More cleanups, avoid multiple parsing of the same HTML content for a …

…label
qgis · May 12, 2020 · 7fc9c55 · 7fc9c55
1 parent a3b9164
commit 7fc9c55
Show file tree

Hide file tree

Showing 9 changed files with 87 additions and 119 deletions.
diff --git a/python/core/auto_generated/labeling/qgspallabeling.sip.in b/python/core/auto_generated/labeling/qgspallabeling.sip.in
@@ -783,7 +783,7 @@ Checks whether a geometry requires preparation before registration with PAL
 .. versionadded:: 2.9
 %End
 
-    static QStringList splitToLines( const QString &text, const QString &wrapCharacter, int autoWrapLength = 0, bool useMaxLineLengthWhenAutoWrapping = true, bool allowHtmlFormatting = false );
+    static QStringList splitToLines( const QString &text, const QString &wrapCharacter, int autoWrapLength = 0, bool useMaxLineLengthWhenAutoWrapping = true );
 %Docstring
 Splits a ``text`` string to a list of separate lines, using a specified wrap character (``wrapCharacter``).
 The text string will be split on either newline characters or the wrap character.

diff --git a/python/core/auto_generated/qgstextrenderer.sip.in b/python/core/auto_generated/qgstextrenderer.sip.in
@@ -1940,15 +1940,6 @@ and background shapes.
       AlignRight,
     };
 
-    static QStringList extractTextBlocksFromHtml( const QString &html );
-%Docstring
-Extracts text blocks from a ``html`` string.
-
-This removes formatting tags, and outputs lines of text which represent the text content of HTML blocks.
-
-.. versionadded:: 3.14
-%End
-
     static int sizeToPixel( double size, const QgsRenderContext &c, QgsUnitTypes::RenderUnit unit, const QgsMapUnitScale &mapUnitScale = QgsMapUnitScale() );
 %Docstring
 Calculates pixel size (considering output size should be in pixel or map units, scale factors and optionally oversampling)

diff --git a/src/core/labeling/qgspallabeling.cpp b/src/core/labeling/qgspallabeling.cpp
@@ -1374,7 +1374,7 @@ bool QgsPalLayerSettings::checkMinimumSizeMM( const QgsRenderContext &ct, const
   return QgsPalLabeling::checkMinimumSizeMM( ct, geom, minSize );
 }
 
-void QgsPalLayerSettings::calculateLabelSize( const QFontMetricsF *fm, const QString &text, double &labelX, double &labelY, const QgsFeature *f, QgsRenderContext *context, double *rotatedLabelX, double *rotatedLabelY )
+void QgsPalLayerSettings::calculateLabelSize( const QFontMetricsF *fm, const QString &text, double &labelX, double &labelY, const QgsFeature *f, QgsRenderContext *context, double *rotatedLabelX, double *rotatedLabelY, QgsTextDocument *document )
 {
   if ( !fm || !f )
   {
@@ -1393,8 +1393,6 @@ void QgsPalLayerSettings::calculateLabelSize( const QFontMetricsF *fm, const QSt
   }
   QgsRenderContext *rc = context ? context : scopedRc.get();
 
-  const bool htmlFormatting = mFormat.allowHtmlFormatting();
-
   QString wrapchr = wrapChar;
   int evalAutoWrapLength = autoWrapLength;
   double multilineH = mFormat.lineHeight();
@@ -1533,7 +1531,19 @@ void QgsPalLayerSettings::calculateLabelSize( const QFontMetricsF *fm, const QSt
 
   double w = 0.0, h = 0.0, rw = 0.0, rh = 0.0;
   double labelHeight = fm->ascent() + fm->descent(); // ignore +1 for baseline
-  const QStringList multiLineSplit = QgsPalLabeling::splitToLines( textCopy, wrapchr, evalAutoWrapLength, useMaxLineLengthForAutoWrap, htmlFormatting );
+
+  QStringList multiLineSplit;
+
+  if ( document )
+  {
+    document->splitLines( wrapchr, evalAutoWrapLength, useMaxLineLengthForAutoWrap );
+    multiLineSplit = document->toPlainText();
+  }
+  else
+  {
+    multiLineSplit = QgsPalLabeling::splitToLines( textCopy, wrapchr, evalAutoWrapLength, useMaxLineLengthForAutoWrap );
+  }
+
   int lines = multiLineSplit.size();
 
   switch ( orientation )
@@ -1889,8 +1899,13 @@ void QgsPalLayerSettings::registerFeature( const QgsFeature &f, QgsRenderContext
   // NOTE: this should come AFTER any option that affects font metrics
   std::unique_ptr<QFontMetricsF> labelFontMetrics( new QFontMetricsF( labelFont ) );
   double labelX, labelY, rotatedLabelX, rotatedLabelY; // will receive label size
-  calculateLabelSize( labelFontMetrics.get(), labelText, labelX, labelY, mCurFeat, &context, &rotatedLabelX, &rotatedLabelY );
 
+  QgsTextDocument doc;
+  if ( format().allowHtmlFormatting() )
+    doc = QgsTextDocument::fromHtml( QStringList() << labelText );
+
+  // also applies the line split to doc!
+  calculateLabelSize( labelFontMetrics.get(), labelText, labelX, labelY, mCurFeat, &context, &rotatedLabelX, &rotatedLabelY, format().allowHtmlFormatting() ? &doc : nullptr );
 
   // maximum angle between curved label characters (hardcoded defaults used in QGIS <2.0)
   //
@@ -2428,6 +2443,7 @@ void QgsPalLayerSettings::registerFeature( const QgsFeature &f, QgsRenderContext
   lf->setAnchorPosition( anchorPosition );
   lf->setFeature( feature );
   lf->setSymbol( symbol );
+  lf->setDocument( doc );
   if ( !qgsDoubleNear( rotatedLabelX, 0.0 ) && !qgsDoubleNear( rotatedLabelY, 0.0 ) )
     lf->setRotatedSize( QSizeF( rotatedLabelX, rotatedLabelY ) );
   mFeatsRegPal++;
@@ -2470,7 +2486,7 @@ void QgsPalLayerSettings::registerFeature( const QgsFeature &f, QgsRenderContext
   // TODO: only for placement which needs character info
   // account for any data defined font metrics adjustments
   lf->calculateInfo( placement == QgsPalLayerSettings::Curved || placement == QgsPalLayerSettings::PerimeterCurved,
-                     labelFontMetrics.get(), xform, maxcharanglein, maxcharangleout, format().allowHtmlFormatting() );
+                     labelFontMetrics.get(), xform, maxcharanglein, maxcharangleout, format().allowHtmlFormatting() ? &doc : nullptr );
   // for labelFeature the LabelInfo is passed to feat when it is registered
 
   // TODO: allow layer-wide feature dist in PAL...?
@@ -3550,49 +3566,33 @@ bool QgsPalLabeling::geometryRequiresPreparation( const QgsGeometry &geometry, Q
   return false;
 }
 
-QStringList QgsPalLabeling::splitToLines( const QString &text, const QString &wrapCharacter, const int autoWrapLength, const bool useMaxLineLengthWhenAutoWrapping, const bool allowHtmlFormatting )
+QStringList QgsPalLabeling::splitToLines( const QString &text, const QString &wrapCharacter, const int autoWrapLength, const bool useMaxLineLengthWhenAutoWrapping )
 {
   QStringList multiLineSplit;
-  auto splitLine = [ & ]( const QString & input )
+  if ( !wrapCharacter.isEmpty() && wrapCharacter != QLatin1String( "\n" ) )
   {
-    QStringList thisParts;
-    if ( !wrapCharacter.isEmpty() && wrapCharacter != QLatin1String( "\n" ) )
+    //wrap on both the wrapchr and new line characters
+    const QStringList lines = text.split( wrapCharacter );
+    for ( const QString &line : lines )
     {
-      //wrap on both the wrapchr and new line characters
-      const QStringList lines = input.split( wrapCharacter );
-      for ( const QString &line : lines )
-      {
-        thisParts.append( line.split( '\n' ) );
-      }
+      multiLineSplit.append( line.split( '\n' ) );
     }
-    else
-    {
-      thisParts = input.split( '\n' );
-    }
-
-    // apply auto wrapping to each manually created line
-    if ( autoWrapLength != 0 )
-    {
-      QStringList autoWrappedLines;
-      autoWrappedLines.reserve( thisParts.count() );
-      for ( const QString &line : qgis::as_const( thisParts ) )
-      {
-        autoWrappedLines.append( QgsStringUtils::wordWrap( line, autoWrapLength, useMaxLineLengthWhenAutoWrapping ).split( '\n' ) );
-      }
-      thisParts = autoWrappedLines;
-    }
-    multiLineSplit.append( thisParts );
-  };
-
-  if ( allowHtmlFormatting )
-  {
-    const QStringList htmlBlocks = QgsTextRenderer::extractTextBlocksFromHtml( text );
-    for ( const QString &block : htmlBlocks )
-      splitLine( block );
   }
   else
   {
-    splitLine( text );
+    multiLineSplit = text.split( '\n' );
+  }
+
+  // apply auto wrapping to each manually created line
+  if ( autoWrapLength != 0 )
+  {
+    QStringList autoWrappedLines;
+    autoWrappedLines.reserve( multiLineSplit.count() );
+    for ( const QString &line : qgis::as_const( multiLineSplit ) )
+    {
+      autoWrappedLines.append( QgsStringUtils::wordWrap( line, autoWrapLength, useMaxLineLengthWhenAutoWrapping ).split( '\n' ) );
+    }
+    multiLineSplit = autoWrappedLines;
   }
   return multiLineSplit;
 }

diff --git a/src/core/labeling/qgspallabeling.h b/src/core/labeling/qgspallabeling.h
@@ -943,7 +943,12 @@ class CORE_EXPORT QgsPalLayerSettings
      * If the text orientation is set to rotation-based, the spaced taken to render
      * vertically oriented text will be written to \a rotatedLabelX and \a rotatedLabelY .
      */
+#ifndef SIP_RUN
+    void calculateLabelSize( const QFontMetricsF *fm, const QString &text, double &labelX, double &labelY, const QgsFeature *f = nullptr, QgsRenderContext *context = nullptr, double *rotatedLabelX SIP_OUT = nullptr, double *rotatedLabelY SIP_OUT = nullptr,
+                             QgsTextDocument *document = nullptr );
+#else
     void calculateLabelSize( const QFontMetricsF *fm, const QString &text, double &labelX, double &labelY, const QgsFeature *f = nullptr, QgsRenderContext *context = nullptr, double *rotatedLabelX SIP_OUT = nullptr, double *rotatedLabelY SIP_OUT = nullptr );
+#endif
 
     /**
      * Register a feature for labeling.
@@ -1301,7 +1306,7 @@ class CORE_EXPORT QgsPalLabeling
      *
      * \since QGIS 2.9
      */
-    static QStringList splitToLines( const QString &text, const QString &wrapCharacter, int autoWrapLength = 0, bool useMaxLineLengthWhenAutoWrapping = true, bool allowHtmlFormatting = false );
+    static QStringList splitToLines( const QString &text, const QString &wrapCharacter, int autoWrapLength = 0, bool useMaxLineLengthWhenAutoWrapping = true );
 
     /**
      * Splits a text string to a list of graphemes, which are the smallest allowable character

diff --git a/src/core/labeling/qgstextlabelfeature.cpp b/src/core/labeling/qgstextlabelfeature.cpp
@@ -52,7 +52,7 @@ bool QgsTextLabelFeature::hasCharacterFormat( int partId ) const
   return partId < mCharacterFormats.size();
 }
 
-void QgsTextLabelFeature::calculateInfo( bool curvedLabeling, QFontMetricsF *fm, const QgsMapToPixel *xform, double maxinangle, double maxoutangle, bool allowHtmlTags )
+void QgsTextLabelFeature::calculateInfo( bool curvedLabeling, QFontMetricsF *fm, const QgsMapToPixel *xform, double maxinangle, double maxoutangle, QgsTextDocument *document )
 {
   if ( mInfo )
     return;
@@ -81,34 +81,19 @@ void QgsTextLabelFeature::calculateInfo( bool curvedLabeling, QFontMetricsF *fm,
   qreal charWidth;
   qreal wordSpaceFix;
 
-
-  if ( allowHtmlTags && curvedLabeling )
+  if ( document && curvedLabeling )
   {
-    QTextDocument doc;
-    doc.setHtml( mLabelText );
-
-    QTextBlock block = doc.firstBlock();
-    while ( true )
+    for ( const QgsTextBlock &block : qgis::as_const( *document ) )
     {
-      auto it = block.begin();
-      QString blockText;
-      while ( !it.atEnd() )
+      for ( const QgsTextFragment &fragment : block )
       {
-        const QTextFragment fragment = it.fragment();
-        if ( fragment.isValid() )
+        const QStringList graphemes = QgsPalLabeling::splitToGraphemes( fragment.text() );
+        for ( const QString &grapheme : graphemes )
         {
-          const QStringList graphemes = QgsPalLabeling::splitToGraphemes( fragment.text() );
-          for ( const QString &grapheme : graphemes )
-          {
-            mClusters.append( grapheme );
-            mCharacterFormats.append( fragment.charFormat() );
-          }
+          mClusters.append( grapheme );
+          mCharacterFormats.append( fragment.characterFormat() );
         }
-        it++;
       }
-      block = block.next();
-      if ( !block.isValid() )
-        break;
     }
   }
   else
@@ -148,3 +133,13 @@ void QgsTextLabelFeature::calculateInfo( bool curvedLabeling, QFontMetricsF *fm,
     mInfo->char_info[i].width = labelWidth;
   }
 }
+
+QgsTextDocument QgsTextLabelFeature::document() const
+{
+  return mDocument;
+}
+
+void QgsTextLabelFeature::setDocument( const QgsTextDocument &document )
+{
+  mDocument = document;
+}
diff --git a/src/core/labeling/qgstextlabelfeature.h b/src/core/labeling/qgstextlabelfeature.h
@@ -60,7 +60,7 @@ class QgsTextLabelFeature : public QgsLabelFeature
     bool hasCharacterFormat( int partId ) const;
 
     //! calculate data for info(). setDefinedFont() must have been called already.
-    void calculateInfo( bool curvedLabeling, QFontMetricsF *fm, const QgsMapToPixel *xform, double maxinangle, double maxoutangle, bool allowHtmlTags );
+    void calculateInfo( bool curvedLabeling, QFontMetricsF *fm, const QgsMapToPixel *xform, double maxinangle, double maxoutangle, QgsTextDocument *document = nullptr );
 
     //! Gets data-defined values
     const QMap< QgsPalLayerSettings::Property, QVariant > &dataDefinedValues() const { return mDataDefinedValues; }
@@ -75,6 +75,20 @@ class QgsTextLabelFeature : public QgsLabelFeature
     //! Metrics of the font for rendering
     QFontMetricsF *labelFontMetrics() { return mFontMetrics; }
 
+    /**
+     * Returns the document for the label.
+     * \see setDocument()
+     * \since QGIS 3.14
+     */
+    QgsTextDocument document() const;
+
+    /**
+     * Sets the \a document for the label.
+     * \see document()
+     * \since QGIS 3.14
+     */
+    void setDocument( const QgsTextDocument &document );
+
   protected:
     //! List of graphemes (used for curved labels)
     QStringList mClusters;
@@ -88,6 +102,8 @@ class QgsTextLabelFeature : public QgsLabelFeature
     //! Stores attribute values for data defined properties
     QMap< QgsPalLayerSettings::Property, QVariant > mDataDefinedValues;
 
+    QgsTextDocument mDocument;
+
 };
 
 #endif //QGSTEXTLABELFEATURE_H
diff --git a/src/core/labeling/qgsvectorlayerlabelprovider.cpp b/src/core/labeling/qgsvectorlayerlabelprovider.cpp
@@ -633,8 +633,7 @@ void QgsVectorLayerLabelProvider::drawLabelPrivate( pal::LabelPosition *label, Q
     }
     else
     {
-      document = QgsTextDocument::fromHtml( QStringList() << txt );
-      document.splitLines( tmpLyr.wrapChar, tmpLyr.autoWrapLength, tmpLyr.useMaxLineLengthForAutoWrap );
+      document = lf->document();
     }
 
     QgsTextRenderer::drawTextInternal( drawType, context, tmpLyr.format(), component, document, labelfm,

diff --git a/src/core/qgstextrenderer.cpp b/src/core/qgstextrenderer.cpp
@@ -2605,38 +2605,6 @@ QPixmap QgsTextFormat::textFormatPreviewPixmap( const QgsTextFormat &format, QSi
   return pixmap;
 }
 
-QStringList QgsTextRenderer::extractTextBlocksFromHtml( const QString &html )
-{
-  QStringList res;
-
-  QTextDocument doc;
-  doc.setHtml( html );
-
-  QTextBlock block = doc.firstBlock();
-  while ( true )
-  {
-    auto it = block.begin();
-    QString blockText;
-    while ( !it.atEnd() )
-    {
-      const QTextFragment fragment = it.fragment();
-      if ( fragment.isValid() )
-      {
-        blockText.append( fragment.text() );
-      }
-      it++;
-    }
-    if ( !blockText.isEmpty() )
-      res << blockText;
-
-    block = block.next();
-    if ( !block.isValid() )
-      break;
-  }
-
-  return res;
-}
-
 int QgsTextRenderer::sizeToPixel( double size, const QgsRenderContext &c, QgsUnitTypes::RenderUnit unit, const QgsMapUnitScale &mapUnitScale )
 {
   return static_cast< int >( c.convertToPainterUnits( size, unit, mapUnitScale ) + 0.5 ); //NOLINT
@@ -4475,6 +4443,9 @@ QgsTextDocument QgsTextDocument::fromHtml( const QStringList &lines )
   document.reserve( lines.size() );
   for ( const QString &line : lines )
   {
+    // QTextDocument is a very heavy way of parsing HTML + css (it's heavily geared toward an editable text document,
+    // and includes a LOT of calculations we don't need, when all we're after is a HTML + CSS style parser).
+    // TODO - try to find an alternative library we can use here
     QTextDocument sourceDoc;
     sourceDoc.setHtml( line );
 

diff --git a/src/core/qgstextrenderer.h b/src/core/qgstextrenderer.h
@@ -1955,15 +1955,6 @@ class CORE_EXPORT QgsTextRenderer
       AlignRight, //!< Right align
     };
 
-    /**
-     * Extracts text blocks from a \a html string.
-     *
-     * This removes formatting tags, and outputs lines of text which represent the text content of HTML blocks.
-     *
-     * \since QGIS 3.14
-     */
-    static QStringList extractTextBlocksFromHtml( const QString &html );
-
     /**
      * Calculates pixel size (considering output size should be in pixel or map units, scale factors and optionally oversampling)
      * \param size size to convert