icuSources/i18n/unicode/ucoleitr.h

   1 /*
   2 *******************************************************************************
   3 *   Copyright (C) 2001-2014, International Business Machines
   4 *   Corporation and others.  All Rights Reserved.
   5 *******************************************************************************
   6 *
   7 * File ucoleitr.h
   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  * DO NOT USE, INTERNAL CONSTANT THAT WAS REMOVED AND THEN
  33  * TEMPORARILY RESTORED TO PREVENT THE BUILD FROM BREAKING.
  34  * This indicates an error has occured during processing or there are no more CEs
  35  * to be returned.
  36  *
  37  * @internal
  38  */
  39 #define UCOL_PROCESSED_NULLORDER        ((int64_t)U_INT64_MAX)
  40 #endif  /* U_HIDE_INTERNAL_API */
  41
  42 #include "unicode/ucol.h"
  43
  44 /**
  45  * The UCollationElements struct.
  46  * For usage in C programs.
  47  * @stable ICU 2.0
  48  */
  49 typedef struct UCollationElements UCollationElements;
  50
  51 /**
  52  * \file
  53  * \brief C API: UCollationElements
  54  *
  55  * The UCollationElements API is used as an iterator to walk through each
  56  * character of an international string. Use the iterator to return the
  57  * ordering priority of the positioned character. The ordering priority of a
  58  * character, which we refer to as a key, defines how a character is collated
  59  * in the given collation object.
  60  * For example, consider the following in Slovak and in traditional Spanish collation:
  61  * <pre>
  62  * .       "ca" -> the first key is key('c') and second key is key('a').
  63  * .       "cha" -> the first key is key('ch') and second key is key('a').
  64  * </pre>
  65  * And in German phonebook collation,
  66  * <pre>
  67  * .       "<ae ligature>b"-> the first key is key('a'), the second key is key('e'), and
  68  * .       the third key is key('b').
  69  * </pre>
  70  * <p>Example of the iterator usage: (without error checking)
  71  * <pre>
  72  * .  void CollationElementIterator_Example()
  73  * .  {
  74  * .      UChar *s;
  75  * .      t_int32 order, primaryOrder;
  76  * .      UCollationElements *c;
  77  * .      UCollatorOld *coll;
  78  * .      UErrorCode success = U_ZERO_ERROR;
  79  * .      s=(UChar*)malloc(sizeof(UChar) * (strlen("This is a test")+1) );
  80  * .      u_uastrcpy(s, "This is a test");
  81  * .      coll = ucol_open(NULL, &success);
  82  * .      c = ucol_openElements(coll, str, u_strlen(str), &status);
  83  * .      order = ucol_next(c, &success);
  84  * .      ucol_reset(c);
  85  * .      order = ucol_prev(c, &success);
  86  * .      free(s);
  87  * .      ucol_close(coll);
  88  * .      ucol_closeElements(c);
  89  * .  }
  90  * </pre>
  91  * <p>
  92  * ucol_next() returns the collation order of the next.
  93  * ucol_prev() returns the collation order of the previous character.
  94  * The Collation Element Iterator moves only in one direction between calls to
  95  * ucol_reset. That is, ucol_next() and ucol_prev can not be inter-used.
  96  * Whenever ucol_prev is to be called after ucol_next() or vice versa,
  97  * ucol_reset has to be called first to reset the status, shifting pointers to
  98  * either the end or the start of the string. Hence at the next call of
  99  * ucol_prev or ucol_next, the first or last collation order will be returned.
 100  * If a change of direction is done without a ucol_reset, the result is
 101  * undefined.
 102  * The result of a forward iterate (ucol_next) and reversed result of the
 103  * backward iterate (ucol_prev) on the same string are equivalent, if
 104  * collation orders with the value 0 are ignored.
 105  * Character based on the comparison level of the collator.  A collation order
 106  * consists of primary order, secondary order and tertiary order.  The data
 107  * type of the collation order is <strong>int32_t</strong>.
 108  *
 109  * @see UCollator
 110  */
 111
 112 /**
 113  * Open the collation elements for a string.
 114  *
 115  * @param coll The collator containing the desired collation rules.
 116  * @param text The text to iterate over.
 117  * @param textLength The number of characters in text, or -1 if null-terminated
 118  * @param status A pointer to a UErrorCode to receive any errors.
 119  * @return a struct containing collation element information
 120  * @stable ICU 2.0
 121  */
 122 U_STABLE UCollationElements* U_EXPORT2
 123 ucol_openElements(const UCollator  *coll,
 124                   const UChar      *text,
 125                         int32_t    textLength,
 126                         UErrorCode *status);
 127
 128
 129 /**
 130  * get a hash code for a key... Not very useful!
 131  * @param key    the given key.
 132  * @param length the size of the key array.
 133  * @return       the hash code.
 134  * @stable ICU 2.0
 135  */
 136 U_STABLE int32_t U_EXPORT2
 137 ucol_keyHashCode(const uint8_t* key, int32_t length);
 138
 139 /**
 140  * Close a UCollationElements.
 141  * Once closed, a UCollationElements may no longer be used.
 142  * @param elems The UCollationElements to close.
 143  * @stable ICU 2.0
 144  */
 145 U_STABLE void U_EXPORT2
 146 ucol_closeElements(UCollationElements *elems);
 147
 148 /**
 149  * Reset the collation elements to their initial state.
 150  * This will move the 'cursor' to the beginning of the text.
 151  * Property settings for collation will be reset to the current status.
 152  * @param elems The UCollationElements to reset.
 153  * @see ucol_next
 154  * @see ucol_previous
 155  * @stable ICU 2.0
 156  */
 157 U_STABLE void U_EXPORT2
 158 ucol_reset(UCollationElements *elems);
 159
 160 /**
 161  * Get the ordering priority of the next collation element in the text.
 162  * A single character may contain more than one collation element.
 163  * @param elems The UCollationElements containing the text.
 164  * @param status A pointer to a UErrorCode to receive any errors.
 165  * @return The next collation elements ordering, otherwise returns NULLORDER
 166  *         if an error has occured or if the end of string has been reached
 167  * @stable ICU 2.0
 168  */
 169 U_STABLE int32_t U_EXPORT2
 170 ucol_next(UCollationElements *elems, UErrorCode *status);
 171
 172 /**
 173  * Get the ordering priority of the previous collation element in the text.
 174  * A single character may contain more than one collation element.
 175  * Note that internally a stack is used to store buffered collation elements.
 176  * @param elems The UCollationElements containing the text.
 177  * @param status A pointer to a UErrorCode to receive any errors. Noteably
 178  *               a U_BUFFER_OVERFLOW_ERROR is returned if the internal stack
 179  *               buffer has been exhausted.
 180  * @return The previous collation elements ordering, otherwise returns
 181  *         NULLORDER if an error has occured or if the start of string has
 182  *         been reached.
 183  * @stable ICU 2.0
 184  */
 185 U_STABLE int32_t U_EXPORT2
 186 ucol_previous(UCollationElements *elems, UErrorCode *status);
 187
 188 #ifndef U_HIDE_INTERNAL_API
 189 /**
 190  * DO NOT USE, INTERNAL FUNCTION THAT WAS REMOVED AND THEN
 191  * TEMPORARILY RESTORED TO PREVENT THE BUILD FROM BREAKING.
 192  * Get the processed ordering priority of the next collation element in the text.
 193  * A single character may contain more than one collation element.
 194  *
 195  * @param elems The UCollationElements containing the text.
 196  * @param ixLow a pointer to an int32_t to receive the iterator index before fetching the CE.
 197  * @param ixHigh a pointer to an int32_t to receive the iterator index after fetching the CE.
 198  * @param status A pointer to an UErrorCode to receive any errors.
 199  * @return The next collation elements ordering, otherwise returns UCOL_PROCESSED_NULLORDER
 200  *         if an error has occured or if the end of string has been reached
 201  *
 202  * @internal
 203  */
 204 U_INTERNAL int64_t U_EXPORT2
 205 ucol_nextProcessed(UCollationElements *elems, int32_t *ixLow, int32_t *ixHigh, UErrorCode *status);
 206
 207 /**
 208  * DO NOT USE, INTERNAL FUNCTION THAT WAS REMOVED AND THEN
 209  * TEMPORARILY RESTORED TO PREVENT THE BUILD FROM BREAKING.
 210  * Get the processed ordering priority of the previous collation element in the text.
 211  * A single character may contain more than one collation element.
 212  * Note that internally a stack is used to store buffered collation elements.
 213  * It is very rare that the stack will overflow, however if such a case is
 214  * encountered, the problem can be solved by increasing the size
 215  * UCOL_EXPAND_CE_BUFFER_SIZE in ucol_imp.h.
 216  *
 217  * @param elems The UCollationElements containing the text.
 218  * @param ixLow A pointer to an int32_t to receive the iterator index after fetching the CE
 219  * @param ixHigh A pointer to an int32_t to receiver the iterator index before fetching the CE
 220  * @param status A pointer to an UErrorCode to receive any errors. Noteably
 221  *               a U_BUFFER_OVERFLOW_ERROR is returned if the internal stack
 222  *               buffer has been exhausted.
 223  * @return The previous collation elements ordering, otherwise returns
 224  *         UCOL_PROCESSED_NULLORDER if an error has occured or if the start of
 225  *         string has been reached.
 226  *
 227  * @internal
 228  */
 229 U_INTERNAL int64_t U_EXPORT2
 230 ucol_previousProcessed(UCollationElements *elems, int32_t *ixLow, int32_t *ixHigh, UErrorCode *status);
 231 #endif  /* U_HIDE_INTERNAL_API */
 232
 233 /**
 234  * Get the maximum length of any expansion sequences that end with the
 235  * specified comparison order.
 236  * This is useful for .... ?
 237  * @param elems The UCollationElements containing the text.
 238  * @param order A collation order returned by previous or next.
 239  * @return maximum size of the expansion sequences ending with the collation
 240  *         element or 1 if collation element does not occur at the end of any
 241  *         expansion sequence
 242  * @stable ICU 2.0
 243  */
 244 U_STABLE int32_t U_EXPORT2
 245 ucol_getMaxExpansion(const UCollationElements *elems, int32_t order);
 246
 247 /**
 248  * Set the text containing the collation elements.
 249  * Property settings for collation will remain the same.
 250  * In order to reset the iterator to the current collation property settings,
 251  * the API reset() has to be called.
 252  * @param elems The UCollationElements to set.
 253  * @param text The source text containing the collation elements.
 254  * @param textLength The length of text, or -1 if null-terminated.
 255  * @param status A pointer to a UErrorCode to receive any errors.
 256  * @see ucol_getText
 257  * @stable ICU 2.0
 258  */
 259 U_STABLE void U_EXPORT2
 260 ucol_setText(      UCollationElements *elems,
 261              const UChar              *text,
 262                    int32_t            textLength,
 263                    UErrorCode         *status);
 264
 265 /**
 266  * Get the offset of the current source character.
 267  * This is an offset into the text of the character containing the current
 268  * collation elements.
 269  * @param elems The UCollationElements to query.
 270  * @return The offset of the current source character.
 271  * @see ucol_setOffset
 272  * @stable ICU 2.0
 273  */
 274 U_STABLE int32_t U_EXPORT2
 275 ucol_getOffset(const UCollationElements *elems);
 276
 277 /**
 278  * Set the offset of the current source character.
 279  * This is an offset into the text of the character to be processed.
 280  * Property settings for collation will remain the same.
 281  * In order to reset the iterator to the current collation property settings,
 282  * the API reset() has to be called.
 283  * @param elems The UCollationElements to set.
 284  * @param offset The desired character offset.
 285  * @param status A pointer to a UErrorCode to receive any errors.
 286  * @see ucol_getOffset
 287  * @stable ICU 2.0
 288  */
 289 U_STABLE void U_EXPORT2
 290 ucol_setOffset(UCollationElements *elems,
 291                int32_t        offset,
 292                UErrorCode         *status);
 293
 294 /**
 295 * Get the primary order of a collation order.
 296 * @param order the collation order
 297 * @return the primary order of a collation order.
 298 * @stable ICU 2.6
 299 */
 300 U_STABLE int32_t U_EXPORT2
 301 ucol_primaryOrder (int32_t order);
 302
 303 /**
 304 * Get the secondary order of a collation order.
 305 * @param order the collation order
 306 * @return the secondary order of a collation order.
 307 * @stable ICU 2.6
 308 */
 309 U_STABLE int32_t U_EXPORT2
 310 ucol_secondaryOrder (int32_t order);
 311
 312 /**
 313 * Get the tertiary order of a collation order.
 314 * @param order the collation order
 315 * @return the tertiary order of a collation order.
 316 * @stable ICU 2.6
 317 */
 318 U_STABLE int32_t U_EXPORT2
 319 ucol_tertiaryOrder (int32_t order);
 320
 321 #endif /* #if !UCONFIG_NO_COLLATION */
 322
 323 #endif