textoutputdev.cc

将pdf文档转换为高质量的html文档

    if (word->xMin < xMin) {
      xMin = word->xMin;
    }
    if (word->xMax > xMax) {
      xMax = word->xMax;
    }
    if (word->yMin < yMin) {
      yMin = word->yMin;
    }
    if (word->yMax > yMax) {
      yMax = word->yMax;
    }
  }
}

double TextLine::primaryDelta(TextLine *line) {
  double delta;

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

int TextLine::primaryCmp(TextLine *line) {
  double cmp;

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

int TextLine::secondaryCmp(TextLine *line) {
  double cmp;

  cmp = (rot == 0 || rot == 3) ? base - line->base : line->base - base;
  return cmp < 0 ? -1 : cmp > 0 ? 1 : 0;
}

int TextLine::cmpYX(TextLine *line) {
  int cmp;

  if ((cmp = secondaryCmp(line))) {
    return cmp;
  }
  return primaryCmp(line);
}

int TextLine::cmpXY(const void *p1, const void *p2) {
  TextLine *line1 = *(TextLine **)p1;
  TextLine *line2 = *(TextLine **)p2;
  int cmp;

  if ((cmp = line1->primaryCmp(line2))) {
    return cmp;
  }
  return line1->secondaryCmp(line2);
}

void TextLine::coalesce(UnicodeMap *uMap) {
  TextWord *word0, *word1;
  double space, delta, minSpace;
  GBool isUnicode;
  char buf[8];
  int i, j;

  if (words->next) {

    // compute the inter-word space threshold
    if (words->len > 1 || words->next->len > 1) {
      minSpace = 0;
    } else {
      minSpace = words->primaryDelta(words->next);
      for (word0 = words->next, word1 = word0->next;
	   word1 && minSpace > 0;
	   word0 = word1, word1 = word0->next) {
	if (word1->len > 1) {
	  minSpace = 0;
	}
	delta = word0->primaryDelta(word1);
	if (delta < minSpace) {
	  minSpace = delta;
	}
      }
    }
    if (minSpace <= 0) {
      space = maxCharSpacing * words->fontSize;
    } else {
      space = maxWideCharSpacingMul * minSpace;
    }

    // merge words
    word0 = words;
    word1 = words->next;
    while (word1) {
      if (word0->primaryDelta(word1) >= space) {
	word0->spaceAfter = gTrue;
	word0 = word1;
	word1 = word1->next;
      } else if (word0->font == word1->font &&
		 fabs(word0->fontSize - word1->fontSize) <
		 maxWordFontSizeDelta * words->fontSize &&
		 word1->charPos == word0->charPos + word0->charLen) {
	word0->merge(word1);
	word0->next = word1->next;
	delete word1;
	word1 = word0->next;
      } else {
	word0 = word1;
	word1 = word1->next;
      }
    }
  }

  // build the line text
  isUnicode = uMap ? uMap->isUnicode() : gFalse;
  len = 0;
  for (word1 = words; word1; word1 = word1->next) {
    len += word1->len;
    if (word1->spaceAfter) {
      ++len;
    }
  }
  text = (Unicode *)gmallocn(len, sizeof(Unicode));
  edge = (double *)gmallocn(len + 1, sizeof(double));
  i = 0;
  for (word1 = words; word1; word1 = word1->next) {
    for (j = 0; j < word1->len; ++j) {
      text[i] = word1->text[j];
      edge[i] = word1->edge[j];
      ++i;
    }
    edge[i] = word1->edge[word1->len];
    if (word1->spaceAfter) {
      text[i] = (Unicode)0x0020;
      ++i;
    }
  }

  // compute convertedLen and set up the col array
  col = (int *)gmallocn(len + 1, sizeof(int));
  convertedLen = 0;
  for (i = 0; i < len; ++i) {
    col[i] = convertedLen;
    if (isUnicode) {
      ++convertedLen;
    } else if (uMap) {
      convertedLen += uMap->mapUnicode(text[i], buf, sizeof(buf));
    }
  }
  col[len] = convertedLen;

  // check for hyphen at end of line
  //~ need to check for other chars used as hyphens
  hyphenated = text[len - 1] == (Unicode)'-';
}

//------------------------------------------------------------------------
// TextLineFrag
//------------------------------------------------------------------------

class TextLineFrag {
public:

  TextLine *line;		// the line object
  int start, len;		// offset and length of this fragment
				//   (in Unicode chars)
  double xMin, xMax;		// bounding box coordinates
  double yMin, yMax;
  double base;			// baseline virtual coordinate
  int col;			// first column

  void init(TextLine *lineA, int startA, int lenA);
  void computeCoords(GBool oneRot);

  static int cmpYXPrimaryRot(const void *p1, const void *p2);
  static int cmpYXLineRot(const void *p1, const void *p2);
  static int cmpXYLineRot(const void *p1, const void *p2);
};

void TextLineFrag::init(TextLine *lineA, int startA, int lenA) {
  line = lineA;
  start = startA;
  len = lenA;
  col = line->col[start];
}

void TextLineFrag::computeCoords(GBool oneRot) {
  TextBlock *blk;
  double d0, d1, d2, d3, d4;

  if (oneRot) {

    switch (line->rot) {
    case 0:
      xMin = line->edge[start];
      xMax = line->edge[start + len];
      yMin = line->yMin;
      yMax = line->yMax;
      break;
    case 1:
      xMin = line->xMin;
      xMax = line->xMax;
      yMin = line->edge[start];
      yMax = line->edge[start + len];
      break;
    case 2:
      xMin = line->edge[start + len];
      xMax = line->edge[start];
      yMin = line->yMin;
      yMax = line->yMax;
      break;
    case 3:
      xMin = line->xMin;
      xMax = line->xMax;
      yMin = line->edge[start + len];
      yMax = line->edge[start];
      break;
    }
    base = line->base;

  } else {

    if (line->rot == 0 && line->blk->page->primaryRot == 0) {

      xMin = line->edge[start];
      xMax = line->edge[start + len];
      yMin = line->yMin;
      yMax = line->yMax;
      base = line->base;

    } else {

      blk = line->blk;
      d0 = line->edge[start];
      d1 = line->edge[start + len];
      d2 = d3 = d4 = 0; // make gcc happy

      switch (line->rot) {
      case 0:
	d2 = line->yMin;
	d3 = line->yMax;
	d4 = line->base;
	d0 = (d0 - blk->xMin) / (blk->xMax - blk->xMin);
	d1 = (d1 - blk->xMin) / (blk->xMax - blk->xMin);
	d2 = (d2 - blk->yMin) / (blk->yMax - blk->yMin);
	d3 = (d3 - blk->yMin) / (blk->yMax - blk->yMin);
	d4 = (d4 - blk->yMin) / (blk->yMax - blk->yMin);
	break;
      case 1:
	d2 = line->xMax;
	d3 = line->xMin;
	d4 = line->base;
	d0 = (d0 - blk->yMin) / (blk->yMax - blk->yMin);
	d1 = (d1 - blk->yMin) / (blk->yMax - blk->yMin);
	d2 = (blk->xMax - d2) / (blk->xMax - blk->xMin);
	d3 = (blk->xMax - d3) / (blk->xMax - blk->xMin);
	d4 = (blk->xMax - d4) / (blk->xMax - blk->xMin);
	break;
      case 2:
	d2 = line->yMax;
	d3 = line->yMin;
	d4 = line->base;
	d0 = (blk->xMax - d0) / (blk->xMax - blk->xMin);
	d1 = (blk->xMax - d1) / (blk->xMax - blk->xMin);
	d2 = (blk->yMax - d2) / (blk->yMax - blk->yMin);
	d3 = (blk->yMax - d3) / (blk->yMax - blk->yMin);
	d4 = (blk->yMax - d4) / (blk->yMax - blk->yMin);
	break;
      case 3:
	d2 = line->xMin;
	d3 = line->xMax;
	d4 = line->base;
	d0 = (blk->yMax - d0) / (blk->yMax - blk->yMin);
	d1 = (blk->yMax - d1) / (blk->yMax - blk->yMin);
	d2 = (d2 - blk->xMin) / (blk->xMax - blk->xMin);
	d3 = (d3 - blk->xMin) / (blk->xMax - blk->xMin);
	d4 = (d4 - blk->xMin) / (blk->xMax - blk->xMin);
	break;
      }

      switch (line->blk->page->primaryRot) {
      case 0:
	xMin = blk->xMin + d0 * (blk->xMax - blk->xMin);
	xMax = blk->xMin + d1 * (blk->xMax - blk->xMin);
	yMin = blk->yMin + d2 * (blk->yMax - blk->yMin);
	yMax = blk->yMin + d3 * (blk->yMax - blk->yMin);
	base = blk->yMin + base * (blk->yMax - blk->yMin);
	break;
      case 1:
	xMin = blk->xMax - d3 * (blk->xMax - blk->xMin);
	xMax = blk->xMax - d2 * (blk->xMax - blk->xMin);
	yMin = blk->yMin + d0 * (blk->yMax - blk->yMin);
	yMax = blk->yMin + d1 * (blk->yMax - blk->yMin);
	base = blk->xMax - d4 * (blk->xMax - blk->xMin);
	break;
      case 2:
	xMin = blk->xMax - d1 * (blk->xMax - blk->xMin);
	xMax = blk->xMax - d0 * (blk->xMax - blk->xMin);
	yMin = blk->yMax - d3 * (blk->yMax - blk->yMin);
	yMax = blk->yMax - d2 * (blk->yMax - blk->yMin);
	base = blk->yMax - d4 * (blk->yMax - blk->yMin);
	break;
      case 3:
	xMin = blk->xMin + d2 * (blk->xMax - blk->xMin);
	xMax = blk->xMin + d3 * (blk->xMax - blk->xMin);
	yMin = blk->yMax - d1 * (blk->yMax - blk->yMin);
	yMax = blk->yMax - d0 * (blk->yMax - blk->yMin);
	base = blk->xMin + d4 * (blk->xMax - blk->xMin);
	break;
      }

    }
  }
}

int TextLineFrag::cmpYXPrimaryRot(const void *p1, const void *p2) {
  TextLineFrag *frag1 = (TextLineFrag *)p1;
  TextLineFrag *frag2 = (TextLineFrag *)p2;
  double cmp;

  cmp = 0; // make gcc happy
  switch (frag1->line->blk->page->primaryRot) {
  case 0:
    if (fabs(cmp = frag1->yMin - frag2->yMin) < 0.01) {
      cmp = frag1->xMin - frag2->xMin;
    }
    break;
  case 1:
    if (fabs(cmp = frag2->xMax - frag1->xMax) < 0.01) {
      cmp = frag1->yMin - frag2->yMin;
    }
    break;
  case 2:
    if (fabs(cmp = frag2->yMin - frag1->yMin) < 0.01) {
      cmp = frag2->xMax - frag1->xMax;
    }
    break;
  case 3:
    if (fabs(cmp = frag1->xMax - frag2->xMax) < 0.01) {
      cmp = frag2->yMax - frag1->yMax;
    }
    break;
  }
  return cmp < 0 ? -1 : cmp > 0 ? 1 : 0;
}

int TextLineFrag::cmpYXLineRot(const void *p1, const void *p2) {
  TextLineFrag *frag1 = (TextLineFrag *)p1;
  TextLineFrag *frag2 = (TextLineFrag *)p2;
  double cmp;

  cmp = 0; // make gcc happy
  switch (frag1->line->rot) {
  case 0:
    if ((cmp = frag1->yMin - frag2->yMin) == 0) {
      cmp = frag1->xMin - frag2->xMin;
    }
    break;
  case 1:
    if ((cmp = frag2->xMax - frag1->xMax) == 0) {
      cmp = frag1->yMin - frag2->yMin;
    }
    break;
  case 2:
    if ((cmp = frag2->yMin - frag1->yMin) == 0) {
      cmp = frag2->xMax - frag1->xMax;
    }
    break;
  case 3:
    if ((cmp = frag1->xMax - frag2->xMax) == 0) {
      cmp = frag2->yMax - frag1->yMax;
    }
    break;
  }
  return cmp < 0 ? -1 : cmp > 0 ? 1 : 0;
}

int TextLineFrag::cmpXYLineRot(const void *p1, const void *p2) {
  TextLineFrag *frag1 = (TextLineFrag *)p1;
  TextLineFrag *frag2 = (TextLineFrag *)p2;
  double cmp;

  cmp = 0; // make gcc happy
  switch (frag1->line->rot) {
  case 0:
    if ((cmp = frag1->xMin - frag2->xMin) == 0) {
      cmp = frag1->yMin - frag2->yMin;
    }
    break;
  case 1:
    if ((cmp = frag1->yMin - frag2->yMin) == 0) {
      cmp = frag2->xMax - frag1->xMax;
    }
    break;
  case 2:
    if ((cmp = frag2->xMax - frag1->xMax) == 0) {
      cmp = frag2->yMin - frag1->yMin;
    }
    break;
  case 3:
    if ((cmp = frag2->yMax - frag1->yMax) == 0) {
      cmp = frag1->xMax - frag2->xMax;
    }
    break;
  }
  return cmp < 0 ? -1 : cmp > 0 ? 1 : 0;
}

//------------------------------------------------------------------------
// TextBlock
//------------------------------------------------------------------------

TextBlock::TextBlock(TextPage *pageA, int rotA) {
  page = pageA;
  rot = rotA;
  xMin = yMin = 0;
  xMax = yMax = -1;
  priMin = 0;
  priMax = page->pageWidth;
  pool = new TextPool();
  lines = NULL;
  curLine = NULL;
  next = NULL;
  stackNext = NULL;
}

TextBlock::~TextBlock() {
  TextLine *line;

  delete pool;
  while (lines) {
    line = lines;
    lines = lines->next;
    delete line;
  }
}

void TextBlock::addWord(TextWord *word) {
  pool->addWord(word);
  if (xMin > xMax) {
    xMin = word->xMin;
    xMax = word->xMax;
    yMin = word->yMin;
    yMax = word->yMax;
  } else {
    if (word->xMin < xMin) {
      xMin = word->xMin;
    }
    if (word->xMax > xMax) {
      xMax = word->xMax;
    }
    if (word->yMin < yMin) {
      yMin = word->yMin;
    }
    if (word->yMax > yMax) {
      yMax = word->yMax;
    }
  }
}

void TextBlock::coalesce(UnicodeMap *uMap) {
  TextWord *word0, *word1, *word2, *bestWord0, *bestWord1, *lastWord;
  TextLine *line, *line0, *line1;
  int poolMinBaseIdx, startBaseIdx, minBaseIdx, maxBaseIdx;
  int baseIdx, bestWordBaseIdx, idx0, idx1;
  double minBase, maxBase;
  double fontSize, delta, priDelta, secDelta;
  TextLine **lineArray;
  GBool found;
  int col1, col2;
  int i, j, k;

  // discard duplicated text (fake boldface, drop shadows)
  for (idx0 = pool->minBaseIdx; idx0 <= pool->maxBaseIdx; ++idx0) {
    word0 = pool->getPool(idx0);
    while (word0) {
      priDelta = dupMaxPriDelta * word0->fontSize;
      secDelta = dupMaxSecDelta * word0->fontSize;
      if (rot == 0 || rot == 3) {
	maxBaseIdx = pool->getBaseIdx(word0->base + secDelta);
      } else {
	maxBaseIdx = pool->getBaseIdx(word0->base - secDelta);
      }
      found = gFalse;
      word1 = word2 = NULL; // make gcc happy
      for (idx1 = idx0; idx1 <= maxBaseIdx; ++idx1) {
	if (idx1 == idx0) {
	  word1 = word0;
	  word2 = word0->next;
	} else {
	  word1 = NULL;
	  word2 = pool->getPool(idx1);
	}
	for (; word2; word1 = word2, word2 = word2->next) {
	  if (word2->len == word0->len &&
	      !memcmp(word2->text, word0->text,