icuSources/common/unicode/ucasemap.h

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*
   4 *******************************************************************************
   5 *
   6 *   Copyright (C) 2005-2012, International Business Machines
   7 *   Corporation and others.  All Rights Reserved.
   8 *
   9 *******************************************************************************
  10 *   file name:  ucasemap.h
  11 *   encoding:   UTF-8
  12 *   tab size:   8 (not used)
  13 *   indentation:4
  14 *
  15 *   created on: 2005may06
  16 *   created by: Markus W. Scherer
  17 *
  18 *   Case mapping service object and functions using it.
  19 */
  20
  21 #ifndef __UCASEMAP_H__
  22 #define __UCASEMAP_H__
  23
  24 #include "unicode/utypes.h"
  25 #include "unicode/localpointer.h"
  26 #include "unicode/stringoptions.h"
  27 #include "unicode/ustring.h"
  28
  29 /**
  30  * \file
  31  * \brief C API: Unicode case mapping functions using a UCaseMap service object.
  32  *
  33  * The service object takes care of memory allocations, data loading, and setup
  34  * for the attributes, as usual.
  35  *
  36  * Currently, the functionality provided here does not overlap with uchar.h
  37  * and ustring.h, except for ucasemap_toTitle().
  38  *
  39  * ucasemap_utf8XYZ() functions operate directly on UTF-8 strings.
  40  */
  41
  42 /**
  43  * UCaseMap is an opaque service object for newer ICU case mapping functions.
  44  * Older functions did not use a service object.
  45  * @stable ICU 3.4
  46  */
  47 struct UCaseMap;
  48 typedef struct UCaseMap UCaseMap; /**< C typedef for struct UCaseMap. @stable ICU 3.4 */
  49
  50 /**
  51  * Open a UCaseMap service object for a locale and a set of options.
  52  * The locale ID and options are preprocessed so that functions using the
  53  * service object need not process them in each call.
  54  *
  55  * @param locale ICU locale ID, used for language-dependent
  56  *               upper-/lower-/title-casing according to the Unicode standard.
  57  *               Usual semantics: ""=root, NULL=default locale, etc.
  58  * @param options Options bit set, used for case folding and string comparisons.
  59  *                Same flags as for u_foldCase(), u_strFoldCase(),
  60  *                u_strCaseCompare(), etc.
  61  *                Use 0 or U_FOLD_CASE_DEFAULT for default behavior.
  62  * @param pErrorCode Must be a valid pointer to an error code value,
  63  *                   which must not indicate a failure before the function call.
  64  * @return Pointer to a UCaseMap service object, if successful.
  65  *
  66  * @see U_FOLD_CASE_DEFAULT
  67  * @see U_FOLD_CASE_EXCLUDE_SPECIAL_I
  68  * @see U_TITLECASE_NO_LOWERCASE
  69  * @see U_TITLECASE_NO_BREAK_ADJUSTMENT
  70  * @stable ICU 3.4
  71  */
  72 U_STABLE UCaseMap * U_EXPORT2
  73 ucasemap_open(const char *locale, uint32_t options, UErrorCode *pErrorCode);
  74
  75 /**
  76  * Close a UCaseMap service object.
  77  * @param csm Object to be closed.
  78  * @stable ICU 3.4
  79  */
  80 U_STABLE void U_EXPORT2
  81 ucasemap_close(UCaseMap *csm);
  82
  83 #if U_SHOW_CPLUSPLUS_API
  84
  85 U_NAMESPACE_BEGIN
  86
  87 /**
  88  * \class LocalUCaseMapPointer
  89  * "Smart pointer" class, closes a UCaseMap via ucasemap_close().
  90  * For most methods see the LocalPointerBase base class.
  91  *
  92  * @see LocalPointerBase
  93  * @see LocalPointer
  94  * @stable ICU 4.4
  95  */
  96 U_DEFINE_LOCAL_OPEN_POINTER(LocalUCaseMapPointer, UCaseMap, ucasemap_close);
  97
  98 U_NAMESPACE_END
  99
 100 #endif // U_SHOW_CPLUSPLUS_API
 101
 102 /**
 103  * Get the locale ID that is used for language-dependent case mappings.
 104  * @param csm UCaseMap service object.
 105  * @return locale ID
 106  * @stable ICU 3.4
 107  */
 108 U_STABLE const char * U_EXPORT2
 109 ucasemap_getLocale(const UCaseMap *csm);
 110
 111 /**
 112  * Get the options bit set that is used for case folding and string comparisons.
 113  * @param csm UCaseMap service object.
 114  * @return options bit set
 115  * @stable ICU 3.4
 116  */
 117 U_STABLE uint32_t U_EXPORT2
 118 ucasemap_getOptions(const UCaseMap *csm);
 119
 120 /**
 121  * Set the locale ID that is used for language-dependent case mappings.
 122  *
 123  * @param csm UCaseMap service object.
 124  * @param locale Locale ID, see ucasemap_open().
 125  * @param pErrorCode Must be a valid pointer to an error code value,
 126  *                   which must not indicate a failure before the function call.
 127  *
 128  * @see ucasemap_open
 129  * @stable ICU 3.4
 130  */
 131 U_STABLE void U_EXPORT2
 132 ucasemap_setLocale(UCaseMap *csm, const char *locale, UErrorCode *pErrorCode);
 133
 134 /**
 135  * Set the options bit set that is used for case folding and string comparisons.
 136  *
 137  * @param csm UCaseMap service object.
 138  * @param options Options bit set, see ucasemap_open().
 139  * @param pErrorCode Must be a valid pointer to an error code value,
 140  *                   which must not indicate a failure before the function call.
 141  *
 142  * @see ucasemap_open
 143  * @stable ICU 3.4
 144  */
 145 U_STABLE void U_EXPORT2
 146 ucasemap_setOptions(UCaseMap *csm, uint32_t options, UErrorCode *pErrorCode);
 147
 148 #if !UCONFIG_NO_BREAK_ITERATION
 149
 150 /**
 151  * Get the break iterator that is used for titlecasing.
 152  * Do not modify the returned break iterator.
 153  * @param csm UCaseMap service object.
 154  * @return titlecasing break iterator
 155  * @stable ICU 3.8
 156  */
 157 U_STABLE const UBreakIterator * U_EXPORT2
 158 ucasemap_getBreakIterator(const UCaseMap *csm);
 159
 160 /**
 161  * Set the break iterator that is used for titlecasing.
 162  * The UCaseMap service object releases a previously set break iterator
 163  * and "adopts" this new one, taking ownership of it.
 164  * It will be released in a subsequent call to ucasemap_setBreakIterator()
 165  * or ucasemap_close().
 166  *
 167  * Break iterator operations are not thread-safe. Therefore, titlecasing
 168  * functions use non-const UCaseMap objects. It is not possible to titlecase
 169  * strings concurrently using the same UCaseMap.
 170  *
 171  * @param csm UCaseMap service object.
 172  * @param iterToAdopt Break iterator to be adopted for titlecasing.
 173  * @param pErrorCode Must be a valid pointer to an error code value,
 174  *                   which must not indicate a failure before the function call.
 175  *
 176  * @see ucasemap_toTitle
 177  * @see ucasemap_utf8ToTitle
 178  * @stable ICU 3.8
 179  */
 180 U_STABLE void U_EXPORT2
 181 ucasemap_setBreakIterator(UCaseMap *csm, UBreakIterator *iterToAdopt, UErrorCode *pErrorCode);
 182
 183 /**
 184  * Titlecase a UTF-16 string. This function is almost a duplicate of u_strToTitle(),
 185  * except that it takes ucasemap_setOptions() into account and has performance
 186  * advantages from being able to use a UCaseMap object for multiple case mapping
 187  * operations, saving setup time.
 188  *
 189  * Casing is locale-dependent and context-sensitive.
 190  * Titlecasing uses a break iterator to find the first characters of words
 191  * that are to be titlecased. It titlecases those characters and lowercases
 192  * all others. (This can be modified with ucasemap_setOptions().)
 193  *
 194  * Note: This function takes a non-const UCaseMap pointer because it will
 195  * open a default break iterator if no break iterator was set yet,
 196  * and effectively call ucasemap_setBreakIterator();
 197  * also because the break iterator is stateful and will be modified during
 198  * the iteration.
 199  *
 200  * The titlecase break iterator can be provided to customize for arbitrary
 201  * styles, using rules and dictionaries beyond the standard iterators.
 202  * The standard titlecase iterator for the root locale implements the
 203  * algorithm of Unicode TR 21.
 204  *
 205  * This function uses only the setText(), first() and next() methods of the
 206  * provided break iterator.
 207  *
 208  * The result may be longer or shorter than the original.
 209  * The source string and the destination buffer must not overlap.
 210  *
 211  * @param csm       UCaseMap service object. This pointer is non-const!
 212  *                  See the note above for details.
 213  * @param dest      A buffer for the result string. The result will be NUL-terminated if
 214  *                  the buffer is large enough.
 215  *                  The contents is undefined in case of failure.
 216  * @param destCapacity The size of the buffer (number of UChars). If it is 0, then
 217  *                  dest may be NULL and the function will only return the length of the result
 218  *                  without writing any of the result string.
 219  * @param src       The original string.
 220  * @param srcLength The length of the original string. If -1, then src must be NUL-terminated.
 221  * @param pErrorCode Must be a valid pointer to an error code value,
 222  *                  which must not indicate a failure before the function call.
 223  * @return The length of the result string, if successful - or in case of a buffer overflow,
 224  *         in which case it will be greater than destCapacity.
 225  *
 226  * @see u_strToTitle
 227  * @stable ICU 3.8
 228  */
 229 U_STABLE int32_t U_EXPORT2
 230 ucasemap_toTitle(UCaseMap *csm,
 231                  UChar *dest, int32_t destCapacity,
 232                  const UChar *src, int32_t srcLength,
 233                  UErrorCode *pErrorCode);
 234
 235 #endif  // UCONFIG_NO_BREAK_ITERATION
 236
 237 /**
 238  * Lowercase the characters in a UTF-8 string.
 239  * Casing is locale-dependent and context-sensitive.
 240  * The result may be longer or shorter than the original.
 241  * The source string and the destination buffer must not overlap.
 242  *
 243  * @param csm       UCaseMap service object.
 244  * @param dest      A buffer for the result string. The result will be NUL-terminated if
 245  *                  the buffer is large enough.
 246  *                  The contents is undefined in case of failure.
 247  * @param destCapacity The size of the buffer (number of bytes). If it is 0, then
 248  *                  dest may be NULL and the function will only return the length of the result
 249  *                  without writing any of the result string.
 250  * @param src       The original string.
 251  * @param srcLength The length of the original string. If -1, then src must be NUL-terminated.
 252  * @param pErrorCode Must be a valid pointer to an error code value,
 253  *                  which must not indicate a failure before the function call.
 254  * @return The length of the result string, if successful - or in case of a buffer overflow,
 255  *         in which case it will be greater than destCapacity.
 256  *
 257  * @see u_strToLower
 258  * @stable ICU 3.4
 259  */
 260 U_STABLE int32_t U_EXPORT2
 261 ucasemap_utf8ToLower(const UCaseMap *csm,
 262                      char *dest, int32_t destCapacity,
 263                      const char *src, int32_t srcLength,
 264                      UErrorCode *pErrorCode);
 265
 266 /**
 267  * Uppercase the characters in a UTF-8 string.
 268  * Casing is locale-dependent and context-sensitive.
 269  * The result may be longer or shorter than the original.
 270  * The source string and the destination buffer must not overlap.
 271  *
 272  * @param csm       UCaseMap service object.
 273  * @param dest      A buffer for the result string. The result will be NUL-terminated if
 274  *                  the buffer is large enough.
 275  *                  The contents is undefined in case of failure.
 276  * @param destCapacity The size of the buffer (number of bytes). If it is 0, then
 277  *                  dest may be NULL and the function will only return the length of the result
 278  *                  without writing any of the result string.
 279  * @param src       The original string.
 280  * @param srcLength The length of the original string. If -1, then src must be NUL-terminated.
 281  * @param pErrorCode Must be a valid pointer to an error code value,
 282  *                  which must not indicate a failure before the function call.
 283  * @return The length of the result string, if successful - or in case of a buffer overflow,
 284  *         in which case it will be greater than destCapacity.
 285  *
 286  * @see u_strToUpper
 287  * @stable ICU 3.4
 288  */
 289 U_STABLE int32_t U_EXPORT2
 290 ucasemap_utf8ToUpper(const UCaseMap *csm,
 291                      char *dest, int32_t destCapacity,
 292                      const char *src, int32_t srcLength,
 293                      UErrorCode *pErrorCode);
 294
 295 #if !UCONFIG_NO_BREAK_ITERATION
 296
 297 /**
 298  * Titlecase a UTF-8 string.
 299  * Casing is locale-dependent and context-sensitive.
 300  * Titlecasing uses a break iterator to find the first characters of words
 301  * that are to be titlecased. It titlecases those characters and lowercases
 302  * all others. (This can be modified with ucasemap_setOptions().)
 303  *
 304  * Note: This function takes a non-const UCaseMap pointer because it will
 305  * open a default break iterator if no break iterator was set yet,
 306  * and effectively call ucasemap_setBreakIterator();
 307  * also because the break iterator is stateful and will be modified during
 308  * the iteration.
 309  *
 310  * The titlecase break iterator can be provided to customize for arbitrary
 311  * styles, using rules and dictionaries beyond the standard iterators.
 312  * The standard titlecase iterator for the root locale implements the
 313  * algorithm of Unicode TR 21.
 314  *
 315  * This function uses only the setUText(), first(), next() and close() methods of the
 316  * provided break iterator.
 317  *
 318  * The result may be longer or shorter than the original.
 319  * The source string and the destination buffer must not overlap.
 320  *
 321  * @param csm       UCaseMap service object. This pointer is non-const!
 322  *                  See the note above for details.
 323  * @param dest      A buffer for the result string. The result will be NUL-terminated if
 324  *                  the buffer is large enough.
 325  *                  The contents is undefined in case of failure.
 326  * @param destCapacity The size of the buffer (number of bytes). If it is 0, then
 327  *                  dest may be NULL and the function will only return the length of the result
 328  *                  without writing any of the result string.
 329  * @param src       The original string.
 330  * @param srcLength The length of the original string. If -1, then src must be NUL-terminated.
 331  * @param pErrorCode Must be a valid pointer to an error code value,
 332  *                  which must not indicate a failure before the function call.
 333  * @return The length of the result string, if successful - or in case of a buffer overflow,
 334  *         in which case it will be greater than destCapacity.
 335  *
 336  * @see u_strToTitle
 337  * @see U_TITLECASE_NO_LOWERCASE
 338  * @see U_TITLECASE_NO_BREAK_ADJUSTMENT
 339  * @stable ICU 3.8
 340  */
 341 U_STABLE int32_t U_EXPORT2
 342 ucasemap_utf8ToTitle(UCaseMap *csm,
 343                     char *dest, int32_t destCapacity,
 344                     const char *src, int32_t srcLength,
 345                     UErrorCode *pErrorCode);
 346
 347 #endif
 348
 349 /**
 350  * Case-folds the characters in a UTF-8 string.
 351  *
 352  * Case-folding is locale-independent and not context-sensitive,
 353  * but there is an option for whether to include or exclude mappings for dotted I
 354  * and dotless i that are marked with 'T' in CaseFolding.txt.
 355  *
 356  * The result may be longer or shorter than the original.
 357  * The source string and the destination buffer must not overlap.
 358  *
 359  * @param csm       UCaseMap service object.
 360  * @param dest      A buffer for the result string. The result will be NUL-terminated if
 361  *                  the buffer is large enough.
 362  *                  The contents is undefined in case of failure.
 363  * @param destCapacity The size of the buffer (number of bytes). If it is 0, then
 364  *                  dest may be NULL and the function will only return the length of the result
 365  *                  without writing any of the result string.
 366  * @param src       The original string.
 367  * @param srcLength The length of the original string. If -1, then src must be NUL-terminated.
 368  * @param pErrorCode Must be a valid pointer to an error code value,
 369  *                  which must not indicate a failure before the function call.
 370  * @return The length of the result string, if successful - or in case of a buffer overflow,
 371  *         in which case it will be greater than destCapacity.
 372  *
 373  * @see u_strFoldCase
 374  * @see ucasemap_setOptions
 375  * @see U_FOLD_CASE_DEFAULT
 376  * @see U_FOLD_CASE_EXCLUDE_SPECIAL_I
 377  * @stable ICU 3.8
 378  */
 379 U_STABLE int32_t U_EXPORT2
 380 ucasemap_utf8FoldCase(const UCaseMap *csm,
 381                       char *dest, int32_t destCapacity,
 382                       const char *src, int32_t srcLength,
 383                       UErrorCode *pErrorCode);
 384
 385 #endif