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: *   file name:  ucol_data.h
michael@0: *   encoding:   US-ASCII
michael@0: *   tab size:   8 (not used)
michael@0: *   indentation:4
michael@0: *
michael@0: *   created on: 2011jul02
michael@0: *   created by: Markus Scherer
michael@0: *
michael@0: * Private implementation header for C/C++ collation.
michael@0: * Some file data structure definitions were moved here from i18n/ucol_imp.h
michael@0: * so that the common library (via ucol_swp.cpp) need not depend on the i18n library at all.
michael@0: *
michael@0: * We do not want to move the collation swapper to the i18n library because
michael@0: * a) the resource bundle swapper depends on it and would have to move too, and
michael@0: * b) we might want to eventually implement runtime data swapping,
michael@0: *    which might (or might not) be easier if all swappers are in the common library.
michael@0: */
michael@0: 
michael@0: #ifndef __UCOL_DATA_H__
michael@0: #define __UCOL_DATA_H__
michael@0: 
michael@0: #include "unicode/utypes.h"
michael@0: 
michael@0: #if !UCONFIG_NO_COLLATION
michael@0: 
michael@0: /* let us know whether reserved fields are reset to zero or junked */
michael@0: #define UCOL_HEADER_MAGIC 0x20030618
michael@0: 
michael@0: typedef struct {
michael@0:       int32_t size;
michael@0:       /* all the offsets are in bytes */
michael@0:       /* to get the address add to the header address and cast properly */
michael@0:       uint32_t options; /* these are the default options for the collator */
michael@0:       uint32_t UCAConsts; /* structure which holds values for indirect positioning and implicit ranges */
michael@0:       uint32_t contractionUCACombos;        /* this one is needed only for UCA, to copy the appropriate contractions */
michael@0:       uint32_t magic;            /* magic number - lets us know whether reserved data is reset or junked */
michael@0:       uint32_t mappingPosition;  /* const uint8_t *mappingPosition; */
michael@0:       uint32_t expansion;        /* uint32_t *expansion;            */
michael@0:       uint32_t contractionIndex; /* UChar *contractionIndex;        */
michael@0:       uint32_t contractionCEs;   /* uint32_t *contractionCEs;       */
michael@0:       uint32_t contractionSize;  /* needed for various closures */
michael@0:       /*int32_t latinOneMapping;*/ /* this is now handled in the trie itself *//* fast track to latin1 chars      */
michael@0: 
michael@0:       uint32_t endExpansionCE;      /* array of last collation element in
michael@0:                                        expansion */
michael@0:       uint32_t expansionCESize;     /* array of maximum expansion size
michael@0:                                        corresponding to the expansion
michael@0:                                        collation elements with last element
michael@0:                                        in endExpansionCE*/
michael@0:       int32_t  endExpansionCECount; /* size of endExpansionCE */
michael@0:       uint32_t unsafeCP;            /* hash table of unsafe code points */
michael@0:       uint32_t contrEndCP;          /* hash table of final code points  */
michael@0:                                     /*   in contractions.               */
michael@0: 
michael@0:       int32_t contractionUCACombosSize;     /* number of UCA contraction items. */
michael@0:                                             /*Length is contractionUCACombosSize*contractionUCACombosWidth*sizeof(UChar) */
michael@0:       UBool jamoSpecial;                    /* is jamoSpecial */
michael@0:       UBool isBigEndian;                    /* is this data big endian? from the UDataInfo header*/
michael@0:       uint8_t charSetFamily;                /* what is the charset family of this data from the UDataInfo header*/
michael@0:       uint8_t contractionUCACombosWidth;    /* width of UCA combos field */
michael@0:       UVersionInfo version;
michael@0:       UVersionInfo UCAVersion;              /* version of the UCA, read from file */
michael@0:       UVersionInfo UCDVersion;              /* UCD version, obtained by u_getUnicodeVersion */
michael@0:       UVersionInfo formatVersion;           /* format version from the UDataInfo header */
michael@0:       uint32_t scriptToLeadByte;            /* offset to script to lead collation byte mapping data */
michael@0:       uint32_t leadByteToScript;            /* offset to lead collation byte to script mapping data */
michael@0:       uint8_t reserved[76];                 /* for future use */
michael@0: } UCATableHeader;
michael@0: 
michael@0: typedef struct {
michael@0:   uint32_t byteSize;
michael@0:   uint32_t tableSize;
michael@0:   uint32_t contsSize;
michael@0:   uint32_t table;
michael@0:   uint32_t conts;
michael@0:   UVersionInfo UCAVersion;              /* version of the UCA, read from file */
michael@0:   uint8_t padding[8];
michael@0: } InverseUCATableHeader;
michael@0: 
michael@0: #endif  /* !UCONFIG_NO_COLLATION */
michael@0: 
michael@0: #endif  /* __UCOL_DATA_H__ */