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, 4 insertions, 15 deletions

diff --git a/cocoa/PlatCocoa.mm b/cocoa/PlatCocoa.mm
index 5f9d788cd..c89a6f3aa 100644
--- a/cocoa/PlatCocoa.mm
+++ b/cocoa/PlatCocoa.mm
@@ -31,6 +31,7 @@
 
 #include "StringCopy.h"
 #include "XPM.h"
+#include "UniConversion.h"
 
 #import "ScintillaView.h"
 #import "ScintillaCocoa.h"
@@ -864,18 +865,6 @@ void SurfaceImpl::DrawTextTransparent(PRectangle rc, Font &font_, XYPOSITION yba
 	textLayout->draw(rc.left, ybase);
 }
 
-static size_t utf8LengthFromLead(unsigned char uch) {
-	if (uch >= (0x80 + 0x40 + 0x20 + 0x10)) {
-		return 4;
-	} else if (uch >= (0x80 + 0x40 + 0x20)) {
-		return 3;
-	} else if (uch >= (0x80)) {
-		return 2;
-	} else {
-		return 1;
-	}
-}
-
 //--------------------------------------------------------------------------------------------------
 
 void SurfaceImpl::MeasureWidths(Font &font_, const char *s, int len, XYPOSITION *positions) {
@@ -892,10 +881,10 @@ void SurfaceImpl::MeasureWidths(Font &font_, const char *s, int len, XYPOSITION
 		const unsigned char *us = reinterpret_cast<const unsigned char *>(s);
 		int i=0;
 		while (ui<fit) {
-			size_t lenChar = utf8LengthFromLead(us[i]);
-			size_t codeUnits = (lenChar < 4) ? 1 : 2;
+			const unsigned int byteCount = UTF8BytesOfLead[us[i]];
+			const int codeUnits = UTF16LengthFromUTF8ByteCount(byteCount);
 			CGFloat xPosition = CTLineGetOffsetForStringIndex(mLine, ui+codeUnits, NULL);
-			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>(xPosition);
 			}
 			ui += codeUnits;