Victor Chang | 7322950 | 2020-09-17 13:39:19 +0100 | [diff] [blame] | 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 | #ifndef _TRANSREG_H |
| 13 | #define _TRANSREG_H |
| 14 | |
| 15 | #include "unicode/utypes.h" |
| 16 | |
| 17 | #if !UCONFIG_NO_TRANSLITERATION |
| 18 | |
| 19 | #include "unicode/uobject.h" |
| 20 | #include "unicode/translit.h" |
| 21 | #include "hash.h" |
| 22 | #include "uvector.h" |
| 23 | |
| 24 | U_NAMESPACE_BEGIN |
| 25 | |
| 26 | class TransliteratorEntry; |
| 27 | class TransliteratorSpec; |
| 28 | class UnicodeString; |
| 29 | |
| 30 | //------------------------------------------------------------------ |
| 31 | // TransliteratorAlias |
| 32 | //------------------------------------------------------------------ |
| 33 | |
| 34 | /** |
| 35 | * A TransliteratorAlias object is returned by get() if the given ID |
| 36 | * actually translates into something else. The caller then invokes |
| 37 | * the create() method on the alias to create the actual |
| 38 | * transliterator, and deletes the alias. |
| 39 | * |
| 40 | * Why all the shenanigans? To prevent circular calls between |
| 41 | * the registry code and the transliterator code that deadlocks. |
| 42 | */ |
| 43 | class TransliteratorAlias : public UMemory { |
| 44 | public: |
| 45 | /** |
| 46 | * Construct a simple alias (type == SIMPLE) |
| 47 | * @param aliasID the given id. |
| 48 | */ |
| 49 | TransliteratorAlias(const UnicodeString& aliasID, const UnicodeSet* compoundFilter); |
| 50 | |
| 51 | /** |
| 52 | * Construct a compound RBT alias (type == COMPOUND) |
| 53 | */ |
| 54 | TransliteratorAlias(const UnicodeString& ID, const UnicodeString& idBlocks, |
| 55 | UVector* adoptedTransliterators, |
| 56 | const UnicodeSet* compoundFilter); |
| 57 | |
| 58 | /** |
| 59 | * Construct a rules alias (type = RULES) |
| 60 | */ |
| 61 | TransliteratorAlias(const UnicodeString& theID, |
| 62 | const UnicodeString& rules, |
| 63 | UTransDirection dir); |
| 64 | |
| 65 | ~TransliteratorAlias(); |
| 66 | |
| 67 | /** |
| 68 | * The whole point of create() is that the caller must invoke |
| 69 | * it when the registry mutex is NOT held, to prevent deadlock. |
| 70 | * It may only be called once. |
| 71 | * |
Victor Chang | ce4bf3c | 2021-01-19 16:34:24 +0000 | [diff] [blame] | 72 | * Note: Only call create() if isRuleBased() returns false. |
Victor Chang | 7322950 | 2020-09-17 13:39:19 +0100 | [diff] [blame] | 73 | * |
| 74 | * This method must be called *outside* of the TransliteratorRegistry |
| 75 | * mutex. |
| 76 | */ |
| 77 | Transliterator* create(UParseError&, UErrorCode&); |
| 78 | |
| 79 | /** |
Victor Chang | ce4bf3c | 2021-01-19 16:34:24 +0000 | [diff] [blame] | 80 | * Return true if this alias is rule-based. If so, the caller |
Victor Chang | 7322950 | 2020-09-17 13:39:19 +0100 | [diff] [blame] | 81 | * must call parse() on it, then call TransliteratorRegistry::reget(). |
| 82 | */ |
| 83 | UBool isRuleBased() const; |
| 84 | |
| 85 | /** |
Victor Chang | ce4bf3c | 2021-01-19 16:34:24 +0000 | [diff] [blame] | 86 | * If isRuleBased() returns true, then the caller must call this |
Victor Chang | 7322950 | 2020-09-17 13:39:19 +0100 | [diff] [blame] | 87 | * method, followed by TransliteratorRegistry::reget(). The latter |
| 88 | * method must be called inside the TransliteratorRegistry mutex. |
| 89 | * |
Victor Chang | ce4bf3c | 2021-01-19 16:34:24 +0000 | [diff] [blame] | 90 | * Note: Only call parse() if isRuleBased() returns true. |
Victor Chang | 7322950 | 2020-09-17 13:39:19 +0100 | [diff] [blame] | 91 | * |
| 92 | * This method must be called *outside* of the TransliteratorRegistry |
| 93 | * mutex, because it can instantiate Transliterators embedded in |
| 94 | * the rules via the "&Latin-Arabic()" syntax. |
| 95 | */ |
| 96 | void parse(TransliteratorParser& parser, |
| 97 | UParseError& pe, UErrorCode& ec) const; |
| 98 | |
| 99 | private: |
| 100 | // We actually come in three flavors: |
| 101 | // 1. Simple alias |
| 102 | // Here aliasID is the alias string. Everything else is |
| 103 | // null, zero, empty. |
| 104 | // 2. CompoundRBT |
| 105 | // Here ID is the ID, aliasID is the idBlock, trans is the |
| 106 | // contained RBT, and idSplitPoint is the offet in aliasID |
| 107 | // where the contained RBT goes. compoundFilter is the |
| 108 | // compound filter, and it is _not_ owned. |
| 109 | // 3. Rules |
| 110 | // Here ID is the ID, aliasID is the rules string. |
| 111 | // idSplitPoint is the UTransDirection. |
| 112 | UnicodeString ID; |
| 113 | UnicodeString aliasesOrRules; |
| 114 | UVector* transes; // owned |
| 115 | const UnicodeSet* compoundFilter; // alias |
| 116 | UTransDirection direction; |
| 117 | enum { SIMPLE, COMPOUND, RULES } type; |
| 118 | |
| 119 | TransliteratorAlias(const TransliteratorAlias &other); // forbid copying of this class |
| 120 | TransliteratorAlias &operator=(const TransliteratorAlias &other); // forbid copying of this class |
| 121 | }; |
| 122 | |
| 123 | |
| 124 | /** |
| 125 | * A registry of system transliterators. This is the data structure |
| 126 | * that implements the mapping between transliterator IDs and the data |
| 127 | * or function pointers used to create the corresponding |
| 128 | * transliterators. There is one instance of the registry that is |
| 129 | * created statically. |
| 130 | * |
| 131 | * The registry consists of a dynamic component -- a hashtable -- and |
| 132 | * a static component -- locale resource bundles. The dynamic store |
| 133 | * is semantically overlaid on the static store, so the static mapping |
| 134 | * can be dynamically overridden. |
| 135 | * |
| 136 | * This is an internal class that is only used by Transliterator. |
| 137 | * Transliterator maintains one static instance of this class and |
| 138 | * delegates all registry-related operations to it. |
| 139 | * |
| 140 | * @author Alan Liu |
| 141 | */ |
| 142 | class TransliteratorRegistry : public UMemory { |
| 143 | |
| 144 | public: |
| 145 | |
| 146 | /** |
| 147 | * Contructor |
| 148 | * @param status Output param set to success/failure code. |
| 149 | */ |
| 150 | TransliteratorRegistry(UErrorCode& status); |
| 151 | |
| 152 | /** |
| 153 | * Nonvirtual destructor -- this class is not subclassable. |
| 154 | */ |
| 155 | ~TransliteratorRegistry(); |
| 156 | |
| 157 | //------------------------------------------------------------------ |
| 158 | // Basic public API |
| 159 | //------------------------------------------------------------------ |
| 160 | |
| 161 | /** |
| 162 | * Given a simple ID (forward direction, no inline filter, not |
| 163 | * compound) attempt to instantiate it from the registry. Return |
| 164 | * 0 on failure. |
| 165 | * |
| 166 | * Return a non-NULL aliasReturn value if the ID points to an alias. |
| 167 | * We cannot instantiate it ourselves because the alias may contain |
| 168 | * filters or compounds, which we do not understand. Caller should |
| 169 | * make aliasReturn NULL before calling. |
| 170 | * @param ID the given ID |
| 171 | * @param aliasReturn output param to receive TransliteratorAlias; |
| 172 | * should be NULL on entry |
| 173 | * @param parseError Struct to recieve information on position |
| 174 | * of error if an error is encountered |
| 175 | * @param status Output param set to success/failure code. |
| 176 | */ |
| 177 | Transliterator* get(const UnicodeString& ID, |
| 178 | TransliteratorAlias*& aliasReturn, |
| 179 | UErrorCode& status); |
| 180 | |
| 181 | /** |
| 182 | * The caller must call this after calling get(), if [a] calling get() |
| 183 | * returns an alias, and [b] the alias is rule based. In that |
| 184 | * situation the caller must call alias->parse() to do the parsing |
| 185 | * OUTSIDE THE REGISTRY MUTEX, then call this method to retry |
| 186 | * instantiating the transliterator. |
| 187 | * |
| 188 | * Note: Another alias might be returned by this method. |
| 189 | * |
| 190 | * This method (like all public methods of this class) must be called |
| 191 | * from within the TransliteratorRegistry mutex. |
| 192 | * |
| 193 | * @param aliasReturn output param to receive TransliteratorAlias; |
| 194 | * should be NULL on entry |
| 195 | */ |
| 196 | Transliterator* reget(const UnicodeString& ID, |
| 197 | TransliteratorParser& parser, |
| 198 | TransliteratorAlias*& aliasReturn, |
| 199 | UErrorCode& status); |
| 200 | |
| 201 | /** |
| 202 | * Register a prototype (adopted). This adds an entry to the |
| 203 | * dynamic store, or replaces an existing entry. Any entry in the |
| 204 | * underlying static locale resource store is masked. |
| 205 | */ |
| 206 | void put(Transliterator* adoptedProto, |
| 207 | UBool visible, |
| 208 | UErrorCode& ec); |
| 209 | |
| 210 | /** |
| 211 | * Register an ID and a factory function pointer. This adds an |
| 212 | * entry to the dynamic store, or replaces an existing entry. Any |
| 213 | * entry in the underlying static locale resource store is masked. |
| 214 | */ |
| 215 | void put(const UnicodeString& ID, |
| 216 | Transliterator::Factory factory, |
| 217 | Transliterator::Token context, |
| 218 | UBool visible, |
| 219 | UErrorCode& ec); |
| 220 | |
| 221 | /** |
| 222 | * Register an ID and a resource name. This adds an entry to the |
| 223 | * dynamic store, or replaces an existing entry. Any entry in the |
| 224 | * underlying static locale resource store is masked. |
| 225 | */ |
| 226 | void put(const UnicodeString& ID, |
| 227 | const UnicodeString& resourceName, |
| 228 | UTransDirection dir, |
| 229 | UBool readonlyResourceAlias, |
| 230 | UBool visible, |
| 231 | UErrorCode& ec); |
| 232 | |
| 233 | /** |
| 234 | * Register an ID and an alias ID. This adds an entry to the |
| 235 | * dynamic store, or replaces an existing entry. Any entry in the |
| 236 | * underlying static locale resource store is masked. |
| 237 | */ |
| 238 | void put(const UnicodeString& ID, |
| 239 | const UnicodeString& alias, |
| 240 | UBool readonlyAliasAlias, |
| 241 | UBool visible, |
| 242 | UErrorCode& ec); |
| 243 | |
| 244 | /** |
| 245 | * Unregister an ID. This removes an entry from the dynamic store |
| 246 | * if there is one. The static locale resource store is |
| 247 | * unaffected. |
| 248 | * @param ID the given ID. |
| 249 | */ |
| 250 | void remove(const UnicodeString& ID); |
| 251 | |
| 252 | //------------------------------------------------------------------ |
| 253 | // Public ID and spec management |
| 254 | //------------------------------------------------------------------ |
| 255 | |
| 256 | /** |
| 257 | * Return a StringEnumeration over the IDs currently registered |
| 258 | * with the system. |
| 259 | * @internal |
| 260 | */ |
| 261 | StringEnumeration* getAvailableIDs() const; |
| 262 | |
| 263 | /** |
| 264 | * == OBSOLETE - remove in ICU 3.4 == |
| 265 | * Return the number of IDs currently registered with the system. |
| 266 | * To retrieve the actual IDs, call getAvailableID(i) with |
| 267 | * i from 0 to countAvailableIDs() - 1. |
| 268 | * @return the number of IDs currently registered with the system. |
| 269 | * @internal |
| 270 | */ |
| 271 | int32_t countAvailableIDs(void) const; |
| 272 | |
| 273 | /** |
| 274 | * == OBSOLETE - remove in ICU 3.4 == |
| 275 | * Return the index-th available ID. index must be between 0 |
| 276 | * and countAvailableIDs() - 1, inclusive. If index is out of |
| 277 | * range, the result of getAvailableID(0) is returned. |
| 278 | * @param index the given index. |
| 279 | * @return the index-th available ID. index must be between 0 |
| 280 | * and countAvailableIDs() - 1, inclusive. If index is out of |
| 281 | * range, the result of getAvailableID(0) is returned. |
| 282 | * @internal |
| 283 | */ |
| 284 | const UnicodeString& getAvailableID(int32_t index) const; |
| 285 | |
| 286 | /** |
| 287 | * Return the number of registered source specifiers. |
| 288 | * @return the number of registered source specifiers. |
| 289 | */ |
| 290 | int32_t countAvailableSources(void) const; |
| 291 | |
| 292 | /** |
| 293 | * Return a registered source specifier. |
| 294 | * @param index which specifier to return, from 0 to n-1, where |
| 295 | * n = countAvailableSources() |
| 296 | * @param result fill-in paramter to receive the source specifier. |
| 297 | * If index is out of range, result will be empty. |
| 298 | * @return reference to result |
| 299 | */ |
| 300 | UnicodeString& getAvailableSource(int32_t index, |
| 301 | UnicodeString& result) const; |
| 302 | |
| 303 | /** |
| 304 | * Return the number of registered target specifiers for a given |
| 305 | * source specifier. |
| 306 | * @param source the given source specifier. |
| 307 | * @return the number of registered target specifiers for a given |
| 308 | * source specifier. |
| 309 | */ |
| 310 | int32_t countAvailableTargets(const UnicodeString& source) const; |
| 311 | |
| 312 | /** |
| 313 | * Return a registered target specifier for a given source. |
| 314 | * @param index which specifier to return, from 0 to n-1, where |
| 315 | * n = countAvailableTargets(source) |
| 316 | * @param source the source specifier |
| 317 | * @param result fill-in paramter to receive the target specifier. |
| 318 | * If source is invalid or if index is out of range, result will |
| 319 | * be empty. |
| 320 | * @return reference to result |
| 321 | */ |
| 322 | UnicodeString& getAvailableTarget(int32_t index, |
| 323 | const UnicodeString& source, |
| 324 | UnicodeString& result) const; |
| 325 | |
| 326 | /** |
| 327 | * Return the number of registered variant specifiers for a given |
| 328 | * source-target pair. There is always at least one variant: If |
| 329 | * just source-target is registered, then the single variant |
| 330 | * NO_VARIANT is returned. If source-target/variant is registered |
| 331 | * then that variant is returned. |
| 332 | * @param source the source specifiers |
| 333 | * @param target the target specifiers |
| 334 | * @return the number of registered variant specifiers for a given |
| 335 | * source-target pair. |
| 336 | */ |
| 337 | int32_t countAvailableVariants(const UnicodeString& source, |
| 338 | const UnicodeString& target) const; |
| 339 | |
| 340 | /** |
| 341 | * Return a registered variant specifier for a given source-target |
| 342 | * pair. If NO_VARIANT is one of the variants, then it will be |
| 343 | * at index 0. |
| 344 | * @param index which specifier to return, from 0 to n-1, where |
| 345 | * n = countAvailableVariants(source, target) |
| 346 | * @param source the source specifier |
| 347 | * @param target the target specifier |
| 348 | * @param result fill-in paramter to receive the variant |
| 349 | * specifier. If source is invalid or if target is invalid or if |
| 350 | * index is out of range, result will be empty. |
| 351 | * @return reference to result |
| 352 | */ |
| 353 | UnicodeString& getAvailableVariant(int32_t index, |
| 354 | const UnicodeString& source, |
| 355 | const UnicodeString& target, |
| 356 | UnicodeString& result) const; |
| 357 | |
| 358 | private: |
| 359 | |
| 360 | //---------------------------------------------------------------- |
| 361 | // Private implementation |
| 362 | //---------------------------------------------------------------- |
| 363 | |
| 364 | TransliteratorEntry* find(const UnicodeString& ID); |
| 365 | |
| 366 | TransliteratorEntry* find(UnicodeString& source, |
| 367 | UnicodeString& target, |
| 368 | UnicodeString& variant); |
| 369 | |
| 370 | TransliteratorEntry* findInDynamicStore(const TransliteratorSpec& src, |
| 371 | const TransliteratorSpec& trg, |
| 372 | const UnicodeString& variant) const; |
| 373 | |
| 374 | TransliteratorEntry* findInStaticStore(const TransliteratorSpec& src, |
| 375 | const TransliteratorSpec& trg, |
| 376 | const UnicodeString& variant); |
| 377 | |
| 378 | static TransliteratorEntry* findInBundle(const TransliteratorSpec& specToOpen, |
| 379 | const TransliteratorSpec& specToFind, |
| 380 | const UnicodeString& variant, |
| 381 | UTransDirection direction); |
| 382 | |
| 383 | void registerEntry(const UnicodeString& source, |
| 384 | const UnicodeString& target, |
| 385 | const UnicodeString& variant, |
| 386 | TransliteratorEntry* adopted, |
| 387 | UBool visible); |
| 388 | |
| 389 | void registerEntry(const UnicodeString& ID, |
| 390 | TransliteratorEntry* adopted, |
| 391 | UBool visible); |
| 392 | |
| 393 | void registerEntry(const UnicodeString& ID, |
| 394 | const UnicodeString& source, |
| 395 | const UnicodeString& target, |
| 396 | const UnicodeString& variant, |
| 397 | TransliteratorEntry* adopted, |
| 398 | UBool visible); |
| 399 | |
| 400 | void registerSTV(const UnicodeString& source, |
| 401 | const UnicodeString& target, |
| 402 | const UnicodeString& variant); |
| 403 | |
| 404 | void removeSTV(const UnicodeString& source, |
| 405 | const UnicodeString& target, |
| 406 | const UnicodeString& variant); |
| 407 | |
| 408 | Transliterator* instantiateEntry(const UnicodeString& ID, |
| 409 | TransliteratorEntry *entry, |
| 410 | TransliteratorAlias*& aliasReturn, |
| 411 | UErrorCode& status); |
| 412 | |
| 413 | /** |
| 414 | * A StringEnumeration over the registered IDs in this object. |
| 415 | */ |
| 416 | class Enumeration : public StringEnumeration { |
| 417 | public: |
| 418 | Enumeration(const TransliteratorRegistry& reg); |
| 419 | virtual ~Enumeration(); |
| 420 | virtual int32_t count(UErrorCode& status) const; |
| 421 | virtual const UnicodeString* snext(UErrorCode& status); |
| 422 | virtual void reset(UErrorCode& status); |
| 423 | static UClassID U_EXPORT2 getStaticClassID(); |
| 424 | virtual UClassID getDynamicClassID() const; |
| 425 | private: |
| 426 | int32_t index; |
| 427 | const TransliteratorRegistry& reg; |
| 428 | }; |
| 429 | friend class Enumeration; |
| 430 | |
| 431 | private: |
| 432 | |
| 433 | /** |
| 434 | * Dynamic registry mapping full IDs to Entry objects. This |
| 435 | * contains both public and internal entities. The visibility is |
| 436 | * controlled by whether an entry is listed in availableIDs and |
| 437 | * specDAG or not. |
| 438 | */ |
| 439 | Hashtable registry; |
| 440 | |
| 441 | /** |
| 442 | * DAG of visible IDs by spec. Hashtable: source => (Hashtable: |
| 443 | * target => variant bitmask) |
| 444 | */ |
| 445 | Hashtable specDAG; |
| 446 | |
| 447 | /** |
| 448 | * Vector of all variant names |
| 449 | */ |
| 450 | UVector variantList; |
| 451 | |
| 452 | /** |
| 453 | * Vector of public full IDs. |
| 454 | */ |
| 455 | UVector availableIDs; |
| 456 | |
| 457 | TransliteratorRegistry(const TransliteratorRegistry &other); // forbid copying of this class |
| 458 | TransliteratorRegistry &operator=(const TransliteratorRegistry &other); // forbid copying of this class |
| 459 | }; |
| 460 | |
| 461 | U_NAMESPACE_END |
| 462 | |
| 463 | U_CFUNC UBool utrans_transliterator_cleanup(void); |
| 464 | |
| 465 | #endif /* #if !UCONFIG_NO_TRANSLITERATION */ |
| 466 | |
| 467 | #endif |
| 468 | //eof |