ucol_elm.h revision b13da9df870a61b11249bf741347908dbea0edd8 
1/*
2*******************************************************************************
3*
4*   Copyright (C) 2000-2006, International Business Machines
5*   Corporation and others.  All Rights Reserved.
6*
7*******************************************************************************
8*   file name:  ucol_elm.h
9*   encoding:   US-ASCII
10*   tab size:   8 (not used)
11*   indentation:4
12*
13*   created 02/22/2001
14*   created by: Vladimir Weinstein
15*
16*   This program reads the Franctional UCA table and generates
17*   internal format for UCA table as well as inverse UCA table.
18*   It then writes binary files containing the data: ucadata.dat
19*   & invuca.dat
20*/
21#ifndef UCOL_UCAELEMS_H
22#define UCOL_UCAELEMS_H
23
24#include "unicode/utypes.h"
25#include "ucol_tok.h"
26
27#if !UCONFIG_NO_COLLATION
28
29#include "ucol_cnt.h"
30#include "ucol_imp.h"
31
32#ifdef UCOL_DEBUG
33#include "cmemory.h"
34#include <stdio.h>
35#endif
36
37U_CDECL_BEGIN
38
39/* This is the maximum trie capacity for the mapping trie.
40Due to current limitations in genuca and the design of UTrie,
41this number can't be more than 256K.
42As of Unicode 5, it currently could safely go to 128K without
43a problem. Normally, less than 32K are tailored.
44*/
45#define UCOL_ELM_TRIE_CAPACITY 0x40000
46
47/* This is the maxmun capacity for temparay combining class
48 * table.  The table will be compacted after scanning all the
49 * Unicode codepoints.
50*/
51#define UCOL_MAX_CM_TAB  0x10000
52
53
54typedef struct {
55    uint32_t *CEs;
56    int32_t position;
57    int32_t size;
58} ExpansionTable;
59
60typedef struct {
61    UChar prefixChars[128];
62    UChar *prefix;
63    uint32_t prefixSize;
64    UChar uchars[128];
65    UChar *cPoints;
66    uint32_t cSize;          /* Number of characters in sequence - for contraction */
67    uint32_t noOfCEs;        /* Number of collation elements                       */
68    uint32_t CEs[128];      /* These are collation elements - there could be more than one - in case of expansion */
69    uint32_t mapCE;         /* This is the value element maps in original table   */
70    uint32_t sizePrim[128];
71    uint32_t sizeSec[128];
72    uint32_t sizeTer[128];
73    UBool variableTop;
74    UBool caseBit;
75    UBool isThai;
76} UCAElements;
77
78typedef struct {
79  uint32_t *endExpansionCE;
80  UBool    *isV;
81  int32_t   position;
82  int32_t   size;
83  uint8_t   maxLSize;
84  uint8_t   maxVSize;
85  uint8_t   maxTSize;
86} MaxJamoExpansionTable;
87
88typedef struct {
89  uint32_t *endExpansionCE;
90  uint8_t  *expansionCESize;
91  int32_t   position;
92  int32_t   size;
93} MaxExpansionTable;
94
95typedef struct {
96    uint16_t   index[256];  /* index of cPoints by combining class 0-255. */
97    UChar      *cPoints;    /* code point array of all combining marks */
98    uint32_t   size;        /* total number of combining marks */
99} CombinClassTable;
100
101typedef struct {
102  /*CompactEIntArray      *mapping; */
103  UNewTrie                 *mapping;
104  ExpansionTable        *expansions;
105  struct CntTable       *contractions;
106  UCATableHeader        *image;
107  UColOptionSet         *options;
108  MaxExpansionTable     *maxExpansions;
109  MaxJamoExpansionTable *maxJamoExpansions;
110  uint8_t               *unsafeCP;
111  uint8_t               *contrEndCP;
112  const UCollator       *UCA;
113  UHashtable      *prefixLookup;
114  CombinClassTable      *cmLookup;  /* combining class lookup for tailoring. */
115} tempUCATable;
116
117typedef struct {
118    UChar cp;
119    uint16_t cClass;   // combining class
120}CompData;
121
122typedef struct {
123    CompData *precomp;
124    int32_t precompLen;
125    UChar *decomp;
126    int32_t decompLen;
127    UChar *comp;
128    int32_t compLen;
129    uint16_t curClass;
130    uint16_t tailoringCM;
131    int32_t  cmPos;
132}tempTailorContext;
133
134U_CAPI tempUCATable * U_EXPORT2 uprv_uca_initTempTable(UCATableHeader *image, UColOptionSet *opts, const UCollator *UCA, UColCETags initTag, UColCETags supplementaryInitTag, UErrorCode *status);
135U_CAPI tempUCATable * U_EXPORT2 uprv_uca_cloneTempTable(tempUCATable *t, UErrorCode *status);
136U_CAPI void U_EXPORT2 uprv_uca_closeTempTable(tempUCATable *t);
137U_CAPI uint32_t U_EXPORT2 uprv_uca_addAnElement(tempUCATable *t, UCAElements *element, UErrorCode *status);
138U_CAPI UCATableHeader * U_EXPORT2 uprv_uca_assembleTable(tempUCATable *t, UErrorCode *status);
139U_CAPI int32_t U_EXPORT2
140uprv_uca_canonicalClosure(tempUCATable *t, UColTokenParser *src, UErrorCode *status);
141#define paddedsize(something) ((something)+((((something)%4)!=0)?(4-(something)%4):0))
142#define headersize (paddedsize(sizeof(UCATableHeader))+paddedsize(sizeof(UColOptionSet)))
143
144U_CDECL_END
145
146#endif /* #if !UCONFIG_NO_COLLATION */
147
148#endif
149