1/*
2*******************************************************************************
3*
4*   Copyright (C) 2002-2010, International Business Machines
5*   Corporation and others.  All Rights Reserved.
6*
7*******************************************************************************
8*   file name:  propsvec.h
9*   encoding:   US-ASCII
10*   tab size:   8 (not used)
11*   indentation:4
12*
13*   created on: 2002feb22
14*   created by: Markus W. Scherer
15*
16*   Store bits (Unicode character properties) in bit set vectors.
17*/
18
19#ifndef __UPROPSVEC_H__
20#define __UPROPSVEC_H__
21
22#include "unicode/utypes.h"
23#include "utrie.h"
24#include "utrie2.h"
25
26U_CDECL_BEGIN
27
28/**
29 * Unicode Properties Vectors associated with code point ranges.
30 *
31 * Rows of uint32_t integers in a contiguous array store
32 * the range limits and the properties vectors.
33 *
34 * Logically, each row has a certain number of uint32_t values,
35 * which is set via the upvec_open() "columns" parameter.
36 *
37 * Internally, two additional columns are stored.
38 * In each internal row,
39 * row[0] contains the start code point and
40 * row[1] contains the limit code point,
41 * which is the start of the next range.
42 *
43 * Initially, there is only one "normal" row for
44 * range [0..0x110000[ with values 0.
45 * There are additional rows for special purposes, see UPVEC_FIRST_SPECIAL_CP.
46 *
47 * It would be possible to store only one range boundary per row,
48 * but self-contained rows allow to later sort them by contents.
49 */
50struct UPropsVectors;
51typedef struct UPropsVectors UPropsVectors;
52
53/*
54 * Special pseudo code points for storing the initialValue and the errorValue,
55 * which are used to initialize a UTrie2 or similar.
56 */
57#define UPVEC_FIRST_SPECIAL_CP 0x110000
58#define UPVEC_INITIAL_VALUE_CP 0x110000
59#define UPVEC_ERROR_VALUE_CP 0x110001
60#define UPVEC_MAX_CP 0x110001
61
62/*
63 * Special pseudo code point used in upvec_compact() signalling the end of
64 * delivering special values and the beginning of delivering real ones.
65 * Stable value, unlike UPVEC_MAX_CP which might grow over time.
66 */
67#define UPVEC_START_REAL_VALUES_CP 0x200000
68
69/*
70 * Open a UPropsVectors object.
71 * @param columns Number of value integers (uint32_t) per row.
72 */
73U_CAPI UPropsVectors * U_EXPORT2
74upvec_open(int32_t columns, UErrorCode *pErrorCode);
75
76U_CAPI void U_EXPORT2
77upvec_close(UPropsVectors *pv);
78
79/*
80 * In rows for code points [start..end], select the column,
81 * reset the mask bits and set the value bits (ANDed with the mask).
82 *
83 * Will set U_NO_WRITE_PERMISSION if called after upvec_compact().
84 */
85U_CAPI void U_EXPORT2
86upvec_setValue(UPropsVectors *pv,
87               UChar32 start, UChar32 end,
88               int32_t column,
89               uint32_t value, uint32_t mask,
90               UErrorCode *pErrorCode);
91
92/*
93 * Logically const but must not be used on the same pv concurrently!
94 * Always returns 0 if called after upvec_compact().
95 */
96U_CAPI uint32_t U_EXPORT2
97upvec_getValue(const UPropsVectors *pv, UChar32 c, int32_t column);
98
99/*
100 * pRangeStart and pRangeEnd can be NULL.
101 * @return NULL if rowIndex out of range and for illegal arguments,
102 *         or if called after upvec_compact()
103 */
104U_CAPI uint32_t * U_EXPORT2
105upvec_getRow(const UPropsVectors *pv, int32_t rowIndex,
106             UChar32 *pRangeStart, UChar32 *pRangeEnd);
107
108/*
109 * Compact the vectors:
110 * - modify the memory
111 * - keep only unique vectors
112 * - store them contiguously from the beginning of the memory
113 * - for each (non-unique) row, call the handler function
114 *
115 * The handler's rowIndex is the index of the row in the compacted
116 * memory block.
117 * (Therefore, it starts at 0 increases in increments of the columns value.)
118 *
119 * In a first phase, only special values are delivered (each exactly once),
120 * with start==end both equalling a special pseudo code point.
121 * Then the handler is called once more with start==end==UPVEC_START_REAL_VALUES_CP
122 * where rowIndex is the length of the compacted array,
123 * and the row is arbitrary (but not NULL).
124 * Then, in the second phase, the handler is called for each row of real values.
125 */
126typedef void U_CALLCONV
127UPVecCompactHandler(void *context,
128                    UChar32 start, UChar32 end,
129                    int32_t rowIndex, uint32_t *row, int32_t columns,
130                    UErrorCode *pErrorCode);
131
132U_CAPI void U_EXPORT2
133upvec_compact(UPropsVectors *pv, UPVecCompactHandler *handler, void *context, UErrorCode *pErrorCode);
134
135/*
136 * Get the vectors array after calling upvec_compact().
137 * The caller must not modify nor release the returned array.
138 * Returns NULL if called before upvec_compact().
139 */
140U_CAPI const uint32_t * U_EXPORT2
141upvec_getArray(const UPropsVectors *pv, int32_t *pRows, int32_t *pColumns);
142
143/*
144 * Get a clone of the vectors array after calling upvec_compact().
145 * The caller owns the returned array and must uprv_free() it.
146 * Returns NULL if called before upvec_compact().
147 */
148U_CAPI uint32_t * U_EXPORT2
149upvec_cloneArray(const UPropsVectors *pv,
150                 int32_t *pRows, int32_t *pColumns, UErrorCode *pErrorCode);
151
152/*
153 * Call upvec_compact(), create a 16-bit UTrie2 with indexes into the compacted
154 * vectors array, and freeze the trie.
155 */
156U_CAPI UTrie2 * U_EXPORT2
157upvec_compactToUTrie2WithRowIndexes(UPropsVectors *pv, UErrorCode *pErrorCode);
158
159struct UPVecToUTrie2Context {
160    UTrie2 *trie;
161    int32_t initialValue;
162    int32_t errorValue;
163    int32_t maxValue;
164};
165typedef struct UPVecToUTrie2Context UPVecToUTrie2Context;
166
167/* context=UPVecToUTrie2Context, creates the trie and stores the rowIndex values */
168U_CAPI void U_CALLCONV
169upvec_compactToUTrie2Handler(void *context,
170                             UChar32 start, UChar32 end,
171                             int32_t rowIndex, uint32_t *row, int32_t columns,
172                             UErrorCode *pErrorCode);
173
174U_CDECL_END
175
176#endif
177