icuSources/i18n/unicode/ucoleitr.h

   1 /*
   2 *******************************************************************************
   3 *   Copyright (C) 2001-2008, 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 /**
  31  * This indicates an error has occured during processing or there are no more CEs
  32  * to be returned.
  33  *
  34  * @internal
  35  */
  36 #define UCOL_PROCESSED_NULLORDER        ((int64_t)U_INT64_MAX)
  37
  38 #include "unicode/ucol.h"
  39
  40 /**
  41  * The UCollationElements struct.
  42  * For usage in C programs.
  43  * @stable ICU 2.0
  44  */
  45 typedef struct UCollationElements UCollationElements;
  46
  47 /**
  48  * \file
  49  * \brief C API: UCollationElements
  50  *
  51  * The UCollationElements API is used as an iterator to walk through each
  52  * character of an international string. Use the iterator to return the
  53  * ordering priority of the positioned character. The ordering priority of a
  54  * character, which we refer to as a key, defines how a character is collated
  55  * in the given collation object.
  56  * For example, consider the following in Spanish:
  57  * <pre>
  58  * .       "ca" -> the first key is key('c') and second key is key('a').
  59  * .       "cha" -> the first key is key('ch') and second key is key('a').
  60  * </pre>
  61  * And in German,
  62  * <pre>
  63  * .       "<ae ligature>b"-> the first key is key('a'), the second key is key('e'), and
  64  * .       the third key is key('b').
  65  * </pre>
  66  * <p>Example of the iterator usage: (without error checking)
  67  * <pre>
  68  * .  void CollationElementIterator_Example()
  69  * .  {
  70  * .      UChar *s;
  71  * .      t_int32 order, primaryOrder;
  72  * .      UCollationElements *c;
  73  * .      UCollatorOld *coll;
  74  * .      UErrorCode success = U_ZERO_ERROR;
  75  * .      s=(UChar*)malloc(sizeof(UChar) * (strlen("This is a test")+1) );
  76  * .      u_uastrcpy(s, "This is a test");
  77  * .      coll = ucol_open(NULL, &success);
  78  * .      c = ucol_openElements(coll, str, u_strlen(str), &status);
  79  * .      order = ucol_next(c, &success);
  80  * .      ucol_reset(c);
  81  * .      order = ucol_prev(c, &success);
  82  * .      free(s);
  83  * .      ucol_close(coll);
  84  * .      ucol_closeElements(c);
  85  * .  }
  86  * </pre>
  87  * <p>
  88  * ucol_next() returns the collation order of the next.
  89  * ucol_prev() returns the collation order of the previous character.
  90  * The Collation Element Iterator moves only in one direction between calls to
  91  * ucol_reset. That is, ucol_next() and ucol_prev can not be inter-used.
  92  * Whenever ucol_prev is to be called after ucol_next() or vice versa,
  93  * ucol_reset has to be called first to reset the status, shifting pointers to
  94  * either the end or the start of the string. Hence at the next call of
  95  * ucol_prev or ucol_next, the first or last collation order will be returned.
  96  * If a change of direction is done without a ucol_reset, the result is
  97  * undefined.
  98  * The result of a forward iterate (ucol_next) and reversed result of the
  99  * backward iterate (ucol_prev) on the same string are equivalent, if
 100  * collation orders with the value UCOL_IGNORABLE are ignored.
 101  * Character based on the comparison level of the collator.  A collation order
 102  * consists of primary order, secondary order and tertiary order.  The data
 103  * type of the collation order is <strong>t_int32</strong>.
 104  *
 105  * @see UCollator
 106  */
 107
 108 /**
 109  * Open the collation elements for a string.
 110  *
 111  * @param coll The collator containing the desired collation rules.
 112  * @param text The text to iterate over.
 113  * @param textLength The number of characters in text, or -1 if null-terminated
 114  * @param status A pointer to an UErrorCode to receive any errors.
 115  * @return a struct containing collation element information
 116  * @stable ICU 2.0
 117  */
 118 U_STABLE UCollationElements* U_EXPORT2
 119 ucol_openElements(const UCollator  *coll,
 120                   const UChar      *text,
 121                         int32_t    textLength,
 122                         UErrorCode *status);
 123
 124 /**
 125  * get a hash code for a key... Not very useful!
 126  * @param key    the given key.
 127  * @param length the size of the key array.
 128  * @return       the hash code.
 129  * @stable ICU 2.0
 130  */
 131 U_STABLE int32_t U_EXPORT2
 132 ucol_keyHashCode(const uint8_t* key, int32_t length);
 133
 134 /**
 135  * Close a UCollationElements.
 136  * Once closed, a UCollationElements may no longer be used.
 137  * @param elems The UCollationElements to close.
 138  * @stable ICU 2.0
 139  */
 140 U_STABLE void U_EXPORT2
 141 ucol_closeElements(UCollationElements *elems);
 142
 143 /**
 144  * Reset the collation elements to their initial state.
 145  * This will move the 'cursor' to the beginning of the text.
 146  * Property settings for collation will be reset to the current status.
 147  * @param elems The UCollationElements to reset.
 148  * @see ucol_next
 149  * @see ucol_previous
 150  * @stable ICU 2.0
 151  */
 152 U_STABLE void U_EXPORT2
 153 ucol_reset(UCollationElements *elems);
 154
 155 /**
 156  * Get the ordering priority of the next collation element in the text.
 157  * A single character may contain more than one collation element.
 158  * @param elems The UCollationElements containing the text.
 159  * @param status A pointer to an UErrorCode to receive any errors.
 160  * @return The next collation elements ordering, otherwise returns NULLORDER
 161  *         if an error has occured or if the end of string has been reached
 162  * @stable ICU 2.0
 163  */
 164 U_STABLE int32_t U_EXPORT2
 165 ucol_next(UCollationElements *elems, UErrorCode *status);
 166
 167 /**
 168  * Get the ordering priority of the previous collation element in the text.
 169  * A single character may contain more than one collation element.
 170  * Note that internally a stack is used to store buffered collation elements.
 171  * It is very rare that the stack will overflow, however if such a case is
 172  * encountered, the problem can be solved by increasing the size
 173  * UCOL_EXPAND_CE_BUFFER_SIZE in ucol_imp.h.
 174  * @param elems The UCollationElements containing the text.
 175  * @param status A pointer to an UErrorCode to receive any errors. Noteably
 176  *               a U_BUFFER_OVERFLOW_ERROR is returned if the internal stack
 177  *               buffer has been exhausted.
 178  * @return The previous collation elements ordering, otherwise returns
 179  *         NULLORDER if an error has occured or if the start of string has
 180  *         been reached.
 181  * @stable ICU 2.0
 182  */
 183 U_STABLE int32_t U_EXPORT2
 184 ucol_previous(UCollationElements *elems, UErrorCode *status);
 185
 186 /**
 187  * Get the processed ordering priority of the next collation element in the text.
 188  * A single character may contain more than one collation element.
 189  *
 190  * @param elems The UCollationElements containing the text.
 191  * @param ixLow a pointer to an int32_t to receive the iterator index before fetching the CE.
 192  * @param ixHigh a pointer to an int32_t to receive the iterator index after fetching the CE.
 193  * @param status A pointer to an UErrorCode to receive any errors.
 194  * @return The next collation elements ordering, otherwise returns UCOL_PROCESSED_NULLORDER
 195  *         if an error has occured or if the end of string has been reached
 196  *
 197  * @internal
 198  */
 199 U_INTERNAL int64_t U_EXPORT2
 200 ucol_nextProcessed(UCollationElements *elems, int32_t *ixLow, int32_t *ixHigh, UErrorCode *status);
 201
 202 /**
 203  * Get the processed ordering priority of the previous collation element in the text.
 204  * A single character may contain more than one collation element.
 205  * Note that internally a stack is used to store buffered collation elements.
 206  * It is very rare that the stack will overflow, however if such a case is
 207  * encountered, the problem can be solved by increasing the size
 208  * UCOL_EXPAND_CE_BUFFER_SIZE in ucol_imp.h.
 209  *
 210  * @param elems The UCollationElements containing the text.
 211  * @param ixLow A pointer to an int32_t to receive the iterator index after fetching the CE
 212  * @param ixHigh A pointer to an int32_t to receiver the iterator index before fetching the CE
 213  * @param status A pointer to an UErrorCode to receive any errors. Noteably
 214  *               a U_BUFFER_OVERFLOW_ERROR is returned if the internal stack
 215  *               buffer has been exhausted.
 216  * @return The previous collation elements ordering, otherwise returns
 217  *         UCOL_PROCESSED_NULLORDER if an error has occured or if the start of
 218  *         string has been reached.
 219  *
 220  * @internal
 221  */
 222 U_INTERNAL int64_t U_EXPORT2
 223 ucol_previousProcessed(UCollationElements *elems, int32_t *ixLow, int32_t *ixHigh, UErrorCode *status);
 224
 225 /**
 226  * Get the maximum length of any expansion sequences that end with the
 227  * specified comparison order.
 228  * This is useful for .... ?
 229  * @param elems The UCollationElements containing the text.
 230  * @param order A collation order returned by previous or next.
 231  * @return maximum size of the expansion sequences ending with the collation
 232  *         element or 1 if collation element does not occur at the end of any
 233  *         expansion sequence
 234  * @stable ICU 2.0
 235  */
 236 U_STABLE int32_t U_EXPORT2
 237 ucol_getMaxExpansion(const UCollationElements *elems, int32_t order);
 238
 239 /**
 240  * Set the text containing the collation elements.
 241  * Property settings for collation will remain the same.
 242  * In order to reset the iterator to the current collation property settings,
 243  * the API reset() has to be called.
 244  * @param elems The UCollationElements to set.
 245  * @param text The source text containing the collation elements.
 246  * @param textLength The length of text, or -1 if null-terminated.
 247  * @param status A pointer to an UErrorCode to receive any errors.
 248  * @see ucol_getText
 249  * @stable ICU 2.0
 250  */
 251 U_STABLE void U_EXPORT2
 252 ucol_setText(      UCollationElements *elems,
 253              const UChar              *text,
 254                    int32_t            textLength,
 255                    UErrorCode         *status);
 256
 257 /**
 258  * Get the offset of the current source character.
 259  * This is an offset into the text of the character containing the current
 260  * collation elements.
 261  * @param elems The UCollationElements to query.
 262  * @return The offset of the current source character.
 263  * @see ucol_setOffset
 264  * @stable ICU 2.0
 265  */
 266 U_STABLE int32_t U_EXPORT2
 267 ucol_getOffset(const UCollationElements *elems);
 268
 269 /**
 270  * Set the offset of the current source character.
 271  * This is an offset into the text of the character to be processed.
 272  * Property settings for collation will remain the same.
 273  * In order to reset the iterator to the current collation property settings,
 274  * the API reset() has to be called.
 275  * @param elems The UCollationElements to set.
 276  * @param offset The desired character offset.
 277  * @param status A pointer to an UErrorCode to receive any errors.
 278  * @see ucol_getOffset
 279  * @stable ICU 2.0
 280  */
 281 U_STABLE void U_EXPORT2
 282 ucol_setOffset(UCollationElements *elems,
 283                int32_t        offset,
 284                UErrorCode         *status);
 285
 286 /**
 287 * Get the primary order of a collation order.
 288 * @param order the collation order
 289 * @return the primary order of a collation order.
 290 * @stable ICU 2.6
 291 */
 292 U_STABLE int32_t U_EXPORT2
 293 ucol_primaryOrder (int32_t order);
 294
 295 /**
 296 * Get the secondary order of a collation order.
 297 * @param order the collation order
 298 * @return the secondary order of a collation order.
 299 * @stable ICU 2.6
 300 */
 301 U_STABLE int32_t U_EXPORT2
 302 ucol_secondaryOrder (int32_t order);
 303
 304 /**
 305 * Get the tertiary order of a collation order.
 306 * @param order the collation order
 307 * @return the tertiary order of a collation order.
 308 * @stable ICU 2.6
 309 */
 310 U_STABLE int32_t U_EXPORT2
 311 ucol_tertiaryOrder (int32_t order);
 312
 313 #endif /* #if !UCONFIG_NO_COLLATION */
 314
 315 #endif