1 | // © 2016 and later: Unicode, Inc. and others. |
2 | // License & terms of use: http://www.unicode.org/copyright.html |
3 | /* |
4 | ********************************************************************** |
5 | * Copyright (C) 1999-2016, International Business Machines |
6 | * Corporation and others. All Rights Reserved. |
7 | ********************************************************************** |
8 | * Date Name Description |
9 | * 11/17/99 aliu Creation. |
10 | ********************************************************************** |
11 | */ |
12 | |
13 | #include "utypeinfo.h" // for 'typeid' to work |
14 | |
15 | #include "unicode/utypes.h" |
16 | |
17 | #if !UCONFIG_NO_TRANSLITERATION |
18 | |
19 | #include "unicode/putil.h" |
20 | #include "unicode/translit.h" |
21 | #include "unicode/locid.h" |
22 | #include "unicode/msgfmt.h" |
23 | #include "unicode/rep.h" |
24 | #include "unicode/resbund.h" |
25 | #include "unicode/unifilt.h" |
26 | #include "unicode/uniset.h" |
27 | #include "unicode/uscript.h" |
28 | #include "unicode/strenum.h" |
29 | #include "unicode/utf16.h" |
30 | #include "cpdtrans.h" |
31 | #include "nultrans.h" |
32 | #include "rbt_data.h" |
33 | #include "rbt_pars.h" |
34 | #include "rbt.h" |
35 | #include "transreg.h" |
36 | #include "name2uni.h" |
37 | #include "nortrans.h" |
38 | #include "remtrans.h" |
39 | #include "titletrn.h" |
40 | #include "tolowtrn.h" |
41 | #include "toupptrn.h" |
42 | #include "uni2name.h" |
43 | #include "brktrans.h" |
44 | #include "esctrn.h" |
45 | #include "unesctrn.h" |
46 | #include "tridpars.h" |
47 | #include "anytrans.h" |
48 | #include "util.h" |
49 | #include "hash.h" |
50 | #include "mutex.h" |
51 | #include "ucln_in.h" |
52 | #include "uassert.h" |
53 | #include "cmemory.h" |
54 | #include "cstring.h" |
55 | #include "uinvchar.h" |
56 | |
57 | static const UChar TARGET_SEP = 0x002D; /*-*/ |
58 | static const UChar ID_DELIM = 0x003B; /*;*/ |
59 | static const UChar VARIANT_SEP = 0x002F; // '/' |
60 | |
61 | /** |
62 | * Prefix for resource bundle key for the display name for a |
63 | * transliterator. The ID is appended to this to form the key. |
64 | * The resource bundle value should be a String. |
65 | */ |
66 | static const char RB_DISPLAY_NAME_PREFIX[] = "%Translit%%" ; |
67 | |
68 | /** |
69 | * Prefix for resource bundle key for the display name for a |
70 | * transliterator SCRIPT. The ID is appended to this to form the key. |
71 | * The resource bundle value should be a String. |
72 | */ |
73 | static const char RB_SCRIPT_DISPLAY_NAME_PREFIX[] = "%Translit%" ; |
74 | |
75 | /** |
76 | * Resource bundle key for display name pattern. |
77 | * The resource bundle value should be a String forming a |
78 | * MessageFormat pattern, e.g.: |
79 | * "{0,choice,0#|1#{1} Transliterator|2#{1} to {2} Transliterator}". |
80 | */ |
81 | static const char RB_DISPLAY_NAME_PATTERN[] = "TransliteratorNamePattern" ; |
82 | |
83 | /** |
84 | * Resource bundle key for the list of RuleBasedTransliterator IDs. |
85 | * The resource bundle value should be a String[] with each element |
86 | * being a valid ID. The ID will be appended to RB_RULE_BASED_PREFIX |
87 | * to obtain the class name in which the RB_RULE key will be sought. |
88 | */ |
89 | static const char RB_RULE_BASED_IDS[] = "RuleBasedTransliteratorIDs" ; |
90 | |
91 | /** |
92 | * The mutex controlling access to registry object. |
93 | */ |
94 | static icu::UMutex registryMutex; |
95 | |
96 | /** |
97 | * System transliterator registry; non-null when initialized. |
98 | */ |
99 | static icu::TransliteratorRegistry* registry = 0; |
100 | |
101 | // Macro to check/initialize the registry. ONLY USE WITHIN |
102 | // MUTEX. Avoids function call when registry is initialized. |
103 | #define HAVE_REGISTRY(status) (registry!=0 || initializeRegistry(status)) |
104 | |
105 | U_NAMESPACE_BEGIN |
106 | |
107 | UOBJECT_DEFINE_ABSTRACT_RTTI_IMPLEMENTATION(Transliterator) |
108 | |
109 | /** |
110 | * Return TRUE if the given UTransPosition is valid for text of |
111 | * the given length. |
112 | */ |
113 | static inline UBool positionIsValid(UTransPosition& index, int32_t len) { |
114 | return !(index.contextStart < 0 || |
115 | index.start < index.contextStart || |
116 | index.limit < index.start || |
117 | index.contextLimit < index.limit || |
118 | len < index.contextLimit); |
119 | } |
120 | |
121 | /** |
122 | * Default constructor. |
123 | * @param theID the string identifier for this transliterator |
124 | * @param theFilter the filter. Any character for which |
125 | * <tt>filter.contains()</tt> returns <tt>FALSE</tt> will not be |
126 | * altered by this transliterator. If <tt>filter</tt> is |
127 | * <tt>null</tt> then no filtering is applied. |
128 | */ |
129 | Transliterator::Transliterator(const UnicodeString& theID, |
130 | UnicodeFilter* adoptedFilter) : |
131 | UObject(), ID(theID), filter(adoptedFilter), |
132 | maximumContextLength(0) |
133 | { |
134 | // NUL-terminate the ID string, which is a non-aliased copy. |
135 | ID.append((UChar)0); |
136 | ID.truncate(ID.length()-1); |
137 | } |
138 | |
139 | /** |
140 | * Destructor. |
141 | */ |
142 | Transliterator::~Transliterator() { |
143 | if (filter) { |
144 | delete filter; |
145 | } |
146 | } |
147 | |
148 | /** |
149 | * Copy constructor. |
150 | */ |
151 | Transliterator::Transliterator(const Transliterator& other) : |
152 | UObject(other), ID(other.ID), filter(0), |
153 | maximumContextLength(other.maximumContextLength) |
154 | { |
155 | // NUL-terminate the ID string, which is a non-aliased copy. |
156 | ID.append((UChar)0); |
157 | ID.truncate(ID.length()-1); |
158 | |
159 | if (other.filter != 0) { |
160 | // We own the filter, so we must have our own copy |
161 | filter = other.filter->clone(); |
162 | } |
163 | } |
164 | |
165 | Transliterator* Transliterator::clone() const { |
166 | return NULL; |
167 | } |
168 | |
169 | /** |
170 | * Assignment operator. |
171 | */ |
172 | Transliterator& Transliterator::operator=(const Transliterator& other) { |
173 | ID = other.ID; |
174 | // NUL-terminate the ID string |
175 | ID.getTerminatedBuffer(); |
176 | |
177 | maximumContextLength = other.maximumContextLength; |
178 | adoptFilter((other.filter == 0) ? 0 : other.filter->clone()); |
179 | return *this; |
180 | } |
181 | |
182 | /** |
183 | * Transliterates a segment of a string. <code>Transliterator</code> API. |
184 | * @param text the string to be transliterated |
185 | * @param start the beginning index, inclusive; <code>0 <= start |
186 | * <= limit</code>. |
187 | * @param limit the ending index, exclusive; <code>start <= limit |
188 | * <= text.length()</code>. |
189 | * @return the new limit index, or -1 |
190 | */ |
191 | int32_t Transliterator::transliterate(Replaceable& text, |
192 | int32_t start, int32_t limit) const { |
193 | if (start < 0 || |
194 | limit < start || |
195 | text.length() < limit) { |
196 | return -1; |
197 | } |
198 | |
199 | UTransPosition offsets; |
200 | offsets.contextStart= start; |
201 | offsets.contextLimit = limit; |
202 | offsets.start = start; |
203 | offsets.limit = limit; |
204 | filteredTransliterate(text, offsets, FALSE, TRUE); |
205 | return offsets.limit; |
206 | } |
207 | |
208 | /** |
209 | * Transliterates an entire string in place. Convenience method. |
210 | * @param text the string to be transliterated |
211 | */ |
212 | void Transliterator::transliterate(Replaceable& text) const { |
213 | transliterate(text, 0, text.length()); |
214 | } |
215 | |
216 | /** |
217 | * Transliterates the portion of the text buffer that can be |
218 | * transliterated unambiguosly after new text has been inserted, |
219 | * typically as a result of a keyboard event. The new text in |
220 | * <code>insertion</code> will be inserted into <code>text</code> |
221 | * at <code>index.contextLimit</code>, advancing |
222 | * <code>index.contextLimit</code> by <code>insertion.length()</code>. |
223 | * Then the transliterator will try to transliterate characters of |
224 | * <code>text</code> between <code>index.start</code> and |
225 | * <code>index.contextLimit</code>. Characters before |
226 | * <code>index.start</code> will not be changed. |
227 | * |
228 | * <p>Upon return, values in <code>index</code> will be updated. |
229 | * <code>index.contextStart</code> will be advanced to the first |
230 | * character that future calls to this method will read. |
231 | * <code>index.start</code> and <code>index.contextLimit</code> will |
232 | * be adjusted to delimit the range of text that future calls to |
233 | * this method may change. |
234 | * |
235 | * <p>Typical usage of this method begins with an initial call |
236 | * with <code>index.contextStart</code> and <code>index.contextLimit</code> |
237 | * set to indicate the portion of <code>text</code> to be |
238 | * transliterated, and <code>index.start == index.contextStart</code>. |
239 | * Thereafter, <code>index</code> can be used without |
240 | * modification in future calls, provided that all changes to |
241 | * <code>text</code> are made via this method. |
242 | * |
243 | * <p>This method assumes that future calls may be made that will |
244 | * insert new text into the buffer. As a result, it only performs |
245 | * unambiguous transliterations. After the last call to this |
246 | * method, there may be untransliterated text that is waiting for |
247 | * more input to resolve an ambiguity. In order to perform these |
248 | * pending transliterations, clients should call {@link |
249 | * #finishKeyboardTransliteration} after the last call to this |
250 | * method has been made. |
251 | * |
252 | * @param text the buffer holding transliterated and untransliterated text |
253 | * @param index an array of three integers. |
254 | * |
255 | * <ul><li><code>index.contextStart</code>: the beginning index, |
256 | * inclusive; <code>0 <= index.contextStart <= index.contextLimit</code>. |
257 | * |
258 | * <li><code>index.contextLimit</code>: the ending index, exclusive; |
259 | * <code>index.contextStart <= index.contextLimit <= text.length()</code>. |
260 | * <code>insertion</code> is inserted at |
261 | * <code>index.contextLimit</code>. |
262 | * |
263 | * <li><code>index.start</code>: the next character to be |
264 | * considered for transliteration; <code>index.contextStart <= |
265 | * index.start <= index.contextLimit</code>. Characters before |
266 | * <code>index.start</code> will not be changed by future calls |
267 | * to this method.</ul> |
268 | * |
269 | * @param insertion text to be inserted and possibly |
270 | * transliterated into the translation buffer at |
271 | * <code>index.contextLimit</code>. If <code>null</code> then no text |
272 | * is inserted. |
273 | * @see #START |
274 | * @see #LIMIT |
275 | * @see #CURSOR |
276 | * @see #handleTransliterate |
277 | * @exception IllegalArgumentException if <code>index</code> |
278 | * is invalid |
279 | */ |
280 | void Transliterator::transliterate(Replaceable& text, |
281 | UTransPosition& index, |
282 | const UnicodeString& insertion, |
283 | UErrorCode &status) const { |
284 | _transliterate(text, index, &insertion, status); |
285 | } |
286 | |
287 | /** |
288 | * Transliterates the portion of the text buffer that can be |
289 | * transliterated unambiguosly after a new character has been |
290 | * inserted, typically as a result of a keyboard event. This is a |
291 | * convenience method; see {@link |
292 | * #transliterate(Replaceable, int[], String)} for details. |
293 | * @param text the buffer holding transliterated and |
294 | * untransliterated text |
295 | * @param index an array of three integers. See {@link |
296 | * #transliterate(Replaceable, int[], String)}. |
297 | * @param insertion text to be inserted and possibly |
298 | * transliterated into the translation buffer at |
299 | * <code>index.contextLimit</code>. |
300 | * @see #transliterate(Replaceable, int[], String) |
301 | */ |
302 | void Transliterator::transliterate(Replaceable& text, |
303 | UTransPosition& index, |
304 | UChar32 insertion, |
305 | UErrorCode& status) const { |
306 | UnicodeString str(insertion); |
307 | _transliterate(text, index, &str, status); |
308 | } |
309 | |
310 | /** |
311 | * Transliterates the portion of the text buffer that can be |
312 | * transliterated unambiguosly. This is a convenience method; see |
313 | * {@link #transliterate(Replaceable, int[], String)} for |
314 | * details. |
315 | * @param text the buffer holding transliterated and |
316 | * untransliterated text |
317 | * @param index an array of three integers. See {@link |
318 | * #transliterate(Replaceable, int[], String)}. |
319 | * @see #transliterate(Replaceable, int[], String) |
320 | */ |
321 | void Transliterator::transliterate(Replaceable& text, |
322 | UTransPosition& index, |
323 | UErrorCode& status) const { |
324 | _transliterate(text, index, 0, status); |
325 | } |
326 | |
327 | /** |
328 | * Finishes any pending transliterations that were waiting for |
329 | * more characters. Clients should call this method as the last |
330 | * call after a sequence of one or more calls to |
331 | * <code>transliterate()</code>. |
332 | * @param text the buffer holding transliterated and |
333 | * untransliterated text. |
334 | * @param index the array of indices previously passed to {@link |
335 | * #transliterate} |
336 | */ |
337 | void Transliterator::finishTransliteration(Replaceable& text, |
338 | UTransPosition& index) const { |
339 | if (!positionIsValid(index, text.length())) { |
340 | return; |
341 | } |
342 | |
343 | filteredTransliterate(text, index, FALSE, TRUE); |
344 | } |
345 | |
346 | /** |
347 | * This internal method does keyboard transliteration. If the |
348 | * 'insertion' is non-null then we append it to 'text' before |
349 | * proceeding. This method calls through to the pure virtual |
350 | * framework method handleTransliterate() to do the actual |
351 | * work. |
352 | */ |
353 | void Transliterator::_transliterate(Replaceable& text, |
354 | UTransPosition& index, |
355 | const UnicodeString* insertion, |
356 | UErrorCode &status) const { |
357 | if (U_FAILURE(status)) { |
358 | return; |
359 | } |
360 | |
361 | if (!positionIsValid(index, text.length())) { |
362 | status = U_ILLEGAL_ARGUMENT_ERROR; |
363 | return; |
364 | } |
365 | |
366 | // int32_t originalStart = index.contextStart; |
367 | if (insertion != 0) { |
368 | text.handleReplaceBetween(index.limit, index.limit, *insertion); |
369 | index.limit += insertion->length(); |
370 | index.contextLimit += insertion->length(); |
371 | } |
372 | |
373 | if (index.limit > 0 && |
374 | U16_IS_LEAD(text.charAt(index.limit - 1))) { |
375 | // Oops, there is a dangling lead surrogate in the buffer. |
376 | // This will break most transliterators, since they will |
377 | // assume it is part of a pair. Don't transliterate until |
378 | // more text comes in. |
379 | return; |
380 | } |
381 | |
382 | filteredTransliterate(text, index, TRUE, TRUE); |
383 | |
384 | #if 0 |
385 | // TODO |
386 | // I CAN'T DO what I'm attempting below now that the Kleene star |
387 | // operator is supported. For example, in the rule |
388 | |
389 | // ([:Lu:]+) { x } > $1; |
390 | |
391 | // what is the maximum context length? getMaximumContextLength() |
392 | // will return 1, but this is just the length of the ante context |
393 | // part of the pattern string -- 1 character, which is a standin |
394 | // for a Quantifier, which contains a StringMatcher, which |
395 | // contains a UnicodeSet. |
396 | |
397 | // There is a complicated way to make this work again, and that's |
398 | // to add a "maximum left context" protocol into the |
399 | // UnicodeMatcher hierarchy. At present I'm not convinced this is |
400 | // worth it. |
401 | |
402 | // --- |
403 | |
404 | // The purpose of the code below is to keep the context small |
405 | // while doing incremental transliteration. When part of the left |
406 | // context (between contextStart and start) is no longer needed, |
407 | // we try to advance contextStart past that portion. We use the |
408 | // maximum context length to do so. |
409 | int32_t newCS = index.start; |
410 | int32_t n = getMaximumContextLength(); |
411 | while (newCS > originalStart && n-- > 0) { |
412 | --newCS; |
413 | newCS -= U16_LENGTH(text.char32At(newCS)) - 1; |
414 | } |
415 | index.contextStart = uprv_max(newCS, originalStart); |
416 | #endif |
417 | } |
418 | |
419 | /** |
420 | * This method breaks up the input text into runs of unfiltered |
421 | * characters. It passes each such run to |
422 | * <subclass>.handleTransliterate(). Subclasses that can handle the |
423 | * filter logic more efficiently themselves may override this method. |
424 | * |
425 | * All transliteration calls in this class go through this method. |
426 | */ |
427 | void Transliterator::filteredTransliterate(Replaceable& text, |
428 | UTransPosition& index, |
429 | UBool incremental, |
430 | UBool rollback) const { |
431 | // Short circuit path for transliterators with no filter in |
432 | // non-incremental mode. |
433 | if (filter == 0 && !rollback) { |
434 | handleTransliterate(text, index, incremental); |
435 | return; |
436 | } |
437 | |
438 | //---------------------------------------------------------------------- |
439 | // This method processes text in two groupings: |
440 | // |
441 | // RUNS -- A run is a contiguous group of characters which are contained |
442 | // in the filter for this transliterator (filter.contains(ch) == TRUE). |
443 | // Text outside of runs may appear as context but it is not modified. |
444 | // The start and limit Position values are narrowed to each run. |
445 | // |
446 | // PASSES (incremental only) -- To make incremental mode work correctly, |
447 | // each run is broken up into n passes, where n is the length (in code |
448 | // points) of the run. Each pass contains the first n characters. If a |
449 | // pass is completely transliterated, it is committed, and further passes |
450 | // include characters after the committed text. If a pass is blocked, |
451 | // and does not transliterate completely, then this method rolls back |
452 | // the changes made during the pass, extends the pass by one code point, |
453 | // and tries again. |
454 | //---------------------------------------------------------------------- |
455 | |
456 | // globalLimit is the limit value for the entire operation. We |
457 | // set index.limit to the end of each unfiltered run before |
458 | // calling handleTransliterate(), so we need to maintain the real |
459 | // value of index.limit here. After each transliteration, we |
460 | // update globalLimit for insertions or deletions that have |
461 | // happened. |
462 | int32_t globalLimit = index.limit; |
463 | |
464 | // If there is a non-null filter, then break the input text up. Say the |
465 | // input text has the form: |
466 | // xxxabcxxdefxx |
467 | // where 'x' represents a filtered character (filter.contains('x') == |
468 | // false). Then we break this up into: |
469 | // xxxabc xxdef xx |
470 | // Each pass through the loop consumes a run of filtered |
471 | // characters (which are ignored) and a subsequent run of |
472 | // unfiltered characters (which are transliterated). |
473 | |
474 | for (;;) { |
475 | |
476 | if (filter != NULL) { |
477 | // Narrow the range to be transliterated to the first segment |
478 | // of unfiltered characters at or after index.start. |
479 | |
480 | // Advance past filtered chars |
481 | UChar32 c; |
482 | while (index.start < globalLimit && |
483 | !filter->contains(c=text.char32At(index.start))) { |
484 | index.start += U16_LENGTH(c); |
485 | } |
486 | |
487 | // Find the end of this run of unfiltered chars |
488 | index.limit = index.start; |
489 | while (index.limit < globalLimit && |
490 | filter->contains(c=text.char32At(index.limit))) { |
491 | index.limit += U16_LENGTH(c); |
492 | } |
493 | } |
494 | |
495 | // Check to see if the unfiltered run is empty. This only |
496 | // happens at the end of the string when all the remaining |
497 | // characters are filtered. |
498 | if (index.limit == index.start) { |
499 | // assert(index.start == globalLimit); |
500 | break; |
501 | } |
502 | |
503 | // Is this run incremental? If there is additional |
504 | // filtered text (if limit < globalLimit) then we pass in |
505 | // an incremental value of FALSE to force the subclass to |
506 | // complete the transliteration for this run. |
507 | UBool isIncrementalRun = |
508 | (index.limit < globalLimit ? FALSE : incremental); |
509 | |
510 | int32_t delta; |
511 | |
512 | // Implement rollback. To understand the need for rollback, |
513 | // consider the following transliterator: |
514 | // |
515 | // "t" is "a > A;" |
516 | // "u" is "A > b;" |
517 | // "v" is a compound of "t; NFD; u" with a filter [:Ll:] |
518 | // |
519 | // Now apply "c" to the input text "a". The result is "b". But if |
520 | // the transliteration is done incrementally, then the NFD holds |
521 | // things up after "t" has already transformed "a" to "A". When |
522 | // finishTransliterate() is called, "A" is _not_ processed because |
523 | // it gets excluded by the [:Ll:] filter, and the end result is "A" |
524 | // -- incorrect. The problem is that the filter is applied to a |
525 | // partially-transliterated result, when we only want it to apply to |
526 | // input text. Although this example hinges on a compound |
527 | // transliterator containing NFD and a specific filter, it can |
528 | // actually happen with any transliterator which may do a partial |
529 | // transformation in incremental mode into characters outside its |
530 | // filter. |
531 | // |
532 | // To handle this, when in incremental mode we supply characters to |
533 | // handleTransliterate() in several passes. Each pass adds one more |
534 | // input character to the input text. That is, for input "ABCD", we |
535 | // first try "A", then "AB", then "ABC", and finally "ABCD". If at |
536 | // any point we block (upon return, start < limit) then we roll |
537 | // back. If at any point we complete the run (upon return start == |
538 | // limit) then we commit that run. |
539 | |
540 | if (rollback && isIncrementalRun) { |
541 | |
542 | int32_t runStart = index.start; |
543 | int32_t runLimit = index.limit; |
544 | int32_t runLength = runLimit - runStart; |
545 | |
546 | // Make a rollback copy at the end of the string |
547 | int32_t rollbackOrigin = text.length(); |
548 | text.copy(runStart, runLimit, rollbackOrigin); |
549 | |
550 | // Variables reflecting the commitment of completely |
551 | // transliterated text. passStart is the runStart, advanced |
552 | // past committed text. rollbackStart is the rollbackOrigin, |
553 | // advanced past rollback text that corresponds to committed |
554 | // text. |
555 | int32_t passStart = runStart; |
556 | int32_t rollbackStart = rollbackOrigin; |
557 | |
558 | // The limit for each pass; we advance by one code point with |
559 | // each iteration. |
560 | int32_t passLimit = index.start; |
561 | |
562 | // Total length, in 16-bit code units, of uncommitted text. |
563 | // This is the length to be rolled back. |
564 | int32_t uncommittedLength = 0; |
565 | |
566 | // Total delta (change in length) for all passes |
567 | int32_t totalDelta = 0; |
568 | |
569 | // PASS MAIN LOOP -- Start with a single character, and extend |
570 | // the text by one character at a time. Roll back partial |
571 | // transliterations and commit complete transliterations. |
572 | for (;;) { |
573 | // Length of additional code point, either one or two |
574 | int32_t charLength = U16_LENGTH(text.char32At(passLimit)); |
575 | passLimit += charLength; |
576 | if (passLimit > runLimit) { |
577 | break; |
578 | } |
579 | uncommittedLength += charLength; |
580 | |
581 | index.limit = passLimit; |
582 | |
583 | // Delegate to subclass for actual transliteration. Upon |
584 | // return, start will be updated to point after the |
585 | // transliterated text, and limit and contextLimit will be |
586 | // adjusted for length changes. |
587 | handleTransliterate(text, index, TRUE); |
588 | |
589 | delta = index.limit - passLimit; // change in length |
590 | |
591 | // We failed to completely transliterate this pass. |
592 | // Roll back the text. Indices remain unchanged; reset |
593 | // them where necessary. |
594 | if (index.start != index.limit) { |
595 | // Find the rollbackStart, adjusted for length changes |
596 | // and the deletion of partially transliterated text. |
597 | int32_t rs = rollbackStart + delta - (index.limit - passStart); |
598 | |
599 | // Delete the partially transliterated text |
600 | text.handleReplaceBetween(passStart, index.limit, UnicodeString()); |
601 | |
602 | // Copy the rollback text back |
603 | text.copy(rs, rs + uncommittedLength, passStart); |
604 | |
605 | // Restore indices to their original values |
606 | index.start = passStart; |
607 | index.limit = passLimit; |
608 | index.contextLimit -= delta; |
609 | } |
610 | |
611 | // We did completely transliterate this pass. Update the |
612 | // commit indices to record how far we got. Adjust indices |
613 | // for length change. |
614 | else { |
615 | // Move the pass indices past the committed text. |
616 | passStart = passLimit = index.start; |
617 | |
618 | // Adjust the rollbackStart for length changes and move |
619 | // it past the committed text. All characters we've |
620 | // processed to this point are committed now, so zero |
621 | // out the uncommittedLength. |
622 | rollbackStart += delta + uncommittedLength; |
623 | uncommittedLength = 0; |
624 | |
625 | // Adjust indices for length changes. |
626 | runLimit += delta; |
627 | totalDelta += delta; |
628 | } |
629 | } |
630 | |
631 | // Adjust overall limit and rollbackOrigin for insertions and |
632 | // deletions. Don't need to worry about contextLimit because |
633 | // handleTransliterate() maintains that. |
634 | rollbackOrigin += totalDelta; |
635 | globalLimit += totalDelta; |
636 | |
637 | // Delete the rollback copy |
638 | text.handleReplaceBetween(rollbackOrigin, rollbackOrigin + runLength, UnicodeString()); |
639 | |
640 | // Move start past committed text |
641 | index.start = passStart; |
642 | } |
643 | |
644 | else { |
645 | // Delegate to subclass for actual transliteration. |
646 | int32_t limit = index.limit; |
647 | handleTransliterate(text, index, isIncrementalRun); |
648 | delta = index.limit - limit; // change in length |
649 | |
650 | // In a properly written transliterator, start == limit after |
651 | // handleTransliterate() returns when incremental is false. |
652 | // Catch cases where the subclass doesn't do this, and throw |
653 | // an exception. (Just pinning start to limit is a bad idea, |
654 | // because what's probably happening is that the subclass |
655 | // isn't transliterating all the way to the end, and it should |
656 | // in non-incremental mode.) |
657 | if (!incremental && index.start != index.limit) { |
658 | // We can't throw an exception, so just fudge things |
659 | index.start = index.limit; |
660 | } |
661 | |
662 | // Adjust overall limit for insertions/deletions. Don't need |
663 | // to worry about contextLimit because handleTransliterate() |
664 | // maintains that. |
665 | globalLimit += delta; |
666 | } |
667 | |
668 | if (filter == NULL || isIncrementalRun) { |
669 | break; |
670 | } |
671 | |
672 | // If we did completely transliterate this |
673 | // run, then repeat with the next unfiltered run. |
674 | } |
675 | |
676 | // Start is valid where it is. Limit needs to be put back where |
677 | // it was, modulo adjustments for deletions/insertions. |
678 | index.limit = globalLimit; |
679 | } |
680 | |
681 | void Transliterator::filteredTransliterate(Replaceable& text, |
682 | UTransPosition& index, |
683 | UBool incremental) const { |
684 | filteredTransliterate(text, index, incremental, FALSE); |
685 | } |
686 | |
687 | /** |
688 | * Method for subclasses to use to set the maximum context length. |
689 | * @see #getMaximumContextLength |
690 | */ |
691 | void Transliterator::setMaximumContextLength(int32_t maxContextLength) { |
692 | maximumContextLength = maxContextLength; |
693 | } |
694 | |
695 | /** |
696 | * Returns a programmatic identifier for this transliterator. |
697 | * If this identifier is passed to <code>getInstance()</code>, it |
698 | * will return this object, if it has been registered. |
699 | * @see #registerInstance |
700 | * @see #getAvailableIDs |
701 | */ |
702 | const UnicodeString& Transliterator::getID(void) const { |
703 | return ID; |
704 | } |
705 | |
706 | /** |
707 | * Returns a name for this transliterator that is appropriate for |
708 | * display to the user in the default locale. See {@link |
709 | * #getDisplayName(Locale)} for details. |
710 | */ |
711 | UnicodeString& U_EXPORT2 Transliterator::getDisplayName(const UnicodeString& ID, |
712 | UnicodeString& result) { |
713 | return getDisplayName(ID, Locale::getDefault(), result); |
714 | } |
715 | |
716 | /** |
717 | * Returns a name for this transliterator that is appropriate for |
718 | * display to the user in the given locale. This name is taken |
719 | * from the locale resource data in the standard manner of the |
720 | * <code>java.text</code> package. |
721 | * |
722 | * <p>If no localized names exist in the system resource bundles, |
723 | * a name is synthesized using a localized |
724 | * <code>MessageFormat</code> pattern from the resource data. The |
725 | * arguments to this pattern are an integer followed by one or two |
726 | * strings. The integer is the number of strings, either 1 or 2. |
727 | * The strings are formed by splitting the ID for this |
728 | * transliterator at the first TARGET_SEP. If there is no TARGET_SEP, then the |
729 | * entire ID forms the only string. |
730 | * @param inLocale the Locale in which the display name should be |
731 | * localized. |
732 | * @see java.text.MessageFormat |
733 | */ |
734 | UnicodeString& U_EXPORT2 Transliterator::getDisplayName(const UnicodeString& id, |
735 | const Locale& inLocale, |
736 | UnicodeString& result) { |
737 | UErrorCode status = U_ZERO_ERROR; |
738 | |
739 | ResourceBundle bundle(U_ICUDATA_TRANSLIT, inLocale, status); |
740 | |
741 | // Suspend checking status until later... |
742 | |
743 | result.truncate(0); |
744 | |
745 | // Normalize the ID |
746 | UnicodeString source, target, variant; |
747 | UBool sawSource; |
748 | TransliteratorIDParser::IDtoSTV(id, source, target, variant, sawSource); |
749 | if (target.length() < 1) { |
750 | // No target; malformed id |
751 | return result; |
752 | } |
753 | if (variant.length() > 0) { // Change "Foo" to "/Foo" |
754 | variant.insert(0, VARIANT_SEP); |
755 | } |
756 | UnicodeString ID(source); |
757 | ID.append(TARGET_SEP).append(target).append(variant); |
758 | |
759 | // build the char* key |
760 | if (uprv_isInvariantUString(ID.getBuffer(), ID.length())) { |
761 | char key[200]; |
762 | uprv_strcpy(key, RB_DISPLAY_NAME_PREFIX); |
763 | int32_t length=(int32_t)uprv_strlen(RB_DISPLAY_NAME_PREFIX); |
764 | ID.extract(0, (int32_t)(sizeof(key)-length), key+length, (int32_t)(sizeof(key)-length), US_INV); |
765 | |
766 | // Try to retrieve a UnicodeString from the bundle. |
767 | UnicodeString resString = bundle.getStringEx(key, status); |
768 | |
769 | if (U_SUCCESS(status) && resString.length() != 0) { |
770 | return result = resString; // [sic] assign & return |
771 | } |
772 | |
773 | #if !UCONFIG_NO_FORMATTING |
774 | // We have failed to get a name from the locale data. This is |
775 | // typical, since most transliterators will not have localized |
776 | // name data. The next step is to retrieve the MessageFormat |
777 | // pattern from the locale data and to use it to synthesize the |
778 | // name from the ID. |
779 | |
780 | status = U_ZERO_ERROR; |
781 | resString = bundle.getStringEx(RB_DISPLAY_NAME_PATTERN, status); |
782 | |
783 | if (U_SUCCESS(status) && resString.length() != 0) { |
784 | MessageFormat msg(resString, inLocale, status); |
785 | // Suspend checking status until later... |
786 | |
787 | // We pass either 2 or 3 Formattable objects to msg. |
788 | Formattable args[3]; |
789 | int32_t nargs; |
790 | args[0].setLong(2); // # of args to follow |
791 | args[1].setString(source); |
792 | args[2].setString(target); |
793 | nargs = 3; |
794 | |
795 | // Use display names for the scripts, if they exist |
796 | UnicodeString s; |
797 | length=(int32_t)uprv_strlen(RB_SCRIPT_DISPLAY_NAME_PREFIX); |
798 | for (int j=1; j<=2; ++j) { |
799 | status = U_ZERO_ERROR; |
800 | uprv_strcpy(key, RB_SCRIPT_DISPLAY_NAME_PREFIX); |
801 | args[j].getString(s); |
802 | if (uprv_isInvariantUString(s.getBuffer(), s.length())) { |
803 | s.extract(0, sizeof(key)-length-1, key+length, (int32_t)sizeof(key)-length-1, US_INV); |
804 | |
805 | resString = bundle.getStringEx(key, status); |
806 | |
807 | if (U_SUCCESS(status)) { |
808 | args[j] = resString; |
809 | } |
810 | } |
811 | } |
812 | |
813 | status = U_ZERO_ERROR; |
814 | FieldPosition pos; // ignored by msg |
815 | msg.format(args, nargs, result, pos, status); |
816 | if (U_SUCCESS(status)) { |
817 | result.append(variant); |
818 | return result; |
819 | } |
820 | } |
821 | #endif |
822 | } |
823 | |
824 | // We should not reach this point unless there is something |
825 | // wrong with the build or the RB_DISPLAY_NAME_PATTERN has |
826 | // been deleted from the root RB_LOCALE_ELEMENTS resource. |
827 | result = ID; |
828 | return result; |
829 | } |
830 | |
831 | /** |
832 | * Returns the filter used by this transliterator, or <tt>null</tt> |
833 | * if this transliterator uses no filter. Caller musn't delete |
834 | * the result! |
835 | */ |
836 | const UnicodeFilter* Transliterator::getFilter(void) const { |
837 | return filter; |
838 | } |
839 | |
840 | /** |
841 | * Returns the filter used by this transliterator, or |
842 | * <tt>NULL</tt> if this transliterator uses no filter. The |
843 | * caller must eventually delete the result. After this call, |
844 | * this transliterator's filter is set to <tt>NULL</tt>. |
845 | */ |
846 | UnicodeFilter* Transliterator::orphanFilter(void) { |
847 | UnicodeFilter *result = filter; |
848 | filter = NULL; |
849 | return result; |
850 | } |
851 | |
852 | /** |
853 | * Changes the filter used by this transliterator. If the filter |
854 | * is set to <tt>null</tt> then no filtering will occur. |
855 | * |
856 | * <p>Callers must take care if a transliterator is in use by |
857 | * multiple threads. The filter should not be changed by one |
858 | * thread while another thread may be transliterating. |
859 | */ |
860 | void Transliterator::adoptFilter(UnicodeFilter* filterToAdopt) { |
861 | delete filter; |
862 | filter = filterToAdopt; |
863 | } |
864 | |
865 | /** |
866 | * Returns this transliterator's inverse. See the class |
867 | * documentation for details. This implementation simply inverts |
868 | * the two entities in the ID and attempts to retrieve the |
869 | * resulting transliterator. That is, if <code>getID()</code> |
870 | * returns "A-B", then this method will return the result of |
871 | * <code>getInstance("B-A")</code>, or <code>null</code> if that |
872 | * call fails. |
873 | * |
874 | * <p>This method does not take filtering into account. The |
875 | * returned transliterator will have no filter. |
876 | * |
877 | * <p>Subclasses with knowledge of their inverse may wish to |
878 | * override this method. |
879 | * |
880 | * @return a transliterator that is an inverse, not necessarily |
881 | * exact, of this transliterator, or <code>null</code> if no such |
882 | * transliterator is registered. |
883 | * @see #registerInstance |
884 | */ |
885 | Transliterator* Transliterator::createInverse(UErrorCode& status) const { |
886 | UParseError parseError; |
887 | return Transliterator::createInstance(ID, UTRANS_REVERSE,parseError,status); |
888 | } |
889 | |
890 | Transliterator* U_EXPORT2 |
891 | Transliterator::createInstance(const UnicodeString& ID, |
892 | UTransDirection dir, |
893 | UErrorCode& status) |
894 | { |
895 | UParseError parseError; |
896 | return createInstance(ID, dir, parseError, status); |
897 | } |
898 | |
899 | /** |
900 | * Returns a <code>Transliterator</code> object given its ID. |
901 | * The ID must be either a system transliterator ID or a ID registered |
902 | * using <code>registerInstance()</code>. |
903 | * |
904 | * @param ID a valid ID, as enumerated by <code>getAvailableIDs()</code> |
905 | * @return A <code>Transliterator</code> object with the given ID |
906 | * @see #registerInstance |
907 | * @see #getAvailableIDs |
908 | * @see #getID |
909 | */ |
910 | Transliterator* U_EXPORT2 |
911 | Transliterator::createInstance(const UnicodeString& ID, |
912 | UTransDirection dir, |
913 | UParseError& parseError, |
914 | UErrorCode& status) |
915 | { |
916 | if (U_FAILURE(status)) { |
917 | return 0; |
918 | } |
919 | |
920 | UnicodeString canonID; |
921 | UVector list(status); |
922 | if (U_FAILURE(status)) { |
923 | return NULL; |
924 | } |
925 | |
926 | UnicodeSet* globalFilter = nullptr; |
927 | // TODO add code for parseError...currently unused, but |
928 | // later may be used by parsing code... |
929 | if (!TransliteratorIDParser::parseCompoundID(ID, dir, canonID, list, globalFilter)) { |
930 | status = U_INVALID_ID; |
931 | delete globalFilter; |
932 | return NULL; |
933 | } |
934 | LocalPointer<UnicodeSet> lpGlobalFilter(globalFilter); |
935 | |
936 | TransliteratorIDParser::instantiateList(list, status); |
937 | if (U_FAILURE(status)) { |
938 | return NULL; |
939 | } |
940 | |
941 | U_ASSERT(list.size() > 0); |
942 | Transliterator* t = NULL; |
943 | |
944 | if (list.size() > 1 || canonID.indexOf(ID_DELIM) >= 0) { |
945 | // [NOTE: If it's a compoundID, we instantiate a CompoundTransliterator even if it only |
946 | // has one child transliterator. This is so that toRules() will return the right thing |
947 | // (without any inactive ID), but our main ID still comes out correct. That is, if we |
948 | // instantiate "(Lower);Latin-Greek;", we want the rules to come out as "::Latin-Greek;" |
949 | // even though the ID is "(Lower);Latin-Greek;". |
950 | t = new CompoundTransliterator(list, parseError, status); |
951 | } |
952 | else { |
953 | t = (Transliterator*)list.elementAt(0); |
954 | } |
955 | // Check null pointer |
956 | if (t != NULL) { |
957 | t->setID(canonID); |
958 | if (lpGlobalFilter.isValid()) { |
959 | t->adoptFilter(lpGlobalFilter.orphan()); |
960 | } |
961 | } |
962 | else if (U_SUCCESS(status)) { |
963 | status = U_MEMORY_ALLOCATION_ERROR; |
964 | } |
965 | return t; |
966 | } |
967 | |
968 | /** |
969 | * Create a transliterator from a basic ID. This is an ID |
970 | * containing only the forward direction source, target, and |
971 | * variant. |
972 | * @param id a basic ID of the form S-T or S-T/V. |
973 | * @return a newly created Transliterator or null if the ID is |
974 | * invalid. |
975 | */ |
976 | Transliterator* Transliterator::createBasicInstance(const UnicodeString& id, |
977 | const UnicodeString* canon) { |
978 | UParseError pe; |
979 | UErrorCode ec = U_ZERO_ERROR; |
980 | TransliteratorAlias* alias = 0; |
981 | Transliterator* t = 0; |
982 | |
983 | umtx_lock(®istryMutex); |
984 | if (HAVE_REGISTRY(ec)) { |
985 | t = registry->get(id, alias, ec); |
986 | } |
987 | umtx_unlock(®istryMutex); |
988 | |
989 | if (U_FAILURE(ec)) { |
990 | delete t; |
991 | delete alias; |
992 | return 0; |
993 | } |
994 | |
995 | // We may have not gotten a transliterator: Because we can't |
996 | // instantiate a transliterator from inside TransliteratorRegistry:: |
997 | // get() (that would deadlock), we sometimes pass back an alias. This |
998 | // contains the data we need to finish the instantiation outside the |
999 | // registry mutex. The alias may, in turn, generate another alias, so |
1000 | // we handle aliases in a loop. The max times through the loop is two. |
1001 | // [alan] |
1002 | while (alias != 0) { |
1003 | U_ASSERT(t==0); |
1004 | // Rule-based aliases are handled with TransliteratorAlias:: |
1005 | // parse(), followed by TransliteratorRegistry::reget(). |
1006 | // Other aliases are handled with TransliteratorAlias::create(). |
1007 | if (alias->isRuleBased()) { |
1008 | // Step 1. parse |
1009 | TransliteratorParser parser(ec); |
1010 | alias->parse(parser, pe, ec); |
1011 | delete alias; |
1012 | alias = 0; |
1013 | |
1014 | // Step 2. reget |
1015 | umtx_lock(®istryMutex); |
1016 | if (HAVE_REGISTRY(ec)) { |
1017 | t = registry->reget(id, parser, alias, ec); |
1018 | } |
1019 | umtx_unlock(®istryMutex); |
1020 | |
1021 | // Step 3. Loop back around! |
1022 | } else { |
1023 | t = alias->create(pe, ec); |
1024 | delete alias; |
1025 | alias = 0; |
1026 | break; |
1027 | } |
1028 | if (U_FAILURE(ec)) { |
1029 | delete t; |
1030 | delete alias; |
1031 | t = NULL; |
1032 | break; |
1033 | } |
1034 | } |
1035 | |
1036 | if (t != NULL && canon != NULL) { |
1037 | t->setID(*canon); |
1038 | } |
1039 | |
1040 | return t; |
1041 | } |
1042 | |
1043 | /** |
1044 | * Returns a <code>Transliterator</code> object constructed from |
1045 | * the given rule string. This will be a RuleBasedTransliterator, |
1046 | * if the rule string contains only rules, or a |
1047 | * CompoundTransliterator, if it contains ID blocks, or a |
1048 | * NullTransliterator, if it contains ID blocks which parse as |
1049 | * empty for the given direction. |
1050 | */ |
1051 | Transliterator* U_EXPORT2 |
1052 | Transliterator::createFromRules(const UnicodeString& ID, |
1053 | const UnicodeString& rules, |
1054 | UTransDirection dir, |
1055 | UParseError& parseError, |
1056 | UErrorCode& status) |
1057 | { |
1058 | Transliterator* t = NULL; |
1059 | |
1060 | TransliteratorParser parser(status); |
1061 | parser.parse(rules, dir, parseError, status); |
1062 | |
1063 | if (U_FAILURE(status)) { |
1064 | return 0; |
1065 | } |
1066 | |
1067 | // NOTE: The logic here matches that in TransliteratorRegistry. |
1068 | if (parser.idBlockVector.size() == 0 && parser.dataVector.size() == 0) { |
1069 | t = new NullTransliterator(); |
1070 | } |
1071 | else if (parser.idBlockVector.size() == 0 && parser.dataVector.size() == 1) { |
1072 | t = new RuleBasedTransliterator(ID, (TransliterationRuleData*)parser.dataVector.orphanElementAt(0), TRUE); |
1073 | } |
1074 | else if (parser.idBlockVector.size() == 1 && parser.dataVector.size() == 0) { |
1075 | // idBlock, no data -- this is an alias. The ID has |
1076 | // been munged from reverse into forward mode, if |
1077 | // necessary, so instantiate the ID in the forward |
1078 | // direction. |
1079 | if (parser.compoundFilter != NULL) { |
1080 | UnicodeString filterPattern; |
1081 | parser.compoundFilter->toPattern(filterPattern, FALSE); |
1082 | t = createInstance(filterPattern + UnicodeString(ID_DELIM) |
1083 | + *((UnicodeString*)parser.idBlockVector.elementAt(0)), UTRANS_FORWARD, parseError, status); |
1084 | } |
1085 | else |
1086 | t = createInstance(*((UnicodeString*)parser.idBlockVector.elementAt(0)), UTRANS_FORWARD, parseError, status); |
1087 | |
1088 | |
1089 | if (t != NULL) { |
1090 | t->setID(ID); |
1091 | } |
1092 | } |
1093 | else { |
1094 | UVector transliterators(status); |
1095 | int32_t passNumber = 1; |
1096 | |
1097 | int32_t limit = parser.idBlockVector.size(); |
1098 | if (parser.dataVector.size() > limit) |
1099 | limit = parser.dataVector.size(); |
1100 | |
1101 | for (int32_t i = 0; i < limit; i++) { |
1102 | if (i < parser.idBlockVector.size()) { |
1103 | UnicodeString* idBlock = (UnicodeString*)parser.idBlockVector.elementAt(i); |
1104 | if (!idBlock->isEmpty()) { |
1105 | Transliterator* temp = createInstance(*idBlock, UTRANS_FORWARD, parseError, status); |
1106 | if (U_FAILURE(status)) { |
1107 | delete temp; |
1108 | return nullptr; |
1109 | } |
1110 | if (temp != NULL && typeid(*temp) != typeid(NullTransliterator)) |
1111 | transliterators.addElement(temp, status); |
1112 | else |
1113 | delete temp; |
1114 | } |
1115 | } |
1116 | if (!parser.dataVector.isEmpty()) { |
1117 | TransliterationRuleData* data = (TransliterationRuleData*)parser.dataVector.orphanElementAt(0); |
1118 | // TODO: Should passNumber be turned into a decimal-string representation (1 -> "1")? |
1119 | RuleBasedTransliterator* temprbt = new RuleBasedTransliterator(UnicodeString(CompoundTransliterator::PASS_STRING) + UnicodeString(passNumber++), |
1120 | data, TRUE); |
1121 | // Check if NULL before adding it to transliterators to avoid future usage of NULL pointer. |
1122 | if (temprbt == NULL) { |
1123 | if (U_SUCCESS(status)) { |
1124 | status = U_MEMORY_ALLOCATION_ERROR; |
1125 | } |
1126 | return t; |
1127 | } |
1128 | transliterators.addElement(temprbt, status); |
1129 | } |
1130 | } |
1131 | |
1132 | t = new CompoundTransliterator(transliterators, passNumber - 1, parseError, status); |
1133 | // Null pointer check |
1134 | if (t != NULL) { |
1135 | t->setID(ID); |
1136 | t->adoptFilter(parser.orphanCompoundFilter()); |
1137 | } |
1138 | } |
1139 | if (U_SUCCESS(status) && t == NULL) { |
1140 | status = U_MEMORY_ALLOCATION_ERROR; |
1141 | } |
1142 | return t; |
1143 | } |
1144 | |
1145 | UnicodeString& Transliterator::toRules(UnicodeString& rulesSource, |
1146 | UBool escapeUnprintable) const { |
1147 | // The base class implementation of toRules munges the ID into |
1148 | // the correct format. That is: foo => ::foo |
1149 | if (escapeUnprintable) { |
1150 | rulesSource.truncate(0); |
1151 | UnicodeString id = getID(); |
1152 | for (int32_t i=0; i<id.length();) { |
1153 | UChar32 c = id.char32At(i); |
1154 | if (!ICU_Utility::escapeUnprintable(rulesSource, c)) { |
1155 | rulesSource.append(c); |
1156 | } |
1157 | i += U16_LENGTH(c); |
1158 | } |
1159 | } else { |
1160 | rulesSource = getID(); |
1161 | } |
1162 | // KEEP in sync with rbt_pars |
1163 | rulesSource.insert(0, UNICODE_STRING_SIMPLE("::" )); |
1164 | rulesSource.append(ID_DELIM); |
1165 | return rulesSource; |
1166 | } |
1167 | |
1168 | int32_t Transliterator::countElements() const { |
1169 | const CompoundTransliterator* ct = dynamic_cast<const CompoundTransliterator*>(this); |
1170 | return ct != NULL ? ct->getCount() : 0; |
1171 | } |
1172 | |
1173 | const Transliterator& Transliterator::getElement(int32_t index, UErrorCode& ec) const { |
1174 | if (U_FAILURE(ec)) { |
1175 | return *this; |
1176 | } |
1177 | const CompoundTransliterator* cpd = dynamic_cast<const CompoundTransliterator*>(this); |
1178 | int32_t n = (cpd == NULL) ? 1 : cpd->getCount(); |
1179 | if (index < 0 || index >= n) { |
1180 | ec = U_INDEX_OUTOFBOUNDS_ERROR; |
1181 | return *this; |
1182 | } else { |
1183 | return (n == 1) ? *this : cpd->getTransliterator(index); |
1184 | } |
1185 | } |
1186 | |
1187 | UnicodeSet& Transliterator::getSourceSet(UnicodeSet& result) const { |
1188 | handleGetSourceSet(result); |
1189 | if (filter != NULL) { |
1190 | UnicodeSet* filterSet = dynamic_cast<UnicodeSet*>(filter); |
1191 | UBool deleteFilterSet = FALSE; |
1192 | // Most, but not all filters will be UnicodeSets. Optimize for |
1193 | // the high-runner case. |
1194 | if (filterSet == NULL) { |
1195 | filterSet = new UnicodeSet(); |
1196 | // Check null pointer |
1197 | if (filterSet == NULL) { |
1198 | return result; |
1199 | } |
1200 | deleteFilterSet = TRUE; |
1201 | filter->addMatchSetTo(*filterSet); |
1202 | } |
1203 | result.retainAll(*filterSet); |
1204 | if (deleteFilterSet) { |
1205 | delete filterSet; |
1206 | } |
1207 | } |
1208 | return result; |
1209 | } |
1210 | |
1211 | void Transliterator::handleGetSourceSet(UnicodeSet& result) const { |
1212 | result.clear(); |
1213 | } |
1214 | |
1215 | UnicodeSet& Transliterator::getTargetSet(UnicodeSet& result) const { |
1216 | return result.clear(); |
1217 | } |
1218 | |
1219 | // For public consumption |
1220 | void U_EXPORT2 Transliterator::registerFactory(const UnicodeString& id, |
1221 | Transliterator::Factory factory, |
1222 | Transliterator::Token context) { |
1223 | Mutex lock(®istryMutex); |
1224 | UErrorCode ec = U_ZERO_ERROR; |
1225 | if (HAVE_REGISTRY(ec)) { |
1226 | _registerFactory(id, factory, context); |
1227 | } |
1228 | } |
1229 | |
1230 | // To be called only by Transliterator subclasses that are called |
1231 | // to register themselves by initializeRegistry(). |
1232 | void Transliterator::_registerFactory(const UnicodeString& id, |
1233 | Transliterator::Factory factory, |
1234 | Transliterator::Token context) { |
1235 | UErrorCode ec = U_ZERO_ERROR; |
1236 | registry->put(id, factory, context, TRUE, ec); |
1237 | } |
1238 | |
1239 | // To be called only by Transliterator subclasses that are called |
1240 | // to register themselves by initializeRegistry(). |
1241 | void Transliterator::_registerSpecialInverse(const UnicodeString& target, |
1242 | const UnicodeString& inverseTarget, |
1243 | UBool bidirectional) { |
1244 | UErrorCode status = U_ZERO_ERROR; |
1245 | TransliteratorIDParser::registerSpecialInverse(target, inverseTarget, bidirectional, status); |
1246 | } |
1247 | |
1248 | /** |
1249 | * Registers a instance <tt>obj</tt> of a subclass of |
1250 | * <code>Transliterator</code> with the system. This object must |
1251 | * implement the <tt>clone()</tt> method. When |
1252 | * <tt>getInstance()</tt> is called with an ID string that is |
1253 | * equal to <tt>obj.getID()</tt>, then <tt>obj.clone()</tt> is |
1254 | * returned. |
1255 | * |
1256 | * @param obj an instance of subclass of |
1257 | * <code>Transliterator</code> that defines <tt>clone()</tt> |
1258 | * @see #getInstance |
1259 | * @see #unregister |
1260 | */ |
1261 | void U_EXPORT2 Transliterator::registerInstance(Transliterator* adoptedPrototype) { |
1262 | Mutex lock(®istryMutex); |
1263 | UErrorCode ec = U_ZERO_ERROR; |
1264 | if (HAVE_REGISTRY(ec)) { |
1265 | _registerInstance(adoptedPrototype); |
1266 | } |
1267 | } |
1268 | |
1269 | void Transliterator::_registerInstance(Transliterator* adoptedPrototype) { |
1270 | UErrorCode ec = U_ZERO_ERROR; |
1271 | registry->put(adoptedPrototype, TRUE, ec); |
1272 | } |
1273 | |
1274 | void U_EXPORT2 Transliterator::registerAlias(const UnicodeString& aliasID, |
1275 | const UnicodeString& realID) { |
1276 | Mutex lock(®istryMutex); |
1277 | UErrorCode ec = U_ZERO_ERROR; |
1278 | if (HAVE_REGISTRY(ec)) { |
1279 | _registerAlias(aliasID, realID); |
1280 | } |
1281 | } |
1282 | |
1283 | void Transliterator::_registerAlias(const UnicodeString& aliasID, |
1284 | const UnicodeString& realID) { |
1285 | UErrorCode ec = U_ZERO_ERROR; |
1286 | registry->put(aliasID, realID, FALSE, TRUE, ec); |
1287 | } |
1288 | |
1289 | /** |
1290 | * Unregisters a transliterator or class. This may be either |
1291 | * a system transliterator or a user transliterator or class. |
1292 | * |
1293 | * @param ID the ID of the transliterator or class |
1294 | * @see #registerInstance |
1295 | |
1296 | */ |
1297 | void U_EXPORT2 Transliterator::unregister(const UnicodeString& ID) { |
1298 | Mutex lock(®istryMutex); |
1299 | UErrorCode ec = U_ZERO_ERROR; |
1300 | if (HAVE_REGISTRY(ec)) { |
1301 | registry->remove(ID); |
1302 | } |
1303 | } |
1304 | |
1305 | /** |
1306 | * == OBSOLETE - remove in ICU 3.4 == |
1307 | * Return the number of IDs currently registered with the system. |
1308 | * To retrieve the actual IDs, call getAvailableID(i) with |
1309 | * i from 0 to countAvailableIDs() - 1. |
1310 | */ |
1311 | int32_t U_EXPORT2 Transliterator::countAvailableIDs(void) { |
1312 | int32_t retVal = 0; |
1313 | Mutex lock(®istryMutex); |
1314 | UErrorCode ec = U_ZERO_ERROR; |
1315 | if (HAVE_REGISTRY(ec)) { |
1316 | retVal = registry->countAvailableIDs(); |
1317 | } |
1318 | return retVal; |
1319 | } |
1320 | |
1321 | /** |
1322 | * == OBSOLETE - remove in ICU 3.4 == |
1323 | * Return the index-th available ID. index must be between 0 |
1324 | * and countAvailableIDs() - 1, inclusive. If index is out of |
1325 | * range, the result of getAvailableID(0) is returned. |
1326 | */ |
1327 | const UnicodeString& U_EXPORT2 Transliterator::getAvailableID(int32_t index) { |
1328 | const UnicodeString* result = NULL; |
1329 | umtx_lock(®istryMutex); |
1330 | UErrorCode ec = U_ZERO_ERROR; |
1331 | if (HAVE_REGISTRY(ec)) { |
1332 | result = ®istry->getAvailableID(index); |
1333 | } |
1334 | umtx_unlock(®istryMutex); |
1335 | U_ASSERT(result != NULL); // fail if no registry |
1336 | return *result; |
1337 | } |
1338 | |
1339 | StringEnumeration* U_EXPORT2 Transliterator::getAvailableIDs(UErrorCode& ec) { |
1340 | if (U_FAILURE(ec)) return NULL; |
1341 | StringEnumeration* result = NULL; |
1342 | umtx_lock(®istryMutex); |
1343 | if (HAVE_REGISTRY(ec)) { |
1344 | result = registry->getAvailableIDs(); |
1345 | } |
1346 | umtx_unlock(®istryMutex); |
1347 | if (result == NULL) { |
1348 | ec = U_INTERNAL_TRANSLITERATOR_ERROR; |
1349 | } |
1350 | return result; |
1351 | } |
1352 | |
1353 | int32_t U_EXPORT2 Transliterator::countAvailableSources(void) { |
1354 | Mutex lock(®istryMutex); |
1355 | UErrorCode ec = U_ZERO_ERROR; |
1356 | return HAVE_REGISTRY(ec) ? _countAvailableSources() : 0; |
1357 | } |
1358 | |
1359 | UnicodeString& U_EXPORT2 Transliterator::getAvailableSource(int32_t index, |
1360 | UnicodeString& result) { |
1361 | Mutex lock(®istryMutex); |
1362 | UErrorCode ec = U_ZERO_ERROR; |
1363 | if (HAVE_REGISTRY(ec)) { |
1364 | _getAvailableSource(index, result); |
1365 | } |
1366 | return result; |
1367 | } |
1368 | |
1369 | int32_t U_EXPORT2 Transliterator::countAvailableTargets(const UnicodeString& source) { |
1370 | Mutex lock(®istryMutex); |
1371 | UErrorCode ec = U_ZERO_ERROR; |
1372 | return HAVE_REGISTRY(ec) ? _countAvailableTargets(source) : 0; |
1373 | } |
1374 | |
1375 | UnicodeString& U_EXPORT2 Transliterator::getAvailableTarget(int32_t index, |
1376 | const UnicodeString& source, |
1377 | UnicodeString& result) { |
1378 | Mutex lock(®istryMutex); |
1379 | UErrorCode ec = U_ZERO_ERROR; |
1380 | if (HAVE_REGISTRY(ec)) { |
1381 | _getAvailableTarget(index, source, result); |
1382 | } |
1383 | return result; |
1384 | } |
1385 | |
1386 | int32_t U_EXPORT2 Transliterator::countAvailableVariants(const UnicodeString& source, |
1387 | const UnicodeString& target) { |
1388 | Mutex lock(®istryMutex); |
1389 | UErrorCode ec = U_ZERO_ERROR; |
1390 | return HAVE_REGISTRY(ec) ? _countAvailableVariants(source, target) : 0; |
1391 | } |
1392 | |
1393 | UnicodeString& U_EXPORT2 Transliterator::getAvailableVariant(int32_t index, |
1394 | const UnicodeString& source, |
1395 | const UnicodeString& target, |
1396 | UnicodeString& result) { |
1397 | Mutex lock(®istryMutex); |
1398 | UErrorCode ec = U_ZERO_ERROR; |
1399 | if (HAVE_REGISTRY(ec)) { |
1400 | _getAvailableVariant(index, source, target, result); |
1401 | } |
1402 | return result; |
1403 | } |
1404 | |
1405 | int32_t Transliterator::_countAvailableSources(void) { |
1406 | return registry->countAvailableSources(); |
1407 | } |
1408 | |
1409 | UnicodeString& Transliterator::_getAvailableSource(int32_t index, |
1410 | UnicodeString& result) { |
1411 | return registry->getAvailableSource(index, result); |
1412 | } |
1413 | |
1414 | int32_t Transliterator::_countAvailableTargets(const UnicodeString& source) { |
1415 | return registry->countAvailableTargets(source); |
1416 | } |
1417 | |
1418 | UnicodeString& Transliterator::_getAvailableTarget(int32_t index, |
1419 | const UnicodeString& source, |
1420 | UnicodeString& result) { |
1421 | return registry->getAvailableTarget(index, source, result); |
1422 | } |
1423 | |
1424 | int32_t Transliterator::_countAvailableVariants(const UnicodeString& source, |
1425 | const UnicodeString& target) { |
1426 | return registry->countAvailableVariants(source, target); |
1427 | } |
1428 | |
1429 | UnicodeString& Transliterator::_getAvailableVariant(int32_t index, |
1430 | const UnicodeString& source, |
1431 | const UnicodeString& target, |
1432 | UnicodeString& result) { |
1433 | return registry->getAvailableVariant(index, source, target, result); |
1434 | } |
1435 | |
1436 | #ifdef U_USE_DEPRECATED_TRANSLITERATOR_API |
1437 | |
1438 | /** |
1439 | * Method for subclasses to use to obtain a character in the given |
1440 | * string, with filtering. |
1441 | * @deprecated the new architecture provides filtering at the top |
1442 | * level. This method will be removed Dec 31 2001. |
1443 | */ |
1444 | UChar Transliterator::filteredCharAt(const Replaceable& text, int32_t i) const { |
1445 | UChar c; |
1446 | const UnicodeFilter* localFilter = getFilter(); |
1447 | return (localFilter == 0) ? text.charAt(i) : |
1448 | (localFilter->contains(c = text.charAt(i)) ? c : (UChar)0xFFFE); |
1449 | } |
1450 | |
1451 | #endif |
1452 | |
1453 | /** |
1454 | * If the registry is initialized, return TRUE. If not, initialize it |
1455 | * and return TRUE. If the registry cannot be initialized, return |
1456 | * FALSE (rare). |
1457 | * |
1458 | * IMPORTANT: Upon entry, registryMutex must be LOCKED. The entire |
1459 | * initialization is done with the lock held. There is NO REASON to |
1460 | * unlock, since no other thread that is waiting on the registryMutex |
1461 | * cannot itself proceed until the registry is initialized. |
1462 | */ |
1463 | UBool Transliterator::initializeRegistry(UErrorCode &status) { |
1464 | if (registry != 0) { |
1465 | return TRUE; |
1466 | } |
1467 | |
1468 | registry = new TransliteratorRegistry(status); |
1469 | if (registry == 0 || U_FAILURE(status)) { |
1470 | delete registry; |
1471 | registry = 0; |
1472 | return FALSE; // can't create registry, no recovery |
1473 | } |
1474 | |
1475 | /* The following code parses the index table located in |
1476 | * icu/data/translit/root.txt. The index is an n x 4 table |
1477 | * that follows this format: |
1478 | * <id>{ |
1479 | * file{ |
1480 | * resource{"<resource>"} |
1481 | * direction{"<direction>"} |
1482 | * } |
1483 | * } |
1484 | * <id>{ |
1485 | * internal{ |
1486 | * resource{"<resource>"} |
1487 | * direction{"<direction"} |
1488 | * } |
1489 | * } |
1490 | * <id>{ |
1491 | * alias{"<getInstanceArg"} |
1492 | * } |
1493 | * <id> is the ID of the system transliterator being defined. These |
1494 | * are public IDs enumerated by Transliterator.getAvailableIDs(), |
1495 | * unless the second field is "internal". |
1496 | * |
1497 | * <resource> is a ResourceReader resource name. Currently these refer |
1498 | * to file names under com/ibm/text/resources. This string is passed |
1499 | * directly to ResourceReader, together with <encoding>. |
1500 | * |
1501 | * <direction> is either "FORWARD" or "REVERSE". |
1502 | * |
1503 | * <getInstanceArg> is a string to be passed directly to |
1504 | * Transliterator.getInstance(). The returned Transliterator object |
1505 | * then has its ID changed to <id> and is returned. |
1506 | * |
1507 | * The extra blank field on "alias" lines is to make the array square. |
1508 | */ |
1509 | //static const char translit_index[] = "translit_index"; |
1510 | |
1511 | UResourceBundle *bundle, *transIDs, *colBund; |
1512 | bundle = ures_open(U_ICUDATA_TRANSLIT, NULL/*open default locale*/, &status); |
1513 | transIDs = ures_getByKey(bundle, RB_RULE_BASED_IDS, 0, &status); |
1514 | const UnicodeString T_PART = UNICODE_STRING_SIMPLE("-t-" ); |
1515 | |
1516 | int32_t row, maxRows; |
1517 | if (U_SUCCESS(status)) { |
1518 | maxRows = ures_getSize(transIDs); |
1519 | for (row = 0; row < maxRows; row++) { |
1520 | colBund = ures_getByIndex(transIDs, row, 0, &status); |
1521 | if (U_SUCCESS(status)) { |
1522 | UnicodeString id(ures_getKey(colBund), -1, US_INV); |
1523 | if(id.indexOf(T_PART) != -1) { |
1524 | ures_close(colBund); |
1525 | continue; |
1526 | } |
1527 | UResourceBundle* res = ures_getNextResource(colBund, NULL, &status); |
1528 | const char* typeStr = ures_getKey(res); |
1529 | UChar type; |
1530 | u_charsToUChars(typeStr, &type, 1); |
1531 | |
1532 | if (U_SUCCESS(status)) { |
1533 | int32_t len = 0; |
1534 | const UChar *resString; |
1535 | switch (type) { |
1536 | case 0x66: // 'f' |
1537 | case 0x69: // 'i' |
1538 | // 'file' or 'internal'; |
1539 | // row[2]=resource, row[3]=direction |
1540 | { |
1541 | |
1542 | resString = ures_getStringByKey(res, "resource" , &len, &status); |
1543 | UBool visible = (type == 0x0066 /*f*/); |
1544 | UTransDirection dir = |
1545 | (ures_getUnicodeStringByKey(res, "direction" , &status).charAt(0) == |
1546 | 0x0046 /*F*/) ? |
1547 | UTRANS_FORWARD : UTRANS_REVERSE; |
1548 | registry->put(id, UnicodeString(TRUE, resString, len), dir, TRUE, visible, status); |
1549 | } |
1550 | break; |
1551 | case 0x61: // 'a' |
1552 | // 'alias'; row[2]=createInstance argument |
1553 | resString = ures_getString(res, &len, &status); |
1554 | registry->put(id, UnicodeString(TRUE, resString, len), TRUE, TRUE, status); |
1555 | break; |
1556 | } |
1557 | } |
1558 | ures_close(res); |
1559 | } |
1560 | ures_close(colBund); |
1561 | } |
1562 | } |
1563 | |
1564 | ures_close(transIDs); |
1565 | ures_close(bundle); |
1566 | |
1567 | // Manually add prototypes that the system knows about to the |
1568 | // cache. This is how new non-rule-based transliterators are |
1569 | // added to the system. |
1570 | |
1571 | // This is to allow for null pointer check |
1572 | NullTransliterator* tempNullTranslit = new NullTransliterator(); |
1573 | LowercaseTransliterator* tempLowercaseTranslit = new LowercaseTransliterator(); |
1574 | UppercaseTransliterator* tempUppercaseTranslit = new UppercaseTransliterator(); |
1575 | TitlecaseTransliterator* tempTitlecaseTranslit = new TitlecaseTransliterator(); |
1576 | UnicodeNameTransliterator* tempUnicodeTranslit = new UnicodeNameTransliterator(); |
1577 | NameUnicodeTransliterator* tempNameUnicodeTranslit = new NameUnicodeTransliterator(); |
1578 | #if !UCONFIG_NO_BREAK_ITERATION |
1579 | // TODO: could or should these transliterators be referenced polymorphically once constructed? |
1580 | BreakTransliterator* tempBreakTranslit = new BreakTransliterator(); |
1581 | #endif |
1582 | // Check for null pointers |
1583 | if (tempNullTranslit == NULL || tempLowercaseTranslit == NULL || tempUppercaseTranslit == NULL || |
1584 | tempTitlecaseTranslit == NULL || tempUnicodeTranslit == NULL || |
1585 | #if !UCONFIG_NO_BREAK_ITERATION |
1586 | tempBreakTranslit == NULL || |
1587 | #endif |
1588 | tempNameUnicodeTranslit == NULL ) |
1589 | { |
1590 | delete tempNullTranslit; |
1591 | delete tempLowercaseTranslit; |
1592 | delete tempUppercaseTranslit; |
1593 | delete tempTitlecaseTranslit; |
1594 | delete tempUnicodeTranslit; |
1595 | delete tempNameUnicodeTranslit; |
1596 | #if !UCONFIG_NO_BREAK_ITERATION |
1597 | delete tempBreakTranslit; |
1598 | #endif |
1599 | // Since there was an error, remove registry |
1600 | delete registry; |
1601 | registry = NULL; |
1602 | |
1603 | status = U_MEMORY_ALLOCATION_ERROR; |
1604 | return 0; |
1605 | } |
1606 | |
1607 | registry->put(tempNullTranslit, TRUE, status); |
1608 | registry->put(tempLowercaseTranslit, TRUE, status); |
1609 | registry->put(tempUppercaseTranslit, TRUE, status); |
1610 | registry->put(tempTitlecaseTranslit, TRUE, status); |
1611 | registry->put(tempUnicodeTranslit, TRUE, status); |
1612 | registry->put(tempNameUnicodeTranslit, TRUE, status); |
1613 | #if !UCONFIG_NO_BREAK_ITERATION |
1614 | registry->put(tempBreakTranslit, FALSE, status); // FALSE means invisible. |
1615 | #endif |
1616 | |
1617 | RemoveTransliterator::registerIDs(); // Must be within mutex |
1618 | EscapeTransliterator::registerIDs(); |
1619 | UnescapeTransliterator::registerIDs(); |
1620 | NormalizationTransliterator::registerIDs(); |
1621 | AnyTransliterator::registerIDs(); |
1622 | |
1623 | _registerSpecialInverse(UNICODE_STRING_SIMPLE("Null" ), |
1624 | UNICODE_STRING_SIMPLE("Null" ), FALSE); |
1625 | _registerSpecialInverse(UNICODE_STRING_SIMPLE("Upper" ), |
1626 | UNICODE_STRING_SIMPLE("Lower" ), TRUE); |
1627 | _registerSpecialInverse(UNICODE_STRING_SIMPLE("Title" ), |
1628 | UNICODE_STRING_SIMPLE("Lower" ), FALSE); |
1629 | |
1630 | ucln_i18n_registerCleanup(UCLN_I18N_TRANSLITERATOR, utrans_transliterator_cleanup); |
1631 | |
1632 | return TRUE; |
1633 | } |
1634 | |
1635 | U_NAMESPACE_END |
1636 | |
1637 | // Defined in transreg.h: |
1638 | |
1639 | /** |
1640 | * Release all static memory held by transliterator. This will |
1641 | * necessarily invalidate any rule-based transliterators held by the |
1642 | * user, because RBTs hold pointers to common data objects. |
1643 | */ |
1644 | U_CFUNC UBool utrans_transliterator_cleanup(void) { |
1645 | U_NAMESPACE_USE |
1646 | TransliteratorIDParser::cleanup(); |
1647 | if (registry) { |
1648 | delete registry; |
1649 | registry = NULL; |
1650 | } |
1651 | return TRUE; |
1652 | } |
1653 | |
1654 | #endif /* #if !UCONFIG_NO_TRANSLITERATION */ |
1655 | |
1656 | //eof |
1657 | |