icuSources/i18n/unicode/ucoleitr.h

   1 /*
   2 *******************************************************************************
   3 *   Copyright (C) 2001-2011, International Business Machines
   4 *   Corporation and others.  All Rights Reserved.
   5 *******************************************************************************
   6 *
   7 * File ucoleitr.cpp
   8 *
   9 * Modification History:
  10 *
  11 * Date        Name        Description
  12 * 02/15/2001  synwee      Modified all methods to process its own function
  13 *                         instead of calling the equivalent c++ api (coleitr.h)
  14 *******************************************************************************/
  15
  16 #ifndef UCOLEITR_H
  17 #define UCOLEITR_H
  18
  19 #include "unicode/utypes.h"
  20
  21 #if !UCONFIG_NO_COLLATION
  22
  23 /**
  24  * This indicates an error has occured during processing or if no more CEs is
  25  * to be returned.
  26  * @stable ICU 2.0
  27  */
  28 #define UCOL_NULLORDER        ((int32_t)0xFFFFFFFF)
  29
  30 #ifndef U_HIDE_INTERNAL_API
  31 /**
  32  * This indicates an error has occured during processing or there are no more CEs
  33  * to be returned.
  34  *
  35  * @internal
  36  */
  37 #define UCOL_PROCESSED_NULLORDER        ((int64_t)U_INT64_MAX)
  38 #endif  /* U_HIDE_INTERNAL_API */
  39
  40 #include "unicode/ucol.h"
  41
  42 /**
  43  * The UCollationElements struct.
  44  * For usage in C programs.
  45  * @stable ICU 2.0
  46  */
  47 typedef struct UCollationElements UCollationElements;
  48
  49 /**
  50  * \file
  51  * \brief C API: UCollationElements
  52  *
  53  * The UCollationElements API is used as an iterator to walk through each
  54  * character of an international string. Use the iterator to return the
  55  * ordering priority of the positioned character. The ordering priority of a
  56  * character, which we refer to as a key, defines how a character is collated
  57  * in the given collation object.
  58  * For example, consider the following in Spanish:
  59  * <pre>
  60  * .       "ca" -> the first key is key('c') and second key is key('a').
  61  * .       "cha" -> the first key is key('ch') and second key is key('a').
  62  * </pre>
  63  * And in German,
  64  * <pre>
  65  * .       "<ae ligature>b"-> the first key is key('a'), the second key is key('e'), and
  66  * .       the third key is key('b').
  67  * </pre>
  68  * <p>Example of the iterator usage: (without error checking)
  69  * <pre>
  70  * .  void CollationElementIterator_Example()
  71  * .  {
  72  * .      UChar *s;
  73  * .      t_int32 order, primaryOrder;
  74  * .      UCollationElements *c;
  75  * .      UCollatorOld *coll;
  76  * .      UErrorCode success = U_ZERO_ERROR;
  77  * .      s=(UChar*)malloc(sizeof(UChar) * (strlen("This is a test")+1) );
  78  * .      u_uastrcpy(s, "This is a test");
  79  * .      coll = ucol_open(NULL, &success);
  80  * .      c = ucol_openElements(coll, str, u_strlen(str), &status);
  81  * .      order = ucol_next(c, &success);
  82  * .      ucol_reset(c);
  83  * .      order = ucol_prev(c, &success);
  84  * .      free(s);
  85  * .      ucol_close(coll);
  86  * .      ucol_closeElements(c);
  87  * .  }
  88  * </pre>
  89  * <p>
  90  * ucol_next() returns the collation order of the next.
  91  * ucol_prev() returns the collation order of the previous character.
  92  * The Collation Element Iterator moves only in one direction between calls to
  93  * ucol_reset. That is, ucol_next() and ucol_prev can not be inter-used.
  94  * Whenever ucol_prev is to be called after ucol_next() or vice versa,
  95  * ucol_reset has to be called first to reset the status, shifting pointers to
  96  * either the end or the start of the string. Hence at the next call of
  97  * ucol_prev or ucol_next, the first or last collation order will be returned.
  98  * If a change of direction is done without a ucol_reset, the result is
  99  * undefined.
 100  * The result of a forward iterate (ucol_next) and reversed result of the
 101  * backward iterate (ucol_prev) on the same string are equivalent, if
 102  * collation orders with the value UCOL_IGNORABLE are ignored.
 103  * Character based on the comparison level of the collator.  A collation order
 104  * consists of primary order, secondary order and tertiary order.  The data
 105  * type of the collation order is <strong>t_int32</strong>.
 106  *
 107  * @see UCollator
 108  */
 109
 110 /**
 111  * Open the collation elements for a string.
 112  *
 113  * @param coll The collator containing the desired collation rules.
 114  * @param text The text to iterate over.
 115  * @param textLength The number of characters in text, or -1 if null-terminated
 116  * @param status A pointer to an UErrorCode to receive any errors.
 117  * @return a struct containing collation element information
 118  * @stable ICU 2.0
 119  */
 120 U_STABLE UCollationElements* U_EXPORT2
 121 ucol_openElements(const UCollator  *coll,
 122                   const UChar      *text,
 123                         int32_t    textLength,
 124                         UErrorCode *status);
 125
 126
 127 /**
 128  * get a hash code for a key... Not very useful!
 129  * @param key    the given key.
 130  * @param length the size of the key array.
 131  * @return       the hash code.
 132  * @stable ICU 2.0
 133  */
 134 U_STABLE int32_t U_EXPORT2
 135 ucol_keyHashCode(const uint8_t* key, int32_t length);
 136
 137 /**
 138  * Close a UCollationElements.
 139  * Once closed, a UCollationElements may no longer be used.
 140  * @param elems The UCollationElements to close.
 141  * @stable ICU 2.0
 142  */
 143 U_STABLE void U_EXPORT2
 144 ucol_closeElements(UCollationElements *elems);
 145
 146 /**
 147  * Reset the collation elements to their initial state.
 148  * This will move the 'cursor' to the beginning of the text.
 149  * Property settings for collation will be reset to the current status.
 150  * @param elems The UCollationElements to reset.
 151  * @see ucol_next
 152  * @see ucol_previous
 153  * @stable ICU 2.0
 154  */
 155 U_STABLE void U_EXPORT2
 156 ucol_reset(UCollationElements *elems);
 157
 158 #ifndef U_HIDE_INTERNAL_API
 159 /**
 160  * Set the collation elements to use implicit ordering for Han
 161  * even if they've been tailored. This will also force Hangul
 162  * syllables to be ordered by decomposing them to their component
 163  * Jamo.
 164  *
 165  * @param elems The UCollationElements containing the text.
 166  * @param status A pointer to a UErrorCode to reveive any errors.
 167  *
 168  * @internal
 169  */
 170 U_INTERNAL void U_EXPORT2
 171 ucol_forceHanImplicit(UCollationElements *elems, UErrorCode *status);
 172 #endif  /* U_HIDE_INTERNAL_API */
 173
 174 /**
 175  * Get the ordering priority of the next collation element in the text.
 176  * A single character may contain more than one collation element.
 177  * @param elems The UCollationElements containing the text.
 178  * @param status A pointer to an UErrorCode to receive any errors.
 179  * @return The next collation elements ordering, otherwise returns NULLORDER
 180  *         if an error has occured or if the end of string has been reached
 181  * @stable ICU 2.0
 182  */
 183 U_STABLE int32_t U_EXPORT2
 184 ucol_next(UCollationElements *elems, UErrorCode *status);
 185
 186 /**
 187  * Get the ordering priority of the previous collation element in the text.
 188  * A single character may contain more than one collation element.
 189  * Note that internally a stack is used to store buffered collation elements.
 190  * It is very rare that the stack will overflow, however if such a case is
 191  * encountered, the problem can be solved by increasing the size
 192  * UCOL_EXPAND_CE_BUFFER_SIZE in ucol_imp.h.
 193  * @param elems The UCollationElements containing the text.
 194  * @param status A pointer to an UErrorCode to receive any errors. Noteably
 195  *               a U_BUFFER_OVERFLOW_ERROR is returned if the internal stack
 196  *               buffer has been exhausted.
 197  * @return The previous collation elements ordering, otherwise returns
 198  *         NULLORDER if an error has occured or if the start of string has
 199  *         been reached.
 200  * @stable ICU 2.0
 201  */
 202 U_STABLE int32_t U_EXPORT2
 203 ucol_previous(UCollationElements *elems, UErrorCode *status);
 204
 205 #ifndef U_HIDE_INTERNAL_API
 206 /**
 207  * Get the processed ordering priority of the next collation element in the text.
 208  * A single character may contain more than one collation element.
 209  *
 210  * @param elems The UCollationElements containing the text.
 211  * @param ixLow a pointer to an int32_t to receive the iterator index before fetching the CE.
 212  * @param ixHigh a pointer to an int32_t to receive the iterator index after fetching the CE.
 213  * @param status A pointer to an UErrorCode to receive any errors.
 214  * @return The next collation elements ordering, otherwise returns UCOL_PROCESSED_NULLORDER
 215  *         if an error has occured or if the end of string has been reached
 216  *
 217  * @internal
 218  */
 219 U_INTERNAL int64_t U_EXPORT2
 220 ucol_nextProcessed(UCollationElements *elems, int32_t *ixLow, int32_t *ixHigh, UErrorCode *status);
 221
 222 /**
 223  * Get the processed ordering priority of the previous collation element in the text.
 224  * A single character may contain more than one collation element.
 225  * Note that internally a stack is used to store buffered collation elements.
 226  * It is very rare that the stack will overflow, however if such a case is
 227  * encountered, the problem can be solved by increasing the size
 228  * UCOL_EXPAND_CE_BUFFER_SIZE in ucol_imp.h.
 229  *
 230  * @param elems The UCollationElements containing the text.
 231  * @param ixLow A pointer to an int32_t to receive the iterator index after fetching the CE
 232  * @param ixHigh A pointer to an int32_t to receiver the iterator index before fetching the CE
 233  * @param status A pointer to an UErrorCode to receive any errors. Noteably
 234  *               a U_BUFFER_OVERFLOW_ERROR is returned if the internal stack
 235  *               buffer has been exhausted.
 236  * @return The previous collation elements ordering, otherwise returns
 237  *         UCOL_PROCESSED_NULLORDER if an error has occured or if the start of
 238  *         string has been reached.
 239  *
 240  * @internal
 241  */
 242 U_INTERNAL int64_t U_EXPORT2
 243 ucol_previousProcessed(UCollationElements *elems, int32_t *ixLow, int32_t *ixHigh, UErrorCode *status);
 244 #endif  /* U_HIDE_INTERNAL_API */
 245
 246 /**
 247  * Get the maximum length of any expansion sequences that end with the
 248  * specified comparison order.
 249  * This is useful for .... ?
 250  * @param elems The UCollationElements containing the text.
 251  * @param order A collation order returned by previous or next.
 252  * @return maximum size of the expansion sequences ending with the collation
 253  *         element or 1 if collation element does not occur at the end of any
 254  *         expansion sequence
 255  * @stable ICU 2.0
 256  */
 257 U_STABLE int32_t U_EXPORT2
 258 ucol_getMaxExpansion(const UCollationElements *elems, int32_t order);
 259
 260 /**
 261  * Set the text containing the collation elements.
 262  * Property settings for collation will remain the same.
 263  * In order to reset the iterator to the current collation property settings,
 264  * the API reset() has to be called.
 265  * @param elems The UCollationElements to set.
 266  * @param text The source text containing the collation elements.
 267  * @param textLength The length of text, or -1 if null-terminated.
 268  * @param status A pointer to an UErrorCode to receive any errors.
 269  * @see ucol_getText
 270  * @stable ICU 2.0
 271  */
 272 U_STABLE void U_EXPORT2
 273 ucol_setText(      UCollationElements *elems,
 274              const UChar              *text,
 275                    int32_t            textLength,
 276                    UErrorCode         *status);
 277
 278 /**
 279  * Get the offset of the current source character.
 280  * This is an offset into the text of the character containing the current
 281  * collation elements.
 282  * @param elems The UCollationElements to query.
 283  * @return The offset of the current source character.
 284  * @see ucol_setOffset
 285  * @stable ICU 2.0
 286  */
 287 U_STABLE int32_t U_EXPORT2
 288 ucol_getOffset(const UCollationElements *elems);
 289
 290 /**
 291  * Set the offset of the current source character.
 292  * This is an offset into the text of the character to be processed.
 293  * Property settings for collation will remain the same.
 294  * In order to reset the iterator to the current collation property settings,
 295  * the API reset() has to be called.
 296  * @param elems The UCollationElements to set.
 297  * @param offset The desired character offset.
 298  * @param status A pointer to an UErrorCode to receive any errors.
 299  * @see ucol_getOffset
 300  * @stable ICU 2.0
 301  */
 302 U_STABLE void U_EXPORT2
 303 ucol_setOffset(UCollationElements *elems,
 304                int32_t        offset,
 305                UErrorCode         *status);
 306
 307 /**
 308 * Get the primary order of a collation order.
 309 * @param order the collation order
 310 * @return the primary order of a collation order.
 311 * @stable ICU 2.6
 312 */
 313 U_STABLE int32_t U_EXPORT2
 314 ucol_primaryOrder (int32_t order);
 315
 316 /**
 317 * Get the secondary order of a collation order.
 318 * @param order the collation order
 319 * @return the secondary order of a collation order.
 320 * @stable ICU 2.6
 321 */
 322 U_STABLE int32_t U_EXPORT2
 323 ucol_secondaryOrder (int32_t order);
 324
 325 /**
 326 * Get the tertiary order of a collation order.
 327 * @param order the collation order
 328 * @return the tertiary order of a collation order.
 329 * @stable ICU 2.6
 330 */
 331 U_STABLE int32_t U_EXPORT2
 332 ucol_tertiaryOrder (int32_t order);
 333
 334 #endif /* #if !UCONFIG_NO_COLLATION */
 335
 336 #endif