Fix editing of text containing surrogate characters.

This relates to issue #162.

Signed-off-by: Michele Calgaro <[email protected]>

4 files changed, 59 insertions, 16 deletions

diff --git a/src/kernel/tqrichtext.cpp b/src/kernel/tqrichtext.cpp
index 3585e470d..98fbc8f53 100644
--- a/src/kernel/tqrichtext.cpp
+++ b/src/kernel/tqrichtext.cpp
@@ -219,12 +219,19 @@ TQTextCursor *TQTextDeleteCommand::execute( TQTextCursor *c )
     cursor.setParagraph( s );
     cursor.setIndex( index );
     int len = text.size();
+    
     if ( c )
 	*c = cursor;
     if ( doc ) {
 	doc->setSelectionStart( TQTextDocument::Temp, cursor );
 	for ( int i = 0; i < len; ++i )
+	{
+	    if (text[i].c.isHighSurrogate() && i < (len - 1) && text[i + 1].c.isLowSurrogate())
+	    {
+		++i; // This is required for correct advancement when handling surrogate pairs
+	    }
 	    cursor.gotoNextLetter();
+	}
 	doc->setSelectionEnd( TQTextDocument::Temp, cursor );
 	doc->removeSelectedText( TQTextDocument::Temp, &cursor );
 	if ( c )
@@ -1248,9 +1255,17 @@ bool TQTextCursor::removePreviousChar()
 {
     tmpX = -1;
     if ( !atParagStart() ) {
-	para->remove( idx-1, 1 );
+	if (para->at(idx - 1)->c.isLowSurrogate() && idx > 1 && para->at(idx - 2)->c.isHighSurrogate())
+	{
+	    para->remove(idx - 2, 2);
+	    idx -= 2;
+	}
+	else
+	{
+	    para->remove(idx - 1, 1);
+	    idx--;
+	}
 	int h = para->rect().height();
-	idx--;
 	// shouldn't be needed, just to make sure.
 	fixCursorPosition();
 	para->format( -1, TRUE );
@@ -4007,6 +4022,7 @@ int TQTextString::width( int idx ) const
 {
      int w = 0;
      TQTextStringChar *c = &at( idx );
+     // '!c->charStop' already takes care of low surrogate chars
      if ( !c->charStop || c->c.unicode() == 0xad || c->c.unicode() == 0x2028 )
 	 return 0;
 #ifndef TQT_NO_TEXTCUSTOMITEM
@@ -4290,10 +4306,10 @@ void TQTextParagraph::format( int start, bool doMove )
     if ( !str || str->length() == 0 || !formatter() )
 	return;
 
-    if ( hasdoc &&
-	 document()->preProcessor() &&
-	 ( needPreProcess || state == -1 ) )
+    if ( hasdoc && document()->preProcessor() && ( needPreProcess || state == -1 ) )
+    {
 	document()->preProcessor()->process( document(), this, invalid <= 0 ? 0 : invalid );
+    }
     needPreProcess = FALSE;
 
     if ( invalid == -1 )
@@ -5782,6 +5798,7 @@ int TQTextFormatterBreakWords::format( TQTextDocument *doc, TQTextParagraph *par
 	    lastChr = c->c;
 	    lastFormat = c->format();
 	}
+
 	bool lastWasOwnLineCustomItem = lastBreak == -2;
 	bool hadBreakableChar = lastBreak != -1;
 	bool lastWasHardBreak = lastChr == TQChar_linesep;
@@ -5791,6 +5808,12 @@ int TQTextFormatterBreakWords::format( TQTextDocument *doc, TQTextParagraph *par
 	    c->format()->setPainter( painter );
 	c = &string->at( i );
 
+	// Skip over low surrogate chars
+	if (c->c.isLowSurrogate())
+	{
+	    continue;
+	}
+
 	if (lastFormat != c->format() && !c->c.isSpace()
 	    && lastFormat->font().italic() && !c->format()->font().italic()) {
 	    int rb = lastFormat->fontMetrics().rightBearing(lastChr);
diff --git a/src/kernel/tqrichtext_p.cpp b/src/kernel/tqrichtext_p.cpp
index e9417d2a9..165d319db 100644
--- a/src/kernel/tqrichtext_p.cpp
+++ b/src/kernel/tqrichtext_p.cpp
@@ -114,15 +114,15 @@ int TQTextCursor::x() const
 	return 0;
     TQTextStringChar *c = para->at( idx );
     int curx = c->x;
-    if ( !c->rightToLeft &&
-	 c->c.isSpace() &&
-	 idx > 0 &&
-	 para->at( idx - 1 )->c != '\t' &&
-	 !c->lineStart &&
-	 ( para->alignment() & TQt::AlignJustify ) == TQt::AlignJustify )
+    if ( !c->rightToLeft && c->c.isSpace() && idx > 0 && para->at( idx - 1 )->c != '\t' &&
+	 !c->lineStart && ( para->alignment() & TQt::AlignJustify ) == TQt::AlignJustify )
+    {
 	curx = para->at( idx - 1 )->x + para->string()->width( idx - 1 );
+    }
     if ( c->rightToLeft )
+    {
 	curx += para->string()->width( idx );
+    }
     return curx;
 }
 
diff --git a/src/widgets/tqlineedit.cpp b/src/widgets/tqlineedit.cpp
index 2267706d4..94cc54bc8 100644
--- a/src/widgets/tqlineedit.cpp
+++ b/src/widgets/tqlineedit.cpp
@@ -860,10 +860,18 @@ void TQLineEdit::backspace()
     if ( d->hasSelectedText() ) {
 	d->removeSelectedText();
     } else if ( d->cursor ) {
+	--d->cursor;
+	if ( d->maskData ) {
+	    d->cursor = d->prevMaskBlank( d->cursor );
+	}
+	// second half of a surrogate, check if we have the first half as well,
+	// if yes delete both at once
+	if (d->cursor > 0 && d->text.at(d->cursor).isLowSurrogate() &&
+	        d->text.at(d->cursor - 1).isHighSurrogate()) {
+	    d->del(true);
 	    --d->cursor;
-	    if ( d->maskData )
-		d->cursor = d->prevMaskBlank( d->cursor );
-	    d->del( TRUE );
+	}
+	d->del(true);
     }
     d->finishChange( priorState );
 }
diff --git a/src/widgets/tqtextedit.cpp b/src/widgets/tqtextedit.cpp
index cc88b3dd9..7ea000f21 100644
--- a/src/widgets/tqtextedit.cpp
+++ b/src/widgets/tqtextedit.cpp
@@ -1782,8 +1782,20 @@ void TQTextEdit::doKeyboardAction( KeyboardAction action )
 		    undoRedoInfo.index = cursor->index();
 		    undoRedoInfo.d->text = TQString::null;
 		}
-		undoRedoInfo.d->text.insert( 0, cursor->paragraph()->at( cursor->index()-1 ), TRUE );
-		undoRedoInfo.index = cursor->index()-1;
+		int cur_idx = cursor->index();
+		const TQTextParagraph *para = cursor->paragraph();
+		if (para->at(cur_idx - 1)->c.isLowSurrogate() && cur_idx > 1 &&
+		       para->at(cur_idx - 2)->c.isHighSurrogate())
+		{
+		    undoRedoInfo.d->text.insert( 0, para->at(cur_idx - 1), TRUE );
+		    undoRedoInfo.d->text.insert( 0, para->at(cur_idx - 2), TRUE );
+		    undoRedoInfo.index = cursor->index() - 2;
+		}
+		else
+		{
+		    undoRedoInfo.d->text.insert( 0, para->at(cur_idx - 1), TRUE );
+		    undoRedoInfo.index = cursor->index() - 1;
+		}
 	    }
 	    cursor->removePreviousChar();
 	    lastFormatted = cursor->paragraph();