[Poppler-bugs] [Bug 30175] New: Bug in void TextPage::visitSelection(TextSelectionVisitor *visitor, PDFRectangle *selection, SelectionStyle style)

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=30175

           Summary: Bug in void
                    TextPage::visitSelection(TextSelectionVisitor
                    *visitor,  PDFRectangle *selection, SelectionStyle
                    style)
           Product: poppler
           Version: unspecified
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: high
         Component: general
        AssignedTo: poppler-bugs at lists.freedesktop.org
        ReportedBy: srinivas.adicherla at gmail.com
                CC: srinivas.adicherla at gmail.com


Created an attachment (id=38683)
 --> (https://bugs.freedesktop.org/attachment.cgi?id=38683)
pdf file

Hi,

    I tried to use the function poppler_page_get_selection_region(), with
passing rectangle as page boundary, style as POPPLER_SELECTION_LINE. But for
some pages its not returning all lines rectangles. So, I started debugging the
code. I found that there is a problem in the function
TextPage::visitSelection() in the TextOutputDev.cc. 


 In the function after finding the best_flow we are looping again to
visitSelection on each block with in the flow. But by assigning the initial
flow in the for loop as (flow=best_flow[start]). This is not looping through
all the flows, for the given boundary. Ofcourse I gave page boundary. But if I
do with initial flow as (flow = flows); its looping through all the flows. But
if I do like this what is the use of finding best_flow. Any suggestions? And I
didn't get why we need to calculate bestflow although we are looping through
all the flows. ( I think may be in case if we don't give the pageboundary as
full page rect) please calrify me?

I attached the pdf file for which the problem is coming. Its giving problem in
8th page.

I pasted the function below. 

 for (flow = best_flow[start]; flow; flow = flow->next) { /* original */
  //for (flow = flows; flow; flow = flow->next) { /* if we make this change its
working*/


TextOutputDev.cc
----------------
void TextPage::visitSelection(TextSelectionVisitor *visitor,
                  PDFRectangle *selection,
                  SelectionStyle style)
{
  PDFRectangle child_selection;
  double x[2], y[2], d, best_d[2];
  double xMin, yMin, xMax, yMax;
  TextFlow *flow, *best_flow[2];
  TextBlock *blk, *best_block[2];
  int i, count = 0, best_count[2], start, stop;
  if (!flows)
    return;

  x[0] = selection->x1;
  y[0] = selection->y1;
  x[1] = selection->x2;
  y[1] = selection->y2;

  xMin = pageWidth;
  yMin = pageHeight;
  xMax = 0.0;
  yMax = 0.0;

  for (i = 0; i < 2; i++) {
    best_block[i] = NULL;
    best_flow[i] = NULL;
    best_count[i] = 0;
    best_d[i] = 0;
  }
  // find the nearest blocks to the selection points
  // using the manhattan distance.
  for (flow = flows; flow; flow = flow->next) {
    for (blk = flow->blocks; blk; blk = blk->next) {
      count++;
      // the first/last blocks in reading order are
      // often not the closest to the page corners;
      // track the corners, force those blocks to
      // be selected if the selection runs across
      // multiple pages.
      xMin = fmin(xMin, blk->xMin);
      yMin = fmin(yMin, blk->yMin);
      xMax = fmax(xMax, blk->xMax);
      yMax = fmax(yMax, blk->yMax);
      for (i = 0; i < 2; i++) {
    d = fmax(blk->xMin - x[i], 0.0) +
      fmax(x[i] - blk->xMax, 0.0) +
      fmax(blk->yMin - y[i], 0.0) +
      fmax(y[i] - blk->yMax, 0.0);
    if (!best_block[i] ||
        d < best_d[i] ||
        (!blk->next && !flow->next &&
         x[i] > xMax && y[i] > yMax)) {
      best_block[i] = blk;
      best_flow[i] = flow;
      best_count[i] = count;
      best_d[i] = d;
    }
      }
    }
  }
  for (i = 0; i < 2; i++) {
    if (primaryLR) {
      if (x[i] < xMin && y[i] < yMin) {
    best_block[i] = flows->blocks;
    best_flow[i] = flows;
    best_count[i] = 1;
      }
    } else {
      if (x[i] > xMax && y[i] < yMin) {
    best_block[i] = flows->blocks;
    best_flow[i] = flows;
    best_count[i] = 1;
      }
    }
  }
  // assert: best is always set.
  if (!best_block[0] || !best_block[1]) {
    return;
  }

  // Now decide which point was first.
  if (best_count[0] < best_count[1] ||
      (best_count[0] == best_count[1] && y[0] < y[1])) {
    start = 0;
    stop = 1;
  } else {
    start = 1;
    stop = 0;
  }

  for (flow = best_flow[start]; flow; flow = flow->next) {
  //for (flow = flows; flow; flow = flow->next) {
   if (flow == best_flow[start]) {
      blk = best_block[start];
    } else {
      blk = flow->blocks;
    }
    for (; blk; blk = blk->next) {
      if (primaryLR) {
    child_selection.x1 = blk->xMin;
    child_selection.x2 = blk->xMax;
      } else {
    child_selection.x1 = blk->xMax;
    child_selection.x2 = blk->xMin;
      }
      child_selection.y1 = blk->yMin;
      child_selection.y2 = blk->yMax;
      if (blk == best_block[start]) {
    child_selection.x1 = fmax(blk->xMin, fmin(blk->xMax, x[start]));
    child_selection.y1 = fmax(blk->yMin, fmin(blk->yMax, y[start]));
      }
      if (blk == best_block[stop]) {
    child_selection.x2 = fmax(blk->xMin, fmin(blk->xMax, x[stop]));
    child_selection.y2 = fmax(blk->yMin, fmin(blk->yMax, y[stop]));
    blk->visitSelection(visitor, &child_selection, style);
    return;
      }
      blk->visitSelection(visitor, &child_selection, style);
    }
  }

}

-- 
Configure bugmail: https://bugs.freedesktop.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.