|
1 /* |
|
2 ********************************************************************** |
|
3 * Copyright (c) 2001-2011, International Business Machines |
|
4 * Corporation and others. All Rights Reserved. |
|
5 ********************************************************************** |
|
6 * Date Name Description |
|
7 * 08/10/2001 aliu Creation. |
|
8 ********************************************************************** |
|
9 */ |
|
10 |
|
11 #include "unicode/utypes.h" |
|
12 |
|
13 #if !UCONFIG_NO_TRANSLITERATION |
|
14 |
|
15 #include "unicode/translit.h" |
|
16 #include "unicode/resbund.h" |
|
17 #include "unicode/uniset.h" |
|
18 #include "unicode/uscript.h" |
|
19 #include "rbt.h" |
|
20 #include "cpdtrans.h" |
|
21 #include "nultrans.h" |
|
22 #include "transreg.h" |
|
23 #include "rbt_data.h" |
|
24 #include "rbt_pars.h" |
|
25 #include "tridpars.h" |
|
26 #include "charstr.h" |
|
27 #include "uassert.h" |
|
28 #include "locutil.h" |
|
29 |
|
30 // Enable the following symbol to add debugging code that tracks the |
|
31 // allocation, deletion, and use of Entry objects. BoundsChecker has |
|
32 // reported dangling pointer errors with these objects, but I have |
|
33 // been unable to confirm them. I suspect BoundsChecker is getting |
|
34 // confused with pointers going into and coming out of a UHashtable, |
|
35 // despite the hinting code that is designed to help it. |
|
36 // #define DEBUG_MEM |
|
37 #ifdef DEBUG_MEM |
|
38 #include <stdio.h> |
|
39 #endif |
|
40 |
|
41 // UChar constants |
|
42 static const UChar LOCALE_SEP = 95; // '_' |
|
43 //static const UChar ID_SEP = 0x002D; /*-*/ |
|
44 //static const UChar VARIANT_SEP = 0x002F; // '/' |
|
45 |
|
46 // String constants |
|
47 static const UChar ANY[] = { 65, 110, 121, 0 }; // Any |
|
48 |
|
49 // empty string |
|
50 #define NO_VARIANT UnicodeString() |
|
51 |
|
52 /** |
|
53 * Resource bundle key for the RuleBasedTransliterator rule. |
|
54 */ |
|
55 //static const char RB_RULE[] = "Rule"; |
|
56 |
|
57 U_NAMESPACE_BEGIN |
|
58 |
|
59 //------------------------------------------------------------------ |
|
60 // Alias |
|
61 //------------------------------------------------------------------ |
|
62 |
|
63 TransliteratorAlias::TransliteratorAlias(const UnicodeString& theAliasID, |
|
64 const UnicodeSet* cpdFilter) : |
|
65 ID(), |
|
66 aliasesOrRules(theAliasID), |
|
67 transes(0), |
|
68 compoundFilter(cpdFilter), |
|
69 direction(UTRANS_FORWARD), |
|
70 type(TransliteratorAlias::SIMPLE) { |
|
71 } |
|
72 |
|
73 TransliteratorAlias::TransliteratorAlias(const UnicodeString& theID, |
|
74 const UnicodeString& idBlocks, |
|
75 UVector* adoptedTransliterators, |
|
76 const UnicodeSet* cpdFilter) : |
|
77 ID(theID), |
|
78 aliasesOrRules(idBlocks), |
|
79 transes(adoptedTransliterators), |
|
80 compoundFilter(cpdFilter), |
|
81 direction(UTRANS_FORWARD), |
|
82 type(TransliteratorAlias::COMPOUND) { |
|
83 } |
|
84 |
|
85 TransliteratorAlias::TransliteratorAlias(const UnicodeString& theID, |
|
86 const UnicodeString& rules, |
|
87 UTransDirection dir) : |
|
88 ID(theID), |
|
89 aliasesOrRules(rules), |
|
90 transes(0), |
|
91 compoundFilter(0), |
|
92 direction(dir), |
|
93 type(TransliteratorAlias::RULES) { |
|
94 } |
|
95 |
|
96 TransliteratorAlias::~TransliteratorAlias() { |
|
97 delete transes; |
|
98 } |
|
99 |
|
100 |
|
101 Transliterator* TransliteratorAlias::create(UParseError& pe, |
|
102 UErrorCode& ec) { |
|
103 if (U_FAILURE(ec)) { |
|
104 return 0; |
|
105 } |
|
106 Transliterator *t = NULL; |
|
107 switch (type) { |
|
108 case SIMPLE: |
|
109 t = Transliterator::createInstance(aliasesOrRules, UTRANS_FORWARD, pe, ec); |
|
110 if(U_FAILURE(ec)){ |
|
111 return 0; |
|
112 } |
|
113 if (compoundFilter != 0) |
|
114 t->adoptFilter((UnicodeSet*)compoundFilter->clone()); |
|
115 break; |
|
116 case COMPOUND: |
|
117 { |
|
118 // the total number of transliterators in the compound is the total number of anonymous transliterators |
|
119 // plus the total number of ID blocks-- we start by assuming the list begins and ends with an ID |
|
120 // block and that each pair anonymous transliterators has an ID block between them. Then we go back |
|
121 // to see whether there really are ID blocks at the beginning and end (by looking for U+FFFF, which |
|
122 // marks the position where an anonymous transliterator goes) and adjust accordingly |
|
123 int32_t anonymousRBTs = transes->size(); |
|
124 int32_t transCount = anonymousRBTs * 2 + 1; |
|
125 if (!aliasesOrRules.isEmpty() && aliasesOrRules[0] == (UChar)(0xffff)) |
|
126 --transCount; |
|
127 if (aliasesOrRules.length() >= 2 && aliasesOrRules[aliasesOrRules.length() - 1] == (UChar)(0xffff)) |
|
128 --transCount; |
|
129 UnicodeString noIDBlock((UChar)(0xffff)); |
|
130 noIDBlock += ((UChar)(0xffff)); |
|
131 int32_t pos = aliasesOrRules.indexOf(noIDBlock); |
|
132 while (pos >= 0) { |
|
133 --transCount; |
|
134 pos = aliasesOrRules.indexOf(noIDBlock, pos + 1); |
|
135 } |
|
136 |
|
137 UVector transliterators(ec); |
|
138 UnicodeString idBlock; |
|
139 int32_t blockSeparatorPos = aliasesOrRules.indexOf((UChar)(0xffff)); |
|
140 while (blockSeparatorPos >= 0) { |
|
141 aliasesOrRules.extract(0, blockSeparatorPos, idBlock); |
|
142 aliasesOrRules.remove(0, blockSeparatorPos + 1); |
|
143 if (!idBlock.isEmpty()) |
|
144 transliterators.addElement(Transliterator::createInstance(idBlock, UTRANS_FORWARD, pe, ec), ec); |
|
145 if (!transes->isEmpty()) |
|
146 transliterators.addElement(transes->orphanElementAt(0), ec); |
|
147 blockSeparatorPos = aliasesOrRules.indexOf((UChar)(0xffff)); |
|
148 } |
|
149 if (!aliasesOrRules.isEmpty()) |
|
150 transliterators.addElement(Transliterator::createInstance(aliasesOrRules, UTRANS_FORWARD, pe, ec), ec); |
|
151 while (!transes->isEmpty()) |
|
152 transliterators.addElement(transes->orphanElementAt(0), ec); |
|
153 |
|
154 if (U_SUCCESS(ec)) { |
|
155 t = new CompoundTransliterator(ID, transliterators, |
|
156 (compoundFilter ? (UnicodeSet*)(compoundFilter->clone()) : 0), |
|
157 anonymousRBTs, pe, ec); |
|
158 if (t == 0) { |
|
159 ec = U_MEMORY_ALLOCATION_ERROR; |
|
160 return 0; |
|
161 } |
|
162 } else { |
|
163 for (int32_t i = 0; i < transliterators.size(); i++) |
|
164 delete (Transliterator*)(transliterators.elementAt(i)); |
|
165 } |
|
166 } |
|
167 break; |
|
168 case RULES: |
|
169 U_ASSERT(FALSE); // don't call create() if isRuleBased() returns TRUE! |
|
170 break; |
|
171 } |
|
172 return t; |
|
173 } |
|
174 |
|
175 UBool TransliteratorAlias::isRuleBased() const { |
|
176 return type == RULES; |
|
177 } |
|
178 |
|
179 void TransliteratorAlias::parse(TransliteratorParser& parser, |
|
180 UParseError& pe, UErrorCode& ec) const { |
|
181 U_ASSERT(type == RULES); |
|
182 if (U_FAILURE(ec)) { |
|
183 return; |
|
184 } |
|
185 |
|
186 parser.parse(aliasesOrRules, direction, pe, ec); |
|
187 } |
|
188 |
|
189 //---------------------------------------------------------------------- |
|
190 // class TransliteratorSpec |
|
191 //---------------------------------------------------------------------- |
|
192 |
|
193 /** |
|
194 * A TransliteratorSpec is a string specifying either a source or a target. In more |
|
195 * general terms, it may also specify a variant, but we only use the |
|
196 * Spec class for sources and targets. |
|
197 * |
|
198 * A Spec may be a locale or a script. If it is a locale, it has a |
|
199 * fallback chain that goes xx_YY_ZZZ -> xx_YY -> xx -> ssss, where |
|
200 * ssss is the script mapping of xx_YY_ZZZ. The Spec API methods |
|
201 * hasFallback(), next(), and reset() iterate over this fallback |
|
202 * sequence. |
|
203 * |
|
204 * The Spec class canonicalizes itself, so the locale is put into |
|
205 * canonical form, or the script is transformed from an abbreviation |
|
206 * to a full name. |
|
207 */ |
|
208 class TransliteratorSpec : public UMemory { |
|
209 public: |
|
210 TransliteratorSpec(const UnicodeString& spec); |
|
211 ~TransliteratorSpec(); |
|
212 |
|
213 const UnicodeString& get() const; |
|
214 UBool hasFallback() const; |
|
215 const UnicodeString& next(); |
|
216 void reset(); |
|
217 |
|
218 UBool isLocale() const; |
|
219 ResourceBundle& getBundle() const; |
|
220 |
|
221 operator const UnicodeString&() const { return get(); } |
|
222 const UnicodeString& getTop() const { return top; } |
|
223 |
|
224 private: |
|
225 void setupNext(); |
|
226 |
|
227 UnicodeString top; |
|
228 UnicodeString spec; |
|
229 UnicodeString nextSpec; |
|
230 UnicodeString scriptName; |
|
231 UBool isSpecLocale; // TRUE if spec is a locale |
|
232 UBool isNextLocale; // TRUE if nextSpec is a locale |
|
233 ResourceBundle* res; |
|
234 |
|
235 TransliteratorSpec(const TransliteratorSpec &other); // forbid copying of this class |
|
236 TransliteratorSpec &operator=(const TransliteratorSpec &other); // forbid copying of this class |
|
237 }; |
|
238 |
|
239 TransliteratorSpec::TransliteratorSpec(const UnicodeString& theSpec) |
|
240 : top(theSpec), |
|
241 res(0) |
|
242 { |
|
243 UErrorCode status = U_ZERO_ERROR; |
|
244 Locale topLoc(""); |
|
245 LocaleUtility::initLocaleFromName(theSpec, topLoc); |
|
246 if (!topLoc.isBogus()) { |
|
247 res = new ResourceBundle(U_ICUDATA_TRANSLIT, topLoc, status); |
|
248 /* test for NULL */ |
|
249 if (res == 0) { |
|
250 return; |
|
251 } |
|
252 if (U_FAILURE(status) || status == U_USING_DEFAULT_WARNING) { |
|
253 delete res; |
|
254 res = 0; |
|
255 } |
|
256 } |
|
257 |
|
258 // Canonicalize script name -or- do locale->script mapping |
|
259 status = U_ZERO_ERROR; |
|
260 static const int32_t capacity = 10; |
|
261 UScriptCode script[capacity]={USCRIPT_INVALID_CODE}; |
|
262 int32_t num = uscript_getCode(CharString().appendInvariantChars(theSpec, status).data(), |
|
263 script, capacity, &status); |
|
264 if (num > 0 && script[0] != USCRIPT_INVALID_CODE) { |
|
265 scriptName = UnicodeString(uscript_getName(script[0]), -1, US_INV); |
|
266 } |
|
267 |
|
268 // Canonicalize top |
|
269 if (res != 0) { |
|
270 // Canonicalize locale name |
|
271 UnicodeString locStr; |
|
272 LocaleUtility::initNameFromLocale(topLoc, locStr); |
|
273 if (!locStr.isBogus()) { |
|
274 top = locStr; |
|
275 } |
|
276 } else if (scriptName.length() != 0) { |
|
277 // We are a script; use canonical name |
|
278 top = scriptName; |
|
279 } |
|
280 |
|
281 // assert(spec != top); |
|
282 reset(); |
|
283 } |
|
284 |
|
285 TransliteratorSpec::~TransliteratorSpec() { |
|
286 delete res; |
|
287 } |
|
288 |
|
289 UBool TransliteratorSpec::hasFallback() const { |
|
290 return nextSpec.length() != 0; |
|
291 } |
|
292 |
|
293 void TransliteratorSpec::reset() { |
|
294 if (spec != top) { |
|
295 spec = top; |
|
296 isSpecLocale = (res != 0); |
|
297 setupNext(); |
|
298 } |
|
299 } |
|
300 |
|
301 void TransliteratorSpec::setupNext() { |
|
302 isNextLocale = FALSE; |
|
303 if (isSpecLocale) { |
|
304 nextSpec = spec; |
|
305 int32_t i = nextSpec.lastIndexOf(LOCALE_SEP); |
|
306 // If i == 0 then we have _FOO, so we fall through |
|
307 // to the scriptName. |
|
308 if (i > 0) { |
|
309 nextSpec.truncate(i); |
|
310 isNextLocale = TRUE; |
|
311 } else { |
|
312 nextSpec = scriptName; // scriptName may be empty |
|
313 } |
|
314 } else { |
|
315 // spec is a script, so we are at the end |
|
316 nextSpec.truncate(0); |
|
317 } |
|
318 } |
|
319 |
|
320 // Protocol: |
|
321 // for(const UnicodeString& s(spec.get()); |
|
322 // spec.hasFallback(); s(spec.next())) { ... |
|
323 |
|
324 const UnicodeString& TransliteratorSpec::next() { |
|
325 spec = nextSpec; |
|
326 isSpecLocale = isNextLocale; |
|
327 setupNext(); |
|
328 return spec; |
|
329 } |
|
330 |
|
331 const UnicodeString& TransliteratorSpec::get() const { |
|
332 return spec; |
|
333 } |
|
334 |
|
335 UBool TransliteratorSpec::isLocale() const { |
|
336 return isSpecLocale; |
|
337 } |
|
338 |
|
339 ResourceBundle& TransliteratorSpec::getBundle() const { |
|
340 return *res; |
|
341 } |
|
342 |
|
343 //---------------------------------------------------------------------- |
|
344 |
|
345 #ifdef DEBUG_MEM |
|
346 |
|
347 // Vector of Entry pointers currently in use |
|
348 static UVector* DEBUG_entries = NULL; |
|
349 |
|
350 static void DEBUG_setup() { |
|
351 if (DEBUG_entries == NULL) { |
|
352 UErrorCode ec = U_ZERO_ERROR; |
|
353 DEBUG_entries = new UVector(ec); |
|
354 } |
|
355 } |
|
356 |
|
357 // Caller must call DEBUG_setup first. Return index of given Entry, |
|
358 // if it is in use (not deleted yet), or -1 if not found. |
|
359 static int DEBUG_findEntry(TransliteratorEntry* e) { |
|
360 for (int i=0; i<DEBUG_entries->size(); ++i) { |
|
361 if (e == (TransliteratorEntry*) DEBUG_entries->elementAt(i)) { |
|
362 return i; |
|
363 } |
|
364 } |
|
365 return -1; |
|
366 } |
|
367 |
|
368 // Track object creation |
|
369 static void DEBUG_newEntry(TransliteratorEntry* e) { |
|
370 DEBUG_setup(); |
|
371 if (DEBUG_findEntry(e) >= 0) { |
|
372 // This should really never happen unless the heap is broken |
|
373 printf("ERROR DEBUG_newEntry duplicate new pointer %08X\n", e); |
|
374 return; |
|
375 } |
|
376 UErrorCode ec = U_ZERO_ERROR; |
|
377 DEBUG_entries->addElement(e, ec); |
|
378 } |
|
379 |
|
380 // Track object deletion |
|
381 static void DEBUG_delEntry(TransliteratorEntry* e) { |
|
382 DEBUG_setup(); |
|
383 int i = DEBUG_findEntry(e); |
|
384 if (i < 0) { |
|
385 printf("ERROR DEBUG_delEntry possible double deletion %08X\n", e); |
|
386 return; |
|
387 } |
|
388 DEBUG_entries->removeElementAt(i); |
|
389 } |
|
390 |
|
391 // Track object usage |
|
392 static void DEBUG_useEntry(TransliteratorEntry* e) { |
|
393 if (e == NULL) return; |
|
394 DEBUG_setup(); |
|
395 int i = DEBUG_findEntry(e); |
|
396 if (i < 0) { |
|
397 printf("ERROR DEBUG_useEntry possible dangling pointer %08X\n", e); |
|
398 } |
|
399 } |
|
400 |
|
401 #else |
|
402 // If we're not debugging then make these macros into NOPs |
|
403 #define DEBUG_newEntry(x) |
|
404 #define DEBUG_delEntry(x) |
|
405 #define DEBUG_useEntry(x) |
|
406 #endif |
|
407 |
|
408 //---------------------------------------------------------------------- |
|
409 // class Entry |
|
410 //---------------------------------------------------------------------- |
|
411 |
|
412 /** |
|
413 * The Entry object stores objects of different types and |
|
414 * singleton objects as placeholders for rule-based transliterators to |
|
415 * be built as needed. Instances of this struct can be placeholders, |
|
416 * can represent prototype transliterators to be cloned, or can |
|
417 * represent TransliteratorData objects. We don't support storing |
|
418 * classes in the registry because we don't have the rtti infrastructure |
|
419 * for it. We could easily add this if there is a need for it in the |
|
420 * future. |
|
421 */ |
|
422 class TransliteratorEntry : public UMemory { |
|
423 public: |
|
424 enum Type { |
|
425 RULES_FORWARD, |
|
426 RULES_REVERSE, |
|
427 LOCALE_RULES, |
|
428 PROTOTYPE, |
|
429 RBT_DATA, |
|
430 COMPOUND_RBT, |
|
431 ALIAS, |
|
432 FACTORY, |
|
433 NONE // Only used for uninitialized entries |
|
434 } entryType; |
|
435 // NOTE: stringArg cannot go inside the union because |
|
436 // it has a copy constructor |
|
437 UnicodeString stringArg; // For RULES_*, ALIAS, COMPOUND_RBT |
|
438 int32_t intArg; // For COMPOUND_RBT, LOCALE_RULES |
|
439 UnicodeSet* compoundFilter; // For COMPOUND_RBT |
|
440 union { |
|
441 Transliterator* prototype; // For PROTOTYPE |
|
442 TransliterationRuleData* data; // For RBT_DATA |
|
443 UVector* dataVector; // For COMPOUND_RBT |
|
444 struct { |
|
445 Transliterator::Factory function; |
|
446 Transliterator::Token context; |
|
447 } factory; // For FACTORY |
|
448 } u; |
|
449 TransliteratorEntry(); |
|
450 ~TransliteratorEntry(); |
|
451 void adoptPrototype(Transliterator* adopted); |
|
452 void setFactory(Transliterator::Factory factory, |
|
453 Transliterator::Token context); |
|
454 |
|
455 private: |
|
456 |
|
457 TransliteratorEntry(const TransliteratorEntry &other); // forbid copying of this class |
|
458 TransliteratorEntry &operator=(const TransliteratorEntry &other); // forbid copying of this class |
|
459 }; |
|
460 |
|
461 TransliteratorEntry::TransliteratorEntry() { |
|
462 u.prototype = 0; |
|
463 compoundFilter = NULL; |
|
464 entryType = NONE; |
|
465 DEBUG_newEntry(this); |
|
466 } |
|
467 |
|
468 TransliteratorEntry::~TransliteratorEntry() { |
|
469 DEBUG_delEntry(this); |
|
470 if (entryType == PROTOTYPE) { |
|
471 delete u.prototype; |
|
472 } else if (entryType == RBT_DATA) { |
|
473 // The data object is shared between instances of RBT. The |
|
474 // entry object owns it. It should only be deleted when the |
|
475 // transliterator component is being cleaned up. Doing so |
|
476 // invalidates any RBTs that the user has instantiated. |
|
477 delete u.data; |
|
478 } else if (entryType == COMPOUND_RBT) { |
|
479 while (u.dataVector != NULL && !u.dataVector->isEmpty()) |
|
480 delete (TransliterationRuleData*)u.dataVector->orphanElementAt(0); |
|
481 delete u.dataVector; |
|
482 } |
|
483 delete compoundFilter; |
|
484 } |
|
485 |
|
486 void TransliteratorEntry::adoptPrototype(Transliterator* adopted) { |
|
487 if (entryType == PROTOTYPE) { |
|
488 delete u.prototype; |
|
489 } |
|
490 entryType = PROTOTYPE; |
|
491 u.prototype = adopted; |
|
492 } |
|
493 |
|
494 void TransliteratorEntry::setFactory(Transliterator::Factory factory, |
|
495 Transliterator::Token context) { |
|
496 if (entryType == PROTOTYPE) { |
|
497 delete u.prototype; |
|
498 } |
|
499 entryType = FACTORY; |
|
500 u.factory.function = factory; |
|
501 u.factory.context = context; |
|
502 } |
|
503 |
|
504 // UObjectDeleter for Hashtable::setValueDeleter |
|
505 U_CDECL_BEGIN |
|
506 static void U_CALLCONV |
|
507 deleteEntry(void* obj) { |
|
508 delete (TransliteratorEntry*) obj; |
|
509 } |
|
510 U_CDECL_END |
|
511 |
|
512 //---------------------------------------------------------------------- |
|
513 // class TransliteratorRegistry: Basic public API |
|
514 //---------------------------------------------------------------------- |
|
515 |
|
516 TransliteratorRegistry::TransliteratorRegistry(UErrorCode& status) : |
|
517 registry(TRUE, status), |
|
518 specDAG(TRUE, status), |
|
519 availableIDs(status) |
|
520 { |
|
521 registry.setValueDeleter(deleteEntry); |
|
522 availableIDs.setDeleter(uprv_deleteUObject); |
|
523 availableIDs.setComparer(uhash_compareCaselessUnicodeString); |
|
524 specDAG.setValueDeleter(uhash_deleteHashtable); |
|
525 } |
|
526 |
|
527 TransliteratorRegistry::~TransliteratorRegistry() { |
|
528 // Through the magic of C++, everything cleans itself up |
|
529 } |
|
530 |
|
531 Transliterator* TransliteratorRegistry::get(const UnicodeString& ID, |
|
532 TransliteratorAlias*& aliasReturn, |
|
533 UErrorCode& status) { |
|
534 U_ASSERT(aliasReturn == NULL); |
|
535 TransliteratorEntry *entry = find(ID); |
|
536 return (entry == 0) ? 0 |
|
537 : instantiateEntry(ID, entry, aliasReturn, status); |
|
538 } |
|
539 |
|
540 Transliterator* TransliteratorRegistry::reget(const UnicodeString& ID, |
|
541 TransliteratorParser& parser, |
|
542 TransliteratorAlias*& aliasReturn, |
|
543 UErrorCode& status) { |
|
544 U_ASSERT(aliasReturn == NULL); |
|
545 TransliteratorEntry *entry = find(ID); |
|
546 |
|
547 if (entry == 0) { |
|
548 // We get to this point if there are two threads, one of which |
|
549 // is instantiating an ID, and another of which is removing |
|
550 // the same ID from the registry, and the timing is just right. |
|
551 return 0; |
|
552 } |
|
553 |
|
554 // The usage model for the caller is that they will first call |
|
555 // reg->get() inside the mutex, they'll get back an alias, they call |
|
556 // alias->isRuleBased(), and if they get TRUE, they call alias->parse() |
|
557 // outside the mutex, then reg->reget() inside the mutex again. A real |
|
558 // mess, but it gets things working for ICU 3.0. [alan]. |
|
559 |
|
560 // Note: It's possible that in between the caller calling |
|
561 // alias->parse() and reg->reget(), that another thread will have |
|
562 // called reg->reget(), and the entry will already have been fixed up. |
|
563 // We have to detect this so we don't stomp over existing entry |
|
564 // data members and potentially leak memory (u.data and compoundFilter). |
|
565 |
|
566 if (entry->entryType == TransliteratorEntry::RULES_FORWARD || |
|
567 entry->entryType == TransliteratorEntry::RULES_REVERSE || |
|
568 entry->entryType == TransliteratorEntry::LOCALE_RULES) { |
|
569 |
|
570 if (parser.idBlockVector.isEmpty() && parser.dataVector.isEmpty()) { |
|
571 entry->u.data = 0; |
|
572 entry->entryType = TransliteratorEntry::ALIAS; |
|
573 entry->stringArg = UNICODE_STRING_SIMPLE("Any-NULL"); |
|
574 } |
|
575 else if (parser.idBlockVector.isEmpty() && parser.dataVector.size() == 1) { |
|
576 entry->u.data = (TransliterationRuleData*)parser.dataVector.orphanElementAt(0); |
|
577 entry->entryType = TransliteratorEntry::RBT_DATA; |
|
578 } |
|
579 else if (parser.idBlockVector.size() == 1 && parser.dataVector.isEmpty()) { |
|
580 entry->stringArg = *(UnicodeString*)(parser.idBlockVector.elementAt(0)); |
|
581 entry->compoundFilter = parser.orphanCompoundFilter(); |
|
582 entry->entryType = TransliteratorEntry::ALIAS; |
|
583 } |
|
584 else { |
|
585 entry->entryType = TransliteratorEntry::COMPOUND_RBT; |
|
586 entry->compoundFilter = parser.orphanCompoundFilter(); |
|
587 entry->u.dataVector = new UVector(status); |
|
588 entry->stringArg.remove(); |
|
589 |
|
590 int32_t limit = parser.idBlockVector.size(); |
|
591 if (parser.dataVector.size() > limit) |
|
592 limit = parser.dataVector.size(); |
|
593 |
|
594 for (int32_t i = 0; i < limit; i++) { |
|
595 if (i < parser.idBlockVector.size()) { |
|
596 UnicodeString* idBlock = (UnicodeString*)parser.idBlockVector.elementAt(i); |
|
597 if (!idBlock->isEmpty()) |
|
598 entry->stringArg += *idBlock; |
|
599 } |
|
600 if (!parser.dataVector.isEmpty()) { |
|
601 TransliterationRuleData* data = (TransliterationRuleData*)parser.dataVector.orphanElementAt(0); |
|
602 entry->u.dataVector->addElement(data, status); |
|
603 entry->stringArg += (UChar)0xffff; // use U+FFFF to mark position of RBTs in ID block |
|
604 } |
|
605 } |
|
606 } |
|
607 } |
|
608 |
|
609 Transliterator *t = |
|
610 instantiateEntry(ID, entry, aliasReturn, status); |
|
611 return t; |
|
612 } |
|
613 |
|
614 void TransliteratorRegistry::put(Transliterator* adoptedProto, |
|
615 UBool visible, |
|
616 UErrorCode& ec) |
|
617 { |
|
618 TransliteratorEntry *entry = new TransliteratorEntry(); |
|
619 if (entry == NULL) { |
|
620 ec = U_MEMORY_ALLOCATION_ERROR; |
|
621 return; |
|
622 } |
|
623 entry->adoptPrototype(adoptedProto); |
|
624 registerEntry(adoptedProto->getID(), entry, visible); |
|
625 } |
|
626 |
|
627 void TransliteratorRegistry::put(const UnicodeString& ID, |
|
628 Transliterator::Factory factory, |
|
629 Transliterator::Token context, |
|
630 UBool visible, |
|
631 UErrorCode& ec) { |
|
632 TransliteratorEntry *entry = new TransliteratorEntry(); |
|
633 if (entry == NULL) { |
|
634 ec = U_MEMORY_ALLOCATION_ERROR; |
|
635 return; |
|
636 } |
|
637 entry->setFactory(factory, context); |
|
638 registerEntry(ID, entry, visible); |
|
639 } |
|
640 |
|
641 void TransliteratorRegistry::put(const UnicodeString& ID, |
|
642 const UnicodeString& resourceName, |
|
643 UTransDirection dir, |
|
644 UBool readonlyResourceAlias, |
|
645 UBool visible, |
|
646 UErrorCode& ec) { |
|
647 TransliteratorEntry *entry = new TransliteratorEntry(); |
|
648 if (entry == NULL) { |
|
649 ec = U_MEMORY_ALLOCATION_ERROR; |
|
650 return; |
|
651 } |
|
652 entry->entryType = (dir == UTRANS_FORWARD) ? TransliteratorEntry::RULES_FORWARD |
|
653 : TransliteratorEntry::RULES_REVERSE; |
|
654 if (readonlyResourceAlias) { |
|
655 entry->stringArg.setTo(TRUE, resourceName.getBuffer(), -1); |
|
656 } |
|
657 else { |
|
658 entry->stringArg = resourceName; |
|
659 } |
|
660 registerEntry(ID, entry, visible); |
|
661 } |
|
662 |
|
663 void TransliteratorRegistry::put(const UnicodeString& ID, |
|
664 const UnicodeString& alias, |
|
665 UBool readonlyAliasAlias, |
|
666 UBool visible, |
|
667 UErrorCode& /*ec*/) { |
|
668 TransliteratorEntry *entry = new TransliteratorEntry(); |
|
669 // Null pointer check |
|
670 if (entry != NULL) { |
|
671 entry->entryType = TransliteratorEntry::ALIAS; |
|
672 if (readonlyAliasAlias) { |
|
673 entry->stringArg.setTo(TRUE, alias.getBuffer(), -1); |
|
674 } |
|
675 else { |
|
676 entry->stringArg = alias; |
|
677 } |
|
678 registerEntry(ID, entry, visible); |
|
679 } |
|
680 } |
|
681 |
|
682 void TransliteratorRegistry::remove(const UnicodeString& ID) { |
|
683 UnicodeString source, target, variant; |
|
684 UBool sawSource; |
|
685 TransliteratorIDParser::IDtoSTV(ID, source, target, variant, sawSource); |
|
686 // Only need to do this if ID.indexOf('-') < 0 |
|
687 UnicodeString id; |
|
688 TransliteratorIDParser::STVtoID(source, target, variant, id); |
|
689 registry.remove(id); |
|
690 removeSTV(source, target, variant); |
|
691 availableIDs.removeElement((void*) &id); |
|
692 } |
|
693 |
|
694 //---------------------------------------------------------------------- |
|
695 // class TransliteratorRegistry: Public ID and spec management |
|
696 //---------------------------------------------------------------------- |
|
697 |
|
698 /** |
|
699 * == OBSOLETE - remove in ICU 3.4 == |
|
700 * Return the number of IDs currently registered with the system. |
|
701 * To retrieve the actual IDs, call getAvailableID(i) with |
|
702 * i from 0 to countAvailableIDs() - 1. |
|
703 */ |
|
704 int32_t TransliteratorRegistry::countAvailableIDs(void) const { |
|
705 return availableIDs.size(); |
|
706 } |
|
707 |
|
708 /** |
|
709 * == OBSOLETE - remove in ICU 3.4 == |
|
710 * Return the index-th available ID. index must be between 0 |
|
711 * and countAvailableIDs() - 1, inclusive. If index is out of |
|
712 * range, the result of getAvailableID(0) is returned. |
|
713 */ |
|
714 const UnicodeString& TransliteratorRegistry::getAvailableID(int32_t index) const { |
|
715 if (index < 0 || index >= availableIDs.size()) { |
|
716 index = 0; |
|
717 } |
|
718 return *(const UnicodeString*) availableIDs[index]; |
|
719 } |
|
720 |
|
721 StringEnumeration* TransliteratorRegistry::getAvailableIDs() const { |
|
722 return new Enumeration(*this); |
|
723 } |
|
724 |
|
725 int32_t TransliteratorRegistry::countAvailableSources(void) const { |
|
726 return specDAG.count(); |
|
727 } |
|
728 |
|
729 UnicodeString& TransliteratorRegistry::getAvailableSource(int32_t index, |
|
730 UnicodeString& result) const { |
|
731 int32_t pos = -1; |
|
732 const UHashElement *e = 0; |
|
733 while (index-- >= 0) { |
|
734 e = specDAG.nextElement(pos); |
|
735 if (e == 0) { |
|
736 break; |
|
737 } |
|
738 } |
|
739 if (e == 0) { |
|
740 result.truncate(0); |
|
741 } else { |
|
742 result = *(UnicodeString*) e->key.pointer; |
|
743 } |
|
744 return result; |
|
745 } |
|
746 |
|
747 int32_t TransliteratorRegistry::countAvailableTargets(const UnicodeString& source) const { |
|
748 Hashtable *targets = (Hashtable*) specDAG.get(source); |
|
749 return (targets == 0) ? 0 : targets->count(); |
|
750 } |
|
751 |
|
752 UnicodeString& TransliteratorRegistry::getAvailableTarget(int32_t index, |
|
753 const UnicodeString& source, |
|
754 UnicodeString& result) const { |
|
755 Hashtable *targets = (Hashtable*) specDAG.get(source); |
|
756 if (targets == 0) { |
|
757 result.truncate(0); // invalid source |
|
758 return result; |
|
759 } |
|
760 int32_t pos = -1; |
|
761 const UHashElement *e = 0; |
|
762 while (index-- >= 0) { |
|
763 e = targets->nextElement(pos); |
|
764 if (e == 0) { |
|
765 break; |
|
766 } |
|
767 } |
|
768 if (e == 0) { |
|
769 result.truncate(0); // invalid index |
|
770 } else { |
|
771 result = *(UnicodeString*) e->key.pointer; |
|
772 } |
|
773 return result; |
|
774 } |
|
775 |
|
776 int32_t TransliteratorRegistry::countAvailableVariants(const UnicodeString& source, |
|
777 const UnicodeString& target) const { |
|
778 Hashtable *targets = (Hashtable*) specDAG.get(source); |
|
779 if (targets == 0) { |
|
780 return 0; |
|
781 } |
|
782 UVector *variants = (UVector*) targets->get(target); |
|
783 // variants may be 0 if the source/target are invalid |
|
784 return (variants == 0) ? 0 : variants->size(); |
|
785 } |
|
786 |
|
787 UnicodeString& TransliteratorRegistry::getAvailableVariant(int32_t index, |
|
788 const UnicodeString& source, |
|
789 const UnicodeString& target, |
|
790 UnicodeString& result) const { |
|
791 Hashtable *targets = (Hashtable*) specDAG.get(source); |
|
792 if (targets == 0) { |
|
793 result.truncate(0); // invalid source |
|
794 return result; |
|
795 } |
|
796 UVector *variants = (UVector*) targets->get(target); |
|
797 if (variants == 0) { |
|
798 result.truncate(0); // invalid target |
|
799 return result; |
|
800 } |
|
801 UnicodeString *v = (UnicodeString*) variants->elementAt(index); |
|
802 if (v == 0) { |
|
803 result.truncate(0); // invalid index |
|
804 } else { |
|
805 result = *v; |
|
806 } |
|
807 return result; |
|
808 } |
|
809 |
|
810 //---------------------------------------------------------------------- |
|
811 // class TransliteratorRegistry::Enumeration |
|
812 //---------------------------------------------------------------------- |
|
813 |
|
814 TransliteratorRegistry::Enumeration::Enumeration(const TransliteratorRegistry& _reg) : |
|
815 index(0), reg(_reg) { |
|
816 } |
|
817 |
|
818 TransliteratorRegistry::Enumeration::~Enumeration() { |
|
819 } |
|
820 |
|
821 int32_t TransliteratorRegistry::Enumeration::count(UErrorCode& /*status*/) const { |
|
822 return reg.availableIDs.size(); |
|
823 } |
|
824 |
|
825 const UnicodeString* TransliteratorRegistry::Enumeration::snext(UErrorCode& status) { |
|
826 // This is sloppy but safe -- if we get out of sync with the underlying |
|
827 // registry, we will still return legal strings, but they might not |
|
828 // correspond to the snapshot at construction time. So there could be |
|
829 // duplicate IDs or omitted IDs if insertions or deletions occur in one |
|
830 // thread while another is iterating. To be more rigorous, add a timestamp, |
|
831 // which is incremented with any modification, and validate this iterator |
|
832 // against the timestamp at construction time. This probably isn't worth |
|
833 // doing as long as there is some possibility of removing this code in favor |
|
834 // of some new code based on Doug's service framework. |
|
835 if (U_FAILURE(status)) { |
|
836 return NULL; |
|
837 } |
|
838 int32_t n = reg.availableIDs.size(); |
|
839 if (index > n) { |
|
840 status = U_ENUM_OUT_OF_SYNC_ERROR; |
|
841 } |
|
842 // index == n is okay -- this means we've reached the end |
|
843 if (index < n) { |
|
844 // Copy the string! This avoids lifetime problems. |
|
845 unistr = *(const UnicodeString*)reg.availableIDs[index++]; |
|
846 return &unistr; |
|
847 } else { |
|
848 return NULL; |
|
849 } |
|
850 } |
|
851 |
|
852 void TransliteratorRegistry::Enumeration::reset(UErrorCode& /*status*/) { |
|
853 index = 0; |
|
854 } |
|
855 |
|
856 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(TransliteratorRegistry::Enumeration) |
|
857 |
|
858 //---------------------------------------------------------------------- |
|
859 // class TransliteratorRegistry: internal |
|
860 //---------------------------------------------------------------------- |
|
861 |
|
862 /** |
|
863 * Convenience method. Calls 6-arg registerEntry(). |
|
864 */ |
|
865 void TransliteratorRegistry::registerEntry(const UnicodeString& source, |
|
866 const UnicodeString& target, |
|
867 const UnicodeString& variant, |
|
868 TransliteratorEntry* adopted, |
|
869 UBool visible) { |
|
870 UnicodeString ID; |
|
871 UnicodeString s(source); |
|
872 if (s.length() == 0) { |
|
873 s.setTo(TRUE, ANY, 3); |
|
874 } |
|
875 TransliteratorIDParser::STVtoID(source, target, variant, ID); |
|
876 registerEntry(ID, s, target, variant, adopted, visible); |
|
877 } |
|
878 |
|
879 /** |
|
880 * Convenience method. Calls 6-arg registerEntry(). |
|
881 */ |
|
882 void TransliteratorRegistry::registerEntry(const UnicodeString& ID, |
|
883 TransliteratorEntry* adopted, |
|
884 UBool visible) { |
|
885 UnicodeString source, target, variant; |
|
886 UBool sawSource; |
|
887 TransliteratorIDParser::IDtoSTV(ID, source, target, variant, sawSource); |
|
888 // Only need to do this if ID.indexOf('-') < 0 |
|
889 UnicodeString id; |
|
890 TransliteratorIDParser::STVtoID(source, target, variant, id); |
|
891 registerEntry(id, source, target, variant, adopted, visible); |
|
892 } |
|
893 |
|
894 /** |
|
895 * Register an entry object (adopted) with the given ID, source, |
|
896 * target, and variant strings. |
|
897 */ |
|
898 void TransliteratorRegistry::registerEntry(const UnicodeString& ID, |
|
899 const UnicodeString& source, |
|
900 const UnicodeString& target, |
|
901 const UnicodeString& variant, |
|
902 TransliteratorEntry* adopted, |
|
903 UBool visible) { |
|
904 UErrorCode status = U_ZERO_ERROR; |
|
905 registry.put(ID, adopted, status); |
|
906 if (visible) { |
|
907 registerSTV(source, target, variant); |
|
908 if (!availableIDs.contains((void*) &ID)) { |
|
909 UnicodeString *newID = (UnicodeString *)ID.clone(); |
|
910 // Check to make sure newID was created. |
|
911 if (newID != NULL) { |
|
912 // NUL-terminate the ID string |
|
913 newID->getTerminatedBuffer(); |
|
914 availableIDs.addElement(newID, status); |
|
915 } |
|
916 } |
|
917 } else { |
|
918 removeSTV(source, target, variant); |
|
919 availableIDs.removeElement((void*) &ID); |
|
920 } |
|
921 } |
|
922 |
|
923 /** |
|
924 * Register a source-target/variant in the specDAG. Variant may be |
|
925 * empty, but source and target must not be. If variant is empty then |
|
926 * the special variant NO_VARIANT is stored in slot zero of the |
|
927 * UVector of variants. |
|
928 */ |
|
929 void TransliteratorRegistry::registerSTV(const UnicodeString& source, |
|
930 const UnicodeString& target, |
|
931 const UnicodeString& variant) { |
|
932 // assert(source.length() > 0); |
|
933 // assert(target.length() > 0); |
|
934 UErrorCode status = U_ZERO_ERROR; |
|
935 Hashtable *targets = (Hashtable*) specDAG.get(source); |
|
936 if (targets == 0) { |
|
937 targets = new Hashtable(TRUE, status); |
|
938 if (U_FAILURE(status) || targets == 0) { |
|
939 return; |
|
940 } |
|
941 targets->setValueDeleter(uprv_deleteUObject); |
|
942 specDAG.put(source, targets, status); |
|
943 } |
|
944 UVector *variants = (UVector*) targets->get(target); |
|
945 if (variants == 0) { |
|
946 variants = new UVector(uprv_deleteUObject, |
|
947 uhash_compareCaselessUnicodeString, status); |
|
948 if (variants == 0) { |
|
949 return; |
|
950 } |
|
951 targets->put(target, variants, status); |
|
952 } |
|
953 // assert(NO_VARIANT == ""); |
|
954 // We add the variant string. If it is the special "no variant" |
|
955 // string, that is, the empty string, we add it at position zero. |
|
956 if (!variants->contains((void*) &variant)) { |
|
957 UnicodeString *tempus; // Used for null pointer check. |
|
958 if (variant.length() > 0) { |
|
959 tempus = new UnicodeString(variant); |
|
960 if (tempus != NULL) { |
|
961 variants->addElement(tempus, status); |
|
962 } |
|
963 } else { |
|
964 tempus = new UnicodeString(); // = NO_VARIANT |
|
965 if (tempus != NULL) { |
|
966 variants->insertElementAt(tempus, 0, status); |
|
967 } |
|
968 } |
|
969 } |
|
970 } |
|
971 |
|
972 /** |
|
973 * Remove a source-target/variant from the specDAG. |
|
974 */ |
|
975 void TransliteratorRegistry::removeSTV(const UnicodeString& source, |
|
976 const UnicodeString& target, |
|
977 const UnicodeString& variant) { |
|
978 // assert(source.length() > 0); |
|
979 // assert(target.length() > 0); |
|
980 // UErrorCode status = U_ZERO_ERROR; |
|
981 Hashtable *targets = (Hashtable*) specDAG.get(source); |
|
982 if (targets == 0) { |
|
983 return; // should never happen for valid s-t/v |
|
984 } |
|
985 UVector *variants = (UVector*) targets->get(target); |
|
986 if (variants == 0) { |
|
987 return; // should never happen for valid s-t/v |
|
988 } |
|
989 variants->removeElement((void*) &variant); |
|
990 if (variants->size() == 0) { |
|
991 targets->remove(target); // should delete variants |
|
992 if (targets->count() == 0) { |
|
993 specDAG.remove(source); // should delete targets |
|
994 } |
|
995 } |
|
996 } |
|
997 |
|
998 /** |
|
999 * Attempt to find a source-target/variant in the dynamic registry |
|
1000 * store. Return 0 on failure. |
|
1001 * |
|
1002 * Caller does NOT own returned object. |
|
1003 */ |
|
1004 TransliteratorEntry* TransliteratorRegistry::findInDynamicStore(const TransliteratorSpec& src, |
|
1005 const TransliteratorSpec& trg, |
|
1006 const UnicodeString& variant) const { |
|
1007 UnicodeString ID; |
|
1008 TransliteratorIDParser::STVtoID(src, trg, variant, ID); |
|
1009 TransliteratorEntry *e = (TransliteratorEntry*) registry.get(ID); |
|
1010 DEBUG_useEntry(e); |
|
1011 return e; |
|
1012 } |
|
1013 |
|
1014 /** |
|
1015 * Attempt to find a source-target/variant in the static locale |
|
1016 * resource store. Do not perform fallback. Return 0 on failure. |
|
1017 * |
|
1018 * On success, create a new entry object, register it in the dynamic |
|
1019 * store, and return a pointer to it, but do not make it public -- |
|
1020 * just because someone requested something, we do not expand the |
|
1021 * available ID list (or spec DAG). |
|
1022 * |
|
1023 * Caller does NOT own returned object. |
|
1024 */ |
|
1025 TransliteratorEntry* TransliteratorRegistry::findInStaticStore(const TransliteratorSpec& src, |
|
1026 const TransliteratorSpec& trg, |
|
1027 const UnicodeString& variant) { |
|
1028 TransliteratorEntry* entry = 0; |
|
1029 if (src.isLocale()) { |
|
1030 entry = findInBundle(src, trg, variant, UTRANS_FORWARD); |
|
1031 } else if (trg.isLocale()) { |
|
1032 entry = findInBundle(trg, src, variant, UTRANS_REVERSE); |
|
1033 } |
|
1034 |
|
1035 // If we found an entry, store it in the Hashtable for next |
|
1036 // time. |
|
1037 if (entry != 0) { |
|
1038 registerEntry(src.getTop(), trg.getTop(), variant, entry, FALSE); |
|
1039 } |
|
1040 |
|
1041 return entry; |
|
1042 } |
|
1043 |
|
1044 // As of 2.0, resource bundle keys cannot contain '_' |
|
1045 static const UChar TRANSLITERATE_TO[] = {84,114,97,110,115,108,105,116,101,114,97,116,101,84,111,0}; // "TransliterateTo" |
|
1046 |
|
1047 static const UChar TRANSLITERATE_FROM[] = {84,114,97,110,115,108,105,116,101,114,97,116,101,70,114,111,109,0}; // "TransliterateFrom" |
|
1048 |
|
1049 static const UChar TRANSLITERATE[] = {84,114,97,110,115,108,105,116,101,114,97,116,101,0}; // "Transliterate" |
|
1050 |
|
1051 /** |
|
1052 * Attempt to find an entry in a single resource bundle. This is |
|
1053 * a one-sided lookup. findInStaticStore() performs up to two such |
|
1054 * lookups, one for the source, and one for the target. |
|
1055 * |
|
1056 * Do not perform fallback. Return 0 on failure. |
|
1057 * |
|
1058 * On success, create a new Entry object, populate it, and return it. |
|
1059 * The caller owns the returned object. |
|
1060 */ |
|
1061 TransliteratorEntry* TransliteratorRegistry::findInBundle(const TransliteratorSpec& specToOpen, |
|
1062 const TransliteratorSpec& specToFind, |
|
1063 const UnicodeString& variant, |
|
1064 UTransDirection direction) |
|
1065 { |
|
1066 UnicodeString utag; |
|
1067 UnicodeString resStr; |
|
1068 int32_t pass; |
|
1069 |
|
1070 for (pass=0; pass<2; ++pass) { |
|
1071 utag.truncate(0); |
|
1072 // First try either TransliteratorTo_xxx or |
|
1073 // TransliterateFrom_xxx, then try the bidirectional |
|
1074 // Transliterate_xxx. This precedence order is arbitrary |
|
1075 // but must be consistent and documented. |
|
1076 if (pass == 0) { |
|
1077 utag.append(direction == UTRANS_FORWARD ? |
|
1078 TRANSLITERATE_TO : TRANSLITERATE_FROM, -1); |
|
1079 } else { |
|
1080 utag.append(TRANSLITERATE, -1); |
|
1081 } |
|
1082 UnicodeString s(specToFind.get()); |
|
1083 utag.append(s.toUpper("")); |
|
1084 UErrorCode status = U_ZERO_ERROR; |
|
1085 ResourceBundle subres(specToOpen.getBundle().get( |
|
1086 CharString().appendInvariantChars(utag, status).data(), status)); |
|
1087 if (U_FAILURE(status) || status == U_USING_DEFAULT_WARNING) { |
|
1088 continue; |
|
1089 } |
|
1090 |
|
1091 s.truncate(0); |
|
1092 if (specToOpen.get() != LocaleUtility::initNameFromLocale(subres.getLocale(), s)) { |
|
1093 continue; |
|
1094 } |
|
1095 |
|
1096 if (variant.length() != 0) { |
|
1097 status = U_ZERO_ERROR; |
|
1098 resStr = subres.getStringEx( |
|
1099 CharString().appendInvariantChars(variant, status).data(), status); |
|
1100 if (U_SUCCESS(status)) { |
|
1101 // Exit loop successfully |
|
1102 break; |
|
1103 } |
|
1104 } else { |
|
1105 // Variant is empty, which means match the first variant listed. |
|
1106 status = U_ZERO_ERROR; |
|
1107 resStr = subres.getStringEx(1, status); |
|
1108 if (U_SUCCESS(status)) { |
|
1109 // Exit loop successfully |
|
1110 break; |
|
1111 } |
|
1112 } |
|
1113 } |
|
1114 |
|
1115 if (pass==2) { |
|
1116 // Failed |
|
1117 return NULL; |
|
1118 } |
|
1119 |
|
1120 // We have succeeded in loading a string from the locale |
|
1121 // resources. Create a new registry entry to hold it and return it. |
|
1122 TransliteratorEntry *entry = new TransliteratorEntry(); |
|
1123 if (entry != 0) { |
|
1124 // The direction is always forward for the |
|
1125 // TransliterateTo_xxx and TransliterateFrom_xxx |
|
1126 // items; those are unidirectional forward rules. |
|
1127 // For the bidirectional Transliterate_xxx items, |
|
1128 // the direction is the value passed in to this |
|
1129 // function. |
|
1130 int32_t dir = (pass == 0) ? UTRANS_FORWARD : direction; |
|
1131 entry->entryType = TransliteratorEntry::LOCALE_RULES; |
|
1132 entry->stringArg = resStr; |
|
1133 entry->intArg = dir; |
|
1134 } |
|
1135 |
|
1136 return entry; |
|
1137 } |
|
1138 |
|
1139 /** |
|
1140 * Convenience method. Calls 3-arg find(). |
|
1141 */ |
|
1142 TransliteratorEntry* TransliteratorRegistry::find(const UnicodeString& ID) { |
|
1143 UnicodeString source, target, variant; |
|
1144 UBool sawSource; |
|
1145 TransliteratorIDParser::IDtoSTV(ID, source, target, variant, sawSource); |
|
1146 return find(source, target, variant); |
|
1147 } |
|
1148 |
|
1149 /** |
|
1150 * Top-level find method. Attempt to find a source-target/variant in |
|
1151 * either the dynamic or the static (locale resource) store. Perform |
|
1152 * fallback. |
|
1153 * |
|
1154 * Lookup sequence for ss_SS_SSS-tt_TT_TTT/v: |
|
1155 * |
|
1156 * ss_SS_SSS-tt_TT_TTT/v -- in hashtable |
|
1157 * ss_SS_SSS-tt_TT_TTT/v -- in ss_SS_SSS (no fallback) |
|
1158 * |
|
1159 * repeat with t = tt_TT_TTT, tt_TT, tt, and tscript |
|
1160 * |
|
1161 * ss_SS_SSS-t/ * |
|
1162 * ss_SS-t/ * |
|
1163 * ss-t/ * |
|
1164 * sscript-t/ * |
|
1165 * |
|
1166 * Here * matches the first variant listed. |
|
1167 * |
|
1168 * Caller does NOT own returned object. Return 0 on failure. |
|
1169 */ |
|
1170 TransliteratorEntry* TransliteratorRegistry::find(UnicodeString& source, |
|
1171 UnicodeString& target, |
|
1172 UnicodeString& variant) { |
|
1173 |
|
1174 TransliteratorSpec src(source); |
|
1175 TransliteratorSpec trg(target); |
|
1176 TransliteratorEntry* entry; |
|
1177 |
|
1178 // Seek exact match in hashtable. Temporary fix for ICU 4.6. |
|
1179 // TODO: The general logic for finding a matching transliterator needs to be reviewed. |
|
1180 // ICU ticket #8089 |
|
1181 UnicodeString ID; |
|
1182 TransliteratorIDParser::STVtoID(source, target, variant, ID); |
|
1183 entry = (TransliteratorEntry*) registry.get(ID); |
|
1184 if (entry != 0) { |
|
1185 // std::string ss; |
|
1186 // std::cout << ID.toUTF8String(ss) << std::endl; |
|
1187 return entry; |
|
1188 } |
|
1189 |
|
1190 if (variant.length() != 0) { |
|
1191 |
|
1192 // Seek exact match in hashtable |
|
1193 entry = findInDynamicStore(src, trg, variant); |
|
1194 if (entry != 0) { |
|
1195 return entry; |
|
1196 } |
|
1197 |
|
1198 // Seek exact match in locale resources |
|
1199 entry = findInStaticStore(src, trg, variant); |
|
1200 if (entry != 0) { |
|
1201 return entry; |
|
1202 } |
|
1203 } |
|
1204 |
|
1205 for (;;) { |
|
1206 src.reset(); |
|
1207 for (;;) { |
|
1208 // Seek match in hashtable |
|
1209 entry = findInDynamicStore(src, trg, NO_VARIANT); |
|
1210 if (entry != 0) { |
|
1211 return entry; |
|
1212 } |
|
1213 |
|
1214 // Seek match in locale resources |
|
1215 entry = findInStaticStore(src, trg, NO_VARIANT); |
|
1216 if (entry != 0) { |
|
1217 return entry; |
|
1218 } |
|
1219 if (!src.hasFallback()) { |
|
1220 break; |
|
1221 } |
|
1222 src.next(); |
|
1223 } |
|
1224 if (!trg.hasFallback()) { |
|
1225 break; |
|
1226 } |
|
1227 trg.next(); |
|
1228 } |
|
1229 |
|
1230 return 0; |
|
1231 } |
|
1232 |
|
1233 /** |
|
1234 * Given an Entry object, instantiate it. Caller owns result. Return |
|
1235 * 0 on failure. |
|
1236 * |
|
1237 * Return a non-empty aliasReturn value if the ID points to an alias. |
|
1238 * We cannot instantiate it ourselves because the alias may contain |
|
1239 * filters or compounds, which we do not understand. Caller should |
|
1240 * make aliasReturn empty before calling. |
|
1241 * |
|
1242 * The entry object is assumed to reside in the dynamic store. It may be |
|
1243 * modified. |
|
1244 */ |
|
1245 Transliterator* TransliteratorRegistry::instantiateEntry(const UnicodeString& ID, |
|
1246 TransliteratorEntry *entry, |
|
1247 TransliteratorAlias* &aliasReturn, |
|
1248 UErrorCode& status) { |
|
1249 Transliterator *t = 0; |
|
1250 U_ASSERT(aliasReturn == 0); |
|
1251 |
|
1252 switch (entry->entryType) { |
|
1253 case TransliteratorEntry::RBT_DATA: |
|
1254 t = new RuleBasedTransliterator(ID, entry->u.data); |
|
1255 if (t == 0) { |
|
1256 status = U_MEMORY_ALLOCATION_ERROR; |
|
1257 } |
|
1258 return t; |
|
1259 case TransliteratorEntry::PROTOTYPE: |
|
1260 t = entry->u.prototype->clone(); |
|
1261 if (t == 0) { |
|
1262 status = U_MEMORY_ALLOCATION_ERROR; |
|
1263 } |
|
1264 return t; |
|
1265 case TransliteratorEntry::ALIAS: |
|
1266 aliasReturn = new TransliteratorAlias(entry->stringArg, entry->compoundFilter); |
|
1267 if (aliasReturn == 0) { |
|
1268 status = U_MEMORY_ALLOCATION_ERROR; |
|
1269 } |
|
1270 return 0; |
|
1271 case TransliteratorEntry::FACTORY: |
|
1272 t = entry->u.factory.function(ID, entry->u.factory.context); |
|
1273 if (t == 0) { |
|
1274 status = U_MEMORY_ALLOCATION_ERROR; |
|
1275 } |
|
1276 return t; |
|
1277 case TransliteratorEntry::COMPOUND_RBT: |
|
1278 { |
|
1279 UVector* rbts = new UVector(entry->u.dataVector->size(), status); |
|
1280 // Check for null pointer |
|
1281 if (rbts == NULL) { |
|
1282 status = U_MEMORY_ALLOCATION_ERROR; |
|
1283 return NULL; |
|
1284 } |
|
1285 int32_t passNumber = 1; |
|
1286 for (int32_t i = 0; U_SUCCESS(status) && i < entry->u.dataVector->size(); i++) { |
|
1287 // TODO: Should passNumber be turned into a decimal-string representation (1 -> "1")? |
|
1288 Transliterator* t = new RuleBasedTransliterator(UnicodeString(CompoundTransliterator::PASS_STRING) + UnicodeString(passNumber++), |
|
1289 (TransliterationRuleData*)(entry->u.dataVector->elementAt(i)), FALSE); |
|
1290 if (t == 0) |
|
1291 status = U_MEMORY_ALLOCATION_ERROR; |
|
1292 else |
|
1293 rbts->addElement(t, status); |
|
1294 } |
|
1295 if (U_FAILURE(status)) { |
|
1296 delete rbts; |
|
1297 return 0; |
|
1298 } |
|
1299 aliasReturn = new TransliteratorAlias(ID, entry->stringArg, rbts, entry->compoundFilter); |
|
1300 } |
|
1301 if (aliasReturn == 0) { |
|
1302 status = U_MEMORY_ALLOCATION_ERROR; |
|
1303 } |
|
1304 return 0; |
|
1305 case TransliteratorEntry::LOCALE_RULES: |
|
1306 aliasReturn = new TransliteratorAlias(ID, entry->stringArg, |
|
1307 (UTransDirection) entry->intArg); |
|
1308 if (aliasReturn == 0) { |
|
1309 status = U_MEMORY_ALLOCATION_ERROR; |
|
1310 } |
|
1311 return 0; |
|
1312 case TransliteratorEntry::RULES_FORWARD: |
|
1313 case TransliteratorEntry::RULES_REVERSE: |
|
1314 // Process the rule data into a TransliteratorRuleData object, |
|
1315 // and possibly also into an ::id header and/or footer. Then |
|
1316 // we modify the registry with the parsed data and retry. |
|
1317 { |
|
1318 TransliteratorParser parser(status); |
|
1319 |
|
1320 // We use the file name, taken from another resource bundle |
|
1321 // 2-d array at static init time, as a locale language. We're |
|
1322 // just using the locale mechanism to map through to a file |
|
1323 // name; this in no way represents an actual locale. |
|
1324 //CharString ch(entry->stringArg); |
|
1325 //UResourceBundle *bundle = ures_openDirect(0, ch, &status); |
|
1326 UnicodeString rules = entry->stringArg; |
|
1327 //ures_close(bundle); |
|
1328 |
|
1329 //if (U_FAILURE(status)) { |
|
1330 // We have a failure of some kind. Remove the ID from the |
|
1331 // registry so we don't keep trying. NOTE: This will throw off |
|
1332 // anyone who is, at the moment, trying to iterate over the |
|
1333 // available IDs. That's acceptable since we should never |
|
1334 // really get here except under installation, configuration, |
|
1335 // or unrecoverable run time memory failures. |
|
1336 // remove(ID); |
|
1337 //} else { |
|
1338 |
|
1339 // If the status indicates a failure, then we don't have any |
|
1340 // rules -- there is probably an installation error. The list |
|
1341 // in the root locale should correspond to all the installed |
|
1342 // transliterators; if it lists something that's not |
|
1343 // installed, we'll get an error from ResourceBundle. |
|
1344 aliasReturn = new TransliteratorAlias(ID, rules, |
|
1345 ((entry->entryType == TransliteratorEntry::RULES_REVERSE) ? |
|
1346 UTRANS_REVERSE : UTRANS_FORWARD)); |
|
1347 if (aliasReturn == 0) { |
|
1348 status = U_MEMORY_ALLOCATION_ERROR; |
|
1349 } |
|
1350 //} |
|
1351 } |
|
1352 return 0; |
|
1353 default: |
|
1354 U_ASSERT(FALSE); // can't get here |
|
1355 return 0; |
|
1356 } |
|
1357 } |
|
1358 U_NAMESPACE_END |
|
1359 |
|
1360 #endif /* #if !UCONFIG_NO_TRANSLITERATION */ |
|
1361 |
|
1362 //eof |