source/libs/poppler/poppler-src/fofi/FoFiIdentifier.cc

   1 //========================================================================
   2 //
   3 // FoFiIdentifier.cc
   4 //
   5 // Copyright 2009 Glyph & Cog, LLC
   6 //
   7 //========================================================================
   8
   9 //========================================================================
  10 //
  11 // Modified under the Poppler project - http://poppler.freedesktop.org
  12 //
  13 // All changes made under the Poppler project to this file are licensed
  14 // under GPL version 2 or later
  15 //
  16 // Copyright (C) 2013 Christoph Duelli <duelli@melosgmbh.de>
  17 //
  18 // To see a description of the changes please see the Changelog file that
  19 // came with your tarball or type make ChangeLog if you are building from git
  20 //
  21 //========================================================================
  22
  23 #ifdef USE_GCC_PRAGMAS
  24 #pragma implementation
  25 #endif
  26
  27 #include <stdio.h>
  28 #include <string.h>
  29 #include <limits.h>
  30 #include "goo/gtypes.h"
  31 #include "FoFiIdentifier.h"
  32
  33 //------------------------------------------------------------------------
  34
  35 namespace { // do not pollute global namespace
  36
  37 class Reader {
  38 public:
  39
  40   virtual ~Reader() {}
  41
  42   // Read one byte.  Returns -1 if past EOF.
  43   virtual int getByte(int pos) = 0;
  44
  45   // Read a big-endian unsigned 16-bit integer.  Fills in *val and
  46   // returns true if successful.
  47   virtual GBool getU16BE(int pos, int *val) = 0;
  48
  49   // Read a big-endian unsigned 32-bit integer.  Fills in *val and
  50   // returns true if successful.
  51   virtual GBool getU32BE(int pos, Guint *val) = 0;
  52
  53   // Read a little-endian unsigned 32-bit integer.  Fills in *val and
  54   // returns true if successful.
  55   virtual GBool getU32LE(int pos, Guint *val) = 0;
  56
  57   // Read a big-endian unsigned <size>-byte integer, where 1 <= size
  58   // <= 4.  Fills in *val and returns true if successful.
  59   virtual GBool getUVarBE(int pos, int size, Guint *val) = 0;
  60
  61   // Compare against a string.  Returns true if equal.
  62   virtual GBool cmp(int pos, const char *s) = 0;
  63
  64 };
  65
  66 //------------------------------------------------------------------------
  67
  68 class MemReader: public Reader {
  69 public:
  70
  71   static MemReader *make(char *bufA, int lenA);
  72   virtual ~MemReader();
  73   virtual int getByte(int pos);
  74   virtual GBool getU16BE(int pos, int *val);
  75   virtual GBool getU32BE(int pos, Guint *val);
  76   virtual GBool getU32LE(int pos, Guint *val);
  77   virtual GBool getUVarBE(int pos, int size, Guint *val);
  78   virtual GBool cmp(int pos, const char *s);
  79
  80 private:
  81
  82   MemReader(char *bufA, int lenA);
  83
  84   char *buf;
  85   int len;
  86 };
  87
  88 MemReader *MemReader::make(char *bufA, int lenA) {
  89   return new MemReader(bufA, lenA);
  90 }
  91
  92 MemReader::MemReader(char *bufA, int lenA) {
  93   buf = bufA;
  94   len = lenA;
  95 }
  96
  97 MemReader::~MemReader() {
  98 }
  99
 100 int MemReader::getByte(int pos) {
 101   if (pos < 0 || pos >= len) {
 102     return -1;
 103   }
 104   return buf[pos] & 0xff;
 105 }
 106
 107 GBool MemReader::getU16BE(int pos, int *val) {
 108   if (pos < 0 || pos > len - 2) {
 109     return gFalse;
 110   }
 111   *val = ((buf[pos] & 0xff) << 8) +
 112          (buf[pos+1] & 0xff);
 113   return gTrue;
 114 }
 115
 116 GBool MemReader::getU32BE(int pos, Guint *val) {
 117   if (pos < 0 || pos > len - 4) {
 118     return gFalse;
 119   }
 120   *val = ((buf[pos] & 0xff) << 24) +
 121          ((buf[pos+1] & 0xff) << 16) +
 122          ((buf[pos+2] & 0xff) << 8) +
 123          (buf[pos+3] & 0xff);
 124   return gTrue;
 125 }
 126
 127 GBool MemReader::getU32LE(int pos, Guint *val) {
 128   if (pos < 0 || pos > len - 4) {
 129     return gFalse;
 130   }
 131   *val = (buf[pos] & 0xff) +
 132          ((buf[pos+1] & 0xff) << 8) +
 133          ((buf[pos+2] & 0xff) << 16) +
 134          ((buf[pos+3] & 0xff) << 24);
 135   return gTrue;
 136 }
 137
 138 GBool MemReader::getUVarBE(int pos, int size, Guint *val) {
 139   int i;
 140
 141   if (size < 1 || size > 4 || pos < 0 || pos > len - size) {
 142     return gFalse;
 143   }
 144   *val = 0;
 145   for (i = 0; i < size; ++i) {
 146     *val = (*val << 8) + (buf[pos + i] & 0xff);
 147   }
 148   return gTrue;
 149 }
 150
 151 GBool MemReader::cmp(int pos, const char *s) {
 152   int n;
 153
 154   n = (int)strlen(s);
 155   if (pos < 0 || len < n || pos > len - n) {
 156     return gFalse;
 157   }
 158   return !memcmp(buf + pos, s, n);
 159 }
 160
 161 //------------------------------------------------------------------------
 162
 163 class FileReader: public Reader {
 164 public:
 165
 166   static FileReader *make(char *fileName);
 167   virtual ~FileReader();
 168   virtual int getByte(int pos);
 169   virtual GBool getU16BE(int pos, int *val);
 170   virtual GBool getU32BE(int pos, Guint *val);
 171   virtual GBool getU32LE(int pos, Guint *val);
 172   virtual GBool getUVarBE(int pos, int size, Guint *val);
 173   virtual GBool cmp(int pos, const char *s);
 174
 175 private:
 176
 177   FileReader(FILE *fA);
 178   GBool fillBuf(int pos, int len);
 179
 180   FILE *f;
 181   char buf[1024];
 182   int bufPos, bufLen;
 183 };
 184
 185 FileReader *FileReader::make(char *fileName) {
 186   FILE *fA;
 187
 188   if (!(fA = fopen(fileName, "rb"))) {
 189     return NULL;
 190   }
 191   return new FileReader(fA);
 192 }
 193
 194 FileReader::FileReader(FILE *fA) {
 195   f = fA;
 196   bufPos = 0;
 197   bufLen = 0;
 198 }
 199
 200 FileReader::~FileReader() {
 201   fclose(f);
 202 }
 203
 204 int FileReader::getByte(int pos) {
 205   if (!fillBuf(pos, 1)) {
 206     return -1;
 207   }
 208   return buf[pos - bufPos] & 0xff;
 209 }
 210
 211 GBool FileReader::getU16BE(int pos, int *val) {
 212   if (!fillBuf(pos, 2)) {
 213     return gFalse;
 214   }
 215   *val = ((buf[pos - bufPos] & 0xff) << 8) +
 216          (buf[pos - bufPos + 1] & 0xff);
 217   return gTrue;
 218 }
 219
 220 GBool FileReader::getU32BE(int pos, Guint *val) {
 221   if (!fillBuf(pos, 4)) {
 222     return gFalse;
 223   }
 224   *val = ((buf[pos - bufPos] & 0xff) << 24) +
 225          ((buf[pos - bufPos + 1] & 0xff) << 16) +
 226          ((buf[pos - bufPos + 2] & 0xff) << 8) +
 227          (buf[pos - bufPos + 3] & 0xff);
 228   return gTrue;
 229 }
 230
 231 GBool FileReader::getU32LE(int pos, Guint *val) {
 232   if (!fillBuf(pos, 4)) {
 233     return gFalse;
 234   }
 235   *val = (buf[pos - bufPos] & 0xff) +
 236          ((buf[pos - bufPos + 1] & 0xff) << 8) +
 237          ((buf[pos - bufPos + 2] & 0xff) << 16) +
 238          ((buf[pos - bufPos + 3] & 0xff) << 24);
 239   return gTrue;
 240 }
 241
 242 GBool FileReader::getUVarBE(int pos, int size, Guint *val) {
 243   int i;
 244
 245   if (size < 1 || size > 4 || !fillBuf(pos, size)) {
 246     return gFalse;
 247   }
 248   *val = 0;
 249   for (i = 0; i < size; ++i) {
 250     *val = (*val << 8) + (buf[pos - bufPos + i] & 0xff);
 251   }
 252   return gTrue;
 253 }
 254
 255 GBool FileReader::cmp(int pos, const char *s) {
 256   int n;
 257
 258   n = (int)strlen(s);
 259   if (!fillBuf(pos, n)) {
 260     return gFalse;
 261   }
 262   return !memcmp(buf - bufPos + pos, s, n);
 263 }
 264
 265 GBool FileReader::fillBuf(int pos, int len) {
 266   if (pos < 0 || len < 0 || len > (int)sizeof(buf) ||
 267       pos > INT_MAX - (int)sizeof(buf)) {
 268     return gFalse;
 269   }
 270   if (pos >= bufPos && pos + len <= bufPos + bufLen) {
 271     return gTrue;
 272   }
 273   if (fseek(f, pos, SEEK_SET)) {
 274     return gFalse;
 275   }
 276   bufPos = pos;
 277   bufLen = (int)fread(buf, 1, sizeof(buf), f);
 278   if (bufLen < len) {
 279     return gFalse;
 280   }
 281   return gTrue;
 282 }
 283
 284 //------------------------------------------------------------------------
 285
 286 class StreamReader: public Reader {
 287 public:
 288
 289   static StreamReader *make(int (*getCharA)(void *data), void *dataA);
 290   virtual ~StreamReader();
 291   virtual int getByte(int pos);
 292   virtual GBool getU16BE(int pos, int *val);
 293   virtual GBool getU32BE(int pos, Guint *val);
 294   virtual GBool getU32LE(int pos, Guint *val);
 295   virtual GBool getUVarBE(int pos, int size, Guint *val);
 296   virtual GBool cmp(int pos, const char *s);
 297
 298 private:
 299
 300   StreamReader(int (*getCharA)(void *data), void *dataA);
 301   GBool fillBuf(int pos, int len);
 302
 303   int (*getChar)(void *data);
 304   void *data;
 305   int streamPos;
 306   char buf[1024];
 307   int bufPos, bufLen;
 308 };
 309
 310 StreamReader *StreamReader::make(int (*getCharA)(void *data), void *dataA) {
 311   return new StreamReader(getCharA, dataA);
 312 }
 313
 314 StreamReader::StreamReader(int (*getCharA)(void *data), void *dataA) {
 315   getChar = getCharA;
 316   data = dataA;
 317   streamPos = 0;
 318   bufPos = 0;
 319   bufLen = 0;
 320 }
 321
 322 StreamReader::~StreamReader() {
 323 }
 324
 325 int StreamReader::getByte(int pos) {
 326   if (!fillBuf(pos, 1)) {
 327     return -1;
 328   }
 329   return buf[pos - bufPos] & 0xff;
 330 }
 331
 332 GBool StreamReader::getU16BE(int pos, int *val) {
 333   if (!fillBuf(pos, 2)) {
 334     return gFalse;
 335   }
 336   *val = ((buf[pos - bufPos] & 0xff) << 8) +
 337          (buf[pos - bufPos + 1] & 0xff);
 338   return gTrue;
 339 }
 340
 341 GBool StreamReader::getU32BE(int pos, Guint *val) {
 342   if (!fillBuf(pos, 4)) {
 343     return gFalse;
 344   }
 345   *val = ((buf[pos - bufPos] & 0xff) << 24) +
 346          ((buf[pos - bufPos + 1] & 0xff) << 16) +
 347          ((buf[pos - bufPos + 2] & 0xff) << 8) +
 348          (buf[pos - bufPos + 3] & 0xff);
 349   return gTrue;
 350 }
 351
 352 GBool StreamReader::getU32LE(int pos, Guint *val) {
 353   if (!fillBuf(pos, 4)) {
 354     return gFalse;
 355   }
 356   *val = (buf[pos - bufPos] & 0xff) +
 357          ((buf[pos - bufPos + 1] & 0xff) << 8) +
 358          ((buf[pos - bufPos + 2] & 0xff) << 16) +
 359          ((buf[pos - bufPos + 3] & 0xff) << 24);
 360   return gTrue;
 361 }
 362
 363 GBool StreamReader::getUVarBE(int pos, int size, Guint *val) {
 364   int i;
 365
 366   if (size < 1 || size > 4 || !fillBuf(pos, size)) {
 367     return gFalse;
 368   }
 369   *val = 0;
 370   for (i = 0; i < size; ++i) {
 371     *val = (*val << 8) + (buf[pos - bufPos + i] & 0xff);
 372   }
 373   return gTrue;
 374 }
 375
 376 GBool StreamReader::cmp(int pos, const char *s) {
 377   int n;
 378
 379   n = (int)strlen(s);
 380   if (!fillBuf(pos, n)) {
 381     return gFalse;
 382   }
 383   return !memcmp(buf - bufPos + pos, s, n);
 384 }
 385
 386 GBool StreamReader::fillBuf(int pos, int len) {
 387   int c;
 388
 389   if (pos < 0 || len < 0 || len > (int)sizeof(buf) ||
 390       pos > INT_MAX - (int)sizeof(buf)) {
 391     return gFalse;
 392   }
 393   if (pos < bufPos) {
 394     return gFalse;
 395   }
 396
 397   // if requested region will not fit in the current buffer...
 398   if (pos + len > bufPos + (int)sizeof(buf)) {
 399
 400     // if the start of the requested data is already in the buffer, move
 401     // it to the start of the buffer
 402     if (pos < bufPos + bufLen) {
 403       bufLen -= pos - bufPos;
 404       memmove(buf, buf + (pos - bufPos), bufLen);
 405       bufPos = pos;
 406
 407     // otherwise discard data from the
 408     // stream until we get to the requested position
 409     } else {
 410       bufPos += bufLen;
 411       bufLen = 0;
 412       while (bufPos < pos) {
 413         if ((c = (*getChar)(data)) < 0) {
 414           return gFalse;
 415         }
 416         ++bufPos;
 417       }
 418     }
 419   }
 420
 421   // read the rest of the requested data
 422   while (bufPos + bufLen < pos + len) {
 423     if ((c = (*getChar)(data)) < 0) {
 424       return gFalse;
 425     }
 426     buf[bufLen++] = (char)c;
 427   }
 428
 429   return gTrue;
 430 }
 431
 432 }
 433
 434 //------------------------------------------------------------------------
 435
 436 static FoFiIdentifierType identify(Reader *reader);
 437 static FoFiIdentifierType identifyOpenType(Reader *reader);
 438 static FoFiIdentifierType identifyCFF(Reader *reader, int start);
 439
 440 FoFiIdentifierType FoFiIdentifier::identifyMem(char *file, int len) {
 441   MemReader *reader;
 442   FoFiIdentifierType type;
 443
 444   if (!(reader = MemReader::make(file, len))) {
 445     return fofiIdError;
 446   }
 447   type = identify(reader);
 448   delete reader;
 449   return type;
 450 }
 451
 452 FoFiIdentifierType FoFiIdentifier::identifyFile(char *fileName) {
 453   FileReader *reader;
 454   FoFiIdentifierType type;
 455
 456   if (!(reader = FileReader::make(fileName))) {
 457     return fofiIdError;
 458   }
 459   type = identify(reader);
 460   delete reader;
 461   return type;
 462 }
 463
 464 FoFiIdentifierType FoFiIdentifier::identifyStream(int (*getChar)(void *data),
 465                                                   void *data) {
 466   StreamReader *reader;
 467   FoFiIdentifierType type;
 468
 469   if (!(reader = StreamReader::make(getChar, data))) {
 470     return fofiIdError;
 471   }
 472   type = identify(reader);
 473   delete reader;
 474   return type;
 475 }
 476
 477 static FoFiIdentifierType identify(Reader *reader) {
 478   Guint n;
 479
 480   //----- PFA
 481   if (reader->cmp(0, "%!PS-AdobeFont-1") ||
 482       reader->cmp(0, "%!FontType1")) {
 483     return fofiIdType1PFA;
 484   }
 485
 486   //----- PFB
 487   if (reader->getByte(0) == 0x80 &&
 488       reader->getByte(1) == 0x01 &&
 489       reader->getU32LE(2, &n)) {
 490     if ((n >= 16 && reader->cmp(6, "%!PS-AdobeFont-1")) ||
 491         (n >= 11 && reader->cmp(6, "%!FontType1"))) {
 492       return fofiIdType1PFB;
 493     }
 494   }
 495
 496   //----- TrueType
 497   if ((reader->getByte(0) == 0x00 &&
 498        reader->getByte(1) == 0x01 &&
 499        reader->getByte(2) == 0x00 &&
 500        reader->getByte(3) == 0x00) ||
 501       (reader->getByte(0) == 0x74 &&    // 'true'
 502        reader->getByte(1) == 0x72 &&
 503        reader->getByte(2) == 0x75 &&
 504        reader->getByte(3) == 0x65)) {
 505     return fofiIdTrueType;
 506   }
 507   if (reader->getByte(0) == 0x74 &&     // 'ttcf'
 508       reader->getByte(1) == 0x74 &&
 509       reader->getByte(2) == 0x63 &&
 510       reader->getByte(3) == 0x66) {
 511     return fofiIdTrueTypeCollection;
 512   }
 513
 514   //----- OpenType
 515   if (reader->getByte(0) == 0x4f &&     // 'OTTO
 516       reader->getByte(1) == 0x54 &&
 517       reader->getByte(2) == 0x54 &&
 518       reader->getByte(3) == 0x4f) {
 519     return identifyOpenType(reader);
 520   }
 521
 522   //----- CFF
 523   if (reader->getByte(0) == 0x01 &&
 524       reader->getByte(1) == 0x00) {
 525     return identifyCFF(reader, 0);
 526   }
 527   // some tools embed CFF fonts with an extra whitespace char at the
 528   // beginning
 529   if (reader->getByte(1) == 0x01 &&
 530       reader->getByte(2) == 0x00) {
 531     return identifyCFF(reader, 1);
 532   }
 533
 534   return fofiIdUnknown;
 535 }
 536
 537 static FoFiIdentifierType identifyOpenType(Reader *reader) {
 538   FoFiIdentifierType type;
 539   Guint offset;
 540   int nTables, i;
 541
 542   if (!reader->getU16BE(4, &nTables)) {
 543     return fofiIdUnknown;
 544   }
 545   for (i = 0; i < nTables; ++i) {
 546     if (reader->cmp(12 + i*16, "CFF ")) {
 547       if (reader->getU32BE(12 + i*16 + 8, &offset) &&
 548           offset < (Guint)INT_MAX) {
 549         type = identifyCFF(reader, (int)offset);
 550         if (type == fofiIdCFF8Bit) {
 551           type = fofiIdOpenTypeCFF8Bit;
 552         } else if (type == fofiIdCFFCID) {
 553           type = fofiIdOpenTypeCFFCID;
 554         }
 555         return type;
 556       }
 557       return fofiIdUnknown;
 558     }
 559   }
 560   return fofiIdUnknown;
 561 }
 562
 563 static FoFiIdentifierType identifyCFF(Reader *reader, int start) {
 564   Guint offset0, offset1;
 565   int hdrSize, offSize0, offSize1, pos, endPos, b0, n, i;
 566
 567   //----- read the header
 568   if (reader->getByte(start) != 0x01 ||
 569       reader->getByte(start + 1) != 0x00) {
 570     return fofiIdUnknown;
 571   }
 572   if ((hdrSize = reader->getByte(start + 2)) < 0) {
 573     return fofiIdUnknown;
 574   }
 575   if ((offSize0 = reader->getByte(start + 3)) < 1 || offSize0 > 4) {
 576     return fofiIdUnknown;
 577   }
 578   pos = start + hdrSize;
 579   if (pos < 0) {
 580     return fofiIdUnknown;
 581   }
 582
 583   //----- skip the name index
 584   if (!reader->getU16BE(pos, &n)) {
 585     return fofiIdUnknown;
 586   }
 587   if (n == 0) {
 588     pos += 2;
 589   } else {
 590     if ((offSize1 = reader->getByte(pos + 2)) < 1 || offSize1 > 4) {
 591       return fofiIdUnknown;
 592     }
 593     if (!reader->getUVarBE(pos + 3 + n * offSize1, offSize1, &offset1) ||
 594         offset1 > (Guint)INT_MAX) {
 595       return fofiIdUnknown;
 596     }
 597     pos += 3 + (n + 1) * offSize1 + (int)offset1 - 1;
 598   }
 599   if (pos < 0) {
 600     return fofiIdUnknown;
 601   }
 602
 603   //----- parse the top dict index
 604   if (!reader->getU16BE(pos, &n) || n < 1) {
 605     return fofiIdUnknown;
 606   }
 607   if ((offSize1 = reader->getByte(pos + 2)) < 1 || offSize1 > 4) {
 608     return fofiIdUnknown;
 609   }
 610   if (!reader->getUVarBE(pos + 3, offSize1, &offset0) ||
 611       offset0 > (Guint)INT_MAX ||
 612       !reader->getUVarBE(pos + 3 + offSize1, offSize1, &offset1) ||
 613       offset1 > (Guint)INT_MAX ||
 614       offset0 > offset1) {
 615     return fofiIdUnknown;
 616   }
 617   pos = pos + 3 + (n + 1) * offSize1 + (int)offset0 - 1;
 618   endPos = pos + 3 + (n + 1) * offSize1 + (int)offset1 - 1;
 619   if (pos < 0 || endPos < 0 || pos > endPos) {
 620     return fofiIdUnknown;
 621   }
 622
 623   //----- parse the top dict, look for ROS as first entry
 624   // for a CID font, the top dict starts with:
 625   //     <int> <int> <int> ROS
 626   for (i = 0; i < 3; ++i) {
 627     b0 = reader->getByte(pos++);
 628     if (b0 == 0x1c) {
 629       pos += 2;
 630     } else if (b0 == 0x1d) {
 631       pos += 4;
 632     } else if (b0 >= 0xf7 && b0 <= 0xfe) {
 633       pos += 1;
 634     } else if (b0 < 0x20 || b0 > 0xf6) {
 635       return fofiIdCFF8Bit;
 636     }
 637     if (pos >= endPos || pos < 0) {
 638       return fofiIdCFF8Bit;
 639     }
 640   }
 641   if (pos + 1 < endPos &&
 642       reader->getByte(pos) == 12 &&
 643       reader->getByte(pos + 1) == 30) {
 644     return fofiIdCFFCID;
 645   } else {
 646     return fofiIdCFF8Bit;
 647   }
 648 }