Add support for unicode surrogate characters to Kate/KWrite

Signed-off-by: Michele Calgaro <michele.calgaro@yahoo.it>

kate/part/katedocument.cpp

@@ -2411,7 +2411,7 @@ bool KateDocument::openFile(TDEIO::Job * job)
 
   // read dir config (if possible and wanted)
   if (!m_reloading)
-	readDirConfig ();
+    readDirConfig ();
 
   // do we have success ?
   bool success = m_buffer->openFile (m_file);
@@ -3135,15 +3135,27 @@ void KateDocument::backspace( KateView *view, const KateTextCursor& c )
   if ((col == 0) && (line == 0))
     return;
 
+  KateTextLine::Ptr tl = m_buffer->plainLine(line);
+  if (!tl)
+  {
+    return;
+  }
+   
+  // Make sure to handle surrogate pairs correctly
+  uint fromCol = col - 1;
+  TQChar prevChar = tl->getChar(col - 1);
+  if (prevChar.isLowSurrogate() && tl->getChar(col - 2).isHighSurrogate())
+  {
+    fromCol = col - 2;
+    prevChar = tl->getChar(col - 2);
+  }
+
   int complement = 0;
   if (col > 0)
   {
     if (config()->configFlags() & KateDocument::cfAutoBrackets)
     {
       // if inside empty (), {}, [], '', "" delete both
-      KateTextLine::Ptr tl = m_buffer->plainLine(line);
-      if(!tl) return;
-      TQChar prevChar = tl->getChar(col-1);
       TQChar nextChar = tl->getChar(col);
 
       if ( (prevChar == '"' && nextChar == '"') ||
@@ -3158,8 +3170,7 @@ void KateDocument::backspace( KateView *view, const KateTextCursor& c )
     if (!(config()->configFlags() & KateDocument::cfBackspaceIndents))
     {
       // ordinary backspace
-      //c.cursor.col--;
-      removeText(line, col-1, line, col+complement);
+      removeText(line, fromCol, line, col+complement);
     }
     else
     {
@@ -3215,9 +3226,23 @@ void KateDocument::del( KateView *view, const KateTextCursor& c )
     return;
   }
 
-  if( c.col() < (int) m_buffer->plainLine(c.line())->length())
+  KateTextLine::Ptr tl = m_buffer->plainLine(c.line());
+  if (!tl)
   {
-    removeText(c.line(), c.col(), c.line(), c.col()+1);
+    return;
+  }
+  uint lineLen = tl->length();
+  uint col = (uint)c.col();
+  if (col < lineLen)
+  {
+    // Make sure to handle surrogate pairs correctly
+    uint toCol = col + 1;
+    if (tl->getChar(col).isHighSurrogate() && tl->getChar(col + 1).isLowSurrogate())
+    {
+      toCol = col + 2;
+    }
+
+    removeText(c.line(), col, c.line(), toCol);
   }
   else if ( (uint)c.line() < lastLine() )
   {

kate/part/katerenderer.cpp

@@ -400,7 +400,7 @@ void KateRenderer::paintTextLine(TQPainter& paint, const KateLineRange* range, i
     uint nextCol = curCol + 1;
 
     // text + attrib data from line
-    const uchar *textAttributes = textLine->attributes ();
+    const uchar *textAttributes = textLine->attributes();
     bool noAttribs = !textAttributes;
 
     // adjust to startcol ;)
@@ -421,26 +421,29 @@ void KateRenderer::paintTextLine(TQPainter& paint, const KateLineRange* range, i
 
     while (curCol - startcol < len)
     {
+      const TQString &textString = textLine->string();
+      int currCharNumCols = textString[curCol].isHighSurrogate() ? 2 : 1;
+      nextCol = curCol + currCharNumCols;
+
       // make sure curPos is updated correctly.
       // ### if uncommented, causes an O(n^2) behaviour
       //Q_ASSERT(curPos == textLine->cursorX(curCol, m_tabWidth));
 
-      TQChar curChar = textLine->string()[curCol];
       // Decide if this character is a tab - we treat the spacing differently
       // TODO: move tab width calculation elsewhere?
-      bool isTab = curChar == TQChar('\t');
+      bool isTab = textString[curCol] == TQChar('\t');
 
       // Determine current syntax highlighting attribute
       // A bit legacy but doesn't need to change
       KateAttribute* curAt = (noAttribs || ((*textAttributes) >= atLen)) ? &attr[0] : &attr[*textAttributes];
 
       // X position calculation. Incorrect for fonts with non-zero leftBearing() and rightBearing() results.
-      // TODO: make internal charWidth() function, use TQFontMetrics::charWidth().
-      xPosAfter += curAt->width(*fs, curChar, m_tabWidth);
+      // TODO: make internal charWidth() function.
+      xPosAfter += curAt->width(*fs, textString, curCol, m_tabWidth);
 
       // Tab special treatment, move to charWidth().
       if (isTab)
-        xPosAfter -= (xPosAfter % curAt->width(*fs, curChar, m_tabWidth));
+        xPosAfter -= (xPosAfter % curAt->width(*fs, textString[curCol], m_tabWidth));
 
       // Only draw after the starting X value
       // Haha, this was always wrong, due to the use of individual char width calculations...?? :(
@@ -459,7 +462,9 @@ void KateRenderer::paintTextLine(TQPainter& paint, const KateLineRange* range, i
         curColor = isSel ? &(curAt->selectedTextColor()) : &(curAt->textColor());
 
         // Incorporate in arbitrary highlighting
-        if (curAt != oldAt || curColor != oldColor || (superRanges.count() && superRanges.currentBoundary() && *(superRanges.currentBoundary()) == currentPos)) {
+        if (curAt != oldAt || curColor != oldColor || (superRanges.count() &&
+                superRanges.currentBoundary() && *(superRanges.currentBoundary()) == currentPos))
+        {
           if (superRanges.count() && superRanges.currentBoundary() && *(superRanges.currentBoundary()) == currentPos)
             customHL = KateArbitraryHighlightRange::merge(superRanges.rangesIncluding(currentPos));
 
@@ -495,10 +500,10 @@ void KateRenderer::paintTextLine(TQPainter& paint, const KateLineRange* range, i
           || (superRanges.count() && superRanges.currentBoundary() && *(superRanges.currentBoundary()) == KateTextCursor(line, nextCol))
 
           // it is the end of the line OR
-          || (curCol - startcol >= len - 1)
+          || ((curCol - startcol) >= (len - currCharNumCols))
 
           // the rest of the line is trailing whitespace OR
-          || (curCol + 1 >= trailingWhitespaceColumn)
+          || (nextCol >= trailingWhitespaceColumn)
 
           // indentation lines OR
           || (showIndentLines() && curCol < lastIndentColumn)
@@ -514,7 +519,7 @@ void KateRenderer::paintTextLine(TQPainter& paint, const KateLineRange* range, i
 
           // the next char is a tab (removed the "and this isn't" because that's dealt with above)
           // i.e. we have to draw the current text so the tab can be rendered as above.
-          || (textLine->string()[nextCol] == TQChar('\t'))
+          || (textString[nextCol] == TQChar('\t'))
 
           // input method edit area
           || ( m_view && (isIMEdit != m_view->isIMEdit( line, nextCol )) )
@@ -557,7 +562,7 @@ void KateRenderer::paintTextLine(TQPainter& paint, const KateLineRange* range, i
 
                 // If this is the last block of text, fill up to the end of the line if the
                 // selection stretches that far
-                if ((curCol >= len - 1) && m_view->lineEndSelected (line, endcol))
+                if (curCol >= (len - currCharNumCols) && m_view->lineEndSelected(line, endcol))
                   width = xEnd - oldXPos;
               }
               else
@@ -632,7 +637,7 @@ void KateRenderer::paintTextLine(TQPainter& paint, const KateLineRange* range, i
           else
           {
             // Here's where the money is...
-            paint.drawText(oldXPos-xStart, y, textLine->string(), blockStartCol, nextCol-blockStartCol);
+            paint.drawText(oldXPos-xStart, y, textString, blockStartCol, nextCol-blockStartCol);
 
             // Draw preedit's underline
             if (isIMEdit) {
@@ -650,7 +655,6 @@ void KateRenderer::paintTextLine(TQPainter& paint, const KateLineRange* range, i
             // variable advancement
             blockStartCol = nextCol;
             oldXPos = xPosAfter;
-            //oldS = s+1;
           }
         } // renderNow
 
@@ -681,9 +685,8 @@ void KateRenderer::paintTextLine(TQPainter& paint, const KateLineRange* range, i
       oldColor = curColor;
 
       // col move
-      curCol++;
-      nextCol++;
-      currentPos.setCol(currentPos.col() + 1);
+      curCol += currCharNumCols;
+      currentPos.setCol(currentPos.col() + currCharNumCols);
 
       // Update the current indentation pos.
       if (isTab)
@@ -756,8 +759,6 @@ uint KateRenderer::textWidth(const KateTextLine::Ptr &textLine, int cursorCol)
     if (z < len) {
       width = a->width(*fs, textString, z, m_tabWidth);
     } else {
-      // DF: commented out. It happens all the time.
-      //Q_ASSERT(!m_doc->wrapCursor());
       width = a->width(*fs, TQChar(' '), m_tabWidth);
     }
 
@@ -765,6 +766,11 @@ uint KateRenderer::textWidth(const KateTextLine::Ptr &textLine, int cursorCol)
 
     if (z < len && unicode[z] == TQChar('\t'))
       x -= x % width;
+
+    if (textString[z].isHighSurrogate())
+    {
+      ++z;
+    }
   }
 
   return x;
@@ -803,7 +809,11 @@ uint KateRenderer::textWidth(const KateTextLine::Ptr &textLine, uint startcol, u
     if (unicode[z] == TQChar('\t'))
       x -= x % width;
 
-    if (unicode[z].isSpace())
+    if (textString[z].isHighSurrogate())
+    {
+      ++z;
+    }
+    else if (unicode[z].isSpace())
     {
       lastWhiteSpace = z+1;
       lastWhiteSpaceX = x;
@@ -876,7 +886,6 @@ uint KateRenderer::textWidth(const KateTextCursor &cursor)
 uint KateRenderer::textWidth( KateTextCursor &cursor, int xPos, uint startCol)
 {
   bool wrapCursor = m_view->wrapCursor();
-  int x, oldX;
 
   KateFontStruct *fs = config()->fontStruct();
 
@@ -890,15 +899,17 @@ uint KateRenderer::textWidth( KateTextCursor &cursor, int xPos, uint startCol)
   const TQChar *unicode = textLine->text();
   const TQString &textString = textLine->string();
 
-  x = oldX = 0;
+  int x = 0;
+  int oldX = 0;
+  int oldZ = 0;
+  int width = 0;
   uint z = startCol;
   while (x < xPos && (!wrapCursor || z < len)) {
     oldX = x;
+    oldZ = z;
 
     KateAttribute* a = attribute(textLine->attribute(z));
 
-    int width = 0;
-
     if (z < len)
       width = a->width(*fs, textString, z, m_tabWidth);
     else
@@ -909,11 +920,11 @@ uint KateRenderer::textWidth( KateTextCursor &cursor, int xPos, uint startCol)
     if (z < len && unicode[z] == TQChar('\t'))
       x -= x % width;
 
-    z++;
+    z += textString[z].isHighSurrogate() ? 2 : 1;
   }
   if (xPos - oldX < x - xPos && z > 0) {
-    z--;
     x = oldX;
+    z = oldZ;
   }
   cursor.setCol(z);
   return x;
@@ -942,23 +953,24 @@ uint KateRenderer::textPos(const KateTextLine::Ptr &textLine, int xPos, uint sta
 
   KateFontStruct *fs = config()->fontStruct();
 
-  int x, oldX;
-  x = oldX = 0;
-
+  int x = 0;
+  int oldX = 0;
+  int oldZ = 0;
   uint z = startCol;
   const uint len = textLine->length();
   const TQString &textString = textLine->string();
 
   while ( (x < xPos)  && (z < len)) {
     oldX = x;
+    oldZ = z;
 
     KateAttribute* a = attribute(textLine->attribute(z));
     x += a->width(*fs, textString, z, m_tabWidth);
 
-    z++;
+    z += textString[z].isHighSurrogate() ? 2 : 1;
   }
   if ( ( (! nearest) || xPos - oldX < x - xPos ) && z > 0 ) {
-    z--;
+    z = oldZ;
    // newXPos = oldX;
   }// else newXPos = x;
   return z;

kate/part/kateviewinternal.cpp

@@ -1091,11 +1091,44 @@ public:
 
 void KateViewInternal::moveChar( Bias bias, bool sel )
 {
+  if (bias == Bias::none)
+  {
+    return;
+  }
+
+  KateTextLine::Ptr tl = m_doc->m_buffer->plainLine(cursor.line());
+  if (!tl)
+  {
+    return;
+  }
+   
+  // Make sure to handle surrogate pairs correctly
+  int offset = 0;
+  int col = cursor.col();
+  if (bias == Bias::left_b)
+  {
+    offset = -1;
+    if (tl->getChar(col - 1).isLowSurrogate() && tl->getChar(col - 2).isHighSurrogate())
+    {
+      offset = -2;
+    }
+  }
+  else
+  {
+    offset = 1;
+    if (tl->getChar(col).isHighSurrogate() && tl->getChar(col + 1).isLowSurrogate())
+    {
+      offset = 2;
+    }
+  }
+
   KateTextCursor c;
-  if ( m_view->wrapCursor() ) {
-    c = WrappingCursor( this, cursor ) += bias;
-  } else {
-    c = BoundedCursor( this, cursor ) += bias;
+  if (m_view->wrapCursor())
+  {
+    c = WrappingCursor( this, cursor ) += offset;
+  } else
+  {
+    c = BoundedCursor( this, cursor ) += offset;
   }
 
   updateSelection( c, sel );