If decoding DBCS text fails, use the MacRoman encoding to ensure something is

visible.

3 files changed, 39 insertions, 11 deletions

diff --git a/cocoa/PlatCocoa.mm b/cocoa/PlatCocoa.mm
index 1141cca44..0e1e1544c 100644
--- a/cocoa/PlatCocoa.mm
+++ b/cocoa/PlatCocoa.mm
@@ -867,10 +867,20 @@ void SurfaceImpl::DrawTextTransparent(PRectangle rc, Font &font_, XYPOSITION yba
 
 void SurfaceImpl::MeasureWidths(Font &font_, std::string_view text, XYPOSITION *positions) {
 	CFStringEncoding encoding = EncodingFromCharacterSet(unicodeMode, FontCharacterSet(font_));
-	textLayout->setText(text, encoding, *TextStyleFromFont(font_));
+	const CFStringEncoding encodingUsed =
+		textLayout->setText(text, encoding, *TextStyleFromFont(font_));
 
 	CTLineRef mLine = textLayout->getCTLine();
-	assert(mLine != NULL);
+	assert(mLine);
+
+	if (encodingUsed != encoding) {
+		// Switched to MacRoman to make work so treat as single byte encoding.
+		for (int i=0; i<text.length(); i++) {
+			CGFloat xPosition = CTLineGetOffsetForStringIndex(mLine, i+1, nullptr);
+			positions[i] = static_cast<XYPOSITION>(xPosition);
+		}
+		return;
+	}
 
 	if (unicodeMode) {
 		// Map the widths given for UTF-16 characters back onto the UTF-8 input string
diff --git a/cocoa/QuartzTextLayout.h b/cocoa/QuartzTextLayout.h
index a3ed910de..62e695525 100644
--- a/cocoa/QuartzTextLayout.h
+++ b/cocoa/QuartzTextLayout.h
@@ -23,35 +23,49 @@ public:
 	}
 
 	~QuartzTextLayout() {
-		if (mString != NULL) {
+		if (mString) {
 			CFRelease(mString);
 			mString = NULL;
 		}
-		if (mLine != NULL) {
+		if (mLine) {
 			CFRelease(mLine);
 			mLine = NULL;
 		}
 	}
 
-	void setText(std::string_view sv, CFStringEncoding encoding, const QuartzTextStyle &r) {
+	CFStringEncoding setText(std::string_view sv, CFStringEncoding encoding, const QuartzTextStyle &r) {
+		// First clear current values in case of failure.
+		if (mString) {
+			CFRelease(mString);
+			mString = NULL;
+		}
+		if (mLine) {
+			CFRelease(mLine);
+			mLine = NULL;
+		}
+
 		const UInt8 *puiBuffer = reinterpret_cast<const UInt8 *>(sv.data());
 		CFStringRef str = CFStringCreateWithBytes(NULL, puiBuffer, sv.length(), encoding, false);
-		if (!str)
-			return;
+		if (!str) {
+			// Failed to decode bytes into string with given encoding so try
+			// MacRoman which should accept any byte.
+			encoding = kCFStringEncodingMacRoman;
+			str = CFStringCreateWithBytes(NULL, puiBuffer, sv.length(), encoding, false);
+		}
+		if (!str) {
+			return encoding;
+		}
 
 		stringLength = CFStringGetLength(str);
 
 		CFMutableDictionaryRef stringAttribs = r.getCTStyle();
 
-		if (mString != NULL)
-			CFRelease(mString);
 		mString = ::CFAttributedStringCreate(NULL, str, stringAttribs);
 
-		if (mLine != NULL)
-			CFRelease(mLine);
 		mLine = ::CTLineCreateWithAttributedString(mString);
 
 		CFRelease(str);
+		return encoding;
 	}
 
 	/** Draw the text layout into a CGContext at the specified position.
diff --git a/doc/ScintillaHistory.html b/doc/ScintillaHistory.html
index bba75339b..030a6de53 100644
--- a/doc/ScintillaHistory.html
+++ b/doc/ScintillaHistory.html
@@ -552,6 +552,10 @@
 	and ensure something is shown.
 	</li>
 	<li>
+	On Cocoa, invalid text in DBCS encodings will be interpreted through the
+	single-byte MacRoman encoding as that will accept any byte.
+	</li>
+	<li>
 	Crashes fixed on macOS for invalid DBCS characters when dragging text,
 	changing case of text, case-insensitive searching, and retrieving text as UTF-8.
 	</li>