Bugfixing of text layout

PdfForQtLib/sources/pdfdrawspacecontroller.cpp

@@ -25,6 +25,7 @@
 #include "pdfcms.h"
 
 #include <QPainter>
+#include <QFontMetrics>
 
 namespace pdf
 {
@@ -622,6 +623,55 @@ void PDFDrawWidgetProxy::draw(QPainter* painter, QRect rect)
                 QMatrix matrix = PDFRenderer::createPagePointToDevicePointMatrix(page, placedRect);
                 compiledPage->draw(painter, page->getCropBox(), matrix, m_features);
 
+                // Draw text blocks/text lines, if it is enabled
+                if (m_features.testFlag(PDFRenderer::DebugTextBlocks))
+                {
+                    const PDFTextLayout& layout = compiledPage->getTextLayout();
+                    const PDFTextBlocks& textBlocks = layout.getTextBlocks();
+
+                    painter->save();
+                    painter->setFont(m_widget->font());
+                    painter->setPen(Qt::red);
+                    painter->setBrush(QColor(255, 0, 0, 128));
+
+                    QFontMetricsF fontMetrics(painter->font(), painter->device());
+                    int blockIndex = 1;
+                    for (const PDFTextBlock& block : textBlocks)
+                    {
+                        QString blockNumber = QString::number(blockIndex++);
+
+                        painter->drawPath(matrix.map(block.getBoundingBox()));
+                        painter->drawText(matrix.map(block.getTopLeft()) - QPointF(fontMetrics.width(blockNumber), 0), blockNumber, Qt::TextSingleLine, 0);
+                    }
+
+                    painter->restore();
+                }
+                if (m_features.testFlag(PDFRenderer::DebugTextLines))
+                {
+                    const PDFTextLayout& layout = compiledPage->getTextLayout();
+                    const PDFTextBlocks& textBlocks = layout.getTextBlocks();
+
+                    painter->save();
+                    painter->setFont(m_widget->font());
+                    painter->setPen(Qt::green);
+                    painter->setBrush(QColor(0, 255, 0, 128));
+
+                    QFontMetricsF fontMetrics(painter->font(), painter->device());
+                    int lineIndex = 1;
+                    for (const PDFTextBlock& block : textBlocks)
+                    {
+                        for (const PDFTextLine& line : block.getLines())
+                        {
+                            QString lineNumber = QString::number(lineIndex++);
+
+                            painter->drawPath(matrix.map(line.getBoundingBox()));
+                            painter->drawText(matrix.map(line.getTopLeft()) - QPointF(fontMetrics.width(lineNumber), 0), lineNumber, Qt::TextSingleLine, 0);
+                        }
+                    }
+
+                    painter->restore();
+                }
+
                 const qint64 drawTimeNS = timer.nsecsElapsed();
 
                 // Draw rendering times

PdfForQtLib/sources/pdfpainter.h

@@ -217,6 +217,9 @@ public:
     /// Returns memory consumption estimate
     qint64 getMemoryConsumptionEstimate() const { return m_memoryConsumptionEstimate; }
 
+    /// Returns text layout of the page
+    const PDFTextLayout& getTextLayout() const { return m_textLayout; }
+
 private:
     struct PathPaintData
     {

PdfForQtLib/sources/pdfrenderer.h

@@ -49,6 +49,8 @@ public:
         IgnoreOptionalContent   = 0x0008,   ///< Ignore optional content (so all is drawn ignoring settings of optional content)
         ClipToCropBox           = 0x0010,   ///< Clip page content to crop box (items outside crop box will not be visible)
         DisplayTimes            = 0x0020,   ///< Display page compile/draw time
+        DebugTextBlocks         = 0x0040,   ///< Debug text block layout algorithm
+        DebugTextLines          = 0x0080,   ///< Debug text line layout algorithm
     };
 
     Q_DECLARE_FLAGS(Features, Feature)

PdfForQtLib/sources/pdftextlayout.cpp

@@ -261,7 +261,7 @@ void PDFTextLayout::performDoLayout(PDFReal angle)
         QRectF aBB = blocks[aIndex].getBoundingBox().boundingRect();
         QRectF bBB = blocks[bIndex].getBoundingBox().boundingRect();
 
-        const bool isOverlappedOnHorizontalAxis = (aBB.right() < bBB.left() && aBB.left() < bBB.right()) || (bBB.right() < aBB.left() && bBB.left() < aBB.right());
+        const bool isOverlappedOnHorizontalAxis = isRectangleHorizontallyOverlapped(aBB, bBB);
         const bool isAoverB = aBB.bottom() > bBB.top();
         return isOverlappedOnHorizontalAxis && isAoverB;
     };
@@ -284,8 +284,8 @@ void PDFTextLayout::performDoLayout(PDFReal angle)
                 QRectF cBB = blocks[i].getBoundingBox().boundingRect();
                 if (cBB.top() >= abBB.top() && cBB.bottom() <= abBB.bottom())
                 {
-                    const bool isAOverlappedOnHorizontalAxis = (aBB.right() < cBB.left() && aBB.left() < cBB.right()) || (cBB.right() < aBB.left() && cBB.left() < aBB.right());
+                    const bool isAOverlappedOnHorizontalAxis = isRectangleHorizontallyOverlapped(aBB, cBB);
-                    const bool isBOverlappedOnHorizontalAxis = (cBB.right() < bBB.left() && cBB.left() < bBB.right()) || (bBB.right() < cBB.left() && bBB.left() < cBB.right());
+                    const bool isBOverlappedOnHorizontalAxis = isRectangleHorizontallyOverlapped(bBB, cBB);
                     if (isAOverlappedOnHorizontalAxis && isBOverlappedOnHorizontalAxis)
                     {
                         return false;
@@ -310,7 +310,7 @@ void PDFTextLayout::performDoLayout(PDFReal angle)
         workBlocks.insert(workBlocks.end(), i);
         for (size_t j = 0; j < blocks.size(); ++j)
         {
-            if (isBeforeByRule1(j, i) || isBeforeByRule2(j, i))
+            if (i != j && (isBeforeByRule1(j, i) || isBeforeByRule2(j, i)))
             {
                 orderingEdges[i].insert(j);
             }
@@ -361,11 +361,13 @@ PDFTextLine::PDFTextLine(TextCharacters characters) :
         boundingBox = boundingBox.united(character.boundingBox.boundingRect());
     }
     m_boundingBox.addRect(boundingBox);
+    m_topLeft = boundingBox.topLeft();
 }
 
 void PDFTextLine::applyTransform(const QMatrix& matrix)
 {
     m_boundingBox = matrix.map(m_boundingBox);
+    m_topLeft = matrix.map(m_topLeft);
     for (TextCharacter& character : m_characters)
     {
         character.applyTransform(matrix);
@@ -383,7 +385,7 @@ PDFTextBlock::PDFTextBlock(PDFTextLines textLines) :
         const PDFReal xR = br.x();
         const PDFReal yL = qRound(bl.y() * 100.0);
         const PDFReal yR = qRound(br.y() * 100.0);
-        return std::tie(yL, xL) < std::tie(yR, xR);
+        return std::tie(-yL, xL) < std::tie(-yR, xR);
     };
     std::sort(m_lines.begin(), m_lines.end(), sortFunction);
 
@@ -393,11 +395,13 @@ PDFTextBlock::PDFTextBlock(PDFTextLines textLines) :
         boundingBox = boundingBox.united(line.getBoundingBox().boundingRect());
     }
     m_boundingBox.addRect(boundingBox);
+    m_topLeft = boundingBox.topLeft();
 }
 
 void PDFTextBlock::applyTransform(const QMatrix& matrix)
 {
     m_boundingBox = matrix.map(m_boundingBox);
+    m_topLeft = matrix.map(m_topLeft);
     for (PDFTextLine& textLine : m_lines)
     {
         textLine.applyTransform(matrix);

PdfForQtLib/sources/pdftextlayout.h

@@ -104,12 +104,14 @@ public:
 
     const TextCharacters& getCharacters() const { return m_characters; }
     const QPainterPath& getBoundingBox() const { return m_boundingBox; }
+    const QPointF& getTopLeft() const { return m_topLeft; }
 
     void applyTransform(const QMatrix& matrix);
 
 private:
     TextCharacters m_characters;
     QPainterPath m_boundingBox;
+    QPointF m_topLeft;
 };
 
 using PDFTextLines = std::vector<PDFTextLine>;
@@ -122,12 +124,14 @@ public:
 
     const PDFTextLines& getLines() const { return m_lines; }
     const QPainterPath& getBoundingBox() const { return m_boundingBox; }
+    const QPointF& getTopLeft() const { return m_topLeft; }
 
     void applyTransform(const QMatrix& matrix);
 
 private:
     PDFTextLines m_lines;
     QPainterPath m_boundingBox;
+    QPointF m_topLeft;
 };
 
 using PDFTextBlocks = std::vector<PDFTextBlock>;
@@ -151,6 +155,9 @@ public:
     /// Returns estimate of number of bytes, which this mesh occupies in memory
     qint64 getMemoryConsumptionEstimate() const;
 
+    /// Returns recognized text blocks
+    const PDFTextBlocks& getTextBlocks() const { return m_blocks; }
+
 private:
     /// Makes layout for particular angle
     void performDoLayout(PDFReal angle);

PdfForQtLib/sources/pdfutils.h

@@ -21,6 +21,7 @@
 
 #include "pdfglobal.h"
 
+#include <QRectF>
 #include <QByteArray>
 #include <QDataStream>
 
@@ -407,6 +408,27 @@ private:
     std::vector<T> m_indices;
 };
 
+template<typename T>
+constexpr bool isIntervalOverlap(T x1_min, T x1_max, T x2_min, T x2_max)
+{
+    // We have two situations, where intervals doesn't overlap:
+    //    1)  |--------|        |---------|
+    //       x1_min   x1_max   x2_min    x2_max
+    //    2)  |--------|        |---------|
+    //       x2_min   x2_max   x1_min    x1_max
+    if (x1_max < x2_min || x2_max < x1_min)
+    {
+        return false;
+    }
+
+    return true;
+}
+
+constexpr bool isRectangleHorizontallyOverlapped(const QRectF& r1, const QRectF& r2)
+{
+    return isIntervalOverlap(r1.left(), r1.right(), r2.left(), r2.right());
+}
+
 }   // namespace pdf
 
 #endif // PDFUTILS_H

PdfForQtViewer/pdfviewermainwindow.cpp

@@ -173,6 +173,8 @@ PDFViewerMainWindow::PDFViewerMainWindow(QWidget* parent) :
     ui->actionRenderOptionTextAntialiasing->setData(pdf::PDFRenderer::TextAntialiasing);
     ui->actionRenderOptionSmoothPictures->setData(pdf::PDFRenderer::SmoothImages);
     ui->actionRenderOptionIgnoreOptionalContentSettings->setData(pdf::PDFRenderer::IgnoreOptionalContent);
+    ui->actionShow_Text_Blocks->setData(pdf::PDFRenderer::DebugTextBlocks);
+    ui->actionShow_Text_Lines->setData(pdf::PDFRenderer::DebugTextLines);
 
     for (QAction* action : getRenderingOptionActions())
     {
@@ -869,7 +871,12 @@ void PDFViewerMainWindow::setPageLayout(pdf::PageLayout pageLayout)
 
 std::vector<QAction*> PDFViewerMainWindow::getRenderingOptionActions() const
 {
-    return { ui->actionRenderOptionAntialiasing, ui->actionRenderOptionTextAntialiasing, ui->actionRenderOptionSmoothPictures, ui->actionRenderOptionIgnoreOptionalContentSettings };
+    return { ui->actionRenderOptionAntialiasing,
+             ui->actionRenderOptionTextAntialiasing,
+             ui->actionRenderOptionSmoothPictures,
+             ui->actionRenderOptionIgnoreOptionalContentSettings,
+             ui->actionShow_Text_Blocks,
+             ui->actionShow_Text_Lines };
 }
 
 QList<QAction*> PDFViewerMainWindow::getActions() const

PdfForQtViewer/pdfviewermainwindow.h

@@ -85,9 +85,7 @@ private slots:
     void on_actionFitPage_triggered();
     void on_actionFitWidth_triggered();
     void on_actionFitHeight_triggered();
-
     void on_actionProperties_triggered();
-
     void on_actionSend_by_E_Mail_triggered();
 
 private:

PdfForQtViewer/pdfviewermainwindow.ui

@@ -89,11 +89,19 @@
     </property>
     <addaction name="actionAbout"/>
    </widget>
+   <widget class="QMenu" name="menuDeveloper">
+    <property name="title">
+     <string>Developer</string>
+    </property>
+    <addaction name="actionShow_Text_Blocks"/>
+    <addaction name="actionShow_Text_Lines"/>
+   </widget>
    <addaction name="menuFile"/>
    <addaction name="menuView"/>
    <addaction name="menuGoTo"/>
    <addaction name="menuTools"/>
    <addaction name="menuHelp"/>
+   <addaction name="menuDeveloper"/>
   </widget>
   <widget class="QToolBar" name="mainToolBar">
    <attribute name="toolBarArea">
@@ -316,6 +324,22 @@
     <string>Send by E-Mail</string>
    </property>
   </action>
+  <action name="actionShow_Text_Blocks">
+   <property name="checkable">
+    <bool>true</bool>
+   </property>
+   <property name="text">
+    <string>Show Text Blocks</string>
+   </property>
+  </action>
+  <action name="actionShow_Text_Lines">
+   <property name="checkable">
+    <bool>true</bool>
+   </property>
+   <property name="text">
+    <string>Show Text Lines</string>
+   </property>
+  </action>
  </widget>
  <layoutdefault spacing="6" margin="11"/>
  <resources>