Fix editing of text containing surrogate characters.fix/unicode-editing

This relates to issue #162.

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

3 files changed, 83 insertions, 59 deletions

diff --git a/src/kernel/tqscriptengine.cpp b/src/kernel/tqscriptengine.cpp
index 588596716..ff05d123b 100644
--- a/src/kernel/tqscriptengine.cpp
+++ b/src/kernel/tqscriptengine.cpp
@@ -288,10 +288,21 @@ static void heuristicSetGlyphAttributes(TQShaperItem *item, const TQChar *uc, in
     Q_ASSERT(item->num_glyphs <= length);
 
     unsigned short *logClusters = item->log_clusters;
-
-    int i;
-    for (i = 0; i < length; ++i)
-        logClusters[i] = i;
+    int glyph_pos = 0;
+    for (int i = 0; i < length; i++)
+    {
+	if (uc[i].isHighSurrogate() && i < (length - 1) && uc[i+1].isLowSurrogate())
+	{
+	    logClusters[i] = glyph_pos;
+	    logClusters[++i] = glyph_pos;
+	}
+	else
+	{
+	    logClusters[i] = glyph_pos;
+	}
+	++glyph_pos;
+    }
+    Q_ASSERT(glyph_pos == item->num_glyphs);
 
     // first char in a run is never (treated as) a mark
     int cStart = 0;
@@ -307,7 +318,7 @@ static void heuristicSetGlyphAttributes(TQShaperItem *item, const TQChar *uc, in
     }
 
     int lastCat = ::category(uc[0]);
-    for (i = 1; i < length; ++i) {
+    for (int i = 1; i < length; ++i) {
         int cat = ::category(uc[i]);
         if (qIsZeroWidthChar(uc[i].unicode())) {
 	    item->attributes[i].mark = FALSE;
diff --git a/src/kernel/tqtextengine.cpp b/src/kernel/tqtextengine.cpp
index 05cdbcc13..5fbf306ab 100644
--- a/src/kernel/tqtextengine.cpp
+++ b/src/kernel/tqtextengine.cpp
@@ -800,72 +800,77 @@ static void calcLineBreaks(const TQString &str, TQCharAttributes *charAttributes
 {
     int len = str.length();
     if (!len)
-        return;
+	return;
 
     const TQChar *uc = str.unicode();
     int cls = lineBreakClass(*uc);
     if (cls >= TQUnicodeTables::LineBreak_CM)
-        cls = TQUnicodeTables::LineBreak_ID;
+	cls = TQUnicodeTables::LineBreak_ID;
 
     charAttributes[0].softBreak = FALSE;
     charAttributes[0].whiteSpace = (cls == TQUnicodeTables::LineBreak_SP);
     charAttributes[0].charStop = TRUE;
 
-    for (int i = 1; i < len; ++i) {
-        int ncls = ::lineBreakClass(uc[i]);
-        int category = ::category(uc[i]);
-        if (category == TQChar::Mark_NonSpacing)
-            goto nsm;
-        
-        if (category == TQChar::Other_Surrogate) {
-            // char stop only on first pair
-            if (uc[i].isHighSurrogate() && i < (len - 1) && uc[i + 1].isLowSurrogate())
-                goto nsm;
-            // ### correctly handle second surrogate
-        }
-
-        if (ncls == TQUnicodeTables::LineBreak_SP) {
-            charAttributes[i].softBreak = FALSE;
-            charAttributes[i].whiteSpace = TRUE;
-            charAttributes[i].charStop = TRUE;
-            cls = ncls;
-            continue;
-        }
+    bool prevIsHighSurrogate = uc[0].isHighSurrogate();
+    for (int i = 1; i < len; ++i)
+    {
+	// Don't stop on low surrogate characters of complete valid pairs
+	if (prevIsHighSurrogate && uc[i].isLowSurrogate())
+	{
+	    prevIsHighSurrogate = false;
+	    charAttributes[i].softBreak = FALSE;
+	    charAttributes[i].whiteSpace = FALSE;
+	    charAttributes[i].charStop = FALSE;
+	    continue;
+	}
+	prevIsHighSurrogate = uc[i].isHighSurrogate();
 
+	int ncls = ::lineBreakClass(uc[i]);
+	int category = ::category(uc[i]);
+	if (category == TQChar::Mark_NonSpacing)
+	{
+	    charAttributes[i].softBreak = FALSE;
+	    charAttributes[i].whiteSpace = FALSE;
+	    charAttributes[i].charStop = FALSE;
+	    continue;
+	}
 
-	if (cls == TQUnicodeTables::LineBreak_SA && ncls == TQUnicodeTables::LineBreak_SA) {
-            // two complex chars (thai or lao), thai_attributes might override, but here
-            // we do a best guess
-            charAttributes[i].softBreak = TRUE;
-            charAttributes[i].whiteSpace = FALSE;
-            charAttributes[i].charStop = TRUE;
-            cls = ncls;
-            continue;
-        }
-        {
-	    int tcls = ncls;
-	    if (tcls >= TQUnicodeTables::LineBreak_SA)
-		tcls = TQUnicodeTables::LineBreak_ID;
-	    if (cls >= TQUnicodeTables::LineBreak_SA)
-		cls = TQUnicodeTables::LineBreak_ID;
-
-	    bool softBreak;
-	    int brk = breakTable[cls][tcls];
-	    if (brk == Ibk)
-		softBreak = (cls == TQUnicodeTables::LineBreak_SP);
-	    else
-		softBreak = (brk == Dbk);
-//        tqDebug("char = %c %04x, cls=%d, ncls=%d, brk=%d soft=%d", uc[i].cell(), uc[i].unicode(), cls, ncls, brk, charAttributes[i].softBreak);
-	    charAttributes[i].softBreak = softBreak;
+	if (ncls == TQUnicodeTables::LineBreak_SP) {
+	    charAttributes[i].softBreak = FALSE;
+	    charAttributes[i].whiteSpace = TRUE;
+	    charAttributes[i].charStop = TRUE;
+	    cls = ncls;
+	    continue;
+	}
+
+	if (cls == TQUnicodeTables::LineBreak_SA && ncls == TQUnicodeTables::LineBreak_SA)
+	{
+	    // two complex chars (thai or lao), thai_attributes might override, but here
+	    // we do a best guess
+	    charAttributes[i].softBreak = TRUE;
 	    charAttributes[i].whiteSpace = FALSE;
 	    charAttributes[i].charStop = TRUE;
 	    cls = ncls;
+	    continue;
 	}
-        continue;
-    nsm:
-        charAttributes[i].softBreak = FALSE;
-        charAttributes[i].whiteSpace = FALSE;
-        charAttributes[i].charStop = FALSE;
+
+	int tcls = ncls;
+	if (tcls >= TQUnicodeTables::LineBreak_SA)
+	    tcls = TQUnicodeTables::LineBreak_ID;
+	if (cls >= TQUnicodeTables::LineBreak_SA)
+	    cls = TQUnicodeTables::LineBreak_ID;
+
+	bool softBreak;
+	int brk = breakTable[cls][tcls];
+	if (brk == Ibk)
+	    softBreak = (cls == TQUnicodeTables::LineBreak_SP);
+	else
+	    softBreak = (brk == Dbk);
+  //        tqDebug("char = %c %04x, cls=%d, ncls=%d, brk=%d soft=%d", uc[i].cell(), uc[i].unicode(), cls, ncls, brk, charAttributes[i].softBreak);
+	charAttributes[i].softBreak = softBreak;
+	charAttributes[i].whiteSpace = FALSE;
+	charAttributes[i].charStop = TRUE;
+	cls = ncls;
     }
 }
 
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 );
 }