1 /*
2 *******************************************************************************
3 *
4 * Copyright (C) 1998-2010, International Business Machines
5 * Corporation and others. All Rights Reserved.
6 *
7 *******************************************************************************
8 *
9 * Private implementation header for C collation
10 * file name: ucol_imp.h
11 * encoding: US-ASCII
12 * tab size: 8 (not used)
13 * indentation:4
14 *
15 * created on: 2000dec11
16 * created by: Vladimir Weinstein
17 *
18 * Modification history
19 * Date Name Comments
20 * 02/16/2001 synwee Added UCOL_GETPREVCE for the use in ucoleitr
21 * 02/27/2001 synwee Added getMaxExpansion data structure in UCollator
22 * 03/02/2001 synwee Added UCOL_IMPLICIT_CE
23 * 03/12/2001 synwee Added pointer start to collIterate.
24 */
25
26 #ifndef UCOL_IMP_H
27 #define UCOL_IMP_H
28
29 #include "unicode/utypes.h"
30
31 #define UCA_DATA_TYPE "icu"
32 #define UCA_DATA_NAME "ucadata"
33 #define INVC_DATA_TYPE "icu"
34 #define INVC_DATA_NAME "invuca"
35
36 /**
37 * Convenience string denoting the Collation data tree
38 * @internal ICU 3.0
39 */
40 #define U_ICUDATA_COLL U_ICUDATA_NAME U_TREE_SEPARATOR_STRING "coll"
41
42 #if !UCONFIG_NO_COLLATION
43
44 #ifdef XP_CPLUSPLUS
45 #include "unicode/normalizer2.h"
46 #include "unicode/unistr.h"
47 #endif
48 #include "unicode/ucol.h"
49 #include "utrie.h"
50 #include "cmemory.h"
51
52 /* This is the internal header file which contains important declarations for
53 * the collation framework.
54 * Ready to use collators are stored as binary images. Both UCA and tailorings
55 * share the same binary format. Individual files (currently only UCA) have a
56 * udata header in front of the image and should be opened using udata_open.
57 * Tailoring images are currently stored inside resource bundles and are intialized
58 * through ucol_open API.
59 *
60 * The following describes the formats for collation binaries
61 * (UCA & tailorings) and for the inverse UCA table.
62 * Substructures are described in the collation design document at
63 * http://source.icu-project.org/repos/icu/icuhtml/trunk/design/collation/ICU_collation_design.htm
64 *
65 * -------------------------------------------------------------
66 *
67 * Here is the format of binary collation image.
68 *
69 * Physical order of structures:
70 * - header (UCATableHeader)
71 * - options (UColOptionSet)
72 * - expansions (CE[])
73 * - contractions (UChar[contractionSize] + CE[contractionSize])
74 * - serialized UTrie with mappings of code points to CEs
75 * - max expansion tables (CE[endExpansionCECount] + uint8_t[endExpansionCECount])
76 * - two bit sets for backward processing in strcoll (identical prefixes)
77 * and for backward CE iteration (each set is uint8_t[UCOL_UNSAFECP_TABLE_SIZE])
78 * - UCA constants (UCAConstants)
79 * - UCA contractions (UChar[contractionUCACombosSize][contractionUCACombosWidth])
80 *
81 * UCATableHeader fields:
82 *
83 * int32_t size; - image size in bytes
84 *
85 * Offsets to interesting data. All offsets are in bytes.
86 * to get the address add to the header address and cast properly.
87 * Some offsets are zero if the corresponding structures are empty.
88 *
89 * Tailoring binaries that only set options and contain no mappings etc.
90 * will have all offsets 0 except for the options and expansion offsets,
91 * which give the position and length of the options array.
92 *
93 * uint32_t options; - offset to default collator options (UColOptionSet *),
94 * a set of 32-bit values. See declaration of UColOptionSet for more details
95 *
96 * uint32_t UCAConsts; - only used (!=0) in UCA image - structure which holds values for indirect positioning and implicit ranges
97 * See declaration of UCAConstants structure. This is a set of unsigned 32-bit values used to store
98 * important constant values that are defined in the UCA and used for building and runtime.
99 *
100 * uint32_t contractionUCACombos; - only used (!=0) in UCA image - list of UCA contractions. This is a zero terminated array of UChar[contractionUCACombosWidth],
101 * containing contractions from the UCA. These are needed in the build process to copy UCA contractions
102 * in case the base contraction symbol is tailored.
103 *
104 * uint32_t magic; - must contain UCOL_HEADER_MAGIC (formatVersion 2.3)
105 *
106 * uint32_t mappingPosition; - offset to UTrie (const uint8_t *mappingPosition). This is a serialized UTrie and should be treated as such.
107 * Used as a primary lookup table for collation elements.
108 *
109 * uint32_t expansion; - offset to expansion table (uint32_t *expansion). This is an array of expansion CEs. Never 0.
110 *
111 * uint32_t contractionIndex; - offset to contraction table (UChar *contractionIndex). Used to look up contraction sequences. Contents
112 * are aligned with the contents of contractionCEs table. 0 if no contractions.
113 *
114 * uint32_t contractionCEs; - offset to resulting contraction CEs (uint32_t *contractionCEs). When a contraction is resolved in the
115 * in the contractionIndex table, the resulting index is used to look up corresponding CE in this table.
116 * 0 if no contractions.
117 * uint32_t contractionSize; - size of contraction table in elements (both Index and CEs).
118 *
119 * Tables described below are used for Boyer-Moore searching algorithm - they define the size of longest expansion
120 * and last CEs in expansions.
121 * uint32_t endExpansionCE; - offset to array of last collation element in expansion (uint32_t *).
122 * Never 0.
123 * uint32_t expansionCESize; - array of maximum expansion sizes (uint8_t *)
124 * int32_t endExpansionCECount; - size of endExpansionCE. See UCOL_GETMAXEXPANSION
125 * for the usage model
126 *
127 * These two offsets point to byte tables that are used in the backup heuristics.
128 * uint32_t unsafeCP; - hash table of unsafe code points (uint8_t *). See ucol_unsafeCP function.
129 * uint32_t contrEndCP; - hash table of final code points in contractions (uint8_t *). See ucol_contractionEndCP.
130 *
131 * int32_t contractionUCACombosSize; - number of UChar[contractionUCACombosWidth] in contractionUCACombos
132 * (formatVersion 2.3)
133 * UBool jamoSpecial; - Jamo special indicator (uint8_t). If TRUE, Jamos are special, so we cannot use simple Hangul decomposition.
134 * UBool isBigEndian; - endianness of this collation binary (formatVersion 2.3)
135 * uint8_t charSetFamily; - charset family of this collation binary (formatVersion 2.3)
136 * uint8_t contractionUCACombosWidth; - number of UChars per UCA contraction in contractionUCACombos (formatVersion 2.3)
137 *
138 * Various version fields
139 * UVersionInfo version; - version 4 uint8_t
140 * UVersionInfo UCAVersion; - version 4 uint8_t
141 * UVersionInfo UCDVersion; - version 4 uint8_t
142 * UVersionInfo formatVersion; - version of the format of the collation binary
143 * same formatVersion as in ucadata.icu's UDataInfo header
144 * (formatVersion 2.3)
145 *
146 * uint32_t offset to the reordering code to lead CE byte remapping table
147 * uint32_t offset to the lead CE byte to reordering code mapping table
148 *
149 * uint8_t reserved[76]; - currently unused
150 *
151 * -------------------------------------------------------------
152 *
153 * Inverse UCA is used for constructing collators from rules. It is always an individual file
154 * and always has a UDataInfo header.
155 * here is the structure:
156 *
157 * uint32_t byteSize; - size of inverse UCA image in bytes
158 * uint32_t tableSize; - length of inverse table (number of uint32_t[3] rows)
159 * uint32_t contsSize; - size of continuation table (number of UChars in table)
160 *
161 * uint32_t table; - offset to inverse table (uint32_t *)
162 * Inverse table contains of rows of 3 uint32_t values. First two values are CE and a possible continuation
163 * the third value is either a code unit (if there is only one code unit for element) or an index to continuation
164 * (number of code units combined with an index).
165 * table. If more than one codepoint have the same CE, continuation table contains code units separated by FFFF and final
166 * code unit sequence for a CE is terminated by FFFE.
167 * uint32_t conts; - offset to continuation table (uint16_t *). Contains code units that transform to a same CE.
168 *
169 * UVersionInfo UCAVersion; - version of the UCA, read from file 4 uint8_t
170 * uint8_t padding[8]; - padding 8 uint8_t
171 * Header is followed by the table and continuation table.
172 */
173
174 /* let us know whether reserved fields are reset to zero or junked */
175 #define UCOL_HEADER_MAGIC 0x20030618
176
177 /* UDataInfo for UCA mapping table */
178 /* dataFormat="UCol" */
179 #define UCA_DATA_FORMAT_0 ((uint8_t)0x55)
180 #define UCA_DATA_FORMAT_1 ((uint8_t)0x43)
181 #define UCA_DATA_FORMAT_2 ((uint8_t)0x6f)
182 #define UCA_DATA_FORMAT_3 ((uint8_t)0x6c)
183
184 #define UCA_FORMAT_VERSION_0 ((uint8_t)3)
185 #define UCA_FORMAT_VERSION_1 ((uint8_t)0)
186 #define UCA_FORMAT_VERSION_2 ((uint8_t)0)
187 #define UCA_FORMAT_VERSION_3 ((uint8_t)0)
188
189 /* UDataInfo for inverse UCA table */
190 /* dataFormat="InvC" */
191 #define INVUCA_DATA_FORMAT_0 ((uint8_t)0x49)
192 #define INVUCA_DATA_FORMAT_1 ((uint8_t)0x6E)
193 #define INVUCA_DATA_FORMAT_2 ((uint8_t)0x76)
194 #define INVUCA_DATA_FORMAT_3 ((uint8_t)0x43)
195
196 #define INVUCA_FORMAT_VERSION_0 ((uint8_t)2)
197 #define INVUCA_FORMAT_VERSION_1 ((uint8_t)1)
198 #define INVUCA_FORMAT_VERSION_2 ((uint8_t)0)
199 #define INVUCA_FORMAT_VERSION_3 ((uint8_t)0)
200
201 /* This is the size of the stack allocated buffer for sortkey generation and similar operations */
202 /* if it is too small, heap allocation will occur.*/
203 /* you can change this value if you need memory - it will affect the performance, though, since we're going to malloc */
204 #define UCOL_MAX_BUFFER 128
205 #define UCOL_PRIMARY_MAX_BUFFER 8*UCOL_MAX_BUFFER
206 #define UCOL_SECONDARY_MAX_BUFFER UCOL_MAX_BUFFER
207 #define UCOL_TERTIARY_MAX_BUFFER UCOL_MAX_BUFFER
208 #define UCOL_CASE_MAX_BUFFER UCOL_MAX_BUFFER/4
209 #define UCOL_QUAD_MAX_BUFFER 2*UCOL_MAX_BUFFER
210
211 #define UCOL_NORMALIZATION_GROWTH 2
212 #define UCOL_NORMALIZATION_MAX_BUFFER UCOL_MAX_BUFFER*UCOL_NORMALIZATION_GROWTH
213
214 /* This writable buffer is used if we encounter Thai and need to reorder the string on the fly */
215 /* Sometimes we already have a writable buffer (like in case of normalized strings). */
216 /*
217 you can change this value to any value >= 4 if you need memory -
218 it will affect the performance, though, since we're going to malloc.
219 Note 3 is the minimum value for Thai collation and 4 is the
220 minimum number for special Jamo
221 */
222 #define UCOL_WRITABLE_BUFFER_SIZE 256
223
224 /* This is the size of the buffer for expansion CE's */
225 /* In reality we should not have to deal with expm sequences longer then 16 */
226 /* you can change this value if you need memory */
227 /* WARNING THIS BUFFER DOES HAVE MALLOC FALLBACK. If you make it too small, you'll get into performance trouble */
228 /* Reasonable small value is around 10, if you don't do Arabic or other funky collations that have long expansion sequence */
229 /* This is the longest expansion sequence we can handle without bombing out */
230 #define UCOL_EXPAND_CE_BUFFER_SIZE 64
231
232 /* This is the size to increase the buffer for expansion CE's */
233 #define UCOL_EXPAND_CE_BUFFER_EXTEND_SIZE 64
234
235
236 /* Unsafe UChar hash table table size. */
237 /* size is 32 bytes for 1 bit for each latin 1 char + some power of two for */
238 /* hashing the rest of the chars. Size in bytes */
239 #define UCOL_UNSAFECP_TABLE_SIZE 1056
240 /* mask value down to "some power of two"-1 */
241 /* number of bits, not num of bytes. */
242 #define UCOL_UNSAFECP_TABLE_MASK 0x1fff
243
244
245 /* flags bits for collIterate.flags */
246 /* */
247 /* NORM - set for incremental normalize of source string */
248 #define UCOL_ITER_NORM 1
249
250 #define UCOL_ITER_HASLEN 2
251
252 /* UCOL_ITER_INNORMBUF - set if the "pos" is in */
253 /* the writable side buffer, handling */
254 /* incrementally normalized characters. */
255 #define UCOL_ITER_INNORMBUF 4
256
257 /* UCOL_ITER_ALLOCATED - set if this iterator has */
258 /* malloced storage to expand a buffer. */
259 #define UCOL_ITER_ALLOCATED 8
260 /* UCOL_HIRAGANA_Q - note if the codepoint was hiragana */
261 #define UCOL_HIRAGANA_Q 16
262 /* UCOL_WAS_HIRAGANA - set to TRUE if there was a Hiragana */
263 /* otherwise set to false */
264 #define UCOL_WAS_HIRAGANA 32
265 /* UCOL_USE_ITERATOR - set this if collIterate uses a */
266 /* character iterator instead of simply accessing string */
267 /* by index */
268 #define UCOL_USE_ITERATOR 64
269
270 #define UCOL_FORCE_HAN_IMPLICIT 128
271
272 #define NFC_ZERO_CC_BLOCK_LIMIT_ 0x300
273
274 #ifdef XP_CPLUSPLUS
275
276 U_NAMESPACE_BEGIN
277
278 typedef struct collIterate : public UMemory {
279 const UChar *string; /* Original string */
280 /* UChar *start; Pointer to the start of the source string. Either points to string
281 or to writableBuffer */
282 const UChar *endp; /* string end ptr. Is undefined for null terminated strings */
283 const UChar *pos; /* This is position in the string. Can be to original or writable buf */
284
285 uint32_t *toReturn; /* This is the CE from CEs buffer that should be returned */
286 uint32_t *CEpos; /* This is the position to which we have stored processed CEs */
287
288 int32_t *offsetReturn; /* This is the offset to return, if non-NULL */
289 int32_t *offsetStore; /* This is the pointer for storing offsets */
290 int32_t offsetRepeatCount; /* Repeat stored offset if non-zero */
291 int32_t offsetRepeatValue; /* offset value to repeat */
292
293 UnicodeString writableBuffer;
294 const UChar *fcdPosition; /* Position in the original string to continue FCD check from. */
295 const UCollator *coll;
296 const Normalizer2 *nfd;
297 uint8_t flags;
298 uint8_t origFlags;
299 uint32_t *extendCEs; /* This is use if CEs is not big enough */
300 int32_t extendCEsSize; /* Holds the size of the dynamic CEs buffer */
301 uint32_t CEs[UCOL_EXPAND_CE_BUFFER_SIZE]; /* This is where we store CEs */
302
303 int32_t *offsetBuffer; /* A dynamic buffer to hold offsets */
304 int32_t offsetBufferSize; /* The size of the offset buffer */
305
306 UCharIterator *iterator;
307 /*int32_t iteratorIndex;*/
308
309 // The offsetBuffer should probably be a UVector32, but helper functions
310 // are an improvement over duplicated code.
311 void appendOffset(int32_t offset, UErrorCode &errorCode);
312 } collIterate;
313
314 U_NAMESPACE_END
315
316 #else
317
318 typedef struct collIterate collIterate;
319
320 #endif
321
322 #define paddedsize(something) ((something)+((((something)%4)!=0)?(4-(something)%4):0))
323 #define headersize (paddedsize(sizeof(UCATableHeader))+paddedsize(sizeof(UColOptionSet)))
324
325 /*
326 struct used internally in getSpecial*CE.
327 data similar to collIterate.
328 */
329 struct collIterateState {
330 const UChar *pos; /* This is position in the string. Can be to original or writable buf */
331 const UChar *returnPos;
332 const UChar *fcdPosition; /* Position in the original string to continue FCD check from. */
333 const UChar *bufferaddress; /* address of the normalization buffer */
334 int32_t buffersize;
335 uint8_t flags;
336 uint8_t origFlags;
337 uint32_t iteratorIndex;
338 int32_t iteratorMove;
339 };
340
341 U_CAPI void U_EXPORT2
342 uprv_init_collIterate(const UCollator *collator,
343 const UChar *sourceString, int32_t sourceLen,
344 U_NAMESPACE_QUALIFIER collIterate *s, UErrorCode *status);
345
346 /* Internal functions for C test code. */
347 U_CAPI U_NAMESPACE_QUALIFIER collIterate * U_EXPORT2
348 uprv_new_collIterate(UErrorCode *status);
349
350 U_CAPI void U_EXPORT2
351 uprv_delete_collIterate(U_NAMESPACE_QUALIFIER collIterate *s);
352
353 /* @return s->pos == s->endp */
354 U_CAPI UBool U_EXPORT2
355 uprv_collIterateAtEnd(U_NAMESPACE_QUALIFIER collIterate *s);
356
357 #ifdef XP_CPLUSPLUS
358
359 U_NAMESPACE_BEGIN
360
361 struct UCollationPCE;
362 typedef struct UCollationPCE UCollationPCE;
363
364 U_NAMESPACE_END
365
366 struct UCollationElements : public U_NAMESPACE_QUALIFIER UMemory
367 {
368 /**
369 * Struct wrapper for source data
370 */
371 U_NAMESPACE_QUALIFIER collIterate iteratordata_;
372 /**
373 * Indicates if this data has been reset.
374 */
375 UBool reset_;
376 /**
377 * Indicates if the data should be deleted.
378 */
379 UBool isWritable;
380
381 /**
382 * Data for getNextProcessed, getPreviousProcessed.
383 */
384 U_NAMESPACE_QUALIFIER UCollationPCE *pce;
385 };
386
387 #else
388 /*opaque type*/
389 struct UCollationElements;
390 #endif
391
392 U_CAPI void U_EXPORT2
393 uprv_init_pce(const struct UCollationElements *elems);
394
395 #define UCOL_LEVELTERMINATOR 1
396
397 /* mask off anything but primary order */
398 #define UCOL_PRIMARYORDERMASK 0xffff0000
399 /* mask off anything but secondary order */
400 #define UCOL_SECONDARYORDERMASK 0x0000ff00
401 /* mask off anything but tertiary order */
402 #define UCOL_TERTIARYORDERMASK 0x000000ff
403 /* primary order shift */
404 #define UCOL_PRIMARYORDERSHIFT 16
405 /* secondary order shift */
406 #define UCOL_SECONDARYORDERSHIFT 8
407
408 #define UCOL_BYTE_SIZE_MASK 0xFF
409
410 #define UCOL_CASE_BYTE_START 0x80
411 #define UCOL_CASE_SHIFT_START 7
412
413 #define UCOL_IGNORABLE 0
414
415 /* get weights from a CE */
416 #define UCOL_PRIMARYORDER(order) (((order) & UCOL_PRIMARYORDERMASK)>> UCOL_PRIMARYORDERSHIFT)
417 #define UCOL_SECONDARYORDER(order) (((order) & UCOL_SECONDARYORDERMASK)>> UCOL_SECONDARYORDERSHIFT)
418 #define UCOL_TERTIARYORDER(order) ((order) & UCOL_TERTIARYORDERMASK)
419
420 /**
421 * Determine if a character is a Thai vowel (which sorts after
422 * its base consonant).
423 */
424 #define UCOL_ISTHAIPREVOWEL(ch) ((((uint32_t)(ch) - 0xe40) <= (0xe44 - 0xe40)) || \
425 (((uint32_t)(ch) - 0xec0) <= (0xec4 - 0xec0)))
426
427 /**
428 * Determine if a character is a Thai base consonant
429 */
430 #define UCOL_ISTHAIBASECONSONANT(ch) ((uint32_t)(ch) - 0xe01) <= (0xe2e - 0xe01)
431
432 #define UCOL_ISJAMO(ch) ((((uint32_t)(ch) - 0x1100) <= (0x1112 - 0x1100)) || \
433 (((uint32_t)(ch) - 0x1161) <= (0x1175 - 0x1161)) || \
434 (((uint32_t)(ch) - 0x11A8) <= (0x11C2 - 0x11A8)))
435
436 /* Han character ranges */
437 #define UCOL_FIRST_HAN 0x4E00
438 #define UCOL_LAST_HAN 0x9FFF
439 #define UCOL_FIRST_HAN_A 0x3400
440 #define UCOL_LAST_HAN_A 0x4DBF
441 #define UCOL_FIRST_HAN_COMPAT 0xFAE0
442 #define UCOL_LAST_HAN_COMPAT 0xFA2F
443
444 /* Han extension B is in plane 2 */
445 #define UCOL_FIRST_HAN_B 0x20000
446 #define UCOL_LAST_HAN_B 0x2A6DF
447
448 /* Hangul range */
449 #define UCOL_FIRST_HANGUL 0xAC00
450 #define UCOL_LAST_HANGUL 0xD7AF
451
452 /* Jamo ranges */
453 #define UCOL_FIRST_L_JAMO 0x1100
454 #define UCOL_FIRST_V_JAMO 0x1161
455 #define UCOL_FIRST_T_JAMO 0x11A8
456 #define UCOL_LAST_T_JAMO 0x11F9
457
458
459 #if 0
460 /* initializes collIterate structure */
461 /* made as macro to speed up things */
462 #define init_collIterate(collator, sourceString, sourceLen, s) { \
463 (s)->start = (s)->string = (s)->pos = (UChar *)(sourceString); \
464 (s)->endp = (sourceLen) == -1 ? NULL :(UChar *)(sourceString)+(sourceLen); \
465 (s)->CEpos = (s)->toReturn = (s)->CEs; \
466 (s)->isThai = TRUE; \
467 (s)->writableBuffer = (s)->stackWritableBuffer; \
468 (s)->writableBufSize = UCOL_WRITABLE_BUFFER_SIZE; \
469 (s)->coll = (collator); \
470 (s)->fcdPosition = 0; \
471 (s)->flags = 0; \
472 if(((collator)->normalizationMode == UCOL_ON)) (s)->flags |= UCOL_ITER_NORM; \
473 }
474 #endif
475
476
477
478 /*
479 * Macro to get the maximum size of an expansion ending with the argument ce.
480 * Used in the Boyer Moore algorithm.
481 * Note for tailoring, the UCA maxexpansion table has been merged.
482 * Hence we only have to search the tailored collator only.
483 * @param coll const UCollator pointer
484 * @param order last collation element of the expansion sequence
485 * @param result size of the longest expansion with argument collation element
486 * as the last element
487 */
488 #define UCOL_GETMAXEXPANSION(coll, order, result) { \
489 const uint32_t *start; \
490 const uint32_t *limit; \
491 const uint32_t *mid; \
492 start = (coll)->endExpansionCE; \
493 limit = (coll)->lastEndExpansionCE; \
494 while (start < limit - 1) { \
495 mid = start + ((limit - start) >> 1); \
496 if ((order) <= *mid) { \
497 limit = mid; \
498 } \
499 else { \
500 start = mid; \
501 } \
502 } \
503 if (*start == order) { \
504 result = *((coll)->expansionCESize + (start - (coll)->endExpansionCE)); \
505 } \
506 else if (*limit == order) { \
507 result = *(coll->expansionCESize + (limit - coll->endExpansionCE)); \
508 } \
509 else if ((order & 0xFFFF) == 0x00C0) { \
510 result = 2; \
511 } \
512 else { \
513 result = 1; \
514 } \
515 }
516
517 U_CFUNC
518 uint32_t ucol_prv_getSpecialCE(const UCollator *coll, UChar ch, uint32_t CE,
519 U_NAMESPACE_QUALIFIER collIterate *source, UErrorCode *status);
520
521 U_CFUNC
522 uint32_t ucol_prv_getSpecialPrevCE(const UCollator *coll, UChar ch, uint32_t CE,
523 U_NAMESPACE_QUALIFIER collIterate *source, UErrorCode *status);
524 U_CAPI uint32_t U_EXPORT2 ucol_getNextCE(const UCollator *coll,
525 U_NAMESPACE_QUALIFIER collIterate *collationSource, UErrorCode *status);
526 U_CFUNC uint32_t U_EXPORT2 ucol_getPrevCE(const UCollator *coll,
527 U_NAMESPACE_QUALIFIER collIterate *collationSource,
528 UErrorCode *status);
529 /* function used by C++ getCollationKey to prevent restarting the calculation */
530 U_CFUNC int32_t
531 ucol_getSortKeyWithAllocation(const UCollator *coll,
532 const UChar *source, int32_t sourceLength,
533 uint8_t **pResult,
534 UErrorCode *pErrorCode);
535
536 /* get some memory */
537 void *ucol_getABuffer(const UCollator *coll, uint32_t size);
538
539 /* worker function for generating sortkeys */
540 U_CFUNC
541 int32_t U_CALLCONV
542 ucol_calcSortKey(const UCollator *coll,
543 const UChar *source,
544 int32_t sourceLength,
545 uint8_t **result,
546 uint32_t resultLength,
547 UBool allocatePrimary,
548 UErrorCode *status);
549
550 U_CFUNC
551 int32_t U_CALLCONV
552 ucol_calcSortKeySimpleTertiary(const UCollator *coll,
553 const UChar *source,
554 int32_t sourceLength,
555 uint8_t **result,
556 uint32_t resultLength,
557 UBool allocatePrimary,
558 UErrorCode *status);
559
560 U_CFUNC
561 int32_t
562 ucol_getSortKeySize(const UCollator *coll, U_NAMESPACE_QUALIFIER collIterate *s,
563 int32_t currentSize, UColAttributeValue strength,
564 int32_t len);
565 /**
566 * Makes a copy of the Collator's rule data. The format is
567 * that of .col files.
568 *
569 * @param length returns the length of the data, in bytes.
570 * @param status the error status
571 * @return memory, owned by the caller, of size 'length' bytes.
572 * @internal INTERNAL USE ONLY
573 */
574 U_CFUNC uint8_t* U_EXPORT2
575 ucol_cloneRuleData(const UCollator *coll, int32_t *length, UErrorCode *status);
576
577 /**
578 * Used to set requested and valid locales on a collator returned by the collator
579 * service.
580 */
581 U_CFUNC void U_EXPORT2
582 ucol_setReqValidLocales(UCollator *coll, char *requestedLocaleToAdopt, char *validLocaleToAdopt, char *actualLocaleToAdopt);
583
584 #define UCOL_SPECIAL_FLAG 0xF0000000
585 #define UCOL_TAG_SHIFT 24
586 #define UCOL_TAG_MASK 0x0F000000
587 #define INIT_EXP_TABLE_SIZE 1024
588 #define UCOL_NOT_FOUND 0xF0000000
589 #define UCOL_EXPANSION 0xF1000000
590 #define UCOL_CONTRACTION 0xF2000000
591 #define UCOL_THAI 0xF3000000
592 #define UCOL_UNMARKED 0x03
593 #define UCOL_NEW_TERTIARYORDERMASK 0x0000003f
594
595 /* Bit mask for primary collation strength. */
596 #define UCOL_PRIMARYMASK 0xFFFF0000
597
598 /* Bit mask for secondary collation strength. */
599 #define UCOL_SECONDARYMASK 0x0000FF00
600
601 /* Bit mask for tertiary collation strength. */
602 #define UCOL_TERTIARYMASK 0x000000FF
603
604 /**
605 * Internal.
606 * This indicates the last element in a UCollationElements has been consumed.
607 * Compare with the UCOL_NULLORDER, UCOL_NULLORDER is returned if error occurs.
608 */
609 #define UCOL_NO_MORE_CES 0x00010101
610 #define UCOL_NO_MORE_CES_PRIMARY 0x00010000
611 #define UCOL_NO_MORE_CES_SECONDARY 0x00000100
612 #define UCOL_NO_MORE_CES_TERTIARY 0x00000001
613
614 #define isSpecial(CE) ((((CE)&UCOL_SPECIAL_FLAG)>>28)==0xF)
615
616 #define UCOL_UPPER_CASE 0x80
617 #define UCOL_MIXED_CASE 0x40
618 #define UCOL_LOWER_CASE 0x00
619
620 #define UCOL_CONTINUATION_MARKER 0xC0
621 #define UCOL_REMOVE_CONTINUATION 0xFFFFFF3F
622
623 #define isContinuation(CE) (((CE) & UCOL_CONTINUATION_MARKER) == UCOL_CONTINUATION_MARKER)
624 #define isFlagged(CE) (((CE) & 0x80) == 0x80)
625 #define isLongPrimary(CE) (((CE) & 0xC0) == 0xC0)
626
627 #define getCETag(CE) (((CE)&UCOL_TAG_MASK)>>UCOL_TAG_SHIFT)
628 #define isContraction(CE) (isSpecial((CE)) && (getCETag((CE)) == CONTRACTION_TAG))
629 #define isPrefix(CE) (isSpecial((CE)) && (getCETag((CE)) == SPEC_PROC_TAG))
630 #define constructContractCE(tag, CE) (UCOL_SPECIAL_FLAG | ((tag)<<UCOL_TAG_SHIFT) | ((CE)&0xFFFFFF))
631 #define constructSpecProcCE(CE) (UCOL_SPECIAL_FLAG | (SPEC_PROC_TAG<<UCOL_TAG_SHIFT) | ((CE)&0xFFFFFF))
632 #define getContractOffset(CE) ((CE)&0xFFFFFF)
633 #define getExpansionOffset(CE) (((CE)&0x00FFFFF0)>>4)
634 #define getExpansionCount(CE) ((CE)&0xF)
635 #define isCEIgnorable(CE) (((CE) & 0xFFFFFFBF) == 0)
636
637 /* StringSearch internal use */
638 #define inNormBuf(coleiter) ((coleiter)->iteratordata_.flags & UCOL_ITER_INNORMBUF)
639 #define isFCDPointerNull(coleiter) ((coleiter)->iteratordata_.fcdPosition == NULL)
640 #define hasExpansion(coleiter) ((coleiter)->iteratordata_.CEpos != (coleiter)->iteratordata_.CEs)
641 #define getExpansionPrefix(coleiter) ((coleiter)->iteratordata_.toReturn - (coleiter)->iteratordata_.CEs)
642 #define setExpansionPrefix(coleiter, offset) ((coleiter)->iteratordata_.CEs + offset)
643 #define getExpansionSuffix(coleiter) ((coleiter)->iteratordata_.CEpos - (coleiter)->iteratordata_.toReturn)
644 #define setExpansionSuffix(coleiter, offset) ((coleiter)->iteratordata_.toReturn = (coleiter)->iteratordata_.CEpos - leftoverces)
645
646 /* This is an enum that lists magic special byte values from the fractional UCA.
647 * See also http://site.icu-project.org/design/collation/bytes */
648 /* TODO: all the #defines that refer to special byte values from the UCA should be changed to point here */
649
650 enum {
651 UCOL_BYTE_ZERO = 0x00,
652 UCOL_BYTE_LEVEL_SEPARATOR = 0x01,
653 UCOL_BYTE_SORTKEY_GLUE = 0x02,
654 UCOL_BYTE_SHIFT_PREFIX = 0x03,
655 UCOL_BYTE_UNSHIFTED_MIN = UCOL_BYTE_SHIFT_PREFIX,
656 UCOL_BYTE_FIRST_TAILORED = 0x04,
657 UCOL_BYTE_COMMON = 0x05,
658 UCOL_BYTE_FIRST_UCA = UCOL_BYTE_COMMON,
659 /* TODO: Make the following values dynamic since they change with almost every UCA version. */
660 UCOL_CODAN_PLACEHOLDER = 0x12,
661 UCOL_BYTE_FIRST_NON_LATIN_PRIMARY = 0x5B,
662 UCOL_BYTE_UNSHIFTED_MAX = 0xFF
663 };
664
665 #if 0
666 #define UCOL_RESET_TOP_VALUE 0x9F000303
667 #define UCOL_FIRST_PRIMARY_IGNORABLE 0x00008705
668 #define UCOL_LAST_PRIMARY_IGNORABLE 0x0000DD05
669 #define UCOL_LAST_PRIMARY_IGNORABLE_CONT 0x000051C0
670 #define UCOL_FIRST_SECONDARY_IGNORABLE 0x00000000
671 #define UCOL_LAST_SECONDARY_IGNORABLE 0x00000500
672 #define UCOL_FIRST_TERTIARY_IGNORABLE 0x00000000
673 #define UCOL_LAST_TERTIARY_IGNORABLE 0x00000000
674 #define UCOL_FIRST_VARIABLE 0x05070505
675 #define UCOL_LAST_VARIABLE 0x179B0505
676 #define UCOL_FIRST_NON_VARIABLE 0x1A200505
677 #define UCOL_LAST_NON_VARIABLE 0x7B41058F
678
679 #define UCOL_NEXT_TOP_VALUE 0xE8960303
680 #define UCOL_NEXT_FIRST_PRIMARY_IGNORABLE 0x00008905
681 #define UCOL_NEXT_LAST_PRIMARY_IGNORABLE 0x03000303
682 #define UCOL_NEXT_FIRST_SECONDARY_IGNORABLE 0x00008705
683 #define UCOL_NEXT_LAST_SECONDARY_IGNORABLE 0x00000500
684 #define UCOL_NEXT_FIRST_TERTIARY_IGNORABLE 0x00000000
685 #define UCOL_NEXT_LAST_TERTIARY_IGNORABLE 0x00000000
686 #define UCOL_NEXT_FIRST_VARIABLE 0x05090505
687 #define UCOL_NEXT_LAST_VARIABLE 0x1A200505
688
689 #define PRIMARY_IMPLICIT_MIN 0xE8000000
690 #define PRIMARY_IMPLICIT_MAX 0xF0000000
691 #endif
692
693 /* These constants can be changed - sortkey size is affected by them */
694 #define UCOL_PROPORTION2 0.5
695 #define UCOL_PROPORTION3 0.667
696
697 /* These values come from the UCA */
698 #define UCOL_COMMON_BOT2 UCOL_BYTE_COMMON
699 #define UCOL_COMMON_TOP2 0x86u
700 #define UCOL_TOTAL2 (UCOL_COMMON_TOP2-UCOL_COMMON_BOT2-1)
701
702 #define UCOL_FLAG_BIT_MASK_CASE_SW_OFF 0x80
703 #define UCOL_FLAG_BIT_MASK_CASE_SW_ON 0x40
704 #define UCOL_COMMON_TOP3_CASE_SW_OFF 0x85
705 #define UCOL_COMMON_TOP3_CASE_SW_LOWER 0x45
706 #define UCOL_COMMON_TOP3_CASE_SW_UPPER 0xC5
707
708 /* These values come from the UCA */
709 #define UCOL_COMMON_BOT3 0x05
710
711 #define UCOL_COMMON_BOTTOM3_CASE_SW_UPPER 0x86;
712 #define UCOL_COMMON_BOTTOM3_CASE_SW_LOWER UCOL_COMMON_BOT3;
713
714 #define UCOL_TOP_COUNT2 (UCOL_PROPORTION2*UCOL_TOTAL2)
715 #define UCOL_BOT_COUNT2 (UCOL_TOTAL2-UCOL_TOP_COUNT2)
716
717
718 #define UCOL_COMMON2 UCOL_COMMON_BOT2
719 #define UCOL_COMMON3_UPPERFIRST 0xC5
720 #define UCOL_COMMON3_NORMAL UCOL_COMMON_BOT3
721
722 #define UCOL_COMMON4 0xFF
723
724 /* constants for case level/case first handling */
725 /* used to instantiate UCollators fields in ucol_updateInternalState */
726 #define UCOL_CASE_SWITCH 0xC0
727 #define UCOL_NO_CASE_SWITCH 0x00
728
729 #define UCOL_REMOVE_CASE 0x3F
730 #define UCOL_KEEP_CASE 0xFF
731
732 #define UCOL_CASE_BIT_MASK 0xC0
733
734 #define UCOL_TERT_CASE_MASK 0xFF
735
736 #define UCOL_ENDOFLATINONERANGE 0xFF
737 #define UCOL_LATINONETABLELEN (UCOL_ENDOFLATINONERANGE+50)
738 #define UCOL_BAIL_OUT_CE 0xFF000000
739
740
741 typedef enum {
742 NOT_FOUND_TAG = 0,
743 EXPANSION_TAG = 1, /* This code point results in an expansion */
744 CONTRACTION_TAG = 2, /* Start of a contraction */
745 THAI_TAG = 3, /* Thai character - do the reordering */
746 CHARSET_TAG = 4, /* Charset processing, not yet implemented */
747 SURROGATE_TAG = 5, /* Lead surrogate that is tailored and doesn't start a contraction */
748 HANGUL_SYLLABLE_TAG = 6, /* AC00-D7AF*/
749 LEAD_SURROGATE_TAG = 7, /* D800-DBFF*/
750 TRAIL_SURROGATE_TAG = 8, /* DC00-DFFF*/
751 CJK_IMPLICIT_TAG = 9, /* 0x3400-0x4DB5, 0x4E00-0x9FA5, 0xF900-0xFA2D*/
752 IMPLICIT_TAG = 10,
753 SPEC_PROC_TAG = 11,
754 /* ICU 2.1 */
755 LONG_PRIMARY_TAG = 12, /* This is a three byte primary with starting secondaries and tertiaries */
756 /* It fits in a single 32 bit CE and is used instead of expansion to save */
757 /* space without affecting the performance (hopefully) */
758
759 DIGIT_TAG = 13, /* COllate Digits As Numbers (CODAN) implementation */
760
761 CE_TAGS_COUNT
762 } UColCETags;
763
764 /*
765 *****************************************************************************************
766 * set to zero
767 * NON_CHARACTER FDD0 - FDEF, FFFE, FFFF, 1FFFE, 1FFFF, 2FFFE, 2FFFF,...e.g. **FFFE, **FFFF
768 ******************************************************************************************
769 */
770
771 typedef struct {
772 uint32_t variableTopValue;
773 /*UColAttributeValue*/ int32_t frenchCollation;
774 /*UColAttributeValue*/ int32_t alternateHandling; /* attribute for handling variable elements*/
775 /*UColAttributeValue*/ int32_t caseFirst; /* who goes first, lower case or uppercase */
776 /*UColAttributeValue*/ int32_t caseLevel; /* do we have an extra case level */
777 /*UColAttributeValue*/ int32_t normalizationMode; /* attribute for normalization */
778 /*UColAttributeValue*/ int32_t strength; /* attribute for strength */
779 /*UColAttributeValue*/ int32_t hiraganaQ; /* attribute for special Hiragana */
780 /*UColAttributeValue*/ int32_t numericCollation; /* attribute for numeric collation */
781 uint32_t reserved[15]; /* for future use */
782 } UColOptionSet;
783
784 typedef struct {
785 uint32_t UCA_FIRST_TERTIARY_IGNORABLE[2]; /*0x00000000*/
786 uint32_t UCA_LAST_TERTIARY_IGNORABLE[2]; /*0x00000000*/
787 uint32_t UCA_FIRST_PRIMARY_IGNORABLE[2]; /*0x00008705*/
788 uint32_t UCA_FIRST_SECONDARY_IGNORABLE[2]; /*0x00000000*/
789 uint32_t UCA_LAST_SECONDARY_IGNORABLE[2]; /*0x00000500*/
790 uint32_t UCA_LAST_PRIMARY_IGNORABLE[2]; /*0x0000DD05*/
791 uint32_t UCA_FIRST_VARIABLE[2]; /*0x05070505*/
792 uint32_t UCA_LAST_VARIABLE[2]; /*0x13CF0505*/
793 uint32_t UCA_FIRST_NON_VARIABLE[2]; /*0x16200505*/
794 uint32_t UCA_LAST_NON_VARIABLE[2]; /*0x767C0505*/
795 uint32_t UCA_RESET_TOP_VALUE[2]; /*0x9F000303*/
796 uint32_t UCA_FIRST_IMPLICIT[2];
797 uint32_t UCA_LAST_IMPLICIT[2];
798 uint32_t UCA_FIRST_TRAILING[2];
799 uint32_t UCA_LAST_TRAILING[2];
800
801 #if 0
802 uint32_t UCA_NEXT_TOP_VALUE[2]; /*0xE8960303*/
803 uint32_t UCA_NEXT_FIRST_PRIMARY_IGNORABLE; /*0x00008905*/
804 uint32_t UCA_NEXT_LAST_PRIMARY_IGNORABLE; /*0x03000303*/
805 uint32_t UCA_NEXT_FIRST_SECONDARY_IGNORABLE; /*0x00008705*/
806 uint32_t UCA_NEXT_LAST_SECONDARY_IGNORABLE; /*0x00000500*/
807 uint32_t UCA_NEXT_FIRST_TERTIARY_IGNORABLE; /*0x00000000*/
808 uint32_t UCA_NEXT_LAST_TERTIARY_IGNORABLE; /*0x00000000*/
809 uint32_t UCA_NEXT_FIRST_VARIABLE; /*0x05090505*/
810 uint32_t UCA_NEXT_LAST_VARIABLE; /*0x16200505*/
811 #endif
812
813 uint32_t UCA_PRIMARY_TOP_MIN;
814 uint32_t UCA_PRIMARY_IMPLICIT_MIN; /*0xE8000000*/
815 uint32_t UCA_PRIMARY_IMPLICIT_MAX; /*0xF0000000*/
816 uint32_t UCA_PRIMARY_TRAILING_MIN; /*0xE8000000*/
817 uint32_t UCA_PRIMARY_TRAILING_MAX; /*0xF0000000*/
818 uint32_t UCA_PRIMARY_SPECIAL_MIN; /*0xE8000000*/
819 uint32_t UCA_PRIMARY_SPECIAL_MAX; /*0xF0000000*/
820 } UCAConstants;
821
822 typedef struct {
823 int32_t size;
824 /* all the offsets are in bytes */
825 /* to get the address add to the header address and cast properly */
826 uint32_t options; /* these are the default options for the collator */
827 uint32_t UCAConsts; /* structure which holds values for indirect positioning and implicit ranges */
828 uint32_t contractionUCACombos; /* this one is needed only for UCA, to copy the appropriate contractions */
829 uint32_t magic; /* magic number - lets us know whether reserved data is reset or junked */
830 uint32_t mappingPosition; /* const uint8_t *mappingPosition; */
831 uint32_t expansion; /* uint32_t *expansion; */
832 uint32_t contractionIndex; /* UChar *contractionIndex; */
833 uint32_t contractionCEs; /* uint32_t *contractionCEs; */
834 uint32_t contractionSize; /* needed for various closures */
835 /*int32_t latinOneMapping;*/ /* this is now handled in the trie itself *//* fast track to latin1 chars */
836
837 uint32_t endExpansionCE; /* array of last collation element in
838 expansion */
839 uint32_t expansionCESize; /* array of maximum expansion size
840 corresponding to the expansion
841 collation elements with last element
842 in endExpansionCE*/
843 int32_t endExpansionCECount; /* size of endExpansionCE */
844 uint32_t unsafeCP; /* hash table of unsafe code points */
845 uint32_t contrEndCP; /* hash table of final code points */
846 /* in contractions. */
847
848 int32_t contractionUCACombosSize; /* number of UCA contraction items. */
849 /*Length is contractionUCACombosSize*contractionUCACombosWidth*sizeof(UChar) */
850 UBool jamoSpecial; /* is jamoSpecial */
851 UBool isBigEndian; /* is this data big endian? from the UDataInfo header*/
852 uint8_t charSetFamily; /* what is the charset family of this data from the UDataInfo header*/
853 uint8_t contractionUCACombosWidth; /* width of UCA combos field */
854 UVersionInfo version;
855 UVersionInfo UCAVersion; /* version of the UCA, read from file */
856 UVersionInfo UCDVersion; /* UCD version, obtained by u_getUnicodeVersion */
857 UVersionInfo formatVersion; /* format version from the UDataInfo header */
858 uint32_t scriptToLeadByte; /* offset to script to lead collation byte mapping data */
859 uint32_t leadByteToScript; /* offset to lead collation byte to script mapping data */
860 uint8_t reserved[76]; /* for future use */
861 } UCATableHeader;
862
863 #define U_UNKNOWN_STATE 0
864 #define U_COLLATOR_STATE 0x01
865 #define U_STATE_LIMIT 0x02
866
867 /* This is the first structure in a state */
868 /* it should be machine independent */
869 typedef struct {
870 /* this structure is supposed to be readable on all the platforms.*/
871 /* first 2 fields hold the size of the structure in a platform independent way */
872 uint8_t sizeLo;
873 uint8_t sizeHi;
874 /* identifying the writing platform */
875 uint8_t isBigEndian;
876 /* see U_CHARSET_FAMILY values in utypes.h */
877 uint8_t charsetFamily;
878 /* version of ICU this state structure comes from */
879 uint8_t icuVersion[4];
880 /* What is the data following this state */
881 uint8_t type;
882 /* more stuff to come, keep it on 16 byte boundary */
883 uint8_t reserved[7];
884 } UStateStruct;
885
886 /* This structure follows UStatusStruct */
887 /* and contains data specific for the collators */
888 /* Endianess needs to be decided before accessing this structure */
889 /* However, it's size IS endianess independent */
890 typedef struct {
891 /* size of this structure */
892 uint8_t sizeLo;
893 uint8_t sizeHi;
894 /* This state is followed by the frozen tailoring */
895 uint8_t containsTailoring;
896 /* This state is followed by the frozen UCA */
897 uint8_t containsUCA;
898 /* Version info - the same one */
899 uint8_t versionInfo[4];
900
901 /* for charset CEs */
902 uint8_t charsetName[32];
903 /* this is the resolved locale name*/
904 uint8_t locale[32];
905
906 /* Attributes. Open ended */
907 /* all the following will be moved to uint32_t because of portability */
908 /* variable top value */
909 uint32_t variableTopValue;
910 /* attribute for handling variable elements*/
911 uint32_t /*UColAttributeValue*/ alternateHandling;
912 /* how to handle secondary weights */
913 uint32_t /*UColAttributeValue*/ frenchCollation;
914 /* who goes first, lower case or uppercase */
915 uint32_t /*UColAttributeValue*/ caseFirst;
916 /* do we have an extra case level */
917 uint32_t /*UColAttributeValue*/ caseLevel;
918 /* attribute for normalization */
919 uint32_t /*UColAttributeValue*/ normalizationMode;
920 /* attribute for strength */
921 uint32_t /*UColAttributeValue*/ strength;
922 /* to be immediately 16 byte aligned */
923 uint8_t reserved[12];
924 } UColStateStruct;
925
926 #define UCOL_INV_SIZEMASK 0xFFF00000
927 #define UCOL_INV_OFFSETMASK 0x000FFFFF
928 #define UCOL_INV_SHIFTVALUE 20
929
930 U_CDECL_BEGIN
931
932 typedef struct {
933 uint32_t byteSize;
934 uint32_t tableSize;
935 uint32_t contsSize;
936 uint32_t table;
937 uint32_t conts;
938 UVersionInfo UCAVersion; /* version of the UCA, read from file */
939 uint8_t padding[8];
940 } InverseUCATableHeader;
941
942 typedef int32_t U_CALLCONV
943 SortKeyGenerator(const UCollator *coll,
944 const UChar *source,
945 int32_t sourceLength,
946 uint8_t **result,
947 uint32_t resultLength,
948 UBool allocatePrimary,
949 UErrorCode *status);
950
951 typedef void U_CALLCONV
952 ResourceCleaner(UCollator *coll);
953
954
955 struct UCollator {
956 UColOptionSet *options;
957 SortKeyGenerator *sortKeyGen;
958 uint32_t *latinOneCEs;
959 char* actualLocale;
960 char* validLocale;
961 char* requestedLocale;
962 const UChar *rules;
963 const UChar *ucaRules;
964 const UCollator *UCA;
965 const UCATableHeader *image;
966 UTrie mapping;
967 const uint32_t *latinOneMapping;
968 const uint32_t *expansion;
969 const UChar *contractionIndex;
970 const uint32_t *contractionCEs;
971
972 const uint32_t *endExpansionCE; /* array of last ces in an expansion ce.
973 corresponds to expansionCESize */
974 const uint32_t *lastEndExpansionCE;/* pointer to the last element in endExpansionCE */
975 const uint8_t *expansionCESize; /* array of the maximum size of a
976 expansion ce with the last ce
977 corresponding to endExpansionCE,
978 terminated with a null */
979 const uint8_t *unsafeCP; /* unsafe code points hashtable */
980 const uint8_t *contrEndCP; /* Contraction ending chars hash table */
981 UChar minUnsafeCP; /* Smallest unsafe Code Point. */
982 UChar minContrEndCP; /* Smallest code point at end of a contraction */
983
984 int32_t rulesLength;
985 int32_t latinOneTableLen;
986
987 uint32_t variableTopValue;
988 UColAttributeValue frenchCollation;
989 UColAttributeValue alternateHandling; /* attribute for handling variable elements*/
990 UColAttributeValue caseFirst; /* who goes first, lower case or uppercase */
991 UColAttributeValue caseLevel; /* do we have an extra case level */
992 UColAttributeValue normalizationMode; /* attribute for normalization */
993 UColAttributeValue strength; /* attribute for strength */
994 UColAttributeValue hiraganaQ; /* attribute for Hiragana */
995 UColAttributeValue numericCollation;
996 UBool variableTopValueisDefault;
997 UBool frenchCollationisDefault;
998 UBool alternateHandlingisDefault; /* attribute for handling variable elements*/
999 UBool caseFirstisDefault; /* who goes first, lower case or uppercase */
1000 UBool caseLevelisDefault; /* do we have an extra case level */
1001 UBool normalizationModeisDefault; /* attribute for normalization */
1002 UBool strengthisDefault; /* attribute for strength */
1003 UBool hiraganaQisDefault; /* attribute for Hiragana */
1004 UBool numericCollationisDefault;
1005 UBool hasRealData; /* some collators have only options, like French, no rules */
1006 /* to speed up things, we use the UCA image, but we don't want it */
1007 /* to run around */
1008
1009 UBool freeOnClose;
1010 UBool freeOptionsOnClose;
1011 UBool freeRulesOnClose;
1012 UBool freeImageOnClose;
1013
1014 UBool latinOneUse;
1015 UBool latinOneRegenTable;
1016 UBool latinOneFailed;
1017
1018 int8_t tertiaryAddition; /* when switching case, we need to add or subtract different values */
1019 uint8_t caseSwitch;
1020 uint8_t tertiaryCommon;
1021 uint8_t tertiaryMask;
1022 uint8_t tertiaryTop; /* Upper range when compressing */
1023 uint8_t tertiaryBottom; /* Upper range when compressing */
1024 uint8_t tertiaryTopCount;
1025 uint8_t tertiaryBottomCount;
1026
1027 UVersionInfo dataVersion; /* Data info of UCA table */
1028 int32_t* reorderCodes;
1029 int32_t reorderCodesLength;
1030 uint8_t* leadBytePermutationTable;
1031 };
1032
1033 U_CDECL_END
1034
1035 /* various internal functions */
1036
1037 /* do not close UCA returned by ucol_initUCA! */
1038 U_CFUNC
1039 UCollator* ucol_initUCA(UErrorCode *status);
1040
1041 U_CFUNC
1042 UCollator* ucol_initCollator(const UCATableHeader *image, UCollator *fillIn, const UCollator *UCA, UErrorCode *status);
1043
1044 U_CFUNC
1045 void ucol_setOptionsFromHeader(UCollator* result, UColOptionSet * opts, UErrorCode *status);
1046
1047 U_CFUNC
1048 UCollator* ucol_open_internal(const char* loc, UErrorCode* status);
1049
1050 #if 0
1051 U_CFUNC
1052 void ucol_putOptionsToHeader(UCollator* result, UColOptionSet * opts, UErrorCode *status);
1053 #endif
1054
1055 U_CFUNC
1056 void ucol_updateInternalState(UCollator *coll, UErrorCode *status);
1057
1058 U_CFUNC uint32_t U_EXPORT2 ucol_getFirstCE(const UCollator *coll, UChar u, UErrorCode *status);
1059 U_CAPI UBool U_EXPORT2 ucol_isTailored(const UCollator *coll, const UChar u, UErrorCode *status);
1060
1061 U_CAPI const InverseUCATableHeader* U_EXPORT2 ucol_initInverseUCA(UErrorCode *status);
1062
1063 U_CAPI void U_EXPORT2
1064 uprv_uca_initImplicitConstants(UErrorCode *status);
1065
1066 U_CAPI uint32_t U_EXPORT2
1067 uprv_uca_getImplicitFromRaw(UChar32 cp);
1068
1069 /*U_CFUNC uint32_t U_EXPORT2
1070 uprv_uca_getImplicitPrimary(UChar32 cp);*/
1071
1072 U_CAPI UChar32 U_EXPORT2
1073 uprv_uca_getRawFromImplicit(uint32_t implicit);
1074
1075 U_CAPI UChar32 U_EXPORT2
1076 uprv_uca_getRawFromCodePoint(UChar32 i);
1077
1078 U_CAPI UChar32 U_EXPORT2
1079 uprv_uca_getCodePointFromRaw(UChar32 i);
1080
1081 typedef const UChar* GetCollationRulesFunction(void* context, const char* locale, const char* type, int32_t* pLength, UErrorCode* status);
1082
1083 U_CAPI UCollator* U_EXPORT2
1084 ucol_openRulesForImport( const UChar *rules,
1085 int32_t rulesLength,
1086 UColAttributeValue normalizationMode,
1087 UCollationStrength strength,
1088 UParseError *parseError,
1089 GetCollationRulesFunction importFunc,
1090 void* context,
1091 UErrorCode *status);
1092
1093
1094 U_CAPI void U_EXPORT2 ucol_buildPermutationTable(UCollator *coll, UErrorCode *status);
1095
1096
1097 #ifdef XP_CPLUSPLUS
1098 /*
1099 * Test whether a character is potentially "unsafe" for use as a collation
1100 * starting point. Unsafe chars are those with combining class != 0 plus
1101 * those that are the 2nd thru nth character in a contraction sequence.
1102 *
1103 * Function is in header file because it's used in both collation and string search,
1104 * and needs to be inline for performance.
1105 */
ucol_unsafeCP(UChar c,const UCollator * coll)1106 static inline UBool ucol_unsafeCP(UChar c, const UCollator *coll) {
1107 int32_t hash;
1108 uint8_t htbyte;
1109
1110 if (c < coll->minUnsafeCP) {
1111 return FALSE;
1112 }
1113
1114 hash = c;
1115 if (hash >= UCOL_UNSAFECP_TABLE_SIZE*8) {
1116 if(UTF_IS_SURROGATE(c)) {
1117 /* Lead or trail surrogate */
1118 /* These are always considered unsafe. */
1119 return TRUE;
1120 }
1121 hash = (hash & UCOL_UNSAFECP_TABLE_MASK) + 256;
1122 }
1123 htbyte = coll->unsafeCP[hash>>3];
1124 return ((htbyte >> (hash & 7)) & 1);
1125 }
1126 #endif /* XP_CPLUSPLUS */
1127
1128 /* The offsetBuffer in collIterate might need to be freed to avoid memory leaks. */
1129 void ucol_freeOffsetBuffer(U_NAMESPACE_QUALIFIER collIterate *s);
1130
1131 #endif /* #if !UCONFIG_NO_COLLATION */
1132
1133 #endif
1134