icuSources/common/ucol_data.h

   1 /*
   2 *******************************************************************************
   3 *   Copyright (C) 2000-2011, International Business Machines
   4 *   Corporation and others.  All Rights Reserved.
   5 *******************************************************************************
   6 *   file name:  ucol_data.h
   7 *   encoding:   US-ASCII
   8 *   tab size:   8 (not used)
   9 *   indentation:4
  10 *
  11 *   created on: 2011jul02
  12 *   created by: Markus Scherer
  13 *
  14 * Private implementation header for C/C++ collation.
  15 * Some file data structure definitions were moved here from i18n/ucol_imp.h
  16 * so that the common library (via ucol_swp.cpp) need not depend on the i18n library at all.
  17 *
  18 * We do not want to move the collation swapper to the i18n library because
  19 * a) the resource bundle swapper depends on it and would have to move too, and
  20 * b) we might want to eventually implement runtime data swapping,
  21 *    which might (or might not) be easier if all swappers are in the common library.
  22 */
  23
  24 #ifndef __UCOL_DATA_H__
  25 #define __UCOL_DATA_H__
  26
  27 #include "unicode/utypes.h"
  28
  29 #if !UCONFIG_NO_COLLATION
  30
  31 /* let us know whether reserved fields are reset to zero or junked */
  32 #define UCOL_HEADER_MAGIC 0x20030618
  33
  34 typedef struct {
  35       int32_t size;
  36       /* all the offsets are in bytes */
  37       /* to get the address add to the header address and cast properly */
  38       uint32_t options; /* these are the default options for the collator */
  39       uint32_t UCAConsts; /* structure which holds values for indirect positioning and implicit ranges */
  40       uint32_t contractionUCACombos;        /* this one is needed only for UCA, to copy the appropriate contractions */
  41       uint32_t magic;            /* magic number - lets us know whether reserved data is reset or junked */
  42       uint32_t mappingPosition;  /* const uint8_t *mappingPosition; */
  43       uint32_t expansion;        /* uint32_t *expansion;            */
  44       uint32_t contractionIndex; /* UChar *contractionIndex;        */
  45       uint32_t contractionCEs;   /* uint32_t *contractionCEs;       */
  46       uint32_t contractionSize;  /* needed for various closures */
  47       /*int32_t latinOneMapping;*/ /* this is now handled in the trie itself *//* fast track to latin1 chars      */
  48
  49       uint32_t endExpansionCE;      /* array of last collation element in
  50                                        expansion */
  51       uint32_t expansionCESize;     /* array of maximum expansion size
  52                                        corresponding to the expansion
  53                                        collation elements with last element
  54                                        in endExpansionCE*/
  55       int32_t  endExpansionCECount; /* size of endExpansionCE */
  56       uint32_t unsafeCP;            /* hash table of unsafe code points */
  57       uint32_t contrEndCP;          /* hash table of final code points  */
  58                                     /*   in contractions.               */
  59
  60       int32_t contractionUCACombosSize;     /* number of UCA contraction items. */
  61                                             /*Length is contractionUCACombosSize*contractionUCACombosWidth*sizeof(UChar) */
  62       UBool jamoSpecial;                    /* is jamoSpecial */
  63       UBool isBigEndian;                    /* is this data big endian? from the UDataInfo header*/
  64       uint8_t charSetFamily;                /* what is the charset family of this data from the UDataInfo header*/
  65       uint8_t contractionUCACombosWidth;    /* width of UCA combos field */
  66       UVersionInfo version;
  67       UVersionInfo UCAVersion;              /* version of the UCA, read from file */
  68       UVersionInfo UCDVersion;              /* UCD version, obtained by u_getUnicodeVersion */
  69       UVersionInfo formatVersion;           /* format version from the UDataInfo header */
  70       uint32_t scriptToLeadByte;            /* offset to script to lead collation byte mapping data */
  71       uint32_t leadByteToScript;            /* offset to lead collation byte to script mapping data */
  72       uint8_t reserved[76];                 /* for future use */
  73 } UCATableHeader;
  74
  75 typedef struct {
  76   uint32_t byteSize;
  77   uint32_t tableSize;
  78   uint32_t contsSize;
  79   uint32_t table;
  80   uint32_t conts;
  81   UVersionInfo UCAVersion;              /* version of the UCA, read from file */
  82   uint8_t padding[8];
  83 } InverseUCATableHeader;
  84
  85 #endif  /* !UCONFIG_NO_COLLATION */
  86
  87 #endif  /* __UCOL_DATA_H__ */