textoutputdev.cc

将pdf文档转换为高质量的html文档

		      word0->len * sizeof(Unicode))) {
	    switch (rot) {
	    case 0:
	    case 2:
	      found = fabs(word0->xMin - word2->xMin) < priDelta &&
		      fabs(word0->xMax - word2->xMax) < priDelta &&
		      fabs(word0->yMin - word2->yMin) < secDelta &&
		      fabs(word0->yMax - word2->yMax) < secDelta;
	      break;
	    case 1:
	    case 3:
	      found = fabs(word0->xMin - word2->xMin) < secDelta &&
		      fabs(word0->xMax - word2->xMax) < secDelta &&
		      fabs(word0->yMin - word2->yMin) < priDelta &&
		      fabs(word0->yMax - word2->yMax) < priDelta;
	      break;
	    }
	  }
	  if (found) {
	    break;
	  }
	}
	if (found) {
	  break;
	}
      }
      if (found) {
	if (word1) {
	  word1->next = word2->next;
	} else {
	  pool->setPool(idx1, word2->next);
	}
	delete word2;
      } else {
	word0 = word0->next;
      }
    }
  }

  // build the lines
  curLine = NULL;
  poolMinBaseIdx = pool->minBaseIdx;
  charCount = 0;
  nLines = 0;
  while (1) {

    // find the first non-empty line in the pool
    for (;
	 poolMinBaseIdx <= pool->maxBaseIdx && !pool->getPool(poolMinBaseIdx);
	 ++poolMinBaseIdx) ;
    if (poolMinBaseIdx > pool->maxBaseIdx) {
      break;
    }

    // look for the left-most word in the first four lines of the
    // pool -- this avoids starting with a superscript word
    startBaseIdx = poolMinBaseIdx;
    for (baseIdx = poolMinBaseIdx + 1;
	 baseIdx < poolMinBaseIdx + 4 && baseIdx <= pool->maxBaseIdx;
	 ++baseIdx) {
      if (!pool->getPool(baseIdx)) {
	continue;
      }
      if (pool->getPool(baseIdx)->primaryCmp(pool->getPool(startBaseIdx))
	  < 0) {
	startBaseIdx = baseIdx;
      }
    }

    // create a new line
    word0 = pool->getPool(startBaseIdx);
    pool->setPool(startBaseIdx, word0->next);
    word0->next = NULL;
    line = new TextLine(this, word0->rot, word0->base);
    line->addWord(word0);
    lastWord = word0;

    // compute the search range
    fontSize = word0->fontSize;
    minBase = word0->base - maxIntraLineDelta * fontSize;
    maxBase = word0->base + maxIntraLineDelta * fontSize;
    minBaseIdx = pool->getBaseIdx(minBase);
    maxBaseIdx = pool->getBaseIdx(maxBase);

    // find the rest of the words in this line
    while (1) {

      // find the left-most word whose baseline is in the range for
      // this line
      bestWordBaseIdx = 0;
      bestWord0 = bestWord1 = NULL;
      for (baseIdx = minBaseIdx; baseIdx <= maxBaseIdx; ++baseIdx) {
	for (word0 = NULL, word1 = pool->getPool(baseIdx);
	     word1;
	     word0 = word1, word1 = word1->next) {
	  if (word1->base >= minBase &&
	      word1->base <= maxBase &&
	      (delta = lastWord->primaryDelta(word1)) >=
	        minCharSpacing * fontSize) {
	    if (delta < maxWordSpacing * fontSize &&
		(!bestWord1 || word1->primaryCmp(bestWord1) < 0)) {
	      bestWordBaseIdx = baseIdx;
	      bestWord0 = word0;
	      bestWord1 = word1;
	    }
	    break;
	  }
	}
      }
      if (!bestWord1) {
	break;
      }

      // remove it from the pool, and add it to the line
      if (bestWord0) {
	bestWord0->next = bestWord1->next;
      } else {
	pool->setPool(bestWordBaseIdx, bestWord1->next);
      }
      bestWord1->next = NULL;
      line->addWord(bestWord1);
      lastWord = bestWord1;
    }

    // add the line
    if (curLine && line->cmpYX(curLine) > 0) {
      line0 = curLine;
      line1 = curLine->next;
    } else {
      line0 = NULL;
      line1 = lines;
    }
    for (;
	 line1 && line->cmpYX(line1) > 0;
	 line0 = line1, line1 = line1->next) ;
    if (line0) {
      line0->next = line;
    } else {
      lines = line;
    }
    line->next = line1;
    curLine = line;
    line->coalesce(uMap);
    charCount += line->len;
    ++nLines;
  }

  // sort lines into xy order for column assignment
  lineArray = (TextLine **)gmallocn(nLines, sizeof(TextLine *));
  for (line = lines, i = 0; line; line = line->next, ++i) {
    lineArray[i] = line;
  }
  qsort(lineArray, nLines, sizeof(TextLine *), &TextLine::cmpXY);

  // column assignment
  nColumns = 0;
  for (i = 0; i < nLines; ++i) {
    line0 = lineArray[i];
    col1 = 0;
    for (j = 0; j < i; ++j) {
      line1 = lineArray[j];
      if (line1->primaryDelta(line0) >= 0) {
	col2 = line1->col[line1->len] + 1;
      } else {
	k = 0; // make gcc happy
	switch (rot) {
	case 0:
	  for (k = 0;
	       k < line1->len &&
		 line0->xMin >= 0.5 * (line1->edge[k] + line1->edge[k+1]);
	       ++k) ;
	  break;
	case 1:
	  for (k = 0;
	       k < line1->len &&
		 line0->yMin >= 0.5 * (line1->edge[k] + line1->edge[k+1]);
	       ++k) ;
	  break;
	case 2:
	  for (k = 0;
	       k < line1->len &&
		 line0->xMax <= 0.5 * (line1->edge[k] + line1->edge[k+1]);
	       ++k) ;
	  break;
	case 3:
	  for (k = 0;
	       k < line1->len &&
		 line0->yMax <= 0.5 * (line1->edge[k] + line1->edge[k+1]);
	       ++k) ;
	  break;
	}
	col2 = line1->col[k];
      }
      if (col2 > col1) {
	col1 = col2;
      }
    }
    for (k = 0; k <= line0->len; ++k) {
      line0->col[k] += col1;
    }
    if (line0->col[line0->len] > nColumns) {
      nColumns = line0->col[line0->len];
    }
  }
  gfree(lineArray);
}

void TextBlock::updatePriMinMax(TextBlock *blk) {
  double newPriMin, newPriMax;
  GBool gotPriMin, gotPriMax;

  gotPriMin = gotPriMax = gFalse;
  newPriMin = newPriMax = 0; // make gcc happy
  switch (page->primaryRot) {
  case 0:
  case 2:
    if (blk->yMin < yMax && blk->yMax > yMin) {
      if (blk->xMin < xMin) {
	newPriMin = blk->xMax;
	gotPriMin = gTrue;
      }
      if (blk->xMax > xMax) {
	newPriMax = blk->xMin;
	gotPriMax = gTrue;
      }
    }
    break;
  case 1:
  case 3:
    if (blk->xMin < xMax && blk->xMax > xMin) {
      if (blk->yMin < yMin) {
	newPriMin = blk->yMax;
	gotPriMin = gTrue;
      }
      if (blk->yMax > yMax) {
	newPriMax = blk->yMin;
	gotPriMax = gTrue;
      }
    }
    break;
  }
  if (gotPriMin) {
    if (newPriMin > xMin) {
      newPriMin = xMin;
    }
    if (newPriMin > priMin) {
      priMin = newPriMin;
    }
  }
  if (gotPriMax) {
    if (newPriMax < xMax) {
      newPriMax = xMax;
    }
    if (newPriMax < priMax) {
      priMax = newPriMax;
    }
  }
}

int TextBlock::cmpXYPrimaryRot(const void *p1, const void *p2) {
  TextBlock *blk1 = *(TextBlock **)p1;
  TextBlock *blk2 = *(TextBlock **)p2;
  double cmp;

  cmp = 0; // make gcc happy
  switch (blk1->page->primaryRot) {
  case 0:
    if ((cmp = blk1->xMin - blk2->xMin) == 0) {
      cmp = blk1->yMin - blk2->yMin;
    }
    break;
  case 1:
    if ((cmp = blk1->yMin - blk2->yMin) == 0) {
      cmp = blk2->xMax - blk1->xMax;
    }
    break;
  case 2:
    if ((cmp = blk2->xMax - blk1->xMax) == 0) {
      cmp = blk2->yMin - blk1->yMin;
    }
    break;
  case 3:
    if ((cmp = blk2->yMax - blk1->yMax) == 0) {
      cmp = blk1->xMax - blk2->xMax;
    }
    break;
  }
  return cmp < 0 ? -1 : cmp > 0 ? 1 : 0;
}

int TextBlock::cmpYXPrimaryRot(const void *p1, const void *p2) {
  TextBlock *blk1 = *(TextBlock **)p1;
  TextBlock *blk2 = *(TextBlock **)p2;
  double cmp;

  cmp = 0; // make gcc happy
  switch (blk1->page->primaryRot) {
  case 0:
    if ((cmp = blk1->yMin - blk2->yMin) == 0) {
      cmp = blk1->xMin - blk2->xMin;
    }
    break;
  case 1:
    if ((cmp = blk2->xMax - blk1->xMax) == 0) {
      cmp = blk1->yMin - blk2->yMin;
    }
    break;
  case 2:
    if ((cmp = blk2->yMin - blk1->yMin) == 0) {
      cmp = blk2->xMax - blk1->xMax;
    }
    break;
  case 3:
    if ((cmp = blk1->xMax - blk2->xMax) == 0) {
      cmp = blk2->yMax - blk1->yMax;
    }
    break;
  }
  return cmp < 0 ? -1 : cmp > 0 ? 1 : 0;
}

int TextBlock::primaryCmp(TextBlock *blk) {
  double cmp;

  cmp = 0; // make gcc happy
  switch (rot) {
  case 0:
    cmp = xMin - blk->xMin;
    break;
  case 1:
    cmp = yMin - blk->yMin;
    break;
  case 2:
    cmp = blk->xMax - xMax;
    break;
  case 3:
    cmp = blk->yMax - yMax;
    break;
  }
  return cmp < 0 ? -1 : cmp > 0 ? 1 : 0;
}

double TextBlock::secondaryDelta(TextBlock *blk) {
  double delta;

  delta = 0; // make gcc happy
  switch (rot) {
  case 0:
    delta = blk->yMin - yMax;
    break;
  case 1:
    delta = xMin - blk->xMax;
    break;
  case 2:
    delta = yMin - blk->yMax;
    break;
  case 3:
    delta = blk->xMin - xMax;
    break;
  }
  return delta;
}

GBool TextBlock::isBelow(TextBlock *blk) {
  GBool below;

  below = gFalse; // make gcc happy
  switch (page->primaryRot) {
  case 0:
    below = xMin >= blk->priMin && xMax <= blk->priMax &&
            yMin > blk->yMin;
    break;
  case 1:
    below = yMin >= blk->priMin && yMax <= blk->priMax &&
            xMax < blk->xMax;
    break;
  case 2:
    below = xMin >= blk->priMin && xMax <= blk->priMax &&
            yMax < blk->yMax;
    break;
  case 3:
    below = yMin >= blk->priMin && yMax <= blk->priMax &&
            xMin > blk->xMin;
    break;
  }

  return below;
}

//------------------------------------------------------------------------
// TextFlow
//------------------------------------------------------------------------

TextFlow::TextFlow(TextPage *pageA, TextBlock *blk) {
  page = pageA;
  xMin = blk->xMin;
  xMax = blk->xMax;
  yMin = blk->yMin;
  yMax = blk->yMax;
  priMin = blk->priMin;
  priMax = blk->priMax;
  blocks = lastBlk = blk;
  next = NULL;
}

TextFlow::~TextFlow() {
  TextBlock *blk;

  while (blocks) {
    blk = blocks;
    blocks = blocks->next;
    delete blk;
  }
}

void TextFlow::addBlock(TextBlock *blk) {
  if (lastBlk) {
    lastBlk->next = blk;
  } else {
    blocks = blk;
  }
  lastBlk = blk;
  if (blk->xMin < xMin) {
    xMin = blk->xMin;
  }
  if (blk->xMax > xMax) {
    xMax = blk->xMax;
  }
  if (blk->yMin < yMin) {
    yMin = blk->yMin;
  }
  if (blk->yMax > yMax) {
    yMax = blk->yMax;
  }
}

GBool TextFlow::blockFits(TextBlock *blk, TextBlock *prevBlk) {
  GBool fits;

  // lower blocks must use smaller fonts
  if (blk->lines->words->fontSize > lastBlk->lines->words->fontSize) {
    return gFalse;
  }

  fits = gFalse; // make gcc happy
  switch (page->primaryRot) {
  case 0:
    fits = blk->xMin >= priMin && blk->xMax <= priMax;
    break;
  case 1:
    fits = blk->yMin >= priMin && blk->yMax <= priMax;
    break;
  case 2:
    fits = blk->xMin >= priMin && blk->xMax <= priMax;
    break;
  case 3:
    fits = blk->yMin >= priMin && blk->yMax <= priMax;
    break;
  }
  return fits;
}

#if TEXTOUT_WORD_LIST

//------------------------------------------------------------------------
// TextWordList
//------------------------------------------------------------------------

TextWordList::TextWordList(TextPage *text, GBool physLayout) {
  TextFlow *flow;
  TextBlock *blk;
  TextLine *line;
  TextWord *word;
  TextWord **wordArray;
  int nWords, i;

  words = new GList();

  if (text->rawOrder) {
    for (word = text->rawWords; word; word = word->next) {
      words->append(word);
    }

  } else if (physLayout) {
    // this is inefficient, but it's also the least useful of these
    // three cases
    nWords = 0;
    for (flow = text->flows; flow; flow = flow->next) {
      for (blk = flow->blocks; blk; blk = blk->next) {
	for (line = blk->lines; line; line = line->next) {
	  for (word = line->words; word; word = word->next) {
	    ++nWords;
	  }
	}
      }
    }
    wordArray = (TextWord **)gmallocn(nWords, sizeof(TextWord *));
    i = 0;
    for (flow = text->flows; flow; flow = flow->next) {
      for (blk = flow->blocks; blk; blk = blk->next) {
	for (line = blk->lines; line; line = line->next) {
	  for (word = line->words; word; word = word->next) {
	    wordArray[i++] = word;
	  }
	}
      }
    }
    qsort(wordArray, nWords, sizeof(TextWord *), &TextWord::cmpYX);
    for (i = 0; i < nWords; ++i) {
      words->append(wordArray[i]);
    }
    gfree(wordArray);

  } else {
    for (flow = text->flows; flow; flow = flow->next) {
      for (blk = flow->blocks; blk; blk = blk->next) {
	for (line = blk->lines; line; line = line->next) {
	  for (word = line->words; word; word = word->next) {
	    words->append(word);
	  }
	}
      }