]>
Commit | Line | Data |
---|---|---|
b75a7d8f A |
1 | /* |
2 | ******************************************************************************* | |
3 | * | |
729e4ab9 | 4 | * Copyright (C) 2000-2010, International Business Machines |
b75a7d8f A |
5 | * Corporation and others. All Rights Reserved. |
6 | * | |
7 | ******************************************************************************* | |
8 | * | |
9 | * File reslist.c | |
10 | * | |
11 | * Modification History: | |
12 | * | |
13 | * Date Name Description | |
14 | * 02/21/00 weiv Creation. | |
15 | ******************************************************************************* | |
16 | */ | |
17 | ||
18 | #include <assert.h> | |
374ca955 | 19 | #include <stdio.h> |
b75a7d8f A |
20 | #include "reslist.h" |
21 | #include "unewdata.h" | |
22 | #include "unicode/ures.h" | |
374ca955 | 23 | #include "unicode/putil.h" |
b75a7d8f A |
24 | #include "errmsg.h" |
25 | ||
729e4ab9 A |
26 | #include "uarrsort.h" |
27 | #include "uinvchar.h" | |
28 | ||
29 | /* | |
30 | * Align binary data at a 16-byte offset from the start of the resource bundle, | |
31 | * to be safe for any data type it may contain. | |
32 | */ | |
b75a7d8f A |
33 | #define BIN_ALIGNMENT 16 |
34 | ||
35 | static UBool gIncludeCopyright = FALSE; | |
729e4ab9 A |
36 | static UBool gUsePoolBundle = FALSE; |
37 | static int32_t gFormatVersion = 2; | |
38 | ||
39 | static UChar gEmptyString = 0; | |
40 | ||
41 | /* How do we store string values? */ | |
42 | enum { | |
43 | STRINGS_UTF16_V1, /* formatVersion 1: int length + UChars + NUL + padding to 4 bytes */ | |
44 | STRINGS_UTF16_V2 /* formatVersion 2: optional length in 1..3 UChars + UChars + NUL */ | |
45 | }; | |
46 | ||
47 | enum { | |
48 | MAX_IMPLICIT_STRING_LENGTH = 40 /* do not store the length explicitly for such strings */ | |
49 | }; | |
b75a7d8f | 50 | |
46f4442e A |
51 | /* |
52 | * res_none() returns the address of kNoResource, | |
53 | * for use in non-error cases when no resource is to be added to the bundle. | |
54 | * (NULL is used in error cases.) | |
55 | */ | |
729e4ab9 | 56 | static const struct SResource kNoResource = { URES_NONE }; |
b75a7d8f | 57 | |
729e4ab9 | 58 | static UDataInfo dataInfo= { |
b75a7d8f A |
59 | sizeof(UDataInfo), |
60 | 0, | |
61 | ||
62 | U_IS_BIG_ENDIAN, | |
63 | U_CHARSET_FAMILY, | |
64 | sizeof(UChar), | |
65 | 0, | |
66 | ||
729e4ab9 A |
67 | {0x52, 0x65, 0x73, 0x42}, /* dataFormat="ResB" */ |
68 | {1, 3, 0, 0}, /* formatVersion */ | |
b75a7d8f A |
69 | {1, 4, 0, 0} /* dataVersion take a look at version inside parsed resb*/ |
70 | }; | |
71 | ||
729e4ab9 A |
72 | static const UVersionInfo gFormatVersions[3] = { /* indexed by a major-formatVersion integer */ |
73 | { 0, 0, 0, 0 }, | |
74 | { 1, 3, 0, 0 }, | |
75 | { 2, 0, 0, 0 } | |
76 | }; | |
77 | ||
b75a7d8f A |
78 | static uint8_t calcPadding(uint32_t size) { |
79 | /* returns space we need to pad */ | |
80 | return (uint8_t) ((size % sizeof(uint32_t)) ? (sizeof(uint32_t) - (size % sizeof(uint32_t))) : 0); | |
81 | ||
82 | } | |
83 | ||
84 | void setIncludeCopyright(UBool val){ | |
85 | gIncludeCopyright=val; | |
86 | } | |
87 | ||
88 | UBool getIncludeCopyright(void){ | |
89 | return gIncludeCopyright; | |
90 | } | |
91 | ||
729e4ab9 A |
92 | void setFormatVersion(int32_t formatVersion) { |
93 | gFormatVersion = formatVersion; | |
94 | } | |
b75a7d8f | 95 | |
729e4ab9 A |
96 | void setUsePoolBundle(UBool use) { |
97 | gUsePoolBundle = use; | |
b75a7d8f A |
98 | } |
99 | ||
729e4ab9 A |
100 | static void |
101 | bundle_compactStrings(struct SRBRoot *bundle, UErrorCode *status); | |
102 | ||
b75a7d8f | 103 | /* Writing Functions */ |
b75a7d8f | 104 | |
729e4ab9 A |
105 | /* |
106 | * type_write16() functions write resource values into f16BitUnits | |
107 | * and determine the resource item word, if possible. | |
108 | */ | |
109 | static void | |
110 | res_write16(struct SRBRoot *bundle, struct SResource *res, | |
111 | UErrorCode *status); | |
b75a7d8f | 112 | |
729e4ab9 A |
113 | /* |
114 | * type_preWrite() functions calculate ("preflight") and advance the *byteOffset | |
115 | * by the size of their data in the binary file and | |
116 | * determine the resource item word. | |
117 | * Most type_preWrite() functions may add any number of bytes, but res_preWrite() | |
118 | * will always pad it to a multiple of 4. | |
119 | * The resource item type may be a related subtype of the fType. | |
120 | * | |
121 | * The type_preWrite() and type_write() functions start and end at the same | |
122 | * byteOffset values. | |
123 | * Prewriting allows bundle_write() to determine the root resource item word, | |
124 | * before actually writing the bundle contents to the file, | |
125 | * which is necessary because the root item is stored at the beginning. | |
126 | */ | |
127 | static void | |
128 | res_preWrite(uint32_t *byteOffset, | |
129 | struct SRBRoot *bundle, struct SResource *res, | |
130 | UErrorCode *status); | |
b75a7d8f | 131 | |
729e4ab9 A |
132 | /* |
133 | * type_write() functions write their data to mem and update the byteOffset | |
134 | * in parallel. | |
135 | * (A kingdom for C++ and polymorphism...) | |
136 | */ | |
137 | static void | |
138 | res_write(UNewDataMemory *mem, uint32_t *byteOffset, | |
139 | struct SRBRoot *bundle, struct SResource *res, | |
140 | UErrorCode *status); | |
b75a7d8f | 141 | |
729e4ab9 A |
142 | static uint16_t * |
143 | reserve16BitUnits(struct SRBRoot *bundle, int32_t length, UErrorCode *status) { | |
b75a7d8f | 144 | if (U_FAILURE(*status)) { |
729e4ab9 A |
145 | return NULL; |
146 | } | |
147 | if ((bundle->f16BitUnitsLength + length) > bundle->f16BitUnitsCapacity) { | |
148 | uint16_t *newUnits; | |
149 | int32_t capacity = 2 * bundle->f16BitUnitsCapacity + length + 1024; | |
150 | capacity &= ~1; /* ensures padding fits if f16BitUnitsLength needs it */ | |
151 | newUnits = (uint16_t *)uprv_malloc(capacity * 2); | |
152 | if (newUnits == NULL) { | |
153 | *status = U_MEMORY_ALLOCATION_ERROR; | |
154 | return NULL; | |
155 | } | |
156 | if (bundle->f16BitUnitsLength > 0) { | |
157 | uprv_memcpy(newUnits, bundle->f16BitUnits, bundle->f16BitUnitsLength * 2); | |
158 | } else { | |
159 | newUnits[0] = 0; | |
160 | bundle->f16BitUnitsLength = 1; | |
161 | } | |
162 | uprv_free(bundle->f16BitUnits); | |
163 | bundle->f16BitUnits = newUnits; | |
164 | bundle->f16BitUnitsCapacity = capacity; | |
b75a7d8f | 165 | } |
729e4ab9 A |
166 | return bundle->f16BitUnits + bundle->f16BitUnitsLength; |
167 | } | |
b75a7d8f | 168 | |
729e4ab9 A |
169 | static int32_t |
170 | makeRes16(uint32_t resWord) { | |
171 | uint32_t type, offset; | |
172 | if (resWord == 0) { | |
173 | return 0; /* empty string */ | |
174 | } | |
175 | type = RES_GET_TYPE(resWord); | |
176 | offset = RES_GET_OFFSET(resWord); | |
177 | if (type == URES_STRING_V2 && offset <= 0xffff) { | |
178 | return (int32_t)offset; | |
179 | } | |
180 | return -1; | |
181 | } | |
b75a7d8f | 182 | |
729e4ab9 A |
183 | static int32_t |
184 | mapKey(struct SRBRoot *bundle, int32_t oldpos) { | |
185 | const KeyMapEntry *map = bundle->fKeyMap; | |
186 | int32_t i, start, limit; | |
187 | ||
188 | /* do a binary search for the old, pre-bundle_compactKeys() key offset */ | |
189 | start = bundle->fPoolBundleKeysCount; | |
190 | limit = start + bundle->fKeysCount; | |
191 | while (start < limit - 1) { | |
192 | i = (start + limit) / 2; | |
193 | if (oldpos < map[i].oldpos) { | |
194 | limit = i; | |
195 | } else { | |
196 | start = i; | |
b75a7d8f | 197 | } |
729e4ab9 A |
198 | } |
199 | assert(oldpos == map[start].oldpos); | |
200 | return map[start].newpos; | |
201 | } | |
b75a7d8f | 202 | |
729e4ab9 A |
203 | static uint16_t |
204 | makeKey16(struct SRBRoot *bundle, int32_t key) { | |
205 | if (key >= 0) { | |
206 | return (uint16_t)key; | |
207 | } else { | |
208 | return (uint16_t)(key + bundle->fLocalKeyLimit); /* offset in the pool bundle */ | |
209 | } | |
210 | } | |
b75a7d8f | 211 | |
729e4ab9 A |
212 | /* |
213 | * Only called for UTF-16 v1 strings and duplicate UTF-16 v2 strings. | |
214 | * For unique UTF-16 v2 strings, res_write16() sees fRes != RES_BOGUS | |
215 | * and exits early. | |
216 | */ | |
217 | static void | |
218 | string_write16(struct SRBRoot *bundle, struct SResource *res, UErrorCode *status) { | |
219 | struct SResource *same; | |
220 | if ((same = res->u.fString.fSame) != NULL) { | |
221 | /* This is a duplicate. */ | |
222 | if (same->fRes == RES_BOGUS) { | |
223 | /* The original has not been visited yet. */ | |
224 | string_write16(bundle, same, status); | |
225 | } | |
226 | res->fRes = same->fRes; | |
227 | res->fWritten = same->fWritten; | |
228 | } | |
229 | } | |
b75a7d8f | 230 | |
729e4ab9 A |
231 | static void |
232 | array_write16(struct SRBRoot *bundle, struct SResource *res, | |
233 | UErrorCode *status) { | |
234 | struct SResource *current; | |
235 | int32_t res16 = 0; | |
b75a7d8f | 236 | |
729e4ab9 A |
237 | if (U_FAILURE(*status)) { |
238 | return; | |
239 | } | |
240 | if (res->u.fArray.fCount == 0 && gFormatVersion > 1) { | |
241 | res->fRes = URES_MAKE_EMPTY_RESOURCE(URES_ARRAY); | |
242 | res->fWritten = TRUE; | |
243 | return; | |
244 | } | |
245 | for (current = res->u.fArray.fFirst; current != NULL; current = current->fNext) { | |
246 | res_write16(bundle, current, status); | |
247 | res16 |= makeRes16(current->fRes); | |
248 | } | |
249 | if (U_SUCCESS(*status) && res->u.fArray.fCount <= 0xffff && res16 >= 0 && gFormatVersion > 1) { | |
250 | uint16_t *p16 = reserve16BitUnits(bundle, 1 + res->u.fArray.fCount, status); | |
251 | if (U_SUCCESS(*status)) { | |
252 | res->fRes = URES_MAKE_RESOURCE(URES_ARRAY16, bundle->f16BitUnitsLength); | |
253 | *p16++ = (uint16_t)res->u.fArray.fCount; | |
254 | for (current = res->u.fArray.fFirst; current != NULL; current = current->fNext) { | |
255 | *p16++ = (uint16_t)makeRes16(current->fRes); | |
256 | } | |
257 | bundle->f16BitUnitsLength += 1 + res->u.fArray.fCount; | |
258 | res->fWritten = TRUE; | |
b75a7d8f | 259 | } |
729e4ab9 A |
260 | } |
261 | } | |
b75a7d8f | 262 | |
729e4ab9 A |
263 | static void |
264 | table_write16(struct SRBRoot *bundle, struct SResource *res, | |
265 | UErrorCode *status) { | |
266 | struct SResource *current; | |
267 | int32_t maxKey = 0, maxPoolKey = 0x80000000; | |
268 | int32_t res16 = 0; | |
269 | UBool hasLocalKeys = FALSE, hasPoolKeys = FALSE; | |
b75a7d8f | 270 | |
729e4ab9 A |
271 | if (U_FAILURE(*status)) { |
272 | return; | |
273 | } | |
274 | if (res->u.fTable.fCount == 0 && gFormatVersion > 1) { | |
275 | res->fRes = URES_MAKE_EMPTY_RESOURCE(URES_TABLE); | |
276 | res->fWritten = TRUE; | |
277 | return; | |
278 | } | |
279 | /* Find the smallest table type that fits the data. */ | |
280 | for (current = res->u.fTable.fFirst; current != NULL; current = current->fNext) { | |
281 | int32_t key; | |
282 | res_write16(bundle, current, status); | |
283 | if (bundle->fKeyMap == NULL) { | |
284 | key = current->fKey; | |
285 | } else { | |
286 | key = current->fKey = mapKey(bundle, current->fKey); | |
287 | } | |
288 | if (key >= 0) { | |
289 | hasLocalKeys = TRUE; | |
290 | if (key > maxKey) { | |
291 | maxKey = key; | |
292 | } | |
293 | } else { | |
294 | hasPoolKeys = TRUE; | |
295 | if (key > maxPoolKey) { | |
296 | maxPoolKey = key; | |
297 | } | |
298 | } | |
299 | res16 |= makeRes16(current->fRes); | |
300 | } | |
301 | if (U_FAILURE(*status)) { | |
302 | return; | |
303 | } | |
304 | if(res->u.fTable.fCount > (uint32_t)bundle->fMaxTableLength) { | |
305 | bundle->fMaxTableLength = res->u.fTable.fCount; | |
306 | } | |
307 | maxPoolKey &= 0x7fffffff; | |
308 | if (res->u.fTable.fCount <= 0xffff && | |
309 | (!hasLocalKeys || maxKey < bundle->fLocalKeyLimit) && | |
310 | (!hasPoolKeys || maxPoolKey < (0x10000 - bundle->fLocalKeyLimit)) | |
311 | ) { | |
312 | if (res16 >= 0 && gFormatVersion > 1) { | |
313 | uint16_t *p16 = reserve16BitUnits(bundle, 1 + res->u.fTable.fCount * 2, status); | |
314 | if (U_SUCCESS(*status)) { | |
315 | /* 16-bit count, key offsets and values */ | |
316 | res->fRes = URES_MAKE_RESOURCE(URES_TABLE16, bundle->f16BitUnitsLength); | |
317 | *p16++ = (uint16_t)res->u.fTable.fCount; | |
318 | for (current = res->u.fTable.fFirst; current != NULL; current = current->fNext) { | |
319 | *p16++ = makeKey16(bundle, current->fKey); | |
320 | } | |
321 | for (current = res->u.fTable.fFirst; current != NULL; current = current->fNext) { | |
322 | *p16++ = (uint16_t)makeRes16(current->fRes); | |
323 | } | |
324 | bundle->f16BitUnitsLength += 1 + res->u.fTable.fCount * 2; | |
325 | res->fWritten = TRUE; | |
326 | } | |
327 | } else { | |
328 | /* 16-bit count, 16-bit key offsets, 32-bit values */ | |
329 | res->u.fTable.fType = URES_TABLE; | |
330 | } | |
b75a7d8f | 331 | } else { |
729e4ab9 A |
332 | /* 32-bit count, key offsets and values */ |
333 | res->u.fTable.fType = URES_TABLE32; | |
b75a7d8f | 334 | } |
b75a7d8f A |
335 | } |
336 | ||
729e4ab9 A |
337 | static void |
338 | res_write16(struct SRBRoot *bundle, struct SResource *res, | |
339 | UErrorCode *status) { | |
340 | if (U_FAILURE(*status) || res == NULL) { | |
341 | return; | |
342 | } | |
343 | if (res->fRes != RES_BOGUS) { | |
344 | /* | |
345 | * The resource item word was already precomputed, which means | |
346 | * no further data needs to be written. | |
347 | * This might be an integer, or an empty or UTF-16 v2 string, | |
348 | * an empty binary, etc. | |
349 | */ | |
350 | return; | |
b75a7d8f | 351 | } |
729e4ab9 A |
352 | switch (res->fType) { |
353 | case URES_STRING: | |
354 | string_write16(bundle, res, status); | |
355 | break; | |
356 | case URES_ARRAY: | |
357 | array_write16(bundle, res, status); | |
358 | break; | |
359 | case URES_TABLE: | |
360 | table_write16(bundle, res, status); | |
361 | break; | |
362 | default: | |
363 | /* Only a few resource types write 16-bit units. */ | |
364 | break; | |
365 | } | |
366 | } | |
b75a7d8f | 367 | |
729e4ab9 A |
368 | /* |
369 | * Only called for UTF-16 v1 strings. | |
370 | * For UTF-16 v2 strings, res_preWrite() sees fRes != RES_BOGUS | |
371 | * and exits early. | |
372 | */ | |
373 | static void | |
374 | string_preWrite(uint32_t *byteOffset, | |
375 | struct SRBRoot *bundle, struct SResource *res, | |
376 | UErrorCode *status) { | |
377 | /* Write the UTF-16 v1 string. */ | |
378 | res->fRes = URES_MAKE_RESOURCE(URES_STRING, *byteOffset >> 2); | |
379 | *byteOffset += 4 + (res->u.fString.fLength + 1) * U_SIZEOF_UCHAR; | |
b75a7d8f A |
380 | } |
381 | ||
729e4ab9 A |
382 | static void |
383 | bin_preWrite(uint32_t *byteOffset, | |
384 | struct SRBRoot *bundle, struct SResource *res, | |
385 | UErrorCode *status) { | |
b75a7d8f | 386 | uint32_t pad = 0; |
729e4ab9 | 387 | uint32_t dataStart = *byteOffset + sizeof(res->u.fBinaryValue.fLength); |
b75a7d8f A |
388 | |
389 | if (dataStart % BIN_ALIGNMENT) { | |
390 | pad = (BIN_ALIGNMENT - dataStart % BIN_ALIGNMENT); | |
729e4ab9 | 391 | *byteOffset += pad; /* pad == 4 or 8 or 12 */ |
b75a7d8f | 392 | } |
729e4ab9 A |
393 | res->fRes = URES_MAKE_RESOURCE(URES_BINARY, *byteOffset >> 2); |
394 | *byteOffset += 4 + res->u.fBinaryValue.fLength; | |
b75a7d8f A |
395 | } |
396 | ||
729e4ab9 A |
397 | static void |
398 | array_preWrite(uint32_t *byteOffset, | |
399 | struct SRBRoot *bundle, struct SResource *res, | |
400 | UErrorCode *status) { | |
401 | struct SResource *current; | |
b75a7d8f | 402 | |
729e4ab9 A |
403 | if (U_FAILURE(*status)) { |
404 | return; | |
405 | } | |
406 | for (current = res->u.fArray.fFirst; current != NULL; current = current->fNext) { | |
407 | res_preWrite(byteOffset, bundle, current, status); | |
408 | } | |
409 | res->fRes = URES_MAKE_RESOURCE(URES_ARRAY, *byteOffset >> 2); | |
410 | *byteOffset += (1 + res->u.fArray.fCount) * 4; | |
411 | } | |
b75a7d8f | 412 | |
729e4ab9 A |
413 | static void |
414 | table_preWrite(uint32_t *byteOffset, | |
415 | struct SRBRoot *bundle, struct SResource *res, | |
416 | UErrorCode *status) { | |
417 | struct SResource *current; | |
b75a7d8f A |
418 | |
419 | if (U_FAILURE(*status)) { | |
729e4ab9 | 420 | return; |
b75a7d8f | 421 | } |
729e4ab9 A |
422 | for (current = res->u.fTable.fFirst; current != NULL; current = current->fNext) { |
423 | res_preWrite(byteOffset, bundle, current, status); | |
424 | } | |
425 | if (res->u.fTable.fType == URES_TABLE) { | |
426 | /* 16-bit count, 16-bit key offsets, 32-bit values */ | |
427 | res->fRes = URES_MAKE_RESOURCE(URES_TABLE, *byteOffset >> 2); | |
428 | *byteOffset += 2 + res->u.fTable.fCount * 6; | |
429 | } else { | |
430 | /* 32-bit count, key offsets and values */ | |
431 | res->fRes = URES_MAKE_RESOURCE(URES_TABLE32, *byteOffset >> 2); | |
432 | *byteOffset += 4 + res->u.fTable.fCount * 8; | |
433 | } | |
434 | } | |
b75a7d8f | 435 | |
729e4ab9 A |
436 | static void |
437 | res_preWrite(uint32_t *byteOffset, | |
438 | struct SRBRoot *bundle, struct SResource *res, | |
439 | UErrorCode *status) { | |
440 | if (U_FAILURE(*status) || res == NULL) { | |
441 | return; | |
442 | } | |
443 | if (res->fRes != RES_BOGUS) { | |
444 | /* | |
445 | * The resource item word was already precomputed, which means | |
446 | * no further data needs to be written. | |
447 | * This might be an integer, or an empty or UTF-16 v2 string, | |
448 | * an empty binary, etc. | |
449 | */ | |
450 | return; | |
451 | } | |
452 | switch (res->fType) { | |
453 | case URES_STRING: | |
454 | string_preWrite(byteOffset, bundle, res, status); | |
455 | break; | |
456 | case URES_ALIAS: | |
457 | res->fRes = URES_MAKE_RESOURCE(URES_ALIAS, *byteOffset >> 2); | |
458 | *byteOffset += 4 + (res->u.fString.fLength + 1) * U_SIZEOF_UCHAR; | |
459 | break; | |
460 | case URES_INT_VECTOR: | |
461 | if (res->u.fIntVector.fCount == 0 && gFormatVersion > 1) { | |
462 | res->fRes = URES_MAKE_EMPTY_RESOURCE(URES_INT_VECTOR); | |
463 | res->fWritten = TRUE; | |
374ca955 | 464 | } else { |
729e4ab9 A |
465 | res->fRes = URES_MAKE_RESOURCE(URES_INT_VECTOR, *byteOffset >> 2); |
466 | *byteOffset += (1 + res->u.fIntVector.fCount) * 4; | |
b75a7d8f | 467 | } |
729e4ab9 A |
468 | break; |
469 | case URES_BINARY: | |
470 | bin_preWrite(byteOffset, bundle, res, status); | |
471 | break; | |
472 | case URES_INT: | |
473 | break; | |
474 | case URES_ARRAY: | |
475 | array_preWrite(byteOffset, bundle, res, status); | |
476 | break; | |
477 | case URES_TABLE: | |
478 | table_preWrite(byteOffset, bundle, res, status); | |
479 | break; | |
480 | default: | |
481 | *status = U_INTERNAL_PROGRAM_ERROR; | |
482 | break; | |
483 | } | |
484 | *byteOffset += calcPadding(*byteOffset); | |
485 | } | |
b75a7d8f | 486 | |
729e4ab9 A |
487 | /* |
488 | * Only called for UTF-16 v1 strings. For UTF-16 v2 strings, | |
489 | * res_write() sees fWritten and exits early. | |
490 | */ | |
491 | static void string_write(UNewDataMemory *mem, uint32_t *byteOffset, | |
492 | struct SRBRoot *bundle, struct SResource *res, | |
493 | UErrorCode *status) { | |
494 | /* Write the UTF-16 v1 string. */ | |
495 | int32_t length = res->u.fString.fLength; | |
496 | udata_write32(mem, length); | |
497 | udata_writeUString(mem, res->u.fString.fChars, length + 1); | |
498 | *byteOffset += 4 + (length + 1) * U_SIZEOF_UCHAR; | |
499 | res->fWritten = TRUE; | |
500 | } | |
b75a7d8f | 501 | |
729e4ab9 A |
502 | static void alias_write(UNewDataMemory *mem, uint32_t *byteOffset, |
503 | struct SRBRoot *bundle, struct SResource *res, | |
504 | UErrorCode *status) { | |
505 | int32_t length = res->u.fString.fLength; | |
506 | udata_write32(mem, length); | |
507 | udata_writeUString(mem, res->u.fString.fChars, length + 1); | |
508 | *byteOffset += 4 + (length + 1) * U_SIZEOF_UCHAR; | |
509 | } | |
b75a7d8f | 510 | |
729e4ab9 A |
511 | static void array_write(UNewDataMemory *mem, uint32_t *byteOffset, |
512 | struct SRBRoot *bundle, struct SResource *res, | |
513 | UErrorCode *status) { | |
514 | uint32_t i; | |
b75a7d8f | 515 | |
729e4ab9 | 516 | struct SResource *current = NULL; |
b75a7d8f | 517 | |
729e4ab9 A |
518 | if (U_FAILURE(*status)) { |
519 | return; | |
520 | } | |
521 | for (i = 0, current = res->u.fArray.fFirst; current != NULL; ++i, current = current->fNext) { | |
522 | res_write(mem, byteOffset, bundle, current, status); | |
523 | } | |
524 | assert(i == res->u.fArray.fCount); | |
374ca955 | 525 | |
729e4ab9 A |
526 | udata_write32(mem, res->u.fArray.fCount); |
527 | for (current = res->u.fArray.fFirst; current != NULL; current = current->fNext) { | |
528 | udata_write32(mem, current->fRes); | |
529 | } | |
530 | *byteOffset += (1 + res->u.fArray.fCount) * 4; | |
531 | } | |
374ca955 | 532 | |
729e4ab9 A |
533 | static void intvector_write(UNewDataMemory *mem, uint32_t *byteOffset, |
534 | struct SRBRoot *bundle, struct SResource *res, | |
535 | UErrorCode *status) { | |
536 | uint32_t i = 0; | |
537 | udata_write32(mem, res->u.fIntVector.fCount); | |
538 | for(i = 0; i<res->u.fIntVector.fCount; i++) { | |
539 | udata_write32(mem, res->u.fIntVector.fArray[i]); | |
540 | } | |
541 | *byteOffset += (1 + res->u.fIntVector.fCount) * 4; | |
542 | } | |
b75a7d8f | 543 | |
729e4ab9 A |
544 | static void bin_write(UNewDataMemory *mem, uint32_t *byteOffset, |
545 | struct SRBRoot *bundle, struct SResource *res, | |
546 | UErrorCode *status) { | |
547 | uint32_t pad = 0; | |
548 | uint32_t dataStart = *byteOffset + sizeof(res->u.fBinaryValue.fLength); | |
b75a7d8f | 549 | |
729e4ab9 A |
550 | if (dataStart % BIN_ALIGNMENT) { |
551 | pad = (BIN_ALIGNMENT - dataStart % BIN_ALIGNMENT); | |
552 | udata_writePadding(mem, pad); /* pad == 4 or 8 or 12 */ | |
553 | *byteOffset += pad; | |
b75a7d8f A |
554 | } |
555 | ||
729e4ab9 A |
556 | udata_write32(mem, res->u.fBinaryValue.fLength); |
557 | if (res->u.fBinaryValue.fLength > 0) { | |
558 | udata_writeBlock(mem, res->u.fBinaryValue.fData, res->u.fBinaryValue.fLength); | |
559 | } | |
560 | *byteOffset += 4 + res->u.fBinaryValue.fLength; | |
b75a7d8f A |
561 | } |
562 | ||
729e4ab9 A |
563 | static void table_write(UNewDataMemory *mem, uint32_t *byteOffset, |
564 | struct SRBRoot *bundle, struct SResource *res, | |
565 | UErrorCode *status) { | |
566 | struct SResource *current; | |
567 | uint32_t i; | |
568 | ||
b75a7d8f | 569 | if (U_FAILURE(*status)) { |
729e4ab9 | 570 | return; |
b75a7d8f | 571 | } |
729e4ab9 A |
572 | for (i = 0, current = res->u.fTable.fFirst; current != NULL; ++i, current = current->fNext) { |
573 | assert(i < res->u.fTable.fCount); | |
574 | res_write(mem, byteOffset, bundle, current, status); | |
575 | } | |
576 | assert(i == res->u.fTable.fCount); | |
b75a7d8f | 577 | |
729e4ab9 A |
578 | if(res->u.fTable.fType == URES_TABLE) { |
579 | udata_write16(mem, (uint16_t)res->u.fTable.fCount); | |
580 | for (current = res->u.fTable.fFirst; current != NULL; current = current->fNext) { | |
581 | udata_write16(mem, makeKey16(bundle, current->fKey)); | |
582 | } | |
583 | *byteOffset += (1 + res->u.fTable.fCount)* 2; | |
584 | if ((res->u.fTable.fCount & 1) == 0) { | |
585 | /* 16-bit count and even number of 16-bit key offsets need padding before 32-bit resource items */ | |
586 | udata_writePadding(mem, 2); | |
587 | *byteOffset += 2; | |
b75a7d8f | 588 | } |
729e4ab9 A |
589 | } else /* URES_TABLE32 */ { |
590 | udata_write32(mem, res->u.fTable.fCount); | |
591 | for (current = res->u.fTable.fFirst; current != NULL; current = current->fNext) { | |
592 | udata_write32(mem, (uint32_t)current->fKey); | |
593 | } | |
594 | *byteOffset += (1 + res->u.fTable.fCount)* 4; | |
595 | } | |
596 | for (current = res->u.fTable.fFirst; current != NULL; current = current->fNext) { | |
597 | udata_write32(mem, current->fRes); | |
b75a7d8f | 598 | } |
729e4ab9 A |
599 | *byteOffset += res->u.fTable.fCount * 4; |
600 | } | |
b75a7d8f | 601 | |
729e4ab9 A |
602 | void res_write(UNewDataMemory *mem, uint32_t *byteOffset, |
603 | struct SRBRoot *bundle, struct SResource *res, | |
604 | UErrorCode *status) { | |
605 | uint8_t paddingSize; | |
606 | ||
607 | if (U_FAILURE(*status) || res == NULL) { | |
608 | return; | |
609 | } | |
610 | if (res->fWritten) { | |
611 | assert(res->fRes != RES_BOGUS); | |
612 | return; | |
613 | } | |
614 | switch (res->fType) { | |
615 | case URES_STRING: | |
616 | string_write (mem, byteOffset, bundle, res, status); | |
617 | break; | |
618 | case URES_ALIAS: | |
619 | alias_write (mem, byteOffset, bundle, res, status); | |
620 | break; | |
621 | case URES_INT_VECTOR: | |
622 | intvector_write (mem, byteOffset, bundle, res, status); | |
623 | break; | |
624 | case URES_BINARY: | |
625 | bin_write (mem, byteOffset, bundle, res, status); | |
626 | break; | |
627 | case URES_INT: | |
628 | break; /* fRes was set by int_open() */ | |
629 | case URES_ARRAY: | |
630 | array_write (mem, byteOffset, bundle, res, status); | |
631 | break; | |
632 | case URES_TABLE: | |
633 | table_write (mem, byteOffset, bundle, res, status); | |
634 | break; | |
635 | default: | |
636 | *status = U_INTERNAL_PROGRAM_ERROR; | |
637 | break; | |
638 | } | |
639 | paddingSize = calcPadding(*byteOffset); | |
640 | if (paddingSize > 0) { | |
641 | udata_writePadding(mem, paddingSize); | |
642 | *byteOffset += paddingSize; | |
643 | } | |
644 | res->fWritten = TRUE; | |
b75a7d8f A |
645 | } |
646 | ||
729e4ab9 A |
647 | void bundle_write(struct SRBRoot *bundle, |
648 | const char *outputDir, const char *outputPkg, | |
649 | char *writtenFilename, int writtenFilenameLen, | |
650 | UErrorCode *status) { | |
b75a7d8f | 651 | UNewDataMemory *mem = NULL; |
729e4ab9 | 652 | uint32_t byteOffset = 0; |
374ca955 | 653 | uint32_t top, size; |
b75a7d8f | 654 | char dataName[1024]; |
374ca955 | 655 | int32_t indexes[URES_INDEX_TOP]; |
b75a7d8f | 656 | |
729e4ab9 A |
657 | bundle_compactKeys(bundle, status); |
658 | /* | |
659 | * Add padding bytes to fKeys so that fKeysTop is 4-aligned. | |
660 | * Safe because the capacity is a multiple of 4. | |
661 | */ | |
662 | while (bundle->fKeysTop & 3) { | |
663 | bundle->fKeys[bundle->fKeysTop++] = (char)0xaa; | |
664 | } | |
665 | /* | |
666 | * In URES_TABLE, use all local key offsets that fit into 16 bits, | |
667 | * and use the remaining 16-bit offsets for pool key offsets | |
668 | * if there are any. | |
669 | * If there are no local keys, then use the whole 16-bit space | |
670 | * for pool key offsets. | |
671 | * Note: This cannot be changed without changing the major formatVersion. | |
672 | */ | |
673 | if (bundle->fKeysBottom < bundle->fKeysTop) { | |
674 | if (bundle->fKeysTop <= 0x10000) { | |
675 | bundle->fLocalKeyLimit = bundle->fKeysTop; | |
676 | } else { | |
677 | bundle->fLocalKeyLimit = 0x10000; | |
678 | } | |
679 | } else { | |
680 | bundle->fLocalKeyLimit = 0; | |
b75a7d8f A |
681 | } |
682 | ||
729e4ab9 A |
683 | bundle_compactStrings(bundle, status); |
684 | res_write16(bundle, bundle->fRoot, status); | |
685 | if (bundle->f16BitUnitsLength & 1) { | |
686 | bundle->f16BitUnits[bundle->f16BitUnitsLength++] = 0xaaaa; /* pad to multiple of 4 bytes */ | |
687 | } | |
688 | /* all keys have been mapped */ | |
689 | uprv_free(bundle->fKeyMap); | |
690 | bundle->fKeyMap = NULL; | |
691 | ||
692 | byteOffset = bundle->fKeysTop + bundle->f16BitUnitsLength * 2; | |
693 | res_preWrite(&byteOffset, bundle, bundle->fRoot, status); | |
694 | ||
695 | /* total size including the root item */ | |
696 | top = byteOffset; | |
697 | ||
b75a7d8f A |
698 | if (U_FAILURE(*status)) { |
699 | return; | |
700 | } | |
701 | ||
729e4ab9 A |
702 | if (writtenFilename && writtenFilenameLen) { |
703 | *writtenFilename = 0; | |
704 | } | |
705 | ||
b75a7d8f A |
706 | if (writtenFilename) { |
707 | int32_t off = 0, len = 0; | |
708 | if (outputDir) { | |
709 | len = (int32_t)uprv_strlen(outputDir); | |
710 | if (len > writtenFilenameLen) { | |
711 | len = writtenFilenameLen; | |
712 | } | |
713 | uprv_strncpy(writtenFilename, outputDir, len); | |
714 | } | |
715 | if (writtenFilenameLen -= len) { | |
716 | off += len; | |
717 | writtenFilename[off] = U_FILE_SEP_CHAR; | |
718 | if (--writtenFilenameLen) { | |
719 | ++off; | |
720 | if(outputPkg != NULL) | |
721 | { | |
722 | uprv_strcpy(writtenFilename+off, outputPkg); | |
374ca955 | 723 | off += (int32_t)uprv_strlen(outputPkg); |
b75a7d8f A |
724 | writtenFilename[off] = '_'; |
725 | ++off; | |
726 | } | |
727 | ||
728 | len = (int32_t)uprv_strlen(bundle->fLocale); | |
729 | if (len > writtenFilenameLen) { | |
730 | len = writtenFilenameLen; | |
731 | } | |
732 | uprv_strncpy(writtenFilename + off, bundle->fLocale, len); | |
733 | if (writtenFilenameLen -= len) { | |
734 | off += len; | |
735 | len = 5; | |
736 | if (len > writtenFilenameLen) { | |
737 | len = writtenFilenameLen; | |
738 | } | |
739 | uprv_strncpy(writtenFilename + off, ".res", len); | |
740 | } | |
741 | } | |
742 | } | |
743 | } | |
744 | ||
745 | if(outputPkg) | |
746 | { | |
747 | uprv_strcpy(dataName, outputPkg); | |
748 | uprv_strcat(dataName, "_"); | |
749 | uprv_strcat(dataName, bundle->fLocale); | |
750 | } | |
751 | else | |
752 | { | |
753 | uprv_strcpy(dataName, bundle->fLocale); | |
754 | } | |
755 | ||
729e4ab9 A |
756 | uprv_memcpy(dataInfo.formatVersion, gFormatVersions + gFormatVersion, sizeof(UVersionInfo)); |
757 | ||
b75a7d8f A |
758 | mem = udata_create(outputDir, "res", dataName, &dataInfo, (gIncludeCopyright==TRUE)? U_COPYRIGHT_STRING:NULL, status); |
759 | if(U_FAILURE(*status)){ | |
760 | return; | |
761 | } | |
b75a7d8f | 762 | |
374ca955 | 763 | /* write the root item */ |
729e4ab9 | 764 | udata_write32(mem, bundle->fRoot->fRes); |
374ca955 A |
765 | |
766 | /* | |
767 | * formatVersion 1.1 (ICU 2.8): | |
768 | * write int32_t indexes[] after root and before the strings | |
769 | * to make it easier to parse resource bundles in icuswap or from Java etc. | |
770 | */ | |
73c04bcf | 771 | uprv_memset(indexes, 0, sizeof(indexes)); |
729e4ab9 A |
772 | indexes[URES_INDEX_LENGTH]= bundle->fIndexLength; |
773 | indexes[URES_INDEX_KEYS_TOP]= bundle->fKeysTop>>2; | |
374ca955 A |
774 | indexes[URES_INDEX_RESOURCES_TOP]= (int32_t)(top>>2); |
775 | indexes[URES_INDEX_BUNDLE_TOP]= indexes[URES_INDEX_RESOURCES_TOP]; | |
776 | indexes[URES_INDEX_MAX_TABLE_LENGTH]= bundle->fMaxTableLength; | |
777 | ||
73c04bcf A |
778 | /* |
779 | * formatVersion 1.2 (ICU 3.6): | |
780 | * write indexes[URES_INDEX_ATTRIBUTES] with URES_ATT_NO_FALLBACK set or not set | |
781 | * the memset() above initialized all indexes[] to 0 | |
782 | */ | |
729e4ab9 | 783 | if (bundle->noFallback) { |
73c04bcf A |
784 | indexes[URES_INDEX_ATTRIBUTES]=URES_ATT_NO_FALLBACK; |
785 | } | |
729e4ab9 A |
786 | /* |
787 | * formatVersion 2.0 (ICU 4.4): | |
788 | * more compact string value storage, optional pool bundle | |
789 | */ | |
790 | if (URES_INDEX_16BIT_TOP < bundle->fIndexLength) { | |
791 | indexes[URES_INDEX_16BIT_TOP] = (bundle->fKeysTop>>2) + (bundle->f16BitUnitsLength>>1); | |
792 | } | |
793 | if (URES_INDEX_POOL_CHECKSUM < bundle->fIndexLength) { | |
794 | if (bundle->fIsPoolBundle) { | |
795 | indexes[URES_INDEX_ATTRIBUTES] |= URES_ATT_IS_POOL_BUNDLE | URES_ATT_NO_FALLBACK; | |
796 | indexes[URES_INDEX_POOL_CHECKSUM] = | |
797 | (int32_t)computeCRC((char *)(bundle->fKeys + bundle->fKeysBottom), | |
798 | (uint32_t)(bundle->fKeysTop - bundle->fKeysBottom), | |
799 | 0); | |
800 | } else if (gUsePoolBundle) { | |
801 | indexes[URES_INDEX_ATTRIBUTES] |= URES_ATT_USES_POOL_BUNDLE; | |
802 | indexes[URES_INDEX_POOL_CHECKSUM] = bundle->fPoolChecksum; | |
803 | } | |
804 | } | |
73c04bcf | 805 | |
374ca955 | 806 | /* write the indexes[] */ |
729e4ab9 | 807 | udata_writeBlock(mem, indexes, bundle->fIndexLength*4); |
374ca955 A |
808 | |
809 | /* write the table key strings */ | |
729e4ab9 A |
810 | udata_writeBlock(mem, bundle->fKeys+bundle->fKeysBottom, |
811 | bundle->fKeysTop-bundle->fKeysBottom); | |
374ca955 | 812 | |
729e4ab9 A |
813 | /* write the v2 UTF-16 strings, URES_TABLE16 and URES_ARRAY16 */ |
814 | udata_writeBlock(mem, bundle->f16BitUnits, bundle->f16BitUnitsLength*2); | |
b75a7d8f | 815 | |
374ca955 | 816 | /* write all of the bundle contents: the root item and its children */ |
729e4ab9 A |
817 | byteOffset = bundle->fKeysTop + bundle->f16BitUnitsLength * 2; |
818 | res_write(mem, &byteOffset, bundle, bundle->fRoot, status); | |
819 | assert(byteOffset == top); | |
b75a7d8f | 820 | |
374ca955 A |
821 | size = udata_finish(mem, status); |
822 | if(top != size) { | |
823 | fprintf(stderr, "genrb error: wrote %u bytes but counted %u\n", | |
824 | (int)size, (int)top); | |
825 | *status = U_INTERNAL_PROGRAM_ERROR; | |
826 | } | |
b75a7d8f A |
827 | } |
828 | ||
829 | /* Opening Functions */ | |
b75a7d8f | 830 | |
729e4ab9 A |
831 | /* gcc 4.2 complained "no previous prototype for res_open" without this prototype... */ |
832 | struct SResource* res_open(struct SRBRoot *bundle, const char *tag, | |
833 | const struct UString* comment, UErrorCode* status); | |
834 | ||
835 | struct SResource* res_open(struct SRBRoot *bundle, const char *tag, | |
836 | const struct UString* comment, UErrorCode* status){ | |
837 | struct SResource *res; | |
838 | int32_t key = bundle_addtag(bundle, tag, status); | |
b75a7d8f A |
839 | if (U_FAILURE(*status)) { |
840 | return NULL; | |
841 | } | |
842 | ||
843 | res = (struct SResource *) uprv_malloc(sizeof(struct SResource)); | |
b75a7d8f A |
844 | if (res == NULL) { |
845 | *status = U_MEMORY_ALLOCATION_ERROR; | |
846 | return NULL; | |
847 | } | |
374ca955 | 848 | uprv_memset(res, 0, sizeof(struct SResource)); |
729e4ab9 A |
849 | res->fKey = key; |
850 | res->fRes = RES_BOGUS; | |
374ca955 | 851 | |
46f4442e | 852 | ustr_init(&res->fComment); |
374ca955 | 853 | if(comment != NULL){ |
46f4442e | 854 | ustr_cpy(&res->fComment, comment, status); |
729e4ab9 A |
855 | if (U_FAILURE(*status)) { |
856 | res_close(res); | |
857 | return NULL; | |
858 | } | |
374ca955 A |
859 | } |
860 | return res; | |
374ca955 | 861 | } |
46f4442e A |
862 | |
863 | struct SResource* res_none() { | |
729e4ab9 | 864 | return (struct SResource*)&kNoResource; |
46f4442e A |
865 | } |
866 | ||
729e4ab9 A |
867 | struct SResource* table_open(struct SRBRoot *bundle, const char *tag, const struct UString* comment, UErrorCode *status) { |
868 | struct SResource *res = res_open(bundle, tag, comment, status); | |
b75a7d8f | 869 | if (U_FAILURE(*status)) { |
b75a7d8f A |
870 | return NULL; |
871 | } | |
374ca955 | 872 | res->fType = URES_TABLE; |
729e4ab9 | 873 | res->u.fTable.fRoot = bundle; |
b75a7d8f A |
874 | return res; |
875 | } | |
876 | ||
73c04bcf | 877 | struct SResource* array_open(struct SRBRoot *bundle, const char *tag, const struct UString* comment, UErrorCode *status) { |
729e4ab9 | 878 | struct SResource *res = res_open(bundle, tag, comment, status); |
374ca955 | 879 | if (U_FAILURE(*status)) { |
b75a7d8f A |
880 | return NULL; |
881 | } | |
b75a7d8f | 882 | res->fType = URES_ARRAY; |
729e4ab9 A |
883 | return res; |
884 | } | |
b75a7d8f | 885 | |
729e4ab9 A |
886 | static int32_t U_CALLCONV |
887 | string_hash(const UHashTok key) { | |
888 | const struct SResource *res = (struct SResource *)key.pointer; | |
889 | return uhash_hashUCharsN(res->u.fString.fChars, res->u.fString.fLength); | |
890 | } | |
b75a7d8f | 891 | |
729e4ab9 A |
892 | static UBool U_CALLCONV |
893 | string_comp(const UHashTok key1, const UHashTok key2) { | |
894 | const struct SResource *res1 = (struct SResource *)key1.pointer; | |
895 | const struct SResource *res2 = (struct SResource *)key2.pointer; | |
896 | return 0 == u_strCompare(res1->u.fString.fChars, res1->u.fString.fLength, | |
897 | res2->u.fString.fChars, res2->u.fString.fLength, | |
898 | FALSE); | |
b75a7d8f A |
899 | } |
900 | ||
374ca955 | 901 | struct SResource *string_open(struct SRBRoot *bundle, char *tag, const UChar *value, int32_t len, const struct UString* comment, UErrorCode *status) { |
729e4ab9 | 902 | struct SResource *res = res_open(bundle, tag, comment, status); |
b75a7d8f A |
903 | if (U_FAILURE(*status)) { |
904 | return NULL; | |
905 | } | |
b75a7d8f | 906 | res->fType = URES_STRING; |
b75a7d8f | 907 | |
729e4ab9 A |
908 | if (len == 0 && gFormatVersion > 1) { |
909 | res->u.fString.fChars = &gEmptyString; | |
910 | res->fRes = 0; | |
911 | res->fWritten = TRUE; | |
912 | return res; | |
b75a7d8f A |
913 | } |
914 | ||
b75a7d8f | 915 | res->u.fString.fLength = len; |
b75a7d8f | 916 | |
729e4ab9 A |
917 | if (gFormatVersion > 1) { |
918 | /* check for duplicates */ | |
919 | res->u.fString.fChars = (UChar *)value; | |
920 | if (bundle->fStringSet == NULL) { | |
921 | UErrorCode localStatus = U_ZERO_ERROR; /* if failure: just don't detect dups */ | |
922 | bundle->fStringSet = uhash_open(string_hash, string_comp, string_comp, &localStatus); | |
923 | } else { | |
924 | res->u.fString.fSame = uhash_get(bundle->fStringSet, res); | |
925 | } | |
b75a7d8f | 926 | } |
729e4ab9 A |
927 | if (res->u.fString.fSame == NULL) { |
928 | /* this is a new string */ | |
929 | res->u.fString.fChars = (UChar *) uprv_malloc(sizeof(UChar) * (len + 1)); | |
b75a7d8f | 930 | |
729e4ab9 A |
931 | if (res->u.fString.fChars == NULL) { |
932 | *status = U_MEMORY_ALLOCATION_ERROR; | |
933 | uprv_free(res); | |
934 | return NULL; | |
935 | } | |
b75a7d8f | 936 | |
729e4ab9 A |
937 | uprv_memcpy(res->u.fString.fChars, value, sizeof(UChar) * len); |
938 | res->u.fString.fChars[len] = 0; | |
939 | if (bundle->fStringSet != NULL) { | |
940 | /* put it into the set for finding duplicates */ | |
941 | uhash_put(bundle->fStringSet, res, res, status); | |
942 | } | |
943 | ||
944 | if (bundle->fStringsForm != STRINGS_UTF16_V1) { | |
945 | if (len <= MAX_IMPLICIT_STRING_LENGTH && !U16_IS_TRAIL(value[0]) && len == u_strlen(value)) { | |
946 | /* | |
947 | * This string will be stored without an explicit length. | |
948 | * Runtime will detect !U16_IS_TRAIL(value[0]) and call u_strlen(). | |
949 | */ | |
950 | res->u.fString.fNumCharsForLength = 0; | |
951 | } else if (len <= 0x3ee) { | |
952 | res->u.fString.fNumCharsForLength = 1; | |
953 | } else if (len <= 0xfffff) { | |
954 | res->u.fString.fNumCharsForLength = 2; | |
955 | } else { | |
956 | res->u.fString.fNumCharsForLength = 3; | |
957 | } | |
958 | bundle->f16BitUnitsLength += res->u.fString.fNumCharsForLength + len + 1; /* +1 for the NUL */ | |
959 | } | |
960 | } else { | |
961 | /* this is a duplicate of fSame */ | |
962 | struct SResource *same = res->u.fString.fSame; | |
963 | res->u.fString.fChars = same->u.fString.fChars; | |
964 | } | |
b75a7d8f A |
965 | return res; |
966 | } | |
967 | ||
968 | /* TODO: make alias_open and string_open use the same code */ | |
374ca955 | 969 | struct SResource *alias_open(struct SRBRoot *bundle, char *tag, UChar *value, int32_t len, const struct UString* comment, UErrorCode *status) { |
729e4ab9 | 970 | struct SResource *res = res_open(bundle, tag, comment, status); |
b75a7d8f A |
971 | if (U_FAILURE(*status)) { |
972 | return NULL; | |
973 | } | |
b75a7d8f | 974 | res->fType = URES_ALIAS; |
729e4ab9 A |
975 | if (len == 0 && gFormatVersion > 1) { |
976 | res->u.fString.fChars = &gEmptyString; | |
977 | res->fRes = URES_MAKE_EMPTY_RESOURCE(URES_ALIAS); | |
978 | res->fWritten = TRUE; | |
979 | return res; | |
b75a7d8f A |
980 | } |
981 | ||
b75a7d8f A |
982 | res->u.fString.fLength = len; |
983 | res->u.fString.fChars = (UChar *) uprv_malloc(sizeof(UChar) * (len + 1)); | |
b75a7d8f A |
984 | if (res->u.fString.fChars == NULL) { |
985 | *status = U_MEMORY_ALLOCATION_ERROR; | |
986 | uprv_free(res); | |
987 | return NULL; | |
988 | } | |
b75a7d8f | 989 | uprv_memcpy(res->u.fString.fChars, value, sizeof(UChar) * (len + 1)); |
b75a7d8f A |
990 | return res; |
991 | } | |
992 | ||
993 | ||
374ca955 | 994 | struct SResource* intvector_open(struct SRBRoot *bundle, char *tag, const struct UString* comment, UErrorCode *status) { |
729e4ab9 | 995 | struct SResource *res = res_open(bundle, tag, comment, status); |
b75a7d8f A |
996 | if (U_FAILURE(*status)) { |
997 | return NULL; | |
998 | } | |
b75a7d8f | 999 | res->fType = URES_INT_VECTOR; |
b75a7d8f A |
1000 | |
1001 | res->u.fIntVector.fCount = 0; | |
1002 | res->u.fIntVector.fArray = (uint32_t *) uprv_malloc(sizeof(uint32_t) * RESLIST_MAX_INT_VECTOR); | |
b75a7d8f A |
1003 | if (res->u.fIntVector.fArray == NULL) { |
1004 | *status = U_MEMORY_ALLOCATION_ERROR; | |
1005 | uprv_free(res); | |
1006 | return NULL; | |
1007 | } | |
b75a7d8f A |
1008 | return res; |
1009 | } | |
1010 | ||
374ca955 | 1011 | struct SResource *int_open(struct SRBRoot *bundle, char *tag, int32_t value, const struct UString* comment, UErrorCode *status) { |
729e4ab9 | 1012 | struct SResource *res = res_open(bundle, tag, comment, status); |
b75a7d8f A |
1013 | if (U_FAILURE(*status)) { |
1014 | return NULL; | |
1015 | } | |
b75a7d8f | 1016 | res->fType = URES_INT; |
b75a7d8f | 1017 | res->u.fIntValue.fValue = value; |
729e4ab9 A |
1018 | res->fRes = URES_MAKE_RESOURCE(URES_INT, value & 0x0FFFFFFF); |
1019 | res->fWritten = TRUE; | |
b75a7d8f A |
1020 | return res; |
1021 | } | |
1022 | ||
374ca955 | 1023 | struct SResource *bin_open(struct SRBRoot *bundle, const char *tag, uint32_t length, uint8_t *data, const char* fileName, const struct UString* comment, UErrorCode *status) { |
729e4ab9 | 1024 | struct SResource *res = res_open(bundle, tag, comment, status); |
b75a7d8f A |
1025 | if (U_FAILURE(*status)) { |
1026 | return NULL; | |
1027 | } | |
b75a7d8f | 1028 | res->fType = URES_BINARY; |
b75a7d8f A |
1029 | |
1030 | res->u.fBinaryValue.fLength = length; | |
1031 | res->u.fBinaryValue.fFileName = NULL; | |
1032 | if(fileName!=NULL && uprv_strcmp(fileName, "") !=0){ | |
1033 | res->u.fBinaryValue.fFileName = (char*) uprv_malloc(sizeof(char) * (uprv_strlen(fileName)+1)); | |
1034 | uprv_strcpy(res->u.fBinaryValue.fFileName,fileName); | |
1035 | } | |
1036 | if (length > 0) { | |
1037 | res->u.fBinaryValue.fData = (uint8_t *) uprv_malloc(sizeof(uint8_t) * length); | |
1038 | ||
1039 | if (res->u.fBinaryValue.fData == NULL) { | |
1040 | *status = U_MEMORY_ALLOCATION_ERROR; | |
1041 | uprv_free(res); | |
1042 | return NULL; | |
1043 | } | |
1044 | ||
1045 | uprv_memcpy(res->u.fBinaryValue.fData, data, length); | |
1046 | } | |
1047 | else { | |
1048 | res->u.fBinaryValue.fData = NULL; | |
729e4ab9 A |
1049 | if (gFormatVersion > 1) { |
1050 | res->fRes = URES_MAKE_EMPTY_RESOURCE(URES_BINARY); | |
1051 | res->fWritten = TRUE; | |
1052 | } | |
b75a7d8f A |
1053 | } |
1054 | ||
b75a7d8f A |
1055 | return res; |
1056 | } | |
1057 | ||
729e4ab9 A |
1058 | struct SRBRoot *bundle_open(const struct UString* comment, UBool isPoolBundle, UErrorCode *status) { |
1059 | struct SRBRoot *bundle; | |
b75a7d8f A |
1060 | |
1061 | if (U_FAILURE(*status)) { | |
1062 | return NULL; | |
1063 | } | |
1064 | ||
1065 | bundle = (struct SRBRoot *) uprv_malloc(sizeof(struct SRBRoot)); | |
b75a7d8f A |
1066 | if (bundle == NULL) { |
1067 | *status = U_MEMORY_ALLOCATION_ERROR; | |
1068 | return 0; | |
1069 | } | |
374ca955 | 1070 | uprv_memset(bundle, 0, sizeof(struct SRBRoot)); |
b75a7d8f | 1071 | |
729e4ab9 A |
1072 | bundle->fKeys = (char *) uprv_malloc(sizeof(char) * KEY_SPACE_SIZE); |
1073 | bundle->fRoot = table_open(bundle, NULL, comment, status); | |
1074 | if (bundle->fKeys == NULL || bundle->fRoot == NULL || U_FAILURE(*status)) { | |
374ca955 A |
1075 | if (U_SUCCESS(*status)) { |
1076 | *status = U_MEMORY_ALLOCATION_ERROR; | |
1077 | } | |
729e4ab9 | 1078 | bundle_close(bundle, status); |
b75a7d8f A |
1079 | return NULL; |
1080 | } | |
1081 | ||
729e4ab9 A |
1082 | bundle->fLocale = NULL; |
1083 | bundle->fKeysCapacity = KEY_SPACE_SIZE; | |
1084 | /* formatVersion 1.1: start fKeysTop after the root item and indexes[] */ | |
1085 | bundle->fIsPoolBundle = isPoolBundle; | |
1086 | if (gUsePoolBundle || isPoolBundle) { | |
1087 | bundle->fIndexLength = URES_INDEX_POOL_CHECKSUM + 1; | |
1088 | } else if (gFormatVersion >= 2) { | |
1089 | bundle->fIndexLength = URES_INDEX_16BIT_TOP + 1; | |
1090 | } else /* formatVersion 1 */ { | |
1091 | bundle->fIndexLength = URES_INDEX_ATTRIBUTES + 1; | |
1092 | } | |
1093 | bundle->fKeysBottom = (1 /* root */ + bundle->fIndexLength) * 4; | |
1094 | uprv_memset(bundle->fKeys, 0, bundle->fKeysBottom); | |
1095 | bundle->fKeysTop = bundle->fKeysBottom; | |
1096 | ||
1097 | if (gFormatVersion == 1) { | |
1098 | bundle->fStringsForm = STRINGS_UTF16_V1; | |
1099 | } else { | |
1100 | bundle->fStringsForm = STRINGS_UTF16_V2; | |
1101 | } | |
1102 | ||
b75a7d8f A |
1103 | return bundle; |
1104 | } | |
1105 | ||
1106 | /* Closing Functions */ | |
46f4442e | 1107 | static void table_close(struct SResource *table) { |
b75a7d8f A |
1108 | struct SResource *current = NULL; |
1109 | struct SResource *prev = NULL; | |
1110 | ||
1111 | current = table->u.fTable.fFirst; | |
1112 | ||
1113 | while (current != NULL) { | |
1114 | prev = current; | |
1115 | current = current->fNext; | |
1116 | ||
46f4442e | 1117 | res_close(prev); |
b75a7d8f A |
1118 | } |
1119 | ||
1120 | table->u.fTable.fFirst = NULL; | |
1121 | } | |
1122 | ||
46f4442e | 1123 | static void array_close(struct SResource *array) { |
b75a7d8f A |
1124 | struct SResource *current = NULL; |
1125 | struct SResource *prev = NULL; | |
73c04bcf A |
1126 | |
1127 | if(array==NULL){ | |
1128 | return; | |
1129 | } | |
b75a7d8f | 1130 | current = array->u.fArray.fFirst; |
73c04bcf | 1131 | |
b75a7d8f A |
1132 | while (current != NULL) { |
1133 | prev = current; | |
1134 | current = current->fNext; | |
1135 | ||
46f4442e | 1136 | res_close(prev); |
b75a7d8f A |
1137 | } |
1138 | array->u.fArray.fFirst = NULL; | |
1139 | } | |
1140 | ||
46f4442e | 1141 | static void string_close(struct SResource *string) { |
729e4ab9 A |
1142 | if (string->u.fString.fChars != NULL && |
1143 | string->u.fString.fChars != &gEmptyString && | |
1144 | string->u.fString.fSame == NULL | |
1145 | ) { | |
b75a7d8f A |
1146 | uprv_free(string->u.fString.fChars); |
1147 | string->u.fString.fChars =NULL; | |
1148 | } | |
1149 | } | |
1150 | ||
46f4442e | 1151 | static void alias_close(struct SResource *alias) { |
b75a7d8f A |
1152 | if (alias->u.fString.fChars != NULL) { |
1153 | uprv_free(alias->u.fString.fChars); | |
1154 | alias->u.fString.fChars =NULL; | |
1155 | } | |
1156 | } | |
1157 | ||
46f4442e | 1158 | static void intvector_close(struct SResource *intvector) { |
b75a7d8f A |
1159 | if (intvector->u.fIntVector.fArray != NULL) { |
1160 | uprv_free(intvector->u.fIntVector.fArray); | |
1161 | intvector->u.fIntVector.fArray =NULL; | |
1162 | } | |
1163 | } | |
1164 | ||
46f4442e | 1165 | static void int_close(struct SResource *intres) { |
b75a7d8f A |
1166 | /* Intentionally left blank */ |
1167 | } | |
1168 | ||
46f4442e | 1169 | static void bin_close(struct SResource *binres) { |
b75a7d8f A |
1170 | if (binres->u.fBinaryValue.fData != NULL) { |
1171 | uprv_free(binres->u.fBinaryValue.fData); | |
1172 | binres->u.fBinaryValue.fData = NULL; | |
1173 | } | |
1174 | } | |
1175 | ||
46f4442e | 1176 | void res_close(struct SResource *res) { |
b75a7d8f A |
1177 | if (res != NULL) { |
1178 | switch(res->fType) { | |
1179 | case URES_STRING: | |
46f4442e | 1180 | string_close(res); |
b75a7d8f A |
1181 | break; |
1182 | case URES_ALIAS: | |
46f4442e | 1183 | alias_close(res); |
b75a7d8f A |
1184 | break; |
1185 | case URES_INT_VECTOR: | |
46f4442e | 1186 | intvector_close(res); |
b75a7d8f A |
1187 | break; |
1188 | case URES_BINARY: | |
46f4442e | 1189 | bin_close(res); |
b75a7d8f A |
1190 | break; |
1191 | case URES_INT: | |
46f4442e | 1192 | int_close(res); |
b75a7d8f A |
1193 | break; |
1194 | case URES_ARRAY: | |
46f4442e | 1195 | array_close(res); |
b75a7d8f | 1196 | break; |
374ca955 | 1197 | case URES_TABLE: |
46f4442e | 1198 | table_close(res); |
b75a7d8f A |
1199 | break; |
1200 | default: | |
1201 | /* Shouldn't happen */ | |
1202 | break; | |
1203 | } | |
1204 | ||
46f4442e | 1205 | ustr_deinit(&res->fComment); |
b75a7d8f A |
1206 | uprv_free(res); |
1207 | } | |
1208 | } | |
1209 | ||
1210 | void bundle_close(struct SRBRoot *bundle, UErrorCode *status) { | |
729e4ab9 A |
1211 | res_close(bundle->fRoot); |
1212 | uprv_free(bundle->fLocale); | |
1213 | uprv_free(bundle->fKeys); | |
1214 | uprv_free(bundle->fKeyMap); | |
1215 | uhash_close(bundle->fStringSet); | |
1216 | uprv_free(bundle->f16BitUnits); | |
1217 | uprv_free(bundle); | |
1218 | } | |
b75a7d8f | 1219 | |
729e4ab9 A |
1220 | void bundle_closeString(struct SRBRoot *bundle, struct SResource *string) { |
1221 | if (bundle->fStringSet != NULL) { | |
1222 | uhash_remove(bundle->fStringSet, string); | |
b75a7d8f | 1223 | } |
729e4ab9 | 1224 | string_close(string); |
b75a7d8f A |
1225 | } |
1226 | ||
1227 | /* Adding Functions */ | |
1228 | void table_add(struct SResource *table, struct SResource *res, int linenumber, UErrorCode *status) { | |
1229 | struct SResource *current = NULL; | |
1230 | struct SResource *prev = NULL; | |
1231 | struct SResTable *list; | |
729e4ab9 | 1232 | const char *resKeyString; |
b75a7d8f A |
1233 | |
1234 | if (U_FAILURE(*status)) { | |
1235 | return; | |
1236 | } | |
46f4442e A |
1237 | if (res == &kNoResource) { |
1238 | return; | |
1239 | } | |
b75a7d8f A |
1240 | |
1241 | /* remember this linenumber to report to the user if there is a duplicate key */ | |
1242 | res->line = linenumber; | |
1243 | ||
1244 | /* here we need to traverse the list */ | |
1245 | list = &(table->u.fTable); | |
b75a7d8f | 1246 | ++(list->fCount); |
b75a7d8f A |
1247 | |
1248 | /* is list still empty? */ | |
1249 | if (list->fFirst == NULL) { | |
1250 | list->fFirst = res; | |
1251 | res->fNext = NULL; | |
1252 | return; | |
1253 | } | |
1254 | ||
729e4ab9 A |
1255 | resKeyString = list->fRoot->fKeys + res->fKey; |
1256 | ||
b75a7d8f A |
1257 | current = list->fFirst; |
1258 | ||
1259 | while (current != NULL) { | |
729e4ab9 A |
1260 | const char *currentKeyString = list->fRoot->fKeys + current->fKey; |
1261 | int diff; | |
1262 | /* | |
1263 | * formatVersion 1: compare key strings in native-charset order | |
1264 | * formatVersion 2 and up: compare key strings in ASCII order | |
1265 | */ | |
1266 | if (gFormatVersion == 1 || U_CHARSET_FAMILY == U_ASCII_FAMILY) { | |
1267 | diff = uprv_strcmp(currentKeyString, resKeyString); | |
1268 | } else { | |
1269 | diff = uprv_compareInvCharsAsAscii(currentKeyString, resKeyString); | |
1270 | } | |
1271 | if (diff < 0) { | |
b75a7d8f A |
1272 | prev = current; |
1273 | current = current->fNext; | |
729e4ab9 | 1274 | } else if (diff > 0) { |
b75a7d8f A |
1275 | /* we're either in front of list, or in middle */ |
1276 | if (prev == NULL) { | |
1277 | /* front of the list */ | |
1278 | list->fFirst = res; | |
1279 | } else { | |
1280 | /* middle of the list */ | |
1281 | prev->fNext = res; | |
1282 | } | |
1283 | ||
1284 | res->fNext = current; | |
1285 | return; | |
1286 | } else { | |
1287 | /* Key already exists! ERROR! */ | |
729e4ab9 | 1288 | error(linenumber, "duplicate key '%s' in table, first appeared at line %d", currentKeyString, current->line); |
b75a7d8f A |
1289 | *status = U_UNSUPPORTED_ERROR; |
1290 | return; | |
1291 | } | |
1292 | } | |
1293 | ||
1294 | /* end of list */ | |
1295 | prev->fNext = res; | |
1296 | res->fNext = NULL; | |
1297 | } | |
1298 | ||
1299 | void array_add(struct SResource *array, struct SResource *res, UErrorCode *status) { | |
1300 | if (U_FAILURE(*status)) { | |
1301 | return; | |
1302 | } | |
1303 | ||
1304 | if (array->u.fArray.fFirst == NULL) { | |
1305 | array->u.fArray.fFirst = res; | |
1306 | array->u.fArray.fLast = res; | |
1307 | } else { | |
1308 | array->u.fArray.fLast->fNext = res; | |
1309 | array->u.fArray.fLast = res; | |
1310 | } | |
1311 | ||
1312 | (array->u.fArray.fCount)++; | |
b75a7d8f A |
1313 | } |
1314 | ||
1315 | void intvector_add(struct SResource *intvector, int32_t value, UErrorCode *status) { | |
1316 | if (U_FAILURE(*status)) { | |
1317 | return; | |
1318 | } | |
1319 | ||
1320 | *(intvector->u.fIntVector.fArray + intvector->u.fIntVector.fCount) = value; | |
1321 | intvector->u.fIntVector.fCount++; | |
b75a7d8f A |
1322 | } |
1323 | ||
1324 | /* Misc Functions */ | |
1325 | ||
1326 | void bundle_setlocale(struct SRBRoot *bundle, UChar *locale, UErrorCode *status) { | |
1327 | ||
1328 | if(U_FAILURE(*status)) { | |
1329 | return; | |
1330 | } | |
1331 | ||
1332 | if (bundle->fLocale!=NULL) { | |
1333 | uprv_free(bundle->fLocale); | |
1334 | } | |
1335 | ||
1336 | bundle->fLocale= (char*) uprv_malloc(sizeof(char) * (u_strlen(locale)+1)); | |
1337 | ||
1338 | if(bundle->fLocale == NULL) { | |
1339 | *status = U_MEMORY_ALLOCATION_ERROR; | |
1340 | return; | |
1341 | } | |
1342 | ||
1343 | /*u_strcpy(bundle->fLocale, locale);*/ | |
1344 | u_UCharsToChars(locale, bundle->fLocale, u_strlen(locale)+1); | |
1345 | ||
1346 | } | |
1347 | ||
729e4ab9 A |
1348 | static const char * |
1349 | getKeyString(const struct SRBRoot *bundle, int32_t key) { | |
1350 | if (key < 0) { | |
1351 | return bundle->fPoolBundleKeys + (key & 0x7fffffff); | |
1352 | } else { | |
1353 | return bundle->fKeys + key; | |
1354 | } | |
1355 | } | |
1356 | ||
1357 | const char * | |
1358 | res_getKeyString(const struct SRBRoot *bundle, const struct SResource *res, char temp[8]) { | |
1359 | if (res->fKey == -1) { | |
1360 | return NULL; | |
1361 | } | |
1362 | return getKeyString(bundle, res->fKey); | |
1363 | } | |
1364 | ||
1365 | const char * | |
1366 | bundle_getKeyBytes(struct SRBRoot *bundle, int32_t *pLength) { | |
1367 | *pLength = bundle->fKeysTop - bundle->fKeysBottom; | |
1368 | return bundle->fKeys + bundle->fKeysBottom; | |
1369 | } | |
374ca955 A |
1370 | |
1371 | int32_t | |
729e4ab9 A |
1372 | bundle_addKeyBytes(struct SRBRoot *bundle, const char *keyBytes, int32_t length, UErrorCode *status) { |
1373 | int32_t keypos; | |
b75a7d8f A |
1374 | |
1375 | if (U_FAILURE(*status)) { | |
374ca955 | 1376 | return -1; |
b75a7d8f | 1377 | } |
729e4ab9 A |
1378 | if (length < 0 || (keyBytes == NULL && length != 0)) { |
1379 | *status = U_ILLEGAL_ARGUMENT_ERROR; | |
374ca955 | 1380 | return -1; |
b75a7d8f | 1381 | } |
729e4ab9 A |
1382 | if (length == 0) { |
1383 | return bundle->fKeysTop; | |
1384 | } | |
b75a7d8f | 1385 | |
729e4ab9 A |
1386 | keypos = bundle->fKeysTop; |
1387 | bundle->fKeysTop += length; | |
1388 | if (bundle->fKeysTop >= bundle->fKeysCapacity) { | |
374ca955 A |
1389 | /* overflow - resize the keys buffer */ |
1390 | bundle->fKeysCapacity += KEY_SPACE_SIZE; | |
1391 | bundle->fKeys = uprv_realloc(bundle->fKeys, bundle->fKeysCapacity); | |
1392 | if(bundle->fKeys == NULL) { | |
1393 | *status = U_MEMORY_ALLOCATION_ERROR; | |
1394 | return -1; | |
1395 | } | |
b75a7d8f A |
1396 | } |
1397 | ||
729e4ab9 A |
1398 | uprv_memcpy(bundle->fKeys + keypos, keyBytes, length); |
1399 | ||
1400 | return keypos; | |
1401 | } | |
1402 | ||
1403 | int32_t | |
1404 | bundle_addtag(struct SRBRoot *bundle, const char *tag, UErrorCode *status) { | |
1405 | int32_t keypos; | |
1406 | ||
1407 | if (U_FAILURE(*status)) { | |
1408 | return -1; | |
1409 | } | |
1410 | ||
1411 | if (tag == NULL) { | |
1412 | /* no error: the root table and array items have no keys */ | |
1413 | return -1; | |
1414 | } | |
b75a7d8f | 1415 | |
729e4ab9 A |
1416 | keypos = bundle_addKeyBytes(bundle, tag, (int32_t)(uprv_strlen(tag) + 1), status); |
1417 | if (U_SUCCESS(*status)) { | |
1418 | ++bundle->fKeysCount; | |
1419 | } | |
b75a7d8f A |
1420 | return keypos; |
1421 | } | |
729e4ab9 A |
1422 | |
1423 | static int32_t | |
1424 | compareInt32(int32_t lPos, int32_t rPos) { | |
1425 | /* | |
1426 | * Compare possibly-negative key offsets. Don't just return lPos - rPos | |
1427 | * because that is prone to negative-integer underflows. | |
1428 | */ | |
1429 | if (lPos < rPos) { | |
1430 | return -1; | |
1431 | } else if (lPos > rPos) { | |
1432 | return 1; | |
1433 | } else { | |
1434 | return 0; | |
1435 | } | |
1436 | } | |
1437 | ||
1438 | static int32_t U_CALLCONV | |
1439 | compareKeySuffixes(const void *context, const void *l, const void *r) { | |
1440 | const struct SRBRoot *bundle=(const struct SRBRoot *)context; | |
1441 | int32_t lPos = ((const KeyMapEntry *)l)->oldpos; | |
1442 | int32_t rPos = ((const KeyMapEntry *)r)->oldpos; | |
1443 | const char *lStart = getKeyString(bundle, lPos); | |
1444 | const char *lLimit = lStart; | |
1445 | const char *rStart = getKeyString(bundle, rPos); | |
1446 | const char *rLimit = rStart; | |
1447 | int32_t diff; | |
1448 | while (*lLimit != 0) { ++lLimit; } | |
1449 | while (*rLimit != 0) { ++rLimit; } | |
1450 | /* compare keys in reverse character order */ | |
1451 | while (lStart < lLimit && rStart < rLimit) { | |
1452 | diff = (int32_t)(uint8_t)*--lLimit - (int32_t)(uint8_t)*--rLimit; | |
1453 | if (diff != 0) { | |
1454 | return diff; | |
1455 | } | |
1456 | } | |
1457 | /* sort equal suffixes by descending key length */ | |
1458 | diff = (int32_t)(rLimit - rStart) - (int32_t)(lLimit - lStart); | |
1459 | if (diff != 0) { | |
1460 | return diff; | |
1461 | } | |
1462 | /* Sort pool bundle keys first (negative oldpos), and otherwise keys in parsing order. */ | |
1463 | return compareInt32(lPos, rPos); | |
1464 | } | |
1465 | ||
1466 | static int32_t U_CALLCONV | |
1467 | compareKeyNewpos(const void *context, const void *l, const void *r) { | |
1468 | return compareInt32(((const KeyMapEntry *)l)->newpos, ((const KeyMapEntry *)r)->newpos); | |
1469 | } | |
1470 | ||
1471 | static int32_t U_CALLCONV | |
1472 | compareKeyOldpos(const void *context, const void *l, const void *r) { | |
1473 | return compareInt32(((const KeyMapEntry *)l)->oldpos, ((const KeyMapEntry *)r)->oldpos); | |
1474 | } | |
1475 | ||
1476 | void | |
1477 | bundle_compactKeys(struct SRBRoot *bundle, UErrorCode *status) { | |
1478 | KeyMapEntry *map; | |
1479 | char *keys; | |
1480 | int32_t i; | |
1481 | int32_t keysCount = bundle->fPoolBundleKeysCount + bundle->fKeysCount; | |
1482 | if (U_FAILURE(*status) || bundle->fKeysCount == 0 || bundle->fKeyMap != NULL) { | |
1483 | return; | |
1484 | } | |
1485 | map = (KeyMapEntry *)uprv_malloc(keysCount * sizeof(KeyMapEntry)); | |
1486 | if (map == NULL) { | |
1487 | *status = U_MEMORY_ALLOCATION_ERROR; | |
1488 | return; | |
1489 | } | |
1490 | keys = (char *)bundle->fPoolBundleKeys; | |
1491 | for (i = 0; i < bundle->fPoolBundleKeysCount; ++i) { | |
1492 | map[i].oldpos = | |
1493 | (int32_t)(keys - bundle->fPoolBundleKeys) | 0x80000000; /* negative oldpos */ | |
1494 | map[i].newpos = 0; | |
1495 | while (*keys != 0) { ++keys; } /* skip the key */ | |
1496 | ++keys; /* skip the NUL */ | |
1497 | } | |
1498 | keys = bundle->fKeys + bundle->fKeysBottom; | |
1499 | for (; i < keysCount; ++i) { | |
1500 | map[i].oldpos = (int32_t)(keys - bundle->fKeys); | |
1501 | map[i].newpos = 0; | |
1502 | while (*keys != 0) { ++keys; } /* skip the key */ | |
1503 | ++keys; /* skip the NUL */ | |
1504 | } | |
1505 | /* Sort the keys so that each one is immediately followed by all of its suffixes. */ | |
1506 | uprv_sortArray(map, keysCount, (int32_t)sizeof(KeyMapEntry), | |
1507 | compareKeySuffixes, bundle, FALSE, status); | |
1508 | /* | |
1509 | * Make suffixes point into earlier, longer strings that contain them | |
1510 | * and mark the old, now unused suffix bytes as deleted. | |
1511 | */ | |
1512 | if (U_SUCCESS(*status)) { | |
1513 | keys = bundle->fKeys; | |
1514 | for (i = 0; i < keysCount;) { | |
1515 | /* | |
1516 | * This key is not a suffix of the previous one; | |
1517 | * keep this one and delete the following ones that are | |
1518 | * suffixes of this one. | |
1519 | */ | |
1520 | const char *key; | |
1521 | const char *keyLimit; | |
1522 | int32_t j = i + 1; | |
1523 | map[i].newpos = map[i].oldpos; | |
1524 | if (j < keysCount && map[j].oldpos < 0) { | |
1525 | /* Key string from the pool bundle, do not delete. */ | |
1526 | i = j; | |
1527 | continue; | |
1528 | } | |
1529 | key = getKeyString(bundle, map[i].oldpos); | |
1530 | for (keyLimit = key; *keyLimit != 0; ++keyLimit) {} | |
1531 | for (; j < keysCount && map[j].oldpos >= 0; ++j) { | |
1532 | const char *k; | |
1533 | char *suffix; | |
1534 | const char *suffixLimit; | |
1535 | int32_t offset; | |
1536 | suffix = keys + map[j].oldpos; | |
1537 | for (suffixLimit = suffix; *suffixLimit != 0; ++suffixLimit) {} | |
1538 | offset = (int32_t)(keyLimit - key) - (suffixLimit - suffix); | |
1539 | if (offset < 0) { | |
1540 | break; /* suffix cannot be longer than the original */ | |
1541 | } | |
1542 | /* Is it a suffix of the earlier, longer key? */ | |
1543 | for (k = keyLimit; suffix < suffixLimit && *--k == *--suffixLimit;) {} | |
1544 | if (suffix == suffixLimit && *k == *suffixLimit) { | |
1545 | map[j].newpos = map[i].oldpos + offset; /* yes, point to the earlier key */ | |
1546 | /* mark the suffix as deleted */ | |
1547 | while (*suffix != 0) { *suffix++ = 1; } | |
1548 | *suffix = 1; | |
1549 | } else { | |
1550 | break; /* not a suffix, restart from here */ | |
1551 | } | |
1552 | } | |
1553 | i = j; | |
1554 | } | |
1555 | /* | |
1556 | * Re-sort by newpos, then modify the key characters array in-place | |
1557 | * to squeeze out unused bytes, and readjust the newpos offsets. | |
1558 | */ | |
1559 | uprv_sortArray(map, keysCount, (int32_t)sizeof(KeyMapEntry), | |
1560 | compareKeyNewpos, NULL, FALSE, status); | |
1561 | if (U_SUCCESS(*status)) { | |
1562 | int32_t oldpos, newpos, limit; | |
1563 | oldpos = newpos = bundle->fKeysBottom; | |
1564 | limit = bundle->fKeysTop; | |
1565 | /* skip key offsets that point into the pool bundle rather than this new bundle */ | |
1566 | for (i = 0; i < keysCount && map[i].newpos < 0; ++i) {} | |
1567 | if (i < keysCount) { | |
1568 | while (oldpos < limit) { | |
1569 | if (keys[oldpos] == 1) { | |
1570 | ++oldpos; /* skip unused bytes */ | |
1571 | } else { | |
1572 | /* adjust the new offsets for keys starting here */ | |
1573 | while (i < keysCount && map[i].newpos == oldpos) { | |
1574 | map[i++].newpos = newpos; | |
1575 | } | |
1576 | /* move the key characters to their new position */ | |
1577 | keys[newpos++] = keys[oldpos++]; | |
1578 | } | |
1579 | } | |
1580 | assert(i == keysCount); | |
1581 | } | |
1582 | bundle->fKeysTop = newpos; | |
1583 | /* Re-sort once more, by old offsets for binary searching. */ | |
1584 | uprv_sortArray(map, keysCount, (int32_t)sizeof(KeyMapEntry), | |
1585 | compareKeyOldpos, NULL, FALSE, status); | |
1586 | if (U_SUCCESS(*status)) { | |
1587 | /* key size reduction by limit - newpos */ | |
1588 | bundle->fKeyMap = map; | |
1589 | map = NULL; | |
1590 | } | |
1591 | } | |
1592 | } | |
1593 | uprv_free(map); | |
1594 | } | |
1595 | ||
1596 | static int32_t U_CALLCONV | |
1597 | compareStringSuffixes(const void *context, const void *l, const void *r) { | |
1598 | struct SResource *left = *((struct SResource **)l); | |
1599 | struct SResource *right = *((struct SResource **)r); | |
1600 | const UChar *lStart = left->u.fString.fChars; | |
1601 | const UChar *lLimit = lStart + left->u.fString.fLength; | |
1602 | const UChar *rStart = right->u.fString.fChars; | |
1603 | const UChar *rLimit = rStart + right->u.fString.fLength; | |
1604 | int32_t diff; | |
1605 | /* compare keys in reverse character order */ | |
1606 | while (lStart < lLimit && rStart < rLimit) { | |
1607 | diff = (int32_t)*--lLimit - (int32_t)*--rLimit; | |
1608 | if (diff != 0) { | |
1609 | return diff; | |
1610 | } | |
1611 | } | |
1612 | /* sort equal suffixes by descending string length */ | |
1613 | return right->u.fString.fLength - left->u.fString.fLength; | |
1614 | } | |
1615 | ||
1616 | static int32_t U_CALLCONV | |
1617 | compareStringLengths(const void *context, const void *l, const void *r) { | |
1618 | struct SResource *left = *((struct SResource **)l); | |
1619 | struct SResource *right = *((struct SResource **)r); | |
1620 | int32_t diff; | |
1621 | /* Make "is suffix of another string" compare greater than a non-suffix. */ | |
1622 | diff = (int)(left->u.fString.fSame != NULL) - (int)(right->u.fString.fSame != NULL); | |
1623 | if (diff != 0) { | |
1624 | return diff; | |
1625 | } | |
1626 | /* sort by ascending string length */ | |
1627 | return left->u.fString.fLength - right->u.fString.fLength; | |
1628 | } | |
1629 | ||
1630 | static int32_t | |
1631 | string_writeUTF16v2(struct SRBRoot *bundle, struct SResource *res, int32_t utf16Length) { | |
1632 | int32_t length = res->u.fString.fLength; | |
1633 | res->fRes = URES_MAKE_RESOURCE(URES_STRING_V2, utf16Length); | |
1634 | res->fWritten = TRUE; | |
1635 | switch(res->u.fString.fNumCharsForLength) { | |
1636 | case 0: | |
1637 | break; | |
1638 | case 1: | |
1639 | bundle->f16BitUnits[utf16Length++] = (uint16_t)(0xdc00 + length); | |
1640 | break; | |
1641 | case 2: | |
1642 | bundle->f16BitUnits[utf16Length] = (uint16_t)(0xdfef + (length >> 16)); | |
1643 | bundle->f16BitUnits[utf16Length + 1] = (uint16_t)length; | |
1644 | utf16Length += 2; | |
1645 | break; | |
1646 | case 3: | |
1647 | bundle->f16BitUnits[utf16Length] = 0xdfff; | |
1648 | bundle->f16BitUnits[utf16Length + 1] = (uint16_t)(length >> 16); | |
1649 | bundle->f16BitUnits[utf16Length + 2] = (uint16_t)length; | |
1650 | utf16Length += 3; | |
1651 | break; | |
1652 | default: | |
1653 | break; /* will not occur */ | |
1654 | } | |
1655 | u_memcpy(bundle->f16BitUnits + utf16Length, res->u.fString.fChars, length + 1); | |
1656 | return utf16Length + length + 1; | |
1657 | } | |
1658 | ||
1659 | static void | |
1660 | bundle_compactStrings(struct SRBRoot *bundle, UErrorCode *status) { | |
1661 | if (U_FAILURE(*status)) { | |
1662 | return; | |
1663 | } | |
1664 | switch(bundle->fStringsForm) { | |
1665 | case STRINGS_UTF16_V2: | |
1666 | if (bundle->f16BitUnitsLength > 0) { | |
1667 | struct SResource **array; | |
1668 | int32_t count = uhash_count(bundle->fStringSet); | |
1669 | int32_t i, pos; | |
1670 | /* | |
1671 | * Allocate enough space for the initial NUL and the UTF-16 v2 strings, | |
1672 | * and some extra for URES_TABLE16 and URES_ARRAY16 values. | |
1673 | * Round down to an even number. | |
1674 | */ | |
1675 | int32_t utf16Length = (bundle->f16BitUnitsLength + 20000) & ~1; | |
1676 | bundle->f16BitUnits = (UChar *)uprv_malloc(utf16Length * U_SIZEOF_UCHAR); | |
1677 | array = (struct SResource **)uprv_malloc(count * sizeof(struct SResource **)); | |
1678 | if (bundle->f16BitUnits == NULL || array == NULL) { | |
1679 | uprv_free(bundle->f16BitUnits); | |
1680 | bundle->f16BitUnits = NULL; | |
1681 | uprv_free(array); | |
1682 | *status = U_MEMORY_ALLOCATION_ERROR; | |
1683 | return; | |
1684 | } | |
1685 | bundle->f16BitUnitsCapacity = utf16Length; | |
1686 | /* insert the initial NUL */ | |
1687 | bundle->f16BitUnits[0] = 0; | |
1688 | utf16Length = 1; | |
1689 | ++bundle->f16BitUnitsLength; | |
1690 | for (pos = -1, i = 0; i < count; ++i) { | |
1691 | array[i] = (struct SResource *)uhash_nextElement(bundle->fStringSet, &pos)->key.pointer; | |
1692 | } | |
1693 | /* Sort the strings so that each one is immediately followed by all of its suffixes. */ | |
1694 | uprv_sortArray(array, count, (int32_t)sizeof(struct SResource **), | |
1695 | compareStringSuffixes, NULL, FALSE, status); | |
1696 | /* | |
1697 | * Make suffixes point into earlier, longer strings that contain them. | |
1698 | * Temporarily use fSame and fSuffixOffset for suffix strings to | |
1699 | * refer to the remaining ones. | |
1700 | */ | |
1701 | if (U_SUCCESS(*status)) { | |
1702 | for (i = 0; i < count;) { | |
1703 | /* | |
1704 | * This string is not a suffix of the previous one; | |
1705 | * write this one and subsume the following ones that are | |
1706 | * suffixes of this one. | |
1707 | */ | |
1708 | struct SResource *res = array[i]; | |
1709 | const UChar *strLimit = res->u.fString.fChars + res->u.fString.fLength; | |
1710 | int32_t j; | |
1711 | for (j = i + 1; j < count; ++j) { | |
1712 | struct SResource *suffixRes = array[j]; | |
1713 | const UChar *s; | |
1714 | const UChar *suffix = suffixRes->u.fString.fChars; | |
1715 | const UChar *suffixLimit = suffix + suffixRes->u.fString.fLength; | |
1716 | int32_t offset = res->u.fString.fLength - suffixRes->u.fString.fLength; | |
1717 | if (offset < 0) { | |
1718 | break; /* suffix cannot be longer than the original */ | |
1719 | } | |
1720 | /* Is it a suffix of the earlier, longer key? */ | |
1721 | for (s = strLimit; suffix < suffixLimit && *--s == *--suffixLimit;) {} | |
1722 | if (suffix == suffixLimit && *s == *suffixLimit) { | |
1723 | if (suffixRes->u.fString.fNumCharsForLength == 0) { | |
1724 | /* yes, point to the earlier string */ | |
1725 | suffixRes->u.fString.fSame = res; | |
1726 | suffixRes->u.fString.fSuffixOffset = offset; | |
1727 | } else { | |
1728 | /* write the suffix by itself if we need explicit length */ | |
1729 | } | |
1730 | } else { | |
1731 | break; /* not a suffix, restart from here */ | |
1732 | } | |
1733 | } | |
1734 | i = j; | |
1735 | } | |
1736 | } | |
1737 | /* | |
1738 | * Re-sort the strings by ascending length (except suffixes last) | |
1739 | * to optimize for URES_TABLE16 and URES_ARRAY16: | |
1740 | * Keep as many as possible within reach of 16-bit offsets. | |
1741 | */ | |
1742 | uprv_sortArray(array, count, (int32_t)sizeof(struct SResource **), | |
1743 | compareStringLengths, NULL, FALSE, status); | |
1744 | if (U_SUCCESS(*status)) { | |
1745 | /* Write the non-suffix strings. */ | |
1746 | for (i = 0; i < count && array[i]->u.fString.fSame == NULL; ++i) { | |
1747 | utf16Length = string_writeUTF16v2(bundle, array[i], utf16Length); | |
1748 | } | |
1749 | /* Write the suffix strings. Make each point to the real string. */ | |
1750 | for (; i < count; ++i) { | |
1751 | struct SResource *res = array[i]; | |
1752 | struct SResource *same = res->u.fString.fSame; | |
1753 | res->fRes = same->fRes + same->u.fString.fNumCharsForLength + res->u.fString.fSuffixOffset; | |
1754 | res->u.fString.fSame = NULL; | |
1755 | res->fWritten = TRUE; | |
1756 | } | |
1757 | } | |
1758 | assert(utf16Length <= bundle->f16BitUnitsLength); | |
1759 | bundle->f16BitUnitsLength = utf16Length; | |
1760 | uprv_free(array); | |
1761 | } | |
1762 | break; | |
1763 | default: | |
1764 | break; | |
1765 | } | |
1766 | } |