Issue #24: Text selection and table selection bug

Pdf4QtLib/sources/pdftextlayout.cpp

@@ -314,7 +314,7 @@ bool PDFTextLayout::isHoveringOverTextBlock(const QPointF& point) const
     return false;
 }
 
-PDFTextSelection PDFTextLayout::createTextSelection(PDFInteger pageIndex, const QPointF& point1, const QPointF& point2, QColor selectionColor)
+PDFTextSelection PDFTextLayout::createTextSelection(PDFInteger pageIndex, const QPointF& point1, const QPointF& point2, QColor selectionColor, bool strictSelection)
 {
     PDFTextSelection selection;
 
@@ -374,12 +374,12 @@ PDFTextSelection PDFTextLayout::createTextSelection(PDFInteger pageIndex, const
                 QRectF boundingBoxPathBBRect = boundingBoxPath.controlPointRect();
 
                 // If start point is above the text block, move start point to the left.
-                if (boundingBoxPathBBRect.bottom() < pointA.y())
+                if (!strictSelection && boundingBoxPathBBRect.bottom() < pointA.y())
                 {
                     pointA.setX(boundingBoxPathBBRect.left());
                     isTopPointAboveText = true;
                 }
-                if (boundingBoxPathBBRect.top() > pointB.y())
+                if (!strictSelection && boundingBoxPathBBRect.top() > pointB.y())
                 {
                     pointB.setX(boundingBoxPathBBRect.right());
                     isBottomPointBelowText = true;
@@ -399,6 +399,16 @@ PDFTextSelection PDFTextLayout::createTextSelection(PDFInteger pageIndex, const
             for (size_t lineId = 0, linesCount = lines.size(); lineId < linesCount; ++lineId)
             {
                 const PDFTextLine& line = lines[lineId];
+                QRectF lineBoundingRect = line.getBoundingBox().boundingRect();
+
+                // We skip lines, which are not in the range (pointB.y(), pointA.y),
+                // i.e. are above or below.
+                if (lineBoundingRect.bottom() < pointB.y() ||
+                    lineBoundingRect.top() > pointA.y())
+                {
+                    continue;
+                }
+
                 const TextCharacters& characters = line.getCharacters();
                 for (size_t characterId = 0, characterCount = characters.size(); characterId < characterCount; ++characterId)
                 {

Pdf4QtLib/sources/pdftextlayout.h

@@ -368,7 +368,12 @@ public:
     /// \param point1 First point
     /// \param point2 Second point
     /// \param selectionColor Selection color
-    PDFTextSelection createTextSelection(PDFInteger pageIndex, const QPointF& point1, const QPointF& point2, QColor selectionColor = Qt::yellow);
+    /// \param strictSelection If true, does not adjust horizontal range when above/below text block
+    PDFTextSelection createTextSelection(PDFInteger pageIndex,
+                                         const QPointF& point1,
+                                         const QPointF& point2,
+                                         QColor selectionColor = Qt::yellow,
+                                         bool strictSelection = false);
 
     /// Returns string from text selection
     /// \param itBegin Iterator (begin range)

Pdf4QtLib/sources/pdfwidgettool.cpp

@@ -1567,46 +1567,15 @@ void PDFSelectTableTool::keyPressEvent(QWidget* widget, QKeyEvent* event)
                 cell.row = rowIndex;
                 cell.column = columnIndex;
                 cell.rectangle = QRectF(left, top, width, height);
+
+                PDFTextSelection textSelection = m_textLayout.createTextSelection(m_pageIndex, cell.rectangle.topLeft(), cell.rectangle.bottomRight(), Qt::yellow, true);
+                cell.text = m_textLayout.getTextFromSelection(textSelection, m_pageIndex).trimmed();
+                cell.text = cell.text.remove(QChar('\n'));
+
                 tableCells.push_back(cell);
             }
         }
 
-        const PDFTextBlocks& textBlocks = m_textLayout.getTextBlocks();
-        for (size_t i = 0; i < textBlocks.size(); ++i)
-        {
-            // Detect, if whole block can be in some text cell
-            const PDFTextBlock& textBlock = textBlocks[i];
-            QRectF textRect = textBlock.getBoundingBox().boundingRect();
-            auto itBlockCell = std::find_if(tableCells.begin(), tableCells.end(), [textRect](const auto& cell) { return cell.rectangle.contains(textRect); });
-            if (itBlockCell != tableCells.end())
-            {
-                // Jakub Melka: whole block is contained in the cell
-                PDFTextSelection blockSelection = m_textLayout.selectBlock(i, m_pageIndex, QColor());
-                QString text = m_textLayout.getTextFromSelection(blockSelection, m_pageIndex);
-                TableCell& cell = *itBlockCell;
-                cell.text = QString("%1 %2").arg(cell.text, text).trimmed();
-                continue;
-            }
-
-            const PDFTextLines& textLines = textBlock.getLines();
-            for (size_t j = 0; j < textLines.size(); ++j)
-            {
-                const PDFTextLine& textLine = textLines[j];
-                QRectF boundingRect = textLine.getBoundingBox().boundingRect();
-
-                auto itLineCell = std::find_if(tableCells.begin(), tableCells.end(), [boundingRect](const auto& cell) { return cell.rectangle.contains(boundingRect); });
-                if (itLineCell != tableCells.end())
-                {
-                    // Jakub Melka: whole block is contained in the cell
-                    PDFTextSelection blockSelection = m_textLayout.selectLineInBlock(i, j, m_pageIndex, QColor());
-                    QString text = m_textLayout.getTextFromSelection(blockSelection, m_pageIndex);
-                    TableCell& cell = *itLineCell;
-                    cell.text = QString("%1 %2").arg(cell.text, text).trimmed();
-                    continue;
-                }
-            }
-        }
-
         if (m_isTransposed)
         {
             auto comparator = [](const TableCell& left, const TableCell right)

RELEASES.txt

@@ -1,6 +1,7 @@
 CURRENT:
  - Issue #21: Table selection tool
  - Issue #22: Solve compilation warnings
+ - Issue #24: Text selection and table selection algorithm improvement
 
 V: 1.2.1 30.6.2022
  - Issue #17: Public key security handler