Add support for surrogate pairs to TQChar API.

This relates to issue #162.
The new code is partially taken from Qt4 but with some local rework.

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

1 files changed, 9 insertions, 14 deletions

diff --git a/src/codecs/tqutfcodec.cpp b/src/codecs/tqutfcodec.cpp
index 1125aa9f3..eba25e505 100644
--- a/src/codecs/tqutfcodec.cpp
+++ b/src/codecs/tqutfcodec.cpp
@@ -64,13 +64,10 @@ TQCString TQUtf8Codec::fromUnicode(const TQString& uc, int& lenInOut) const
  	    if ( u < 0x0800 ) {
 		*cursor++ = 0xc0 | ((uchar) (u >> 6));
  	    } else {
-		if (u >= 0xd800 && u < 0xdc00 && i < l-1) {
-		    unsigned short low = ch[1].unicode();
-		    if (low >= 0xdc00 && low < 0xe000) {
-			++ch;
-			++i;
-			u = (u - 0xd800)*0x400 + (low - 0xdc00) + 0x10000;
-		    }
+		if (ch[0].isHighSurrogate() && i < (l - 1) && ch[1].isLowSurrogate()) {
+		    u = TQChar::surrogateToUcs4(ch[0], ch[1]);
+		    ++ch;
+		    ++i;
 		}
 		if (u > 0xffff) {
 		    // see TQString::fromUtf8() and TQString::utf8() for explanations
@@ -179,16 +176,14 @@ public:
 		    uc = (uc << 6) | (ch & 0x3f);
 		    need--;
 		    if ( !need ) {
-			if (uc > 0xffff) {
+			if (TQChar::requiresSurrogates(uc)) {
 			    // surrogate pair
-			    uc -= 0x10000;
-			    unsigned short high = uc/0x400 + 0xd800;
-			    unsigned short low = uc%0x400 + 0xdc00;
-			    *qch++ = TQChar(high);
-			    *qch++ = TQChar(low);
+			    *qch++ = TQChar(TQChar::highSurrogate(uc));
+			    *qch++ = TQChar(TQChar::lowSurrogate(uc));
 			    headerDone = TRUE;
 			} else if ((uc < min_uc) || (uc >= 0xd800 && uc <= 0xdfff) || (uc >= 0xfffe)) {
-                            *qch++ = TQChar::replacement;
+			    // overlong sequence, UTF16 surrogate or BOM
+			    *qch++ = TQChar::replacement;
 			} else {
 			    if (headerDone || TQChar(uc) != TQChar::byteOrderMark)
 				*qch++ = uc;