pdf2swf/xpdf/UnicodeMap.cc

   1 //========================================================================
   2 //
   3 // UnicodeMap.cc
   4 //
   5 // Copyright 2001-2003 Glyph & Cog, LLC
   6 //
   7 //========================================================================
   8
   9 #include <aconf.h>
  10
  11 #ifdef USE_GCC_PRAGMAS
  12 #pragma implementation
  13 #endif
  14
  15 #include <stdio.h>
  16 #include <string.h>
  17 #include "gmem.h"
  18 #include "gfile.h"
  19 #include "GString.h"
  20 #include "GList.h"
  21 #include "Error.h"
  22 #include "GlobalParams.h"
  23 #include "UnicodeMap.h"
  24
  25 //------------------------------------------------------------------------
  26
  27 #define maxExtCode 16
  28
  29 struct UnicodeMapExt {
  30   Unicode u;                    // Unicode char
  31   char code[maxExtCode];
  32   Guint nBytes;
  33 };
  34
  35 //------------------------------------------------------------------------
  36
  37 UnicodeMap *UnicodeMap::parse(GString *encodingNameA) {
  38   FILE *f;
  39   UnicodeMap *map;
  40   UnicodeMapRange *range;
  41   UnicodeMapExt *eMap;
  42   int size, eMapsSize;
  43   char buf[256];
  44   int line, nBytes, i, x;
  45   char *tok1, *tok2, *tok3;
  46
  47   if (!(f = globalParams->getUnicodeMapFile(encodingNameA))) {
  48     error(-1, "Couldn't find unicodeMap file for the '%s' encoding",
  49           encodingNameA->getCString());
  50     return NULL;
  51   }
  52
  53   map = new UnicodeMap(encodingNameA->copy());
  54
  55   size = 8;
  56   map->ranges = (UnicodeMapRange *)gmalloc(size * sizeof(UnicodeMapRange));
  57   eMapsSize = 0;
  58
  59   line = 1;
  60   while (getLine(buf, sizeof(buf), f)) {
  61     if ((tok1 = strtok(buf, " \t\r\n")) &&
  62         (tok2 = strtok(NULL, " \t\r\n"))) {
  63       if (!(tok3 = strtok(NULL, " \t\r\n"))) {
  64         tok3 = tok2;
  65         tok2 = tok1;
  66       }
  67       nBytes = strlen(tok3) / 2;
  68       if (nBytes <= 4) {
  69         if (map->len == size) {
  70           size *= 2;
  71           map->ranges = (UnicodeMapRange *)
  72             grealloc(map->ranges, size * sizeof(UnicodeMapRange));
  73         }
  74         range = &map->ranges[map->len];
  75         sscanf(tok1, "%x", &range->start);
  76         sscanf(tok2, "%x", &range->end);
  77         sscanf(tok3, "%x", &range->code);
  78         range->nBytes = nBytes;
  79         ++map->len;
  80       } else if (tok2 == tok1) {
  81         if (map->eMapsLen == eMapsSize) {
  82           eMapsSize += 16;
  83           map->eMaps = (UnicodeMapExt *)
  84             grealloc(map->eMaps, eMapsSize * sizeof(UnicodeMapExt));
  85         }
  86         eMap = &map->eMaps[map->eMapsLen];
  87         sscanf(tok1, "%x", &eMap->u);
  88         for (i = 0; i < nBytes; ++i) {
  89           sscanf(tok3 + i*2, "%2x", &x);
  90           eMap->code[i] = (char)x;
  91         }
  92         eMap->nBytes = nBytes;
  93         ++map->eMapsLen;
  94       } else {
  95         error(-1, "Bad line (%d) in unicodeMap file for the '%s' encoding",
  96               line, encodingNameA->getCString());
  97       }
  98     } else {
  99       error(-1, "Bad line (%d) in unicodeMap file for the '%s' encoding",
 100             line, encodingNameA->getCString());
 101     }
 102     ++line;
 103   }
 104
 105   fclose(f);
 106
 107   return map;
 108 }
 109
 110 UnicodeMap::UnicodeMap(GString *encodingNameA) {
 111   encodingName = encodingNameA;
 112   unicodeOut = gFalse;
 113   kind = unicodeMapUser;
 114   ranges = NULL;
 115   len = 0;
 116   eMaps = NULL;
 117   eMapsLen = 0;
 118   refCnt = 1;
 119 #if MULTITHREADED
 120   gInitMutex(&mutex);
 121 #endif
 122 }
 123
 124 UnicodeMap::UnicodeMap(char *encodingNameA, GBool unicodeOutA,
 125                        UnicodeMapRange *rangesA, int lenA) {
 126   encodingName = new GString(encodingNameA);
 127   unicodeOut = unicodeOutA;
 128   kind = unicodeMapResident;
 129   ranges = rangesA;
 130   len = lenA;
 131   eMaps = NULL;
 132   eMapsLen = 0;
 133   refCnt = 1;
 134 #if MULTITHREADED
 135   gInitMutex(&mutex);
 136 #endif
 137 }
 138
 139 UnicodeMap::UnicodeMap(char *encodingNameA, GBool unicodeOutA,
 140                        UnicodeMapFunc funcA) {
 141   encodingName = new GString(encodingNameA);
 142   unicodeOut = unicodeOutA;
 143   kind = unicodeMapFunc;
 144   func = funcA;
 145   eMaps = NULL;
 146   eMapsLen = 0;
 147   refCnt = 1;
 148 #if MULTITHREADED
 149   gInitMutex(&mutex);
 150 #endif
 151 }
 152
 153 UnicodeMap::~UnicodeMap() {
 154   delete encodingName;
 155   if (kind == unicodeMapUser && ranges) {
 156     gfree(ranges);
 157   }
 158   if (eMaps) {
 159     gfree(eMaps);
 160   }
 161 #if MULTITHREADED
 162   gDestroyMutex(&mutex);
 163 #endif
 164 }
 165
 166 void UnicodeMap::incRefCnt() {
 167 #if MULTITHREADED
 168   gLockMutex(&mutex);
 169 #endif
 170   ++refCnt;
 171 #if MULTITHREADED
 172   gUnlockMutex(&mutex);
 173 #endif
 174 }
 175
 176 void UnicodeMap::decRefCnt() {
 177   GBool done;
 178
 179 #if MULTITHREADED
 180   gLockMutex(&mutex);
 181 #endif
 182   done = --refCnt == 0;
 183 #if MULTITHREADED
 184   gUnlockMutex(&mutex);
 185 #endif
 186   if (done) {
 187     delete this;
 188   }
 189 }
 190
 191 GBool UnicodeMap::match(GString *encodingNameA) {
 192   return !encodingName->cmp(encodingNameA);
 193 }
 194
 195 int UnicodeMap::mapUnicode(Unicode u, char *buf, int bufSize) {
 196   int a, b, m, n, i, j;
 197   Guint code;
 198
 199   if (kind == unicodeMapFunc) {
 200     return (*func)(u, buf, bufSize);
 201   }
 202
 203   a = 0;
 204   b = len;
 205   if (u >= ranges[a].start) {
 206     // invariant: ranges[a].start <= u < ranges[b].start
 207     while (b - a > 1) {
 208       m = (a + b) / 2;
 209       if (u >= ranges[m].start) {
 210         a = m;
 211       } else if (u < ranges[m].start) {
 212         b = m;
 213       }
 214     }
 215     if (u <= ranges[a].end) {
 216       n = ranges[a].nBytes;
 217       if (n > bufSize) {
 218         return 0;
 219       }
 220       code = ranges[a].code + (u - ranges[a].start);
 221       for (i = n - 1; i >= 0; --i) {
 222         buf[i] = (char)(code & 0xff);
 223         code >>= 8;
 224       }
 225       return n;
 226     }
 227   }
 228
 229   for (i = 0; i < eMapsLen; ++i) {
 230     if (eMaps[i].u == u) {
 231       n = eMaps[i].nBytes;
 232       for (j = 0; j < n; ++j) {
 233         buf[j] = eMaps[i].code[j];
 234       }
 235       return n;
 236     }
 237   }
 238
 239   return 0;
 240 }
 241
 242 //------------------------------------------------------------------------
 243
 244 UnicodeMapCache::UnicodeMapCache() {
 245   int i;
 246
 247   for (i = 0; i < unicodeMapCacheSize; ++i) {
 248     cache[i] = NULL;
 249   }
 250 }
 251
 252 UnicodeMapCache::~UnicodeMapCache() {
 253   int i;
 254
 255   for (i = 0; i < unicodeMapCacheSize; ++i) {
 256     if (cache[i]) {
 257       cache[i]->decRefCnt();
 258     }
 259   }
 260 }
 261
 262 UnicodeMap *UnicodeMapCache::getUnicodeMap(GString *encodingName) {
 263   UnicodeMap *map;
 264   int i, j;
 265
 266   if (cache[0] && cache[0]->match(encodingName)) {
 267     cache[0]->incRefCnt();
 268     return cache[0];
 269   }
 270   for (i = 1; i < unicodeMapCacheSize; ++i) {
 271     if (cache[i] && cache[i]->match(encodingName)) {
 272       map = cache[i];
 273       for (j = i; j >= 1; --j) {
 274         cache[j] = cache[j - 1];
 275       }
 276       cache[0] = map;
 277       map->incRefCnt();
 278       return map;
 279     }
 280   }
 281   if ((map = UnicodeMap::parse(encodingName))) {
 282     if (cache[unicodeMapCacheSize - 1]) {
 283       cache[unicodeMapCacheSize - 1]->decRefCnt();
 284     }
 285     for (j = unicodeMapCacheSize - 1; j >= 1; --j) {
 286       cache[j] = cache[j - 1];
 287     }
 288     cache[0] = map;
 289     map->incRefCnt();
 290     return map;
 291   }
 292   return NULL;
 293 }