The Tor Browser / file comparison

comparison: intl/icu/source/i18n/unicode/ucol.h

intl/icu/source/i18n/unicode/ucol.h

branch
TOR_BUG_3246
changeset 4
fc2d59ddac77
equal deleted inserted replaced
-1:000000000000 0:f1096e8174d9
1 /*
2 *******************************************************************************
3 * Copyright (c) 1996-2013, International Business Machines Corporation and others.
4 * All Rights Reserved.
5 *******************************************************************************
6 */
7
8 #ifndef UCOL_H
9 #define UCOL_H
10
11 #include "unicode/utypes.h"
12
13 #if !UCONFIG_NO_COLLATION
14
15 #include "unicode/unorm.h"
16 #include "unicode/localpointer.h"
17 #include "unicode/parseerr.h"
18 #include "unicode/uloc.h"
19 #include "unicode/uset.h"
20 #include "unicode/uscript.h"
21
22 /**
23 * \file
24 * \brief C API: Collator
25 *
26 * <h2> Collator C API </h2>
27 *
28 * The C API for Collator performs locale-sensitive
29 * string comparison. You use this service to build
30 * searching and sorting routines for natural language text.
31 * <em>Important: </em>The ICU collation service has been reimplemented
32 * in order to achieve better performance and UCA compliance.
33 * For details, see the
34 * <a href="http://source.icu-project.org/repos/icu/icuhtml/trunk/design/collation/ICU_collation_design.htm">
35 * collation design document</a>.
36 * <p>
37 * For more information about the collation service see
38 * <a href="http://icu-project.org/userguide/Collate_Intro.html">the users guide</a>.
39 * <p>
40 * Collation service provides correct sorting orders for most locales supported in ICU.
41 * If specific data for a locale is not available, the orders eventually falls back
42 * to the <a href="http://www.unicode.org/unicode/reports/tr10/">UCA sort order</a>.
43 * <p>
44 * Sort ordering may be customized by providing your own set of rules. For more on
45 * this subject see the
46 * <a href="http://icu-project.org/userguide/Collate_Customization.html">
47 * Collation customization</a> section of the users guide.
48 * <p>
49 * @see UCollationResult
50 * @see UNormalizationMode
51 * @see UCollationStrength
52 * @see UCollationElements
53 */
54
55 /** A collator.
56 * For usage in C programs.
57 */
58 struct UCollator;
59 /** structure representing a collator object instance
60 * @stable ICU 2.0
61 */
62 typedef struct UCollator UCollator;
63
64
65 /**
66 * UCOL_LESS is returned if source string is compared to be less than target
67 * string in the ucol_strcoll() method.
68 * UCOL_EQUAL is returned if source string is compared to be equal to target
69 * string in the ucol_strcoll() method.
70 * UCOL_GREATER is returned if source string is compared to be greater than
71 * target string in the ucol_strcoll() method.
72 * @see ucol_strcoll()
73 * <p>
74 * Possible values for a comparison result
75 * @stable ICU 2.0
76 */
77 typedef enum {
78 /** string a == string b */
79 UCOL_EQUAL = 0,
80 /** string a > string b */
81 UCOL_GREATER = 1,
82 /** string a < string b */
83 UCOL_LESS = -1
84 } UCollationResult ;
85
86
87 /** Enum containing attribute values for controling collation behavior.
88 * Here are all the allowable values. Not every attribute can take every value. The only
89 * universal value is UCOL_DEFAULT, which resets the attribute value to the predefined
90 * value for that locale
91 * @stable ICU 2.0
92 */
93 typedef enum {
94 /** accepted by most attributes */
95 UCOL_DEFAULT = -1,
96
97 /** Primary collation strength */
98 UCOL_PRIMARY = 0,
99 /** Secondary collation strength */
100 UCOL_SECONDARY = 1,
101 /** Tertiary collation strength */
102 UCOL_TERTIARY = 2,
103 /** Default collation strength */
104 UCOL_DEFAULT_STRENGTH = UCOL_TERTIARY,
105 UCOL_CE_STRENGTH_LIMIT,
106 /** Quaternary collation strength */
107 UCOL_QUATERNARY=3,
108 /** Identical collation strength */
109 UCOL_IDENTICAL=15,
110 UCOL_STRENGTH_LIMIT,
111
112 /** Turn the feature off - works for UCOL_FRENCH_COLLATION,
113 UCOL_CASE_LEVEL, UCOL_HIRAGANA_QUATERNARY_MODE
114 & UCOL_DECOMPOSITION_MODE*/
115 UCOL_OFF = 16,
116 /** Turn the feature on - works for UCOL_FRENCH_COLLATION,
117 UCOL_CASE_LEVEL, UCOL_HIRAGANA_QUATERNARY_MODE
118 & UCOL_DECOMPOSITION_MODE*/
119 UCOL_ON = 17,
120
121 /** Valid for UCOL_ALTERNATE_HANDLING. Alternate handling will be shifted */
122 UCOL_SHIFTED = 20,
123 /** Valid for UCOL_ALTERNATE_HANDLING. Alternate handling will be non ignorable */
124 UCOL_NON_IGNORABLE = 21,
125
126 /** Valid for UCOL_CASE_FIRST -
127 lower case sorts before upper case */
128 UCOL_LOWER_FIRST = 24,
129 /** upper case sorts before lower case */
130 UCOL_UPPER_FIRST = 25,
131
132 UCOL_ATTRIBUTE_VALUE_COUNT
133
134 } UColAttributeValue;
135
136 /**
137 * Enum containing the codes for reordering segments of the collation table that are not script
138 * codes. These reordering codes are to be used in conjunction with the script codes.
139 * @see ucol_getReorderCodes
140 * @see ucol_setReorderCodes
141 * @see ucol_getEquivalentReorderCodes
142 * @see UScriptCode
143 * @stable ICU 4.8
144 */
145 typedef enum {
146 /**
147 * A special reordering code that is used to specify the default
148 * reordering codes for a locale.
149 * @stable ICU 4.8
150 */
151 UCOL_REORDER_CODE_DEFAULT = -1,
152 /**
153 * A special reordering code that is used to specify no reordering codes.
154 * @stable ICU 4.8
155 */
156 UCOL_REORDER_CODE_NONE = USCRIPT_UNKNOWN,
157 /**
158 * A special reordering code that is used to specify all other codes used for
159 * reordering except for the codes lised as UColReorderCode values and those
160 * listed explicitly in a reordering.
161 * @stable ICU 4.8
162 */
163 UCOL_REORDER_CODE_OTHERS = USCRIPT_UNKNOWN,
164 /**
165 * Characters with the space property.
166 * This is equivalent to the rule value "space".
167 * @stable ICU 4.8
168 */
169 UCOL_REORDER_CODE_SPACE = 0x1000,
170 /**
171 * The first entry in the enumeration of reordering groups. This is intended for use in
172 * range checking and enumeration of the reorder codes.
173 * @stable ICU 4.8
174 */
175 UCOL_REORDER_CODE_FIRST = UCOL_REORDER_CODE_SPACE,
176 /**
177 * Characters with the punctuation property.
178 * This is equivalent to the rule value "punct".
179 * @stable ICU 4.8
180 */
181 UCOL_REORDER_CODE_PUNCTUATION = 0x1001,
182 /**
183 * Characters with the symbol property.
184 * This is equivalent to the rule value "symbol".
185 * @stable ICU 4.8
186 */
187 UCOL_REORDER_CODE_SYMBOL = 0x1002,
188 /**
189 * Characters with the currency property.
190 * This is equivalent to the rule value "currency".
191 * @stable ICU 4.8
192 */
193 UCOL_REORDER_CODE_CURRENCY = 0x1003,
194 /**
195 * Characters with the digit property.
196 * This is equivalent to the rule value "digit".
197 * @stable ICU 4.8
198 */
199 UCOL_REORDER_CODE_DIGIT = 0x1004,
200 /**
201 * The limit of the reorder codes. This is intended for use in range checking
202 * and enumeration of the reorder codes.
203 * @stable ICU 4.8
204 */
205 UCOL_REORDER_CODE_LIMIT = 0x1005
206 } UColReorderCode;
207
208 /**
209 * Base letter represents a primary difference. Set comparison
210 * level to UCOL_PRIMARY to ignore secondary and tertiary differences.
211 * Use this to set the strength of a Collator object.
212 * Example of primary difference, "abc" &lt; "abd"
213 *
214 * Diacritical differences on the same base letter represent a secondary
215 * difference. Set comparison level to UCOL_SECONDARY to ignore tertiary
216 * differences. Use this to set the strength of a Collator object.
217 * Example of secondary difference, "&auml;" >> "a".
218 *
219 * Uppercase and lowercase versions of the same character represents a
220 * tertiary difference. Set comparison level to UCOL_TERTIARY to include
221 * all comparison differences. Use this to set the strength of a Collator
222 * object.
223 * Example of tertiary difference, "abc" &lt;&lt;&lt; "ABC".
224 *
225 * Two characters are considered "identical" when they have the same
226 * unicode spellings. UCOL_IDENTICAL.
227 * For example, "&auml;" == "&auml;".
228 *
229 * UCollationStrength is also used to determine the strength of sort keys
230 * generated from UCollator objects
231 * These values can be now found in the UColAttributeValue enum.
232 * @stable ICU 2.0
233 **/
234 typedef UColAttributeValue UCollationStrength;
235
236 /** Attributes that collation service understands. All the attributes can take UCOL_DEFAULT
237 * value, as well as the values specific to each one.
238 * @stable ICU 2.0
239 */
240 typedef enum {
241 /** Attribute for direction of secondary weights - used in Canadian French.
242 * Acceptable values are UCOL_ON, which results in secondary weights
243 * being considered backwards and UCOL_OFF which treats secondary
244 * weights in the order they appear.
245 * @stable ICU 2.0
246 */
247 UCOL_FRENCH_COLLATION,
248 /** Attribute for handling variable elements.
249 * Acceptable values are UCOL_NON_IGNORABLE (default)
250 * which treats all the codepoints with non-ignorable
251 * primary weights in the same way,
252 * and UCOL_SHIFTED which causes codepoints with primary
253 * weights that are equal or below the variable top value
254 * to be ignored on primary level and moved to the quaternary
255 * level.
256 * @stable ICU 2.0
257 */
258 UCOL_ALTERNATE_HANDLING,
259 /** Controls the ordering of upper and lower case letters.
260 * Acceptable values are UCOL_OFF (default), which orders
261 * upper and lower case letters in accordance to their tertiary
262 * weights, UCOL_UPPER_FIRST which forces upper case letters to
263 * sort before lower case letters, and UCOL_LOWER_FIRST which does
264 * the opposite.
265 * @stable ICU 2.0
266 */
267 UCOL_CASE_FIRST,
268 /** Controls whether an extra case level (positioned before the third
269 * level) is generated or not. Acceptable values are UCOL_OFF (default),
270 * when case level is not generated, and UCOL_ON which causes the case
271 * level to be generated. Contents of the case level are affected by
272 * the value of UCOL_CASE_FIRST attribute. A simple way to ignore
273 * accent differences in a string is to set the strength to UCOL_PRIMARY
274 * and enable case level.
275 * @stable ICU 2.0
276 */
277 UCOL_CASE_LEVEL,
278 /** Controls whether the normalization check and necessary normalizations
279 * are performed. When set to UCOL_OFF (default) no normalization check
280 * is performed. The correctness of the result is guaranteed only if the
281 * input data is in so-called FCD form (see users manual for more info).
282 * When set to UCOL_ON, an incremental check is performed to see whether
283 * the input data is in the FCD form. If the data is not in the FCD form,
284 * incremental NFD normalization is performed.
285 * @stable ICU 2.0
286 */
287 UCOL_NORMALIZATION_MODE,
288 /** An alias for UCOL_NORMALIZATION_MODE attribute.
289 * @stable ICU 2.0
290 */
291 UCOL_DECOMPOSITION_MODE = UCOL_NORMALIZATION_MODE,
292 /** The strength attribute. Can be either UCOL_PRIMARY, UCOL_SECONDARY,
293 * UCOL_TERTIARY, UCOL_QUATERNARY or UCOL_IDENTICAL. The usual strength
294 * for most locales (except Japanese) is tertiary. Quaternary strength
295 * is useful when combined with shifted setting for alternate handling
296 * attribute and for JIS x 4061 collation, when it is used to distinguish
297 * between Katakana and Hiragana (this is achieved by setting the
298 * UCOL_HIRAGANA_QUATERNARY mode to on. Otherwise, quaternary level
299 * is affected only by the number of non ignorable code points in
300 * the string. Identical strength is rarely useful, as it amounts
301 * to codepoints of the NFD form of the string.
302 * @stable ICU 2.0
303 */
304 UCOL_STRENGTH,
305 #ifndef U_HIDE_DEPRECATED_API
306 /** When turned on, this attribute positions Hiragana before all
307 * non-ignorables on quaternary level This is a sneaky way to produce JIS
308 * sort order.
309 *
310 * This attribute is an implementation detail of the CLDR Japanese tailoring.
311 * The implementation might change to use a different mechanism
312 * to achieve the same Japanese sort order.
313 * Since ICU 50, this attribute is not settable any more via API functions.
314 * @deprecated ICU 50 Implementation detail, cannot be set via API, might be removed from implementation.
315 */
316 UCOL_HIRAGANA_QUATERNARY_MODE = UCOL_STRENGTH + 1,
317 #endif /* U_HIDE_DEPRECATED_API */
318 /** When turned on, this attribute generates a collation key
319 * for the numeric value of substrings of digits.
320 * This is a way to get '100' to sort AFTER '2'. Note that the longest
321 * digit substring that can be treated as a single collation element is
322 * 254 digits (not counting leading zeros). If a digit substring is
323 * longer than that, the digits beyond the limit will be treated as a
324 * separate digit substring associated with a separate collation element.
325 * @stable ICU 2.8
326 */
327 UCOL_NUMERIC_COLLATION = UCOL_STRENGTH + 2,
328 /**
329 * The number of UColAttribute constants.
330 * @stable ICU 2.0
331 */
332 UCOL_ATTRIBUTE_COUNT
333 } UColAttribute;
334
335 /** Options for retrieving the rule string
336 * @stable ICU 2.0
337 */
338 typedef enum {
339 /**
340 * Retrieves the tailoring rules only.
341 * Same as calling the version of getRules() without UColRuleOption.
342 * @stable ICU 2.0
343 */
344 UCOL_TAILORING_ONLY,
345 /**
346 * Retrieves the "UCA rules" concatenated with the tailoring rules.
347 * The "UCA rules" are an <i>approximation</i> of the root collator's sort order.
348 * They are almost never used or useful at runtime and can be removed from the data.
349 * See http://userguide.icu-project.org/collation/customization#TOC-Building-on-Existing-Locales
350 * @stable ICU 2.0
351 */
352 UCOL_FULL_RULES
353 } UColRuleOption ;
354
355 /**
356 * Open a UCollator for comparing strings.
357 * The UCollator pointer is used in all the calls to the Collation
358 * service. After finished, collator must be disposed of by calling
359 * {@link #ucol_close }.
360 * @param loc The locale containing the required collation rules.
361 * Special values for locales can be passed in -
362 * if NULL is passed for the locale, the default locale
363 * collation rules will be used. If empty string ("") or
364 * "root" are passed, UCA rules will be used.
365 * @param status A pointer to an UErrorCode to receive any errors
366 * @return A pointer to a UCollator, or 0 if an error occurred.
367 * @see ucol_openRules
368 * @see ucol_safeClone
369 * @see ucol_close
370 * @stable ICU 2.0
371 */
372 U_STABLE UCollator* U_EXPORT2
373 ucol_open(const char *loc, UErrorCode *status);
374
375 /**
376 * Produce an UCollator instance according to the rules supplied.
377 * The rules are used to change the default ordering, defined in the
378 * UCA in a process called tailoring. The resulting UCollator pointer
379 * can be used in the same way as the one obtained by {@link #ucol_strcoll }.
380 * @param rules A string describing the collation rules. For the syntax
381 * of the rules please see users guide.
382 * @param rulesLength The length of rules, or -1 if null-terminated.
383 * @param normalizationMode The normalization mode: One of
384 * UCOL_OFF (expect the text to not need normalization),
385 * UCOL_ON (normalize), or
386 * UCOL_DEFAULT (set the mode according to the rules)
387 * @param strength The default collation strength; one of UCOL_PRIMARY, UCOL_SECONDARY,
388 * UCOL_TERTIARY, UCOL_IDENTICAL,UCOL_DEFAULT_STRENGTH - can be also set in the rules.
389 * @param parseError A pointer to UParseError to recieve information about errors
390 * occurred during parsing. This argument can currently be set
391 * to NULL, but at users own risk. Please provide a real structure.
392 * @param status A pointer to an UErrorCode to receive any errors
393 * @return A pointer to a UCollator. It is not guaranteed that NULL be returned in case
394 * of error - please use status argument to check for errors.
395 * @see ucol_open
396 * @see ucol_safeClone
397 * @see ucol_close
398 * @stable ICU 2.0
399 */
400 U_STABLE UCollator* U_EXPORT2
401 ucol_openRules( const UChar *rules,
402 int32_t rulesLength,
403 UColAttributeValue normalizationMode,
404 UCollationStrength strength,
405 UParseError *parseError,
406 UErrorCode *status);
407
408 /**
409 * Open a collator defined by a short form string.
410 * The structure and the syntax of the string is defined in the "Naming collators"
411 * section of the users guide:
412 * http://icu-project.org/userguide/Collate_Concepts.html#Naming_Collators
413 * Attributes are overriden by the subsequent attributes. So, for "S2_S3", final
414 * strength will be 3. 3066bis locale overrides individual locale parts.
415 * The call to this function is equivalent to a call to ucol_open, followed by a
416 * series of calls to ucol_setAttribute and ucol_setVariableTop.
417 * @param definition A short string containing a locale and a set of attributes.
418 * Attributes not explicitly mentioned are left at the default
419 * state for a locale.
420 * @param parseError if not NULL, structure that will get filled with error's pre
421 * and post context in case of error.
422 * @param forceDefaults if FALSE, the settings that are the same as the collator
423 * default settings will not be applied (for example, setting
424 * French secondary on a French collator would not be executed).
425 * If TRUE, all the settings will be applied regardless of the
426 * collator default value. If the definition
427 * strings are to be cached, should be set to FALSE.
428 * @param status Error code. Apart from regular error conditions connected to
429 * instantiating collators (like out of memory or similar), this
430 * API will return an error if an invalid attribute or attribute/value
431 * combination is specified.
432 * @return A pointer to a UCollator or 0 if an error occured (including an
433 * invalid attribute).
434 * @see ucol_open
435 * @see ucol_setAttribute
436 * @see ucol_setVariableTop
437 * @see ucol_getShortDefinitionString
438 * @see ucol_normalizeShortDefinitionString
439 * @stable ICU 3.0
440 *
441 */
442 U_STABLE UCollator* U_EXPORT2
443 ucol_openFromShortString( const char *definition,
444 UBool forceDefaults,
445 UParseError *parseError,
446 UErrorCode *status);
447
448 #ifndef U_HIDE_DEPRECATED_API
449 /**
450 * Get a set containing the contractions defined by the collator. The set includes
451 * both the UCA contractions and the contractions defined by the collator. This set
452 * will contain only strings. If a tailoring explicitly suppresses contractions from
453 * the UCA (like Russian), removed contractions will not be in the resulting set.
454 * @param coll collator
455 * @param conts the set to hold the result. It gets emptied before
456 * contractions are added.
457 * @param status to hold the error code
458 * @return the size of the contraction set
459 *
460 * @deprecated ICU 3.4, use ucol_getContractionsAndExpansions instead
461 */
462 U_DEPRECATED int32_t U_EXPORT2
463 ucol_getContractions( const UCollator *coll,
464 USet *conts,
465 UErrorCode *status);
466 #endif /* U_HIDE_DEPRECATED_API */
467
468 /**
469 * Get a set containing the expansions defined by the collator. The set includes
470 * both the UCA expansions and the expansions defined by the tailoring
471 * @param coll collator
472 * @param contractions if not NULL, the set to hold the contractions
473 * @param expansions if not NULL, the set to hold the expansions
474 * @param addPrefixes add the prefix contextual elements to contractions
475 * @param status to hold the error code
476 *
477 * @stable ICU 3.4
478 */
479 U_STABLE void U_EXPORT2
480 ucol_getContractionsAndExpansions( const UCollator *coll,
481 USet *contractions, USet *expansions,
482 UBool addPrefixes, UErrorCode *status);
483
484 /**
485 * Close a UCollator.
486 * Once closed, a UCollator should not be used. Every open collator should
487 * be closed. Otherwise, a memory leak will result.
488 * @param coll The UCollator to close.
489 * @see ucol_open
490 * @see ucol_openRules
491 * @see ucol_safeClone
492 * @stable ICU 2.0
493 */
494 U_STABLE void U_EXPORT2
495 ucol_close(UCollator *coll);
496
497 #if U_SHOW_CPLUSPLUS_API
498
499 U_NAMESPACE_BEGIN
500
501 /**
502 * \class LocalUCollatorPointer
503 * "Smart pointer" class, closes a UCollator via ucol_close().
504 * For most methods see the LocalPointerBase base class.
505 *
506 * @see LocalPointerBase
507 * @see LocalPointer
508 * @stable ICU 4.4
509 */
510 U_DEFINE_LOCAL_OPEN_POINTER(LocalUCollatorPointer, UCollator, ucol_close);
511
512 U_NAMESPACE_END
513
514 #endif
515
516 /**
517 * Compare two strings.
518 * The strings will be compared using the options already specified.
519 * @param coll The UCollator containing the comparison rules.
520 * @param source The source string.
521 * @param sourceLength The length of source, or -1 if null-terminated.
522 * @param target The target string.
523 * @param targetLength The length of target, or -1 if null-terminated.
524 * @return The result of comparing the strings; one of UCOL_EQUAL,
525 * UCOL_GREATER, UCOL_LESS
526 * @see ucol_greater
527 * @see ucol_greaterOrEqual
528 * @see ucol_equal
529 * @stable ICU 2.0
530 */
531 U_STABLE UCollationResult U_EXPORT2
532 ucol_strcoll( const UCollator *coll,
533 const UChar *source,
534 int32_t sourceLength,
535 const UChar *target,
536 int32_t targetLength);
537
538 /**
539 * Compare two strings in UTF-8.
540 * The strings will be compared using the options already specified.
541 * Note: When input string contains malformed a UTF-8 byte sequence,
542 * this function treats these bytes as REPLACEMENT CHARACTER (U+FFFD).
543 * @param coll The UCollator containing the comparison rules.
544 * @param source The source UTF-8 string.
545 * @param sourceLength The length of source, or -1 if null-terminated.
546 * @param target The target UTF-8 string.
547 * @param targetLength The length of target, or -1 if null-terminated.
548 * @param status A pointer to an UErrorCode to receive any errors
549 * @return The result of comparing the strings; one of UCOL_EQUAL,
550 * UCOL_GREATER, UCOL_LESS
551 * @see ucol_greater
552 * @see ucol_greaterOrEqual
553 * @see ucol_equal
554 * @stable ICU 50
555 */
556 U_STABLE UCollationResult U_EXPORT2
557 ucol_strcollUTF8(
558 const UCollator *coll,
559 const char *source,
560 int32_t sourceLength,
561 const char *target,
562 int32_t targetLength,
563 UErrorCode *status);
564
565 /**
566 * Determine if one string is greater than another.
567 * This function is equivalent to {@link #ucol_strcoll } == UCOL_GREATER
568 * @param coll The UCollator containing the comparison rules.
569 * @param source The source string.
570 * @param sourceLength The length of source, or -1 if null-terminated.
571 * @param target The target string.
572 * @param targetLength The length of target, or -1 if null-terminated.
573 * @return TRUE if source is greater than target, FALSE otherwise.
574 * @see ucol_strcoll
575 * @see ucol_greaterOrEqual
576 * @see ucol_equal
577 * @stable ICU 2.0
578 */
579 U_STABLE UBool U_EXPORT2
580 ucol_greater(const UCollator *coll,
581 const UChar *source, int32_t sourceLength,
582 const UChar *target, int32_t targetLength);
583
584 /**
585 * Determine if one string is greater than or equal to another.
586 * This function is equivalent to {@link #ucol_strcoll } != UCOL_LESS
587 * @param coll The UCollator containing the comparison rules.
588 * @param source The source string.
589 * @param sourceLength The length of source, or -1 if null-terminated.
590 * @param target The target string.
591 * @param targetLength The length of target, or -1 if null-terminated.
592 * @return TRUE if source is greater than or equal to target, FALSE otherwise.
593 * @see ucol_strcoll
594 * @see ucol_greater
595 * @see ucol_equal
596 * @stable ICU 2.0
597 */
598 U_STABLE UBool U_EXPORT2
599 ucol_greaterOrEqual(const UCollator *coll,
600 const UChar *source, int32_t sourceLength,
601 const UChar *target, int32_t targetLength);
602
603 /**
604 * Compare two strings for equality.
605 * This function is equivalent to {@link #ucol_strcoll } == UCOL_EQUAL
606 * @param coll The UCollator containing the comparison rules.
607 * @param source The source string.
608 * @param sourceLength The length of source, or -1 if null-terminated.
609 * @param target The target string.
610 * @param targetLength The length of target, or -1 if null-terminated.
611 * @return TRUE if source is equal to target, FALSE otherwise
612 * @see ucol_strcoll
613 * @see ucol_greater
614 * @see ucol_greaterOrEqual
615 * @stable ICU 2.0
616 */
617 U_STABLE UBool U_EXPORT2
618 ucol_equal(const UCollator *coll,
619 const UChar *source, int32_t sourceLength,
620 const UChar *target, int32_t targetLength);
621
622 /**
623 * Compare two UTF-8 encoded trings.
624 * The strings will be compared using the options already specified.
625 * @param coll The UCollator containing the comparison rules.
626 * @param sIter The source string iterator.
627 * @param tIter The target string iterator.
628 * @return The result of comparing the strings; one of UCOL_EQUAL,
629 * UCOL_GREATER, UCOL_LESS
630 * @param status A pointer to an UErrorCode to receive any errors
631 * @see ucol_strcoll
632 * @stable ICU 2.6
633 */
634 U_STABLE UCollationResult U_EXPORT2
635 ucol_strcollIter( const UCollator *coll,
636 UCharIterator *sIter,
637 UCharIterator *tIter,
638 UErrorCode *status);
639
640 /**
641 * Get the collation strength used in a UCollator.
642 * The strength influences how strings are compared.
643 * @param coll The UCollator to query.
644 * @return The collation strength; one of UCOL_PRIMARY, UCOL_SECONDARY,
645 * UCOL_TERTIARY, UCOL_QUATERNARY, UCOL_IDENTICAL
646 * @see ucol_setStrength
647 * @stable ICU 2.0
648 */
649 U_STABLE UCollationStrength U_EXPORT2
650 ucol_getStrength(const UCollator *coll);
651
652 /**
653 * Set the collation strength used in a UCollator.
654 * The strength influences how strings are compared.
655 * @param coll The UCollator to set.
656 * @param strength The desired collation strength; one of UCOL_PRIMARY,
657 * UCOL_SECONDARY, UCOL_TERTIARY, UCOL_QUATERNARY, UCOL_IDENTICAL, UCOL_DEFAULT
658 * @see ucol_getStrength
659 * @stable ICU 2.0
660 */
661 U_STABLE void U_EXPORT2
662 ucol_setStrength(UCollator *coll,
663 UCollationStrength strength);
664
665 /**
666 * Retrieves the reordering codes for this collator.
667 * These reordering codes are a combination of UScript codes and UColReorderCode entries.
668 * @param coll The UCollator to query.
669 * @param dest The array to fill with the script ordering.
670 * @param destCapacity The length of dest. If it is 0, then dest may be NULL and the function
671 * will only return the length of the result without writing any of the result string (pre-flighting).
672 * @param pErrorCode Must be a valid pointer to an error code value, which must not indicate a
673 * failure before the function call.
674 * @return The number of reordering codes written to the dest array.
675 * @see ucol_setReorderCodes
676 * @see ucol_getEquivalentReorderCodes
677 * @see UScriptCode
678 * @see UColReorderCode
679 * @stable ICU 4.8
680 */
681 U_STABLE int32_t U_EXPORT2
682 ucol_getReorderCodes(const UCollator* coll,
683 int32_t* dest,
684 int32_t destCapacity,
685 UErrorCode *pErrorCode);
686 /**
687 * Sets the reordering codes for this collator.
688 * Collation reordering allows scripts and some other defined blocks of characters
689 * to be moved relative to each other as a block. This reordering is done on top of
690 * the DUCET/CLDR standard collation order. Reordering can specify groups to be placed
691 * at the start and/or the end of the collation order. These groups are specified using
692 * UScript codes and UColReorderCode entries.
693 * <p>By default, reordering codes specified for the start of the order are placed in the
694 * order given after a group of "special" non-script blocks. These special groups of characters
695 * are space, punctuation, symbol, currency, and digit. These special groups are represented with
696 * UColReorderCode entries. Script groups can be intermingled with
697 * these special non-script blocks if those special blocks are explicitly specified in the reordering.
698 * <p>The special code OTHERS stands for any script that is not explicitly
699 * mentioned in the list of reordering codes given. Anything that is after OTHERS
700 * will go at the very end of the reordering in the order given.
701 * <p>The special reorder code DEFAULT will reset the reordering for this collator
702 * to the default for this collator. The default reordering may be the DUCET/CLDR order or may be a reordering that
703 * was specified when this collator was created from resource data or from rules. The
704 * DEFAULT code <b>must</b> be the sole code supplied when it used. If not
705 * that will result in an U_ILLEGAL_ARGUMENT_ERROR being set.
706 * <p>The special reorder code NONE will remove any reordering for this collator.
707 * The result of setting no reordering will be to have the DUCET/CLDR ordering used. The
708 * NONE code <b>must</b> be the sole code supplied when it used.
709 * @param coll The UCollator to set.
710 * @param reorderCodes An array of script codes in the new order. This can be NULL if the
711 * length is also set to 0. An empty array will clear any reordering codes on the collator.
712 * @param reorderCodesLength The length of reorderCodes.
713 * @param pErrorCode Must be a valid pointer to an error code value, which must not indicate a
714 * failure before the function call.
715 * @see ucol_getReorderCodes
716 * @see ucol_getEquivalentReorderCodes
717 * @see UScriptCode
718 * @see UColReorderCode
719 * @stable ICU 4.8
720 */
721 U_STABLE void U_EXPORT2
722 ucol_setReorderCodes(UCollator* coll,
723 const int32_t* reorderCodes,
724 int32_t reorderCodesLength,
725 UErrorCode *pErrorCode);
726
727 /**
728 * Retrieves the reorder codes that are grouped with the given reorder code. Some reorder
729 * codes will be grouped and must reorder together.
730 * @param reorderCode The reorder code to determine equivalence for.
731 * @param dest The array to fill with the script ordering.
732 * @param destCapacity The length of dest. If it is 0, then dest may be NULL and the function
733 * will only return the length of the result without writing any of the result string (pre-flighting).
734 * @param pErrorCode Must be a valid pointer to an error code value, which must not indicate
735 * a failure before the function call.
736 * @return The number of reordering codes written to the dest array.
737 * @see ucol_setReorderCodes
738 * @see ucol_getReorderCodes
739 * @see UScriptCode
740 * @see UColReorderCode
741 * @stable ICU 4.8
742 */
743 U_STABLE int32_t U_EXPORT2
744 ucol_getEquivalentReorderCodes(int32_t reorderCode,
745 int32_t* dest,
746 int32_t destCapacity,
747 UErrorCode *pErrorCode);
748
749 /**
750 * Get the display name for a UCollator.
751 * The display name is suitable for presentation to a user.
752 * @param objLoc The locale of the collator in question.
753 * @param dispLoc The locale for display.
754 * @param result A pointer to a buffer to receive the attribute.
755 * @param resultLength The maximum size of result.
756 * @param status A pointer to an UErrorCode to receive any errors
757 * @return The total buffer size needed; if greater than resultLength,
758 * the output was truncated.
759 * @stable ICU 2.0
760 */
761 U_STABLE int32_t U_EXPORT2
762 ucol_getDisplayName( const char *objLoc,
763 const char *dispLoc,
764 UChar *result,
765 int32_t resultLength,
766 UErrorCode *status);
767
768 /**
769 * Get a locale for which collation rules are available.
770 * A UCollator in a locale returned by this function will perform the correct
771 * collation for the locale.
772 * @param localeIndex The index of the desired locale.
773 * @return A locale for which collation rules are available, or 0 if none.
774 * @see ucol_countAvailable
775 * @stable ICU 2.0
776 */
777 U_STABLE const char* U_EXPORT2
778 ucol_getAvailable(int32_t localeIndex);
779
780 /**
781 * Determine how many locales have collation rules available.
782 * This function is most useful as determining the loop ending condition for
783 * calls to {@link #ucol_getAvailable }.
784 * @return The number of locales for which collation rules are available.
785 * @see ucol_getAvailable
786 * @stable ICU 2.0
787 */
788 U_STABLE int32_t U_EXPORT2
789 ucol_countAvailable(void);
790
791 #if !UCONFIG_NO_SERVICE
792 /**
793 * Create a string enumerator of all locales for which a valid
794 * collator may be opened.
795 * @param status input-output error code
796 * @return a string enumeration over locale strings. The caller is
797 * responsible for closing the result.
798 * @stable ICU 3.0
799 */
800 U_STABLE UEnumeration* U_EXPORT2
801 ucol_openAvailableLocales(UErrorCode *status);
802 #endif
803
804 /**
805 * Create a string enumerator of all possible keywords that are relevant to
806 * collation. At this point, the only recognized keyword for this
807 * service is "collation".
808 * @param status input-output error code
809 * @return a string enumeration over locale strings. The caller is
810 * responsible for closing the result.
811 * @stable ICU 3.0
812 */
813 U_STABLE UEnumeration* U_EXPORT2
814 ucol_getKeywords(UErrorCode *status);
815
816 /**
817 * Given a keyword, create a string enumeration of all values
818 * for that keyword that are currently in use.
819 * @param keyword a particular keyword as enumerated by
820 * ucol_getKeywords. If any other keyword is passed in, *status is set
821 * to U_ILLEGAL_ARGUMENT_ERROR.
822 * @param status input-output error code
823 * @return a string enumeration over collation keyword values, or NULL
824 * upon error. The caller is responsible for closing the result.
825 * @stable ICU 3.0
826 */
827 U_STABLE UEnumeration* U_EXPORT2
828 ucol_getKeywordValues(const char *keyword, UErrorCode *status);
829
830 /**
831 * Given a key and a locale, returns an array of string values in a preferred
832 * order that would make a difference. These are all and only those values where
833 * the open (creation) of the service with the locale formed from the input locale
834 * plus input keyword and that value has different behavior than creation with the
835 * input locale alone.
836 * @param key one of the keys supported by this service. For now, only
837 * "collation" is supported.
838 * @param locale the locale
839 * @param commonlyUsed if set to true it will return only commonly used values
840 * with the given locale in preferred order. Otherwise,
841 * it will return all the available values for the locale.
842 * @param status error status
843 * @return a string enumeration over keyword values for the given key and the locale.
844 * @stable ICU 4.2
845 */
846 U_STABLE UEnumeration* U_EXPORT2
847 ucol_getKeywordValuesForLocale(const char* key,
848 const char* locale,
849 UBool commonlyUsed,
850 UErrorCode* status);
851
852 /**
853 * Return the functionally equivalent locale for the given
854 * requested locale, with respect to given keyword, for the
855 * collation service. If two locales return the same result, then
856 * collators instantiated for these locales will behave
857 * equivalently. The converse is not always true; two collators
858 * may in fact be equivalent, but return different results, due to
859 * internal details. The return result has no other meaning than
860 * that stated above, and implies nothing as to the relationship
861 * between the two locales. This is intended for use by
862 * applications who wish to cache collators, or otherwise reuse
863 * collators when possible. The functional equivalent may change
864 * over time. For more information, please see the <a
865 * href="http://icu-project.org/userguide/locale.html#services">
866 * Locales and Services</a> section of the ICU User Guide.
867 * @param result fillin for the functionally equivalent locale
868 * @param resultCapacity capacity of the fillin buffer
869 * @param keyword a particular keyword as enumerated by
870 * ucol_getKeywords.
871 * @param locale the requested locale
872 * @param isAvailable if non-NULL, pointer to a fillin parameter that
873 * indicates whether the requested locale was 'available' to the
874 * collation service. A locale is defined as 'available' if it
875 * physically exists within the collation locale data.
876 * @param status pointer to input-output error code
877 * @return the actual buffer size needed for the locale. If greater
878 * than resultCapacity, the returned full name will be truncated and
879 * an error code will be returned.
880 * @stable ICU 3.0
881 */
882 U_STABLE int32_t U_EXPORT2
883 ucol_getFunctionalEquivalent(char* result, int32_t resultCapacity,
884 const char* keyword, const char* locale,
885 UBool* isAvailable, UErrorCode* status);
886
887 /**
888 * Get the collation tailoring rules from a UCollator.
889 * The rules will follow the rule syntax.
890 * @param coll The UCollator to query.
891 * @param length
892 * @return The collation tailoring rules.
893 * @stable ICU 2.0
894 */
895 U_STABLE const UChar* U_EXPORT2
896 ucol_getRules( const UCollator *coll,
897 int32_t *length);
898
899 /** Get the short definition string for a collator. This API harvests the collator's
900 * locale and the attribute set and produces a string that can be used for opening
901 * a collator with the same properties using the ucol_openFromShortString API.
902 * This string will be normalized.
903 * The structure and the syntax of the string is defined in the "Naming collators"
904 * section of the users guide:
905 * http://icu-project.org/userguide/Collate_Concepts.html#Naming_Collators
906 * This API supports preflighting.
907 * @param coll a collator
908 * @param locale a locale that will appear as a collators locale in the resulting
909 * short string definition. If NULL, the locale will be harvested
910 * from the collator.
911 * @param buffer space to hold the resulting string
912 * @param capacity capacity of the buffer
913 * @param status for returning errors. All the preflighting errors are featured
914 * @return length of the resulting string
915 * @see ucol_openFromShortString
916 * @see ucol_normalizeShortDefinitionString
917 * @stable ICU 3.0
918 */
919 U_STABLE int32_t U_EXPORT2
920 ucol_getShortDefinitionString(const UCollator *coll,
921 const char *locale,
922 char *buffer,
923 int32_t capacity,
924 UErrorCode *status);
925
926 /** Verifies and normalizes short definition string.
927 * Normalized short definition string has all the option sorted by the argument name,
928 * so that equivalent definition strings are the same.
929 * This API supports preflighting.
930 * @param source definition string
931 * @param destination space to hold the resulting string
932 * @param capacity capacity of the buffer
933 * @param parseError if not NULL, structure that will get filled with error's pre
934 * and post context in case of error.
935 * @param status Error code. This API will return an error if an invalid attribute
936 * or attribute/value combination is specified. All the preflighting
937 * errors are also featured
938 * @return length of the resulting normalized string.
939 *
940 * @see ucol_openFromShortString
941 * @see ucol_getShortDefinitionString
942 *
943 * @stable ICU 3.0
944 */
945
946 U_STABLE int32_t U_EXPORT2
947 ucol_normalizeShortDefinitionString(const char *source,
948 char *destination,
949 int32_t capacity,
950 UParseError *parseError,
951 UErrorCode *status);
952
953
954 /**
955 * Get a sort key for a string from a UCollator.
956 * Sort keys may be compared using <TT>strcmp</TT>.
957 *
958 * Like ICU functions that write to an output buffer, the buffer contents
959 * is undefined if the buffer capacity (resultLength parameter) is too small.
960 * Unlike ICU functions that write a string to an output buffer,
961 * the terminating zero byte is counted in the sort key length.
962 * @param coll The UCollator containing the collation rules.
963 * @param source The string to transform.
964 * @param sourceLength The length of source, or -1 if null-terminated.
965 * @param result A pointer to a buffer to receive the attribute.
966 * @param resultLength The maximum size of result.
967 * @return The size needed to fully store the sort key.
968 * If there was an internal error generating the sort key,
969 * a zero value is returned.
970 * @see ucol_keyHashCode
971 * @stable ICU 2.0
972 */
973 U_STABLE int32_t U_EXPORT2
974 ucol_getSortKey(const UCollator *coll,
975 const UChar *source,
976 int32_t sourceLength,
977 uint8_t *result,
978 int32_t resultLength);
979
980
981 /** Gets the next count bytes of a sort key. Caller needs
982 * to preserve state array between calls and to provide
983 * the same type of UCharIterator set with the same string.
984 * The destination buffer provided must be big enough to store
985 * the number of requested bytes.
986 *
987 * The generated sort key may or may not be compatible with
988 * sort keys generated using ucol_getSortKey().
989 * @param coll The UCollator containing the collation rules.
990 * @param iter UCharIterator containing the string we need
991 * the sort key to be calculated for.
992 * @param state Opaque state of sortkey iteration.
993 * @param dest Buffer to hold the resulting sortkey part
994 * @param count number of sort key bytes required.
995 * @param status error code indicator.
996 * @return the actual number of bytes of a sortkey. It can be
997 * smaller than count if we have reached the end of
998 * the sort key.
999 * @stable ICU 2.6
1000 */
1001 U_STABLE int32_t U_EXPORT2
1002 ucol_nextSortKeyPart(const UCollator *coll,
1003 UCharIterator *iter,
1004 uint32_t state[2],
1005 uint8_t *dest, int32_t count,
1006 UErrorCode *status);
1007
1008 /** enum that is taken by ucol_getBound API
1009 * See below for explanation
1010 * do not change the values assigned to the
1011 * members of this enum. Underlying code
1012 * depends on them having these numbers
1013 * @stable ICU 2.0
1014 */
1015 typedef enum {
1016 /** lower bound */
1017 UCOL_BOUND_LOWER = 0,
1018 /** upper bound that will match strings of exact size */
1019 UCOL_BOUND_UPPER = 1,
1020 /** upper bound that will match all the strings that have the same initial substring as the given string */
1021 UCOL_BOUND_UPPER_LONG = 2,
1022 UCOL_BOUND_VALUE_COUNT
1023 } UColBoundMode;
1024
1025 /**
1026 * Produce a bound for a given sortkey and a number of levels.
1027 * Return value is always the number of bytes needed, regardless of
1028 * whether the result buffer was big enough or even valid.<br>
1029 * Resulting bounds can be used to produce a range of strings that are
1030 * between upper and lower bounds. For example, if bounds are produced
1031 * for a sortkey of string "smith", strings between upper and lower
1032 * bounds with one level would include "Smith", "SMITH", "sMiTh".<br>
1033 * There are two upper bounds that can be produced. If UCOL_BOUND_UPPER
1034 * is produced, strings matched would be as above. However, if bound
1035 * produced using UCOL_BOUND_UPPER_LONG is used, the above example will
1036 * also match "Smithsonian" and similar.<br>
1037 * For more on usage, see example in cintltst/capitst.c in procedure
1038 * TestBounds.
1039 * Sort keys may be compared using <TT>strcmp</TT>.
1040 * @param source The source sortkey.
1041 * @param sourceLength The length of source, or -1 if null-terminated.
1042 * (If an unmodified sortkey is passed, it is always null
1043 * terminated).
1044 * @param boundType Type of bound required. It can be UCOL_BOUND_LOWER, which
1045 * produces a lower inclusive bound, UCOL_BOUND_UPPER, that
1046 * produces upper bound that matches strings of the same length
1047 * or UCOL_BOUND_UPPER_LONG that matches strings that have the
1048 * same starting substring as the source string.
1049 * @param noOfLevels Number of levels required in the resulting bound (for most
1050 * uses, the recommended value is 1). See users guide for
1051 * explanation on number of levels a sortkey can have.
1052 * @param result A pointer to a buffer to receive the resulting sortkey.
1053 * @param resultLength The maximum size of result.
1054 * @param status Used for returning error code if something went wrong. If the
1055 * number of levels requested is higher than the number of levels
1056 * in the source key, a warning (U_SORT_KEY_TOO_SHORT_WARNING) is
1057 * issued.
1058 * @return The size needed to fully store the bound.
1059 * @see ucol_keyHashCode
1060 * @stable ICU 2.1
1061 */
1062 U_STABLE int32_t U_EXPORT2
1063 ucol_getBound(const uint8_t *source,
1064 int32_t sourceLength,
1065 UColBoundMode boundType,
1066 uint32_t noOfLevels,
1067 uint8_t *result,
1068 int32_t resultLength,
1069 UErrorCode *status);
1070
1071 /**
1072 * Gets the version information for a Collator. Version is currently
1073 * an opaque 32-bit number which depends, among other things, on major
1074 * versions of the collator tailoring and UCA.
1075 * @param coll The UCollator to query.
1076 * @param info the version # information, the result will be filled in
1077 * @stable ICU 2.0
1078 */
1079 U_STABLE void U_EXPORT2
1080 ucol_getVersion(const UCollator* coll, UVersionInfo info);
1081
1082 /**
1083 * Gets the UCA version information for a Collator. Version is the
1084 * UCA version number (3.1.1, 4.0).
1085 * @param coll The UCollator to query.
1086 * @param info the version # information, the result will be filled in
1087 * @stable ICU 2.8
1088 */
1089 U_STABLE void U_EXPORT2
1090 ucol_getUCAVersion(const UCollator* coll, UVersionInfo info);
1091
1092 /**
1093 * Merges two sort keys. The levels are merged with their corresponding counterparts
1094 * (primaries with primaries, secondaries with secondaries etc.). Between the values
1095 * from the same level a separator is inserted.
1096 *
1097 * This is useful, for example, for combining sort keys from first and last names
1098 * to sort such pairs.
1099 * It is possible to merge multiple sort keys by consecutively merging
1100 * another one with the intermediate result.
1101 *
1102 * The length of the merge result is the sum of the lengths of the input sort keys.
1103 *
1104 * Example (uncompressed):
1105 * <pre>191B1D 01 050505 01 910505 00
1106 * 1F2123 01 050505 01 910505 00</pre>
1107 * will be merged as
1108 * <pre>191B1D 02 1F2123 01 050505 02 050505 01 910505 02 910505 00</pre>
1109 *
1110 * If the destination buffer is not big enough, then its contents are undefined.
1111 * If any of source lengths are zero or any of the source pointers are NULL/undefined,
1112 * the result is of size zero.
1113 *
1114 * @param src1 the first sort key
1115 * @param src1Length the length of the first sort key, including the zero byte at the end;
1116 * can be -1 if the function is to find the length
1117 * @param src2 the second sort key
1118 * @param src2Length the length of the second sort key, including the zero byte at the end;
1119 * can be -1 if the function is to find the length
1120 * @param dest the buffer where the merged sort key is written,
1121 * can be NULL if destCapacity==0
1122 * @param destCapacity the number of bytes in the dest buffer
1123 * @return the length of the merged sort key, src1Length+src2Length;
1124 * can be larger than destCapacity, or 0 if an error occurs (only for illegal arguments),
1125 * in which cases the contents of dest is undefined
1126 * @stable ICU 2.0
1127 */
1128 U_STABLE int32_t U_EXPORT2
1129 ucol_mergeSortkeys(const uint8_t *src1, int32_t src1Length,
1130 const uint8_t *src2, int32_t src2Length,
1131 uint8_t *dest, int32_t destCapacity);
1132
1133 /**
1134 * Universal attribute setter
1135 * @param coll collator which attributes are to be changed
1136 * @param attr attribute type
1137 * @param value attribute value
1138 * @param status to indicate whether the operation went on smoothly or there were errors
1139 * @see UColAttribute
1140 * @see UColAttributeValue
1141 * @see ucol_getAttribute
1142 * @stable ICU 2.0
1143 */
1144 U_STABLE void U_EXPORT2
1145 ucol_setAttribute(UCollator *coll, UColAttribute attr, UColAttributeValue value, UErrorCode *status);
1146
1147 /**
1148 * Universal attribute getter
1149 * @param coll collator which attributes are to be changed
1150 * @param attr attribute type
1151 * @return attribute value
1152 * @param status to indicate whether the operation went on smoothly or there were errors
1153 * @see UColAttribute
1154 * @see UColAttributeValue
1155 * @see ucol_setAttribute
1156 * @stable ICU 2.0
1157 */
1158 U_STABLE UColAttributeValue U_EXPORT2
1159 ucol_getAttribute(const UCollator *coll, UColAttribute attr, UErrorCode *status);
1160
1161 /** Variable top
1162 * is a two byte primary value which causes all the codepoints with primary values that
1163 * are less or equal than the variable top to be shifted when alternate handling is set
1164 * to UCOL_SHIFTED.
1165 * Sets the variable top to a collation element value of a string supplied.
1166 * @param coll collator which variable top needs to be changed
1167 * @param varTop one or more (if contraction) UChars to which the variable top should be set
1168 * @param len length of variable top string. If -1 it is considered to be zero terminated.
1169 * @param status error code. If error code is set, the return value is undefined.
1170 * Errors set by this function are: <br>
1171 * U_CE_NOT_FOUND_ERROR if more than one character was passed and there is no such
1172 * a contraction<br>
1173 * U_PRIMARY_TOO_LONG_ERROR if the primary for the variable top has more than two bytes
1174 * @return a 32 bit value containing the value of the variable top in upper 16 bits.
1175 * Lower 16 bits are undefined
1176 * @see ucol_getVariableTop
1177 * @see ucol_restoreVariableTop
1178 * @stable ICU 2.0
1179 */
1180 U_STABLE uint32_t U_EXPORT2
1181 ucol_setVariableTop(UCollator *coll,
1182 const UChar *varTop, int32_t len,
1183 UErrorCode *status);
1184
1185 /**
1186 * Gets the variable top value of a Collator.
1187 * Lower 16 bits are undefined and should be ignored.
1188 * @param coll collator which variable top needs to be retrieved
1189 * @param status error code (not changed by function). If error code is set,
1190 * the return value is undefined.
1191 * @return the variable top value of a Collator.
1192 * @see ucol_setVariableTop
1193 * @see ucol_restoreVariableTop
1194 * @stable ICU 2.0
1195 */
1196 U_STABLE uint32_t U_EXPORT2 ucol_getVariableTop(const UCollator *coll, UErrorCode *status);
1197
1198 /**
1199 * Sets the variable top to a collation element value supplied. Variable top is
1200 * set to the upper 16 bits.
1201 * Lower 16 bits are ignored.
1202 * @param coll collator which variable top needs to be changed
1203 * @param varTop CE value, as returned by ucol_setVariableTop or ucol)getVariableTop
1204 * @param status error code (not changed by function)
1205 * @see ucol_getVariableTop
1206 * @see ucol_setVariableTop
1207 * @stable ICU 2.0
1208 */
1209 U_STABLE void U_EXPORT2
1210 ucol_restoreVariableTop(UCollator *coll, const uint32_t varTop, UErrorCode *status);
1211
1212 /**
1213 * Thread safe cloning operation. The result is a clone of a given collator.
1214 * @param coll collator to be cloned
1215 * @param stackBuffer <em>Deprecated functionality as of ICU 52, use NULL.</em><br>
1216 * user allocated space for the new clone.
1217 * If NULL new memory will be allocated.
1218 * If buffer is not large enough, new memory will be allocated.
1219 * Clients can use the U_COL_SAFECLONE_BUFFERSIZE.
1220 * @param pBufferSize <em>Deprecated functionality as of ICU 52, use NULL or 1.</em><br>
1221 * pointer to size of allocated space.
1222 * If *pBufferSize == 0, a sufficient size for use in cloning will
1223 * be returned ('pre-flighting')
1224 * If *pBufferSize is not enough for a stack-based safe clone,
1225 * new memory will be allocated.
1226 * @param status to indicate whether the operation went on smoothly or there were errors
1227 * An informational status value, U_SAFECLONE_ALLOCATED_ERROR, is used if any
1228 * allocations were necessary.
1229 * @return pointer to the new clone
1230 * @see ucol_open
1231 * @see ucol_openRules
1232 * @see ucol_close
1233 * @stable ICU 2.0
1234 */
1235 U_STABLE UCollator* U_EXPORT2
1236 ucol_safeClone(const UCollator *coll,
1237 void *stackBuffer,
1238 int32_t *pBufferSize,
1239 UErrorCode *status);
1240
1241 #ifndef U_HIDE_DEPRECATED_API
1242
1243 /** default memory size for the new clone.
1244 * @deprecated ICU 52. Do not rely on ucol_safeClone() cloning into any provided buffer.
1245 */
1246 #define U_COL_SAFECLONE_BUFFERSIZE 1
1247
1248 #endif /* U_HIDE_DEPRECATED_API */
1249
1250 /**
1251 * Returns current rules. Delta defines whether full rules are returned or just the tailoring.
1252 * Returns number of UChars needed to store rules. If buffer is NULL or bufferLen is not enough
1253 * to store rules, will store up to available space.
1254 *
1255 * ucol_getRules() should normally be used instead.
1256 * See http://userguide.icu-project.org/collation/customization#TOC-Building-on-Existing-Locales
1257 * @param coll collator to get the rules from
1258 * @param delta one of UCOL_TAILORING_ONLY, UCOL_FULL_RULES.
1259 * @param buffer buffer to store the result in. If NULL, you'll get no rules.
1260 * @param bufferLen length of buffer to store rules in. If less than needed you'll get only the part that fits in.
1261 * @return current rules
1262 * @stable ICU 2.0
1263 * @see UCOL_FULL_RULES
1264 */
1265 U_STABLE int32_t U_EXPORT2
1266 ucol_getRulesEx(const UCollator *coll, UColRuleOption delta, UChar *buffer, int32_t bufferLen);
1267
1268 #ifndef U_HIDE_DEPRECATED_API
1269 /**
1270 * gets the locale name of the collator. If the collator
1271 * is instantiated from the rules, then this function returns
1272 * NULL.
1273 * @param coll The UCollator for which the locale is needed
1274 * @param type You can choose between requested, valid and actual
1275 * locale. For description see the definition of
1276 * ULocDataLocaleType in uloc.h
1277 * @param status error code of the operation
1278 * @return real locale name from which the collation data comes.
1279 * If the collator was instantiated from rules, returns
1280 * NULL.
1281 * @deprecated ICU 2.8 Use ucol_getLocaleByType instead
1282 */
1283 U_DEPRECATED const char * U_EXPORT2
1284 ucol_getLocale(const UCollator *coll, ULocDataLocaleType type, UErrorCode *status);
1285 #endif /* U_HIDE_DEPRECATED_API */
1286
1287 /**
1288 * gets the locale name of the collator. If the collator
1289 * is instantiated from the rules, then this function returns
1290 * NULL.
1291 * @param coll The UCollator for which the locale is needed
1292 * @param type You can choose between requested, valid and actual
1293 * locale. For description see the definition of
1294 * ULocDataLocaleType in uloc.h
1295 * @param status error code of the operation
1296 * @return real locale name from which the collation data comes.
1297 * If the collator was instantiated from rules, returns
1298 * NULL.
1299 * @stable ICU 2.8
1300 */
1301 U_STABLE const char * U_EXPORT2
1302 ucol_getLocaleByType(const UCollator *coll, ULocDataLocaleType type, UErrorCode *status);
1303
1304 /**
1305 * Get an Unicode set that contains all the characters and sequences tailored in
1306 * this collator. The result must be disposed of by using uset_close.
1307 * @param coll The UCollator for which we want to get tailored chars
1308 * @param status error code of the operation
1309 * @return a pointer to newly created USet. Must be be disposed by using uset_close
1310 * @see ucol_openRules
1311 * @see uset_close
1312 * @stable ICU 2.4
1313 */
1314 U_STABLE USet * U_EXPORT2
1315 ucol_getTailoredSet(const UCollator *coll, UErrorCode *status);
1316
1317 #ifndef U_HIDE_INTERNAL_API
1318 /**
1319 * Universal attribute getter that returns UCOL_DEFAULT if the value is default
1320 * @param coll collator which attributes are to be changed
1321 * @param attr attribute type
1322 * @return attribute value or UCOL_DEFAULT if the value is default
1323 * @param status to indicate whether the operation went on smoothly or there were errors
1324 * @see UColAttribute
1325 * @see UColAttributeValue
1326 * @see ucol_setAttribute
1327 * @internal ICU 3.0
1328 */
1329 U_INTERNAL UColAttributeValue U_EXPORT2
1330 ucol_getAttributeOrDefault(const UCollator *coll, UColAttribute attr, UErrorCode *status);
1331
1332 /** Check whether two collators are equal. Collators are considered equal if they
1333 * will sort strings the same. This means that both the current attributes and the
1334 * rules must be equivalent. Currently used for RuleBasedCollator::operator==.
1335 * @param source first collator
1336 * @param target second collator
1337 * @return TRUE or FALSE
1338 * @internal ICU 3.0
1339 */
1340 U_INTERNAL UBool U_EXPORT2
1341 ucol_equals(const UCollator *source, const UCollator *target);
1342
1343 /** Calculates the set of unsafe code points, given a collator.
1344 * A character is unsafe if you could append any character and cause the ordering to alter significantly.
1345 * Collation sorts in normalized order, so anything that rearranges in normalization can cause this.
1346 * Thus if you have a character like a_umlaut, and you add a lower_dot to it,
1347 * then it normalizes to a_lower_dot + umlaut, and sorts differently.
1348 * @param coll Collator
1349 * @param unsafe a fill-in set to receive the unsafe points
1350 * @param status for catching errors
1351 * @return number of elements in the set
1352 * @internal ICU 3.0
1353 */
1354 U_INTERNAL int32_t U_EXPORT2
1355 ucol_getUnsafeSet( const UCollator *coll,
1356 USet *unsafe,
1357 UErrorCode *status);
1358
1359 /** Reset UCA's static pointers. You don't want to use this, unless your static memory can go away.
1360 * @internal ICU 3.2.1
1361 */
1362 U_INTERNAL void U_EXPORT2
1363 ucol_forgetUCA(void);
1364
1365 /** Touches all resources needed for instantiating a collator from a short string definition,
1366 * thus filling up the cache.
1367 * @param definition A short string containing a locale and a set of attributes.
1368 * Attributes not explicitly mentioned are left at the default
1369 * state for a locale.
1370 * @param parseError if not NULL, structure that will get filled with error's pre
1371 * and post context in case of error.
1372 * @param forceDefaults if FALSE, the settings that are the same as the collator
1373 * default settings will not be applied (for example, setting
1374 * French secondary on a French collator would not be executed).
1375 * If TRUE, all the settings will be applied regardless of the
1376 * collator default value. If the definition
1377 * strings are to be cached, should be set to FALSE.
1378 * @param status Error code. Apart from regular error conditions connected to
1379 * instantiating collators (like out of memory or similar), this
1380 * API will return an error if an invalid attribute or attribute/value
1381 * combination is specified.
1382 * @see ucol_openFromShortString
1383 * @internal ICU 3.2.1
1384 */
1385 U_INTERNAL void U_EXPORT2
1386 ucol_prepareShortStringOpen( const char *definition,
1387 UBool forceDefaults,
1388 UParseError *parseError,
1389 UErrorCode *status);
1390 #endif /* U_HIDE_INTERNAL_API */
1391
1392 /** Creates a binary image of a collator. This binary image can be stored and
1393 * later used to instantiate a collator using ucol_openBinary.
1394 * This API supports preflighting.
1395 * @param coll Collator
1396 * @param buffer a fill-in buffer to receive the binary image
1397 * @param capacity capacity of the destination buffer
1398 * @param status for catching errors
1399 * @return size of the image
1400 * @see ucol_openBinary
1401 * @stable ICU 3.2
1402 */
1403 U_STABLE int32_t U_EXPORT2
1404 ucol_cloneBinary(const UCollator *coll,
1405 uint8_t *buffer, int32_t capacity,
1406 UErrorCode *status);
1407
1408 /** Opens a collator from a collator binary image created using
1409 * ucol_cloneBinary. Binary image used in instantiation of the
1410 * collator remains owned by the user and should stay around for
1411 * the lifetime of the collator. The API also takes a base collator
1412 * which usualy should be UCA.
1413 * @param bin binary image owned by the user and required through the
1414 * lifetime of the collator
1415 * @param length size of the image. If negative, the API will try to
1416 * figure out the length of the image
1417 * @param base fallback collator, usually UCA. Base is required to be
1418 * present through the lifetime of the collator. Currently
1419 * it cannot be NULL.
1420 * @param status for catching errors
1421 * @return newly created collator
1422 * @see ucol_cloneBinary
1423 * @stable ICU 3.2
1424 */
1425 U_STABLE UCollator* U_EXPORT2
1426 ucol_openBinary(const uint8_t *bin, int32_t length,
1427 const UCollator *base,
1428 UErrorCode *status);
1429
1430
1431 #endif /* #if !UCONFIG_NO_COLLATION */
1432
1433 #endif

Mercurial Repository: The Tor Browser

Europalab SCM Repository Server

mercurial