1 /* 2 ****************************************************************************** 3 * Copyright (C) 1999-2015, International Business Machines 4 * Corporation and others. All Rights Reserved. 5 ****************************************************************************** 6 * file name: uresdata.h 7 * encoding: US-ASCII 8 * tab size: 8 (not used) 9 * indentation:4 10 * 11 * created on: 1999dec08 12 * created by: Markus W. Scherer 13 * 06/24/02 weiv Added support for resource sharing 14 */ 15 16 #ifndef __RESDATA_H__ 17 #define __RESDATA_H__ 18 19 #include "unicode/utypes.h" 20 #include "unicode/udata.h" 21 #include "unicode/ures.h" 22 #include "putilimp.h" 23 #include "udataswp.h" 24 25 /** 26 * Numeric constants for internal-only types of resource items. 27 * These must use different numeric values than UResType constants 28 * because they are used together. 29 * Internal types are never returned by ures_getType(). 30 */ 31 typedef enum { 32 /** Include a negative value so that the compiler uses the same int type as for UResType. */ 33 URES_INTERNAL_NONE=-1, 34 35 /** Resource type constant for tables with 32-bit count, key offsets and values. */ 36 URES_TABLE32=4, 37 38 /** 39 * Resource type constant for tables with 16-bit count, key offsets and values. 40 * All values are URES_STRING_V2 strings. 41 */ 42 URES_TABLE16=5, 43 44 /** Resource type constant for 16-bit Unicode strings in formatVersion 2. */ 45 URES_STRING_V2=6, 46 47 /** 48 * Resource type constant for arrays with 16-bit count and values. 49 * All values are URES_STRING_V2 strings. 50 */ 51 URES_ARRAY16=9 52 53 /* Resource type 15 is not defined but effectively used by RES_BOGUS=0xffffffff. */ 54 } UResInternalType; 55 56 /* 57 * A Resource is a 32-bit value that has 2 bit fields: 58 * 31..28 4-bit type, see enum below 59 * 27..0 28-bit four-byte-offset or value according to the type 60 */ 61 typedef uint32_t Resource; 62 63 #define RES_BOGUS 0xffffffff 64 #define RES_MAX_OFFSET 0x0fffffff 65 66 #define RES_GET_TYPE(res) ((int32_t)((res)>>28UL)) 67 #define RES_GET_OFFSET(res) ((res)&0x0fffffff) 68 #define RES_GET_POINTER(pRoot, res) ((pRoot)+RES_GET_OFFSET(res)) 69 70 /* get signed and unsigned integer values directly from the Resource handle */ 71 #if U_SIGNED_RIGHT_SHIFT_IS_ARITHMETIC 72 # define RES_GET_INT(res) (((int32_t)((res)<<4L))>>4L) 73 #else 74 # define RES_GET_INT(res) (int32_t)(((res)&0x08000000) ? (res)|0xf0000000 : (res)&0x07ffffff) 75 #endif 76 77 #define RES_GET_UINT(res) ((res)&0x0fffffff) 78 79 #define URES_IS_ARRAY(type) ((int32_t)(type)==URES_ARRAY || (int32_t)(type)==URES_ARRAY16) 80 #define URES_IS_TABLE(type) ((int32_t)(type)==URES_TABLE || (int32_t)(type)==URES_TABLE16 || (int32_t)(type)==URES_TABLE32) 81 #define URES_IS_CONTAINER(type) (URES_IS_TABLE(type) || URES_IS_ARRAY(type)) 82 83 #define URES_MAKE_RESOURCE(type, offset) (((Resource)(type)<<28)|(Resource)(offset)) 84 #define URES_MAKE_EMPTY_RESOURCE(type) ((Resource)(type)<<28) 85 86 /* indexes[] value names; indexes are generally 32-bit (Resource) indexes */ 87 enum { 88 /** 89 * [0] contains the length of indexes[] 90 * which is at most URES_INDEX_TOP of the latest format version 91 * 92 * formatVersion==1: all bits contain the length of indexes[] 93 * but the length is much less than 0xff; 94 * formatVersion>1: 95 * only bits 7..0 contain the length of indexes[], 96 * bits 31..8 are reserved and set to 0 97 * formatVersion>=3: 98 * bits 31..8 poolStringIndexLimit bits 23..0 99 */ 100 URES_INDEX_LENGTH, 101 /** 102 * [1] contains the top of the key strings, 103 * same as the bottom of resources or UTF-16 strings, rounded up 104 */ 105 URES_INDEX_KEYS_TOP, 106 /** [2] contains the top of all resources */ 107 URES_INDEX_RESOURCES_TOP, 108 /** 109 * [3] contains the top of the bundle, 110 * in case it were ever different from [2] 111 */ 112 URES_INDEX_BUNDLE_TOP, 113 /** [4] max. length of any table */ 114 URES_INDEX_MAX_TABLE_LENGTH, 115 /** 116 * [5] attributes bit set, see URES_ATT_* (new in formatVersion 1.2) 117 * 118 * formatVersion>=3: 119 * bits 31..16 poolStringIndex16Limit 120 * bits 15..12 poolStringIndexLimit bits 27..24 121 */ 122 URES_INDEX_ATTRIBUTES, 123 /** 124 * [6] top of the 16-bit units (UTF-16 string v2 UChars, URES_TABLE16, URES_ARRAY16), 125 * rounded up (new in formatVersion 2.0, ICU 4.4) 126 */ 127 URES_INDEX_16BIT_TOP, 128 /** [7] checksum of the pool bundle (new in formatVersion 2.0, ICU 4.4) */ 129 URES_INDEX_POOL_CHECKSUM, 130 URES_INDEX_TOP 131 }; 132 133 /* 134 * Nofallback attribute, attribute bit 0 in indexes[URES_INDEX_ATTRIBUTES]. 135 * New in formatVersion 1.2 (ICU 3.6). 136 * 137 * If set, then this resource bundle is a standalone bundle. 138 * If not set, then the bundle participates in locale fallback, eventually 139 * all the way to the root bundle. 140 * If indexes[] is missing or too short, then the attribute cannot be determined 141 * reliably. Dependency checking should ignore such bundles, and loading should 142 * use fallbacks. 143 */ 144 #define URES_ATT_NO_FALLBACK 1 145 146 /* 147 * Attributes for bundles that are, or use, a pool bundle. 148 * A pool bundle provides key strings that are shared among several other bundles 149 * to reduce their total size. 150 * New in formatVersion 2 (ICU 4.4). 151 */ 152 #define URES_ATT_IS_POOL_BUNDLE 2 153 #define URES_ATT_USES_POOL_BUNDLE 4 154 155 /* 156 * File format for .res resource bundle files 157 * 158 * ICU 56: New in formatVersion 3 compared with 2: ------------- 159 * 160 * Resource bundles can optionally use shared string-v2 values 161 * stored in the pool bundle. 162 * If so, then the indexes[] contain two new values 163 * in previously-unused bits of existing indexes[] slots: 164 * - poolStringIndexLimit: 165 * String-v2 offsets (in 32-bit Resource words) below this limit 166 * point to pool bundle string-v2 values. 167 * - poolStringIndex16Limit: 168 * Resource16 string-v2 offsets below this limit 169 * point to pool bundle string-v2 values. 170 * Guarantee: poolStringIndex16Limit <= poolStringIndexLimit 171 * 172 * The local bundle's poolStringIndexLimit is greater than 173 * any pool bundle string index used in the local bundle. 174 * The poolStringIndexLimit should not be greater than 175 * the maximum possible pool bundle string index. 176 * 177 * The maximum possible pool bundle string index is the index to the last non-NUL 178 * pool string character, due to suffix sharing. 179 * 180 * In the pool bundle, there is no structure that lists the strings. 181 * (The root resource is an empty Table.) 182 * If the strings need to be enumerated (as genrb --usePoolBundle does), 183 * then iterate through the pool bundle's 16-bit-units array from the beginning. 184 * Stop at the end of the array, or when an explicit or implicit string length 185 * would lead beyond the end of the array, 186 * or when an apparent string is not NUL-terminated. 187 * (Future genrb version might terminate the strings with 188 * what looks like a large explicit string length.) 189 * 190 * ICU 4.4: New in formatVersion 2 compared with 1.3: ------------- 191 * 192 * Three new resource types -- String-v2, Table16 and Array16 -- have their 193 * values stored in a new array of 16-bit units between the table key strings 194 * and the start of the other resources. 195 * 196 * genrb eliminates duplicates among Unicode string-v2 values. 197 * Multiple Unicode strings may use the same offset and string data, 198 * or a short string may point to the suffix of a longer string. ("Suffix sharing") 199 * For example, one string "abc" may be reused for another string "bc" by pointing 200 * to the second character. (Short strings-v2 are NUL-terminated 201 * and not preceded by an explicit length value.) 202 * 203 * It is allowed for all resource types to share values. 204 * The swapper code (ures_swap()) has been modified so that it swaps each item 205 * exactly once. 206 * 207 * A resource bundle may use a special pool bundle. Some or all of the table key strings 208 * of the using-bundle are omitted, and the key string offsets for such key strings refer 209 * to offsets in the pool bundle. 210 * The using-bundle's and the pool-bundle's indexes[URES_INDEX_POOL_CHECKSUM] values 211 * must match. 212 * Two bits in indexes[URES_INDEX_ATTRIBUTES] indicate whether a resource bundle 213 * is or uses a pool bundle. 214 * 215 * Table key strings must be compared in ASCII order, even if they are not 216 * stored in ASCII. 217 * 218 * New in formatVersion 1.3 compared with 1.2: ------------- 219 * 220 * genrb eliminates duplicates among key strings. 221 * Multiple table items may share one key string, or one item may point 222 * to the suffix of another's key string. ("Suffix sharing") 223 * For example, one key "abc" may be reused for another key "bc" by pointing 224 * to the second character. (Key strings are NUL-terminated.) 225 * 226 * ------------- 227 * 228 * An ICU4C resource bundle file (.res) is a binary, memory-mappable file 229 * with nested, hierarchical data structures. 230 * It physically contains the following: 231 * 232 * Resource root; -- 32-bit Resource item, root item for this bundle's tree; 233 * currently, the root item must be a table or table32 resource item 234 * int32_t indexes[indexes[0]]; -- array of indexes for friendly 235 * reading and swapping; see URES_INDEX_* above 236 * new in formatVersion 1.1 (ICU 2.8) 237 * char keys[]; -- characters for key strings 238 * (formatVersion 1.0: up to 65k of characters; 1.1: <2G) 239 * (minus the space for root and indexes[]), 240 * which consist of invariant characters (ASCII/EBCDIC) and are NUL-terminated; 241 * padded to multiple of 4 bytes for 4-alignment of the following data 242 * uint16_t 16BitUnits[]; -- resources that are stored entirely as sequences of 16-bit units 243 * (new in formatVersion 2/ICU 4.4) 244 * data is indexed by the offset values in 16-bit resource types, 245 * with offset 0 pointing to the beginning of this array; 246 * there is a 0 at offset 0, for empty resources; 247 * padded to multiple of 4 bytes for 4-alignment of the following data 248 * data; -- data directly and indirectly indexed by the root item; 249 * the structure is determined by walking the tree 250 * 251 * Each resource bundle item has a 32-bit Resource handle (see typedef above) 252 * which contains the item type number in its upper 4 bits (31..28) and either 253 * an offset or a direct value in its lower 28 bits (27..0). 254 * The order of items is undefined and only determined by walking the tree. 255 * Leaves of the tree may be stored first or last or anywhere in between, 256 * and it is in theory possible to have unreferenced holes in the file. 257 * 258 * 16-bit-unit values: 259 * Starting with formatVersion 2/ICU 4.4, some resources are stored in a special 260 * array of 16-bit units. Each resource value is a sequence of 16-bit units, 261 * with no per-resource padding to a 4-byte boundary. 262 * 16-bit container types (Table16 and Array16) contain Resource16 values 263 * which are offsets to String-v2 resources in the same 16-bit-units array. 264 * 265 * Direct values: 266 * - Empty Unicode strings have an offset value of 0 in the Resource handle itself. 267 * - Starting with formatVersion 2/ICU 4.4, an offset value of 0 for 268 * _any_ resource type indicates an empty value. 269 * - Integer values are 28-bit values stored in the Resource handle itself; 270 * the interpretation of unsigned vs. signed integers is up to the application. 271 * 272 * All other types and values use 28-bit offsets to point to the item's data. 273 * The offset is an index to the first 32-bit word of the value, relative to the 274 * start of the resource data (i.e., the root item handle is at offset 0). 275 * To get byte offsets, the offset is multiplied by 4 (or shifted left by 2 bits). 276 * All resource item values are 4-aligned. 277 * 278 * New in formatVersion 2/ICU 4.4: Some types use offsets into the 16-bit-units array, 279 * indexing 16-bit units in that array. 280 * 281 * The structures (memory layouts) for the values for each item type are listed 282 * in the table below. 283 * 284 * Nested, hierarchical structures: ------------- 285 * 286 * Table items contain key-value pairs where the keys are offsets to char * key strings. 287 * The values of these pairs are either Resource handles or 288 * offsets into the 16-bit-units array, depending on the table type. 289 * 290 * Array items are simple vectors of Resource handles, 291 * or of offsets into the 16-bit-units array, depending on the array type. 292 * 293 * Table key string offsets: ------- 294 * 295 * Key string offsets are relative to the start of the resource data (of the root handle), 296 * i.e., the first string has an offset of 4+sizeof(indexes). 297 * (After the 4-byte root handle and after the indexes array.) 298 * 299 * If the resource bundle uses a pool bundle, then some key strings are stored 300 * in the pool bundle rather than in the local bundle itself. 301 * - In a Table or Table16, the 16-bit key string offset is local if it is 302 * less than indexes[URES_INDEX_KEYS_TOP]<<2. 303 * Otherwise, subtract indexes[URES_INDEX_KEYS_TOP]<<2 to get the offset into 304 * the pool bundle key strings. 305 * - In a Table32, the 32-bit key string offset is local if it is non-negative. 306 * Otherwise, reset bit 31 to get the pool key string offset. 307 * 308 * Unlike the local offset, the pool key offset is relative to 309 * the start of the key strings, not to the start of the bundle. 310 * 311 * An alias item is special (and new in ICU 2.4): -------------- 312 * 313 * Its memory layout is just like for a UnicodeString, but at runtime it resolves to 314 * another resource bundle's item according to the path in the string. 315 * This is used to share items across bundles that are in different lookup/fallback 316 * chains (e.g., large collation data among zh_TW and zh_HK). 317 * This saves space (for large items) and maintenance effort (less duplication of data). 318 * 319 * -------------------------------------------------------------------------- 320 * 321 * Resource types: 322 * 323 * Most resources have their values stored at four-byte offsets from the start 324 * of the resource data. These values are at least 4-aligned. 325 * Some resource values are stored directly in the offset field of the Resource itself. 326 * See UResType in unicode/ures.h for enumeration constants for Resource types. 327 * 328 * Some resources have their values stored as sequences of 16-bit units, 329 * at 2-byte offsets from the start of a contiguous 16-bit-unit array between 330 * the table key strings and the other resources. (new in formatVersion 2/ICU 4.4) 331 * At offset 0 of that array is a 16-bit zero value for empty 16-bit resources. 332 * 333 * Resource16 values in Table16 and Array16 are 16-bit offsets to String-v2 334 * resources, with the offsets relative to the start of the 16-bit-units array. 335 * Starting with formatVersion 3/ICU 56, if offset<poolStringIndex16Limit 336 * then use the pool bundle's 16-bit-units array, 337 * otherwise subtract that limit and use the local 16-bit-units array. 338 * 339 * Type Name Memory layout of values 340 * (in parentheses: scalar, non-offset values) 341 * 342 * 0 Unicode String: int32_t length, UChar[length], (UChar)0, (padding) 343 * or (empty string ("") if offset==0) 344 * 1 Binary: int32_t length, uint8_t[length], (padding) 345 * - the start of the bytes is 16-aligned - 346 * 2 Table: uint16_t count, uint16_t keyStringOffsets[count], (uint16_t padding), Resource[count] 347 * 3 Alias: (physically same value layout as string, new in ICU 2.4) 348 * 4 Table32: int32_t count, int32_t keyStringOffsets[count], Resource[count] 349 * (new in formatVersion 1.1/ICU 2.8) 350 * 5 Table16: uint16_t count, uint16_t keyStringOffsets[count], Resource16[count] 351 * (stored in the 16-bit-units array; new in formatVersion 2/ICU 4.4) 352 * 6 Unicode String-v2:UChar[length], (UChar)0; length determined by the first UChar: 353 * - if first is not a trail surrogate, then the length is implicit 354 * and u_strlen() needs to be called 355 * - if first<0xdfef then length=first&0x3ff (and skip first) 356 * - if first<0xdfff then length=((first-0xdfef)<<16) | second UChar 357 * - if first==0xdfff then length=((second UChar)<<16) | third UChar 358 * (stored in the 16-bit-units array; new in formatVersion 2/ICU 4.4) 359 * 360 * Starting with formatVersion 3/ICU 56, if offset<poolStringIndexLimit 361 * then use the pool bundle's 16-bit-units array, 362 * otherwise subtract that limit and use the local 16-bit-units array. 363 * (Note different limits for Resource16 vs. Resource.) 364 * 365 * 7 Integer: (28-bit offset is integer value) 366 * 8 Array: int32_t count, Resource[count] 367 * 9 Array16: uint16_t count, Resource16[count] 368 * (stored in the 16-bit-units array; new in formatVersion 2/ICU 4.4) 369 * 14 Integer Vector: int32_t length, int32_t[length] 370 * 15 Reserved: This value denotes special purpose resources and is for internal use. 371 * 372 * Note that there are 3 types with data vector values: 373 * - Vectors of 8-bit bytes stored as type Binary. 374 * - Vectors of 16-bit words stored as type Unicode String or Unicode String-v2 375 * (no value restrictions, all values 0..ffff allowed!). 376 * - Vectors of 32-bit words stored as type Integer Vector. 377 */ 378 379 /* 380 * Structure for a single, memory-mapped ResourceBundle. 381 */ 382 typedef struct { 383 UDataMemory *data; 384 const int32_t *pRoot; 385 const uint16_t *p16BitUnits; 386 const char *poolBundleKeys; 387 Resource rootRes; 388 int32_t localKeyLimit; 389 const uint16_t *poolBundleStrings; 390 int32_t poolStringIndexLimit; 391 int32_t poolStringIndex16Limit; 392 UBool noFallback; /* see URES_ATT_NO_FALLBACK */ 393 UBool isPoolBundle; 394 UBool usesPoolBundle; 395 UBool useNativeStrcmp; 396 } ResourceData; 397 398 /* 399 * Read a resource bundle from memory. 400 */ 401 U_INTERNAL void U_EXPORT2 402 res_read(ResourceData *pResData, 403 const UDataInfo *pInfo, const void *inBytes, int32_t length, 404 UErrorCode *errorCode); 405 406 /* 407 * Load a resource bundle file. 408 * The ResourceData structure must be allocated externally. 409 */ 410 U_CFUNC void 411 res_load(ResourceData *pResData, 412 const char *path, const char *name, UErrorCode *errorCode); 413 414 /* 415 * Release a resource bundle file. 416 * This does not release the ResourceData structure itself. 417 */ 418 U_CFUNC void 419 res_unload(ResourceData *pResData); 420 421 U_INTERNAL UResType U_EXPORT2 422 res_getPublicType(Resource res); 423 424 /* 425 * Return a pointer to a zero-terminated, const UChar* string 426 * and set its length in *pLength. 427 * Returns NULL if not found. 428 */ 429 U_INTERNAL const UChar * U_EXPORT2 430 res_getString(const ResourceData *pResData, Resource res, int32_t *pLength); 431 432 U_INTERNAL const UChar * U_EXPORT2 433 res_getAlias(const ResourceData *pResData, Resource res, int32_t *pLength); 434 435 U_INTERNAL const uint8_t * U_EXPORT2 436 res_getBinary(const ResourceData *pResData, Resource res, int32_t *pLength); 437 438 U_INTERNAL const int32_t * U_EXPORT2 439 res_getIntVector(const ResourceData *pResData, Resource res, int32_t *pLength); 440 441 U_INTERNAL Resource U_EXPORT2 442 res_getResource(const ResourceData *pResData, const char *key); 443 444 U_INTERNAL int32_t U_EXPORT2 445 res_countArrayItems(const ResourceData *pResData, Resource res); 446 447 U_INTERNAL Resource U_EXPORT2 448 res_getArrayItem(const ResourceData *pResData, Resource array, int32_t indexS); 449 450 U_INTERNAL Resource U_EXPORT2 451 res_getTableItemByIndex(const ResourceData *pResData, Resource table, int32_t indexS, const char ** key); 452 453 U_INTERNAL Resource U_EXPORT2 454 res_getTableItemByKey(const ResourceData *pResData, Resource table, int32_t *indexS, const char* * key); 455 456 /** 457 * Iterates over the path and stops when a scalar resource is found. 458 * Follows aliases. 459 * Modifies the contents of *path (replacing separators with NULs), 460 * and also moves *path forward while it finds items. 461 * 462 * @param path input: "CollationElements/Sequence" or "zoneStrings/3/2" etc.; 463 * output: points to the part that has not yet been processed 464 */ 465 U_CFUNC Resource res_findResource(const ResourceData *pResData, Resource r, 466 char** path, const char** key); 467 468 #ifdef __cplusplus 469 470 #include "resource.h" 471 472 U_NAMESPACE_BEGIN 473 474 class ResourceDataValue : public ResourceValue { 475 public: ResourceDataValue()476 ResourceDataValue() : pResData(NULL), res(URES_NONE) {} 477 virtual ~ResourceDataValue(); 478 setData(const ResourceData * data)479 void setData(const ResourceData *data) { pResData = data; } setResource(Resource r)480 void setResource(Resource r) { res = r; } 481 482 virtual UResType getType() const; 483 virtual const UChar *getString(int32_t &length, UErrorCode &errorCode) const; 484 virtual const UChar *getAliasString(int32_t &length, UErrorCode &errorCode) const; 485 virtual int32_t getInt(UErrorCode &errorCode) const; 486 virtual uint32_t getUInt(UErrorCode &errorCode) const; 487 virtual const int32_t *getIntVector(int32_t &length, UErrorCode &errorCode) const; 488 virtual const uint8_t *getBinary(int32_t &length, UErrorCode &errorCode) const; 489 490 const ResourceData *pResData; 491 492 private: 493 Resource res; 494 }; 495 496 U_NAMESPACE_END 497 498 /** 499 * @param value will be set during enumeration; input contents is ignored 500 * @param sink receives all table item key-value pairs 501 */ 502 U_CFUNC void 503 ures_getAllTableItems(const ResourceData *pResData, Resource table, 504 icu::ResourceDataValue &value, icu::ResourceTableSink &sink, 505 UErrorCode &errorCode); 506 507 /** 508 * @param value will be set during enumeration; input contents is ignored 509 * @param sink receives all array item values 510 */ 511 U_CFUNC void 512 ures_getAllArrayItems(const ResourceData *pResData, Resource array, 513 icu::ResourceDataValue &value, icu::ResourceArraySink &sink, 514 UErrorCode &errorCode); 515 516 #endif /* __cplusplus */ 517 518 /** 519 * Swap an ICU resource bundle. See udataswp.h. 520 * @internal 521 */ 522 U_CAPI int32_t U_EXPORT2 523 ures_swap(const UDataSwapper *ds, 524 const void *inData, int32_t length, void *outData, 525 UErrorCode *pErrorCode); 526 527 #endif 528