ucoleitr.h revision c73f511526464f8e56c242df80552e9b0d94ae3d 
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#include "unicode/ucol.h"
31
32/**
33 * The UCollationElements struct.
34 * For usage in C programs.
35 * @stable ICU 2.0
36 */
37typedef struct UCollationElements UCollationElements;
38
39/**
40 * \file
41 * \brief C API: UCollationElements
42 *
43 * The UCollationElements API is used as an iterator to walk through each
44 * character of an international string. Use the iterator to return the
45 * ordering priority of the positioned character. The ordering priority of a
46 * character, which we refer to as a key, defines how a character is collated
47 * in the given collation object.
48 * For example, consider the following in Slovak and in traditional Spanish collation:
49 * <pre>
50 * .       "ca" -> the first key is key('c') and second key is key('a').
51 * .       "cha" -> the first key is key('ch') and second key is key('a').
52 * </pre>
53 * And in German phonebook collation,
54 * <pre>
55 * .       "<ae ligature>b"-> the first key is key('a'), the second key is key('e'), and
56 * .       the third key is key('b').
57 * </pre>
58 * <p>Example of the iterator usage: (without error checking)
59 * <pre>
60 * .  void CollationElementIterator_Example()
61 * .  {
62 * .      UChar *s;
63 * .      t_int32 order, primaryOrder;
64 * .      UCollationElements *c;
65 * .      UCollatorOld *coll;
66 * .      UErrorCode success = U_ZERO_ERROR;
67 * .      s=(UChar*)malloc(sizeof(UChar) * (strlen("This is a test")+1) );
68 * .      u_uastrcpy(s, "This is a test");
69 * .      coll = ucol_open(NULL, &success);
70 * .      c = ucol_openElements(coll, str, u_strlen(str), &status);
71 * .      order = ucol_next(c, &success);
72 * .      ucol_reset(c);
73 * .      order = ucol_prev(c, &success);
74 * .      free(s);
75 * .      ucol_close(coll);
76 * .      ucol_closeElements(c);
77 * .  }
78 * </pre>
79 * <p>
80 * ucol_next() returns the collation order of the next.
81 * ucol_prev() returns the collation order of the previous character.
82 * The Collation Element Iterator moves only in one direction between calls to
83 * ucol_reset. That is, ucol_next() and ucol_prev can not be inter-used.
84 * Whenever ucol_prev is to be called after ucol_next() or vice versa,
85 * ucol_reset has to be called first to reset the status, shifting pointers to
86 * either the end or the start of the string. Hence at the next call of
87 * ucol_prev or ucol_next, the first or last collation order will be returned.
88 * If a change of direction is done without a ucol_reset, the result is
89 * undefined.
90 * The result of a forward iterate (ucol_next) and reversed result of the
91 * backward iterate (ucol_prev) on the same string are equivalent, if
92 * collation orders with the value 0 are ignored.
93 * Character based on the comparison level of the collator.  A collation order
94 * consists of primary order, secondary order and tertiary order.  The data
95 * type of the collation order is <strong>int32_t</strong>.
96 *
97 * @see UCollator
98 */
99
100/**
101 * Open the collation elements for a string.
102 *
103 * @param coll The collator containing the desired collation rules.
104 * @param text The text to iterate over.
105 * @param textLength The number of characters in text, or -1 if null-terminated
106 * @param status A pointer to a UErrorCode to receive any errors.
107 * @return a struct containing collation element information
108 * @stable ICU 2.0
109 */
110U_STABLE UCollationElements* U_EXPORT2
111ucol_openElements(const UCollator  *coll,
112                  const UChar      *text,
113                        int32_t    textLength,
114                        UErrorCode *status);
115
116
117/**
118 * get a hash code for a key... Not very useful!
119 * @param key    the given key.
120 * @param length the size of the key array.
121 * @return       the hash code.
122 * @stable ICU 2.0
123 */
124U_STABLE int32_t U_EXPORT2
125ucol_keyHashCode(const uint8_t* key, int32_t length);
126
127/**
128 * Close a UCollationElements.
129 * Once closed, a UCollationElements may no longer be used.
130 * @param elems The UCollationElements to close.
131 * @stable ICU 2.0
132 */
133U_STABLE void U_EXPORT2
134ucol_closeElements(UCollationElements *elems);
135
136/**
137 * Reset the collation elements to their initial state.
138 * This will move the 'cursor' to the beginning of the text.
139 * Property settings for collation will be reset to the current status.
140 * @param elems The UCollationElements to reset.
141 * @see ucol_next
142 * @see ucol_previous
143 * @stable ICU 2.0
144 */
145U_STABLE void U_EXPORT2
146ucol_reset(UCollationElements *elems);
147
148/**
149 * Get the ordering priority of the next collation element in the text.
150 * A single character may contain more than one collation element.
151 * @param elems The UCollationElements containing the text.
152 * @param status A pointer to a UErrorCode to receive any errors.
153 * @return The next collation elements ordering, otherwise returns NULLORDER
154 *         if an error has occured or if the end of string has been reached
155 * @stable ICU 2.0
156 */
157U_STABLE int32_t U_EXPORT2
158ucol_next(UCollationElements *elems, UErrorCode *status);
159
160/**
161 * Get the ordering priority of the previous collation element in the text.
162 * A single character may contain more than one collation element.
163 * Note that internally a stack is used to store buffered collation elements.
164 * @param elems The UCollationElements containing the text.
165 * @param status A pointer to a UErrorCode to receive any errors. Noteably
166 *               a U_BUFFER_OVERFLOW_ERROR is returned if the internal stack
167 *               buffer has been exhausted.
168 * @return The previous collation elements ordering, otherwise returns
169 *         NULLORDER if an error has occured or if the start of string has
170 *         been reached.
171 * @stable ICU 2.0
172 */
173U_STABLE int32_t U_EXPORT2
174ucol_previous(UCollationElements *elems, UErrorCode *status);
175
176/**
177 * Get the maximum length of any expansion sequences that end with the
178 * specified comparison order.
179 * This is useful for .... ?
180 * @param elems The UCollationElements containing the text.
181 * @param order A collation order returned by previous or next.
182 * @return maximum size of the expansion sequences ending with the collation
183 *         element or 1 if collation element does not occur at the end of any
184 *         expansion sequence
185 * @stable ICU 2.0
186 */
187U_STABLE int32_t U_EXPORT2
188ucol_getMaxExpansion(const UCollationElements *elems, int32_t order);
189
190/**
191 * Set the text containing the collation elements.
192 * Property settings for collation will remain the same.
193 * In order to reset the iterator to the current collation property settings,
194 * the API reset() has to be called.
195 * @param elems The UCollationElements to set.
196 * @param text The source text containing the collation elements.
197 * @param textLength The length of text, or -1 if null-terminated.
198 * @param status A pointer to a UErrorCode to receive any errors.
199 * @see ucol_getText
200 * @stable ICU 2.0
201 */
202U_STABLE void U_EXPORT2
203ucol_setText(      UCollationElements *elems,
204             const UChar              *text,
205                   int32_t            textLength,
206                   UErrorCode         *status);
207
208/**
209 * Get the offset of the current source character.
210 * This is an offset into the text of the character containing the current
211 * collation elements.
212 * @param elems The UCollationElements to query.
213 * @return The offset of the current source character.
214 * @see ucol_setOffset
215 * @stable ICU 2.0
216 */
217U_STABLE int32_t U_EXPORT2
218ucol_getOffset(const UCollationElements *elems);
219
220/**
221 * Set the offset of the current source character.
222 * This is an offset into the text of the character to be processed.
223 * Property settings for collation will remain the same.
224 * In order to reset the iterator to the current collation property settings,
225 * the API reset() has to be called.
226 * @param elems The UCollationElements to set.
227 * @param offset The desired character offset.
228 * @param status A pointer to a UErrorCode to receive any errors.
229 * @see ucol_getOffset
230 * @stable ICU 2.0
231 */
232U_STABLE void U_EXPORT2
233ucol_setOffset(UCollationElements *elems,
234               int32_t        offset,
235               UErrorCode         *status);
236
237/**
238* Get the primary order of a collation order.
239* @param order the collation order
240* @return the primary order of a collation order.
241* @stable ICU 2.6
242*/
243U_STABLE int32_t U_EXPORT2
244ucol_primaryOrder (int32_t order);
245
246/**
247* Get the secondary order of a collation order.
248* @param order the collation order
249* @return the secondary order of a collation order.
250* @stable ICU 2.6
251*/
252U_STABLE int32_t U_EXPORT2
253ucol_secondaryOrder (int32_t order);
254
255/**
256* Get the tertiary order of a collation order.
257* @param order the collation order
258* @return the tertiary order of a collation order.
259* @stable ICU 2.6
260*/
261U_STABLE int32_t U_EXPORT2
262ucol_tertiaryOrder (int32_t order);
263
264#endif /* #if !UCONFIG_NO_COLLATION */
265
266#endif
267