1 | // © 2016 and later: Unicode, Inc. and others. |
2 | // License & terms of use: http://www.unicode.org/copyright.html |
3 | /* |
4 | ********************************************************************** |
5 | * Copyright (c) 2001-2014, International Business Machines |
6 | * Corporation and others. All Rights Reserved. |
7 | ********************************************************************** |
8 | * Date Name Description |
9 | * 08/10/2001 aliu Creation. |
10 | ********************************************************************** |
11 | */ |
12 | |
13 | #include "unicode/utypes.h" |
14 | |
15 | #if !UCONFIG_NO_TRANSLITERATION |
16 | |
17 | #include "unicode/translit.h" |
18 | #include "unicode/resbund.h" |
19 | #include "unicode/uniset.h" |
20 | #include "unicode/uscript.h" |
21 | #include "rbt.h" |
22 | #include "cpdtrans.h" |
23 | #include "nultrans.h" |
24 | #include "transreg.h" |
25 | #include "rbt_data.h" |
26 | #include "rbt_pars.h" |
27 | #include "tridpars.h" |
28 | #include "charstr.h" |
29 | #include "uassert.h" |
30 | #include "locutil.h" |
31 | |
32 | // Enable the following symbol to add debugging code that tracks the |
33 | // allocation, deletion, and use of Entry objects. BoundsChecker has |
34 | // reported dangling pointer errors with these objects, but I have |
35 | // been unable to confirm them. I suspect BoundsChecker is getting |
36 | // confused with pointers going into and coming out of a UHashtable, |
37 | // despite the hinting code that is designed to help it. |
38 | // #define DEBUG_MEM |
39 | #ifdef DEBUG_MEM |
40 | #include <stdio.h> |
41 | #endif |
42 | |
43 | // UChar constants |
44 | static const UChar LOCALE_SEP = 95; // '_' |
45 | //static const UChar ID_SEP = 0x002D; /*-*/ |
46 | //static const UChar VARIANT_SEP = 0x002F; // '/' |
47 | |
48 | // String constants |
49 | static const UChar ANY[] = { 0x41, 0x6E, 0x79, 0 }; // Any |
50 | static const UChar LAT[] = { 0x4C, 0x61, 0x74, 0 }; // Lat |
51 | |
52 | // empty string |
53 | #define NO_VARIANT UnicodeString() |
54 | |
55 | // initial estimate for specDAG size |
56 | // ICU 60 Transliterator::countAvailableSources() |
57 | #define SPECDAG_INIT_SIZE 149 |
58 | |
59 | // initial estimate for number of variant names |
60 | #define VARIANT_LIST_INIT_SIZE 11 |
61 | #define VARIANT_LIST_MAX_SIZE 31 |
62 | |
63 | // initial estimate for availableIDs count (default estimate is 8 => multiple reallocs) |
64 | // ICU 60 Transliterator::countAvailableIDs() |
65 | #define AVAILABLE_IDS_INIT_SIZE 641 |
66 | |
67 | // initial estimate for number of targets for source "Any", "Lat" |
68 | // ICU 60 Transliterator::countAvailableTargets("Any")/("Latn") |
69 | #define ANY_TARGETS_INIT_SIZE 125 |
70 | #define LAT_TARGETS_INIT_SIZE 23 |
71 | |
72 | /** |
73 | * Resource bundle key for the RuleBasedTransliterator rule. |
74 | */ |
75 | //static const char RB_RULE[] = "Rule"; |
76 | |
77 | U_NAMESPACE_BEGIN |
78 | |
79 | //------------------------------------------------------------------ |
80 | // Alias |
81 | //------------------------------------------------------------------ |
82 | |
83 | TransliteratorAlias::TransliteratorAlias(const UnicodeString& theAliasID, |
84 | const UnicodeSet* cpdFilter) : |
85 | ID(), |
86 | aliasesOrRules(theAliasID), |
87 | transes(0), |
88 | compoundFilter(cpdFilter), |
89 | direction(UTRANS_FORWARD), |
90 | type(TransliteratorAlias::SIMPLE) { |
91 | } |
92 | |
93 | TransliteratorAlias::TransliteratorAlias(const UnicodeString& theID, |
94 | const UnicodeString& idBlocks, |
95 | UVector* adoptedTransliterators, |
96 | const UnicodeSet* cpdFilter) : |
97 | ID(theID), |
98 | aliasesOrRules(idBlocks), |
99 | transes(adoptedTransliterators), |
100 | compoundFilter(cpdFilter), |
101 | direction(UTRANS_FORWARD), |
102 | type(TransliteratorAlias::COMPOUND) { |
103 | } |
104 | |
105 | TransliteratorAlias::TransliteratorAlias(const UnicodeString& theID, |
106 | const UnicodeString& rules, |
107 | UTransDirection dir) : |
108 | ID(theID), |
109 | aliasesOrRules(rules), |
110 | transes(0), |
111 | compoundFilter(0), |
112 | direction(dir), |
113 | type(TransliteratorAlias::RULES) { |
114 | } |
115 | |
116 | TransliteratorAlias::~TransliteratorAlias() { |
117 | delete transes; |
118 | } |
119 | |
120 | |
121 | Transliterator* TransliteratorAlias::create(UParseError& pe, |
122 | UErrorCode& ec) { |
123 | if (U_FAILURE(ec)) { |
124 | return 0; |
125 | } |
126 | Transliterator *t = NULL; |
127 | switch (type) { |
128 | case SIMPLE: |
129 | t = Transliterator::createInstance(aliasesOrRules, UTRANS_FORWARD, pe, ec); |
130 | if(U_FAILURE(ec)){ |
131 | return 0; |
132 | } |
133 | if (compoundFilter != 0) |
134 | t->adoptFilter(compoundFilter->clone()); |
135 | break; |
136 | case COMPOUND: |
137 | { |
138 | // the total number of transliterators in the compound is the total number of anonymous transliterators |
139 | // plus the total number of ID blocks-- we start by assuming the list begins and ends with an ID |
140 | // block and that each pair anonymous transliterators has an ID block between them. Then we go back |
141 | // to see whether there really are ID blocks at the beginning and end (by looking for U+FFFF, which |
142 | // marks the position where an anonymous transliterator goes) and adjust accordingly |
143 | int32_t anonymousRBTs = transes->size(); |
144 | int32_t transCount = anonymousRBTs * 2 + 1; |
145 | if (!aliasesOrRules.isEmpty() && aliasesOrRules[0] == (UChar)(0xffff)) |
146 | --transCount; |
147 | if (aliasesOrRules.length() >= 2 && aliasesOrRules[aliasesOrRules.length() - 1] == (UChar)(0xffff)) |
148 | --transCount; |
149 | UnicodeString noIDBlock((UChar)(0xffff)); |
150 | noIDBlock += ((UChar)(0xffff)); |
151 | int32_t pos = aliasesOrRules.indexOf(noIDBlock); |
152 | while (pos >= 0) { |
153 | --transCount; |
154 | pos = aliasesOrRules.indexOf(noIDBlock, pos + 1); |
155 | } |
156 | |
157 | UVector transliterators(ec); |
158 | UnicodeString idBlock; |
159 | int32_t blockSeparatorPos = aliasesOrRules.indexOf((UChar)(0xffff)); |
160 | while (blockSeparatorPos >= 0) { |
161 | aliasesOrRules.extract(0, blockSeparatorPos, idBlock); |
162 | aliasesOrRules.remove(0, blockSeparatorPos + 1); |
163 | if (!idBlock.isEmpty()) |
164 | transliterators.addElement(Transliterator::createInstance(idBlock, UTRANS_FORWARD, pe, ec), ec); |
165 | if (!transes->isEmpty()) |
166 | transliterators.addElement(transes->orphanElementAt(0), ec); |
167 | blockSeparatorPos = aliasesOrRules.indexOf((UChar)(0xffff)); |
168 | } |
169 | if (!aliasesOrRules.isEmpty()) |
170 | transliterators.addElement(Transliterator::createInstance(aliasesOrRules, UTRANS_FORWARD, pe, ec), ec); |
171 | while (!transes->isEmpty()) |
172 | transliterators.addElement(transes->orphanElementAt(0), ec); |
173 | |
174 | if (U_SUCCESS(ec)) { |
175 | t = new CompoundTransliterator(ID, transliterators, |
176 | (compoundFilter ? compoundFilter->clone() : nullptr), |
177 | anonymousRBTs, pe, ec); |
178 | if (t == 0) { |
179 | ec = U_MEMORY_ALLOCATION_ERROR; |
180 | return 0; |
181 | } |
182 | } else { |
183 | for (int32_t i = 0; i < transliterators.size(); i++) |
184 | delete (Transliterator*)(transliterators.elementAt(i)); |
185 | } |
186 | } |
187 | break; |
188 | case RULES: |
189 | UPRV_UNREACHABLE; // don't call create() if isRuleBased() returns TRUE! |
190 | } |
191 | return t; |
192 | } |
193 | |
194 | UBool TransliteratorAlias::isRuleBased() const { |
195 | return type == RULES; |
196 | } |
197 | |
198 | void TransliteratorAlias::parse(TransliteratorParser& parser, |
199 | UParseError& pe, UErrorCode& ec) const { |
200 | U_ASSERT(type == RULES); |
201 | if (U_FAILURE(ec)) { |
202 | return; |
203 | } |
204 | |
205 | parser.parse(aliasesOrRules, direction, pe, ec); |
206 | } |
207 | |
208 | //---------------------------------------------------------------------- |
209 | // class TransliteratorSpec |
210 | //---------------------------------------------------------------------- |
211 | |
212 | /** |
213 | * A TransliteratorSpec is a string specifying either a source or a target. In more |
214 | * general terms, it may also specify a variant, but we only use the |
215 | * Spec class for sources and targets. |
216 | * |
217 | * A Spec may be a locale or a script. If it is a locale, it has a |
218 | * fallback chain that goes xx_YY_ZZZ -> xx_YY -> xx -> ssss, where |
219 | * ssss is the script mapping of xx_YY_ZZZ. The Spec API methods |
220 | * hasFallback(), next(), and reset() iterate over this fallback |
221 | * sequence. |
222 | * |
223 | * The Spec class canonicalizes itself, so the locale is put into |
224 | * canonical form, or the script is transformed from an abbreviation |
225 | * to a full name. |
226 | */ |
227 | class TransliteratorSpec : public UMemory { |
228 | public: |
229 | TransliteratorSpec(const UnicodeString& spec); |
230 | ~TransliteratorSpec(); |
231 | |
232 | const UnicodeString& get() const; |
233 | UBool hasFallback() const; |
234 | const UnicodeString& next(); |
235 | void reset(); |
236 | |
237 | UBool isLocale() const; |
238 | ResourceBundle& getBundle() const; |
239 | |
240 | operator const UnicodeString&() const { return get(); } |
241 | const UnicodeString& getTop() const { return top; } |
242 | |
243 | private: |
244 | void setupNext(); |
245 | |
246 | UnicodeString top; |
247 | UnicodeString spec; |
248 | UnicodeString nextSpec; |
249 | UnicodeString scriptName; |
250 | UBool isSpecLocale; // TRUE if spec is a locale |
251 | UBool isNextLocale; // TRUE if nextSpec is a locale |
252 | ResourceBundle* res; |
253 | |
254 | TransliteratorSpec(const TransliteratorSpec &other); // forbid copying of this class |
255 | TransliteratorSpec &operator=(const TransliteratorSpec &other); // forbid copying of this class |
256 | }; |
257 | |
258 | TransliteratorSpec::TransliteratorSpec(const UnicodeString& theSpec) |
259 | : top(theSpec), |
260 | res(0) |
261 | { |
262 | UErrorCode status = U_ZERO_ERROR; |
263 | Locale topLoc("" ); |
264 | LocaleUtility::initLocaleFromName(theSpec, topLoc); |
265 | if (!topLoc.isBogus()) { |
266 | res = new ResourceBundle(U_ICUDATA_TRANSLIT, topLoc, status); |
267 | /* test for NULL */ |
268 | if (res == 0) { |
269 | return; |
270 | } |
271 | if (U_FAILURE(status) || status == U_USING_DEFAULT_WARNING) { |
272 | delete res; |
273 | res = 0; |
274 | } |
275 | } |
276 | |
277 | // Canonicalize script name -or- do locale->script mapping |
278 | status = U_ZERO_ERROR; |
279 | static const int32_t capacity = 10; |
280 | UScriptCode script[capacity]={USCRIPT_INVALID_CODE}; |
281 | int32_t num = uscript_getCode(CharString().appendInvariantChars(theSpec, status).data(), |
282 | script, capacity, &status); |
283 | if (num > 0 && script[0] != USCRIPT_INVALID_CODE) { |
284 | scriptName = UnicodeString(uscript_getName(script[0]), -1, US_INV); |
285 | } |
286 | |
287 | // Canonicalize top |
288 | if (res != 0) { |
289 | // Canonicalize locale name |
290 | UnicodeString locStr; |
291 | LocaleUtility::initNameFromLocale(topLoc, locStr); |
292 | if (!locStr.isBogus()) { |
293 | top = locStr; |
294 | } |
295 | } else if (scriptName.length() != 0) { |
296 | // We are a script; use canonical name |
297 | top = scriptName; |
298 | } |
299 | |
300 | // assert(spec != top); |
301 | reset(); |
302 | } |
303 | |
304 | TransliteratorSpec::~TransliteratorSpec() { |
305 | delete res; |
306 | } |
307 | |
308 | UBool TransliteratorSpec::hasFallback() const { |
309 | return nextSpec.length() != 0; |
310 | } |
311 | |
312 | void TransliteratorSpec::reset() { |
313 | if (spec != top) { |
314 | spec = top; |
315 | isSpecLocale = (res != 0); |
316 | setupNext(); |
317 | } |
318 | } |
319 | |
320 | void TransliteratorSpec::setupNext() { |
321 | isNextLocale = FALSE; |
322 | if (isSpecLocale) { |
323 | nextSpec = spec; |
324 | int32_t i = nextSpec.lastIndexOf(LOCALE_SEP); |
325 | // If i == 0 then we have _FOO, so we fall through |
326 | // to the scriptName. |
327 | if (i > 0) { |
328 | nextSpec.truncate(i); |
329 | isNextLocale = TRUE; |
330 | } else { |
331 | nextSpec = scriptName; // scriptName may be empty |
332 | } |
333 | } else { |
334 | // spec is a script, so we are at the end |
335 | nextSpec.truncate(0); |
336 | } |
337 | } |
338 | |
339 | // Protocol: |
340 | // for(const UnicodeString& s(spec.get()); |
341 | // spec.hasFallback(); s(spec.next())) { ... |
342 | |
343 | const UnicodeString& TransliteratorSpec::next() { |
344 | spec = nextSpec; |
345 | isSpecLocale = isNextLocale; |
346 | setupNext(); |
347 | return spec; |
348 | } |
349 | |
350 | const UnicodeString& TransliteratorSpec::get() const { |
351 | return spec; |
352 | } |
353 | |
354 | UBool TransliteratorSpec::isLocale() const { |
355 | return isSpecLocale; |
356 | } |
357 | |
358 | ResourceBundle& TransliteratorSpec::getBundle() const { |
359 | return *res; |
360 | } |
361 | |
362 | //---------------------------------------------------------------------- |
363 | |
364 | #ifdef DEBUG_MEM |
365 | |
366 | // Vector of Entry pointers currently in use |
367 | static UVector* DEBUG_entries = NULL; |
368 | |
369 | static void DEBUG_setup() { |
370 | if (DEBUG_entries == NULL) { |
371 | UErrorCode ec = U_ZERO_ERROR; |
372 | DEBUG_entries = new UVector(ec); |
373 | } |
374 | } |
375 | |
376 | // Caller must call DEBUG_setup first. Return index of given Entry, |
377 | // if it is in use (not deleted yet), or -1 if not found. |
378 | static int DEBUG_findEntry(TransliteratorEntry* e) { |
379 | for (int i=0; i<DEBUG_entries->size(); ++i) { |
380 | if (e == (TransliteratorEntry*) DEBUG_entries->elementAt(i)) { |
381 | return i; |
382 | } |
383 | } |
384 | return -1; |
385 | } |
386 | |
387 | // Track object creation |
388 | static void DEBUG_newEntry(TransliteratorEntry* e) { |
389 | DEBUG_setup(); |
390 | if (DEBUG_findEntry(e) >= 0) { |
391 | // This should really never happen unless the heap is broken |
392 | printf("ERROR DEBUG_newEntry duplicate new pointer %08X\n" , e); |
393 | return; |
394 | } |
395 | UErrorCode ec = U_ZERO_ERROR; |
396 | DEBUG_entries->addElement(e, ec); |
397 | } |
398 | |
399 | // Track object deletion |
400 | static void DEBUG_delEntry(TransliteratorEntry* e) { |
401 | DEBUG_setup(); |
402 | int i = DEBUG_findEntry(e); |
403 | if (i < 0) { |
404 | printf("ERROR DEBUG_delEntry possible double deletion %08X\n" , e); |
405 | return; |
406 | } |
407 | DEBUG_entries->removeElementAt(i); |
408 | } |
409 | |
410 | // Track object usage |
411 | static void DEBUG_useEntry(TransliteratorEntry* e) { |
412 | if (e == NULL) return; |
413 | DEBUG_setup(); |
414 | int i = DEBUG_findEntry(e); |
415 | if (i < 0) { |
416 | printf("ERROR DEBUG_useEntry possible dangling pointer %08X\n" , e); |
417 | } |
418 | } |
419 | |
420 | #else |
421 | // If we're not debugging then make these macros into NOPs |
422 | #define DEBUG_newEntry(x) |
423 | #define DEBUG_delEntry(x) |
424 | #define DEBUG_useEntry(x) |
425 | #endif |
426 | |
427 | //---------------------------------------------------------------------- |
428 | // class Entry |
429 | //---------------------------------------------------------------------- |
430 | |
431 | /** |
432 | * The Entry object stores objects of different types and |
433 | * singleton objects as placeholders for rule-based transliterators to |
434 | * be built as needed. Instances of this struct can be placeholders, |
435 | * can represent prototype transliterators to be cloned, or can |
436 | * represent TransliteratorData objects. We don't support storing |
437 | * classes in the registry because we don't have the rtti infrastructure |
438 | * for it. We could easily add this if there is a need for it in the |
439 | * future. |
440 | */ |
441 | class TransliteratorEntry : public UMemory { |
442 | public: |
443 | enum Type { |
444 | RULES_FORWARD, |
445 | RULES_REVERSE, |
446 | LOCALE_RULES, |
447 | PROTOTYPE, |
448 | RBT_DATA, |
449 | COMPOUND_RBT, |
450 | ALIAS, |
451 | FACTORY, |
452 | NONE // Only used for uninitialized entries |
453 | } entryType; |
454 | // NOTE: stringArg cannot go inside the union because |
455 | // it has a copy constructor |
456 | UnicodeString stringArg; // For RULES_*, ALIAS, COMPOUND_RBT |
457 | int32_t intArg; // For COMPOUND_RBT, LOCALE_RULES |
458 | UnicodeSet* compoundFilter; // For COMPOUND_RBT |
459 | union { |
460 | Transliterator* prototype; // For PROTOTYPE |
461 | TransliterationRuleData* data; // For RBT_DATA |
462 | UVector* dataVector; // For COMPOUND_RBT |
463 | struct { |
464 | Transliterator::Factory function; |
465 | Transliterator::Token context; |
466 | } factory; // For FACTORY |
467 | } u; |
468 | TransliteratorEntry(); |
469 | ~TransliteratorEntry(); |
470 | void adoptPrototype(Transliterator* adopted); |
471 | void setFactory(Transliterator::Factory factory, |
472 | Transliterator::Token context); |
473 | |
474 | private: |
475 | |
476 | TransliteratorEntry(const TransliteratorEntry &other); // forbid copying of this class |
477 | TransliteratorEntry &operator=(const TransliteratorEntry &other); // forbid copying of this class |
478 | }; |
479 | |
480 | TransliteratorEntry::TransliteratorEntry() { |
481 | u.prototype = 0; |
482 | compoundFilter = NULL; |
483 | entryType = NONE; |
484 | DEBUG_newEntry(this); |
485 | } |
486 | |
487 | TransliteratorEntry::~TransliteratorEntry() { |
488 | DEBUG_delEntry(this); |
489 | if (entryType == PROTOTYPE) { |
490 | delete u.prototype; |
491 | } else if (entryType == RBT_DATA) { |
492 | // The data object is shared between instances of RBT. The |
493 | // entry object owns it. It should only be deleted when the |
494 | // transliterator component is being cleaned up. Doing so |
495 | // invalidates any RBTs that the user has instantiated. |
496 | delete u.data; |
497 | } else if (entryType == COMPOUND_RBT) { |
498 | while (u.dataVector != NULL && !u.dataVector->isEmpty()) |
499 | delete (TransliterationRuleData*)u.dataVector->orphanElementAt(0); |
500 | delete u.dataVector; |
501 | } |
502 | delete compoundFilter; |
503 | } |
504 | |
505 | void TransliteratorEntry::adoptPrototype(Transliterator* adopted) { |
506 | if (entryType == PROTOTYPE) { |
507 | delete u.prototype; |
508 | } |
509 | entryType = PROTOTYPE; |
510 | u.prototype = adopted; |
511 | } |
512 | |
513 | void TransliteratorEntry::setFactory(Transliterator::Factory factory, |
514 | Transliterator::Token context) { |
515 | if (entryType == PROTOTYPE) { |
516 | delete u.prototype; |
517 | } |
518 | entryType = FACTORY; |
519 | u.factory.function = factory; |
520 | u.factory.context = context; |
521 | } |
522 | |
523 | // UObjectDeleter for Hashtable::setValueDeleter |
524 | U_CDECL_BEGIN |
525 | static void U_CALLCONV |
526 | deleteEntry(void* obj) { |
527 | delete (TransliteratorEntry*) obj; |
528 | } |
529 | U_CDECL_END |
530 | |
531 | //---------------------------------------------------------------------- |
532 | // class TransliteratorRegistry: Basic public API |
533 | //---------------------------------------------------------------------- |
534 | |
535 | TransliteratorRegistry::TransliteratorRegistry(UErrorCode& status) : |
536 | registry(TRUE, status), |
537 | specDAG(TRUE, SPECDAG_INIT_SIZE, status), |
538 | variantList(VARIANT_LIST_INIT_SIZE, status), |
539 | availableIDs(AVAILABLE_IDS_INIT_SIZE, status) |
540 | { |
541 | registry.setValueDeleter(deleteEntry); |
542 | variantList.setDeleter(uprv_deleteUObject); |
543 | variantList.setComparer(uhash_compareCaselessUnicodeString); |
544 | UnicodeString *emptyString = new UnicodeString(); |
545 | if (emptyString != NULL) { |
546 | variantList.addElement(emptyString, status); |
547 | } |
548 | availableIDs.setDeleter(uprv_deleteUObject); |
549 | availableIDs.setComparer(uhash_compareCaselessUnicodeString); |
550 | specDAG.setValueDeleter(uhash_deleteHashtable); |
551 | } |
552 | |
553 | TransliteratorRegistry::~TransliteratorRegistry() { |
554 | // Through the magic of C++, everything cleans itself up |
555 | } |
556 | |
557 | Transliterator* TransliteratorRegistry::get(const UnicodeString& ID, |
558 | TransliteratorAlias*& aliasReturn, |
559 | UErrorCode& status) { |
560 | U_ASSERT(aliasReturn == NULL); |
561 | TransliteratorEntry *entry = find(ID); |
562 | return (entry == 0) ? 0 |
563 | : instantiateEntry(ID, entry, aliasReturn, status); |
564 | } |
565 | |
566 | Transliterator* TransliteratorRegistry::reget(const UnicodeString& ID, |
567 | TransliteratorParser& parser, |
568 | TransliteratorAlias*& aliasReturn, |
569 | UErrorCode& status) { |
570 | U_ASSERT(aliasReturn == NULL); |
571 | TransliteratorEntry *entry = find(ID); |
572 | |
573 | if (entry == 0) { |
574 | // We get to this point if there are two threads, one of which |
575 | // is instantiating an ID, and another of which is removing |
576 | // the same ID from the registry, and the timing is just right. |
577 | return 0; |
578 | } |
579 | |
580 | // The usage model for the caller is that they will first call |
581 | // reg->get() inside the mutex, they'll get back an alias, they call |
582 | // alias->isRuleBased(), and if they get TRUE, they call alias->parse() |
583 | // outside the mutex, then reg->reget() inside the mutex again. A real |
584 | // mess, but it gets things working for ICU 3.0. [alan]. |
585 | |
586 | // Note: It's possible that in between the caller calling |
587 | // alias->parse() and reg->reget(), that another thread will have |
588 | // called reg->reget(), and the entry will already have been fixed up. |
589 | // We have to detect this so we don't stomp over existing entry |
590 | // data members and potentially leak memory (u.data and compoundFilter). |
591 | |
592 | if (entry->entryType == TransliteratorEntry::RULES_FORWARD || |
593 | entry->entryType == TransliteratorEntry::RULES_REVERSE || |
594 | entry->entryType == TransliteratorEntry::LOCALE_RULES) { |
595 | |
596 | if (parser.idBlockVector.isEmpty() && parser.dataVector.isEmpty()) { |
597 | entry->u.data = 0; |
598 | entry->entryType = TransliteratorEntry::ALIAS; |
599 | entry->stringArg = UNICODE_STRING_SIMPLE("Any-NULL" ); |
600 | } |
601 | else if (parser.idBlockVector.isEmpty() && parser.dataVector.size() == 1) { |
602 | entry->u.data = (TransliterationRuleData*)parser.dataVector.orphanElementAt(0); |
603 | entry->entryType = TransliteratorEntry::RBT_DATA; |
604 | } |
605 | else if (parser.idBlockVector.size() == 1 && parser.dataVector.isEmpty()) { |
606 | entry->stringArg = *(UnicodeString*)(parser.idBlockVector.elementAt(0)); |
607 | entry->compoundFilter = parser.orphanCompoundFilter(); |
608 | entry->entryType = TransliteratorEntry::ALIAS; |
609 | } |
610 | else { |
611 | entry->entryType = TransliteratorEntry::COMPOUND_RBT; |
612 | entry->compoundFilter = parser.orphanCompoundFilter(); |
613 | entry->u.dataVector = new UVector(status); |
614 | entry->stringArg.remove(); |
615 | |
616 | int32_t limit = parser.idBlockVector.size(); |
617 | if (parser.dataVector.size() > limit) |
618 | limit = parser.dataVector.size(); |
619 | |
620 | for (int32_t i = 0; i < limit; i++) { |
621 | if (i < parser.idBlockVector.size()) { |
622 | UnicodeString* idBlock = (UnicodeString*)parser.idBlockVector.elementAt(i); |
623 | if (!idBlock->isEmpty()) |
624 | entry->stringArg += *idBlock; |
625 | } |
626 | if (!parser.dataVector.isEmpty()) { |
627 | TransliterationRuleData* data = (TransliterationRuleData*)parser.dataVector.orphanElementAt(0); |
628 | entry->u.dataVector->addElement(data, status); |
629 | entry->stringArg += (UChar)0xffff; // use U+FFFF to mark position of RBTs in ID block |
630 | } |
631 | } |
632 | } |
633 | } |
634 | |
635 | Transliterator *t = |
636 | instantiateEntry(ID, entry, aliasReturn, status); |
637 | return t; |
638 | } |
639 | |
640 | void TransliteratorRegistry::put(Transliterator* adoptedProto, |
641 | UBool visible, |
642 | UErrorCode& ec) |
643 | { |
644 | TransliteratorEntry *entry = new TransliteratorEntry(); |
645 | if (entry == NULL) { |
646 | ec = U_MEMORY_ALLOCATION_ERROR; |
647 | return; |
648 | } |
649 | entry->adoptPrototype(adoptedProto); |
650 | registerEntry(adoptedProto->getID(), entry, visible); |
651 | } |
652 | |
653 | void TransliteratorRegistry::put(const UnicodeString& ID, |
654 | Transliterator::Factory factory, |
655 | Transliterator::Token context, |
656 | UBool visible, |
657 | UErrorCode& ec) { |
658 | TransliteratorEntry *entry = new TransliteratorEntry(); |
659 | if (entry == NULL) { |
660 | ec = U_MEMORY_ALLOCATION_ERROR; |
661 | return; |
662 | } |
663 | entry->setFactory(factory, context); |
664 | registerEntry(ID, entry, visible); |
665 | } |
666 | |
667 | void TransliteratorRegistry::put(const UnicodeString& ID, |
668 | const UnicodeString& resourceName, |
669 | UTransDirection dir, |
670 | UBool readonlyResourceAlias, |
671 | UBool visible, |
672 | UErrorCode& ec) { |
673 | TransliteratorEntry *entry = new TransliteratorEntry(); |
674 | if (entry == NULL) { |
675 | ec = U_MEMORY_ALLOCATION_ERROR; |
676 | return; |
677 | } |
678 | entry->entryType = (dir == UTRANS_FORWARD) ? TransliteratorEntry::RULES_FORWARD |
679 | : TransliteratorEntry::RULES_REVERSE; |
680 | if (readonlyResourceAlias) { |
681 | entry->stringArg.setTo(TRUE, resourceName.getBuffer(), -1); |
682 | } |
683 | else { |
684 | entry->stringArg = resourceName; |
685 | } |
686 | registerEntry(ID, entry, visible); |
687 | } |
688 | |
689 | void TransliteratorRegistry::put(const UnicodeString& ID, |
690 | const UnicodeString& alias, |
691 | UBool readonlyAliasAlias, |
692 | UBool visible, |
693 | UErrorCode& /*ec*/) { |
694 | TransliteratorEntry *entry = new TransliteratorEntry(); |
695 | // Null pointer check |
696 | if (entry != NULL) { |
697 | entry->entryType = TransliteratorEntry::ALIAS; |
698 | if (readonlyAliasAlias) { |
699 | entry->stringArg.setTo(TRUE, alias.getBuffer(), -1); |
700 | } |
701 | else { |
702 | entry->stringArg = alias; |
703 | } |
704 | registerEntry(ID, entry, visible); |
705 | } |
706 | } |
707 | |
708 | void TransliteratorRegistry::remove(const UnicodeString& ID) { |
709 | UnicodeString source, target, variant; |
710 | UBool sawSource; |
711 | TransliteratorIDParser::IDtoSTV(ID, source, target, variant, sawSource); |
712 | // Only need to do this if ID.indexOf('-') < 0 |
713 | UnicodeString id; |
714 | TransliteratorIDParser::STVtoID(source, target, variant, id); |
715 | registry.remove(id); |
716 | removeSTV(source, target, variant); |
717 | availableIDs.removeElement((void*) &id); |
718 | } |
719 | |
720 | //---------------------------------------------------------------------- |
721 | // class TransliteratorRegistry: Public ID and spec management |
722 | //---------------------------------------------------------------------- |
723 | |
724 | /** |
725 | * == OBSOLETE - remove in ICU 3.4 == |
726 | * Return the number of IDs currently registered with the system. |
727 | * To retrieve the actual IDs, call getAvailableID(i) with |
728 | * i from 0 to countAvailableIDs() - 1. |
729 | */ |
730 | int32_t TransliteratorRegistry::countAvailableIDs(void) const { |
731 | return availableIDs.size(); |
732 | } |
733 | |
734 | /** |
735 | * == OBSOLETE - remove in ICU 3.4 == |
736 | * Return the index-th available ID. index must be between 0 |
737 | * and countAvailableIDs() - 1, inclusive. If index is out of |
738 | * range, the result of getAvailableID(0) is returned. |
739 | */ |
740 | const UnicodeString& TransliteratorRegistry::getAvailableID(int32_t index) const { |
741 | if (index < 0 || index >= availableIDs.size()) { |
742 | index = 0; |
743 | } |
744 | return *(const UnicodeString*) availableIDs[index]; |
745 | } |
746 | |
747 | StringEnumeration* TransliteratorRegistry::getAvailableIDs() const { |
748 | return new Enumeration(*this); |
749 | } |
750 | |
751 | int32_t TransliteratorRegistry::countAvailableSources(void) const { |
752 | return specDAG.count(); |
753 | } |
754 | |
755 | UnicodeString& TransliteratorRegistry::getAvailableSource(int32_t index, |
756 | UnicodeString& result) const { |
757 | int32_t pos = UHASH_FIRST; |
758 | const UHashElement *e = 0; |
759 | while (index-- >= 0) { |
760 | e = specDAG.nextElement(pos); |
761 | if (e == 0) { |
762 | break; |
763 | } |
764 | } |
765 | if (e == 0) { |
766 | result.truncate(0); |
767 | } else { |
768 | result = *(UnicodeString*) e->key.pointer; |
769 | } |
770 | return result; |
771 | } |
772 | |
773 | int32_t TransliteratorRegistry::countAvailableTargets(const UnicodeString& source) const { |
774 | Hashtable *targets = (Hashtable*) specDAG.get(source); |
775 | return (targets == 0) ? 0 : targets->count(); |
776 | } |
777 | |
778 | UnicodeString& TransliteratorRegistry::getAvailableTarget(int32_t index, |
779 | const UnicodeString& source, |
780 | UnicodeString& result) const { |
781 | Hashtable *targets = (Hashtable*) specDAG.get(source); |
782 | if (targets == 0) { |
783 | result.truncate(0); // invalid source |
784 | return result; |
785 | } |
786 | int32_t pos = UHASH_FIRST; |
787 | const UHashElement *e = 0; |
788 | while (index-- >= 0) { |
789 | e = targets->nextElement(pos); |
790 | if (e == 0) { |
791 | break; |
792 | } |
793 | } |
794 | if (e == 0) { |
795 | result.truncate(0); // invalid index |
796 | } else { |
797 | result = *(UnicodeString*) e->key.pointer; |
798 | } |
799 | return result; |
800 | } |
801 | |
802 | int32_t TransliteratorRegistry::countAvailableVariants(const UnicodeString& source, |
803 | const UnicodeString& target) const { |
804 | Hashtable *targets = (Hashtable*) specDAG.get(source); |
805 | if (targets == 0) { |
806 | return 0; |
807 | } |
808 | uint32_t varMask = targets->geti(target); |
809 | int32_t varCount = 0; |
810 | while (varMask > 0) { |
811 | if (varMask & 1) { |
812 | varCount++; |
813 | } |
814 | varMask >>= 1; |
815 | } |
816 | return varCount; |
817 | } |
818 | |
819 | UnicodeString& TransliteratorRegistry::getAvailableVariant(int32_t index, |
820 | const UnicodeString& source, |
821 | const UnicodeString& target, |
822 | UnicodeString& result) const { |
823 | Hashtable *targets = (Hashtable*) specDAG.get(source); |
824 | if (targets == 0) { |
825 | result.truncate(0); // invalid source |
826 | return result; |
827 | } |
828 | uint32_t varMask = targets->geti(target); |
829 | int32_t varCount = 0; |
830 | int32_t varListIndex = 0; |
831 | while (varMask > 0) { |
832 | if (varMask & 1) { |
833 | if (varCount == index) { |
834 | UnicodeString *v = (UnicodeString*) variantList.elementAt(varListIndex); |
835 | if (v != NULL) { |
836 | result = *v; |
837 | return result; |
838 | } |
839 | break; |
840 | } |
841 | varCount++; |
842 | } |
843 | varMask >>= 1; |
844 | varListIndex++; |
845 | } |
846 | result.truncate(0); // invalid target or index |
847 | return result; |
848 | } |
849 | |
850 | //---------------------------------------------------------------------- |
851 | // class TransliteratorRegistry::Enumeration |
852 | //---------------------------------------------------------------------- |
853 | |
854 | TransliteratorRegistry::Enumeration::Enumeration(const TransliteratorRegistry& _reg) : |
855 | index(0), reg(_reg) { |
856 | } |
857 | |
858 | TransliteratorRegistry::Enumeration::~Enumeration() { |
859 | } |
860 | |
861 | int32_t TransliteratorRegistry::Enumeration::count(UErrorCode& /*status*/) const { |
862 | return reg.availableIDs.size(); |
863 | } |
864 | |
865 | const UnicodeString* TransliteratorRegistry::Enumeration::snext(UErrorCode& status) { |
866 | // This is sloppy but safe -- if we get out of sync with the underlying |
867 | // registry, we will still return legal strings, but they might not |
868 | // correspond to the snapshot at construction time. So there could be |
869 | // duplicate IDs or omitted IDs if insertions or deletions occur in one |
870 | // thread while another is iterating. To be more rigorous, add a timestamp, |
871 | // which is incremented with any modification, and validate this iterator |
872 | // against the timestamp at construction time. This probably isn't worth |
873 | // doing as long as there is some possibility of removing this code in favor |
874 | // of some new code based on Doug's service framework. |
875 | if (U_FAILURE(status)) { |
876 | return NULL; |
877 | } |
878 | int32_t n = reg.availableIDs.size(); |
879 | if (index > n) { |
880 | status = U_ENUM_OUT_OF_SYNC_ERROR; |
881 | } |
882 | // index == n is okay -- this means we've reached the end |
883 | if (index < n) { |
884 | // Copy the string! This avoids lifetime problems. |
885 | unistr = *(const UnicodeString*)reg.availableIDs[index++]; |
886 | return &unistr; |
887 | } else { |
888 | return NULL; |
889 | } |
890 | } |
891 | |
892 | void TransliteratorRegistry::Enumeration::reset(UErrorCode& /*status*/) { |
893 | index = 0; |
894 | } |
895 | |
896 | UOBJECT_DEFINE_RTTI_IMPLEMENTATION(TransliteratorRegistry::Enumeration) |
897 | |
898 | //---------------------------------------------------------------------- |
899 | // class TransliteratorRegistry: internal |
900 | //---------------------------------------------------------------------- |
901 | |
902 | /** |
903 | * Convenience method. Calls 6-arg registerEntry(). |
904 | */ |
905 | void TransliteratorRegistry::registerEntry(const UnicodeString& source, |
906 | const UnicodeString& target, |
907 | const UnicodeString& variant, |
908 | TransliteratorEntry* adopted, |
909 | UBool visible) { |
910 | UnicodeString ID; |
911 | UnicodeString s(source); |
912 | if (s.length() == 0) { |
913 | s.setTo(TRUE, ANY, 3); |
914 | } |
915 | TransliteratorIDParser::STVtoID(source, target, variant, ID); |
916 | registerEntry(ID, s, target, variant, adopted, visible); |
917 | } |
918 | |
919 | /** |
920 | * Convenience method. Calls 6-arg registerEntry(). |
921 | */ |
922 | void TransliteratorRegistry::registerEntry(const UnicodeString& ID, |
923 | TransliteratorEntry* adopted, |
924 | UBool visible) { |
925 | UnicodeString source, target, variant; |
926 | UBool sawSource; |
927 | TransliteratorIDParser::IDtoSTV(ID, source, target, variant, sawSource); |
928 | // Only need to do this if ID.indexOf('-') < 0 |
929 | UnicodeString id; |
930 | TransliteratorIDParser::STVtoID(source, target, variant, id); |
931 | registerEntry(id, source, target, variant, adopted, visible); |
932 | } |
933 | |
934 | /** |
935 | * Register an entry object (adopted) with the given ID, source, |
936 | * target, and variant strings. |
937 | */ |
938 | void TransliteratorRegistry::registerEntry(const UnicodeString& ID, |
939 | const UnicodeString& source, |
940 | const UnicodeString& target, |
941 | const UnicodeString& variant, |
942 | TransliteratorEntry* adopted, |
943 | UBool visible) { |
944 | UErrorCode status = U_ZERO_ERROR; |
945 | registry.put(ID, adopted, status); |
946 | if (visible) { |
947 | registerSTV(source, target, variant); |
948 | if (!availableIDs.contains((void*) &ID)) { |
949 | UnicodeString *newID = ID.clone(); |
950 | // Check to make sure newID was created. |
951 | if (newID != NULL) { |
952 | // NUL-terminate the ID string |
953 | newID->getTerminatedBuffer(); |
954 | availableIDs.addElement(newID, status); |
955 | } |
956 | } |
957 | } else { |
958 | removeSTV(source, target, variant); |
959 | availableIDs.removeElement((void*) &ID); |
960 | } |
961 | } |
962 | |
963 | /** |
964 | * Register a source-target/variant in the specDAG. Variant may be |
965 | * empty, but source and target must not be. |
966 | */ |
967 | void TransliteratorRegistry::registerSTV(const UnicodeString& source, |
968 | const UnicodeString& target, |
969 | const UnicodeString& variant) { |
970 | // assert(source.length() > 0); |
971 | // assert(target.length() > 0); |
972 | UErrorCode status = U_ZERO_ERROR; |
973 | Hashtable *targets = (Hashtable*) specDAG.get(source); |
974 | if (targets == 0) { |
975 | int32_t size = 3; |
976 | if (source.compare(ANY,3) == 0) { |
977 | size = ANY_TARGETS_INIT_SIZE; |
978 | } else if (source.compare(LAT,3) == 0) { |
979 | size = LAT_TARGETS_INIT_SIZE; |
980 | } |
981 | targets = new Hashtable(TRUE, size, status); |
982 | if (U_FAILURE(status) || targets == NULL) { |
983 | return; |
984 | } |
985 | specDAG.put(source, targets, status); |
986 | } |
987 | int32_t variantListIndex = variantList.indexOf((void*) &variant, 0); |
988 | if (variantListIndex < 0) { |
989 | if (variantList.size() >= VARIANT_LIST_MAX_SIZE) { |
990 | // can't handle any more variants |
991 | return; |
992 | } |
993 | UnicodeString *variantEntry = new UnicodeString(variant); |
994 | if (variantEntry != NULL) { |
995 | variantList.addElement(variantEntry, status); |
996 | if (U_SUCCESS(status)) { |
997 | variantListIndex = variantList.size() - 1; |
998 | } |
999 | } |
1000 | if (variantListIndex < 0) { |
1001 | return; |
1002 | } |
1003 | } |
1004 | uint32_t addMask = 1 << variantListIndex; |
1005 | uint32_t varMask = targets->geti(target); |
1006 | targets->puti(target, varMask | addMask, status); |
1007 | } |
1008 | |
1009 | /** |
1010 | * Remove a source-target/variant from the specDAG. |
1011 | */ |
1012 | void TransliteratorRegistry::removeSTV(const UnicodeString& source, |
1013 | const UnicodeString& target, |
1014 | const UnicodeString& variant) { |
1015 | // assert(source.length() > 0); |
1016 | // assert(target.length() > 0); |
1017 | UErrorCode status = U_ZERO_ERROR; |
1018 | Hashtable *targets = (Hashtable*) specDAG.get(source); |
1019 | if (targets == NULL) { |
1020 | return; // should never happen for valid s-t/v |
1021 | } |
1022 | uint32_t varMask = targets->geti(target); |
1023 | if (varMask == 0) { |
1024 | return; // should never happen for valid s-t/v |
1025 | } |
1026 | int32_t variantListIndex = variantList.indexOf((void*) &variant, 0); |
1027 | if (variantListIndex < 0) { |
1028 | return; // should never happen for valid s-t/v |
1029 | } |
1030 | int32_t remMask = 1 << variantListIndex; |
1031 | varMask &= (~remMask); |
1032 | if (varMask != 0) { |
1033 | targets->puti(target, varMask, status); |
1034 | } else { |
1035 | targets->remove(target); // should delete variants |
1036 | if (targets->count() == 0) { |
1037 | specDAG.remove(source); // should delete targets |
1038 | } |
1039 | } |
1040 | } |
1041 | |
1042 | /** |
1043 | * Attempt to find a source-target/variant in the dynamic registry |
1044 | * store. Return 0 on failure. |
1045 | * |
1046 | * Caller does NOT own returned object. |
1047 | */ |
1048 | TransliteratorEntry* TransliteratorRegistry::findInDynamicStore(const TransliteratorSpec& src, |
1049 | const TransliteratorSpec& trg, |
1050 | const UnicodeString& variant) const { |
1051 | UnicodeString ID; |
1052 | TransliteratorIDParser::STVtoID(src, trg, variant, ID); |
1053 | TransliteratorEntry *e = (TransliteratorEntry*) registry.get(ID); |
1054 | DEBUG_useEntry(e); |
1055 | return e; |
1056 | } |
1057 | |
1058 | /** |
1059 | * Attempt to find a source-target/variant in the static locale |
1060 | * resource store. Do not perform fallback. Return 0 on failure. |
1061 | * |
1062 | * On success, create a new entry object, register it in the dynamic |
1063 | * store, and return a pointer to it, but do not make it public -- |
1064 | * just because someone requested something, we do not expand the |
1065 | * available ID list (or spec DAG). |
1066 | * |
1067 | * Caller does NOT own returned object. |
1068 | */ |
1069 | TransliteratorEntry* TransliteratorRegistry::findInStaticStore(const TransliteratorSpec& src, |
1070 | const TransliteratorSpec& trg, |
1071 | const UnicodeString& variant) { |
1072 | TransliteratorEntry* entry = 0; |
1073 | if (src.isLocale()) { |
1074 | entry = findInBundle(src, trg, variant, UTRANS_FORWARD); |
1075 | } else if (trg.isLocale()) { |
1076 | entry = findInBundle(trg, src, variant, UTRANS_REVERSE); |
1077 | } |
1078 | |
1079 | // If we found an entry, store it in the Hashtable for next |
1080 | // time. |
1081 | if (entry != 0) { |
1082 | registerEntry(src.getTop(), trg.getTop(), variant, entry, FALSE); |
1083 | } |
1084 | |
1085 | return entry; |
1086 | } |
1087 | |
1088 | // As of 2.0, resource bundle keys cannot contain '_' |
1089 | static const UChar TRANSLITERATE_TO[] = {84,114,97,110,115,108,105,116,101,114,97,116,101,84,111,0}; // "TransliterateTo" |
1090 | |
1091 | static const UChar TRANSLITERATE_FROM[] = {84,114,97,110,115,108,105,116,101,114,97,116,101,70,114,111,109,0}; // "TransliterateFrom" |
1092 | |
1093 | static const UChar TRANSLITERATE[] = {84,114,97,110,115,108,105,116,101,114,97,116,101,0}; // "Transliterate" |
1094 | |
1095 | /** |
1096 | * Attempt to find an entry in a single resource bundle. This is |
1097 | * a one-sided lookup. findInStaticStore() performs up to two such |
1098 | * lookups, one for the source, and one for the target. |
1099 | * |
1100 | * Do not perform fallback. Return 0 on failure. |
1101 | * |
1102 | * On success, create a new Entry object, populate it, and return it. |
1103 | * The caller owns the returned object. |
1104 | */ |
1105 | TransliteratorEntry* TransliteratorRegistry::findInBundle(const TransliteratorSpec& specToOpen, |
1106 | const TransliteratorSpec& specToFind, |
1107 | const UnicodeString& variant, |
1108 | UTransDirection direction) |
1109 | { |
1110 | UnicodeString utag; |
1111 | UnicodeString resStr; |
1112 | int32_t pass; |
1113 | |
1114 | for (pass=0; pass<2; ++pass) { |
1115 | utag.truncate(0); |
1116 | // First try either TransliteratorTo_xxx or |
1117 | // TransliterateFrom_xxx, then try the bidirectional |
1118 | // Transliterate_xxx. This precedence order is arbitrary |
1119 | // but must be consistent and documented. |
1120 | if (pass == 0) { |
1121 | utag.append(direction == UTRANS_FORWARD ? |
1122 | TRANSLITERATE_TO : TRANSLITERATE_FROM, -1); |
1123 | } else { |
1124 | utag.append(TRANSLITERATE, -1); |
1125 | } |
1126 | UnicodeString s(specToFind.get()); |
1127 | utag.append(s.toUpper("" )); |
1128 | UErrorCode status = U_ZERO_ERROR; |
1129 | ResourceBundle subres(specToOpen.getBundle().get( |
1130 | CharString().appendInvariantChars(utag, status).data(), status)); |
1131 | if (U_FAILURE(status) || status == U_USING_DEFAULT_WARNING) { |
1132 | continue; |
1133 | } |
1134 | |
1135 | s.truncate(0); |
1136 | if (specToOpen.get() != LocaleUtility::initNameFromLocale(subres.getLocale(), s)) { |
1137 | continue; |
1138 | } |
1139 | |
1140 | if (variant.length() != 0) { |
1141 | status = U_ZERO_ERROR; |
1142 | resStr = subres.getStringEx( |
1143 | CharString().appendInvariantChars(variant, status).data(), status); |
1144 | if (U_SUCCESS(status)) { |
1145 | // Exit loop successfully |
1146 | break; |
1147 | } |
1148 | } else { |
1149 | // Variant is empty, which means match the first variant listed. |
1150 | status = U_ZERO_ERROR; |
1151 | resStr = subres.getStringEx(1, status); |
1152 | if (U_SUCCESS(status)) { |
1153 | // Exit loop successfully |
1154 | break; |
1155 | } |
1156 | } |
1157 | } |
1158 | |
1159 | if (pass==2) { |
1160 | // Failed |
1161 | return NULL; |
1162 | } |
1163 | |
1164 | // We have succeeded in loading a string from the locale |
1165 | // resources. Create a new registry entry to hold it and return it. |
1166 | TransliteratorEntry *entry = new TransliteratorEntry(); |
1167 | if (entry != 0) { |
1168 | // The direction is always forward for the |
1169 | // TransliterateTo_xxx and TransliterateFrom_xxx |
1170 | // items; those are unidirectional forward rules. |
1171 | // For the bidirectional Transliterate_xxx items, |
1172 | // the direction is the value passed in to this |
1173 | // function. |
1174 | int32_t dir = (pass == 0) ? UTRANS_FORWARD : direction; |
1175 | entry->entryType = TransliteratorEntry::LOCALE_RULES; |
1176 | entry->stringArg = resStr; |
1177 | entry->intArg = dir; |
1178 | } |
1179 | |
1180 | return entry; |
1181 | } |
1182 | |
1183 | /** |
1184 | * Convenience method. Calls 3-arg find(). |
1185 | */ |
1186 | TransliteratorEntry* TransliteratorRegistry::find(const UnicodeString& ID) { |
1187 | UnicodeString source, target, variant; |
1188 | UBool sawSource; |
1189 | TransliteratorIDParser::IDtoSTV(ID, source, target, variant, sawSource); |
1190 | return find(source, target, variant); |
1191 | } |
1192 | |
1193 | /** |
1194 | * Top-level find method. Attempt to find a source-target/variant in |
1195 | * either the dynamic or the static (locale resource) store. Perform |
1196 | * fallback. |
1197 | * |
1198 | * Lookup sequence for ss_SS_SSS-tt_TT_TTT/v: |
1199 | * |
1200 | * ss_SS_SSS-tt_TT_TTT/v -- in hashtable |
1201 | * ss_SS_SSS-tt_TT_TTT/v -- in ss_SS_SSS (no fallback) |
1202 | * |
1203 | * repeat with t = tt_TT_TTT, tt_TT, tt, and tscript |
1204 | * |
1205 | * ss_SS_SSS-t/ * |
1206 | * ss_SS-t/ * |
1207 | * ss-t/ * |
1208 | * sscript-t/ * |
1209 | * |
1210 | * Here * matches the first variant listed. |
1211 | * |
1212 | * Caller does NOT own returned object. Return 0 on failure. |
1213 | */ |
1214 | TransliteratorEntry* TransliteratorRegistry::find(UnicodeString& source, |
1215 | UnicodeString& target, |
1216 | UnicodeString& variant) { |
1217 | |
1218 | TransliteratorSpec src(source); |
1219 | TransliteratorSpec trg(target); |
1220 | TransliteratorEntry* entry; |
1221 | |
1222 | // Seek exact match in hashtable. Temporary fix for ICU 4.6. |
1223 | // TODO: The general logic for finding a matching transliterator needs to be reviewed. |
1224 | // ICU ticket #8089 |
1225 | UnicodeString ID; |
1226 | TransliteratorIDParser::STVtoID(source, target, variant, ID); |
1227 | entry = (TransliteratorEntry*) registry.get(ID); |
1228 | if (entry != 0) { |
1229 | // std::string ss; |
1230 | // std::cout << ID.toUTF8String(ss) << std::endl; |
1231 | return entry; |
1232 | } |
1233 | |
1234 | if (variant.length() != 0) { |
1235 | |
1236 | // Seek exact match in hashtable |
1237 | entry = findInDynamicStore(src, trg, variant); |
1238 | if (entry != 0) { |
1239 | return entry; |
1240 | } |
1241 | |
1242 | // Seek exact match in locale resources |
1243 | entry = findInStaticStore(src, trg, variant); |
1244 | if (entry != 0) { |
1245 | return entry; |
1246 | } |
1247 | } |
1248 | |
1249 | for (;;) { |
1250 | src.reset(); |
1251 | for (;;) { |
1252 | // Seek match in hashtable |
1253 | entry = findInDynamicStore(src, trg, NO_VARIANT); |
1254 | if (entry != 0) { |
1255 | return entry; |
1256 | } |
1257 | |
1258 | // Seek match in locale resources |
1259 | entry = findInStaticStore(src, trg, NO_VARIANT); |
1260 | if (entry != 0) { |
1261 | return entry; |
1262 | } |
1263 | if (!src.hasFallback()) { |
1264 | break; |
1265 | } |
1266 | src.next(); |
1267 | } |
1268 | if (!trg.hasFallback()) { |
1269 | break; |
1270 | } |
1271 | trg.next(); |
1272 | } |
1273 | |
1274 | return 0; |
1275 | } |
1276 | |
1277 | /** |
1278 | * Given an Entry object, instantiate it. Caller owns result. Return |
1279 | * 0 on failure. |
1280 | * |
1281 | * Return a non-empty aliasReturn value if the ID points to an alias. |
1282 | * We cannot instantiate it ourselves because the alias may contain |
1283 | * filters or compounds, which we do not understand. Caller should |
1284 | * make aliasReturn empty before calling. |
1285 | * |
1286 | * The entry object is assumed to reside in the dynamic store. It may be |
1287 | * modified. |
1288 | */ |
1289 | Transliterator* TransliteratorRegistry::instantiateEntry(const UnicodeString& ID, |
1290 | TransliteratorEntry *entry, |
1291 | TransliteratorAlias* &aliasReturn, |
1292 | UErrorCode& status) { |
1293 | Transliterator *t = 0; |
1294 | U_ASSERT(aliasReturn == 0); |
1295 | |
1296 | switch (entry->entryType) { |
1297 | case TransliteratorEntry::RBT_DATA: |
1298 | t = new RuleBasedTransliterator(ID, entry->u.data); |
1299 | if (t == 0) { |
1300 | status = U_MEMORY_ALLOCATION_ERROR; |
1301 | } |
1302 | return t; |
1303 | case TransliteratorEntry::PROTOTYPE: |
1304 | t = entry->u.prototype->clone(); |
1305 | if (t == 0) { |
1306 | status = U_MEMORY_ALLOCATION_ERROR; |
1307 | } |
1308 | return t; |
1309 | case TransliteratorEntry::ALIAS: |
1310 | aliasReturn = new TransliteratorAlias(entry->stringArg, entry->compoundFilter); |
1311 | if (aliasReturn == 0) { |
1312 | status = U_MEMORY_ALLOCATION_ERROR; |
1313 | } |
1314 | return 0; |
1315 | case TransliteratorEntry::FACTORY: |
1316 | t = entry->u.factory.function(ID, entry->u.factory.context); |
1317 | if (t == 0) { |
1318 | status = U_MEMORY_ALLOCATION_ERROR; |
1319 | } |
1320 | return t; |
1321 | case TransliteratorEntry::COMPOUND_RBT: |
1322 | { |
1323 | UVector* rbts = new UVector(entry->u.dataVector->size(), status); |
1324 | // Check for null pointer |
1325 | if (rbts == NULL) { |
1326 | status = U_MEMORY_ALLOCATION_ERROR; |
1327 | return NULL; |
1328 | } |
1329 | int32_t passNumber = 1; |
1330 | for (int32_t i = 0; U_SUCCESS(status) && i < entry->u.dataVector->size(); i++) { |
1331 | // TODO: Should passNumber be turned into a decimal-string representation (1 -> "1")? |
1332 | Transliterator* tl = new RuleBasedTransliterator(UnicodeString(CompoundTransliterator::PASS_STRING) + UnicodeString(passNumber++), |
1333 | (TransliterationRuleData*)(entry->u.dataVector->elementAt(i)), FALSE); |
1334 | if (tl == 0) |
1335 | status = U_MEMORY_ALLOCATION_ERROR; |
1336 | else |
1337 | rbts->addElement(tl, status); |
1338 | } |
1339 | if (U_FAILURE(status)) { |
1340 | delete rbts; |
1341 | return 0; |
1342 | } |
1343 | aliasReturn = new TransliteratorAlias(ID, entry->stringArg, rbts, entry->compoundFilter); |
1344 | } |
1345 | if (aliasReturn == 0) { |
1346 | status = U_MEMORY_ALLOCATION_ERROR; |
1347 | } |
1348 | return 0; |
1349 | case TransliteratorEntry::LOCALE_RULES: |
1350 | aliasReturn = new TransliteratorAlias(ID, entry->stringArg, |
1351 | (UTransDirection) entry->intArg); |
1352 | if (aliasReturn == 0) { |
1353 | status = U_MEMORY_ALLOCATION_ERROR; |
1354 | } |
1355 | return 0; |
1356 | case TransliteratorEntry::RULES_FORWARD: |
1357 | case TransliteratorEntry::RULES_REVERSE: |
1358 | // Process the rule data into a TransliteratorRuleData object, |
1359 | // and possibly also into an ::id header and/or footer. Then |
1360 | // we modify the registry with the parsed data and retry. |
1361 | { |
1362 | TransliteratorParser parser(status); |
1363 | |
1364 | // We use the file name, taken from another resource bundle |
1365 | // 2-d array at static init time, as a locale language. We're |
1366 | // just using the locale mechanism to map through to a file |
1367 | // name; this in no way represents an actual locale. |
1368 | //CharString ch(entry->stringArg); |
1369 | //UResourceBundle *bundle = ures_openDirect(0, ch, &status); |
1370 | UnicodeString rules = entry->stringArg; |
1371 | //ures_close(bundle); |
1372 | |
1373 | //if (U_FAILURE(status)) { |
1374 | // We have a failure of some kind. Remove the ID from the |
1375 | // registry so we don't keep trying. NOTE: This will throw off |
1376 | // anyone who is, at the moment, trying to iterate over the |
1377 | // available IDs. That's acceptable since we should never |
1378 | // really get here except under installation, configuration, |
1379 | // or unrecoverable run time memory failures. |
1380 | // remove(ID); |
1381 | //} else { |
1382 | |
1383 | // If the status indicates a failure, then we don't have any |
1384 | // rules -- there is probably an installation error. The list |
1385 | // in the root locale should correspond to all the installed |
1386 | // transliterators; if it lists something that's not |
1387 | // installed, we'll get an error from ResourceBundle. |
1388 | aliasReturn = new TransliteratorAlias(ID, rules, |
1389 | ((entry->entryType == TransliteratorEntry::RULES_REVERSE) ? |
1390 | UTRANS_REVERSE : UTRANS_FORWARD)); |
1391 | if (aliasReturn == 0) { |
1392 | status = U_MEMORY_ALLOCATION_ERROR; |
1393 | } |
1394 | //} |
1395 | } |
1396 | return 0; |
1397 | default: |
1398 | UPRV_UNREACHABLE; // can't get here |
1399 | } |
1400 | } |
1401 | U_NAMESPACE_END |
1402 | |
1403 | #endif /* #if !UCONFIG_NO_TRANSLITERATION */ |
1404 | |
1405 | //eof |
1406 | |