if (dest) { |
2014 | 0 | utext_replace(dest, 0, utext_nativeLength(dest), buffer.getAlias(), length, status); |
2015 | 0 | return dest; |
2016 | 0 | } |
2017 | | |
2018 | | // Caller did not provide a preexisting UText. |
2019 | | // Open a new one, and have it adopt the text buffer storage. |
2020 | 0 | if (U_FAILURE(*status)) { |
2021 | 0 | return NULL; |
2022 | 0 | } |
2023 | 0 | int32_t ownedLength = 0; |
2024 | 0 | UChar *ownedBuf = buffer.orphanOrClone(length+1, ownedLength); |
2025 | 0 | if (ownedBuf == NULL) { |
2026 | 0 | *status = U_MEMORY_ALLOCATION_ERROR; |
2027 | 0 | return NULL; |
2028 | 0 | } |
2029 | 0 | UText *result = utext_openUChars(NULL, ownedBuf, length, status); |
2030 | 0 | if (U_FAILURE(*status)) { |
2031 | 0 | uprv_free(ownedBuf); |
2032 | 0 | return NULL; |
2033 | 0 | } |
2034 | 0 | result->providerProperties |= (1 << UTEXT_PROVIDER_OWNS_TEXT); |
2035 | 0 | return result; |
2036 | 0 | } |
2037 | | |
2038 | | |
2039 | | //--------------------------------------------------------------------- |
2040 | | // |
2041 | | // split |
2042 | | // |
2043 | | //--------------------------------------------------------------------- |
2044 | | int32_t RegexMatcher::split(const UnicodeString &input, |
2045 | | UnicodeString dest[], |
2046 | | int32_t destCapacity, |
2047 | | UErrorCode &status) |
2048 | 0 | { |
2049 | 0 | UText inputText = UTEXT_INITIALIZER; |
2050 | 0 | utext_openConstUnicodeString(&inputText, &input, &status); |
2051 | 0 | if (U_FAILURE(status)) { |
2052 | 0 | return 0; |
2053 | 0 | } |
2054 | | |
2055 | 0 | UText **destText = (UText **)uprv_malloc(sizeof(UText*)*destCapacity); |
2056 | 0 | if (destText == NULL) { |
2057 | 0 | status = U_MEMORY_ALLOCATION_ERROR; |
2058 | 0 | return 0; |
2059 | 0 | } |
2060 | 0 | int32_t i; |
2061 | 0 | for (i = 0; i < destCapacity; i++) { |
2062 | 0 | destText[i] = utext_openUnicodeString(NULL, &dest[i], &status); |
2063 | 0 | } |
2064 | |
|
2065 | 0 | int32_t fieldCount = split(&inputText, destText, destCapacity, status); |
2066 | |
|
2067 | 0 | for (i = 0; i < destCapacity; i++) { |
2068 | 0 | utext_close(destText[i]); |
2069 | 0 | } |
2070 | |
|
2071 | 0 | uprv_free(destText); |
2072 | 0 | utext_close(&inputText); |
2073 | 0 | return fieldCount; |
2074 | 0 | } |
2075 | | |
2076 | | // |
2077 | | // split, UText mode |
2078 | | // |
2079 | | int32_t RegexMatcher::split(UText *input, |
2080 | | UText *dest[], |
2081 | | int32_t destCapacity, |
2082 | | UErrorCode &status) |
2083 | 0 | { |
2084 | | // |
2085 | | // Check arguments for validity |
2086 | | // |
2087 | 0 | if (U_FAILURE(status)) { |
2088 | 0 | return 0; |
2089 | 0 | } |
2090 | | |
2091 | 0 | if (destCapacity < 1) { |
2092 | 0 | status = U_ILLEGAL_ARGUMENT_ERROR; |
2093 | 0 | return 0; |
2094 | 0 | } |
2095 | | |
2096 | | // |
2097 | | // Reset for the input text |
2098 | | // |
2099 | 0 | reset(input); |
2100 | 0 | int64_t nextOutputStringStart = 0; |
2101 | 0 | if (fActiveLimit == 0) { |
2102 | 0 | return 0; |
2103 | 0 | } |
2104 | | |
2105 | | // |
2106 | | // Loop through the input text, searching for the delimiter pattern |
2107 | | // |
2108 | 0 | int32_t i; |
2109 | 0 | int32_t numCaptureGroups = fPattern->fGroupMap->size(); |
2110 | 0 | for (i=0; ; i++) { |
2111 | 0 | if (i>=destCapacity-1) { |
2112 | | // There is one or zero output string left. |
2113 | | // Fill the last output string with whatever is left from the input, then exit the loop. |
2114 | | // ( i will be == destCapacity if we filled the output array while processing |
2115 | | // capture groups of the delimiter expression, in which case we will discard the |
2116 | | // last capture group saved in favor of the unprocessed remainder of the |
2117 | | // input string.) |
2118 | 0 | i = destCapacity-1; |
2119 | 0 | if (fActiveLimit > nextOutputStringStart) { |
2120 | 0 | if (UTEXT_FULL_TEXT_IN_CHUNK(input, fInputLength)) { |
2121 | 0 | if (dest[i]) { |
2122 | 0 | utext_replace(dest[i], 0, utext_nativeLength(dest[i]), |
2123 | 0 | input->chunkContents+nextOutputStringStart, |
2124 | 0 | (int32_t)(fActiveLimit-nextOutputStringStart), &status); |
2125 | 0 | } else { |
2126 | 0 | UText remainingText = UTEXT_INITIALIZER; |
2127 | 0 | utext_openUChars(&remainingText, input->chunkContents+nextOutputStringStart, |
2128 | 0 | fActiveLimit-nextOutputStringStart, &status); |
2129 | 0 | dest[i] = utext_clone(NULL, &remainingText, true, false, &status); |
2130 | 0 | utext_close(&remainingText); |
2131 | 0 | } |
2132 | 0 | } else { |
2133 | 0 | UErrorCode lengthStatus = U_ZERO_ERROR; |
2134 | 0 | int32_t remaining16Length = |
2135 | 0 | utext_extract(input, nextOutputStringStart, fActiveLimit, NULL, 0, &lengthStatus); |
2136 | 0 | UChar *remainingChars = (UChar *)uprv_malloc(sizeof(UChar)*(remaining16Length+1)); |
2137 | 0 | if (remainingChars == NULL) { |
2138 | 0 | status = U_MEMORY_ALLOCATION_ERROR; |
2139 | 0 | break; |
2140 | 0 | } |
2141 | | |
2142 | 0 | utext_extract(input, nextOutputStringStart, fActiveLimit, remainingChars, remaining16Length+1, &status); |
2143 | 0 | if (dest[i]) { |
2144 | 0 | utext_replace(dest[i], 0, utext_nativeLength(dest[i]), remainingChars, remaining16Length, &status); |
2145 | 0 | } else { |
2146 | 0 | UText remainingText = UTEXT_INITIALIZER; |
2147 | 0 | utext_openUChars(&remainingText, remainingChars, remaining16Length, &status); |
2148 | 0 | dest[i] = utext_clone(NULL, &remainingText, true, false, &status); |
2149 | 0 | utext_close(&remainingText); |
2150 | 0 | } |
2151 | |
|
2152 | 0 | uprv_free(remainingChars); |
2153 | 0 | } |
2154 | 0 | } |
2155 | 0 | break; |
2156 | 0 | } |
2157 | 0 | if (find()) { |
2158 | | // We found another delimiter. Move everything from where we started looking |
2159 | | // up until the start of the delimiter into the next output string. |
2160 | 0 | if (UTEXT_FULL_TEXT_IN_CHUNK(input, fInputLength)) { |
2161 | 0 | if (dest[i]) { |
2162 | 0 | utext_replace(dest[i], 0, utext_nativeLength(dest[i]), |
2163 | 0 | input->chunkContents+nextOutputStringStart, |
2164 | 0 | (int32_t)(fMatchStart-nextOutputStringStart), &status); |
2165 | 0 | } else { |
2166 | 0 | UText remainingText = UTEXT_INITIALIZER; |
2167 | 0 | utext_openUChars(&remainingText, input->chunkContents+nextOutputStringStart, |
2168 | 0 | fMatchStart-nextOutputStringStart, &status); |
2169 | 0 | dest[i] = utext_clone(NULL, &remainingText, true, false, &status); |
2170 | 0 | utext_close(&remainingText); |
2171 | 0 | } |
2172 | 0 | } else { |
2173 | 0 | UErrorCode lengthStatus = U_ZERO_ERROR; |
2174 | 0 | int32_t remaining16Length = utext_extract(input, nextOutputStringStart, fMatchStart, NULL, 0, &lengthStatus); |
2175 | 0 | UChar *remainingChars = (UChar *)uprv_malloc(sizeof(UChar)*(remaining16Length+1)); |
2176 | 0 | if (remainingChars == NULL) { |
2177 | 0 | status = U_MEMORY_ALLOCATION_ERROR; |
2178 | 0 | break; |
2179 | 0 | } |
2180 | 0 | utext_extract(input, nextOutputStringStart, fMatchStart, remainingChars, remaining16Length+1, &status); |
2181 | 0 | if (dest[i]) { |
2182 | 0 | utext_replace(dest[i], 0, utext_nativeLength(dest[i]), remainingChars, remaining16Length, &status); |
2183 | 0 | } else { |
2184 | 0 | UText remainingText = UTEXT_INITIALIZER; |
2185 | 0 | utext_openUChars(&remainingText, remainingChars, remaining16Length, &status); |
2186 | 0 | dest[i] = utext_clone(NULL, &remainingText, true, false, &status); |
2187 | 0 | utext_close(&remainingText); |
2188 | 0 | } |
2189 | |
|
2190 | 0 | uprv_free(remainingChars); |
2191 | 0 | } |
2192 | 0 | nextOutputStringStart = fMatchEnd; |
2193 | | |
2194 | | // If the delimiter pattern has capturing parentheses, the captured |
2195 | | // text goes out into the next n destination strings. |
2196 | 0 | int32_t groupNum; |
2197 | 0 | for (groupNum=1; groupNum<=numCaptureGroups; groupNum++) { |
2198 | 0 | if (i >= destCapacity-2) { |
2199 | | // Never fill the last available output string with capture group text. |
2200 | | // It will filled with the last field, the remainder of the |
2201 | | // unsplit input text. |
2202 | 0 | break; |
2203 | 0 | } |
2204 | 0 | i++; |
2205 | 0 | dest[i] = utext_extract_replace(fInputText, dest[i], |
2206 | 0 | start64(groupNum, status), end64(groupNum, status), &status); |
2207 | 0 | } |
2208 | |
|
2209 | 0 | if (nextOutputStringStart == fActiveLimit) { |
2210 | | // The delimiter was at the end of the string. We're done, but first |
2211 | | // we output one last empty string, for the empty field following |
2212 | | // the delimiter at the end of input. |
2213 | 0 | if (i+1 < destCapacity) { |
2214 | 0 | ++i; |
2215 | 0 | if (dest[i] == NULL) { |
2216 | 0 | dest[i] = utext_openUChars(NULL, NULL, 0, &status); |
2217 | 0 | } else { |
2218 | 0 | static const UChar emptyString[] = {(UChar)0}; |
2219 | 0 | utext_replace(dest[i], 0, utext_nativeLength(dest[i]), emptyString, 0, &status); |
2220 | 0 | } |
2221 | 0 | } |
2222 | 0 | break; |
2223 | |
|
2224 | 0 | } |
2225 | 0 | } |
2226 | 0 | else |
2227 | 0 | { |
2228 | | // We ran off the end of the input while looking for the next delimiter. |
2229 | | // All the remaining text goes into the current output string. |
2230 | 0 | if (UTEXT_FULL_TEXT_IN_CHUNK(input, fInputLength)) { |
2231 | 0 | if (dest[i]) { |
2232 | 0 | utext_replace(dest[i], 0, utext_nativeLength(dest[i]), |
2233 | 0 | input->chunkContents+nextOutputStringStart, |
2234 | 0 | (int32_t)(fActiveLimit-nextOutputStringStart), &status); |
2235 | 0 | } else { |
2236 | 0 | UText remainingText = UTEXT_INITIALIZER; |
2237 | 0 | utext_openUChars(&remainingText, input->chunkContents+nextOutputStringStart, |
2238 | 0 | fActiveLimit-nextOutputStringStart, &status); |
2239 | 0 | dest[i] = utext_clone(NULL, &remainingText, true, false, &status); |
2240 | 0 | utext_close(&remainingText); |
2241 | 0 | } |
2242 | 0 | } else { |
2243 | 0 | UErrorCode lengthStatus = U_ZERO_ERROR; |
2244 | 0 | int32_t remaining16Length = utext_extract(input, nextOutputStringStart, fActiveLimit, NULL, 0, &lengthStatus); |
2245 | 0 | UChar *remainingChars = (UChar *)uprv_malloc(sizeof(UChar)*(remaining16Length+1)); |
2246 | 0 | if (remainingChars == NULL) { |
2247 | 0 | status = U_MEMORY_ALLOCATION_ERROR; |
2248 | 0 | break; |
2249 | 0 | } |
2250 | | |
2251 | 0 | utext_extract(input, nextOutputStringStart, fActiveLimit, remainingChars, remaining16Length+1, &status); |
2252 | 0 | if (dest[i]) { |
2253 | 0 | utext_replace(dest[i], 0, utext_nativeLength(dest[i]), remainingChars, remaining16Length, &status); |
2254 | 0 | } else { |
2255 | 0 | UText remainingText = UTEXT_INITIALIZER; |
2256 | 0 | utext_openUChars(&remainingText, remainingChars, remaining16Length, &status); |
2257 | 0 | dest[i] = utext_clone(NULL, &remainingText, true, false, &status); |
2258 | 0 | utext_close(&remainingText); |
2259 | 0 | } |
2260 | |
|
2261 | 0 | uprv_free(remainingChars); |
2262 | 0 | } |
2263 | 0 | break; |
2264 | 0 | } |
2265 | 0 | if (U_FAILURE(status)) { |
2266 | 0 | break; |
2267 | 0 | } |
2268 | 0 | } // end of for loop |
2269 | 0 | return i+1; |
2270 | 0 | } |
2271 | | |
2272 | | |
2273 | | //-------------------------------------------------------------------------------- |
2274 | | // |
2275 | | // start |
2276 | | // |
2277 | | //-------------------------------------------------------------------------------- |
2278 | 0 | int32_t RegexMatcher::start(UErrorCode &status) const { |
2279 | 0 | return start(0, status); |
2280 | 0 | } |
2281 | | |
2282 | 0 | int64_t RegexMatcher::start64(UErrorCode &status) const { |
2283 | 0 | return start64(0, status); |
2284 | 0 | } |
2285 | | |
2286 | | //-------------------------------------------------------------------------------- |
2287 | | // |
2288 | | // start(int32_t group, UErrorCode &status) |
2289 | | // |
2290 | | //-------------------------------------------------------------------------------- |
2291 | | |
2292 | 0 | int64_t RegexMatcher::start64(int32_t group, UErrorCode &status) const { |
2293 | 0 | if (U_FAILURE(status)) { |
2294 | 0 | return -1; |
2295 | 0 | } |
2296 | 0 | if (U_FAILURE(fDeferredStatus)) { |
2297 | 0 | status = fDeferredStatus; |
2298 | 0 | return -1; |
2299 | 0 | } |
2300 | 0 | if (fMatch == false) { |
2301 | 0 | status = U_REGEX_INVALID_STATE; |
2302 | 0 | return -1; |
2303 | 0 | } |
2304 | 0 | if (group < 0 || group > fPattern->fGroupMap->size()) { |
2305 | 0 | status = U_INDEX_OUTOFBOUNDS_ERROR; |
2306 | 0 | return -1; |
2307 | 0 | } |
2308 | 0 | int64_t s; |
2309 | 0 | if (group == 0) { |
2310 | 0 | s = fMatchStart; |
2311 | 0 | } else { |
2312 | 0 | int32_t groupOffset = fPattern->fGroupMap->elementAti(group-1); |
2313 | 0 | U_ASSERT(groupOffset < fPattern->fFrameSize); |
2314 | 0 | U_ASSERT(groupOffset >= 0); |
2315 | 0 | s = fFrame->fExtra[groupOffset]; |
2316 | 0 | } |
2317 | |
|
2318 | 0 | return s; |
2319 | 0 | } |
2320 | | |
2321 | | |
2322 | 0 | int32_t RegexMatcher::start(int32_t group, UErrorCode &status) const { |
2323 | 0 | return (int32_t)start64(group, status); |
2324 | 0 | } |
2325 | | |
2326 | | //-------------------------------------------------------------------------------- |
2327 | | // |
2328 | | // useAnchoringBounds |
2329 | | // |
2330 | | //-------------------------------------------------------------------------------- |
2331 | 0 | RegexMatcher &RegexMatcher::useAnchoringBounds(UBool b) { |
2332 | 0 | fAnchoringBounds = b; |
2333 | 0 | fAnchorStart = (fAnchoringBounds ? fRegionStart : 0); |
2334 | 0 | fAnchorLimit = (fAnchoringBounds ? fRegionLimit : fInputLength); |
2335 | 0 | return *this; |
2336 | 0 | } |
2337 | | |
2338 | | |
2339 | | //-------------------------------------------------------------------------------- |
2340 | | // |
2341 | | // useTransparentBounds |
2342 | | // |
2343 | | //-------------------------------------------------------------------------------- |
2344 | 0 | RegexMatcher &RegexMatcher::useTransparentBounds(UBool b) { |
2345 | 0 | fTransparentBounds = b; |
2346 | 0 | fLookStart = (fTransparentBounds ? 0 : fRegionStart); |
2347 | 0 | fLookLimit = (fTransparentBounds ? fInputLength : fRegionLimit); |
2348 | 0 | return *this; |
2349 | 0 | } |
2350 | | |
2351 | | //-------------------------------------------------------------------------------- |
2352 | | // |
2353 | | // setTimeLimit |
2354 | | // |
2355 | | //-------------------------------------------------------------------------------- |
2356 | 0 | void RegexMatcher::setTimeLimit(int32_t limit, UErrorCode &status) { |
2357 | 0 | if (U_FAILURE(status)) { |
2358 | 0 | return; |
2359 | 0 | } |
2360 | 0 | if (U_FAILURE(fDeferredStatus)) { |
2361 | 0 | status = fDeferredStatus; |
2362 | 0 | return; |
2363 | 0 | } |
2364 | 0 | if (limit < 0) { |
2365 | 0 | status = U_ILLEGAL_ARGUMENT_ERROR; |
2366 | 0 | return; |
2367 | 0 | } |
2368 | 0 | fTimeLimit = limit; |
2369 | 0 | } |
2370 | | |
2371 | | |
2372 | | //-------------------------------------------------------------------------------- |
2373 | | // |
2374 | | // getTimeLimit |
2375 | | // |
2376 | | //-------------------------------------------------------------------------------- |
2377 | 0 | int32_t RegexMatcher::getTimeLimit() const { |
2378 | 0 | return fTimeLimit; |
2379 | 0 | } |
2380 | | |
2381 | | |
2382 | | //-------------------------------------------------------------------------------- |
2383 | | // |
2384 | | // setStackLimit |
2385 | | // |
2386 | | //-------------------------------------------------------------------------------- |
2387 | 3.94k | void RegexMatcher::setStackLimit(int32_t limit, UErrorCode &status) { |
2388 | 3.94k | if (U_FAILURE(status)) { |
2389 | 0 | return; |
2390 | 0 | } |
2391 | 3.94k | if (U_FAILURE(fDeferredStatus)) { |
2392 | 0 | status = fDeferredStatus; |
2393 | 0 | return; |
2394 | 0 | } |
2395 | 3.94k | if (limit < 0) { |
2396 | 0 | status = U_ILLEGAL_ARGUMENT_ERROR; |
2397 | 0 | return; |
2398 | 0 | } |
2399 | | |
2400 | | // Reset the matcher. This is needed here in case there is a current match |
2401 | | // whose final stack frame (containing the match results, pointed to by fFrame) |
2402 | | // would be lost by resizing to a smaller stack size. |
2403 | 3.94k | reset(); |
2404 | | |
2405 | 3.94k | if (limit == 0) { |
2406 | | // Unlimited stack expansion |
2407 | 0 | fStack->setMaxCapacity(0); |
2408 | 3.94k | } else { |
2409 | | // Change the units of the limit from bytes to ints, and bump the size up |
2410 | | // to be big enough to hold at least one stack frame for the pattern, |
2411 | | // if it isn't there already. |
2412 | 3.94k | int32_t adjustedLimit = limit / sizeof(int32_t); |
2413 | 3.94k | if (adjustedLimit < fPattern->fFrameSize) { |
2414 | 0 | adjustedLimit = fPattern->fFrameSize; |
2415 | 0 | } |
2416 | 3.94k | fStack->setMaxCapacity(adjustedLimit); |
2417 | 3.94k | } |
2418 | 3.94k | fStackLimit = limit; |
2419 | 3.94k | } |
2420 | | |
2421 | | |
2422 | | //-------------------------------------------------------------------------------- |
2423 | | // |
2424 | | // getStackLimit |
2425 | | // |
2426 | | //-------------------------------------------------------------------------------- |
2427 | 0 | int32_t RegexMatcher::getStackLimit() const { |
2428 | 0 | return fStackLimit; |
2429 | 0 | } |
2430 | | |
2431 | | |
2432 | | //-------------------------------------------------------------------------------- |
2433 | | // |
2434 | | // setMatchCallback |
2435 | | // |
2436 | | //-------------------------------------------------------------------------------- |
2437 | | void RegexMatcher::setMatchCallback(URegexMatchCallback *callback, |
2438 | | const void *context, |
2439 | 0 | UErrorCode &status) { |
2440 | 0 | if (U_FAILURE(status)) { |
2441 | 0 | return; |
2442 | 0 | } |
2443 | 0 | fCallbackFn = callback; |
2444 | 0 | fCallbackContext = context; |
2445 | 0 | } |
2446 | | |
2447 | | |
2448 | | //-------------------------------------------------------------------------------- |
2449 | | // |
2450 | | // getMatchCallback |
2451 | | // |
2452 | | //-------------------------------------------------------------------------------- |
2453 | | void RegexMatcher::getMatchCallback(URegexMatchCallback *&callback, |
2454 | | const void *&context, |
2455 | 0 | UErrorCode &status) { |
2456 | 0 | if (U_FAILURE(status)) { |
2457 | 0 | return; |
2458 | 0 | } |
2459 | 0 | callback = fCallbackFn; |
2460 | 0 | context = fCallbackContext; |
2461 | 0 | } |
2462 | | |
2463 | | |
2464 | | //-------------------------------------------------------------------------------- |
2465 | | // |
2466 | | // setMatchCallback |
2467 | | // |
2468 | | //-------------------------------------------------------------------------------- |
2469 | | void RegexMatcher::setFindProgressCallback(URegexFindProgressCallback *callback, |
2470 | | const void *context, |
2471 | 0 | UErrorCode &status) { |
2472 | 0 | if (U_FAILURE(status)) { |
2473 | 0 | return; |
2474 | 0 | } |
2475 | 0 | fFindProgressCallbackFn = callback; |
2476 | 0 | fFindProgressCallbackContext = context; |
2477 | 0 | } |
2478 | | |
2479 | | |
2480 | | //-------------------------------------------------------------------------------- |
2481 | | // |
2482 | | // getMatchCallback |
2483 | | // |
2484 | | //-------------------------------------------------------------------------------- |
2485 | | void RegexMatcher::getFindProgressCallback(URegexFindProgressCallback *&callback, |
2486 | | const void *&context, |
2487 | 0 | UErrorCode &status) { |
2488 | 0 | if (U_FAILURE(status)) { |
2489 | 0 | return; |
2490 | 0 | } |
2491 | 0 | callback = fFindProgressCallbackFn; |
2492 | 0 | context = fFindProgressCallbackContext; |
2493 | 0 | } |
2494 | | |
2495 | | |
2496 | | //================================================================================ |
2497 | | // |
2498 | | // Code following this point in this file is the internal |
2499 | | // Match Engine Implementation. |
2500 | | // |
2501 | | //================================================================================ |
2502 | | |
2503 | | |
2504 | | //-------------------------------------------------------------------------------- |
2505 | | // |
2506 | | // resetStack |
2507 | | // Discard any previous contents of the state save stack, and initialize a |
2508 | | // new stack frame to all -1. The -1s are needed for capture group limits, |
2509 | | // where they indicate that a group has not yet matched anything. |
2510 | | //-------------------------------------------------------------------------------- |
2511 | 0 | REStackFrame *RegexMatcher::resetStack() { |
2512 | | // Discard any previous contents of the state save stack, and initialize a |
2513 | | // new stack frame with all -1 data. The -1s are needed for capture group limits, |
2514 | | // where they indicate that a group has not yet matched anything. |
2515 | 0 | fStack->removeAllElements(); |
2516 | |
|
2517 | 0 | REStackFrame *iFrame = (REStackFrame *)fStack->reserveBlock(fPattern->fFrameSize, fDeferredStatus); |
2518 | 0 | if(U_FAILURE(fDeferredStatus)) { |
2519 | 0 | return NULL; |
2520 | 0 | } |
2521 | | |
2522 | 0 | int32_t i; |
2523 | 0 | for (i=0; i<fPattern->fFrameSize-RESTACKFRAME_HDRCOUNT; i++) { |
2524 | 0 | iFrame->fExtra[i] = -1; |
2525 | 0 | } |
2526 | 0 | return iFrame; |
2527 | 0 | } |
2528 | | |
2529 | | |
2530 | | |
2531 | | //-------------------------------------------------------------------------------- |
2532 | | // |
2533 | | // isWordBoundary |
2534 | | // in perl, "xab..cd..", \b is true at positions 0,3,5,7 |
2535 | | // For us, |
2536 | | // If the current char is a combining mark, |
2537 | | // \b is false. |
2538 | | // Else Scan backwards to the first non-combining char. |
2539 | | // We are at a boundary if the this char and the original chars are |
2540 | | // opposite in membership in \w set |
2541 | | // |
2542 | | // parameters: pos - the current position in the input buffer |
2543 | | // |
2544 | | // TODO: double-check edge cases at region boundaries. |
2545 | | // |
2546 | | //-------------------------------------------------------------------------------- |
2547 | 0 | UBool RegexMatcher::isWordBoundary(int64_t pos) { |
2548 | 0 | UBool isBoundary = false; |
2549 | 0 | UBool cIsWord = false; |
2550 | |
|
2551 | 0 | if (pos >= fLookLimit) { |
2552 | 0 | fHitEnd = true; |
2553 | 0 | } else { |
2554 | | // Determine whether char c at current position is a member of the word set of chars. |
2555 | | // If we're off the end of the string, behave as though we're not at a word char. |
2556 | 0 | UTEXT_SETNATIVEINDEX(fInputText, pos); |
2557 | 0 | UChar32 c = UTEXT_CURRENT32(fInputText); |
2558 | 0 | if (u_hasBinaryProperty(c, UCHAR_GRAPHEME_EXTEND) || u_charType(c) == U_FORMAT_CHAR) { |
2559 | | // Current char is a combining one. Not a boundary. |
2560 | 0 | return false; |
2561 | 0 | } |
2562 | 0 | cIsWord = RegexStaticSets::gStaticSets->fPropSets[URX_ISWORD_SET].contains(c); |
2563 | 0 | } |
2564 | | |
2565 | | // Back up until we come to a non-combining char, determine whether |
2566 | | // that char is a word char. |
2567 | 0 | UBool prevCIsWord = false; |
2568 | 0 | for (;;) { |
2569 | 0 | if (UTEXT_GETNATIVEINDEX(fInputText) <= fLookStart) { |
2570 | 0 | break; |
2571 | 0 | } |
2572 | 0 | UChar32 prevChar = UTEXT_PREVIOUS32(fInputText); |
2573 | 0 | if (!(u_hasBinaryProperty(prevChar, UCHAR_GRAPHEME_EXTEND) |
2574 | 0 | || u_charType(prevChar) == U_FORMAT_CHAR)) { |
2575 | 0 | prevCIsWord = RegexStaticSets::gStaticSets->fPropSets[URX_ISWORD_SET].contains(prevChar); |
2576 | 0 | break; |
2577 | 0 | } |
2578 | 0 | } |
2579 | 0 | isBoundary = cIsWord ^ prevCIsWord; |
2580 | 0 | return isBoundary; |
2581 | 0 | } |
2582 | | |
2583 | 0 | UBool RegexMatcher::isChunkWordBoundary(int32_t pos) { |
2584 | 0 | UBool isBoundary = false; |
2585 | 0 | UBool cIsWord = false; |
2586 | |
|
2587 | 0 | const UChar *inputBuf = fInputText->chunkContents; |
2588 | |
|
2589 | 0 | if (pos >= fLookLimit) { |
2590 | 0 | fHitEnd = true; |
2591 | 0 | } else { |
2592 | | // Determine whether char c at current position is a member of the word set of chars. |
2593 | | // If we're off the end of the string, behave as though we're not at a word char. |
2594 | 0 | UChar32 c; |
2595 | 0 | U16_GET(inputBuf, fLookStart, pos, fLookLimit, c); |
2596 | 0 | if (u_hasBinaryProperty(c, UCHAR_GRAPHEME_EXTEND) || u_charType(c) == U_FORMAT_CHAR) { |
2597 | | // Current char is a combining one. Not a boundary. |
2598 | 0 | return false; |
2599 | 0 | } |
2600 | 0 | cIsWord = RegexStaticSets::gStaticSets->fPropSets[URX_ISWORD_SET].contains(c); |
2601 | 0 | } |
2602 | | |
2603 | | // Back up until we come to a non-combining char, determine whether |
2604 | | // that char is a word char. |
2605 | 0 | UBool prevCIsWord = false; |
2606 | 0 | for (;;) { |
2607 | 0 | if (pos <= fLookStart) { |
2608 | 0 | break; |
2609 | 0 | } |
2610 | 0 | UChar32 prevChar; |
2611 | 0 | U16_PREV(inputBuf, fLookStart, pos, prevChar); |
2612 | 0 | if (!(u_hasBinaryProperty(prevChar, UCHAR_GRAPHEME_EXTEND) |
2613 | 0 | || u_charType(prevChar) == U_FORMAT_CHAR)) { |
2614 | 0 | prevCIsWord = RegexStaticSets::gStaticSets->fPropSets[URX_ISWORD_SET].contains(prevChar); |
2615 | 0 | break; |
2616 | 0 | } |
2617 | 0 | } |
2618 | 0 | isBoundary = cIsWord ^ prevCIsWord; |
2619 | 0 | return isBoundary; |
2620 | 0 | } |
2621 | | |
2622 | | //-------------------------------------------------------------------------------- |
2623 | | // |
2624 | | // isUWordBoundary |
2625 | | // |
2626 | | // Test for a word boundary using RBBI word break. |
2627 | | // |
2628 | | // parameters: pos - the current position in the input buffer |
2629 | | // |
2630 | | //-------------------------------------------------------------------------------- |
2631 | 0 | UBool RegexMatcher::isUWordBoundary(int64_t pos, UErrorCode &status) { |
2632 | 0 | UBool returnVal = false; |
2633 | |
|
2634 | 0 | #if UCONFIG_NO_BREAK_ITERATION==0 |
2635 | | // Note: this point will never be reached if break iteration is configured out. |
2636 | | // Regex patterns that would require this function will fail to compile. |
2637 | | |
2638 | | // If we haven't yet created a break iterator for this matcher, do it now. |
2639 | 0 | if (fWordBreakItr == nullptr) { |
2640 | 0 | fWordBreakItr = BreakIterator::createWordInstance(Locale::getEnglish(), status); |
2641 | 0 | if (U_FAILURE(status)) { |
2642 | 0 | return false; |
2643 | 0 | } |
2644 | 0 | fWordBreakItr->setText(fInputText, status); |
2645 | 0 | } |
2646 | | |
2647 | | // Note: zero width boundary tests like \b see through transparent region bounds, |
2648 | | // which is why fLookLimit is used here, rather than fActiveLimit. |
2649 | 0 | if (pos >= fLookLimit) { |
2650 | 0 | fHitEnd = true; |
2651 | 0 | returnVal = true; // With Unicode word rules, only positions within the interior of "real" |
2652 | | // words are not boundaries. All non-word chars stand by themselves, |
2653 | | // with word boundaries on both sides. |
2654 | 0 | } else { |
2655 | 0 | returnVal = fWordBreakItr->isBoundary((int32_t)pos); |
2656 | 0 | } |
2657 | 0 | #endif |
2658 | 0 | return returnVal; |
2659 | 0 | } |
2660 | | |
2661 | | |
2662 | 0 | int64_t RegexMatcher::followingGCBoundary(int64_t pos, UErrorCode &status) { |
2663 | 0 | int64_t result = pos; |
2664 | |
|
2665 | 0 | #if UCONFIG_NO_BREAK_ITERATION==0 |
2666 | | // Note: this point will never be reached if break iteration is configured out. |
2667 | | // Regex patterns that would require this function will fail to compile. |
2668 | | |
2669 | | // If we haven't yet created a break iterator for this matcher, do it now. |
2670 | 0 | if (fGCBreakItr == nullptr) { |
2671 | 0 | fGCBreakItr = BreakIterator::createCharacterInstance(Locale::getEnglish(), status); |
2672 | 0 | if (U_FAILURE(status)) { |
2673 | 0 | return pos; |
2674 | 0 | } |
2675 | 0 | fGCBreakItr->setText(fInputText, status); |
2676 | 0 | } |
2677 | 0 | result = fGCBreakItr->following(pos); |
2678 | 0 | if (result == BreakIterator::DONE) { |
2679 | 0 | result = pos; |
2680 | 0 | } |
2681 | 0 | #endif |
2682 | 0 | return result; |
2683 | 0 | } |
2684 | | |
2685 | | //-------------------------------------------------------------------------------- |
2686 | | // |
2687 | | // IncrementTime This function is called once each TIMER_INITIAL_VALUE state |
2688 | | // saves. Increment the "time" counter, and call the |
2689 | | // user callback function if there is one installed. |
2690 | | // |
2691 | | // If the match operation needs to be aborted, either for a time-out |
2692 | | // or because the user callback asked for it, just set an error status. |
2693 | | // The engine will pick that up and stop in its outer loop. |
2694 | | // |
2695 | | //-------------------------------------------------------------------------------- |
2696 | 0 | void RegexMatcher::IncrementTime(UErrorCode &status) { |
2697 | 0 | fTickCounter = TIMER_INITIAL_VALUE; |
2698 | 0 | fTime++; |
2699 | 0 | if (fCallbackFn != NULL) { |
2700 | 0 | if ((*fCallbackFn)(fCallbackContext, fTime) == false) { |
2701 | 0 | status = U_REGEX_STOPPED_BY_CALLER; |
2702 | 0 | return; |
2703 | 0 | } |
2704 | 0 | } |
2705 | 0 | if (fTimeLimit > 0 && fTime >= fTimeLimit) { |
2706 | 0 | status = U_REGEX_TIME_OUT; |
2707 | 0 | } |
2708 | 0 | } |
2709 | | |
2710 | | //-------------------------------------------------------------------------------- |
2711 | | // |
2712 | | // StateSave |
2713 | | // Make a new stack frame, initialized as a copy of the current stack frame. |
2714 | | // Set the pattern index in the original stack frame from the operand value |
2715 | | // in the opcode. Execution of the engine continues with the state in |
2716 | | // the newly created stack frame |
2717 | | // |
2718 | | // Note that reserveBlock() may grow the stack, resulting in the |
2719 | | // whole thing being relocated in memory. |
2720 | | // |
2721 | | // Parameters: |
2722 | | // fp The top frame pointer when called. At return, a new |
2723 | | // fame will be present |
2724 | | // savePatIdx An index into the compiled pattern. Goes into the original |
2725 | | // (not new) frame. If execution ever back-tracks out of the |
2726 | | // new frame, this will be where we continue from in the pattern. |
2727 | | // Return |
2728 | | // The new frame pointer. |
2729 | | // |
2730 | | //-------------------------------------------------------------------------------- |
2731 | 0 | inline REStackFrame *RegexMatcher::StateSave(REStackFrame *fp, int64_t savePatIdx, UErrorCode &status) { |
2732 | 0 | if (U_FAILURE(status)) { |
2733 | 0 | return fp; |
2734 | 0 | } |
2735 | | // push storage for a new frame. |
2736 | 0 | int64_t *newFP = fStack->reserveBlock(fFrameSize, status); |
2737 | 0 | if (U_FAILURE(status)) { |
2738 | | // Failure on attempted stack expansion. |
2739 | | // Stack function set some other error code, change it to a more |
2740 | | // specific one for regular expressions. |
2741 | 0 | status = U_REGEX_STACK_OVERFLOW; |
2742 | | // We need to return a writable stack frame, so just return the |
2743 | | // previous frame. The match operation will stop quickly |
2744 | | // because of the error status, after which the frame will never |
2745 | | // be looked at again. |
2746 | 0 | return fp; |
2747 | 0 | } |
2748 | 0 | fp = (REStackFrame *)(newFP - fFrameSize); // in case of realloc of stack. |
2749 | | |
2750 | | // New stack frame = copy of old top frame. |
2751 | 0 | int64_t *source = (int64_t *)fp; |
2752 | 0 | int64_t *dest = newFP; |
2753 | 0 | for (;;) { |
2754 | 0 | *dest++ = *source++; |
2755 | 0 | if (source == newFP) { |
2756 | 0 | break; |
2757 | 0 | } |
2758 | 0 | } |
2759 | |
|
2760 | 0 | fTickCounter--; |
2761 | 0 | if (fTickCounter <= 0) { |
2762 | 0 | IncrementTime(status); // Re-initializes fTickCounter |
2763 | 0 | } |
2764 | 0 | fp->fPatIdx = savePatIdx; |
2765 | 0 | return (REStackFrame *)newFP; |
2766 | 0 | } |
2767 | | |
2768 | | #if defined(REGEX_DEBUG) |
2769 | | namespace { |
2770 | | UnicodeString StringFromUText(UText *ut) { |
2771 | | UnicodeString result; |
2772 | | for (UChar32 c = utext_next32From(ut, 0); c != U_SENTINEL; c = UTEXT_NEXT32(ut)) { |
2773 | | result.append(c); |
2774 | | } |
2775 | | return result; |
2776 | | } |
2777 | | } |
2778 | | #endif // REGEX_DEBUG |
2779 | | |
2780 | | |
2781 | | //-------------------------------------------------------------------------------- |
2782 | | // |
2783 | | // MatchAt This is the actual matching engine. |
2784 | | // |
2785 | | // startIdx: begin matching a this index. |
2786 | | // toEnd: if true, match must extend to end of the input region |
2787 | | // |
2788 | | //-------------------------------------------------------------------------------- |
2789 | 0 | void RegexMatcher::MatchAt(int64_t startIdx, UBool toEnd, UErrorCode &status) { |
2790 | 0 | UBool isMatch = false; // True if the we have a match. |
2791 | |
|
2792 | 0 | int64_t backSearchIndex = U_INT64_MAX; // used after greedy single-character matches for searching backwards |
2793 | |
|
2794 | 0 | int32_t op; // Operation from the compiled pattern, split into |
2795 | 0 | int32_t opType; // the opcode |
2796 | 0 | int32_t opValue; // and the operand value. |
2797 | |
|
2798 | | #ifdef REGEX_RUN_DEBUG |
2799 | | if (fTraceDebug) { |
2800 | | printf("MatchAt(startIdx=%ld)\n", startIdx); |
2801 | | printf("Original Pattern: \"%s\"\n", CStr(StringFromUText(fPattern->fPattern))()); |
2802 | | printf("Input String: \"%s\"\n\n", CStr(StringFromUText(fInputText))()); |
2803 | | } |
2804 | | #endif |
2805 | |
|
2806 | 0 | if (U_FAILURE(status)) { |
2807 | 0 | return; |
2808 | 0 | } |
2809 | | |
2810 | | // Cache frequently referenced items from the compiled pattern |
2811 | | // |
2812 | 0 | int64_t *pat = fPattern->fCompiledPat->getBuffer(); |
2813 | |
|
2814 | 0 | const UChar *litText = fPattern->fLiteralText.getBuffer(); |
2815 | 0 | UVector *fSets = fPattern->fSets; |
2816 | |
|
2817 | 0 | fFrameSize = fPattern->fFrameSize; |
2818 | 0 | REStackFrame *fp = resetStack(); |
2819 | 0 | if (U_FAILURE(fDeferredStatus)) { |
2820 | 0 | status = fDeferredStatus; |
2821 | 0 | return; |
2822 | 0 | } |
2823 | | |
2824 | 0 | fp->fPatIdx = 0; |
2825 | 0 | fp->fInputIdx = startIdx; |
2826 | | |
2827 | | // Zero out the pattern's static data |
2828 | 0 | int32_t i; |
2829 | 0 | for (i = 0; i<fPattern->fDataSize; i++) { |
2830 | 0 | fData[i] = 0; |
2831 | 0 | } |
2832 | | |
2833 | | // |
2834 | | // Main loop for interpreting the compiled pattern. |
2835 | | // One iteration of the loop per pattern operation performed. |
2836 | | // |
2837 | 0 | for (;;) { |
2838 | 0 | op = (int32_t)pat[fp->fPatIdx]; |
2839 | 0 | opType = URX_TYPE(op); |
2840 | 0 | opValue = URX_VAL(op); |
2841 | | #ifdef REGEX_RUN_DEBUG |
2842 | | if (fTraceDebug) { |
2843 | | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
2844 | | printf("inputIdx=%ld inputChar=%x sp=%3ld activeLimit=%ld ", fp->fInputIdx, |
2845 | | UTEXT_CURRENT32(fInputText), (int64_t *)fp-fStack->getBuffer(), fActiveLimit); |
2846 | | fPattern->dumpOp(fp->fPatIdx); |
2847 | | } |
2848 | | #endif |
2849 | 0 | fp->fPatIdx++; |
2850 | |
|
2851 | 0 | switch (opType) { |
2852 | | |
2853 | | |
2854 | 0 | case URX_NOP: |
2855 | 0 | break; |
2856 | | |
2857 | | |
2858 | 0 | case URX_BACKTRACK: |
2859 | | // Force a backtrack. In some circumstances, the pattern compiler |
2860 | | // will notice that the pattern can't possibly match anything, and will |
2861 | | // emit one of these at that point. |
2862 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
2863 | 0 | break; |
2864 | | |
2865 | | |
2866 | 0 | case URX_ONECHAR: |
2867 | 0 | if (fp->fInputIdx < fActiveLimit) { |
2868 | 0 | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
2869 | 0 | UChar32 c = UTEXT_NEXT32(fInputText); |
2870 | 0 | if (c == opValue) { |
2871 | 0 | fp->fInputIdx = UTEXT_GETNATIVEINDEX(fInputText); |
2872 | 0 | break; |
2873 | 0 | } |
2874 | 0 | } else { |
2875 | 0 | fHitEnd = true; |
2876 | 0 | } |
2877 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
2878 | 0 | break; |
2879 | | |
2880 | | |
2881 | 0 | case URX_STRING: |
2882 | 0 | { |
2883 | | // Test input against a literal string. |
2884 | | // Strings require two slots in the compiled pattern, one for the |
2885 | | // offset to the string text, and one for the length. |
2886 | |
|
2887 | 0 | int32_t stringStartIdx = opValue; |
2888 | 0 | op = (int32_t)pat[fp->fPatIdx]; // Fetch the second operand |
2889 | 0 | fp->fPatIdx++; |
2890 | 0 | opType = URX_TYPE(op); |
2891 | 0 | int32_t stringLen = URX_VAL(op); |
2892 | 0 | U_ASSERT(opType == URX_STRING_LEN); |
2893 | 0 | U_ASSERT(stringLen >= 2); |
2894 | |
|
2895 | 0 | const UChar *patternString = litText+stringStartIdx; |
2896 | 0 | int32_t patternStringIndex = 0; |
2897 | 0 | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
2898 | 0 | UChar32 inputChar; |
2899 | 0 | UChar32 patternChar; |
2900 | 0 | UBool success = true; |
2901 | 0 | while (patternStringIndex < stringLen) { |
2902 | 0 | if (UTEXT_GETNATIVEINDEX(fInputText) >= fActiveLimit) { |
2903 | 0 | success = false; |
2904 | 0 | fHitEnd = true; |
2905 | 0 | break; |
2906 | 0 | } |
2907 | 0 | inputChar = UTEXT_NEXT32(fInputText); |
2908 | 0 | U16_NEXT(patternString, patternStringIndex, stringLen, patternChar); |
2909 | 0 | if (patternChar != inputChar) { |
2910 | 0 | success = false; |
2911 | 0 | break; |
2912 | 0 | } |
2913 | 0 | } |
2914 | |
|
2915 | 0 | if (success) { |
2916 | 0 | fp->fInputIdx = UTEXT_GETNATIVEINDEX(fInputText); |
2917 | 0 | } else { |
2918 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
2919 | 0 | } |
2920 | 0 | } |
2921 | 0 | break; |
2922 | | |
2923 | | |
2924 | 0 | case URX_STATE_SAVE: |
2925 | 0 | fp = StateSave(fp, opValue, status); |
2926 | 0 | break; |
2927 | | |
2928 | | |
2929 | 0 | case URX_END: |
2930 | | // The match loop will exit via this path on a successful match, |
2931 | | // when we reach the end of the pattern. |
2932 | 0 | if (toEnd && fp->fInputIdx != fActiveLimit) { |
2933 | | // The pattern matched, but not to the end of input. Try some more. |
2934 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
2935 | 0 | break; |
2936 | 0 | } |
2937 | 0 | isMatch = true; |
2938 | 0 | goto breakFromLoop; |
2939 | | |
2940 | | // Start and End Capture stack frame variables are laid out out like this: |
2941 | | // fp->fExtra[opValue] - The start of a completed capture group |
2942 | | // opValue+1 - The end of a completed capture group |
2943 | | // opValue+2 - the start of a capture group whose end |
2944 | | // has not yet been reached (and might not ever be). |
2945 | 0 | case URX_START_CAPTURE: |
2946 | 0 | U_ASSERT(opValue >= 0 && opValue < fFrameSize-3); |
2947 | 0 | fp->fExtra[opValue+2] = fp->fInputIdx; |
2948 | 0 | break; |
2949 | | |
2950 | | |
2951 | 0 | case URX_END_CAPTURE: |
2952 | 0 | U_ASSERT(opValue >= 0 && opValue < fFrameSize-3); |
2953 | 0 | U_ASSERT(fp->fExtra[opValue+2] >= 0); // Start pos for this group must be set. |
2954 | 0 | fp->fExtra[opValue] = fp->fExtra[opValue+2]; // Tentative start becomes real. |
2955 | 0 | fp->fExtra[opValue+1] = fp->fInputIdx; // End position |
2956 | 0 | U_ASSERT(fp->fExtra[opValue] <= fp->fExtra[opValue+1]); |
2957 | 0 | break; |
2958 | | |
2959 | | |
2960 | 0 | case URX_DOLLAR: // $, test for End of line |
2961 | | // or for position before new line at end of input |
2962 | 0 | { |
2963 | 0 | if (fp->fInputIdx >= fAnchorLimit) { |
2964 | | // We really are at the end of input. Success. |
2965 | 0 | fHitEnd = true; |
2966 | 0 | fRequireEnd = true; |
2967 | 0 | break; |
2968 | 0 | } |
2969 | | |
2970 | 0 | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
2971 | | |
2972 | | // If we are positioned just before a new-line that is located at the |
2973 | | // end of input, succeed. |
2974 | 0 | UChar32 c = UTEXT_NEXT32(fInputText); |
2975 | 0 | if (UTEXT_GETNATIVEINDEX(fInputText) >= fAnchorLimit) { |
2976 | 0 | if (isLineTerminator(c)) { |
2977 | | // If not in the middle of a CR/LF sequence |
2978 | 0 | if ( !(c==0x0a && fp->fInputIdx>fAnchorStart && ((void)UTEXT_PREVIOUS32(fInputText), UTEXT_PREVIOUS32(fInputText))==0x0d)) { |
2979 | | // At new-line at end of input. Success |
2980 | 0 | fHitEnd = true; |
2981 | 0 | fRequireEnd = true; |
2982 | |
|
2983 | 0 | break; |
2984 | 0 | } |
2985 | 0 | } |
2986 | 0 | } else { |
2987 | 0 | UChar32 nextC = UTEXT_NEXT32(fInputText); |
2988 | 0 | if (c == 0x0d && nextC == 0x0a && UTEXT_GETNATIVEINDEX(fInputText) >= fAnchorLimit) { |
2989 | 0 | fHitEnd = true; |
2990 | 0 | fRequireEnd = true; |
2991 | 0 | break; // At CR/LF at end of input. Success |
2992 | 0 | } |
2993 | 0 | } |
2994 | | |
2995 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
2996 | 0 | } |
2997 | 0 | break; |
2998 | | |
2999 | | |
3000 | 0 | case URX_DOLLAR_D: // $, test for End of Line, in UNIX_LINES mode. |
3001 | 0 | if (fp->fInputIdx >= fAnchorLimit) { |
3002 | | // Off the end of input. Success. |
3003 | 0 | fHitEnd = true; |
3004 | 0 | fRequireEnd = true; |
3005 | 0 | break; |
3006 | 0 | } else { |
3007 | 0 | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
3008 | 0 | UChar32 c = UTEXT_NEXT32(fInputText); |
3009 | | // Either at the last character of input, or off the end. |
3010 | 0 | if (c == 0x0a && UTEXT_GETNATIVEINDEX(fInputText) == fAnchorLimit) { |
3011 | 0 | fHitEnd = true; |
3012 | 0 | fRequireEnd = true; |
3013 | 0 | break; |
3014 | 0 | } |
3015 | 0 | } |
3016 | | |
3017 | | // Not at end of input. Back-track out. |
3018 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3019 | 0 | break; |
3020 | | |
3021 | | |
3022 | 0 | case URX_DOLLAR_M: // $, test for End of line in multi-line mode |
3023 | 0 | { |
3024 | 0 | if (fp->fInputIdx >= fAnchorLimit) { |
3025 | | // We really are at the end of input. Success. |
3026 | 0 | fHitEnd = true; |
3027 | 0 | fRequireEnd = true; |
3028 | 0 | break; |
3029 | 0 | } |
3030 | | // If we are positioned just before a new-line, succeed. |
3031 | | // It makes no difference where the new-line is within the input. |
3032 | 0 | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
3033 | 0 | UChar32 c = UTEXT_CURRENT32(fInputText); |
3034 | 0 | if (isLineTerminator(c)) { |
3035 | | // At a line end, except for the odd chance of being in the middle of a CR/LF sequence |
3036 | | // In multi-line mode, hitting a new-line just before the end of input does not |
3037 | | // set the hitEnd or requireEnd flags |
3038 | 0 | if ( !(c==0x0a && fp->fInputIdx>fAnchorStart && UTEXT_PREVIOUS32(fInputText)==0x0d)) { |
3039 | 0 | break; |
3040 | 0 | } |
3041 | 0 | } |
3042 | | // not at a new line. Fail. |
3043 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3044 | 0 | } |
3045 | 0 | break; |
3046 | | |
3047 | | |
3048 | 0 | case URX_DOLLAR_MD: // $, test for End of line in multi-line and UNIX_LINES mode |
3049 | 0 | { |
3050 | 0 | if (fp->fInputIdx >= fAnchorLimit) { |
3051 | | // We really are at the end of input. Success. |
3052 | 0 | fHitEnd = true; |
3053 | 0 | fRequireEnd = true; // Java set requireEnd in this case, even though |
3054 | 0 | break; // adding a new-line would not lose the match. |
3055 | 0 | } |
3056 | | // If we are not positioned just before a new-line, the test fails; backtrack out. |
3057 | | // It makes no difference where the new-line is within the input. |
3058 | 0 | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
3059 | 0 | if (UTEXT_CURRENT32(fInputText) != 0x0a) { |
3060 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3061 | 0 | } |
3062 | 0 | } |
3063 | 0 | break; |
3064 | | |
3065 | | |
3066 | 0 | case URX_CARET: // ^, test for start of line |
3067 | 0 | if (fp->fInputIdx != fAnchorStart) { |
3068 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3069 | 0 | } |
3070 | 0 | break; |
3071 | | |
3072 | | |
3073 | 0 | case URX_CARET_M: // ^, test for start of line in mulit-line mode |
3074 | 0 | { |
3075 | 0 | if (fp->fInputIdx == fAnchorStart) { |
3076 | | // We are at the start input. Success. |
3077 | 0 | break; |
3078 | 0 | } |
3079 | | // Check whether character just before the current pos is a new-line |
3080 | | // unless we are at the end of input |
3081 | 0 | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
3082 | 0 | UChar32 c = UTEXT_PREVIOUS32(fInputText); |
3083 | 0 | if ((fp->fInputIdx < fAnchorLimit) && isLineTerminator(c)) { |
3084 | | // It's a new-line. ^ is true. Success. |
3085 | | // TODO: what should be done with positions between a CR and LF? |
3086 | 0 | break; |
3087 | 0 | } |
3088 | | // Not at the start of a line. Fail. |
3089 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3090 | 0 | } |
3091 | 0 | break; |
3092 | | |
3093 | | |
3094 | 0 | case URX_CARET_M_UNIX: // ^, test for start of line in mulit-line + Unix-line mode |
3095 | 0 | { |
3096 | 0 | U_ASSERT(fp->fInputIdx >= fAnchorStart); |
3097 | 0 | if (fp->fInputIdx <= fAnchorStart) { |
3098 | | // We are at the start input. Success. |
3099 | 0 | break; |
3100 | 0 | } |
3101 | | // Check whether character just before the current pos is a new-line |
3102 | 0 | U_ASSERT(fp->fInputIdx <= fAnchorLimit); |
3103 | 0 | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
3104 | 0 | UChar32 c = UTEXT_PREVIOUS32(fInputText); |
3105 | 0 | if (c != 0x0a) { |
3106 | | // Not at the start of a line. Back-track out. |
3107 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3108 | 0 | } |
3109 | 0 | } |
3110 | 0 | break; |
3111 | | |
3112 | 0 | case URX_BACKSLASH_B: // Test for word boundaries |
3113 | 0 | { |
3114 | 0 | UBool success = isWordBoundary(fp->fInputIdx); |
3115 | 0 | success ^= (UBool)(opValue != 0); // flip sense for \B |
3116 | 0 | if (!success) { |
3117 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3118 | 0 | } |
3119 | 0 | } |
3120 | 0 | break; |
3121 | | |
3122 | | |
3123 | 0 | case URX_BACKSLASH_BU: // Test for word boundaries, Unicode-style |
3124 | 0 | { |
3125 | 0 | UBool success = isUWordBoundary(fp->fInputIdx, status); |
3126 | 0 | success ^= (UBool)(opValue != 0); // flip sense for \B |
3127 | 0 | if (!success) { |
3128 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3129 | 0 | } |
3130 | 0 | } |
3131 | 0 | break; |
3132 | | |
3133 | | |
3134 | 0 | case URX_BACKSLASH_D: // Test for decimal digit |
3135 | 0 | { |
3136 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
3137 | 0 | fHitEnd = true; |
3138 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3139 | 0 | break; |
3140 | 0 | } |
3141 | | |
3142 | 0 | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
3143 | |
|
3144 | 0 | UChar32 c = UTEXT_NEXT32(fInputText); |
3145 | 0 | int8_t ctype = u_charType(c); // TODO: make a unicode set for this. Will be faster. |
3146 | 0 | UBool success = (ctype == U_DECIMAL_DIGIT_NUMBER); |
3147 | 0 | success ^= (UBool)(opValue != 0); // flip sense for \D |
3148 | 0 | if (success) { |
3149 | 0 | fp->fInputIdx = UTEXT_GETNATIVEINDEX(fInputText); |
3150 | 0 | } else { |
3151 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3152 | 0 | } |
3153 | 0 | } |
3154 | 0 | break; |
3155 | | |
3156 | | |
3157 | 0 | case URX_BACKSLASH_G: // Test for position at end of previous match |
3158 | 0 | if (!((fMatch && fp->fInputIdx==fMatchEnd) || (fMatch==false && fp->fInputIdx==fActiveStart))) { |
3159 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3160 | 0 | } |
3161 | 0 | break; |
3162 | | |
3163 | | |
3164 | 0 | case URX_BACKSLASH_H: // Test for \h, horizontal white space. |
3165 | 0 | { |
3166 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
3167 | 0 | fHitEnd = true; |
3168 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3169 | 0 | break; |
3170 | 0 | } |
3171 | 0 | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
3172 | 0 | UChar32 c = UTEXT_NEXT32(fInputText); |
3173 | 0 | int8_t ctype = u_charType(c); |
3174 | 0 | UBool success = (ctype == U_SPACE_SEPARATOR || c == 9); // SPACE_SEPARATOR || TAB |
3175 | 0 | success ^= (UBool)(opValue != 0); // flip sense for \H |
3176 | 0 | if (success) { |
3177 | 0 | fp->fInputIdx = UTEXT_GETNATIVEINDEX(fInputText); |
3178 | 0 | } else { |
3179 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3180 | 0 | } |
3181 | 0 | } |
3182 | 0 | break; |
3183 | | |
3184 | | |
3185 | 0 | case URX_BACKSLASH_R: // Test for \R, any line break sequence. |
3186 | 0 | { |
3187 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
3188 | 0 | fHitEnd = true; |
3189 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3190 | 0 | break; |
3191 | 0 | } |
3192 | 0 | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
3193 | 0 | UChar32 c = UTEXT_NEXT32(fInputText); |
3194 | 0 | if (isLineTerminator(c)) { |
3195 | 0 | if (c == 0x0d && utext_current32(fInputText) == 0x0a) { |
3196 | 0 | utext_next32(fInputText); |
3197 | 0 | } |
3198 | 0 | fp->fInputIdx = UTEXT_GETNATIVEINDEX(fInputText); |
3199 | 0 | } else { |
3200 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3201 | 0 | } |
3202 | 0 | } |
3203 | 0 | break; |
3204 | | |
3205 | | |
3206 | 0 | case URX_BACKSLASH_V: // \v, any single line ending character. |
3207 | 0 | { |
3208 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
3209 | 0 | fHitEnd = true; |
3210 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3211 | 0 | break; |
3212 | 0 | } |
3213 | 0 | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
3214 | 0 | UChar32 c = UTEXT_NEXT32(fInputText); |
3215 | 0 | UBool success = isLineTerminator(c); |
3216 | 0 | success ^= (UBool)(opValue != 0); // flip sense for \V |
3217 | 0 | if (success) { |
3218 | 0 | fp->fInputIdx = UTEXT_GETNATIVEINDEX(fInputText); |
3219 | 0 | } else { |
3220 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3221 | 0 | } |
3222 | 0 | } |
3223 | 0 | break; |
3224 | | |
3225 | | |
3226 | 0 | case URX_BACKSLASH_X: |
3227 | | // Match a Grapheme, as defined by Unicode UAX 29. |
3228 | | |
3229 | | // Fail if at end of input |
3230 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
3231 | 0 | fHitEnd = true; |
3232 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3233 | 0 | break; |
3234 | 0 | } |
3235 | | |
3236 | 0 | fp->fInputIdx = followingGCBoundary(fp->fInputIdx, status); |
3237 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
3238 | 0 | fHitEnd = true; |
3239 | 0 | fp->fInputIdx = fActiveLimit; |
3240 | 0 | } |
3241 | 0 | break; |
3242 | | |
3243 | | |
3244 | 0 | case URX_BACKSLASH_Z: // Test for end of Input |
3245 | 0 | if (fp->fInputIdx < fAnchorLimit) { |
3246 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3247 | 0 | } else { |
3248 | 0 | fHitEnd = true; |
3249 | 0 | fRequireEnd = true; |
3250 | 0 | } |
3251 | 0 | break; |
3252 | | |
3253 | | |
3254 | | |
3255 | 0 | case URX_STATIC_SETREF: |
3256 | 0 | { |
3257 | | // Test input character against one of the predefined sets |
3258 | | // (Word Characters, for example) |
3259 | | // The high bit of the op value is a flag for the match polarity. |
3260 | | // 0: success if input char is in set. |
3261 | | // 1: success if input char is not in set. |
3262 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
3263 | 0 | fHitEnd = true; |
3264 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3265 | 0 | break; |
3266 | 0 | } |
3267 | | |
3268 | 0 | UBool success = ((opValue & URX_NEG_SET) == URX_NEG_SET); |
3269 | 0 | opValue &= ~URX_NEG_SET; |
3270 | 0 | U_ASSERT(opValue > 0 && opValue < URX_LAST_SET); |
3271 | |
|
3272 | 0 | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
3273 | 0 | UChar32 c = UTEXT_NEXT32(fInputText); |
3274 | 0 | if (c < 256) { |
3275 | 0 | Regex8BitSet &s8 = RegexStaticSets::gStaticSets->fPropSets8[opValue]; |
3276 | 0 | if (s8.contains(c)) { |
3277 | 0 | success = !success; |
3278 | 0 | } |
3279 | 0 | } else { |
3280 | 0 | const UnicodeSet &s = RegexStaticSets::gStaticSets->fPropSets[opValue]; |
3281 | 0 | if (s.contains(c)) { |
3282 | 0 | success = !success; |
3283 | 0 | } |
3284 | 0 | } |
3285 | 0 | if (success) { |
3286 | 0 | fp->fInputIdx = UTEXT_GETNATIVEINDEX(fInputText); |
3287 | 0 | } else { |
3288 | | // the character wasn't in the set. |
3289 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3290 | 0 | } |
3291 | 0 | } |
3292 | 0 | break; |
3293 | | |
3294 | | |
3295 | 0 | case URX_STAT_SETREF_N: |
3296 | 0 | { |
3297 | | // Test input character for NOT being a member of one of |
3298 | | // the predefined sets (Word Characters, for example) |
3299 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
3300 | 0 | fHitEnd = true; |
3301 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3302 | 0 | break; |
3303 | 0 | } |
3304 | | |
3305 | 0 | U_ASSERT(opValue > 0 && opValue < URX_LAST_SET); |
3306 | |
|
3307 | 0 | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
3308 | |
|
3309 | 0 | UChar32 c = UTEXT_NEXT32(fInputText); |
3310 | 0 | if (c < 256) { |
3311 | 0 | Regex8BitSet &s8 = RegexStaticSets::gStaticSets->fPropSets8[opValue]; |
3312 | 0 | if (s8.contains(c) == false) { |
3313 | 0 | fp->fInputIdx = UTEXT_GETNATIVEINDEX(fInputText); |
3314 | 0 | break; |
3315 | 0 | } |
3316 | 0 | } else { |
3317 | 0 | const UnicodeSet &s = RegexStaticSets::gStaticSets->fPropSets[opValue]; |
3318 | 0 | if (s.contains(c) == false) { |
3319 | 0 | fp->fInputIdx = UTEXT_GETNATIVEINDEX(fInputText); |
3320 | 0 | break; |
3321 | 0 | } |
3322 | 0 | } |
3323 | | // the character wasn't in the set. |
3324 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3325 | 0 | } |
3326 | 0 | break; |
3327 | | |
3328 | | |
3329 | 0 | case URX_SETREF: |
3330 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
3331 | 0 | fHitEnd = true; |
3332 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3333 | 0 | break; |
3334 | 0 | } else { |
3335 | 0 | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
3336 | | |
3337 | | // There is input left. Pick up one char and test it for set membership. |
3338 | 0 | UChar32 c = UTEXT_NEXT32(fInputText); |
3339 | 0 | U_ASSERT(opValue > 0 && opValue < fSets->size()); |
3340 | 0 | if (c<256) { |
3341 | 0 | Regex8BitSet *s8 = &fPattern->fSets8[opValue]; |
3342 | 0 | if (s8->contains(c)) { |
3343 | 0 | fp->fInputIdx = UTEXT_GETNATIVEINDEX(fInputText); |
3344 | 0 | break; |
3345 | 0 | } |
3346 | 0 | } else { |
3347 | 0 | UnicodeSet *s = (UnicodeSet *)fSets->elementAt(opValue); |
3348 | 0 | if (s->contains(c)) { |
3349 | | // The character is in the set. A Match. |
3350 | 0 | fp->fInputIdx = UTEXT_GETNATIVEINDEX(fInputText); |
3351 | 0 | break; |
3352 | 0 | } |
3353 | 0 | } |
3354 | | |
3355 | | // the character wasn't in the set. |
3356 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3357 | 0 | } |
3358 | 0 | break; |
3359 | | |
3360 | | |
3361 | 0 | case URX_DOTANY: |
3362 | 0 | { |
3363 | | // . matches anything, but stops at end-of-line. |
3364 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
3365 | | // At end of input. Match failed. Backtrack out. |
3366 | 0 | fHitEnd = true; |
3367 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3368 | 0 | break; |
3369 | 0 | } |
3370 | | |
3371 | 0 | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
3372 | | |
3373 | | // There is input left. Advance over one char, unless we've hit end-of-line |
3374 | 0 | UChar32 c = UTEXT_NEXT32(fInputText); |
3375 | 0 | if (isLineTerminator(c)) { |
3376 | | // End of line in normal mode. . does not match. |
3377 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3378 | 0 | break; |
3379 | 0 | } |
3380 | 0 | fp->fInputIdx = UTEXT_GETNATIVEINDEX(fInputText); |
3381 | 0 | } |
3382 | 0 | break; |
3383 | | |
3384 | | |
3385 | 0 | case URX_DOTANY_ALL: |
3386 | 0 | { |
3387 | | // ., in dot-matches-all (including new lines) mode |
3388 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
3389 | | // At end of input. Match failed. Backtrack out. |
3390 | 0 | fHitEnd = true; |
3391 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3392 | 0 | break; |
3393 | 0 | } |
3394 | | |
3395 | 0 | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
3396 | | |
3397 | | // There is input left. Advance over one char, except if we are |
3398 | | // at a cr/lf, advance over both of them. |
3399 | 0 | UChar32 c; |
3400 | 0 | c = UTEXT_NEXT32(fInputText); |
3401 | 0 | fp->fInputIdx = UTEXT_GETNATIVEINDEX(fInputText); |
3402 | 0 | if (c==0x0d && fp->fInputIdx < fActiveLimit) { |
3403 | | // In the case of a CR/LF, we need to advance over both. |
3404 | 0 | UChar32 nextc = UTEXT_CURRENT32(fInputText); |
3405 | 0 | if (nextc == 0x0a) { |
3406 | 0 | (void)UTEXT_NEXT32(fInputText); |
3407 | 0 | fp->fInputIdx = UTEXT_GETNATIVEINDEX(fInputText); |
3408 | 0 | } |
3409 | 0 | } |
3410 | 0 | } |
3411 | 0 | break; |
3412 | | |
3413 | | |
3414 | 0 | case URX_DOTANY_UNIX: |
3415 | 0 | { |
3416 | | // '.' operator, matches all, but stops at end-of-line. |
3417 | | // UNIX_LINES mode, so 0x0a is the only recognized line ending. |
3418 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
3419 | | // At end of input. Match failed. Backtrack out. |
3420 | 0 | fHitEnd = true; |
3421 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3422 | 0 | break; |
3423 | 0 | } |
3424 | | |
3425 | 0 | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
3426 | | |
3427 | | // There is input left. Advance over one char, unless we've hit end-of-line |
3428 | 0 | UChar32 c = UTEXT_NEXT32(fInputText); |
3429 | 0 | if (c == 0x0a) { |
3430 | | // End of line in normal mode. '.' does not match the \n |
3431 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3432 | 0 | } else { |
3433 | 0 | fp->fInputIdx = UTEXT_GETNATIVEINDEX(fInputText); |
3434 | 0 | } |
3435 | 0 | } |
3436 | 0 | break; |
3437 | | |
3438 | | |
3439 | 0 | case URX_JMP: |
3440 | 0 | fp->fPatIdx = opValue; |
3441 | 0 | break; |
3442 | | |
3443 | 0 | case URX_FAIL: |
3444 | 0 | isMatch = false; |
3445 | 0 | goto breakFromLoop; |
3446 | | |
3447 | 0 | case URX_JMP_SAV: |
3448 | 0 | U_ASSERT(opValue < fPattern->fCompiledPat->size()); |
3449 | 0 | fp = StateSave(fp, fp->fPatIdx, status); // State save to loc following current |
3450 | 0 | fp->fPatIdx = opValue; // Then JMP. |
3451 | 0 | break; |
3452 | | |
3453 | 0 | case URX_JMP_SAV_X: |
3454 | | // This opcode is used with (x)+, when x can match a zero length string. |
3455 | | // Same as JMP_SAV, except conditional on the match having made forward progress. |
3456 | | // Destination of the JMP must be a URX_STO_INP_LOC, from which we get the |
3457 | | // data address of the input position at the start of the loop. |
3458 | 0 | { |
3459 | 0 | U_ASSERT(opValue > 0 && opValue < fPattern->fCompiledPat->size()); |
3460 | 0 | int32_t stoOp = (int32_t)pat[opValue-1]; |
3461 | 0 | U_ASSERT(URX_TYPE(stoOp) == URX_STO_INP_LOC); |
3462 | 0 | int32_t frameLoc = URX_VAL(stoOp); |
3463 | 0 | U_ASSERT(frameLoc >= 0 && frameLoc < fFrameSize); |
3464 | 0 | int64_t prevInputIdx = fp->fExtra[frameLoc]; |
3465 | 0 | U_ASSERT(prevInputIdx <= fp->fInputIdx); |
3466 | 0 | if (prevInputIdx < fp->fInputIdx) { |
3467 | | // The match did make progress. Repeat the loop. |
3468 | 0 | fp = StateSave(fp, fp->fPatIdx, status); // State save to loc following current |
3469 | 0 | fp->fPatIdx = opValue; |
3470 | 0 | fp->fExtra[frameLoc] = fp->fInputIdx; |
3471 | 0 | } |
3472 | | // If the input position did not advance, we do nothing here, |
3473 | | // execution will fall out of the loop. |
3474 | 0 | } |
3475 | 0 | break; |
3476 | | |
3477 | 0 | case URX_CTR_INIT: |
3478 | 0 | { |
3479 | 0 | U_ASSERT(opValue >= 0 && opValue < fFrameSize-2); |
3480 | 0 | fp->fExtra[opValue] = 0; // Set the loop counter variable to zero |
3481 | | |
3482 | | // Pick up the three extra operands that CTR_INIT has, and |
3483 | | // skip the pattern location counter past |
3484 | 0 | int32_t instrOperandLoc = (int32_t)fp->fPatIdx; |
3485 | 0 | fp->fPatIdx += 3; |
3486 | 0 | int32_t loopLoc = URX_VAL(pat[instrOperandLoc]); |
3487 | 0 | int32_t minCount = (int32_t)pat[instrOperandLoc+1]; |
3488 | 0 | int32_t maxCount = (int32_t)pat[instrOperandLoc+2]; |
3489 | 0 | U_ASSERT(minCount>=0); |
3490 | 0 | U_ASSERT(maxCount>=minCount || maxCount==-1); |
3491 | 0 | U_ASSERT(loopLoc>=fp->fPatIdx); |
3492 | |
|
3493 | 0 | if (minCount == 0) { |
3494 | 0 | fp = StateSave(fp, loopLoc+1, status); |
3495 | 0 | } |
3496 | 0 | if (maxCount == -1) { |
3497 | 0 | fp->fExtra[opValue+1] = fp->fInputIdx; // For loop breaking. |
3498 | 0 | } else if (maxCount == 0) { |
3499 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3500 | 0 | } |
3501 | 0 | } |
3502 | 0 | break; |
3503 | | |
3504 | 0 | case URX_CTR_LOOP: |
3505 | 0 | { |
3506 | 0 | U_ASSERT(opValue>0 && opValue < fp->fPatIdx-2); |
3507 | 0 | int32_t initOp = (int32_t)pat[opValue]; |
3508 | 0 | U_ASSERT(URX_TYPE(initOp) == URX_CTR_INIT); |
3509 | 0 | int64_t *pCounter = &fp->fExtra[URX_VAL(initOp)]; |
3510 | 0 | int32_t minCount = (int32_t)pat[opValue+2]; |
3511 | 0 | int32_t maxCount = (int32_t)pat[opValue+3]; |
3512 | 0 | (*pCounter)++; |
3513 | 0 | if ((uint64_t)*pCounter >= (uint32_t)maxCount && maxCount != -1) { |
3514 | 0 | U_ASSERT(*pCounter == maxCount); |
3515 | 0 | break; |
3516 | 0 | } |
3517 | 0 | if (*pCounter >= minCount) { |
3518 | 0 | if (maxCount == -1) { |
3519 | | // Loop has no hard upper bound. |
3520 | | // Check that it is progressing through the input, break if it is not. |
3521 | 0 | int64_t *pLastInputIdx = &fp->fExtra[URX_VAL(initOp) + 1]; |
3522 | 0 | if (fp->fInputIdx == *pLastInputIdx) { |
3523 | 0 | break; |
3524 | 0 | } else { |
3525 | 0 | *pLastInputIdx = fp->fInputIdx; |
3526 | 0 | } |
3527 | 0 | } |
3528 | 0 | fp = StateSave(fp, fp->fPatIdx, status); |
3529 | 0 | } else { |
3530 | | // Increment time-out counter. (StateSave() does it if count >= minCount) |
3531 | 0 | fTickCounter--; |
3532 | 0 | if (fTickCounter <= 0) { |
3533 | 0 | IncrementTime(status); // Re-initializes fTickCounter |
3534 | 0 | } |
3535 | 0 | } |
3536 | | |
3537 | 0 | fp->fPatIdx = opValue + 4; // Loop back. |
3538 | 0 | } |
3539 | 0 | break; |
3540 | | |
3541 | 0 | case URX_CTR_INIT_NG: |
3542 | 0 | { |
3543 | | // Initialize a non-greedy loop |
3544 | 0 | U_ASSERT(opValue >= 0 && opValue < fFrameSize-2); |
3545 | 0 | fp->fExtra[opValue] = 0; // Set the loop counter variable to zero |
3546 | | |
3547 | | // Pick up the three extra operands that CTR_INIT_NG has, and |
3548 | | // skip the pattern location counter past |
3549 | 0 | int32_t instrOperandLoc = (int32_t)fp->fPatIdx; |
3550 | 0 | fp->fPatIdx += 3; |
3551 | 0 | int32_t loopLoc = URX_VAL(pat[instrOperandLoc]); |
3552 | 0 | int32_t minCount = (int32_t)pat[instrOperandLoc+1]; |
3553 | 0 | int32_t maxCount = (int32_t)pat[instrOperandLoc+2]; |
3554 | 0 | U_ASSERT(minCount>=0); |
3555 | 0 | U_ASSERT(maxCount>=minCount || maxCount==-1); |
3556 | 0 | U_ASSERT(loopLoc>fp->fPatIdx); |
3557 | 0 | if (maxCount == -1) { |
3558 | 0 | fp->fExtra[opValue+1] = fp->fInputIdx; // Save initial input index for loop breaking. |
3559 | 0 | } |
3560 | |
|
3561 | 0 | if (minCount == 0) { |
3562 | 0 | if (maxCount != 0) { |
3563 | 0 | fp = StateSave(fp, fp->fPatIdx, status); |
3564 | 0 | } |
3565 | 0 | fp->fPatIdx = loopLoc+1; // Continue with stuff after repeated block |
3566 | 0 | } |
3567 | 0 | } |
3568 | 0 | break; |
3569 | | |
3570 | 0 | case URX_CTR_LOOP_NG: |
3571 | 0 | { |
3572 | | // Non-greedy {min, max} loops |
3573 | 0 | U_ASSERT(opValue>0 && opValue < fp->fPatIdx-2); |
3574 | 0 | int32_t initOp = (int32_t)pat[opValue]; |
3575 | 0 | U_ASSERT(URX_TYPE(initOp) == URX_CTR_INIT_NG); |
3576 | 0 | int64_t *pCounter = &fp->fExtra[URX_VAL(initOp)]; |
3577 | 0 | int32_t minCount = (int32_t)pat[opValue+2]; |
3578 | 0 | int32_t maxCount = (int32_t)pat[opValue+3]; |
3579 | |
|
3580 | 0 | (*pCounter)++; |
3581 | 0 | if ((uint64_t)*pCounter >= (uint32_t)maxCount && maxCount != -1) { |
3582 | | // The loop has matched the maximum permitted number of times. |
3583 | | // Break out of here with no action. Matching will |
3584 | | // continue with the following pattern. |
3585 | 0 | U_ASSERT(*pCounter == maxCount); |
3586 | 0 | break; |
3587 | 0 | } |
3588 | | |
3589 | 0 | if (*pCounter < minCount) { |
3590 | | // We haven't met the minimum number of matches yet. |
3591 | | // Loop back for another one. |
3592 | 0 | fp->fPatIdx = opValue + 4; // Loop back. |
3593 | | // Increment time-out counter. (StateSave() does it if count >= minCount) |
3594 | 0 | fTickCounter--; |
3595 | 0 | if (fTickCounter <= 0) { |
3596 | 0 | IncrementTime(status); // Re-initializes fTickCounter |
3597 | 0 | } |
3598 | 0 | } else { |
3599 | | // We do have the minimum number of matches. |
3600 | | |
3601 | | // If there is no upper bound on the loop iterations, check that the input index |
3602 | | // is progressing, and stop the loop if it is not. |
3603 | 0 | if (maxCount == -1) { |
3604 | 0 | int64_t *pLastInputIdx = &fp->fExtra[URX_VAL(initOp) + 1]; |
3605 | 0 | if (fp->fInputIdx == *pLastInputIdx) { |
3606 | 0 | break; |
3607 | 0 | } |
3608 | 0 | *pLastInputIdx = fp->fInputIdx; |
3609 | 0 | } |
3610 | | |
3611 | | // Loop Continuation: we will fall into the pattern following the loop |
3612 | | // (non-greedy, don't execute loop body first), but first do |
3613 | | // a state save to the top of the loop, so that a match failure |
3614 | | // in the following pattern will try another iteration of the loop. |
3615 | 0 | fp = StateSave(fp, opValue + 4, status); |
3616 | 0 | } |
3617 | 0 | } |
3618 | 0 | break; |
3619 | | |
3620 | 0 | case URX_STO_SP: |
3621 | 0 | U_ASSERT(opValue >= 0 && opValue < fPattern->fDataSize); |
3622 | 0 | fData[opValue] = fStack->size(); |
3623 | 0 | break; |
3624 | | |
3625 | 0 | case URX_LD_SP: |
3626 | 0 | { |
3627 | 0 | U_ASSERT(opValue >= 0 && opValue < fPattern->fDataSize); |
3628 | 0 | int32_t newStackSize = (int32_t)fData[opValue]; |
3629 | 0 | U_ASSERT(newStackSize <= fStack->size()); |
3630 | 0 | int64_t *newFP = fStack->getBuffer() + newStackSize - fFrameSize; |
3631 | 0 | if (newFP == (int64_t *)fp) { |
3632 | 0 | break; |
3633 | 0 | } |
3634 | 0 | int32_t j; |
3635 | 0 | for (j=0; j<fFrameSize; j++) { |
3636 | 0 | newFP[j] = ((int64_t *)fp)[j]; |
3637 | 0 | } |
3638 | 0 | fp = (REStackFrame *)newFP; |
3639 | 0 | fStack->setSize(newStackSize); |
3640 | 0 | } |
3641 | 0 | break; |
3642 | | |
3643 | 0 | case URX_BACKREF: |
3644 | 0 | { |
3645 | 0 | U_ASSERT(opValue < fFrameSize); |
3646 | 0 | int64_t groupStartIdx = fp->fExtra[opValue]; |
3647 | 0 | int64_t groupEndIdx = fp->fExtra[opValue+1]; |
3648 | 0 | U_ASSERT(groupStartIdx <= groupEndIdx); |
3649 | 0 | if (groupStartIdx < 0) { |
3650 | | // This capture group has not participated in the match thus far, |
3651 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); // FAIL, no match. |
3652 | 0 | break; |
3653 | 0 | } |
3654 | 0 | UTEXT_SETNATIVEINDEX(fAltInputText, groupStartIdx); |
3655 | 0 | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
3656 | | |
3657 | | // Note: if the capture group match was of an empty string the backref |
3658 | | // match succeeds. Verified by testing: Perl matches succeed |
3659 | | // in this case, so we do too. |
3660 | |
|
3661 | 0 | UBool success = true; |
3662 | 0 | for (;;) { |
3663 | 0 | if (utext_getNativeIndex(fAltInputText) >= groupEndIdx) { |
3664 | 0 | success = true; |
3665 | 0 | break; |
3666 | 0 | } |
3667 | 0 | if (utext_getNativeIndex(fInputText) >= fActiveLimit) { |
3668 | 0 | success = false; |
3669 | 0 | fHitEnd = true; |
3670 | 0 | break; |
3671 | 0 | } |
3672 | 0 | UChar32 captureGroupChar = utext_next32(fAltInputText); |
3673 | 0 | UChar32 inputChar = utext_next32(fInputText); |
3674 | 0 | if (inputChar != captureGroupChar) { |
3675 | 0 | success = false; |
3676 | 0 | break; |
3677 | 0 | } |
3678 | 0 | } |
3679 | |
|
3680 | 0 | if (success) { |
3681 | 0 | fp->fInputIdx = UTEXT_GETNATIVEINDEX(fInputText); |
3682 | 0 | } else { |
3683 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3684 | 0 | } |
3685 | 0 | } |
3686 | 0 | break; |
3687 | | |
3688 | | |
3689 | | |
3690 | 0 | case URX_BACKREF_I: |
3691 | 0 | { |
3692 | 0 | U_ASSERT(opValue < fFrameSize); |
3693 | 0 | int64_t groupStartIdx = fp->fExtra[opValue]; |
3694 | 0 | int64_t groupEndIdx = fp->fExtra[opValue+1]; |
3695 | 0 | U_ASSERT(groupStartIdx <= groupEndIdx); |
3696 | 0 | if (groupStartIdx < 0) { |
3697 | | // This capture group has not participated in the match thus far, |
3698 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); // FAIL, no match. |
3699 | 0 | break; |
3700 | 0 | } |
3701 | 0 | utext_setNativeIndex(fAltInputText, groupStartIdx); |
3702 | 0 | utext_setNativeIndex(fInputText, fp->fInputIdx); |
3703 | 0 | CaseFoldingUTextIterator captureGroupItr(*fAltInputText); |
3704 | 0 | CaseFoldingUTextIterator inputItr(*fInputText); |
3705 | | |
3706 | | // Note: if the capture group match was of an empty string the backref |
3707 | | // match succeeds. Verified by testing: Perl matches succeed |
3708 | | // in this case, so we do too. |
3709 | |
|
3710 | 0 | UBool success = true; |
3711 | 0 | for (;;) { |
3712 | 0 | if (!captureGroupItr.inExpansion() && utext_getNativeIndex(fAltInputText) >= groupEndIdx) { |
3713 | 0 | success = true; |
3714 | 0 | break; |
3715 | 0 | } |
3716 | 0 | if (!inputItr.inExpansion() && utext_getNativeIndex(fInputText) >= fActiveLimit) { |
3717 | 0 | success = false; |
3718 | 0 | fHitEnd = true; |
3719 | 0 | break; |
3720 | 0 | } |
3721 | 0 | UChar32 captureGroupChar = captureGroupItr.next(); |
3722 | 0 | UChar32 inputChar = inputItr.next(); |
3723 | 0 | if (inputChar != captureGroupChar) { |
3724 | 0 | success = false; |
3725 | 0 | break; |
3726 | 0 | } |
3727 | 0 | } |
3728 | |
|
3729 | 0 | if (success && inputItr.inExpansion()) { |
3730 | | // We obtained a match by consuming part of a string obtained from |
3731 | | // case-folding a single code point of the input text. |
3732 | | // This does not count as an overall match. |
3733 | 0 | success = false; |
3734 | 0 | } |
3735 | |
|
3736 | 0 | if (success) { |
3737 | 0 | fp->fInputIdx = UTEXT_GETNATIVEINDEX(fInputText); |
3738 | 0 | } else { |
3739 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3740 | 0 | } |
3741 | |
|
3742 | 0 | } |
3743 | 0 | break; |
3744 | | |
3745 | 0 | case URX_STO_INP_LOC: |
3746 | 0 | { |
3747 | 0 | U_ASSERT(opValue >= 0 && opValue < fFrameSize); |
3748 | 0 | fp->fExtra[opValue] = fp->fInputIdx; |
3749 | 0 | } |
3750 | 0 | break; |
3751 | | |
3752 | 0 | case URX_JMPX: |
3753 | 0 | { |
3754 | 0 | int32_t instrOperandLoc = (int32_t)fp->fPatIdx; |
3755 | 0 | fp->fPatIdx += 1; |
3756 | 0 | int32_t dataLoc = URX_VAL(pat[instrOperandLoc]); |
3757 | 0 | U_ASSERT(dataLoc >= 0 && dataLoc < fFrameSize); |
3758 | 0 | int64_t savedInputIdx = fp->fExtra[dataLoc]; |
3759 | 0 | U_ASSERT(savedInputIdx <= fp->fInputIdx); |
3760 | 0 | if (savedInputIdx < fp->fInputIdx) { |
3761 | 0 | fp->fPatIdx = opValue; // JMP |
3762 | 0 | } else { |
3763 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); // FAIL, no progress in loop. |
3764 | 0 | } |
3765 | 0 | } |
3766 | 0 | break; |
3767 | | |
3768 | 0 | case URX_LA_START: |
3769 | 0 | { |
3770 | | // Entering a look around block. |
3771 | | // Save Stack Ptr, Input Pos. |
3772 | 0 | U_ASSERT(opValue>=0 && opValue+3<fPattern->fDataSize); |
3773 | 0 | fData[opValue] = fStack->size(); |
3774 | 0 | fData[opValue+1] = fp->fInputIdx; |
3775 | 0 | fData[opValue+2] = fActiveStart; |
3776 | 0 | fData[opValue+3] = fActiveLimit; |
3777 | 0 | fActiveStart = fLookStart; // Set the match region change for |
3778 | 0 | fActiveLimit = fLookLimit; // transparent bounds. |
3779 | 0 | } |
3780 | 0 | break; |
3781 | | |
3782 | 0 | case URX_LA_END: |
3783 | 0 | { |
3784 | | // Leaving a look-ahead block. |
3785 | | // restore Stack Ptr, Input Pos to positions they had on entry to block. |
3786 | 0 | U_ASSERT(opValue>=0 && opValue+3<fPattern->fDataSize); |
3787 | 0 | int32_t stackSize = fStack->size(); |
3788 | 0 | int32_t newStackSize =(int32_t)fData[opValue]; |
3789 | 0 | U_ASSERT(stackSize >= newStackSize); |
3790 | 0 | if (stackSize > newStackSize) { |
3791 | | // Copy the current top frame back to the new (cut back) top frame. |
3792 | | // This makes the capture groups from within the look-ahead |
3793 | | // expression available. |
3794 | 0 | int64_t *newFP = fStack->getBuffer() + newStackSize - fFrameSize; |
3795 | 0 | int32_t j; |
3796 | 0 | for (j=0; j<fFrameSize; j++) { |
3797 | 0 | newFP[j] = ((int64_t *)fp)[j]; |
3798 | 0 | } |
3799 | 0 | fp = (REStackFrame *)newFP; |
3800 | 0 | fStack->setSize(newStackSize); |
3801 | 0 | } |
3802 | 0 | fp->fInputIdx = fData[opValue+1]; |
3803 | | |
3804 | | // Restore the active region bounds in the input string; they may have |
3805 | | // been changed because of transparent bounds on a Region. |
3806 | 0 | fActiveStart = fData[opValue+2]; |
3807 | 0 | fActiveLimit = fData[opValue+3]; |
3808 | 0 | U_ASSERT(fActiveStart >= 0); |
3809 | 0 | U_ASSERT(fActiveLimit <= fInputLength); |
3810 | 0 | } |
3811 | 0 | break; |
3812 | | |
3813 | 0 | case URX_ONECHAR_I: |
3814 | | // Case insensitive one char. The char from the pattern is already case folded. |
3815 | | // Input text is not, but case folding the input can not reduce two or more code |
3816 | | // points to one. |
3817 | 0 | if (fp->fInputIdx < fActiveLimit) { |
3818 | 0 | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
3819 | |
|
3820 | 0 | UChar32 c = UTEXT_NEXT32(fInputText); |
3821 | 0 | if (u_foldCase(c, U_FOLD_CASE_DEFAULT) == opValue) { |
3822 | 0 | fp->fInputIdx = UTEXT_GETNATIVEINDEX(fInputText); |
3823 | 0 | break; |
3824 | 0 | } |
3825 | 0 | } else { |
3826 | 0 | fHitEnd = true; |
3827 | 0 | } |
3828 | | |
3829 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3830 | 0 | break; |
3831 | | |
3832 | 0 | case URX_STRING_I: |
3833 | 0 | { |
3834 | | // Case-insensitive test input against a literal string. |
3835 | | // Strings require two slots in the compiled pattern, one for the |
3836 | | // offset to the string text, and one for the length. |
3837 | | // The compiled string has already been case folded. |
3838 | 0 | { |
3839 | 0 | const UChar *patternString = litText + opValue; |
3840 | 0 | int32_t patternStringIdx = 0; |
3841 | |
|
3842 | 0 | op = (int32_t)pat[fp->fPatIdx]; |
3843 | 0 | fp->fPatIdx++; |
3844 | 0 | opType = URX_TYPE(op); |
3845 | 0 | opValue = URX_VAL(op); |
3846 | 0 | U_ASSERT(opType == URX_STRING_LEN); |
3847 | 0 | int32_t patternStringLen = opValue; // Length of the string from the pattern. |
3848 | | |
3849 | |
|
3850 | 0 | UChar32 cPattern; |
3851 | 0 | UChar32 cText; |
3852 | 0 | UBool success = true; |
3853 | |
|
3854 | 0 | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
3855 | 0 | CaseFoldingUTextIterator inputIterator(*fInputText); |
3856 | 0 | while (patternStringIdx < patternStringLen) { |
3857 | 0 | if (!inputIterator.inExpansion() && UTEXT_GETNATIVEINDEX(fInputText) >= fActiveLimit) { |
3858 | 0 | success = false; |
3859 | 0 | fHitEnd = true; |
3860 | 0 | break; |
3861 | 0 | } |
3862 | 0 | U16_NEXT(patternString, patternStringIdx, patternStringLen, cPattern); |
3863 | 0 | cText = inputIterator.next(); |
3864 | 0 | if (cText != cPattern) { |
3865 | 0 | success = false; |
3866 | 0 | break; |
3867 | 0 | } |
3868 | 0 | } |
3869 | 0 | if (inputIterator.inExpansion()) { |
3870 | 0 | success = false; |
3871 | 0 | } |
3872 | |
|
3873 | 0 | if (success) { |
3874 | 0 | fp->fInputIdx = UTEXT_GETNATIVEINDEX(fInputText); |
3875 | 0 | } else { |
3876 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3877 | 0 | } |
3878 | 0 | } |
3879 | 0 | } |
3880 | 0 | break; |
3881 | | |
3882 | 0 | case URX_LB_START: |
3883 | 0 | { |
3884 | | // Entering a look-behind block. |
3885 | | // Save Stack Ptr, Input Pos and active input region. |
3886 | | // TODO: implement transparent bounds. Ticket #6067 |
3887 | 0 | U_ASSERT(opValue>=0 && opValue+4<fPattern->fDataSize); |
3888 | 0 | fData[opValue] = fStack->size(); |
3889 | 0 | fData[opValue+1] = fp->fInputIdx; |
3890 | | // Save input string length, then reset to pin any matches to end at |
3891 | | // the current position. |
3892 | 0 | fData[opValue+2] = fActiveStart; |
3893 | 0 | fData[opValue+3] = fActiveLimit; |
3894 | 0 | fActiveStart = fRegionStart; |
3895 | 0 | fActiveLimit = fp->fInputIdx; |
3896 | | // Init the variable containing the start index for attempted matches. |
3897 | 0 | fData[opValue+4] = -1; |
3898 | 0 | } |
3899 | 0 | break; |
3900 | | |
3901 | | |
3902 | 0 | case URX_LB_CONT: |
3903 | 0 | { |
3904 | | // Positive Look-Behind, at top of loop checking for matches of LB expression |
3905 | | // at all possible input starting positions. |
3906 | | |
3907 | | // Fetch the min and max possible match lengths. They are the operands |
3908 | | // of this op in the pattern. |
3909 | 0 | int32_t minML = (int32_t)pat[fp->fPatIdx++]; |
3910 | 0 | int32_t maxML = (int32_t)pat[fp->fPatIdx++]; |
3911 | 0 | if (!UTEXT_USES_U16(fInputText)) { |
3912 | | // utf-8 fix to maximum match length. The pattern compiler assumes utf-16. |
3913 | | // The max length need not be exact; it just needs to be >= actual maximum. |
3914 | 0 | maxML *= 3; |
3915 | 0 | } |
3916 | 0 | U_ASSERT(minML <= maxML); |
3917 | 0 | U_ASSERT(minML >= 0); |
3918 | | |
3919 | | // Fetch (from data) the last input index where a match was attempted. |
3920 | 0 | U_ASSERT(opValue>=0 && opValue+4<fPattern->fDataSize); |
3921 | 0 | int64_t &lbStartIdx = fData[opValue+4]; |
3922 | 0 | if (lbStartIdx < 0) { |
3923 | | // First time through loop. |
3924 | 0 | lbStartIdx = fp->fInputIdx - minML; |
3925 | 0 | if (lbStartIdx > 0) { |
3926 | | // move index to a code point boundary, if it's not on one already. |
3927 | 0 | UTEXT_SETNATIVEINDEX(fInputText, lbStartIdx); |
3928 | 0 | lbStartIdx = UTEXT_GETNATIVEINDEX(fInputText); |
3929 | 0 | } |
3930 | 0 | } else { |
3931 | | // 2nd through nth time through the loop. |
3932 | | // Back up start position for match by one. |
3933 | 0 | if (lbStartIdx == 0) { |
3934 | 0 | (lbStartIdx)--; |
3935 | 0 | } else { |
3936 | 0 | UTEXT_SETNATIVEINDEX(fInputText, lbStartIdx); |
3937 | 0 | (void)UTEXT_PREVIOUS32(fInputText); |
3938 | 0 | lbStartIdx = UTEXT_GETNATIVEINDEX(fInputText); |
3939 | 0 | } |
3940 | 0 | } |
3941 | |
|
3942 | 0 | if (lbStartIdx < 0 || lbStartIdx < fp->fInputIdx - maxML) { |
3943 | | // We have tried all potential match starting points without |
3944 | | // getting a match. Backtrack out, and out of the |
3945 | | // Look Behind altogether. |
3946 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3947 | 0 | fActiveStart = fData[opValue+2]; |
3948 | 0 | fActiveLimit = fData[opValue+3]; |
3949 | 0 | U_ASSERT(fActiveStart >= 0); |
3950 | 0 | U_ASSERT(fActiveLimit <= fInputLength); |
3951 | 0 | break; |
3952 | 0 | } |
3953 | | |
3954 | | // Save state to this URX_LB_CONT op, so failure to match will repeat the loop. |
3955 | | // (successful match will fall off the end of the loop.) |
3956 | 0 | fp = StateSave(fp, fp->fPatIdx-3, status); |
3957 | 0 | fp->fInputIdx = lbStartIdx; |
3958 | 0 | } |
3959 | 0 | break; |
3960 | | |
3961 | 0 | case URX_LB_END: |
3962 | | // End of a look-behind block, after a successful match. |
3963 | 0 | { |
3964 | 0 | U_ASSERT(opValue>=0 && opValue+4<fPattern->fDataSize); |
3965 | 0 | if (fp->fInputIdx != fActiveLimit) { |
3966 | | // The look-behind expression matched, but the match did not |
3967 | | // extend all the way to the point that we are looking behind from. |
3968 | | // FAIL out of here, which will take us back to the LB_CONT, which |
3969 | | // will retry the match starting at another position or fail |
3970 | | // the look-behind altogether, whichever is appropriate. |
3971 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
3972 | 0 | break; |
3973 | 0 | } |
3974 | | |
3975 | | // Look-behind match is good. Restore the original input string region, |
3976 | | // which had been truncated to pin the end of the lookbehind match to the |
3977 | | // position being looked-behind. |
3978 | 0 | fActiveStart = fData[opValue+2]; |
3979 | 0 | fActiveLimit = fData[opValue+3]; |
3980 | 0 | U_ASSERT(fActiveStart >= 0); |
3981 | 0 | U_ASSERT(fActiveLimit <= fInputLength); |
3982 | 0 | } |
3983 | 0 | break; |
3984 | | |
3985 | | |
3986 | 0 | case URX_LBN_CONT: |
3987 | 0 | { |
3988 | | // Negative Look-Behind, at top of loop checking for matches of LB expression |
3989 | | // at all possible input starting positions. |
3990 | | |
3991 | | // Fetch the extra parameters of this op. |
3992 | 0 | int32_t minML = (int32_t)pat[fp->fPatIdx++]; |
3993 | 0 | int32_t maxML = (int32_t)pat[fp->fPatIdx++]; |
3994 | 0 | if (!UTEXT_USES_U16(fInputText)) { |
3995 | | // utf-8 fix to maximum match length. The pattern compiler assumes utf-16. |
3996 | | // The max length need not be exact; it just needs to be >= actual maximum. |
3997 | 0 | maxML *= 3; |
3998 | 0 | } |
3999 | 0 | int32_t continueLoc = (int32_t)pat[fp->fPatIdx++]; |
4000 | 0 | continueLoc = URX_VAL(continueLoc); |
4001 | 0 | U_ASSERT(minML <= maxML); |
4002 | 0 | U_ASSERT(minML >= 0); |
4003 | 0 | U_ASSERT(continueLoc > fp->fPatIdx); |
4004 | | |
4005 | | // Fetch (from data) the last input index where a match was attempted. |
4006 | 0 | U_ASSERT(opValue>=0 && opValue+4<fPattern->fDataSize); |
4007 | 0 | int64_t &lbStartIdx = fData[opValue+4]; |
4008 | 0 | if (lbStartIdx < 0) { |
4009 | | // First time through loop. |
4010 | 0 | lbStartIdx = fp->fInputIdx - minML; |
4011 | 0 | if (lbStartIdx > 0) { |
4012 | | // move index to a code point boundary, if it's not on one already. |
4013 | 0 | UTEXT_SETNATIVEINDEX(fInputText, lbStartIdx); |
4014 | 0 | lbStartIdx = UTEXT_GETNATIVEINDEX(fInputText); |
4015 | 0 | } |
4016 | 0 | } else { |
4017 | | // 2nd through nth time through the loop. |
4018 | | // Back up start position for match by one. |
4019 | 0 | if (lbStartIdx == 0) { |
4020 | 0 | (lbStartIdx)--; |
4021 | 0 | } else { |
4022 | 0 | UTEXT_SETNATIVEINDEX(fInputText, lbStartIdx); |
4023 | 0 | (void)UTEXT_PREVIOUS32(fInputText); |
4024 | 0 | lbStartIdx = UTEXT_GETNATIVEINDEX(fInputText); |
4025 | 0 | } |
4026 | 0 | } |
4027 | |
|
4028 | 0 | if (lbStartIdx < 0 || lbStartIdx < fp->fInputIdx - maxML) { |
4029 | | // We have tried all potential match starting points without |
4030 | | // getting a match, which means that the negative lookbehind as |
4031 | | // a whole has succeeded. Jump forward to the continue location |
4032 | 0 | fActiveStart = fData[opValue+2]; |
4033 | 0 | fActiveLimit = fData[opValue+3]; |
4034 | 0 | U_ASSERT(fActiveStart >= 0); |
4035 | 0 | U_ASSERT(fActiveLimit <= fInputLength); |
4036 | 0 | fp->fPatIdx = continueLoc; |
4037 | 0 | break; |
4038 | 0 | } |
4039 | | |
4040 | | // Save state to this URX_LB_CONT op, so failure to match will repeat the loop. |
4041 | | // (successful match will cause a FAIL out of the loop altogether.) |
4042 | 0 | fp = StateSave(fp, fp->fPatIdx-4, status); |
4043 | 0 | fp->fInputIdx = lbStartIdx; |
4044 | 0 | } |
4045 | 0 | break; |
4046 | | |
4047 | 0 | case URX_LBN_END: |
4048 | | // End of a negative look-behind block, after a successful match. |
4049 | 0 | { |
4050 | 0 | U_ASSERT(opValue>=0 && opValue+4<fPattern->fDataSize); |
4051 | 0 | if (fp->fInputIdx != fActiveLimit) { |
4052 | | // The look-behind expression matched, but the match did not |
4053 | | // extend all the way to the point that we are looking behind from. |
4054 | | // FAIL out of here, which will take us back to the LB_CONT, which |
4055 | | // will retry the match starting at another position or succeed |
4056 | | // the look-behind altogether, whichever is appropriate. |
4057 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4058 | 0 | break; |
4059 | 0 | } |
4060 | | |
4061 | | // Look-behind expression matched, which means look-behind test as |
4062 | | // a whole Fails |
4063 | | |
4064 | | // Restore the original input string length, which had been truncated |
4065 | | // inorder to pin the end of the lookbehind match |
4066 | | // to the position being looked-behind. |
4067 | 0 | fActiveStart = fData[opValue+2]; |
4068 | 0 | fActiveLimit = fData[opValue+3]; |
4069 | 0 | U_ASSERT(fActiveStart >= 0); |
4070 | 0 | U_ASSERT(fActiveLimit <= fInputLength); |
4071 | | |
4072 | | // Restore original stack position, discarding any state saved |
4073 | | // by the successful pattern match. |
4074 | 0 | U_ASSERT(opValue>=0 && opValue+1<fPattern->fDataSize); |
4075 | 0 | int32_t newStackSize = (int32_t)fData[opValue]; |
4076 | 0 | U_ASSERT(fStack->size() > newStackSize); |
4077 | 0 | fStack->setSize(newStackSize); |
4078 | | |
4079 | | // FAIL, which will take control back to someplace |
4080 | | // prior to entering the look-behind test. |
4081 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4082 | 0 | } |
4083 | 0 | break; |
4084 | | |
4085 | | |
4086 | 0 | case URX_LOOP_SR_I: |
4087 | | // Loop Initialization for the optimized implementation of |
4088 | | // [some character set]* |
4089 | | // This op scans through all matching input. |
4090 | | // The following LOOP_C op emulates stack unwinding if the following pattern fails. |
4091 | 0 | { |
4092 | 0 | U_ASSERT(opValue > 0 && opValue < fSets->size()); |
4093 | 0 | Regex8BitSet *s8 = &fPattern->fSets8[opValue]; |
4094 | 0 | UnicodeSet *s = (UnicodeSet *)fSets->elementAt(opValue); |
4095 | | |
4096 | | // Loop through input, until either the input is exhausted or |
4097 | | // we reach a character that is not a member of the set. |
4098 | 0 | int64_t ix = fp->fInputIdx; |
4099 | 0 | UTEXT_SETNATIVEINDEX(fInputText, ix); |
4100 | 0 | for (;;) { |
4101 | 0 | if (ix >= fActiveLimit) { |
4102 | 0 | fHitEnd = true; |
4103 | 0 | break; |
4104 | 0 | } |
4105 | 0 | UChar32 c = UTEXT_NEXT32(fInputText); |
4106 | 0 | if (c<256) { |
4107 | 0 | if (s8->contains(c) == false) { |
4108 | 0 | break; |
4109 | 0 | } |
4110 | 0 | } else { |
4111 | 0 | if (s->contains(c) == false) { |
4112 | 0 | break; |
4113 | 0 | } |
4114 | 0 | } |
4115 | 0 | ix = UTEXT_GETNATIVEINDEX(fInputText); |
4116 | 0 | } |
4117 | | |
4118 | | // If there were no matching characters, skip over the loop altogether. |
4119 | | // The loop doesn't run at all, a * op always succeeds. |
4120 | 0 | if (ix == fp->fInputIdx) { |
4121 | 0 | fp->fPatIdx++; // skip the URX_LOOP_C op. |
4122 | 0 | break; |
4123 | 0 | } |
4124 | | |
4125 | | // Peek ahead in the compiled pattern, to the URX_LOOP_C that |
4126 | | // must follow. It's operand is the stack location |
4127 | | // that holds the starting input index for the match of this [set]* |
4128 | 0 | int32_t loopcOp = (int32_t)pat[fp->fPatIdx]; |
4129 | 0 | U_ASSERT(URX_TYPE(loopcOp) == URX_LOOP_C); |
4130 | 0 | int32_t stackLoc = URX_VAL(loopcOp); |
4131 | 0 | U_ASSERT(stackLoc >= 0 && stackLoc < fFrameSize); |
4132 | 0 | fp->fExtra[stackLoc] = fp->fInputIdx; |
4133 | 0 | fp->fInputIdx = ix; |
4134 | | |
4135 | | // Save State to the URX_LOOP_C op that follows this one, |
4136 | | // so that match failures in the following code will return to there. |
4137 | | // Then bump the pattern idx so the LOOP_C is skipped on the way out of here. |
4138 | 0 | fp = StateSave(fp, fp->fPatIdx, status); |
4139 | 0 | fp->fPatIdx++; |
4140 | 0 | } |
4141 | 0 | break; |
4142 | | |
4143 | | |
4144 | 0 | case URX_LOOP_DOT_I: |
4145 | | // Loop Initialization for the optimized implementation of .* |
4146 | | // This op scans through all remaining input. |
4147 | | // The following LOOP_C op emulates stack unwinding if the following pattern fails. |
4148 | 0 | { |
4149 | | // Loop through input until the input is exhausted (we reach an end-of-line) |
4150 | | // In DOTALL mode, we can just go straight to the end of the input. |
4151 | 0 | int64_t ix; |
4152 | 0 | if ((opValue & 1) == 1) { |
4153 | | // Dot-matches-All mode. Jump straight to the end of the string. |
4154 | 0 | ix = fActiveLimit; |
4155 | 0 | fHitEnd = true; |
4156 | 0 | } else { |
4157 | | // NOT DOT ALL mode. Line endings do not match '.' |
4158 | | // Scan forward until a line ending or end of input. |
4159 | 0 | ix = fp->fInputIdx; |
4160 | 0 | UTEXT_SETNATIVEINDEX(fInputText, ix); |
4161 | 0 | for (;;) { |
4162 | 0 | if (ix >= fActiveLimit) { |
4163 | 0 | fHitEnd = true; |
4164 | 0 | break; |
4165 | 0 | } |
4166 | 0 | UChar32 c = UTEXT_NEXT32(fInputText); |
4167 | 0 | if ((c & 0x7f) <= 0x29) { // Fast filter of non-new-line-s |
4168 | 0 | if ((c == 0x0a) || // 0x0a is newline in both modes. |
4169 | 0 | (((opValue & 2) == 0) && // IF not UNIX_LINES mode |
4170 | 0 | isLineTerminator(c))) { |
4171 | | // char is a line ending. Exit the scanning loop. |
4172 | 0 | break; |
4173 | 0 | } |
4174 | 0 | } |
4175 | 0 | ix = UTEXT_GETNATIVEINDEX(fInputText); |
4176 | 0 | } |
4177 | 0 | } |
4178 | | |
4179 | | // If there were no matching characters, skip over the loop altogether. |
4180 | | // The loop doesn't run at all, a * op always succeeds. |
4181 | 0 | if (ix == fp->fInputIdx) { |
4182 | 0 | fp->fPatIdx++; // skip the URX_LOOP_C op. |
4183 | 0 | break; |
4184 | 0 | } |
4185 | | |
4186 | | // Peek ahead in the compiled pattern, to the URX_LOOP_C that |
4187 | | // must follow. It's operand is the stack location |
4188 | | // that holds the starting input index for the match of this .* |
4189 | 0 | int32_t loopcOp = (int32_t)pat[fp->fPatIdx]; |
4190 | 0 | U_ASSERT(URX_TYPE(loopcOp) == URX_LOOP_C); |
4191 | 0 | int32_t stackLoc = URX_VAL(loopcOp); |
4192 | 0 | U_ASSERT(stackLoc >= 0 && stackLoc < fFrameSize); |
4193 | 0 | fp->fExtra[stackLoc] = fp->fInputIdx; |
4194 | 0 | fp->fInputIdx = ix; |
4195 | | |
4196 | | // Save State to the URX_LOOP_C op that follows this one, |
4197 | | // so that match failures in the following code will return to there. |
4198 | | // Then bump the pattern idx so the LOOP_C is skipped on the way out of here. |
4199 | 0 | fp = StateSave(fp, fp->fPatIdx, status); |
4200 | 0 | fp->fPatIdx++; |
4201 | 0 | } |
4202 | 0 | break; |
4203 | | |
4204 | | |
4205 | 0 | case URX_LOOP_C: |
4206 | 0 | { |
4207 | 0 | U_ASSERT(opValue>=0 && opValue<fFrameSize); |
4208 | 0 | backSearchIndex = fp->fExtra[opValue]; |
4209 | 0 | U_ASSERT(backSearchIndex <= fp->fInputIdx); |
4210 | 0 | if (backSearchIndex == fp->fInputIdx) { |
4211 | | // We've backed up the input idx to the point that the loop started. |
4212 | | // The loop is done. Leave here without saving state. |
4213 | | // Subsequent failures won't come back here. |
4214 | 0 | break; |
4215 | 0 | } |
4216 | | // Set up for the next iteration of the loop, with input index |
4217 | | // backed up by one from the last time through, |
4218 | | // and a state save to this instruction in case the following code fails again. |
4219 | | // (We're going backwards because this loop emulates stack unwinding, not |
4220 | | // the initial scan forward.) |
4221 | 0 | U_ASSERT(fp->fInputIdx > 0); |
4222 | 0 | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
4223 | 0 | UChar32 prevC = UTEXT_PREVIOUS32(fInputText); |
4224 | 0 | fp->fInputIdx = UTEXT_GETNATIVEINDEX(fInputText); |
4225 | |
|
4226 | 0 | UChar32 twoPrevC = UTEXT_PREVIOUS32(fInputText); |
4227 | 0 | if (prevC == 0x0a && |
4228 | 0 | fp->fInputIdx > backSearchIndex && |
4229 | 0 | twoPrevC == 0x0d) { |
4230 | 0 | int32_t prevOp = (int32_t)pat[fp->fPatIdx-2]; |
4231 | 0 | if (URX_TYPE(prevOp) == URX_LOOP_DOT_I) { |
4232 | | // .*, stepping back over CRLF pair. |
4233 | 0 | fp->fInputIdx = UTEXT_GETNATIVEINDEX(fInputText); |
4234 | 0 | } |
4235 | 0 | } |
4236 | | |
4237 | |
|
4238 | 0 | fp = StateSave(fp, fp->fPatIdx-1, status); |
4239 | 0 | } |
4240 | 0 | break; |
4241 | | |
4242 | | |
4243 | | |
4244 | 0 | default: |
4245 | | // Trouble. The compiled pattern contains an entry with an |
4246 | | // unrecognized type tag. |
4247 | 0 | UPRV_UNREACHABLE_ASSERT; |
4248 | | // Unknown opcode type in opType = URX_TYPE(pat[fp->fPatIdx]). But we have |
4249 | | // reports of this in production code, don't use UPRV_UNREACHABLE_EXIT. |
4250 | | // See ICU-21669. |
4251 | 0 | status = U_INTERNAL_PROGRAM_ERROR; |
4252 | 0 | } |
4253 | | |
4254 | 0 | if (U_FAILURE(status)) { |
4255 | 0 | isMatch = false; |
4256 | 0 | break; |
4257 | 0 | } |
4258 | 0 | } |
4259 | | |
4260 | 0 | breakFromLoop: |
4261 | 0 | fMatch = isMatch; |
4262 | 0 | if (isMatch) { |
4263 | 0 | fLastMatchEnd = fMatchEnd; |
4264 | 0 | fMatchStart = startIdx; |
4265 | 0 | fMatchEnd = fp->fInputIdx; |
4266 | 0 | } |
4267 | |
|
4268 | | #ifdef REGEX_RUN_DEBUG |
4269 | | if (fTraceDebug) { |
4270 | | if (isMatch) { |
4271 | | printf("Match. start=%ld end=%ld\n\n", fMatchStart, fMatchEnd); |
4272 | | } else { |
4273 | | printf("No match\n\n"); |
4274 | | } |
4275 | | } |
4276 | | #endif |
4277 | |
|
4278 | 0 | fFrame = fp; // The active stack frame when the engine stopped. |
4279 | | // Contains the capture group results that we need to |
4280 | | // access later. |
4281 | 0 | return; |
4282 | 0 | } |
4283 | | |
4284 | | |
4285 | | //-------------------------------------------------------------------------------- |
4286 | | // |
4287 | | // MatchChunkAt This is the actual matching engine. Like MatchAt, but with the |
4288 | | // assumption that the entire string is available in the UText's |
4289 | | // chunk buffer. For now, that means we can use int32_t indexes, |
4290 | | // except for anything that needs to be saved (like group starts |
4291 | | // and ends). |
4292 | | // |
4293 | | // startIdx: begin matching a this index. |
4294 | | // toEnd: if true, match must extend to end of the input region |
4295 | | // |
4296 | | //-------------------------------------------------------------------------------- |
4297 | 0 | void RegexMatcher::MatchChunkAt(int32_t startIdx, UBool toEnd, UErrorCode &status) { |
4298 | 0 | UBool isMatch = false; // True if the we have a match. |
4299 | |
|
4300 | 0 | int32_t backSearchIndex = INT32_MAX; // used after greedy single-character matches for searching backwards |
4301 | |
|
4302 | 0 | int32_t op; // Operation from the compiled pattern, split into |
4303 | 0 | int32_t opType; // the opcode |
4304 | 0 | int32_t opValue; // and the operand value. |
4305 | |
|
4306 | | #ifdef REGEX_RUN_DEBUG |
4307 | | if (fTraceDebug) { |
4308 | | printf("MatchAt(startIdx=%d)\n", startIdx); |
4309 | | printf("Original Pattern: \"%s\"\n", CStr(StringFromUText(fPattern->fPattern))()); |
4310 | | printf("Input String: \"%s\"\n\n", CStr(StringFromUText(fInputText))()); |
4311 | | } |
4312 | | #endif |
4313 | |
|
4314 | 0 | if (U_FAILURE(status)) { |
4315 | 0 | return; |
4316 | 0 | } |
4317 | | |
4318 | | // Cache frequently referenced items from the compiled pattern |
4319 | | // |
4320 | 0 | int64_t *pat = fPattern->fCompiledPat->getBuffer(); |
4321 | |
|
4322 | 0 | const UChar *litText = fPattern->fLiteralText.getBuffer(); |
4323 | 0 | UVector *fSets = fPattern->fSets; |
4324 | |
|
4325 | 0 | const UChar *inputBuf = fInputText->chunkContents; |
4326 | |
|
4327 | 0 | fFrameSize = fPattern->fFrameSize; |
4328 | 0 | REStackFrame *fp = resetStack(); |
4329 | 0 | if (U_FAILURE(fDeferredStatus)) { |
4330 | 0 | status = fDeferredStatus; |
4331 | 0 | return; |
4332 | 0 | } |
4333 | | |
4334 | 0 | fp->fPatIdx = 0; |
4335 | 0 | fp->fInputIdx = startIdx; |
4336 | | |
4337 | | // Zero out the pattern's static data |
4338 | 0 | int32_t i; |
4339 | 0 | for (i = 0; i<fPattern->fDataSize; i++) { |
4340 | 0 | fData[i] = 0; |
4341 | 0 | } |
4342 | | |
4343 | | // |
4344 | | // Main loop for interpreting the compiled pattern. |
4345 | | // One iteration of the loop per pattern operation performed. |
4346 | | // |
4347 | 0 | for (;;) { |
4348 | 0 | op = (int32_t)pat[fp->fPatIdx]; |
4349 | 0 | opType = URX_TYPE(op); |
4350 | 0 | opValue = URX_VAL(op); |
4351 | | #ifdef REGEX_RUN_DEBUG |
4352 | | if (fTraceDebug) { |
4353 | | UTEXT_SETNATIVEINDEX(fInputText, fp->fInputIdx); |
4354 | | printf("inputIdx=%ld inputChar=%x sp=%3ld activeLimit=%ld ", fp->fInputIdx, |
4355 | | UTEXT_CURRENT32(fInputText), (int64_t *)fp-fStack->getBuffer(), fActiveLimit); |
4356 | | fPattern->dumpOp(fp->fPatIdx); |
4357 | | } |
4358 | | #endif |
4359 | 0 | fp->fPatIdx++; |
4360 | |
|
4361 | 0 | switch (opType) { |
4362 | | |
4363 | | |
4364 | 0 | case URX_NOP: |
4365 | 0 | break; |
4366 | | |
4367 | | |
4368 | 0 | case URX_BACKTRACK: |
4369 | | // Force a backtrack. In some circumstances, the pattern compiler |
4370 | | // will notice that the pattern can't possibly match anything, and will |
4371 | | // emit one of these at that point. |
4372 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4373 | 0 | break; |
4374 | | |
4375 | | |
4376 | 0 | case URX_ONECHAR: |
4377 | 0 | if (fp->fInputIdx < fActiveLimit) { |
4378 | 0 | UChar32 c; |
4379 | 0 | U16_NEXT(inputBuf, fp->fInputIdx, fActiveLimit, c); |
4380 | 0 | if (c == opValue) { |
4381 | 0 | break; |
4382 | 0 | } |
4383 | 0 | } else { |
4384 | 0 | fHitEnd = true; |
4385 | 0 | } |
4386 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4387 | 0 | break; |
4388 | | |
4389 | | |
4390 | 0 | case URX_STRING: |
4391 | 0 | { |
4392 | | // Test input against a literal string. |
4393 | | // Strings require two slots in the compiled pattern, one for the |
4394 | | // offset to the string text, and one for the length. |
4395 | 0 | int32_t stringStartIdx = opValue; |
4396 | 0 | int32_t stringLen; |
4397 | |
|
4398 | 0 | op = (int32_t)pat[fp->fPatIdx]; // Fetch the second operand |
4399 | 0 | fp->fPatIdx++; |
4400 | 0 | opType = URX_TYPE(op); |
4401 | 0 | stringLen = URX_VAL(op); |
4402 | 0 | U_ASSERT(opType == URX_STRING_LEN); |
4403 | 0 | U_ASSERT(stringLen >= 2); |
4404 | |
|
4405 | 0 | const UChar * pInp = inputBuf + fp->fInputIdx; |
4406 | 0 | const UChar * pInpLimit = inputBuf + fActiveLimit; |
4407 | 0 | const UChar * pPat = litText+stringStartIdx; |
4408 | 0 | const UChar * pEnd = pInp + stringLen; |
4409 | 0 | UBool success = true; |
4410 | 0 | while (pInp < pEnd) { |
4411 | 0 | if (pInp >= pInpLimit) { |
4412 | 0 | fHitEnd = true; |
4413 | 0 | success = false; |
4414 | 0 | break; |
4415 | 0 | } |
4416 | 0 | if (*pInp++ != *pPat++) { |
4417 | 0 | success = false; |
4418 | 0 | break; |
4419 | 0 | } |
4420 | 0 | } |
4421 | |
|
4422 | 0 | if (success) { |
4423 | 0 | fp->fInputIdx += stringLen; |
4424 | 0 | } else { |
4425 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4426 | 0 | } |
4427 | 0 | } |
4428 | 0 | break; |
4429 | | |
4430 | | |
4431 | 0 | case URX_STATE_SAVE: |
4432 | 0 | fp = StateSave(fp, opValue, status); |
4433 | 0 | break; |
4434 | | |
4435 | | |
4436 | 0 | case URX_END: |
4437 | | // The match loop will exit via this path on a successful match, |
4438 | | // when we reach the end of the pattern. |
4439 | 0 | if (toEnd && fp->fInputIdx != fActiveLimit) { |
4440 | | // The pattern matched, but not to the end of input. Try some more. |
4441 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4442 | 0 | break; |
4443 | 0 | } |
4444 | 0 | isMatch = true; |
4445 | 0 | goto breakFromLoop; |
4446 | | |
4447 | | // Start and End Capture stack frame variables are laid out out like this: |
4448 | | // fp->fExtra[opValue] - The start of a completed capture group |
4449 | | // opValue+1 - The end of a completed capture group |
4450 | | // opValue+2 - the start of a capture group whose end |
4451 | | // has not yet been reached (and might not ever be). |
4452 | 0 | case URX_START_CAPTURE: |
4453 | 0 | U_ASSERT(opValue >= 0 && opValue < fFrameSize-3); |
4454 | 0 | fp->fExtra[opValue+2] = fp->fInputIdx; |
4455 | 0 | break; |
4456 | | |
4457 | | |
4458 | 0 | case URX_END_CAPTURE: |
4459 | 0 | U_ASSERT(opValue >= 0 && opValue < fFrameSize-3); |
4460 | 0 | U_ASSERT(fp->fExtra[opValue+2] >= 0); // Start pos for this group must be set. |
4461 | 0 | fp->fExtra[opValue] = fp->fExtra[opValue+2]; // Tentative start becomes real. |
4462 | 0 | fp->fExtra[opValue+1] = fp->fInputIdx; // End position |
4463 | 0 | U_ASSERT(fp->fExtra[opValue] <= fp->fExtra[opValue+1]); |
4464 | 0 | break; |
4465 | | |
4466 | | |
4467 | 0 | case URX_DOLLAR: // $, test for End of line |
4468 | | // or for position before new line at end of input |
4469 | 0 | if (fp->fInputIdx < fAnchorLimit-2) { |
4470 | | // We are no where near the end of input. Fail. |
4471 | | // This is the common case. Keep it first. |
4472 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4473 | 0 | break; |
4474 | 0 | } |
4475 | 0 | if (fp->fInputIdx >= fAnchorLimit) { |
4476 | | // We really are at the end of input. Success. |
4477 | 0 | fHitEnd = true; |
4478 | 0 | fRequireEnd = true; |
4479 | 0 | break; |
4480 | 0 | } |
4481 | | |
4482 | | // If we are positioned just before a new-line that is located at the |
4483 | | // end of input, succeed. |
4484 | 0 | if (fp->fInputIdx == fAnchorLimit-1) { |
4485 | 0 | UChar32 c; |
4486 | 0 | U16_GET(inputBuf, fAnchorStart, fp->fInputIdx, fAnchorLimit, c); |
4487 | |
|
4488 | 0 | if (isLineTerminator(c)) { |
4489 | 0 | if ( !(c==0x0a && fp->fInputIdx>fAnchorStart && inputBuf[fp->fInputIdx-1]==0x0d)) { |
4490 | | // At new-line at end of input. Success |
4491 | 0 | fHitEnd = true; |
4492 | 0 | fRequireEnd = true; |
4493 | 0 | break; |
4494 | 0 | } |
4495 | 0 | } |
4496 | 0 | } else if (fp->fInputIdx == fAnchorLimit-2 && |
4497 | 0 | inputBuf[fp->fInputIdx]==0x0d && inputBuf[fp->fInputIdx+1]==0x0a) { |
4498 | 0 | fHitEnd = true; |
4499 | 0 | fRequireEnd = true; |
4500 | 0 | break; // At CR/LF at end of input. Success |
4501 | 0 | } |
4502 | | |
4503 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4504 | |
|
4505 | 0 | break; |
4506 | | |
4507 | | |
4508 | 0 | case URX_DOLLAR_D: // $, test for End of Line, in UNIX_LINES mode. |
4509 | 0 | if (fp->fInputIdx >= fAnchorLimit-1) { |
4510 | | // Either at the last character of input, or off the end. |
4511 | 0 | if (fp->fInputIdx == fAnchorLimit-1) { |
4512 | | // At last char of input. Success if it's a new line. |
4513 | 0 | if (inputBuf[fp->fInputIdx] == 0x0a) { |
4514 | 0 | fHitEnd = true; |
4515 | 0 | fRequireEnd = true; |
4516 | 0 | break; |
4517 | 0 | } |
4518 | 0 | } else { |
4519 | | // Off the end of input. Success. |
4520 | 0 | fHitEnd = true; |
4521 | 0 | fRequireEnd = true; |
4522 | 0 | break; |
4523 | 0 | } |
4524 | 0 | } |
4525 | | |
4526 | | // Not at end of input. Back-track out. |
4527 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4528 | 0 | break; |
4529 | | |
4530 | | |
4531 | 0 | case URX_DOLLAR_M: // $, test for End of line in multi-line mode |
4532 | 0 | { |
4533 | 0 | if (fp->fInputIdx >= fAnchorLimit) { |
4534 | | // We really are at the end of input. Success. |
4535 | 0 | fHitEnd = true; |
4536 | 0 | fRequireEnd = true; |
4537 | 0 | break; |
4538 | 0 | } |
4539 | | // If we are positioned just before a new-line, succeed. |
4540 | | // It makes no difference where the new-line is within the input. |
4541 | 0 | UChar32 c = inputBuf[fp->fInputIdx]; |
4542 | 0 | if (isLineTerminator(c)) { |
4543 | | // At a line end, except for the odd chance of being in the middle of a CR/LF sequence |
4544 | | // In multi-line mode, hitting a new-line just before the end of input does not |
4545 | | // set the hitEnd or requireEnd flags |
4546 | 0 | if ( !(c==0x0a && fp->fInputIdx>fAnchorStart && inputBuf[fp->fInputIdx-1]==0x0d)) { |
4547 | 0 | break; |
4548 | 0 | } |
4549 | 0 | } |
4550 | | // not at a new line. Fail. |
4551 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4552 | 0 | } |
4553 | 0 | break; |
4554 | | |
4555 | | |
4556 | 0 | case URX_DOLLAR_MD: // $, test for End of line in multi-line and UNIX_LINES mode |
4557 | 0 | { |
4558 | 0 | if (fp->fInputIdx >= fAnchorLimit) { |
4559 | | // We really are at the end of input. Success. |
4560 | 0 | fHitEnd = true; |
4561 | 0 | fRequireEnd = true; // Java set requireEnd in this case, even though |
4562 | 0 | break; // adding a new-line would not lose the match. |
4563 | 0 | } |
4564 | | // If we are not positioned just before a new-line, the test fails; backtrack out. |
4565 | | // It makes no difference where the new-line is within the input. |
4566 | 0 | if (inputBuf[fp->fInputIdx] != 0x0a) { |
4567 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4568 | 0 | } |
4569 | 0 | } |
4570 | 0 | break; |
4571 | | |
4572 | | |
4573 | 0 | case URX_CARET: // ^, test for start of line |
4574 | 0 | if (fp->fInputIdx != fAnchorStart) { |
4575 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4576 | 0 | } |
4577 | 0 | break; |
4578 | | |
4579 | | |
4580 | 0 | case URX_CARET_M: // ^, test for start of line in mulit-line mode |
4581 | 0 | { |
4582 | 0 | if (fp->fInputIdx == fAnchorStart) { |
4583 | | // We are at the start input. Success. |
4584 | 0 | break; |
4585 | 0 | } |
4586 | | // Check whether character just before the current pos is a new-line |
4587 | | // unless we are at the end of input |
4588 | 0 | UChar c = inputBuf[fp->fInputIdx - 1]; |
4589 | 0 | if ((fp->fInputIdx < fAnchorLimit) && |
4590 | 0 | isLineTerminator(c)) { |
4591 | | // It's a new-line. ^ is true. Success. |
4592 | | // TODO: what should be done with positions between a CR and LF? |
4593 | 0 | break; |
4594 | 0 | } |
4595 | | // Not at the start of a line. Fail. |
4596 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4597 | 0 | } |
4598 | 0 | break; |
4599 | | |
4600 | | |
4601 | 0 | case URX_CARET_M_UNIX: // ^, test for start of line in mulit-line + Unix-line mode |
4602 | 0 | { |
4603 | 0 | U_ASSERT(fp->fInputIdx >= fAnchorStart); |
4604 | 0 | if (fp->fInputIdx <= fAnchorStart) { |
4605 | | // We are at the start input. Success. |
4606 | 0 | break; |
4607 | 0 | } |
4608 | | // Check whether character just before the current pos is a new-line |
4609 | 0 | U_ASSERT(fp->fInputIdx <= fAnchorLimit); |
4610 | 0 | UChar c = inputBuf[fp->fInputIdx - 1]; |
4611 | 0 | if (c != 0x0a) { |
4612 | | // Not at the start of a line. Back-track out. |
4613 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4614 | 0 | } |
4615 | 0 | } |
4616 | 0 | break; |
4617 | | |
4618 | 0 | case URX_BACKSLASH_B: // Test for word boundaries |
4619 | 0 | { |
4620 | 0 | UBool success = isChunkWordBoundary((int32_t)fp->fInputIdx); |
4621 | 0 | success ^= (UBool)(opValue != 0); // flip sense for \B |
4622 | 0 | if (!success) { |
4623 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4624 | 0 | } |
4625 | 0 | } |
4626 | 0 | break; |
4627 | | |
4628 | | |
4629 | 0 | case URX_BACKSLASH_BU: // Test for word boundaries, Unicode-style |
4630 | 0 | { |
4631 | 0 | UBool success = isUWordBoundary(fp->fInputIdx, status); |
4632 | 0 | success ^= (UBool)(opValue != 0); // flip sense for \B |
4633 | 0 | if (!success) { |
4634 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4635 | 0 | } |
4636 | 0 | } |
4637 | 0 | break; |
4638 | | |
4639 | | |
4640 | 0 | case URX_BACKSLASH_D: // Test for decimal digit |
4641 | 0 | { |
4642 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
4643 | 0 | fHitEnd = true; |
4644 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4645 | 0 | break; |
4646 | 0 | } |
4647 | | |
4648 | 0 | UChar32 c; |
4649 | 0 | U16_NEXT(inputBuf, fp->fInputIdx, fActiveLimit, c); |
4650 | 0 | int8_t ctype = u_charType(c); // TODO: make a unicode set for this. Will be faster. |
4651 | 0 | UBool success = (ctype == U_DECIMAL_DIGIT_NUMBER); |
4652 | 0 | success ^= (UBool)(opValue != 0); // flip sense for \D |
4653 | 0 | if (!success) { |
4654 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4655 | 0 | } |
4656 | 0 | } |
4657 | 0 | break; |
4658 | | |
4659 | | |
4660 | 0 | case URX_BACKSLASH_G: // Test for position at end of previous match |
4661 | 0 | if (!((fMatch && fp->fInputIdx==fMatchEnd) || (fMatch==false && fp->fInputIdx==fActiveStart))) { |
4662 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4663 | 0 | } |
4664 | 0 | break; |
4665 | | |
4666 | | |
4667 | 0 | case URX_BACKSLASH_H: // Test for \h, horizontal white space. |
4668 | 0 | { |
4669 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
4670 | 0 | fHitEnd = true; |
4671 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4672 | 0 | break; |
4673 | 0 | } |
4674 | 0 | UChar32 c; |
4675 | 0 | U16_NEXT(inputBuf, fp->fInputIdx, fActiveLimit, c); |
4676 | 0 | int8_t ctype = u_charType(c); |
4677 | 0 | UBool success = (ctype == U_SPACE_SEPARATOR || c == 9); // SPACE_SEPARATOR || TAB |
4678 | 0 | success ^= (UBool)(opValue != 0); // flip sense for \H |
4679 | 0 | if (!success) { |
4680 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4681 | 0 | } |
4682 | 0 | } |
4683 | 0 | break; |
4684 | | |
4685 | | |
4686 | 0 | case URX_BACKSLASH_R: // Test for \R, any line break sequence. |
4687 | 0 | { |
4688 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
4689 | 0 | fHitEnd = true; |
4690 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4691 | 0 | break; |
4692 | 0 | } |
4693 | 0 | UChar32 c; |
4694 | 0 | U16_NEXT(inputBuf, fp->fInputIdx, fActiveLimit, c); |
4695 | 0 | if (isLineTerminator(c)) { |
4696 | 0 | if (c == 0x0d && fp->fInputIdx < fActiveLimit) { |
4697 | | // Check for CR/LF sequence. Consume both together when found. |
4698 | 0 | UChar c2; |
4699 | 0 | U16_NEXT(inputBuf, fp->fInputIdx, fActiveLimit, c2); |
4700 | 0 | if (c2 != 0x0a) { |
4701 | 0 | U16_PREV(inputBuf, 0, fp->fInputIdx, c2); |
4702 | 0 | } |
4703 | 0 | } |
4704 | 0 | } else { |
4705 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4706 | 0 | } |
4707 | 0 | } |
4708 | 0 | break; |
4709 | | |
4710 | | |
4711 | 0 | case URX_BACKSLASH_V: // Any single code point line ending. |
4712 | 0 | { |
4713 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
4714 | 0 | fHitEnd = true; |
4715 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4716 | 0 | break; |
4717 | 0 | } |
4718 | 0 | UChar32 c; |
4719 | 0 | U16_NEXT(inputBuf, fp->fInputIdx, fActiveLimit, c); |
4720 | 0 | UBool success = isLineTerminator(c); |
4721 | 0 | success ^= (UBool)(opValue != 0); // flip sense for \V |
4722 | 0 | if (!success) { |
4723 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4724 | 0 | } |
4725 | 0 | } |
4726 | 0 | break; |
4727 | | |
4728 | | |
4729 | 0 | case URX_BACKSLASH_X: |
4730 | | // Match a Grapheme, as defined by Unicode UAX 29. |
4731 | | |
4732 | | // Fail if at end of input |
4733 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
4734 | 0 | fHitEnd = true; |
4735 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4736 | 0 | break; |
4737 | 0 | } |
4738 | | |
4739 | 0 | fp->fInputIdx = followingGCBoundary(fp->fInputIdx, status); |
4740 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
4741 | 0 | fHitEnd = true; |
4742 | 0 | fp->fInputIdx = fActiveLimit; |
4743 | 0 | } |
4744 | 0 | break; |
4745 | | |
4746 | | |
4747 | 0 | case URX_BACKSLASH_Z: // Test for end of Input |
4748 | 0 | if (fp->fInputIdx < fAnchorLimit) { |
4749 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4750 | 0 | } else { |
4751 | 0 | fHitEnd = true; |
4752 | 0 | fRequireEnd = true; |
4753 | 0 | } |
4754 | 0 | break; |
4755 | | |
4756 | | |
4757 | | |
4758 | 0 | case URX_STATIC_SETREF: |
4759 | 0 | { |
4760 | | // Test input character against one of the predefined sets |
4761 | | // (Word Characters, for example) |
4762 | | // The high bit of the op value is a flag for the match polarity. |
4763 | | // 0: success if input char is in set. |
4764 | | // 1: success if input char is not in set. |
4765 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
4766 | 0 | fHitEnd = true; |
4767 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4768 | 0 | break; |
4769 | 0 | } |
4770 | | |
4771 | 0 | UBool success = ((opValue & URX_NEG_SET) == URX_NEG_SET); |
4772 | 0 | opValue &= ~URX_NEG_SET; |
4773 | 0 | U_ASSERT(opValue > 0 && opValue < URX_LAST_SET); |
4774 | |
|
4775 | 0 | UChar32 c; |
4776 | 0 | U16_NEXT(inputBuf, fp->fInputIdx, fActiveLimit, c); |
4777 | 0 | if (c < 256) { |
4778 | 0 | Regex8BitSet &s8 = RegexStaticSets::gStaticSets->fPropSets8[opValue]; |
4779 | 0 | if (s8.contains(c)) { |
4780 | 0 | success = !success; |
4781 | 0 | } |
4782 | 0 | } else { |
4783 | 0 | const UnicodeSet &s = RegexStaticSets::gStaticSets->fPropSets[opValue]; |
4784 | 0 | if (s.contains(c)) { |
4785 | 0 | success = !success; |
4786 | 0 | } |
4787 | 0 | } |
4788 | 0 | if (!success) { |
4789 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4790 | 0 | } |
4791 | 0 | } |
4792 | 0 | break; |
4793 | | |
4794 | | |
4795 | 0 | case URX_STAT_SETREF_N: |
4796 | 0 | { |
4797 | | // Test input character for NOT being a member of one of |
4798 | | // the predefined sets (Word Characters, for example) |
4799 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
4800 | 0 | fHitEnd = true; |
4801 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4802 | 0 | break; |
4803 | 0 | } |
4804 | | |
4805 | 0 | U_ASSERT(opValue > 0 && opValue < URX_LAST_SET); |
4806 | |
|
4807 | 0 | UChar32 c; |
4808 | 0 | U16_NEXT(inputBuf, fp->fInputIdx, fActiveLimit, c); |
4809 | 0 | if (c < 256) { |
4810 | 0 | Regex8BitSet &s8 = RegexStaticSets::gStaticSets->fPropSets8[opValue]; |
4811 | 0 | if (s8.contains(c) == false) { |
4812 | 0 | break; |
4813 | 0 | } |
4814 | 0 | } else { |
4815 | 0 | const UnicodeSet &s = RegexStaticSets::gStaticSets->fPropSets[opValue]; |
4816 | 0 | if (s.contains(c) == false) { |
4817 | 0 | break; |
4818 | 0 | } |
4819 | 0 | } |
4820 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4821 | 0 | } |
4822 | 0 | break; |
4823 | | |
4824 | | |
4825 | 0 | case URX_SETREF: |
4826 | 0 | { |
4827 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
4828 | 0 | fHitEnd = true; |
4829 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4830 | 0 | break; |
4831 | 0 | } |
4832 | | |
4833 | 0 | U_ASSERT(opValue > 0 && opValue < fSets->size()); |
4834 | | |
4835 | | // There is input left. Pick up one char and test it for set membership. |
4836 | 0 | UChar32 c; |
4837 | 0 | U16_NEXT(inputBuf, fp->fInputIdx, fActiveLimit, c); |
4838 | 0 | if (c<256) { |
4839 | 0 | Regex8BitSet *s8 = &fPattern->fSets8[opValue]; |
4840 | 0 | if (s8->contains(c)) { |
4841 | | // The character is in the set. A Match. |
4842 | 0 | break; |
4843 | 0 | } |
4844 | 0 | } else { |
4845 | 0 | UnicodeSet *s = (UnicodeSet *)fSets->elementAt(opValue); |
4846 | 0 | if (s->contains(c)) { |
4847 | | // The character is in the set. A Match. |
4848 | 0 | break; |
4849 | 0 | } |
4850 | 0 | } |
4851 | | |
4852 | | // the character wasn't in the set. |
4853 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4854 | 0 | } |
4855 | 0 | break; |
4856 | | |
4857 | | |
4858 | 0 | case URX_DOTANY: |
4859 | 0 | { |
4860 | | // . matches anything, but stops at end-of-line. |
4861 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
4862 | | // At end of input. Match failed. Backtrack out. |
4863 | 0 | fHitEnd = true; |
4864 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4865 | 0 | break; |
4866 | 0 | } |
4867 | | |
4868 | | // There is input left. Advance over one char, unless we've hit end-of-line |
4869 | 0 | UChar32 c; |
4870 | 0 | U16_NEXT(inputBuf, fp->fInputIdx, fActiveLimit, c); |
4871 | 0 | if (isLineTerminator(c)) { |
4872 | | // End of line in normal mode. . does not match. |
4873 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4874 | 0 | break; |
4875 | 0 | } |
4876 | 0 | } |
4877 | 0 | break; |
4878 | | |
4879 | | |
4880 | 0 | case URX_DOTANY_ALL: |
4881 | 0 | { |
4882 | | // . in dot-matches-all (including new lines) mode |
4883 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
4884 | | // At end of input. Match failed. Backtrack out. |
4885 | 0 | fHitEnd = true; |
4886 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4887 | 0 | break; |
4888 | 0 | } |
4889 | | |
4890 | | // There is input left. Advance over one char, except if we are |
4891 | | // at a cr/lf, advance over both of them. |
4892 | 0 | UChar32 c; |
4893 | 0 | U16_NEXT(inputBuf, fp->fInputIdx, fActiveLimit, c); |
4894 | 0 | if (c==0x0d && fp->fInputIdx < fActiveLimit) { |
4895 | | // In the case of a CR/LF, we need to advance over both. |
4896 | 0 | if (inputBuf[fp->fInputIdx] == 0x0a) { |
4897 | 0 | U16_FWD_1(inputBuf, fp->fInputIdx, fActiveLimit); |
4898 | 0 | } |
4899 | 0 | } |
4900 | 0 | } |
4901 | 0 | break; |
4902 | | |
4903 | | |
4904 | 0 | case URX_DOTANY_UNIX: |
4905 | 0 | { |
4906 | | // '.' operator, matches all, but stops at end-of-line. |
4907 | | // UNIX_LINES mode, so 0x0a is the only recognized line ending. |
4908 | 0 | if (fp->fInputIdx >= fActiveLimit) { |
4909 | | // At end of input. Match failed. Backtrack out. |
4910 | 0 | fHitEnd = true; |
4911 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4912 | 0 | break; |
4913 | 0 | } |
4914 | | |
4915 | | // There is input left. Advance over one char, unless we've hit end-of-line |
4916 | 0 | UChar32 c; |
4917 | 0 | U16_NEXT(inputBuf, fp->fInputIdx, fActiveLimit, c); |
4918 | 0 | if (c == 0x0a) { |
4919 | | // End of line in normal mode. '.' does not match the \n |
4920 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4921 | 0 | } |
4922 | 0 | } |
4923 | 0 | break; |
4924 | | |
4925 | | |
4926 | 0 | case URX_JMP: |
4927 | 0 | fp->fPatIdx = opValue; |
4928 | 0 | break; |
4929 | | |
4930 | 0 | case URX_FAIL: |
4931 | 0 | isMatch = false; |
4932 | 0 | goto breakFromLoop; |
4933 | | |
4934 | 0 | case URX_JMP_SAV: |
4935 | 0 | U_ASSERT(opValue < fPattern->fCompiledPat->size()); |
4936 | 0 | fp = StateSave(fp, fp->fPatIdx, status); // State save to loc following current |
4937 | 0 | fp->fPatIdx = opValue; // Then JMP. |
4938 | 0 | break; |
4939 | | |
4940 | 0 | case URX_JMP_SAV_X: |
4941 | | // This opcode is used with (x)+, when x can match a zero length string. |
4942 | | // Same as JMP_SAV, except conditional on the match having made forward progress. |
4943 | | // Destination of the JMP must be a URX_STO_INP_LOC, from which we get the |
4944 | | // data address of the input position at the start of the loop. |
4945 | 0 | { |
4946 | 0 | U_ASSERT(opValue > 0 && opValue < fPattern->fCompiledPat->size()); |
4947 | 0 | int32_t stoOp = (int32_t)pat[opValue-1]; |
4948 | 0 | U_ASSERT(URX_TYPE(stoOp) == URX_STO_INP_LOC); |
4949 | 0 | int32_t frameLoc = URX_VAL(stoOp); |
4950 | 0 | U_ASSERT(frameLoc >= 0 && frameLoc < fFrameSize); |
4951 | 0 | int32_t prevInputIdx = (int32_t)fp->fExtra[frameLoc]; |
4952 | 0 | U_ASSERT(prevInputIdx <= fp->fInputIdx); |
4953 | 0 | if (prevInputIdx < fp->fInputIdx) { |
4954 | | // The match did make progress. Repeat the loop. |
4955 | 0 | fp = StateSave(fp, fp->fPatIdx, status); // State save to loc following current |
4956 | 0 | fp->fPatIdx = opValue; |
4957 | 0 | fp->fExtra[frameLoc] = fp->fInputIdx; |
4958 | 0 | } |
4959 | | // If the input position did not advance, we do nothing here, |
4960 | | // execution will fall out of the loop. |
4961 | 0 | } |
4962 | 0 | break; |
4963 | | |
4964 | 0 | case URX_CTR_INIT: |
4965 | 0 | { |
4966 | 0 | U_ASSERT(opValue >= 0 && opValue < fFrameSize-2); |
4967 | 0 | fp->fExtra[opValue] = 0; // Set the loop counter variable to zero |
4968 | | |
4969 | | // Pick up the three extra operands that CTR_INIT has, and |
4970 | | // skip the pattern location counter past |
4971 | 0 | int32_t instrOperandLoc = (int32_t)fp->fPatIdx; |
4972 | 0 | fp->fPatIdx += 3; |
4973 | 0 | int32_t loopLoc = URX_VAL(pat[instrOperandLoc]); |
4974 | 0 | int32_t minCount = (int32_t)pat[instrOperandLoc+1]; |
4975 | 0 | int32_t maxCount = (int32_t)pat[instrOperandLoc+2]; |
4976 | 0 | U_ASSERT(minCount>=0); |
4977 | 0 | U_ASSERT(maxCount>=minCount || maxCount==-1); |
4978 | 0 | U_ASSERT(loopLoc>=fp->fPatIdx); |
4979 | |
|
4980 | 0 | if (minCount == 0) { |
4981 | 0 | fp = StateSave(fp, loopLoc+1, status); |
4982 | 0 | } |
4983 | 0 | if (maxCount == -1) { |
4984 | 0 | fp->fExtra[opValue+1] = fp->fInputIdx; // For loop breaking. |
4985 | 0 | } else if (maxCount == 0) { |
4986 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
4987 | 0 | } |
4988 | 0 | } |
4989 | 0 | break; |
4990 | | |
4991 | 0 | case URX_CTR_LOOP: |
4992 | 0 | { |
4993 | 0 | U_ASSERT(opValue>0 && opValue < fp->fPatIdx-2); |
4994 | 0 | int32_t initOp = (int32_t)pat[opValue]; |
4995 | 0 | U_ASSERT(URX_TYPE(initOp) == URX_CTR_INIT); |
4996 | 0 | int64_t *pCounter = &fp->fExtra[URX_VAL(initOp)]; |
4997 | 0 | int32_t minCount = (int32_t)pat[opValue+2]; |
4998 | 0 | int32_t maxCount = (int32_t)pat[opValue+3]; |
4999 | 0 | (*pCounter)++; |
5000 | 0 | if ((uint64_t)*pCounter >= (uint32_t)maxCount && maxCount != -1) { |
5001 | 0 | U_ASSERT(*pCounter == maxCount); |
5002 | 0 | break; |
5003 | 0 | } |
5004 | 0 | if (*pCounter >= minCount) { |
5005 | 0 | if (maxCount == -1) { |
5006 | | // Loop has no hard upper bound. |
5007 | | // Check that it is progressing through the input, break if it is not. |
5008 | 0 | int64_t *pLastInputIdx = &fp->fExtra[URX_VAL(initOp) + 1]; |
5009 | 0 | if (fp->fInputIdx == *pLastInputIdx) { |
5010 | 0 | break; |
5011 | 0 | } else { |
5012 | 0 | *pLastInputIdx = fp->fInputIdx; |
5013 | 0 | } |
5014 | 0 | } |
5015 | 0 | fp = StateSave(fp, fp->fPatIdx, status); |
5016 | 0 | } else { |
5017 | | // Increment time-out counter. (StateSave() does it if count >= minCount) |
5018 | 0 | fTickCounter--; |
5019 | 0 | if (fTickCounter <= 0) { |
5020 | 0 | IncrementTime(status); // Re-initializes fTickCounter |
5021 | 0 | } |
5022 | 0 | } |
5023 | 0 | fp->fPatIdx = opValue + 4; // Loop back. |
5024 | 0 | } |
5025 | 0 | break; |
5026 | | |
5027 | 0 | case URX_CTR_INIT_NG: |
5028 | 0 | { |
5029 | | // Initialize a non-greedy loop |
5030 | 0 | U_ASSERT(opValue >= 0 && opValue < fFrameSize-2); |
5031 | 0 | fp->fExtra[opValue] = 0; // Set the loop counter variable to zero |
5032 | | |
5033 | | // Pick up the three extra operands that CTR_INIT_NG has, and |
5034 | | // skip the pattern location counter past |
5035 | 0 | int32_t instrOperandLoc = (int32_t)fp->fPatIdx; |
5036 | 0 | fp->fPatIdx += 3; |
5037 | 0 | int32_t loopLoc = URX_VAL(pat[instrOperandLoc]); |
5038 | 0 | int32_t minCount = (int32_t)pat[instrOperandLoc+1]; |
5039 | 0 | int32_t maxCount = (int32_t)pat[instrOperandLoc+2]; |
5040 | 0 | U_ASSERT(minCount>=0); |
5041 | 0 | U_ASSERT(maxCount>=minCount || maxCount==-1); |
5042 | 0 | U_ASSERT(loopLoc>fp->fPatIdx); |
5043 | 0 | if (maxCount == -1) { |
5044 | 0 | fp->fExtra[opValue+1] = fp->fInputIdx; // Save initial input index for loop breaking. |
5045 | 0 | } |
5046 | |
|
5047 | 0 | if (minCount == 0) { |
5048 | 0 | if (maxCount != 0) { |
5049 | 0 | fp = StateSave(fp, fp->fPatIdx, status); |
5050 | 0 | } |
5051 | 0 | fp->fPatIdx = loopLoc+1; // Continue with stuff after repeated block |
5052 | 0 | } |
5053 | 0 | } |
5054 | 0 | break; |
5055 | | |
5056 | 0 | case URX_CTR_LOOP_NG: |
5057 | 0 | { |
5058 | | // Non-greedy {min, max} loops |
5059 | 0 | U_ASSERT(opValue>0 && opValue < fp->fPatIdx-2); |
5060 | 0 | int32_t initOp = (int32_t)pat[opValue]; |
5061 | 0 | U_ASSERT(URX_TYPE(initOp) == URX_CTR_INIT_NG); |
5062 | 0 | int64_t *pCounter = &fp->fExtra[URX_VAL(initOp)]; |
5063 | 0 | int32_t minCount = (int32_t)pat[opValue+2]; |
5064 | 0 | int32_t maxCount = (int32_t)pat[opValue+3]; |
5065 | |
|
5066 | 0 | (*pCounter)++; |
5067 | 0 | if ((uint64_t)*pCounter >= (uint32_t)maxCount && maxCount != -1) { |
5068 | | // The loop has matched the maximum permitted number of times. |
5069 | | // Break out of here with no action. Matching will |
5070 | | // continue with the following pattern. |
5071 | 0 | U_ASSERT(*pCounter == maxCount); |
5072 | 0 | break; |
5073 | 0 | } |
5074 | | |
5075 | 0 | if (*pCounter < minCount) { |
5076 | | // We haven't met the minimum number of matches yet. |
5077 | | // Loop back for another one. |
5078 | 0 | fp->fPatIdx = opValue + 4; // Loop back. |
5079 | 0 | fTickCounter--; |
5080 | 0 | if (fTickCounter <= 0) { |
5081 | 0 | IncrementTime(status); // Re-initializes fTickCounter |
5082 | 0 | } |
5083 | 0 | } else { |
5084 | | // We do have the minimum number of matches. |
5085 | | |
5086 | | // If there is no upper bound on the loop iterations, check that the input index |
5087 | | // is progressing, and stop the loop if it is not. |
5088 | 0 | if (maxCount == -1) { |
5089 | 0 | int64_t *pLastInputIdx = &fp->fExtra[URX_VAL(initOp) + 1]; |
5090 | 0 | if (fp->fInputIdx == *pLastInputIdx) { |
5091 | 0 | break; |
5092 | 0 | } |
5093 | 0 | *pLastInputIdx = fp->fInputIdx; |
5094 | 0 | } |
5095 | | |
5096 | | // Loop Continuation: we will fall into the pattern following the loop |
5097 | | // (non-greedy, don't execute loop body first), but first do |
5098 | | // a state save to the top of the loop, so that a match failure |
5099 | | // in the following pattern will try another iteration of the loop. |
5100 | 0 | fp = StateSave(fp, opValue + 4, status); |
5101 | 0 | } |
5102 | 0 | } |
5103 | 0 | break; |
5104 | | |
5105 | 0 | case URX_STO_SP: |
5106 | 0 | U_ASSERT(opValue >= 0 && opValue < fPattern->fDataSize); |
5107 | 0 | fData[opValue] = fStack->size(); |
5108 | 0 | break; |
5109 | | |
5110 | 0 | case URX_LD_SP: |
5111 | 0 | { |
5112 | 0 | U_ASSERT(opValue >= 0 && opValue < fPattern->fDataSize); |
5113 | 0 | int32_t newStackSize = (int32_t)fData[opValue]; |
5114 | 0 | U_ASSERT(newStackSize <= fStack->size()); |
5115 | 0 | int64_t *newFP = fStack->getBuffer() + newStackSize - fFrameSize; |
5116 | 0 | if (newFP == (int64_t *)fp) { |
5117 | 0 | break; |
5118 | 0 | } |
5119 | 0 | int32_t j; |
5120 | 0 | for (j=0; j<fFrameSize; j++) { |
5121 | 0 | newFP[j] = ((int64_t *)fp)[j]; |
5122 | 0 | } |
5123 | 0 | fp = (REStackFrame *)newFP; |
5124 | 0 | fStack->setSize(newStackSize); |
5125 | 0 | } |
5126 | 0 | break; |
5127 | | |
5128 | 0 | case URX_BACKREF: |
5129 | 0 | { |
5130 | 0 | U_ASSERT(opValue < fFrameSize); |
5131 | 0 | int64_t groupStartIdx = fp->fExtra[opValue]; |
5132 | 0 | int64_t groupEndIdx = fp->fExtra[opValue+1]; |
5133 | 0 | U_ASSERT(groupStartIdx <= groupEndIdx); |
5134 | 0 | int64_t inputIndex = fp->fInputIdx; |
5135 | 0 | if (groupStartIdx < 0) { |
5136 | | // This capture group has not participated in the match thus far, |
5137 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); // FAIL, no match. |
5138 | 0 | break; |
5139 | 0 | } |
5140 | 0 | UBool success = true; |
5141 | 0 | for (int64_t groupIndex = groupStartIdx; groupIndex < groupEndIdx; ++groupIndex,++inputIndex) { |
5142 | 0 | if (inputIndex >= fActiveLimit) { |
5143 | 0 | success = false; |
5144 | 0 | fHitEnd = true; |
5145 | 0 | break; |
5146 | 0 | } |
5147 | 0 | if (inputBuf[groupIndex] != inputBuf[inputIndex]) { |
5148 | 0 | success = false; |
5149 | 0 | break; |
5150 | 0 | } |
5151 | 0 | } |
5152 | 0 | if (success && groupStartIdx < groupEndIdx && U16_IS_LEAD(inputBuf[groupEndIdx-1]) && |
5153 | 0 | inputIndex < fActiveLimit && U16_IS_TRAIL(inputBuf[inputIndex])) { |
5154 | | // Capture group ended with an unpaired lead surrogate. |
5155 | | // Back reference is not permitted to match lead only of a surrogatge pair. |
5156 | 0 | success = false; |
5157 | 0 | } |
5158 | 0 | if (success) { |
5159 | 0 | fp->fInputIdx = inputIndex; |
5160 | 0 | } else { |
5161 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
5162 | 0 | } |
5163 | 0 | } |
5164 | 0 | break; |
5165 | | |
5166 | 0 | case URX_BACKREF_I: |
5167 | 0 | { |
5168 | 0 | U_ASSERT(opValue < fFrameSize); |
5169 | 0 | int64_t groupStartIdx = fp->fExtra[opValue]; |
5170 | 0 | int64_t groupEndIdx = fp->fExtra[opValue+1]; |
5171 | 0 | U_ASSERT(groupStartIdx <= groupEndIdx); |
5172 | 0 | if (groupStartIdx < 0) { |
5173 | | // This capture group has not participated in the match thus far, |
5174 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); // FAIL, no match. |
5175 | 0 | break; |
5176 | 0 | } |
5177 | 0 | CaseFoldingUCharIterator captureGroupItr(inputBuf, groupStartIdx, groupEndIdx); |
5178 | 0 | CaseFoldingUCharIterator inputItr(inputBuf, fp->fInputIdx, fActiveLimit); |
5179 | | |
5180 | | // Note: if the capture group match was of an empty string the backref |
5181 | | // match succeeds. Verified by testing: Perl matches succeed |
5182 | | // in this case, so we do too. |
5183 | |
|
5184 | 0 | UBool success = true; |
5185 | 0 | for (;;) { |
5186 | 0 | UChar32 captureGroupChar = captureGroupItr.next(); |
5187 | 0 | if (captureGroupChar == U_SENTINEL) { |
5188 | 0 | success = true; |
5189 | 0 | break; |
5190 | 0 | } |
5191 | 0 | UChar32 inputChar = inputItr.next(); |
5192 | 0 | if (inputChar == U_SENTINEL) { |
5193 | 0 | success = false; |
5194 | 0 | fHitEnd = true; |
5195 | 0 | break; |
5196 | 0 | } |
5197 | 0 | if (inputChar != captureGroupChar) { |
5198 | 0 | success = false; |
5199 | 0 | break; |
5200 | 0 | } |
5201 | 0 | } |
5202 | |
|
5203 | 0 | if (success && inputItr.inExpansion()) { |
5204 | | // We obtained a match by consuming part of a string obtained from |
5205 | | // case-folding a single code point of the input text. |
5206 | | // This does not count as an overall match. |
5207 | 0 | success = false; |
5208 | 0 | } |
5209 | |
|
5210 | 0 | if (success) { |
5211 | 0 | fp->fInputIdx = inputItr.getIndex(); |
5212 | 0 | } else { |
5213 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
5214 | 0 | } |
5215 | 0 | } |
5216 | 0 | break; |
5217 | | |
5218 | 0 | case URX_STO_INP_LOC: |
5219 | 0 | { |
5220 | 0 | U_ASSERT(opValue >= 0 && opValue < fFrameSize); |
5221 | 0 | fp->fExtra[opValue] = fp->fInputIdx; |
5222 | 0 | } |
5223 | 0 | break; |
5224 | | |
5225 | 0 | case URX_JMPX: |
5226 | 0 | { |
5227 | 0 | int32_t instrOperandLoc = (int32_t)fp->fPatIdx; |
5228 | 0 | fp->fPatIdx += 1; |
5229 | 0 | int32_t dataLoc = URX_VAL(pat[instrOperandLoc]); |
5230 | 0 | U_ASSERT(dataLoc >= 0 && dataLoc < fFrameSize); |
5231 | 0 | int32_t savedInputIdx = (int32_t)fp->fExtra[dataLoc]; |
5232 | 0 | U_ASSERT(savedInputIdx <= fp->fInputIdx); |
5233 | 0 | if (savedInputIdx < fp->fInputIdx) { |
5234 | 0 | fp->fPatIdx = opValue; // JMP |
5235 | 0 | } else { |
5236 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); // FAIL, no progress in loop. |
5237 | 0 | } |
5238 | 0 | } |
5239 | 0 | break; |
5240 | | |
5241 | 0 | case URX_LA_START: |
5242 | 0 | { |
5243 | | // Entering a look around block. |
5244 | | // Save Stack Ptr, Input Pos. |
5245 | 0 | U_ASSERT(opValue>=0 && opValue+3<fPattern->fDataSize); |
5246 | 0 | fData[opValue] = fStack->size(); |
5247 | 0 | fData[opValue+1] = fp->fInputIdx; |
5248 | 0 | fData[opValue+2] = fActiveStart; |
5249 | 0 | fData[opValue+3] = fActiveLimit; |
5250 | 0 | fActiveStart = fLookStart; // Set the match region change for |
5251 | 0 | fActiveLimit = fLookLimit; // transparent bounds. |
5252 | 0 | } |
5253 | 0 | break; |
5254 | | |
5255 | 0 | case URX_LA_END: |
5256 | 0 | { |
5257 | | // Leaving a look around block. |
5258 | | // restore Stack Ptr, Input Pos to positions they had on entry to block. |
5259 | 0 | U_ASSERT(opValue>=0 && opValue+3<fPattern->fDataSize); |
5260 | 0 | int32_t stackSize = fStack->size(); |
5261 | 0 | int32_t newStackSize = (int32_t)fData[opValue]; |
5262 | 0 | U_ASSERT(stackSize >= newStackSize); |
5263 | 0 | if (stackSize > newStackSize) { |
5264 | | // Copy the current top frame back to the new (cut back) top frame. |
5265 | | // This makes the capture groups from within the look-ahead |
5266 | | // expression available. |
5267 | 0 | int64_t *newFP = fStack->getBuffer() + newStackSize - fFrameSize; |
5268 | 0 | int32_t j; |
5269 | 0 | for (j=0; j<fFrameSize; j++) { |
5270 | 0 | newFP[j] = ((int64_t *)fp)[j]; |
5271 | 0 | } |
5272 | 0 | fp = (REStackFrame *)newFP; |
5273 | 0 | fStack->setSize(newStackSize); |
5274 | 0 | } |
5275 | 0 | fp->fInputIdx = fData[opValue+1]; |
5276 | | |
5277 | | // Restore the active region bounds in the input string; they may have |
5278 | | // been changed because of transparent bounds on a Region. |
5279 | 0 | fActiveStart = fData[opValue+2]; |
5280 | 0 | fActiveLimit = fData[opValue+3]; |
5281 | 0 | U_ASSERT(fActiveStart >= 0); |
5282 | 0 | U_ASSERT(fActiveLimit <= fInputLength); |
5283 | 0 | } |
5284 | 0 | break; |
5285 | | |
5286 | 0 | case URX_ONECHAR_I: |
5287 | 0 | if (fp->fInputIdx < fActiveLimit) { |
5288 | 0 | UChar32 c; |
5289 | 0 | U16_NEXT(inputBuf, fp->fInputIdx, fActiveLimit, c); |
5290 | 0 | if (u_foldCase(c, U_FOLD_CASE_DEFAULT) == opValue) { |
5291 | 0 | break; |
5292 | 0 | } |
5293 | 0 | } else { |
5294 | 0 | fHitEnd = true; |
5295 | 0 | } |
5296 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
5297 | 0 | break; |
5298 | | |
5299 | 0 | case URX_STRING_I: |
5300 | | // Case-insensitive test input against a literal string. |
5301 | | // Strings require two slots in the compiled pattern, one for the |
5302 | | // offset to the string text, and one for the length. |
5303 | | // The compiled string has already been case folded. |
5304 | 0 | { |
5305 | 0 | const UChar *patternString = litText + opValue; |
5306 | |
|
5307 | 0 | op = (int32_t)pat[fp->fPatIdx]; |
5308 | 0 | fp->fPatIdx++; |
5309 | 0 | opType = URX_TYPE(op); |
5310 | 0 | opValue = URX_VAL(op); |
5311 | 0 | U_ASSERT(opType == URX_STRING_LEN); |
5312 | 0 | int32_t patternStringLen = opValue; // Length of the string from the pattern. |
5313 | |
|
5314 | 0 | UChar32 cText; |
5315 | 0 | UChar32 cPattern; |
5316 | 0 | UBool success = true; |
5317 | 0 | int32_t patternStringIdx = 0; |
5318 | 0 | CaseFoldingUCharIterator inputIterator(inputBuf, fp->fInputIdx, fActiveLimit); |
5319 | 0 | while (patternStringIdx < patternStringLen) { |
5320 | 0 | U16_NEXT(patternString, patternStringIdx, patternStringLen, cPattern); |
5321 | 0 | cText = inputIterator.next(); |
5322 | 0 | if (cText != cPattern) { |
5323 | 0 | success = false; |
5324 | 0 | if (cText == U_SENTINEL) { |
5325 | 0 | fHitEnd = true; |
5326 | 0 | } |
5327 | 0 | break; |
5328 | 0 | } |
5329 | 0 | } |
5330 | 0 | if (inputIterator.inExpansion()) { |
5331 | 0 | success = false; |
5332 | 0 | } |
5333 | |
|
5334 | 0 | if (success) { |
5335 | 0 | fp->fInputIdx = inputIterator.getIndex(); |
5336 | 0 | } else { |
5337 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
5338 | 0 | } |
5339 | 0 | } |
5340 | 0 | break; |
5341 | | |
5342 | 0 | case URX_LB_START: |
5343 | 0 | { |
5344 | | // Entering a look-behind block. |
5345 | | // Save Stack Ptr, Input Pos and active input region. |
5346 | | // TODO: implement transparent bounds. Ticket #6067 |
5347 | 0 | U_ASSERT(opValue>=0 && opValue+4<fPattern->fDataSize); |
5348 | 0 | fData[opValue] = fStack->size(); |
5349 | 0 | fData[opValue+1] = fp->fInputIdx; |
5350 | | // Save input string length, then reset to pin any matches to end at |
5351 | | // the current position. |
5352 | 0 | fData[opValue+2] = fActiveStart; |
5353 | 0 | fData[opValue+3] = fActiveLimit; |
5354 | 0 | fActiveStart = fRegionStart; |
5355 | 0 | fActiveLimit = fp->fInputIdx; |
5356 | | // Init the variable containing the start index for attempted matches. |
5357 | 0 | fData[opValue+4] = -1; |
5358 | 0 | } |
5359 | 0 | break; |
5360 | | |
5361 | | |
5362 | 0 | case URX_LB_CONT: |
5363 | 0 | { |
5364 | | // Positive Look-Behind, at top of loop checking for matches of LB expression |
5365 | | // at all possible input starting positions. |
5366 | | |
5367 | | // Fetch the min and max possible match lengths. They are the operands |
5368 | | // of this op in the pattern. |
5369 | 0 | int32_t minML = (int32_t)pat[fp->fPatIdx++]; |
5370 | 0 | int32_t maxML = (int32_t)pat[fp->fPatIdx++]; |
5371 | 0 | U_ASSERT(minML <= maxML); |
5372 | 0 | U_ASSERT(minML >= 0); |
5373 | | |
5374 | | // Fetch (from data) the last input index where a match was attempted. |
5375 | 0 | U_ASSERT(opValue>=0 && opValue+4<fPattern->fDataSize); |
5376 | 0 | int64_t &lbStartIdx = fData[opValue+4]; |
5377 | 0 | if (lbStartIdx < 0) { |
5378 | | // First time through loop. |
5379 | 0 | lbStartIdx = fp->fInputIdx - minML; |
5380 | 0 | if (lbStartIdx > 0 && lbStartIdx < fInputLength) { |
5381 | 0 | U16_SET_CP_START(inputBuf, 0, lbStartIdx); |
5382 | 0 | } |
5383 | 0 | } else { |
5384 | | // 2nd through nth time through the loop. |
5385 | | // Back up start position for match by one. |
5386 | 0 | if (lbStartIdx == 0) { |
5387 | 0 | lbStartIdx--; |
5388 | 0 | } else { |
5389 | 0 | U16_BACK_1(inputBuf, 0, lbStartIdx); |
5390 | 0 | } |
5391 | 0 | } |
5392 | |
|
5393 | 0 | if (lbStartIdx < 0 || lbStartIdx < fp->fInputIdx - maxML) { |
5394 | | // We have tried all potential match starting points without |
5395 | | // getting a match. Backtrack out, and out of the |
5396 | | // Look Behind altogether. |
5397 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
5398 | 0 | fActiveStart = fData[opValue+2]; |
5399 | 0 | fActiveLimit = fData[opValue+3]; |
5400 | 0 | U_ASSERT(fActiveStart >= 0); |
5401 | 0 | U_ASSERT(fActiveLimit <= fInputLength); |
5402 | 0 | break; |
5403 | 0 | } |
5404 | | |
5405 | | // Save state to this URX_LB_CONT op, so failure to match will repeat the loop. |
5406 | | // (successful match will fall off the end of the loop.) |
5407 | 0 | fp = StateSave(fp, fp->fPatIdx-3, status); |
5408 | 0 | fp->fInputIdx = lbStartIdx; |
5409 | 0 | } |
5410 | 0 | break; |
5411 | | |
5412 | 0 | case URX_LB_END: |
5413 | | // End of a look-behind block, after a successful match. |
5414 | 0 | { |
5415 | 0 | U_ASSERT(opValue>=0 && opValue+4<fPattern->fDataSize); |
5416 | 0 | if (fp->fInputIdx != fActiveLimit) { |
5417 | | // The look-behind expression matched, but the match did not |
5418 | | // extend all the way to the point that we are looking behind from. |
5419 | | // FAIL out of here, which will take us back to the LB_CONT, which |
5420 | | // will retry the match starting at another position or fail |
5421 | | // the look-behind altogether, whichever is appropriate. |
5422 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
5423 | 0 | break; |
5424 | 0 | } |
5425 | | |
5426 | | // Look-behind match is good. Restore the original input string region, |
5427 | | // which had been truncated to pin the end of the lookbehind match to the |
5428 | | // position being looked-behind. |
5429 | 0 | fActiveStart = fData[opValue+2]; |
5430 | 0 | fActiveLimit = fData[opValue+3]; |
5431 | 0 | U_ASSERT(fActiveStart >= 0); |
5432 | 0 | U_ASSERT(fActiveLimit <= fInputLength); |
5433 | 0 | } |
5434 | 0 | break; |
5435 | | |
5436 | | |
5437 | 0 | case URX_LBN_CONT: |
5438 | 0 | { |
5439 | | // Negative Look-Behind, at top of loop checking for matches of LB expression |
5440 | | // at all possible input starting positions. |
5441 | | |
5442 | | // Fetch the extra parameters of this op. |
5443 | 0 | int32_t minML = (int32_t)pat[fp->fPatIdx++]; |
5444 | 0 | int32_t maxML = (int32_t)pat[fp->fPatIdx++]; |
5445 | 0 | int32_t continueLoc = (int32_t)pat[fp->fPatIdx++]; |
5446 | 0 | continueLoc = URX_VAL(continueLoc); |
5447 | 0 | U_ASSERT(minML <= maxML); |
5448 | 0 | U_ASSERT(minML >= 0); |
5449 | 0 | U_ASSERT(continueLoc > fp->fPatIdx); |
5450 | | |
5451 | | // Fetch (from data) the last input index where a match was attempted. |
5452 | 0 | U_ASSERT(opValue>=0 && opValue+4<fPattern->fDataSize); |
5453 | 0 | int64_t &lbStartIdx = fData[opValue+4]; |
5454 | 0 | if (lbStartIdx < 0) { |
5455 | | // First time through loop. |
5456 | 0 | lbStartIdx = fp->fInputIdx - minML; |
5457 | 0 | if (lbStartIdx > 0 && lbStartIdx < fInputLength) { |
5458 | 0 | U16_SET_CP_START(inputBuf, 0, lbStartIdx); |
5459 | 0 | } |
5460 | 0 | } else { |
5461 | | // 2nd through nth time through the loop. |
5462 | | // Back up start position for match by one. |
5463 | 0 | if (lbStartIdx == 0) { |
5464 | 0 | lbStartIdx--; // Because U16_BACK is unsafe starting at 0. |
5465 | 0 | } else { |
5466 | 0 | U16_BACK_1(inputBuf, 0, lbStartIdx); |
5467 | 0 | } |
5468 | 0 | } |
5469 | |
|
5470 | 0 | if (lbStartIdx < 0 || lbStartIdx < fp->fInputIdx - maxML) { |
5471 | | // We have tried all potential match starting points without |
5472 | | // getting a match, which means that the negative lookbehind as |
5473 | | // a whole has succeeded. Jump forward to the continue location |
5474 | 0 | fActiveStart = fData[opValue+2]; |
5475 | 0 | fActiveLimit = fData[opValue+3]; |
5476 | 0 | U_ASSERT(fActiveStart >= 0); |
5477 | 0 | U_ASSERT(fActiveLimit <= fInputLength); |
5478 | 0 | fp->fPatIdx = continueLoc; |
5479 | 0 | break; |
5480 | 0 | } |
5481 | | |
5482 | | // Save state to this URX_LB_CONT op, so failure to match will repeat the loop. |
5483 | | // (successful match will cause a FAIL out of the loop altogether.) |
5484 | 0 | fp = StateSave(fp, fp->fPatIdx-4, status); |
5485 | 0 | fp->fInputIdx = lbStartIdx; |
5486 | 0 | } |
5487 | 0 | break; |
5488 | | |
5489 | 0 | case URX_LBN_END: |
5490 | | // End of a negative look-behind block, after a successful match. |
5491 | 0 | { |
5492 | 0 | U_ASSERT(opValue>=0 && opValue+4<fPattern->fDataSize); |
5493 | 0 | if (fp->fInputIdx != fActiveLimit) { |
5494 | | // The look-behind expression matched, but the match did not |
5495 | | // extend all the way to the point that we are looking behind from. |
5496 | | // FAIL out of here, which will take us back to the LB_CONT, which |
5497 | | // will retry the match starting at another position or succeed |
5498 | | // the look-behind altogether, whichever is appropriate. |
5499 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
5500 | 0 | break; |
5501 | 0 | } |
5502 | | |
5503 | | // Look-behind expression matched, which means look-behind test as |
5504 | | // a whole Fails |
5505 | | |
5506 | | // Restore the original input string length, which had been truncated |
5507 | | // inorder to pin the end of the lookbehind match |
5508 | | // to the position being looked-behind. |
5509 | 0 | fActiveStart = fData[opValue+2]; |
5510 | 0 | fActiveLimit = fData[opValue+3]; |
5511 | 0 | U_ASSERT(fActiveStart >= 0); |
5512 | 0 | U_ASSERT(fActiveLimit <= fInputLength); |
5513 | | |
5514 | | // Restore original stack position, discarding any state saved |
5515 | | // by the successful pattern match. |
5516 | 0 | U_ASSERT(opValue>=0 && opValue+1<fPattern->fDataSize); |
5517 | 0 | int32_t newStackSize = (int32_t)fData[opValue]; |
5518 | 0 | U_ASSERT(fStack->size() > newStackSize); |
5519 | 0 | fStack->setSize(newStackSize); |
5520 | | |
5521 | | // FAIL, which will take control back to someplace |
5522 | | // prior to entering the look-behind test. |
5523 | 0 | fp = (REStackFrame *)fStack->popFrame(fFrameSize); |
5524 | 0 | } |
5525 | 0 | break; |
5526 | | |
5527 | | |
5528 | 0 | case URX_LOOP_SR_I: |
5529 | | // Loop Initialization for the optimized implementation of |
5530 | | // [some character set]* |
5531 | | // This op scans through all matching input. |
5532 | | // The following LOOP_C op emulates stack unwinding if the following pattern fails. |
5533 | 0 | { |
5534 | 0 | U_ASSERT(opValue > 0 && opValue < fSets->size()); |
5535 | 0 | Regex8BitSet *s8 = &fPattern->fSets8[opValue]; |
5536 | 0 | UnicodeSet *s = (UnicodeSet *)fSets->elementAt(opValue); |
5537 | | |
5538 | | // Loop through input, until either the input is exhausted or |
5539 | | // we reach a character that is not a member of the set. |
5540 | 0 | int32_t ix = (int32_t)fp->fInputIdx; |
5541 | 0 | for (;;) { |
5542 | 0 | if (ix >= fActiveLimit) { |
5543 | 0 | fHitEnd = true; |
5544 | 0 | break; |
5545 | 0 | } |
5546 | 0 | UChar32 c; |
5547 | 0 | U16_NEXT(inputBuf, ix, fActiveLimit, c); |
5548 | 0 | if (c<256) { |
5549 | 0 | if (s8->contains(c) == false) { |
5550 | 0 | U16_BACK_1(inputBuf, 0, ix); |
5551 | 0 | break; |
5552 | 0 | } |
5553 | 0 | } else { |
5554 | 0 | if (s->contains(c) == false) { |
5555 | 0 | U16_BACK_1(inputBuf, 0, ix); |
5556 | 0 | break; |
5557 | 0 | } |
5558 | 0 | } |
5559 | 0 | } |
5560 | | |
5561 | | // If there were no matching characters, skip over the loop altogether. |
5562 | | // The loop doesn't run at all, a * op always succeeds. |
5563 | 0 | if (ix == fp->fInputIdx) { |
5564 | 0 | fp->fPatIdx++; // skip the URX_LOOP_C op. |
5565 | 0 | break; |
5566 | 0 | } |
5567 | | |
5568 | | // Peek ahead in the compiled pattern, to the URX_LOOP_C that |
5569 | | // must follow. It's operand is the stack location |
5570 | | // that holds the starting input index for the match of this [set]* |
5571 | 0 | int32_t loopcOp = (int32_t)pat[fp->fPatIdx]; |
5572 | 0 | U_ASSERT(URX_TYPE(loopcOp) == URX_LOOP_C); |
5573 | 0 | int32_t stackLoc = URX_VAL(loopcOp); |
5574 | 0 | U_ASSERT(stackLoc >= 0 && stackLoc < fFrameSize); |
5575 | 0 | fp->fExtra[stackLoc] = fp->fInputIdx; |
5576 | 0 | fp->fInputIdx = ix; |
5577 | | |
5578 | | // Save State to the URX_LOOP_C op that follows this one, |
5579 | | // so that match failures in the following code will return to there. |
5580 | | // Then bump the pattern idx so the LOOP_C is skipped on the way out of here. |
5581 | 0 | fp = StateSave(fp, fp->fPatIdx, status); |
5582 | 0 | fp->fPatIdx++; |
5583 | 0 | } |
5584 | 0 | break; |
5585 | | |
5586 | | |
5587 | 0 | case URX_LOOP_DOT_I: |
5588 | | // Loop Initialization for the optimized implementation of .* |
5589 | | // This op scans through all remaining input. |
5590 | | // The following LOOP_C op emulates stack unwinding if the following pattern fails. |
5591 | 0 | { |
5592 | | // Loop through input until the input is exhausted (we reach an end-of-line) |
5593 | | // In DOTALL mode, we can just go straight to the end of the input. |
5594 | 0 | int32_t ix; |
5595 | 0 | if ((opValue & 1) == 1) { |
5596 | | // Dot-matches-All mode. Jump straight to the end of the string. |
5597 | 0 | ix = (int32_t)fActiveLimit; |
5598 | 0 | fHitEnd = true; |
5599 | 0 | } else { |
5600 | | // NOT DOT ALL mode. Line endings do not match '.' |
5601 | | // Scan forward until a line ending or end of input. |
5602 | 0 | ix = (int32_t)fp->fInputIdx; |
5603 | 0 | for (;;) { |
5604 | 0 | if (ix >= fActiveLimit) { |
5605 | 0 | fHitEnd = true; |
5606 | 0 | break; |
5607 | 0 | } |
5608 | 0 | UChar32 c; |
5609 | 0 | U16_NEXT(inputBuf, ix, fActiveLimit, c); // c = inputBuf[ix++] |
5610 | 0 | if ((c & 0x7f) <= 0x29) { // Fast filter of non-new-line-s |
5611 | 0 | if ((c == 0x0a) || // 0x0a is newline in both modes. |
5612 | 0 | (((opValue & 2) == 0) && // IF not UNIX_LINES mode |
5613 | 0 | isLineTerminator(c))) { |
5614 | | // char is a line ending. Put the input pos back to the |
5615 | | // line ending char, and exit the scanning loop. |
5616 | 0 | U16_BACK_1(inputBuf, 0, ix); |
5617 | 0 | break; |
5618 | 0 | } |
5619 | 0 | } |
5620 | 0 | } |
5621 | 0 | } |
5622 | | |
5623 | | // If there were no matching characters, skip over the loop altogether. |
5624 | | // The loop doesn't run at all, a * op always succeeds. |
5625 | 0 | if (ix == fp->fInputIdx) { |
5626 | 0 | fp->fPatIdx++; // skip the URX_LOOP_C op. |
5627 | 0 | break; |
5628 | 0 | } |
5629 | | |
5630 | | // Peek ahead in the compiled pattern, to the URX_LOOP_C that |
5631 | | // must follow. It's operand is the stack location |
5632 | | // that holds the starting input index for the match of this .* |
5633 | 0 | int32_t loopcOp = (int32_t)pat[fp->fPatIdx]; |
5634 | 0 | U_ASSERT(URX_TYPE(loopcOp) == URX_LOOP_C); |
5635 | 0 | int32_t stackLoc = URX_VAL(loopcOp); |
5636 | 0 | U_ASSERT(stackLoc >= 0 && stackLoc < fFrameSize); |
5637 | 0 | fp->fExtra[stackLoc] = fp->fInputIdx; |
5638 | 0 | fp->fInputIdx = ix; |
5639 | | |
5640 | | // Save State to the URX_LOOP_C op that follows this one, |
5641 | | // so that match failures in the following code will return to there. |
5642 | | // Then bump the pattern idx so the LOOP_C is skipped on the way out of here. |
5643 | 0 | fp = StateSave(fp, fp->fPatIdx, status); |
5644 | 0 | fp->fPatIdx++; |
5645 | 0 | } |
5646 | 0 | break; |
5647 | | |
5648 | | |
5649 | 0 | case URX_LOOP_C: |
5650 | 0 | { |
5651 | 0 | U_ASSERT(opValue>=0 && opValue<fFrameSize); |
5652 | 0 | backSearchIndex = (int32_t)fp->fExtra[opValue]; |
5653 | 0 | U_ASSERT(backSearchIndex <= fp->fInputIdx); |
5654 | 0 | if (backSearchIndex == fp->fInputIdx) { |
5655 | | // We've backed up the input idx to the point that the loop started. |
5656 | | // The loop is done. Leave here without saving state. |
5657 | | // Subsequent failures won't come back here. |
5658 | 0 | break; |
5659 | 0 | } |
5660 | | // Set up for the next iteration of the loop, with input index |
5661 | | // backed up by one from the last time through, |
5662 | | // and a state save to this instruction in case the following code fails again. |
5663 | | // (We're going backwards because this loop emulates stack unwinding, not |
5664 | | // the initial scan forward.) |
5665 | 0 | U_ASSERT(fp->fInputIdx > 0); |
5666 | 0 | UChar32 prevC; |
5667 | 0 | U16_PREV(inputBuf, 0, fp->fInputIdx, prevC); // !!!: should this 0 be one of f*Limit? |
5668 | |
|
5669 | 0 | if (prevC == 0x0a && |
5670 | 0 | fp->fInputIdx > backSearchIndex && |
5671 | 0 | inputBuf[fp->fInputIdx-1] == 0x0d) { |
5672 | 0 | int32_t prevOp = (int32_t)pat[fp->fPatIdx-2]; |
5673 | 0 | if (URX_TYPE(prevOp) == URX_LOOP_DOT_I) { |
5674 | | // .*, stepping back over CRLF pair. |
5675 | 0 | U16_BACK_1(inputBuf, 0, fp->fInputIdx); |
5676 | 0 | } |
5677 | 0 | } |
5678 | | |
5679 | |
|
5680 | 0 | fp = StateSave(fp, fp->fPatIdx-1, status); |
5681 | 0 | } |
5682 | 0 | break; |
5683 | | |
5684 | | |
5685 | | |
5686 | 0 | default: |
5687 | | // Trouble. The compiled pattern contains an entry with an |
5688 | | // unrecognized type tag. |
5689 | 0 | UPRV_UNREACHABLE_ASSERT; |
5690 | | // Unknown opcode type in opType = URX_TYPE(pat[fp->fPatIdx]). But we have |
5691 | | // reports of this in production code, don't use UPRV_UNREACHABLE_EXIT. |
5692 | | // See ICU-21669. |
5693 | 0 | status = U_INTERNAL_PROGRAM_ERROR; |
5694 | 0 | } |
5695 | | |
5696 | 0 | if (U_FAILURE(status)) { |
5697 | 0 | isMatch = false; |
5698 | 0 | break; |
5699 | 0 | } |
5700 | 0 | } |
5701 | | |
5702 | 0 | breakFromLoop: |
5703 | 0 | fMatch = isMatch; |
5704 | 0 | if (isMatch) { |
5705 | 0 | fLastMatchEnd = fMatchEnd; |
5706 | 0 | fMatchStart = startIdx; |
5707 | 0 | fMatchEnd = fp->fInputIdx; |
5708 | 0 | } |
5709 | |
|
5710 | | #ifdef REGEX_RUN_DEBUG |
5711 | | if (fTraceDebug) { |
5712 | | if (isMatch) { |
5713 | | printf("Match. start=%ld end=%ld\n\n", fMatchStart, fMatchEnd); |
5714 | | } else { |
5715 | | printf("No match\n\n"); |
5716 | | } |
5717 | | } |
5718 | | #endif |
5719 | |
|
5720 | 0 | fFrame = fp; // The active stack frame when the engine stopped. |
5721 | | // Contains the capture group results that we need to |
5722 | | // access later. |
5723 | |
|
5724 | 0 | return; |
5725 | 0 | } |
5726 | | |
5727 | | |
5728 | | UOBJECT_DEFINE_RTTI_IMPLEMENTATION(RegexMatcher) |
5729 | | |
5730 | | U_NAMESPACE_END |
5731 | | |
5732 | | #endif // !UCONFIG_NO_REGULAR_EXPRESSIONS |
5733 | | |