michael@0: /*
michael@0: *******************************************************************************
michael@0: *
michael@0: *   Copyright (C) 2000-2008, International Business Machines
michael@0: *   Corporation and others.  All Rights Reserved.
michael@0: *
michael@0: *******************************************************************************
michael@0: *   file name:  genmbcs.h
michael@0: *   encoding:   US-ASCII
michael@0: *   tab size:   8 (not used)
michael@0: *   indentation:4
michael@0: *
michael@0: *   created on: 2000jul10
michael@0: *   created by: Markus W. Scherer
michael@0: */
michael@0: 
michael@0: #ifndef __GENMBCS_H__
michael@0: #define __GENMBCS_H__
michael@0: 
michael@0: #include "makeconv.h"
michael@0: 
michael@0: enum {
michael@0:     /*
michael@0:      * TODO: Consider using ucnvmbcs.h constants.
michael@0:      * However, not all values need to be exactly the same, for example
michael@0:      * the xxx_UTF8_MAX values may be different. (Especially SBCS_UTF8_MAX
michael@0:      * may be higher in makeconv than in the runtime code because that
michael@0:      * affects only a small number of .cnv files [if any] but all
michael@0:      * runtime UConverterSharedData objects.
michael@0:      */
michael@0:     MBCS_STAGE_2_SHIFT=4,
michael@0:     MBCS_STAGE_2_BLOCK_SIZE=0x40,       /* =64=1<<6 for 6 bits in stage 2 */
michael@0:     MBCS_STAGE_2_BLOCK_SIZE_SHIFT=6,    /* log2(MBCS_STAGE_2_BLOCK_SIZE) */
michael@0:     MBCS_STAGE_2_BLOCK_MASK=0x3f,       /* for after shifting by MBCS_STAGE_2_SHIFT */
michael@0:     MBCS_STAGE_1_SHIFT=10,
michael@0:     MBCS_STAGE_1_BMP_SIZE=0x40, /* 0x10000>>MBCS_STAGE_1_SHIFT, or 16 for one entry per 1k code points on the BMP */
michael@0:     MBCS_STAGE_1_SIZE=0x440,    /* 0x110000>>MBCS_STAGE_1_SHIFT, or 17*64 for one entry per 1k code points */
michael@0:     MBCS_STAGE_2_SIZE=0xfbc0,   /* 0x10000-MBCS_STAGE_1_SIZE: stages 1 & 2 share a 16-bit-indexed array */
michael@0:     MBCS_MAX_STAGE_2_TOP=MBCS_STAGE_2_SIZE,
michael@0:     MBCS_STAGE_2_MAX_BLOCKS=MBCS_STAGE_2_SIZE>>MBCS_STAGE_2_BLOCK_SIZE_SHIFT,
michael@0: 
michael@0:     MBCS_STAGE_2_ALL_UNASSIGNED_INDEX=0, /* stage 1 entry for the all-unassigned stage 2 block */
michael@0:     MBCS_STAGE_2_FIRST_ASSIGNED=MBCS_STAGE_2_BLOCK_SIZE, /* start of the first stage 2 block after the all-unassigned one */
michael@0: 
michael@0:     MBCS_STAGE_3_BLOCK_SIZE=16,         /* =16=1<<4 for 4 bits in stage 3 */
michael@0:     MBCS_STAGE_3_BLOCK_MASK=0xf,
michael@0:     MBCS_STAGE_3_FIRST_ASSIGNED=MBCS_STAGE_3_BLOCK_SIZE, /* start of the first stage 3 block after the all-unassigned one */
michael@0: 
michael@0:     MBCS_STAGE_3_GRANULARITY=16,        /* =1<<4: MBCS stage 2 indexes are shifted left 4 */
michael@0:     MBCS_STAGE_3_SBCS_SIZE=0x10000,     /* max 64k mappings for SBCS */
michael@0:     MBCS_STAGE_3_MBCS_SIZE=0x10000*MBCS_STAGE_3_GRANULARITY, /* max mappings for MBCS */
michael@0: 
michael@0:     /*
michael@0:      * SBCS_UTF8_MAX: Maximum code point with UTF-8-friendly SBCS data structures.
michael@0:      * Possible values are 0x01ff..0xffff, in steps of 0x100.
michael@0:      *
michael@0:      * Unlike for MBCS, this constant only affects the stage 3 block allocation size;
michael@0:      * there is no additional stage 1/2 table stored in the .cnv file.
michael@0:      * The max value should be at least 0x7ff to cover 2-byte UTF-8.
michael@0:      * 0xfff also covers a number other small scripts which have legacy charsets
michael@0:      * (like Thai).
michael@0:      * Higher values up to 0x1fff are harmless and potentially useful because
michael@0:      * that covers small-script blocks which usually have either dense mappings
michael@0:      * or no mappings at all.
michael@0:      * Starting at U+2000, there are mostly symbols and format characters
michael@0:      * with a low density of SBCS mappings, which would result in more wasted
michael@0:      * stage 3 entries with the larger block size.
michael@0:      */
michael@0:     SBCS_UTF8_MAX=0x1fff,
michael@0: 
michael@0:     /*
michael@0:      * MBCS_UTF8_MAX: Maximum code point with UTF-8-friendly MBCS data structures.
michael@0:      * Possible values are 0x01ff..0xffff, in steps of 0x100.
michael@0:      *
michael@0:      * Note that with 0xffff, MBCSAddFromUnicode() may overflow the additional UTF-8 stage table
michael@0:      * with extreme input data. The function checks for this overflow.
michael@0:      *
michael@0:      * 0xd7ff is chosen for the majority of common characters including Unihan and Hangul.
michael@0:      * At U+d800 there are mostly surrogates, private use codes, compatibility characters, etc.
michael@0:      * Larger values cause slightly larger MBCS .cnv files.
michael@0:      */
michael@0:     MBCS_UTF8_MAX=0xd7ff,
michael@0:     MBCS_UTF8_LIMIT=MBCS_UTF8_MAX+1,    /* =0xd800 */
michael@0: 
michael@0:     MBCS_UTF8_STAGE_SHIFT=6,
michael@0:     MBCS_UTF8_STAGE_3_BLOCK_SIZE=0x40,  /* =64=1<<6 for 6 bits from last trail byte */
michael@0:     MBCS_UTF8_STAGE_3_BLOCK_MASK=0x3f,
michael@0: 
michael@0:     /* size of the single-stage table for up to U+d7ff (used instead of stage1/2) */
michael@0:     MBCS_UTF8_STAGE_SIZE=MBCS_UTF8_LIMIT>>MBCS_UTF8_STAGE_SHIFT, /* =0x360 */
michael@0: 
michael@0:     MBCS_FROM_U_EXT_FLAG=0x10,          /* UCMapping.f bit for base table mappings that fit into the base toU table */
michael@0:     MBCS_FROM_U_EXT_MASK=0x0f,          /* but need to go into the extension fromU table */
michael@0: 
michael@0:     /* =4 number of regular stage 3 blocks for final UTF-8 trail byte */
michael@0:     MBCS_UTF8_STAGE_3_BLOCKS=MBCS_UTF8_STAGE_3_BLOCK_SIZE/MBCS_STAGE_3_BLOCK_SIZE,
michael@0: 
michael@0:     MBCS_MAX_FALLBACK_COUNT=8192
michael@0: };
michael@0: 
michael@0: U_CFUNC NewConverter *
michael@0: MBCSOpen(UCMFile *ucm);
michael@0: 
michael@0: struct MBCSData;
michael@0: typedef struct MBCSData MBCSData;
michael@0: 
michael@0: /*
michael@0:  * Get a dummy MBCSData for use with MBCSOkForBaseFromUnicode()
michael@0:  * for creating an extension-only file.
michael@0:  * Assume maxCharLength>1.
michael@0:  */
michael@0: U_CFUNC const MBCSData *
michael@0: MBCSGetDummy(void);
michael@0: 
michael@0: /* Test if a 1:1 mapping fits into the MBCS base table's fromUnicode structure. */
michael@0: U_CFUNC UBool
michael@0: MBCSOkForBaseFromUnicode(const MBCSData *mbcsData,
michael@0:                          const uint8_t *bytes, int32_t length,
michael@0:                          UChar32 c, int8_t flag);
michael@0: 
michael@0: U_CFUNC NewConverter *
michael@0: CnvExtOpen(UCMFile *ucm);
michael@0: 
michael@0: #endif /* __GENMBCS_H__ */