Followup 2dc5d95, add unit tests

Author: nyalldawson

python/core/qgspallabeling.sip

@@ -772,6 +772,15 @@ class QgsPalLabeling : QgsLabelingEngineInterface
      * @note added in QGIS 2.9
      */
     static QStringList splitToLines( const QString& text, const QString& wrapCharacter );
+    
+    /** Splits a text string to a list of graphemes, which are the smallest allowable character
+     * divisions in the string. This accounts for scripts were individual characters are not
+     * allowed to be split apart (eg Arabic and Indic based scripts)
+     * @param text string to split
+     * @returns list of graphemes
+     * @note added in QGIS 2.10
+     */
+    static QStringList splitToGraphemes( const QString& text );
 
   protected:
     // update temporary QgsPalLayerSettings with any data defined text style values

src/core/qgspalgeometry.h

@@ -2,9 +2,9 @@
 #define QGSPALGEOMETRY_H
 
 #include "qgsgeometry.h"
+#include "qgspallabeling.h"
 #include <pal/feature.h>
 #include <pal/palgeometry.h>
-#include <QTextBoundaryFinder>
 
 using namespace pal;
 
@@ -90,14 +90,7 @@ class QgsPalGeometry : public PalGeometry
       qreal wordSpaceFix;
 
       //split string by valid grapheme boundaries - required for certain scripts (see #6883)
-      QTextBoundaryFinder boundaryFinder( QTextBoundaryFinder::Grapheme, mText );
-      int currentBoundary = -1;
-      int previousBoundary = 0;
-      while (( currentBoundary = boundaryFinder.toNextBoundary() ) > 0 )
-      {
-        mClusters << mText.mid( previousBoundary, currentBoundary - previousBoundary );
-        previousBoundary = currentBoundary;
-      }
+      mClusters = QgsPalLabeling::splitToGraphemes( mText );
 
       mInfo = new pal::LabelInfo( mClusters.count(), labelHeight, maxinangle, maxoutangle );
       for ( int i = 0; i < mClusters.count(); i++ )

src/core/qgspallabeling.cpp

@@ -3412,6 +3412,20 @@ QStringList QgsPalLabeling::splitToLines( const QString &text, const QString &wr
   return multiLineSplit;
 }
 
+QStringList QgsPalLabeling::splitToGraphemes( const QString &text )
+{
+  QStringList graphemes;
+  QTextBoundaryFinder boundaryFinder( QTextBoundaryFinder::Grapheme, text );
+  int currentBoundary = -1;
+  int previousBoundary = 0;
+  while (( currentBoundary = boundaryFinder.toNextBoundary() ) > 0 )
+  {
+    graphemes << text.mid( previousBoundary, currentBoundary - previousBoundary );
+    previousBoundary = currentBoundary;
+  }
+  return graphemes;
+}
+
 QgsGeometry* QgsPalLabeling::prepareGeometry( const QgsGeometry* geometry, const QgsRenderContext& context, const QgsCoordinateTransform* ct, double minSize, QgsGeometry* clipGeometry )
 {
   if ( !geometry )

src/core/qgspallabeling.h

@@ -843,6 +843,15 @@ class CORE_EXPORT QgsPalLabeling : public QgsLabelingEngineInterface
      */
     static QStringList splitToLines( const QString& text, const QString& wrapCharacter );
 
+    /** Splits a text string to a list of graphemes, which are the smallest allowable character
+     * divisions in the string. This accounts for scripts were individual characters are not
+     * allowed to be split apart (eg Arabic and Indic based scripts)
+     * @param text string to split
+     * @returns list of graphemes
+     * @note added in QGIS 2.10
+     */
+    static QStringList splitToGraphemes( const QString& text );
+
   protected:
     // update temporary QgsPalLayerSettings with any data defined text style values
     void dataDefinedTextStyle( QgsPalLayerSettings& tmpLyr,

tests/src/core/testqgspallabeling.cpp

@@ -30,6 +30,7 @@ class TestQgsPalLabeling: public QObject
     void init();// will be called before each testfunction is executed.
     void cleanup();// will be called after every testfunction.
     void wrapChar();//test wrapping text lines
+    void graphemes(); //test splitting strings to graphemes
 
   private:
 };
@@ -63,5 +64,19 @@ void TestQgsPalLabeling::wrapChar()
   QCOMPARE( QgsPalLabeling::splitToLines( "no\nmatching\nchars", QString( "#" ) ), QStringList() << "no" << "matching" << "chars" );
 }
 
+void TestQgsPalLabeling::graphemes()
+{
+  QCOMPARE( QgsPalLabeling::splitToGraphemes( QString() ) , QStringList() );
+  QCOMPARE( QgsPalLabeling::splitToGraphemes( "abcd" ) , QStringList() << "a" << "b" << "c" << "d" );
+  QCOMPARE( QgsPalLabeling::splitToGraphemes( "ab cd" ) , QStringList() << "a" << "b" << " " << "c" << "d" );
+  QCOMPARE( QgsPalLabeling::splitToGraphemes( "ab cd " ) , QStringList() << "a" << "b" << " " << "c" << "d" << " " );
+  QCOMPARE( QgsPalLabeling::splitToGraphemes( QString::fromUtf8( "\u179F\u17D2\u178F\u17D2\u179A\u17B8\u179B\u17D2" ) ) , QStringList() << QString::fromUtf8( "\u179F\u17D2\u178F\u17D2\u179A\u17B8" ) << QString::fromUtf8( "\u179B\u17D2" ) );
+  QCOMPARE( QgsPalLabeling::splitToGraphemes( QString::fromUtf8( "\u1780\u17D2\u179A\u17BB\u1798\u17A2\u1784\u17D2\u1782\u1780\u17B6\u179A\u179F\u17B7\u1791\u17D2\u1792\u17B7\u1798\u1793\u17BB\u179F\u17D2\u179F" ) ) ,
+            QStringList() << QString::fromUtf8( "\u1780\u17D2\u179A\u17BB" ) << QString::fromUtf8( "\u1798" ) << QString::fromUtf8( "\u17A2" )
+            << QString::fromUtf8( "\u1784\u17D2\u1782" ) << QString::fromUtf8( "\u1780\u17B6" ) << QString::fromUtf8( "\u179A" )
+            << QString::fromUtf8( "\u179F\u17B7" ) << QString::fromUtf8( "\u1791\u17D2\u1792\u17B7" ) << QString::fromUtf8( "\u1798" )
+            << QString::fromUtf8( "\u1793\u17BB" ) << QString::fromUtf8( "\u179F\u17D2\u179F" ) );
+}
+
 QTEST_MAIN( TestQgsPalLabeling )
 #include "testqgspallabeling.moc"