michael@0: /*
michael@0: *******************************************************************************
michael@0: *
michael@0: *   Copyright (C) 2000-2011, International Business Machines
michael@0: *   Corporation and others.  All Rights Reserved.
michael@0: *
michael@0: *******************************************************************************
michael@0: *   file name:  ucol_elm.h
michael@0: *   encoding:   US-ASCII
michael@0: *   tab size:   8 (not used)
michael@0: *   indentation:4
michael@0: *
michael@0: *   created 02/22/2001
michael@0: *   created by: Vladimir Weinstein
michael@0: *
michael@0: *   This program reads the Franctional UCA table and generates
michael@0: *   internal format for UCA table as well as inverse UCA table.
michael@0: *   It then writes binary files containing the data: ucadata.dat 
michael@0: *   & invuca.dat
michael@0: */
michael@0: #ifndef UCOL_UCAELEMS_H
michael@0: #define UCOL_UCAELEMS_H
michael@0: 
michael@0: #include "unicode/utypes.h"
michael@0: #include "unicode/uniset.h"
michael@0: #include "ucol_tok.h"
michael@0: 
michael@0: #if !UCONFIG_NO_COLLATION
michael@0: 
michael@0: #include "ucol_imp.h"
michael@0: 
michael@0: #ifdef UCOL_DEBUG
michael@0: #include "cmemory.h"
michael@0: #include <stdio.h>
michael@0: #endif
michael@0: 
michael@0: U_CDECL_BEGIN
michael@0: 
michael@0: /* This is the maximum trie capacity for the mapping trie.
michael@0: Due to current limitations in genuca and the design of UTrie,
michael@0: this number can't be more than 256K.
michael@0: As of Unicode 5, it currently could safely go to 128K without
michael@0: a problem. Normally, less than 32K are tailored.
michael@0: */
michael@0: #define UCOL_ELM_TRIE_CAPACITY 0x40000
michael@0: 
michael@0: /* This is the maxmun capacity for temparay combining class 
michael@0:  * table.  The table will be compacted after scanning all the
michael@0:  * Unicode codepoints.
michael@0: */
michael@0: #define UCOL_MAX_CM_TAB  0x10000
michael@0: 
michael@0: 
michael@0: typedef struct {
michael@0:     uint32_t *CEs;
michael@0:     int32_t position;
michael@0:     int32_t size;
michael@0: } ExpansionTable;
michael@0: 
michael@0: typedef struct {
michael@0:     UChar prefixChars[128];
michael@0:     UChar *prefix;
michael@0:     uint32_t prefixSize;
michael@0:     UChar uchars[128];
michael@0:     UChar *cPoints;
michael@0:     uint32_t cSize;          /* Number of characters in sequence - for contraction */
michael@0:     uint32_t noOfCEs;        /* Number of collation elements                       */
michael@0:     uint32_t CEs[128];      /* These are collation elements - there could be more than one - in case of expansion */
michael@0:     uint32_t mapCE;         /* This is the value element maps in original table   */
michael@0:     uint32_t sizePrim[128];
michael@0:     uint32_t sizeSec[128];
michael@0:     uint32_t sizeTer[128];
michael@0:     UBool caseBit;
michael@0:     UBool isThai;
michael@0: } UCAElements;
michael@0: 
michael@0: typedef struct {
michael@0:   uint32_t *endExpansionCE;
michael@0:   UBool    *isV;
michael@0:   int32_t   position;
michael@0:   int32_t   size;
michael@0:   uint8_t   maxLSize;
michael@0:   uint8_t   maxVSize;
michael@0:   uint8_t   maxTSize;
michael@0: } MaxJamoExpansionTable;
michael@0: 
michael@0: typedef struct {
michael@0:   uint32_t *endExpansionCE;
michael@0:   uint8_t  *expansionCESize;
michael@0:   int32_t   position;
michael@0:   int32_t   size;
michael@0: } MaxExpansionTable;
michael@0: 
michael@0: typedef struct {
michael@0:     uint16_t   index[256];  /* index of cPoints by combining class 0-255. */
michael@0:     UChar      *cPoints;    /* code point array of all combining marks */
michael@0:     uint32_t   size;        /* total number of combining marks */
michael@0: } CombinClassTable;
michael@0: 
michael@0: typedef struct {
michael@0:   /*CompactEIntArray      *mapping; */
michael@0:   UNewTrie                 *mapping; 
michael@0:   ExpansionTable        *expansions; 
michael@0:   struct CntTable       *contractions;
michael@0:   UCATableHeader        *image;
michael@0:   UColOptionSet         *options;
michael@0:   MaxExpansionTable     *maxExpansions;
michael@0:   MaxJamoExpansionTable *maxJamoExpansions;
michael@0:   uint8_t               *unsafeCP;
michael@0:   uint8_t               *contrEndCP;
michael@0:   const UCollator       *UCA;
michael@0:   UHashtable      *prefixLookup;
michael@0:   CombinClassTable      *cmLookup;  /* combining class lookup for tailoring. */
michael@0: } tempUCATable; 
michael@0: 
michael@0: typedef struct {
michael@0:     UChar cp;
michael@0:     uint16_t cClass;   // combining class
michael@0: }CompData;
michael@0: 
michael@0: typedef struct {
michael@0:     CompData *precomp;
michael@0:     int32_t precompLen;
michael@0:     UChar *decomp;
michael@0:     int32_t decompLen;
michael@0:     UChar *comp;
michael@0:     int32_t compLen;
michael@0:     uint16_t curClass;
michael@0:     uint16_t tailoringCM;
michael@0:     int32_t  cmPos;
michael@0: }tempTailorContext;
michael@0: 
michael@0: U_CAPI tempUCATable * U_EXPORT2 uprv_uca_initTempTable(UCATableHeader *image, UColOptionSet *opts, const UCollator *UCA, UColCETags initTag, UColCETags supplementaryInitTag, UErrorCode *status);
michael@0: U_CAPI void U_EXPORT2 uprv_uca_closeTempTable(tempUCATable *t);
michael@0: U_CAPI uint32_t U_EXPORT2 uprv_uca_addAnElement(tempUCATable *t, UCAElements *element, UErrorCode *status);
michael@0: U_CAPI UCATableHeader * U_EXPORT2 uprv_uca_assembleTable(tempUCATable *t, UErrorCode *status);
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: uprv_uca_canonicalClosure(tempUCATable *t, UColTokenParser *src,
michael@0:                           icu::UnicodeSet *closed, UErrorCode *status);
michael@0: 
michael@0: U_CDECL_END
michael@0: 
michael@0: #endif /* #if !UCONFIG_NO_COLLATION */
michael@0: 
michael@0: #endif