icuSources/tools/toolutil/ucm.h

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