include/unicode/ucoleitr.h

0001 // © 2016 and later: Unicode, Inc. and others.
0002 // License & terms of use: http://www.unicode.org/copyright.html
0003 /*
0004 *******************************************************************************
0005 *   Copyright (C) 2001-2014, International Business Machines
0006 *   Corporation and others.  All Rights Reserved.
0007 *******************************************************************************
0008 *
0009 * File ucoleitr.h
0010 *
0011 * Modification History:
0012 *
0013 * Date        Name        Description
0014 * 02/15/2001  synwee      Modified all methods to process its own function
0015 *                         instead of calling the equivalent c++ api (coleitr.h)
0016 *******************************************************************************/
0017
0018 #ifndef UCOLEITR_H
0019 #define UCOLEITR_H
0020
0021 #include "unicode/utypes.h"
0022
0023 #if !UCONFIG_NO_COLLATION
0024
0025 /**
0026  * This indicates an error has occurred during processing or if no more CEs is
0027  * to be returned.
0028  * @stable ICU 2.0
0029  */
0030 #define UCOL_NULLORDER        ((int32_t)0xFFFFFFFF)
0031
0032 #include "unicode/ucol.h"
0033
0034 /**
0035  * The UCollationElements struct.
0036  * For usage in C programs.
0037  * @stable ICU 2.0
0038  */
0039 typedef struct UCollationElements UCollationElements;
0040
0041 /**
0042  * \file
0043  * \brief C API: UCollationElements
0044  *
0045  * The UCollationElements API is used as an iterator to walk through each
0046  * character of an international string. Use the iterator to return the
0047  * ordering priority of the positioned character. The ordering priority of a
0048  * character, which we refer to as a key, defines how a character is collated
0049  * in the given collation object.
0050  * For example, consider the following in Slovak and in traditional Spanish collation:
0051  * <pre>
0052  * .       "ca" -> the first key is key('c') and second key is key('a').
0053  * .       "cha" -> the first key is key('ch') and second key is key('a').
0054  * </pre>
0055  * And in German phonebook collation,
0056  * <pre>
0057  * .       "<ae ligature>b"-> the first key is key('a'), the second key is key('e'), and
0058  * .       the third key is key('b').
0059  * </pre>
0060  * <p>Example of the iterator usage: (without error checking)
0061  * <pre>
0062  * .  void CollationElementIterator_Example()
0063  * .  {
0064  * .      UChar *s;
0065  * .      t_int32 order, primaryOrder;
0066  * .      UCollationElements *c;
0067  * .      UCollatorOld *coll;
0068  * .      UErrorCode success = U_ZERO_ERROR;
0069  * .      str=(UChar*)malloc(sizeof(UChar) * (strlen("This is a test")+1) );
0070  * .      u_uastrcpy(str, "This is a test");
0071  * .      coll = ucol_open(NULL, &success);
0072  * .      c = ucol_openElements(coll, str, u_strlen(str), &status);
0073  * .      order = ucol_next(c, &success);
0074  * .      ucol_reset(c);
0075  * .      order = ucol_prev(c, &success);
0076  * .      free(str);
0077  * .      ucol_close(coll);
0078  * .      ucol_closeElements(c);
0079  * .  }
0080  * </pre>
0081  * <p>
0082  * ucol_next() returns the collation order of the next.
0083  * ucol_prev() returns the collation order of the previous character.
0084  * The Collation Element Iterator moves only in one direction between calls to
0085  * ucol_reset. That is, ucol_next() and ucol_prev can not be inter-used.
0086  * Whenever ucol_prev is to be called after ucol_next() or vice versa,
0087  * ucol_reset has to be called first to reset the status, shifting pointers to
0088  * either the end or the start of the string. Hence at the next call of
0089  * ucol_prev or ucol_next, the first or last collation order will be returned.
0090  * If a change of direction is done without a ucol_reset, the result is
0091  * undefined.
0092  * The result of a forward iterate (ucol_next) and reversed result of the
0093  * backward iterate (ucol_prev) on the same string are equivalent, if
0094  * collation orders with the value 0 are ignored.
0095  * Character based on the comparison level of the collator.  A collation order
0096  * consists of primary order, secondary order and tertiary order.  The data
0097  * type of the collation order is <strong>int32_t</strong>.
0098  *
0099  * @see UCollator
0100  */
0101
0102 /**
0103  * Open the collation elements for a string.
0104  *
0105  * The UCollationElements retains a pointer to the supplied text.
0106  * The caller must not modify or delete the text while the UCollationElements
0107  * object is used to iterate over this text.
0108  *
0109  * @param coll The collator containing the desired collation rules.
0110  * @param text The text to iterate over.
0111  * @param textLength The number of characters in text, or -1 if null-terminated
0112  * @param status A pointer to a UErrorCode to receive any errors.
0113  * @return a struct containing collation element information
0114  * @stable ICU 2.0
0115  */
0116 U_CAPI UCollationElements* U_EXPORT2
0117 ucol_openElements(const UCollator  *coll,
0118                   const UChar      *text,
0119                         int32_t    textLength,
0120                         UErrorCode *status);
0121
0122 /**
0123  * get a hash code for a key... Not very useful!
0124  * @param key    the given key.
0125  * @param length the size of the key array.
0126  * @return       the hash code.
0127  * @stable ICU 2.0
0128  */
0129 U_CAPI int32_t U_EXPORT2
0130 ucol_keyHashCode(const uint8_t* key, int32_t length);
0131
0132 /**
0133  * Close a UCollationElements.
0134  * Once closed, a UCollationElements may no longer be used.
0135  * @param elems The UCollationElements to close.
0136  * @stable ICU 2.0
0137  */
0138 U_CAPI void U_EXPORT2
0139 ucol_closeElements(UCollationElements *elems);
0140
0141 /**
0142  * Reset the collation elements to their initial state.
0143  * This will move the 'cursor' to the beginning of the text.
0144  * Property settings for collation will be reset to the current status.
0145  * @param elems The UCollationElements to reset.
0146  * @see ucol_next
0147  * @see ucol_previous
0148  * @stable ICU 2.0
0149  */
0150 U_CAPI void U_EXPORT2
0151 ucol_reset(UCollationElements *elems);
0152
0153 /**
0154  * Get the ordering priority of the next collation element in the text.
0155  * A single character may contain more than one collation element.
0156  * @param elems The UCollationElements containing the text.
0157  * @param status A pointer to a UErrorCode to receive any errors.
0158  * @return The next collation elements ordering, otherwise returns UCOL_NULLORDER
0159  *         if an error has occurred or if the end of string has been reached
0160  * @stable ICU 2.0
0161  */
0162 U_CAPI int32_t U_EXPORT2
0163 ucol_next(UCollationElements *elems, UErrorCode *status);
0164
0165 /**
0166  * Get the ordering priority of the previous collation element in the text.
0167  * A single character may contain more than one collation element.
0168  * Note that internally a stack is used to store buffered collation elements.
0169  * @param elems The UCollationElements containing the text.
0170  * @param status A pointer to a UErrorCode to receive any errors. Notably
0171  *               a U_BUFFER_OVERFLOW_ERROR is returned if the internal stack
0172  *               buffer has been exhausted.
0173  * @return The previous collation elements ordering, otherwise returns
0174  *         UCOL_NULLORDER if an error has occurred or if the start of string has
0175  *         been reached.
0176  * @stable ICU 2.0
0177  */
0178 U_CAPI int32_t U_EXPORT2
0179 ucol_previous(UCollationElements *elems, UErrorCode *status);
0180
0181 /**
0182  * Get the maximum length of any expansion sequences that end with the
0183  * specified comparison order.
0184  * This is useful for .... ?
0185  * @param elems The UCollationElements containing the text.
0186  * @param order A collation order returned by previous or next.
0187  * @return maximum size of the expansion sequences ending with the collation
0188  *         element or 1 if collation element does not occur at the end of any
0189  *         expansion sequence
0190  * @stable ICU 2.0
0191  */
0192 U_CAPI int32_t U_EXPORT2
0193 ucol_getMaxExpansion(const UCollationElements *elems, int32_t order);
0194
0195 /**
0196  * Set the text containing the collation elements.
0197  * Property settings for collation will remain the same.
0198  * In order to reset the iterator to the current collation property settings,
0199  * the API reset() has to be called.
0200  *
0201  * The UCollationElements retains a pointer to the supplied text.
0202  * The caller must not modify or delete the text while the UCollationElements
0203  * object is used to iterate over this text.
0204  *
0205  * @param elems The UCollationElements to set.
0206  * @param text The source text containing the collation elements.
0207  * @param textLength The length of text, or -1 if null-terminated.
0208  * @param status A pointer to a UErrorCode to receive any errors.
0209  * @see ucol_getText
0210  * @stable ICU 2.0
0211  */
0212 U_CAPI void U_EXPORT2
0213 ucol_setText(      UCollationElements *elems,
0214              const UChar              *text,
0215                    int32_t            textLength,
0216                    UErrorCode         *status);
0217
0218 /**
0219  * Get the offset of the current source character.
0220  * This is an offset into the text of the character containing the current
0221  * collation elements.
0222  * @param elems The UCollationElements to query.
0223  * @return The offset of the current source character.
0224  * @see ucol_setOffset
0225  * @stable ICU 2.0
0226  */
0227 U_CAPI int32_t U_EXPORT2
0228 ucol_getOffset(const UCollationElements *elems);
0229
0230 /**
0231  * Set the offset of the current source character.
0232  * This is an offset into the text of the character to be processed.
0233  * Property settings for collation will remain the same.
0234  * In order to reset the iterator to the current collation property settings,
0235  * the API reset() has to be called.
0236  * @param elems The UCollationElements to set.
0237  * @param offset The desired character offset.
0238  * @param status A pointer to a UErrorCode to receive any errors.
0239  * @see ucol_getOffset
0240  * @stable ICU 2.0
0241  */
0242 U_CAPI void U_EXPORT2
0243 ucol_setOffset(UCollationElements *elems,
0244                int32_t             offset,
0245                UErrorCode         *status);
0246
0247 /**
0248 * Get the primary order of a collation order.
0249 * @param order the collation order
0250 * @return the primary order of a collation order.
0251 * @stable ICU 2.6
0252 */
0253 U_CAPI int32_t U_EXPORT2
0254 ucol_primaryOrder (int32_t order);
0255
0256 /**
0257 * Get the secondary order of a collation order.
0258 * @param order the collation order
0259 * @return the secondary order of a collation order.
0260 * @stable ICU 2.6
0261 */
0262 U_CAPI int32_t U_EXPORT2
0263 ucol_secondaryOrder (int32_t order);
0264
0265 /**
0266 * Get the tertiary order of a collation order.
0267 * @param order the collation order
0268 * @return the tertiary order of a collation order.
0269 * @stable ICU 2.6
0270 */
0271 U_CAPI int32_t U_EXPORT2
0272 ucol_tertiaryOrder (int32_t order);
0273
0274 #endif /* #if !UCONFIG_NO_COLLATION */
0275
0276 #endif