Add method to extract blocks from HTML text for labeling

qgis · May 12, 2020 · 87296f2 · 87296f2
1 parent 2cbea02
commit 87296f2
Show file tree

Hide file tree

Showing 4 changed files with 58 additions and 0 deletions.
diff --git a/python/core/auto_generated/qgstextrenderer.sip.in b/python/core/auto_generated/qgstextrenderer.sip.in
@@ -1940,6 +1940,15 @@ 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/qgstextrenderer.cpp b/src/core/qgstextrenderer.cpp
@@ -29,6 +29,8 @@
 #include "qgspainterswapper.h"
 #include <QFontDatabase>
 #include <QDesktopWidget>
+#include <QTextDocument>
+#include <QTextBlock>
 
 Q_GUI_EXPORT extern int qt_defaultDpiX();
 Q_GUI_EXPORT extern int qt_defaultDpiY();
@@ -2603,6 +2605,37 @@ 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 )
 {

diff --git a/src/core/qgstextrenderer.h b/src/core/qgstextrenderer.h
@@ -1670,6 +1670,15 @@ 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

diff --git a/tests/src/python/test_qgstextrenderer.py b/tests/src/python/test_qgstextrenderer.py
@@ -61,6 +61,13 @@ def tearDown(self):
         with open(report_file_path, 'a') as report_file:
             report_file.write(self.report)
 
+    def testExtractBlocksFromHtml(self):
+        self.assertEqual(QgsTextRenderer.extractTextBlocksFromHtml(''), [])
+        self.assertEqual(QgsTextRenderer.extractTextBlocksFromHtml('abc'), ['abc'])
+        self.assertEqual(QgsTextRenderer.extractTextBlocksFromHtml('abc\ndef'), ['abc def'])
+        self.assertEqual(QgsTextRenderer.extractTextBlocksFromHtml('abc<b>def</b>'), ['abcdef'])
+        self.assertEqual(QgsTextRenderer.extractTextBlocksFromHtml('abc<div><b>def</b><div>ghi</div></div>'), ['abc', 'def', 'ghi'])
+
     def createBufferSettings(self):
         s = QgsTextBufferSettings()
         s.setEnabled(True)