icuSources/common/utrie2_impl.h

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*
   4 ******************************************************************************
   5 *
   6 *   Copyright (C) 2001-2008, International Business Machines
   7 *   Corporation and others.  All Rights Reserved.
   8 *
   9 ******************************************************************************
  10 *   file name:  utrie2_impl.h
  11 *   encoding:   UTF-8
  12 *   tab size:   8 (not used)
  13 *   indentation:4
  14 *
  15 *   created on: 2008sep26 (split off from utrie2.c)
  16 *   created by: Markus W. Scherer
  17 *
  18 *   Definitions needed for both runtime and builder code for UTrie2,
  19 *   used by utrie2.c and utrie2_builder.c.
  20 */
  21
  22 #ifndef __UTRIE2_IMPL_H__
  23 #define __UTRIE2_IMPL_H__
  24
  25 #include "utrie2.h"
  26
  27 /* Public UTrie2 API implementation ----------------------------------------- */
  28
  29 /*
  30  * These definitions are mostly needed by utrie2.c,
  31  * but also by utrie2_serialize() and utrie2_swap().
  32  */
  33
  34 /*
  35  * UTrie and UTrie2 signature values,
  36  * in platform endianness and opposite endianness.
  37  */
  38 #define UTRIE_SIG       0x54726965
  39 #define UTRIE_OE_SIG    0x65697254
  40
  41 #define UTRIE2_SIG      0x54726932
  42 #define UTRIE2_OE_SIG   0x32697254
  43
  44 /**
  45  * Trie data structure in serialized form:
  46  *
  47  * UTrie2Header header;
  48  * uint16_t index[header.index2Length];
  49  * uint16_t data[header.shiftedDataLength<<2];  -- or uint32_t data[...]
  50  * @internal
  51  */
  52 typedef struct UTrie2Header {
  53     /** "Tri2" in big-endian US-ASCII (0x54726932) */
  54     uint32_t signature;
  55
  56     /**
  57      * options bit field:
  58      * 15.. 4   reserved (0)
  59      *  3.. 0   UTrie2ValueBits valueBits
  60      */
  61     uint16_t options;
  62
  63     /** UTRIE2_INDEX_1_OFFSET..UTRIE2_MAX_INDEX_LENGTH */
  64     uint16_t indexLength;
  65
  66     /** (UTRIE2_DATA_START_OFFSET..UTRIE2_MAX_DATA_LENGTH)>>UTRIE2_INDEX_SHIFT */
  67     uint16_t shiftedDataLength;
  68
  69     /** Null index and data blocks, not shifted. */
  70     uint16_t index2NullOffset, dataNullOffset;
  71
  72     /**
  73      * First code point of the single-value range ending with U+10ffff,
  74      * rounded up and then shifted right by UTRIE2_SHIFT_1.
  75      */
  76     uint16_t shiftedHighStart;
  77 } UTrie2Header;
  78
  79 /**
  80  * Constants for use with UTrie2Header.options.
  81  * @internal
  82  */
  83 enum {
  84     /** Mask to get the UTrie2ValueBits valueBits from options. */
  85     UTRIE2_OPTIONS_VALUE_BITS_MASK=0xf
  86 };
  87
  88 /* Building a trie ---------------------------------------------------------- */
  89
  90 /*
  91  * These definitions are mostly needed by utrie2_builder.c, but also by
  92  * utrie2_get32() and utrie2_enum().
  93  */
  94
  95 enum {
  96     /**
  97      * At build time, leave a gap in the index-2 table,
  98      * at least as long as the maximum lengths of the 2-byte UTF-8 index-2 table
  99      * and the supplementary index-1 table.
 100      * Round up to UTRIE2_INDEX_2_BLOCK_LENGTH for proper compacting.
 101      */
 102     UNEWTRIE2_INDEX_GAP_OFFSET=UTRIE2_INDEX_2_BMP_LENGTH,
 103     UNEWTRIE2_INDEX_GAP_LENGTH=
 104         ((UTRIE2_UTF8_2B_INDEX_2_LENGTH+UTRIE2_MAX_INDEX_1_LENGTH)+UTRIE2_INDEX_2_MASK)&
 105         ~UTRIE2_INDEX_2_MASK,
 106
 107     /**
 108      * Maximum length of the build-time index-2 array.
 109      * Maximum number of Unicode code points (0x110000) shifted right by UTRIE2_SHIFT_2,
 110      * plus the part of the index-2 table for lead surrogate code points,
 111      * plus the build-time index gap,
 112      * plus the null index-2 block.
 113      */
 114     UNEWTRIE2_MAX_INDEX_2_LENGTH=
 115         (0x110000>>UTRIE2_SHIFT_2)+
 116         UTRIE2_LSCP_INDEX_2_LENGTH+
 117         UNEWTRIE2_INDEX_GAP_LENGTH+
 118         UTRIE2_INDEX_2_BLOCK_LENGTH,
 119
 120     UNEWTRIE2_INDEX_1_LENGTH=0x110000>>UTRIE2_SHIFT_1
 121 };
 122
 123 /**
 124  * Maximum length of the build-time data array.
 125  * One entry per 0x110000 code points, plus the illegal-UTF-8 block and the null block,
 126  * plus values for the 0x400 surrogate code units.
 127  */
 128 #define UNEWTRIE2_MAX_DATA_LENGTH (0x110000+0x40+0x40+0x400)
 129
 130 /*
 131  * Build-time trie structure.
 132  *
 133  * Just using a boolean flag for "repeat use" could lead to data array overflow
 134  * because we would not be able to detect when a data block becomes unused.
 135  * It also leads to orphan data blocks that are kept through serialization.
 136  *
 137  * Need to use reference counting for data blocks,
 138  * and allocDataBlock() needs to look for a free block before increasing dataLength.
 139  *
 140  * This scheme seems like overkill for index-2 blocks since the whole index array is
 141  * preallocated anyway (unlike the growable data array).
 142  * Just allocating multiple index-2 blocks as needed.
 143  */
 144 struct UNewTrie2 {
 145     int32_t index1[UNEWTRIE2_INDEX_1_LENGTH];
 146     int32_t index2[UNEWTRIE2_MAX_INDEX_2_LENGTH];
 147     uint32_t *data;
 148
 149     uint32_t initialValue, errorValue;
 150     int32_t index2Length, dataCapacity, dataLength;
 151     int32_t firstFreeBlock;
 152     int32_t index2NullOffset, dataNullOffset;
 153     UChar32 highStart;
 154     UBool isCompacted;
 155
 156     /**
 157      * Multi-purpose per-data-block table.
 158      *
 159      * Before compacting:
 160      *
 161      * Per-data-block reference counters/free-block list.
 162      *  0: unused
 163      * >0: reference counter (number of index-2 entries pointing here)
 164      * <0: next free data block in free-block list
 165      *
 166      * While compacting:
 167      *
 168      * Map of adjusted indexes, used in compactData() and compactIndex2().
 169      * Maps from original indexes to new ones.
 170      */
 171     int32_t map[UNEWTRIE2_MAX_DATA_LENGTH>>UTRIE2_SHIFT_2];
 172 };
 173
 174 #endif