1 | // © 2017 and later: Unicode, Inc. and others. |
2 | // License & terms of use: http://www.unicode.org/copyright.html |
3 | |
4 | // edits.cpp |
5 | // created: 2017feb08 Markus W. Scherer |
6 | |
7 | #include "unicode/edits.h" |
8 | #include "unicode/unistr.h" |
9 | #include "unicode/utypes.h" |
10 | #include "cmemory.h" |
11 | #include "uassert.h" |
12 | #include "util.h" |
13 | |
14 | U_NAMESPACE_BEGIN |
15 | |
16 | namespace { |
17 | |
18 | // 0000uuuuuuuuuuuu records u+1 unchanged text units. |
19 | const int32_t MAX_UNCHANGED_LENGTH = 0x1000; |
20 | const int32_t MAX_UNCHANGED = MAX_UNCHANGED_LENGTH - 1; |
21 | |
22 | // 0mmmnnnccccccccc with m=1..6 records ccc+1 replacements of m:n text units. |
23 | const int32_t MAX_SHORT_CHANGE_OLD_LENGTH = 6; |
24 | const int32_t MAX_SHORT_CHANGE_NEW_LENGTH = 7; |
25 | const int32_t SHORT_CHANGE_NUM_MASK = 0x1ff; |
26 | const int32_t MAX_SHORT_CHANGE = 0x6fff; |
27 | |
28 | // 0111mmmmmmnnnnnn records a replacement of m text units with n. |
29 | // m or n = 61: actual length follows in the next edits array unit. |
30 | // m or n = 62..63: actual length follows in the next two edits array units. |
31 | // Bit 30 of the actual length is in the head unit. |
32 | // Trailing units have bit 15 set. |
33 | const int32_t LENGTH_IN_1TRAIL = 61; |
34 | const int32_t LENGTH_IN_2TRAIL = 62; |
35 | |
36 | } // namespace |
37 | |
38 | void Edits::releaseArray() noexcept { |
39 | if (array != stackArray) { |
40 | uprv_free(array); |
41 | } |
42 | } |
43 | |
44 | Edits &Edits::copyArray(const Edits &other) { |
45 | if (U_FAILURE(errorCode_)) { |
46 | length = delta = numChanges = 0; |
47 | return *this; |
48 | } |
49 | if (length > capacity) { |
50 | uint16_t *newArray = (uint16_t *)uprv_malloc((size_t)length * 2); |
51 | if (newArray == nullptr) { |
52 | length = delta = numChanges = 0; |
53 | errorCode_ = U_MEMORY_ALLOCATION_ERROR; |
54 | return *this; |
55 | } |
56 | releaseArray(); |
57 | array = newArray; |
58 | capacity = length; |
59 | } |
60 | if (length > 0) { |
61 | uprv_memcpy(array, other.array, (size_t)length * 2); |
62 | } |
63 | return *this; |
64 | } |
65 | |
66 | Edits &Edits::moveArray(Edits &src) noexcept { |
67 | if (U_FAILURE(errorCode_)) { |
68 | length = delta = numChanges = 0; |
69 | return *this; |
70 | } |
71 | releaseArray(); |
72 | if (length > STACK_CAPACITY) { |
73 | array = src.array; |
74 | capacity = src.capacity; |
75 | src.array = src.stackArray; |
76 | src.capacity = STACK_CAPACITY; |
77 | src.reset(); |
78 | return *this; |
79 | } |
80 | array = stackArray; |
81 | capacity = STACK_CAPACITY; |
82 | if (length > 0) { |
83 | uprv_memcpy(array, src.array, (size_t)length * 2); |
84 | } |
85 | return *this; |
86 | } |
87 | |
88 | Edits &Edits::operator=(const Edits &other) { |
89 | if (this == &other) { return *this; } // self-assignment: no-op |
90 | length = other.length; |
91 | delta = other.delta; |
92 | numChanges = other.numChanges; |
93 | errorCode_ = other.errorCode_; |
94 | return copyArray(other); |
95 | } |
96 | |
97 | Edits &Edits::operator=(Edits &&src) noexcept { |
98 | length = src.length; |
99 | delta = src.delta; |
100 | numChanges = src.numChanges; |
101 | errorCode_ = src.errorCode_; |
102 | return moveArray(src); |
103 | } |
104 | |
105 | Edits::~Edits() { |
106 | releaseArray(); |
107 | } |
108 | |
109 | void Edits::reset() noexcept { |
110 | length = delta = numChanges = 0; |
111 | errorCode_ = U_ZERO_ERROR; |
112 | } |
113 | |
114 | void Edits::addUnchanged(int32_t unchangedLength) { |
115 | if(U_FAILURE(errorCode_) || unchangedLength == 0) { return; } |
116 | if(unchangedLength < 0) { |
117 | errorCode_ = U_ILLEGAL_ARGUMENT_ERROR; |
118 | return; |
119 | } |
120 | // Merge into previous unchanged-text record, if any. |
121 | int32_t last = lastUnit(); |
122 | if(last < MAX_UNCHANGED) { |
123 | int32_t remaining = MAX_UNCHANGED - last; |
124 | if (remaining >= unchangedLength) { |
125 | setLastUnit(last + unchangedLength); |
126 | return; |
127 | } |
128 | setLastUnit(MAX_UNCHANGED); |
129 | unchangedLength -= remaining; |
130 | } |
131 | // Split large lengths into multiple units. |
132 | while(unchangedLength >= MAX_UNCHANGED_LENGTH) { |
133 | append(MAX_UNCHANGED); |
134 | unchangedLength -= MAX_UNCHANGED_LENGTH; |
135 | } |
136 | // Write a small (remaining) length. |
137 | if(unchangedLength > 0) { |
138 | append(unchangedLength - 1); |
139 | } |
140 | } |
141 | |
142 | void Edits::addReplace(int32_t oldLength, int32_t newLength) { |
143 | if(U_FAILURE(errorCode_)) { return; } |
144 | if(oldLength < 0 || newLength < 0) { |
145 | errorCode_ = U_ILLEGAL_ARGUMENT_ERROR; |
146 | return; |
147 | } |
148 | if (oldLength == 0 && newLength == 0) { |
149 | return; |
150 | } |
151 | ++numChanges; |
152 | int32_t newDelta = newLength - oldLength; |
153 | if (newDelta != 0) { |
154 | if ((newDelta > 0 && delta >= 0 && newDelta > (INT32_MAX - delta)) || |
155 | (newDelta < 0 && delta < 0 && newDelta < (INT32_MIN - delta))) { |
156 | // Integer overflow or underflow. |
157 | errorCode_ = U_INDEX_OUTOFBOUNDS_ERROR; |
158 | return; |
159 | } |
160 | delta += newDelta; |
161 | } |
162 | |
163 | if(0 < oldLength && oldLength <= MAX_SHORT_CHANGE_OLD_LENGTH && |
164 | newLength <= MAX_SHORT_CHANGE_NEW_LENGTH) { |
165 | // Merge into previous same-lengths short-replacement record, if any. |
166 | int32_t u = (oldLength << 12) | (newLength << 9); |
167 | int32_t last = lastUnit(); |
168 | if(MAX_UNCHANGED < last && last < MAX_SHORT_CHANGE && |
169 | (last & ~SHORT_CHANGE_NUM_MASK) == u && |
170 | (last & SHORT_CHANGE_NUM_MASK) < SHORT_CHANGE_NUM_MASK) { |
171 | setLastUnit(last + 1); |
172 | return; |
173 | } |
174 | append(u); |
175 | return; |
176 | } |
177 | |
178 | int32_t head = 0x7000; |
179 | if (oldLength < LENGTH_IN_1TRAIL && newLength < LENGTH_IN_1TRAIL) { |
180 | head |= oldLength << 6; |
181 | head |= newLength; |
182 | append(head); |
183 | } else if ((capacity - length) >= 5 || growArray()) { |
184 | int32_t limit = length + 1; |
185 | if(oldLength < LENGTH_IN_1TRAIL) { |
186 | head |= oldLength << 6; |
187 | } else if(oldLength <= 0x7fff) { |
188 | head |= LENGTH_IN_1TRAIL << 6; |
189 | array[limit++] = (uint16_t)(0x8000 | oldLength); |
190 | } else { |
191 | head |= (LENGTH_IN_2TRAIL + (oldLength >> 30)) << 6; |
192 | array[limit++] = (uint16_t)(0x8000 | (oldLength >> 15)); |
193 | array[limit++] = (uint16_t)(0x8000 | oldLength); |
194 | } |
195 | if(newLength < LENGTH_IN_1TRAIL) { |
196 | head |= newLength; |
197 | } else if(newLength <= 0x7fff) { |
198 | head |= LENGTH_IN_1TRAIL; |
199 | array[limit++] = (uint16_t)(0x8000 | newLength); |
200 | } else { |
201 | head |= LENGTH_IN_2TRAIL + (newLength >> 30); |
202 | array[limit++] = (uint16_t)(0x8000 | (newLength >> 15)); |
203 | array[limit++] = (uint16_t)(0x8000 | newLength); |
204 | } |
205 | array[length] = (uint16_t)head; |
206 | length = limit; |
207 | } |
208 | } |
209 | |
210 | void Edits::append(int32_t r) { |
211 | if(length < capacity || growArray()) { |
212 | array[length++] = (uint16_t)r; |
213 | } |
214 | } |
215 | |
216 | UBool Edits::growArray() { |
217 | int32_t newCapacity; |
218 | if (array == stackArray) { |
219 | newCapacity = 2000; |
220 | } else if (capacity == INT32_MAX) { |
221 | // Not U_BUFFER_OVERFLOW_ERROR because that could be confused on a string transform API |
222 | // with a result-string-buffer overflow. |
223 | errorCode_ = U_INDEX_OUTOFBOUNDS_ERROR; |
224 | return false; |
225 | } else if (capacity >= (INT32_MAX / 2)) { |
226 | newCapacity = INT32_MAX; |
227 | } else { |
228 | newCapacity = 2 * capacity; |
229 | } |
230 | // Grow by at least 5 units so that a maximal change record will fit. |
231 | if ((newCapacity - capacity) < 5) { |
232 | errorCode_ = U_INDEX_OUTOFBOUNDS_ERROR; |
233 | return false; |
234 | } |
235 | uint16_t *newArray = (uint16_t *)uprv_malloc((size_t)newCapacity * 2); |
236 | if (newArray == nullptr) { |
237 | errorCode_ = U_MEMORY_ALLOCATION_ERROR; |
238 | return false; |
239 | } |
240 | uprv_memcpy(newArray, array, (size_t)length * 2); |
241 | releaseArray(); |
242 | array = newArray; |
243 | capacity = newCapacity; |
244 | return true; |
245 | } |
246 | |
247 | UBool Edits::copyErrorTo(UErrorCode &outErrorCode) const { |
248 | if (U_FAILURE(outErrorCode)) { return true; } |
249 | if (U_SUCCESS(errorCode_)) { return false; } |
250 | outErrorCode = errorCode_; |
251 | return true; |
252 | } |
253 | |
254 | Edits &Edits::mergeAndAppend(const Edits &ab, const Edits &bc, UErrorCode &errorCode) { |
255 | if (copyErrorTo(errorCode)) { return *this; } |
256 | // Picture string a --(Edits ab)--> string b --(Edits bc)--> string c. |
257 | // Parallel iteration over both Edits. |
258 | Iterator abIter = ab.getFineIterator(); |
259 | Iterator bcIter = bc.getFineIterator(); |
260 | UBool abHasNext = true, bcHasNext = true; |
261 | // Copy iterator state into local variables, so that we can modify and subdivide spans. |
262 | // ab old & new length, bc old & new length |
263 | int32_t aLength = 0, ab_bLength = 0, bc_bLength = 0, cLength = 0; |
264 | // When we have different-intermediate-length changes, we accumulate a larger change. |
265 | int32_t pending_aLength = 0, pending_cLength = 0; |
266 | for (;;) { |
267 | // At this point, for each of the two iterators: |
268 | // Either we are done with the locally cached current edit, |
269 | // and its intermediate-string length has been reset, |
270 | // or we will continue to work with a truncated remainder of this edit. |
271 | // |
272 | // If the current edit is done, and the iterator has not yet reached the end, |
273 | // then we fetch the next edit. This is true for at least one of the iterators. |
274 | // |
275 | // Normally it does not matter whether we fetch from ab and then bc or vice versa. |
276 | // However, the result is observably different when |
277 | // ab deletions meet bc insertions at the same intermediate-string index. |
278 | // Some users expect the bc insertions to come first, so we fetch from bc first. |
279 | if (bc_bLength == 0) { |
280 | if (bcHasNext && (bcHasNext = bcIter.next(errorCode)) != 0) { |
281 | bc_bLength = bcIter.oldLength(); |
282 | cLength = bcIter.newLength(); |
283 | if (bc_bLength == 0) { |
284 | // insertion |
285 | if (ab_bLength == 0 || !abIter.hasChange()) { |
286 | addReplace(pending_aLength, pending_cLength + cLength); |
287 | pending_aLength = pending_cLength = 0; |
288 | } else { |
289 | pending_cLength += cLength; |
290 | } |
291 | continue; |
292 | } |
293 | } |
294 | // else see if the other iterator is done, too. |
295 | } |
296 | if (ab_bLength == 0) { |
297 | if (abHasNext && (abHasNext = abIter.next(errorCode)) != 0) { |
298 | aLength = abIter.oldLength(); |
299 | ab_bLength = abIter.newLength(); |
300 | if (ab_bLength == 0) { |
301 | // deletion |
302 | if (bc_bLength == bcIter.oldLength() || !bcIter.hasChange()) { |
303 | addReplace(pending_aLength + aLength, pending_cLength); |
304 | pending_aLength = pending_cLength = 0; |
305 | } else { |
306 | pending_aLength += aLength; |
307 | } |
308 | continue; |
309 | } |
310 | } else if (bc_bLength == 0) { |
311 | // Both iterators are done at the same time: |
312 | // The intermediate-string lengths match. |
313 | break; |
314 | } else { |
315 | // The ab output string is shorter than the bc input string. |
316 | if (!copyErrorTo(errorCode)) { |
317 | errorCode = U_ILLEGAL_ARGUMENT_ERROR; |
318 | } |
319 | return *this; |
320 | } |
321 | } |
322 | if (bc_bLength == 0) { |
323 | // The bc input string is shorter than the ab output string. |
324 | if (!copyErrorTo(errorCode)) { |
325 | errorCode = U_ILLEGAL_ARGUMENT_ERROR; |
326 | } |
327 | return *this; |
328 | } |
329 | // Done fetching: ab_bLength > 0 && bc_bLength > 0 |
330 | |
331 | // The current state has two parts: |
332 | // - Past: We accumulate a longer ac edit in the "pending" variables. |
333 | // - Current: We have copies of the current ab/bc edits in local variables. |
334 | // At least one side is newly fetched. |
335 | // One side might be a truncated remainder of an edit we fetched earlier. |
336 | |
337 | if (!abIter.hasChange() && !bcIter.hasChange()) { |
338 | // An unchanged span all the way from string a to string c. |
339 | if (pending_aLength != 0 || pending_cLength != 0) { |
340 | addReplace(pending_aLength, pending_cLength); |
341 | pending_aLength = pending_cLength = 0; |
342 | } |
343 | int32_t unchangedLength = aLength <= cLength ? aLength : cLength; |
344 | addUnchanged(unchangedLength); |
345 | ab_bLength = aLength -= unchangedLength; |
346 | bc_bLength = cLength -= unchangedLength; |
347 | // At least one of the unchanged spans is now empty. |
348 | continue; |
349 | } |
350 | if (!abIter.hasChange() && bcIter.hasChange()) { |
351 | // Unchanged a->b but changed b->c. |
352 | if (ab_bLength >= bc_bLength) { |
353 | // Split the longer unchanged span into change + remainder. |
354 | addReplace(pending_aLength + bc_bLength, pending_cLength + cLength); |
355 | pending_aLength = pending_cLength = 0; |
356 | aLength = ab_bLength -= bc_bLength; |
357 | bc_bLength = 0; |
358 | continue; |
359 | } |
360 | // Handle the shorter unchanged span below like a change. |
361 | } else if (abIter.hasChange() && !bcIter.hasChange()) { |
362 | // Changed a->b and then unchanged b->c. |
363 | if (ab_bLength <= bc_bLength) { |
364 | // Split the longer unchanged span into change + remainder. |
365 | addReplace(pending_aLength + aLength, pending_cLength + ab_bLength); |
366 | pending_aLength = pending_cLength = 0; |
367 | cLength = bc_bLength -= ab_bLength; |
368 | ab_bLength = 0; |
369 | continue; |
370 | } |
371 | // Handle the shorter unchanged span below like a change. |
372 | } else { // both abIter.hasChange() && bcIter.hasChange() |
373 | if (ab_bLength == bc_bLength) { |
374 | // Changes on both sides up to the same position. Emit & reset. |
375 | addReplace(pending_aLength + aLength, pending_cLength + cLength); |
376 | pending_aLength = pending_cLength = 0; |
377 | ab_bLength = bc_bLength = 0; |
378 | continue; |
379 | } |
380 | } |
381 | // Accumulate the a->c change, reset the shorter side, |
382 | // keep a remainder of the longer one. |
383 | pending_aLength += aLength; |
384 | pending_cLength += cLength; |
385 | if (ab_bLength < bc_bLength) { |
386 | bc_bLength -= ab_bLength; |
387 | cLength = ab_bLength = 0; |
388 | } else { // ab_bLength > bc_bLength |
389 | ab_bLength -= bc_bLength; |
390 | aLength = bc_bLength = 0; |
391 | } |
392 | } |
393 | if (pending_aLength != 0 || pending_cLength != 0) { |
394 | addReplace(pending_aLength, pending_cLength); |
395 | } |
396 | copyErrorTo(errorCode); |
397 | return *this; |
398 | } |
399 | |
400 | Edits::Iterator::Iterator(const uint16_t *a, int32_t len, UBool oc, UBool crs) : |
401 | array(a), index(0), length(len), remaining(0), |
402 | onlyChanges_(oc), coarse(crs), |
403 | dir(0), changed(false), oldLength_(0), newLength_(0), |
404 | srcIndex(0), replIndex(0), destIndex(0) {} |
405 | |
406 | int32_t Edits::Iterator::readLength(int32_t head) { |
407 | if (head < LENGTH_IN_1TRAIL) { |
408 | return head; |
409 | } else if (head < LENGTH_IN_2TRAIL) { |
410 | U_ASSERT(index < length); |
411 | U_ASSERT(array[index] >= 0x8000); |
412 | return array[index++] & 0x7fff; |
413 | } else { |
414 | U_ASSERT((index + 2) <= length); |
415 | U_ASSERT(array[index] >= 0x8000); |
416 | U_ASSERT(array[index + 1] >= 0x8000); |
417 | int32_t len = ((head & 1) << 30) | |
418 | ((int32_t)(array[index] & 0x7fff) << 15) | |
419 | (array[index + 1] & 0x7fff); |
420 | index += 2; |
421 | return len; |
422 | } |
423 | } |
424 | |
425 | void Edits::Iterator::updateNextIndexes() { |
426 | srcIndex += oldLength_; |
427 | if (changed) { |
428 | replIndex += newLength_; |
429 | } |
430 | destIndex += newLength_; |
431 | } |
432 | |
433 | void Edits::Iterator::updatePreviousIndexes() { |
434 | srcIndex -= oldLength_; |
435 | if (changed) { |
436 | replIndex -= newLength_; |
437 | } |
438 | destIndex -= newLength_; |
439 | } |
440 | |
441 | UBool Edits::Iterator::noNext() { |
442 | // No change before or beyond the string. |
443 | dir = 0; |
444 | changed = false; |
445 | oldLength_ = newLength_ = 0; |
446 | return false; |
447 | } |
448 | |
449 | UBool Edits::Iterator::next(UBool onlyChanges, UErrorCode &errorCode) { |
450 | // Forward iteration: Update the string indexes to the limit of the current span, |
451 | // and post-increment-read array units to assemble a new span. |
452 | // Leaves the array index one after the last unit of that span. |
453 | if (U_FAILURE(errorCode)) { return false; } |
454 | // We have an errorCode in case we need to start guarding against integer overflows. |
455 | // It is also convenient for caller loops if we bail out when an error was set elsewhere. |
456 | if (dir > 0) { |
457 | updateNextIndexes(); |
458 | } else { |
459 | if (dir < 0) { |
460 | // Turn around from previous() to next(). |
461 | // Post-increment-read the same span again. |
462 | if (remaining > 0) { |
463 | // Fine-grained iterator: |
464 | // Stay on the current one of a sequence of compressed changes. |
465 | ++index; // next() rests on the index after the sequence unit. |
466 | dir = 1; |
467 | return true; |
468 | } |
469 | } |
470 | dir = 1; |
471 | } |
472 | if (remaining >= 1) { |
473 | // Fine-grained iterator: Continue a sequence of compressed changes. |
474 | if (remaining > 1) { |
475 | --remaining; |
476 | return true; |
477 | } |
478 | remaining = 0; |
479 | } |
480 | if (index >= length) { |
481 | return noNext(); |
482 | } |
483 | int32_t u = array[index++]; |
484 | if (u <= MAX_UNCHANGED) { |
485 | // Combine adjacent unchanged ranges. |
486 | changed = false; |
487 | oldLength_ = u + 1; |
488 | while (index < length && (u = array[index]) <= MAX_UNCHANGED) { |
489 | ++index; |
490 | oldLength_ += u + 1; |
491 | } |
492 | newLength_ = oldLength_; |
493 | if (onlyChanges) { |
494 | updateNextIndexes(); |
495 | if (index >= length) { |
496 | return noNext(); |
497 | } |
498 | // already fetched u > MAX_UNCHANGED at index |
499 | ++index; |
500 | } else { |
501 | return true; |
502 | } |
503 | } |
504 | changed = true; |
505 | if (u <= MAX_SHORT_CHANGE) { |
506 | int32_t oldLen = u >> 12; |
507 | int32_t newLen = (u >> 9) & MAX_SHORT_CHANGE_NEW_LENGTH; |
508 | int32_t num = (u & SHORT_CHANGE_NUM_MASK) + 1; |
509 | if (coarse) { |
510 | oldLength_ = num * oldLen; |
511 | newLength_ = num * newLen; |
512 | } else { |
513 | // Split a sequence of changes that was compressed into one unit. |
514 | oldLength_ = oldLen; |
515 | newLength_ = newLen; |
516 | if (num > 1) { |
517 | remaining = num; // This is the first of two or more changes. |
518 | } |
519 | return true; |
520 | } |
521 | } else { |
522 | U_ASSERT(u <= 0x7fff); |
523 | oldLength_ = readLength((u >> 6) & 0x3f); |
524 | newLength_ = readLength(u & 0x3f); |
525 | if (!coarse) { |
526 | return true; |
527 | } |
528 | } |
529 | // Combine adjacent changes. |
530 | while (index < length && (u = array[index]) > MAX_UNCHANGED) { |
531 | ++index; |
532 | if (u <= MAX_SHORT_CHANGE) { |
533 | int32_t num = (u & SHORT_CHANGE_NUM_MASK) + 1; |
534 | oldLength_ += (u >> 12) * num; |
535 | newLength_ += ((u >> 9) & MAX_SHORT_CHANGE_NEW_LENGTH) * num; |
536 | } else { |
537 | U_ASSERT(u <= 0x7fff); |
538 | oldLength_ += readLength((u >> 6) & 0x3f); |
539 | newLength_ += readLength(u & 0x3f); |
540 | } |
541 | } |
542 | return true; |
543 | } |
544 | |
545 | UBool Edits::Iterator::previous(UErrorCode &errorCode) { |
546 | // Backward iteration: Pre-decrement-read array units to assemble a new span, |
547 | // then update the string indexes to the start of that span. |
548 | // Leaves the array index on the head unit of that span. |
549 | if (U_FAILURE(errorCode)) { return false; } |
550 | // We have an errorCode in case we need to start guarding against integer overflows. |
551 | // It is also convenient for caller loops if we bail out when an error was set elsewhere. |
552 | if (dir >= 0) { |
553 | if (dir > 0) { |
554 | // Turn around from next() to previous(). |
555 | // Set the string indexes to the span limit and |
556 | // pre-decrement-read the same span again. |
557 | if (remaining > 0) { |
558 | // Fine-grained iterator: |
559 | // Stay on the current one of a sequence of compressed changes. |
560 | --index; // previous() rests on the sequence unit. |
561 | dir = -1; |
562 | return true; |
563 | } |
564 | updateNextIndexes(); |
565 | } |
566 | dir = -1; |
567 | } |
568 | if (remaining > 0) { |
569 | // Fine-grained iterator: Continue a sequence of compressed changes. |
570 | int32_t u = array[index]; |
571 | U_ASSERT(MAX_UNCHANGED < u && u <= MAX_SHORT_CHANGE); |
572 | if (remaining <= (u & SHORT_CHANGE_NUM_MASK)) { |
573 | ++remaining; |
574 | updatePreviousIndexes(); |
575 | return true; |
576 | } |
577 | remaining = 0; |
578 | } |
579 | if (index <= 0) { |
580 | return noNext(); |
581 | } |
582 | int32_t u = array[--index]; |
583 | if (u <= MAX_UNCHANGED) { |
584 | // Combine adjacent unchanged ranges. |
585 | changed = false; |
586 | oldLength_ = u + 1; |
587 | while (index > 0 && (u = array[index - 1]) <= MAX_UNCHANGED) { |
588 | --index; |
589 | oldLength_ += u + 1; |
590 | } |
591 | newLength_ = oldLength_; |
592 | // No need to handle onlyChanges as long as previous() is called only from findIndex(). |
593 | updatePreviousIndexes(); |
594 | return true; |
595 | } |
596 | changed = true; |
597 | if (u <= MAX_SHORT_CHANGE) { |
598 | int32_t oldLen = u >> 12; |
599 | int32_t newLen = (u >> 9) & MAX_SHORT_CHANGE_NEW_LENGTH; |
600 | int32_t num = (u & SHORT_CHANGE_NUM_MASK) + 1; |
601 | if (coarse) { |
602 | oldLength_ = num * oldLen; |
603 | newLength_ = num * newLen; |
604 | } else { |
605 | // Split a sequence of changes that was compressed into one unit. |
606 | oldLength_ = oldLen; |
607 | newLength_ = newLen; |
608 | if (num > 1) { |
609 | remaining = 1; // This is the last of two or more changes. |
610 | } |
611 | updatePreviousIndexes(); |
612 | return true; |
613 | } |
614 | } else { |
615 | if (u <= 0x7fff) { |
616 | // The change is encoded in u alone. |
617 | oldLength_ = readLength((u >> 6) & 0x3f); |
618 | newLength_ = readLength(u & 0x3f); |
619 | } else { |
620 | // Back up to the head of the change, read the lengths, |
621 | // and reset the index to the head again. |
622 | U_ASSERT(index > 0); |
623 | while ((u = array[--index]) > 0x7fff) {} |
624 | U_ASSERT(u > MAX_SHORT_CHANGE); |
625 | int32_t headIndex = index++; |
626 | oldLength_ = readLength((u >> 6) & 0x3f); |
627 | newLength_ = readLength(u & 0x3f); |
628 | index = headIndex; |
629 | } |
630 | if (!coarse) { |
631 | updatePreviousIndexes(); |
632 | return true; |
633 | } |
634 | } |
635 | // Combine adjacent changes. |
636 | while (index > 0 && (u = array[index - 1]) > MAX_UNCHANGED) { |
637 | --index; |
638 | if (u <= MAX_SHORT_CHANGE) { |
639 | int32_t num = (u & SHORT_CHANGE_NUM_MASK) + 1; |
640 | oldLength_ += (u >> 12) * num; |
641 | newLength_ += ((u >> 9) & MAX_SHORT_CHANGE_NEW_LENGTH) * num; |
642 | } else if (u <= 0x7fff) { |
643 | // Read the lengths, and reset the index to the head again. |
644 | int32_t headIndex = index++; |
645 | oldLength_ += readLength((u >> 6) & 0x3f); |
646 | newLength_ += readLength(u & 0x3f); |
647 | index = headIndex; |
648 | } |
649 | } |
650 | updatePreviousIndexes(); |
651 | return true; |
652 | } |
653 | |
654 | int32_t Edits::Iterator::findIndex(int32_t i, UBool findSource, UErrorCode &errorCode) { |
655 | if (U_FAILURE(errorCode) || i < 0) { return -1; } |
656 | int32_t spanStart, spanLength; |
657 | if (findSource) { // find source index |
658 | spanStart = srcIndex; |
659 | spanLength = oldLength_; |
660 | } else { // find destination index |
661 | spanStart = destIndex; |
662 | spanLength = newLength_; |
663 | } |
664 | if (i < spanStart) { |
665 | if (i >= (spanStart / 2)) { |
666 | // Search backwards. |
667 | for (;;) { |
668 | UBool hasPrevious = previous(errorCode); |
669 | U_ASSERT(hasPrevious); // because i>=0 and the first span starts at 0 |
670 | (void)hasPrevious; // avoid unused-variable warning |
671 | spanStart = findSource ? srcIndex : destIndex; |
672 | if (i >= spanStart) { |
673 | // The index is in the current span. |
674 | return 0; |
675 | } |
676 | if (remaining > 0) { |
677 | // Is the index in one of the remaining compressed edits? |
678 | // spanStart is the start of the current span, first of the remaining ones. |
679 | spanLength = findSource ? oldLength_ : newLength_; |
680 | int32_t u = array[index]; |
681 | U_ASSERT(MAX_UNCHANGED < u && u <= MAX_SHORT_CHANGE); |
682 | int32_t num = (u & SHORT_CHANGE_NUM_MASK) + 1 - remaining; |
683 | int32_t len = num * spanLength; |
684 | if (i >= (spanStart - len)) { |
685 | int32_t n = ((spanStart - i - 1) / spanLength) + 1; |
686 | // 1 <= n <= num |
687 | srcIndex -= n * oldLength_; |
688 | replIndex -= n * newLength_; |
689 | destIndex -= n * newLength_; |
690 | remaining += n; |
691 | return 0; |
692 | } |
693 | // Skip all of these edits at once. |
694 | srcIndex -= num * oldLength_; |
695 | replIndex -= num * newLength_; |
696 | destIndex -= num * newLength_; |
697 | remaining = 0; |
698 | } |
699 | } |
700 | } |
701 | // Reset the iterator to the start. |
702 | dir = 0; |
703 | index = remaining = oldLength_ = newLength_ = srcIndex = replIndex = destIndex = 0; |
704 | } else if (i < (spanStart + spanLength)) { |
705 | // The index is in the current span. |
706 | return 0; |
707 | } |
708 | while (next(false, errorCode)) { |
709 | if (findSource) { |
710 | spanStart = srcIndex; |
711 | spanLength = oldLength_; |
712 | } else { |
713 | spanStart = destIndex; |
714 | spanLength = newLength_; |
715 | } |
716 | if (i < (spanStart + spanLength)) { |
717 | // The index is in the current span. |
718 | return 0; |
719 | } |
720 | if (remaining > 1) { |
721 | // Is the index in one of the remaining compressed edits? |
722 | // spanStart is the start of the current span, first of the remaining ones. |
723 | int32_t len = remaining * spanLength; |
724 | if (i < (spanStart + len)) { |
725 | int32_t n = (i - spanStart) / spanLength; // 1 <= n <= remaining - 1 |
726 | srcIndex += n * oldLength_; |
727 | replIndex += n * newLength_; |
728 | destIndex += n * newLength_; |
729 | remaining -= n; |
730 | return 0; |
731 | } |
732 | // Make next() skip all of these edits at once. |
733 | oldLength_ *= remaining; |
734 | newLength_ *= remaining; |
735 | remaining = 0; |
736 | } |
737 | } |
738 | return 1; |
739 | } |
740 | |
741 | int32_t Edits::Iterator::destinationIndexFromSourceIndex(int32_t i, UErrorCode &errorCode) { |
742 | int32_t where = findIndex(i, true, errorCode); |
743 | if (where < 0) { |
744 | // Error or before the string. |
745 | return 0; |
746 | } |
747 | if (where > 0 || i == srcIndex) { |
748 | // At or after string length, or at start of the found span. |
749 | return destIndex; |
750 | } |
751 | if (changed) { |
752 | // In a change span, map to its end. |
753 | return destIndex + newLength_; |
754 | } else { |
755 | // In an unchanged span, offset 1:1 within it. |
756 | return destIndex + (i - srcIndex); |
757 | } |
758 | } |
759 | |
760 | int32_t Edits::Iterator::sourceIndexFromDestinationIndex(int32_t i, UErrorCode &errorCode) { |
761 | int32_t where = findIndex(i, false, errorCode); |
762 | if (where < 0) { |
763 | // Error or before the string. |
764 | return 0; |
765 | } |
766 | if (where > 0 || i == destIndex) { |
767 | // At or after string length, or at start of the found span. |
768 | return srcIndex; |
769 | } |
770 | if (changed) { |
771 | // In a change span, map to its end. |
772 | return srcIndex + oldLength_; |
773 | } else { |
774 | // In an unchanged span, offset within it. |
775 | return srcIndex + (i - destIndex); |
776 | } |
777 | } |
778 | |
779 | UnicodeString& Edits::Iterator::toString(UnicodeString& sb) const { |
780 | sb.append(u"{ src[" , -1); |
781 | ICU_Utility::appendNumber(sb, srcIndex); |
782 | sb.append(u".." , -1); |
783 | ICU_Utility::appendNumber(sb, srcIndex + oldLength_); |
784 | if (changed) { |
785 | sb.append(u"] ⇝ dest[" , -1); |
786 | } else { |
787 | sb.append(u"] ≡ dest[" , -1); |
788 | } |
789 | ICU_Utility::appendNumber(sb, destIndex); |
790 | sb.append(u".." , -1); |
791 | ICU_Utility::appendNumber(sb, destIndex + newLength_); |
792 | if (changed) { |
793 | sb.append(u"], repl[" , -1); |
794 | ICU_Utility::appendNumber(sb, replIndex); |
795 | sb.append(u".." , -1); |
796 | ICU_Utility::appendNumber(sb, replIndex + newLength_); |
797 | sb.append(u"] }" , -1); |
798 | } else { |
799 | sb.append(u"] (no-change) }" , -1); |
800 | } |
801 | return sb; |
802 | } |
803 | |
804 | U_NAMESPACE_END |
805 | |