icuSources/tools/toolutil/ucm.h

   1 /*
   2  *******************************************************************************
   3  *   Copyright (C) 2003-2007, International Business Machines
   4  *   Corporation and others.  All Rights Reserved.
   5  *******************************************************************************
   6  *   file name:  ucm.h
   7  *   encoding:   US-ASCII
   8  *   tab size:   8 (not used)
   9  *   indentation:4
  10  *
  11  *   created on: 2003jun20
  12  *   created by: Markus W. Scherer
  13  *
  14  *   Definitions for the .ucm file parser and handler module ucm.c.
  15  */
  16
  17 #ifndef __UCM_H__
  18 #define __UCM_H__
  19
  20 #include "unicode/utypes.h"
  21 #include "ucnvmbcs.h"
  22 #include "ucnv_ext.h"
  23 #include "filestrm.h"
  24 #include <stdio.h>
  25
  26 #if !UCONFIG_NO_CONVERSION
  27
  28 U_CDECL_BEGIN
  29
  30 /* constants for UCMapping.moveFlag */
  31 enum {
  32     UCM_MOVE_TO_EXT=1,
  33     UCM_REMOVE_MAPPING=2
  34 };
  35
  36 /*
  37  * Per-mapping data structure
  38  *
  39  * u if uLen==1: Unicode code point
  40  *   else index to uLen code points
  41  * b if bLen<=4: up to 4 bytes
  42  *   else index to bLen bytes
  43  * uLen number of code points
  44  * bLen number of words containing left-justified bytes
  45  * bIsMultipleChars indicates that the bytes contain more than one sequence
  46  *                  according to the state table
  47  * f flag for roundtrip (0), fallback (1), sub mapping (2), reverse fallback (3)
  48  *   same values as in the source file after |
  49  */
  50 typedef struct UCMapping {
  51     UChar32 u;
  52     union {
  53         uint32_t index;
  54         uint8_t bytes[4];
  55     } b;
  56     int8_t uLen, bLen, f, moveFlag;
  57 } UCMapping;
  58
  59 /* constants for UCMTable.flagsType */
  60 enum {
  61     UCM_FLAGS_INITIAL,  /* no mappings parsed yet */
  62     UCM_FLAGS_EXPLICIT, /* .ucm file has mappings with | fallback indicators */
  63     UCM_FLAGS_IMPLICIT, /* .ucm file has mappings without | fallback indicators, later wins */
  64     UCM_FLAGS_MIXED     /* both implicit and explicit */
  65 };
  66
  67 typedef struct UCMTable {
  68     UCMapping *mappings;
  69     int32_t mappingsCapacity, mappingsLength;
  70
  71     UChar32 *codePoints;
  72     int32_t codePointsCapacity, codePointsLength;
  73
  74     uint8_t *bytes;
  75     int32_t bytesCapacity, bytesLength;
  76
  77     /* index map for mapping by bytes first */
  78     int32_t *reverseMap;
  79
  80     uint8_t unicodeMask;
  81     int8_t flagsType; /* UCM_FLAGS_INITIAL etc. */
  82     UBool isSorted;
  83 } UCMTable;
  84
  85 enum {
  86     MBCS_STATE_FLAG_DIRECT=1,
  87     MBCS_STATE_FLAG_SURROGATES,
  88
  89     MBCS_STATE_FLAG_READY=16
  90 };
  91
  92 typedef struct UCMStates {
  93     int32_t stateTable[MBCS_MAX_STATE_COUNT][256];
  94     uint32_t stateFlags[MBCS_MAX_STATE_COUNT],
  95              stateOffsetSum[MBCS_MAX_STATE_COUNT];
  96
  97     int32_t countStates, minCharLength, maxCharLength, countToUCodeUnits;
  98     int8_t conversionType, outputType;
  99 } UCMStates;
 100
 101 typedef struct UCMFile {
 102     UCMTable *base, *ext;
 103     UCMStates states;
 104
 105     char baseName[UCNV_MAX_CONVERTER_NAME_LENGTH];
 106 } UCMFile;
 107
 108 /* simple accesses ---------------------------------------------------------- */
 109
 110 #define UCM_GET_CODE_POINTS(t, m) \
 111     (((m)->uLen==1) ? &(m)->u : (t)->codePoints+(m)->u)
 112
 113 #define UCM_GET_BYTES(t, m) \
 114     (((m)->bLen<=4) ? (m)->b.bytes : (t)->bytes+(m)->b.index)
 115
 116 /* APIs --------------------------------------------------------------------- */
 117
 118 U_CAPI UCMFile * U_EXPORT2
 119 ucm_open(void);
 120
 121 U_CAPI void U_EXPORT2
 122 ucm_close(UCMFile *ucm);
 123
 124 U_CAPI UBool U_EXPORT2
 125 ucm_parseHeaderLine(UCMFile *ucm,
 126                     char *line, char **pKey, char **pValue);
 127
 128 /* @return -1 illegal bytes  0 suitable for base table  1 needs to go into extension table */
 129 U_CAPI int32_t U_EXPORT2
 130 ucm_mappingType(UCMStates *baseStates,
 131                 UCMapping *m,
 132                 UChar32 codePoints[UCNV_EXT_MAX_UCHARS],
 133                 uint8_t bytes[UCNV_EXT_MAX_BYTES]);
 134
 135 /* add a mapping to the base or extension table as appropriate */
 136 U_CAPI UBool U_EXPORT2
 137 ucm_addMappingAuto(UCMFile *ucm, UBool forBase, UCMStates *baseStates,
 138                    UCMapping *m,
 139                    UChar32 codePoints[UCNV_EXT_MAX_UCHARS],
 140                    uint8_t bytes[UCNV_EXT_MAX_BYTES]);
 141
 142 U_CAPI UBool U_EXPORT2
 143 ucm_addMappingFromLine(UCMFile *ucm, const char *line, UBool forBase, UCMStates *baseStates);
 144
 145
 146 U_CAPI UCMTable * U_EXPORT2
 147 ucm_openTable(void);
 148
 149 U_CAPI void U_EXPORT2
 150 ucm_closeTable(UCMTable *table);
 151
 152 U_CAPI void U_EXPORT2
 153 ucm_resetTable(UCMTable *table);
 154
 155 U_CAPI void U_EXPORT2
 156 ucm_sortTable(UCMTable *t);
 157
 158 /*
 159  * Remove mappings with their move flag set from the base table
 160  * and move some of them (with UCM_MOVE_TO_EXT) to the extension table.
 161  */
 162 U_CAPI void U_EXPORT2
 163 ucm_moveMappings(UCMTable *base, UCMTable *ext);
 164
 165 /**
 166  * Read a table from a .ucm file, from after the CHARMAP line to
 167  * including the END CHARMAP line.
 168  */
 169 U_CAPI void U_EXPORT2
 170 ucm_readTable(UCMFile *ucm, FileStream* convFile,
 171               UBool forBase, UCMStates *baseStates,
 172               UErrorCode *pErrorCode);
 173
 174 /**
 175  * Check the validity of mappings against a base table's states;
 176  * necessary for extension-only tables that were read before their base tables.
 177  */
 178 U_CAPI UBool U_EXPORT2
 179 ucm_checkValidity(UCMTable *ext, UCMStates *baseStates);
 180
 181 /**
 182  * Check a base table against an extension table.
 183  * Set the moveTarget!=NULL if it is possible to move mappings from the base.
 184  * This is the case where base and extension tables are parsed from a single file
 185  * (moveTarget==ext)
 186  * or when delta file mappings are subtracted from a base table.
 187  *
 188  * When a base table cannot be modified because a delta file is parsed in makeconv,
 189  * then set moveTarget=NULL.
 190  *
 191  * if(intersectBase) then mappings that exist in the base table but not in
 192  * the extension table are moved to moveTarget instead of showing an error.
 193  *
 194  * Special mode:
 195  * If intersectBase==2 for a DBCS extension table, then SBCS mappings are
 196  * not moved out of the base unless their Unicode input requires it.
 197  * This helps ucmkbase generate base tables for DBCS-only extension .cnv files.
 198  *
 199  * For both tables in the same file, the extension table is automatically
 200  * built.
 201  * For separate files, the extension file can use a complete mapping table (.ucm file),
 202  * so that common mappings need not be stripped out manually.
 203  *
 204  *
 205  * Sort both tables, and then for each mapping direction:
 206  *
 207  * If intersectBase is TRUE and the base table contains a mapping
 208  * that does not exist in the extension table, then this mapping is moved
 209  * to moveTarget.
 210  *
 211  * - otherwise -
 212  *
 213  * If the base table contains a mapping for which the input sequence is
 214  * the same as the extension input, then
 215  * - if the output is the same: remove the extension mapping
 216  * - else: error
 217  *
 218  * If the base table contains a mapping for which the input sequence is
 219  * a prefix of the extension input, then
 220  * - if moveTarget!=NULL: move the base mapping to the moveTarget table
 221  * - else: error
 222  *
 223  * @return FALSE in case of an irreparable error
 224  */
 225 U_CAPI UBool U_EXPORT2
 226 ucm_checkBaseExt(UCMStates *baseStates, UCMTable *base, UCMTable *ext,
 227                  UCMTable *moveTarget, UBool intersectBase);
 228
 229 U_CAPI void U_EXPORT2
 230 ucm_printTable(UCMTable *table, FILE *f, UBool byUnicode);
 231
 232 U_CAPI void U_EXPORT2
 233 ucm_printMapping(UCMTable *table, UCMapping *m, FILE *f);
 234
 235
 236 U_CAPI void U_EXPORT2
 237 ucm_addState(UCMStates *states, const char *s);
 238
 239 U_CAPI void U_EXPORT2
 240 ucm_processStates(UCMStates *states);
 241
 242 U_CAPI int32_t U_EXPORT2
 243 ucm_countChars(UCMStates *states,
 244                const uint8_t *bytes, int32_t length);
 245
 246
 247 U_CAPI int8_t U_EXPORT2
 248 ucm_parseBytes(uint8_t bytes[UCNV_EXT_MAX_BYTES], const char *line, const char **ps);
 249
 250 U_CAPI UBool U_EXPORT2
 251 ucm_parseMappingLine(UCMapping *m,
 252                      UChar32 codePoints[UCNV_EXT_MAX_UCHARS],
 253                      uint8_t bytes[UCNV_EXT_MAX_BYTES],
 254                      const char *line);
 255
 256 U_CAPI void U_EXPORT2
 257 ucm_addMapping(UCMTable *table,
 258                UCMapping *m,
 259                UChar32 codePoints[UCNV_EXT_MAX_UCHARS],
 260                uint8_t bytes[UCNV_EXT_MAX_BYTES]);
 261
 262 /* very makeconv-specific functions ----------------------------------------- */
 263
 264 /* finalize and optimize states after the toUnicode mappings are processed */
 265 U_CAPI void U_EXPORT2
 266 ucm_optimizeStates(UCMStates *states,
 267                    uint16_t **pUnicodeCodeUnits,
 268                    _MBCSToUFallback *toUFallbacks, int32_t countToUFallbacks,
 269                    UBool verbose);
 270
 271 /* moved here because it is used inside ucmstate.c */
 272 U_CAPI int32_t U_EXPORT2
 273 ucm_findFallback(_MBCSToUFallback *toUFallbacks, int32_t countToUFallbacks,
 274                  uint32_t offset);
 275
 276 /* very rptp2ucm-specific functions ----------------------------------------- */
 277
 278 /*
 279  * Input: Separate tables with mappings from/to Unicode,
 280  * subchar and subchar1 (0 if none).
 281  * All mappings must have flag 0.
 282  *
 283  * Output: fromUTable will contain the union of mappings with the correct
 284  * precision flags, and be sorted.
 285  */
 286 U_CAPI void U_EXPORT2
 287 ucm_mergeTables(UCMTable *fromUTable, UCMTable *toUTable,
 288                 const uint8_t *subchar, int32_t subcharLength,
 289                 uint8_t subchar1);
 290
 291 U_CAPI UBool U_EXPORT2
 292 ucm_separateMappings(UCMFile *ucm, UBool isSISO);
 293
 294 U_CDECL_END
 295
 296 #endif
 297
 298 #endif
 299