Feature [feature-requests:#1211]. Use pre-computed table for UTF8BytesOfLead.

Friendlier treatment of invalid UTF-8.
Add tests for UniConversion handling invalid UTF-8. Simplify UTF8Classify tests.

1 files changed, 8 insertions, 11 deletions

diff --git a/win32/PlatWin.cxx b/win32/PlatWin.cxx
index 9e89e2f84..79970a969 100644
--- a/win32/PlatWin.cxx
+++ b/win32/PlatWin.cxx
@@ -951,12 +951,14 @@ void SurfaceGDI::MeasureWidths(Font &font_, const char *s, int len, XYPOSITION *
 			return;
 		}
 		// Map the widths given for UTF-16 characters back onto the UTF-8 input string
+		const unsigned char *us = reinterpret_cast<const unsigned char *>(s);
 		for (int ui = 0; ui < fit; ui++) {
-			const unsigned int lenChar = UTF8BytesOfLead[static_cast<unsigned char>(s[i])];
-			if (lenChar == 4) {	// Non-BMP
+			const unsigned char uch = us[i];
+			const unsigned int byteCount = UTF8BytesOfLead[uch];
+			if (byteCount == 4) {	// Non-BMP
 				ui++;
 			}
-			for (unsigned int bytePos=0; (bytePos<lenChar) && (i<len); bytePos++) {
+			for (unsigned int bytePos=0; (bytePos<byteCount) && (i<len); bytePos++) {
 				positions[i++] = static_cast<XYPOSITION>(poses.buffer[ui]);
 			}
 		}
@@ -1623,16 +1625,11 @@ void SurfaceD2D::MeasureWidths(Font &font_, const char *s, int len, XYPOSITION *
 		int i=0;
 		while (ui<tbuf.tlen) {
 			const unsigned char uch = us[i];
-			unsigned int lenChar = 1;
-			if (uch >= (0x80 + 0x40 + 0x20 + 0x10)) {
-				lenChar = 4;
+			const unsigned int byteCount = UTF8BytesOfLead[uch];
+			if (byteCount == 4) {	// Non-BMP
 				ui++;
-			} else if (uch >= (0x80 + 0x40 + 0x20)) {
-				lenChar = 3;
-			} else if (uch >= (0x80)) {
-				lenChar = 2;
 			}
-			for (unsigned int bytePos=0; (bytePos<lenChar) && (i<len); bytePos++) {
+			for (unsigned int bytePos=0; (bytePos<byteCount) && (i<len); bytePos++) {
 				positions[i++] = poses.buffer[ui];
 			}
 			ui++;