jbig2stream.cc

来自「source code: Covert TXT to PDF」· CC 代码 · 共 2,563 行 · 第 1/5 页

      readTextRegionSeg(segNum, gFalse, gFalse, segLength, refSegs, nRefSegs);
      break;
    case 6:
      readTextRegionSeg(segNum, gTrue, gFalse, segLength, refSegs, nRefSegs);
      break;
    case 7:
      readTextRegionSeg(segNum, gTrue, gTrue, segLength, refSegs, nRefSegs);
      break;
    case 16:
      readPatternDictSeg(segNum, segLength);
      break;
    case 20:
      readHalftoneRegionSeg(segNum, gFalse, gFalse, segLength,
			    refSegs, nRefSegs);
      break;
    case 22:
      readHalftoneRegionSeg(segNum, gTrue, gFalse, segLength,
			    refSegs, nRefSegs);
      break;
    case 23:
      readHalftoneRegionSeg(segNum, gTrue, gTrue, segLength,
			    refSegs, nRefSegs);
      break;
    case 36:
      readGenericRegionSeg(segNum, gFalse, gFalse, segLength);
      break;
    case 38:
      readGenericRegionSeg(segNum, gTrue, gFalse, segLength);
      break;
    case 39:
      readGenericRegionSeg(segNum, gTrue, gTrue, segLength);
      break;
    case 40:
      readGenericRefinementRegionSeg(segNum, gFalse, gFalse, segLength,
				     refSegs, nRefSegs);
      break;
    case 42:
      readGenericRefinementRegionSeg(segNum, gTrue, gFalse, segLength,
				     refSegs, nRefSegs);
      break;
    case 43:
      readGenericRefinementRegionSeg(segNum, gTrue, gTrue, segLength,
				     refSegs, nRefSegs);
      break;
    case 48:
      readPageInfoSeg(segLength);
      break;
    case 50:
      readEndOfStripeSeg(segLength);
      break;
    case 52:
      readProfilesSeg(segLength);
      break;
    case 53:
      readCodeTableSeg(segNum, segLength);
      break;
    case 62:
      readExtensionSeg(segLength);
      break;
    default:
      error(getPos(), "Unknown segment type in JBIG2 stream");
      for (i = 0; i < segLength; ++i) {
	if ((c1 = curStr->getChar()) == EOF) {
	  goto eofError2;
	}
      }
      break;
    }

    gfree(refSegs);
  }

  return;

 eofError2:
  gfree(refSegs);
 eofError1:
  error(getPos(), "Unexpected EOF in JBIG2 stream");
}

void JBIG2Stream::readSymbolDictSeg(Guint segNum, Guint length,
				    Guint *refSegs, Guint nRefSegs) {
  JBIG2SymbolDict *symbolDict;
  JBIG2HuffmanTable *huffDHTable, *huffDWTable;
  JBIG2HuffmanTable *huffBMSizeTable, *huffAggInstTable;
  JBIG2Segment *seg;
  GList *codeTables;
  JBIG2SymbolDict *inputSymbolDict;
  Guint flags, sdTemplate, sdrTemplate, huff, refAgg;
  Guint huffDH, huffDW, huffBMSize, huffAggInst;
  Guint contextUsed, contextRetained;
  int sdATX[4], sdATY[4], sdrATX[2], sdrATY[2];
  Guint numExSyms, numNewSyms, numInputSyms, symCodeLen;
  JBIG2Bitmap **bitmaps;
  JBIG2Bitmap *collBitmap, *refBitmap;
  Guint *symWidths;
  Guint symHeight, symWidth, totalWidth, x, symID;
  int dh, dw, refAggNum, refDX, refDY, bmSize;
  GBool ex;
  int run, cnt;
  Guint i, j, k;
  Guchar *p;

  // symbol dictionary flags
  if (!readUWord(&flags)) {
    goto eofError;
  }
  sdTemplate = (flags >> 10) & 3;
  sdrTemplate = (flags >> 12) & 1;
  huff = flags & 1;
  refAgg = (flags >> 1) & 1;
  huffDH = (flags >> 2) & 3;
  huffDW = (flags >> 4) & 3;
  huffBMSize = (flags >> 6) & 1;
  huffAggInst = (flags >> 7) & 1;
  contextUsed = (flags >> 8) & 1;
  contextRetained = (flags >> 9) & 1;

  // symbol dictionary AT flags
  if (!huff) {
    if (sdTemplate == 0) {
      if (!readByte(&sdATX[0]) ||
	  !readByte(&sdATY[0]) ||
	  !readByte(&sdATX[1]) ||
	  !readByte(&sdATY[1]) ||
	  !readByte(&sdATX[2]) ||
	  !readByte(&sdATY[2]) ||
	  !readByte(&sdATX[3]) ||
	  !readByte(&sdATY[3])) {
	goto eofError;
      }
    } else {
      if (!readByte(&sdATX[0]) ||
	  !readByte(&sdATY[0])) {
	goto eofError;
      }
    }
  }

  // symbol dictionary refinement AT flags
  if (refAgg && !sdrTemplate) {
    if (!readByte(&sdrATX[0]) ||
	!readByte(&sdrATY[0]) ||
	!readByte(&sdrATX[1]) ||
	!readByte(&sdrATY[1])) {
      goto eofError;
    }
  }

  // SDNUMEXSYMS and SDNUMNEWSYMS
  if (!readULong(&numExSyms) || !readULong(&numNewSyms)) {
    goto eofError;
  }

  // get referenced segments: input symbol dictionaries and code tables
  codeTables = new GList();
  numInputSyms = 0;
  for (i = 0; i < nRefSegs; ++i) {
    seg = findSegment(refSegs[i]);
    if (seg->getType() == jbig2SegSymbolDict) {
      numInputSyms += ((JBIG2SymbolDict *)seg)->getSize();
    } else if (seg->getType() == jbig2SegCodeTable) {
      codeTables->append(seg);
    }
  }

  // compute symbol code length
  symCodeLen = 0;
  i = 1;
  while (i < numInputSyms + numNewSyms) {
    ++symCodeLen;
    i <<= 1;
  }

  // get the input symbol bitmaps
  bitmaps = (JBIG2Bitmap **)gmalloc((numInputSyms + numNewSyms) *
				    sizeof(JBIG2Bitmap *));
  k = 0;
  inputSymbolDict = NULL;
  for (i = 0; i < nRefSegs; ++i) {
    seg = findSegment(refSegs[i]);
    if (seg->getType() == jbig2SegSymbolDict) {
      inputSymbolDict = (JBIG2SymbolDict *)seg;
      for (j = 0; j < inputSymbolDict->getSize(); ++j) {
	bitmaps[k++] = inputSymbolDict->getBitmap(j);
      }
    }
  }

  // get the Huffman tables
  huffDHTable = huffDWTable = NULL; // make gcc happy
  huffBMSizeTable = huffAggInstTable = NULL; // make gcc happy
  i = 0;
  if (huff) {
    if (huffDH == 0) {
      huffDHTable = huffTableD;
    } else if (huffDH == 1) {
      huffDHTable = huffTableE;
    } else {
      huffDHTable = ((JBIG2CodeTable *)codeTables->get(i++))->getHuffTable();
    }
    if (huffDW == 0) {
      huffDWTable = huffTableB;
    } else if (huffDW == 1) {
      huffDWTable = huffTableC;
    } else {
      huffDWTable = ((JBIG2CodeTable *)codeTables->get(i++))->getHuffTable();
    }
    if (huffBMSize == 0) {
      huffBMSizeTable = huffTableA;
    } else {
      huffBMSizeTable =
	  ((JBIG2CodeTable *)codeTables->get(i++))->getHuffTable();
    }
    if (huffAggInst == 0) {
      huffAggInstTable = huffTableA;
    } else {
      huffAggInstTable =
	  ((JBIG2CodeTable *)codeTables->get(i++))->getHuffTable();
    }
  }
  delete codeTables;

  // set up the Huffman decoder
  if (huff) {
    huffDecoder->reset();

  // set up the arithmetic decoder
  } else {
    if (contextUsed && inputSymbolDict) {
      resetGenericStats(sdTemplate, inputSymbolDict->getGenericRegionStats());
      if (refAgg) {
	resetRefinementStats(sdrTemplate,
			     inputSymbolDict->getRefinementRegionStats());
      }
    } else {
      resetGenericStats(sdTemplate, NULL);
      if (refAgg) {
	resetRefinementStats(sdrTemplate, NULL);
      }
    }
    resetIntStats(symCodeLen);
    arithDecoder->start();
  }

  // allocate symbol widths storage
  symWidths = NULL;
  if (huff && !refAgg) {
    symWidths = (Guint *)gmalloc(numNewSyms * sizeof(Guint));
  }

  symHeight = 0;
  i = 0;
  while (i < numNewSyms) {

    // read the height class delta height
    if (huff) {
      huffDecoder->decodeInt(&dh, huffDHTable);
    } else {
      arithDecoder->decodeInt(&dh, iadhStats);
    }
    symHeight += dh;
    symWidth = 0;
    totalWidth = 0;
    j = i;

    // read the symbols in this height class
    while (1) {

      // read the delta width
      if (huff) {
	if (!huffDecoder->decodeInt(&dw, huffDWTable)) {
	  break;
	}
      } else {
	if (!arithDecoder->decodeInt(&dw, iadwStats)) {
	  break;
	}
      }
      symWidth += dw;

      // using a collective bitmap, so don't read a bitmap here
      if (huff && !refAgg) {
	symWidths[i] = symWidth;
	totalWidth += symWidth;

      // refinement/aggregate coding
      } else if (refAgg) {
	if (huff) {
	  if (!huffDecoder->decodeInt(&refAggNum, huffAggInstTable)) {
	    break;
	  }
	} else {
	  if (!arithDecoder->decodeInt(&refAggNum, iaaiStats)) {
	    break;
	  }
	}
	if (refAggNum == 1) {
	  if (huff) {
	    symID = huffDecoder->readBits(symCodeLen);
	    huffDecoder->decodeInt(&refDX, huffTableO);
	    huffDecoder->decodeInt(&refDY, huffTableO);
	    huffDecoder->decodeInt(&bmSize, huffTableA);
	    huffDecoder->reset();
	    arithDecoder->start();
	  } else {
	    symID = arithDecoder->decodeIAID(symCodeLen, iaidStats);
	    arithDecoder->decodeInt(&refDX, iardxStats);
	    arithDecoder->decodeInt(&refDY, iardyStats);
	  }
	  refBitmap = bitmaps[symID];
	  bitmaps[numInputSyms + i] =
	      readGenericRefinementRegion(symWidth, symHeight,
					  sdrTemplate, gFalse,
					  refBitmap, refDX, refDY,
					  sdrATX, sdrATY);
	  //~ do we need to use the bmSize value here (in Huffman mode)?
	} else {
	  bitmaps[numInputSyms + i] =
	      readTextRegion(huff, gTrue, symWidth, symHeight,
			     refAggNum, 0, numInputSyms + i, NULL,
			     symCodeLen, bitmaps, 0, 0, 0, 1, 0,
			     huffTableF, huffTableH, huffTableK, huffTableO,
			     huffTableO, huffTableO, huffTableO, huffTableA,
			     sdrTemplate, sdrATX, sdrATY);
	}

      // non-ref/agg coding
      } else {
	bitmaps[numInputSyms + i] =
	    readGenericBitmap(gFalse, symWidth, symHeight,
			      sdTemplate, gFalse, gFalse, NULL,
			      sdATX, sdATY, 0);
      }

      ++i;
    }

    // read the collective bitmap
    if (huff && !refAgg) {
      huffDecoder->decodeInt(&bmSize, huffBMSizeTable);
      if (huff) {
	huffDecoder->reset();
      }
      if (bmSize == 0) {
	collBitmap = new JBIG2Bitmap(0, totalWidth, symHeight);
	bmSize = symHeight * ((totalWidth + 7) >> 3);
	p = collBitmap->getDataPtr();
	for (k = 0; k < (Guint)bmSize; ++k) {
	  *p++ = str->getChar();
	}
      } else {
	collBitmap = readGenericBitmap(gTrue, totalWidth, symHeight,
				       0, gFalse, gFalse, NULL, NULL, NULL,
				       bmSize);
      }
      x = 0;
      for (; j < i; ++j) {
	bitmaps[numInputSyms + j] =
	    collBitmap->getSlice(x, 0, symWidths[j], symHeight);
	x += symWidths[j];
      }
      delete collBitmap;
    }
  }

  // create the symbol dict object
  symbolDict = new JBIG2SymbolDict(segNum, numExSyms);

  // exported symbol list
  i = j = 0;
  ex = gFalse;
  while (i < numInputSyms + numNewSyms) {
    if (huff) {
      huffDecoder->decodeInt(&run, huffTableA);
    } else {
      arithDecoder->decodeInt(&run, iaexStats);
    }
    if (ex) {
      for (cnt = 0; cnt < run; ++cnt) {
	symbolDict->setBitmap(j++, bitmaps[i++]->copy());
      }
    } else {
      i += run;
    }
    ex = !ex;
  }
  
  for (i = 0; i < numNewSyms; ++i) {
    delete bitmaps[numInputSyms + i];
  }
  gfree(bitmaps);
  if (symWidths) {
    gfree(symWidths);
  }

  // save the arithmetic decoder stats
  if (!huff && contextRetained) {
    symbolDict->setGenericRegionStats(genericRegionStats->copy());
    if (refAgg) {
      symbolDict->setRefinementRegionStats(refinementRegionStats->copy());
    }
  }

  // store the new symbol dict
  segments->append(symbolDict);

  return;

 eofError:
  error(getPos(), "Unexpected EOF in JBIG2 stream");
}

void JBIG2Stream::readTextRegionSeg(Guint segNum, GBool imm,
				    GBool lossless, Guint length,
				    Guint *refSegs, Guint nRefSegs) {
  JBIG2Bitmap *bitmap;
  JBIG2HuffmanTable runLengthTab[36];
  JBIG2HuffmanTable *symCodeTab;
  JBIG2HuffmanTable *huffFSTable, *huffDSTable, *huffDTTable;
  JBIG2HuffmanTable *huffRDWTable, *huffRDHTable;
  JBIG2HuffmanTable *huffRDXTable, *huffRDYTable, *huffRSizeTable;
  JBIG2Segment *seg;
  GList *codeTables;
  JBIG2SymbolDict *symbolDict;
  JBIG2Bitmap **syms;
  Guint w, h, x, y, segInfoFlags, extCombOp;
  Guint flags, huff, refine, logStrips, refCorner, transposed;
  Guint combOp, defPixel, sOffset, templ;
  Guint huffFlags, huffFS, huffDS, huffDT;
  Guint huffRDW, huffRDH, huffRDX, huffRDY, huffRSize;
  Guint numInstances, numSyms, symCodeLen;
  int atx[2], aty[2];
  Guint i, k, kk;
  int j;

  // region segment info field
  if (!readULong(&w) || !readULong(&h) ||
      !readULong(&x) || !readULong(&y) ||
      !readUByte(&segInfoFlags)) {
    goto eofError;
  }
  extCombOp = segInfoFlags & 7;

  // rest of the text region header
  if (!readUWord(&flags)) {
    goto eofError;
  }
  huff = flags & 1;
  refine = (flags >> 1) & 1;
  logStrips = (flags >> 2) & 3;
  refCorner = (flags >> 4) & 3;
  transposed = (flags >> 6) & 1;
  combOp = (flags >> 7) & 3;
  defPixel = (flags >> 9) & 1;
  sOffset = (flags >> 10) & 0x1f;
  templ = (flags >> 15) & 1;
  huffFS = huffDS = huffDT = 0; // make gcc happy
  huffRDW = huffRDH = huffRDX = huffRDY = huffRSize = 0; // make gcc happy
  if (huff) {
    if (!readUWord(&huffFlags)) {
      goto eofError;
    }
    huffFS = huffFlags & 3;
    huffDS = (huffFlags >> 2) & 3;
    huffDT = (huffFlags >> 4) & 3;
    huffRDW = (huffFlags >> 6) & 3;
    huffRDH = (huffFlags >> 8) & 3;
    huffRDX = (huffFlags >> 10) & 3;
    huffRDY = (huffFlags >> 12) & 3;
    huffRSize = (huffFlags >> 14) & 1;
  }
  if (refine && templ == 0) {
    if (!readByte(&atx[0]) || !readByte(&aty[0]) ||
	!readByte(&atx[1]) || !readByte(&aty[1])) {
      goto eofError;
    }
  }
  if (!readULong(&numInstances)) {
    goto eofError;
  }

  // get symbol dictionaries and tables
  codeTables = new GList();
  numSyms = 0;
  for (i = 0; i < nRefSegs; ++i) {
    seg = findSegment(refSegs[i]);
    if (seg->getType() == jbig2SegSymbolDict) {
      numSyms += ((JBIG2SymbolDict *)seg)->getSize();
    } else if (seg->getType() == jbig2SegCodeTable) {
      codeTables->append(seg);
    }
  }
  symCodeLen = 0;
  i = 1;
  while (i < numSyms) {
    ++symCodeLen;
    i <<= 1;
  }

  // get the symbol bitmaps
  syms = (JBIG2Bitmap **)gmalloc(numSyms * sizeof(JBIG2Bitmap *));
  kk = 0;
  for (i = 0; i < nRefSegs; ++i) {
    seg = findSegment(refSegs[i]);
    if (seg->getType() == jbig2SegSymbolDict) {
      symbolDict = (JBIG2SymbolDict *)seg;
      for (k = 0; k < symbolDict->getSize(); ++k) {
	syms[kk++] = symbolDict->getBitmap(k);
      }
    }
  }