Fix doubling spaces on wrapped strings

Author: spoonmilk

binaryninjaapi.h

@@ -14275,6 +14275,7 @@ namespace BinaryNinja {
 		size_t minimumContentLength;
 		size_t tabWidth;
 		size_t maximumAnnotationLength;
+		size_t stringWrappingWidth;
 		std::string languageName;
 		std::string commentStartString;
 		std::string commentEndString;

binaryninjacore.h

@@ -3670,6 +3670,7 @@ extern "C"
 		size_t minimumContentLength;
 		size_t tabWidth;
 		size_t maximumAnnotationLength;
+		size_t stringWrappingWidth;
 		char* languageName;
 		char* commentStartString;
 		char* commentEndString;

formatter/generic/genericformatter.cpp

@@ -203,6 +203,8 @@ struct ItemLayoutStackEntry
 	size_t additionalContinuationIndentation;
 	size_t desiredWidth;
 	size_t desiredContinuationWidth;
+	size_t desiredStringWidth;
+	size_t desiredStringContinuationWidth;
 	bool newLineOnReenteringScope;
 };
 
@@ -318,7 +320,18 @@ static vector<InstructionTextToken> ParseStringToken(
             ConstructToken(start, curEnd);
             curStart = curEnd;
         }
-        else if (c == ',' || c == '.' || c == ':' || c == ';' || isspace(c))
+    	else if (isspace(c))
+    	{
+    		// Flush before whitespace
+    		flushToken(curStart, curEnd);
+
+    		size_t start = curEnd;
+    		while (curEnd < tail && isspace(src[curEnd]))
+    			curEnd++;
+    		ConstructToken(start, curEnd);
+    		curStart = curEnd;
+    	}
+        else if (c == ',' || c == '.' || c == ':' || c == ';')
         {
         	// Flush before punctuation
         	flushToken(curStart, curEnd);
@@ -731,6 +744,24 @@ vector<DisassemblyTextLine> GenericLineFormatter::FormatLines(
 				desiredContinuationWidth = remainingWidth;
 		}
 
+		// Compute target string width for this line
+		size_t desiredStringWidth = settings.stringWrappingWidth;
+		if (indentation < settings.desiredLineLength)
+		{
+			size_t remainingStringWidth = desiredStringWidth - indentation;
+			if (remainingStringWidth > desiredStringWidth)
+				desiredStringWidth = remainingStringWidth;
+		}
+
+		// Compute target width for continuation string wrapping lines
+		size_t desiredStringContinuationWidth = settings.stringWrappingWidth;
+		if (continuationIndentation < settings.desiredLineLength)
+		{
+			size_t remainingStringWidth = desiredStringContinuationWidth - continuationIndentation;
+			if (remainingStringWidth > desiredStringContinuationWidth)
+				desiredStringContinuationWidth = remainingStringWidth;
+		}
+
 		// Gather the indentation tokens at the beginning of the line
 		vector<InstructionTextToken> indentationTokens = currentLine.GetAddressAndIndentationTokens();
 		size_t tokenIndex = indentationTokens.size();
@@ -831,7 +862,7 @@ vector<DisassemblyTextLine> GenericLineFormatter::FormatLines(
 					string trimmedSubText = TrimString(subToken.text);
 					if (trimmedSubText.empty())
 						items.push_back(Item {StringWhitespace, {}, {subToken}, 0});
-					if (trimmedSubText[0] == '%')
+					else if (trimmedSubText[0] == '%')
 						items.push_back(Item {FormatSpecifier, {}, {subToken}, 0});
 					else if (!trimmedSubText.empty() && trimmedSubText[0] == '\\')
 						items.push_back(Item {EscapeSequence, {}, {subToken}, 0});
@@ -892,7 +923,7 @@ vector<DisassemblyTextLine> GenericLineFormatter::FormatLines(
 		bool firstTokenOfLine = true;
 
 		stack<ItemLayoutStackEntry> layoutStack;
-		layoutStack.push({items, additionalContinuationIndentation, desiredWidth, desiredContinuationWidth, false});
+		layoutStack.push({items, additionalContinuationIndentation, desiredWidth, desiredContinuationWidth, desiredStringWidth, desiredStringContinuationWidth, false});
 
 		auto newLine = [&]() {
 			if (!firstTokenOfLine)
@@ -917,6 +948,7 @@ vector<DisassemblyTextLine> GenericLineFormatter::FormatLines(
 			outputLine.tokens.emplace_back(TextToken, string(additionalContinuationIndentation, ' '));
 			currentWidth = 0;
 			desiredWidth = desiredContinuationWidth;
+			desiredStringWidth = desiredStringContinuationWidth;
 			firstTokenOfLine = true;
 		};
 
@@ -929,6 +961,8 @@ vector<DisassemblyTextLine> GenericLineFormatter::FormatLines(
 			additionalContinuationIndentation = layoutStackEntry.additionalContinuationIndentation;
 			desiredWidth = layoutStackEntry.desiredWidth;
 			desiredContinuationWidth = layoutStackEntry.desiredContinuationWidth;
+			desiredStringWidth = layoutStackEntry.desiredStringWidth;
+			desiredStringContinuationWidth = layoutStackEntry.desiredStringContinuationWidth;
 
 			// Check to see if the scope we are returning to needs a new line. This is used when an argument
 			// spans multiple lines. The rest of the arguments are placed on separate lines from the long argument.
@@ -937,13 +971,14 @@ vector<DisassemblyTextLine> GenericLineFormatter::FormatLines(
 
 			for (auto item = items.begin(); item != items.end();)
 			{
-				if (item->type == StringComponent && currentWidth + item->width > desiredWidth)
+				if (item->type == StringComponent && currentWidth + item->width > desiredStringWidth)
 				{
 					// If a string is too wide to fit on the current line, create a newline
 					// without additional indentation
 					newLine();
+					continue;
 				}
-				else if (currentWidth + item->width > desiredWidth && item->type != StringWhitespace)
+				if (currentWidth + item->width > desiredWidth && item->type != StringWhitespace)
 				{
 					// Current item is too wide to fit on the current line, will need to start a new line.
 					// Whitespace is allowed to be too wide; we push it on as the preceding word is wrapped.
@@ -962,7 +997,7 @@ vector<DisassemblyTextLine> GenericLineFormatter::FormatLines(
 							if (next != items.end())
 							{
 								layoutStack.push({vector(next, items.end()), additionalContinuationIndentation,
-									desiredWidth, desiredContinuationWidth, true});
+									desiredWidth, desiredContinuationWidth, desiredStringWidth, desiredStringContinuationWidth, true});
 							}
 
 							newLine();
@@ -973,8 +1008,13 @@ vector<DisassemblyTextLine> GenericLineFormatter::FormatLines(
 							else
 								desiredContinuationWidth -= settings.tabWidth;
 
+							if (desiredStringContinuationWidth < settings.minimumContentLength + settings.tabWidth)
+								desiredStringContinuationWidth = settings.minimumContentLength;
+							else
+								desiredStringContinuationWidth -= settings.tabWidth;
+
 							layoutStack.push({item->items, additionalContinuationIndentation, desiredWidth,
-								desiredContinuationWidth, false});
+								desiredContinuationWidth, desiredStringWidth, desiredStringContinuationWidth, false});
 							break;
 						}
 
@@ -984,10 +1024,10 @@ vector<DisassemblyTextLine> GenericLineFormatter::FormatLines(
 							if (next != items.end())
 							{
 								layoutStack.push({vector(next, items.end()), additionalContinuationIndentation,
-									desiredWidth, desiredContinuationWidth, false});
+									desiredWidth, desiredContinuationWidth, desiredStringWidth, desiredStringContinuationWidth, false});
 							}
 							layoutStack.push({item->items, additionalContinuationIndentation, desiredWidth,
-								desiredContinuationWidth, false});
+								desiredContinuationWidth, desiredStringWidth, desiredStringContinuationWidth, false});
 							break;
 						}
 

lineformatter.cpp

@@ -54,6 +54,7 @@ LineFormatterSettings LineFormatterSettings::FromAPIObject(const BNLineFormatter
 	result.minimumContentLength = settings->minimumContentLength;
 	result.tabWidth = settings->tabWidth;
 	result.maximumAnnotationLength = settings->maximumAnnotationLength;
+	result.stringWrappingWidth = settings->stringWrappingWidth;
 	result.languageName = settings->languageName;
 	result.commentStartString = settings->commentStartString;
 	result.commentEndString = settings->commentEndString;
@@ -71,6 +72,7 @@ BNLineFormatterSettings LineFormatterSettings::ToAPIObject() const
 	result.minimumContentLength = minimumContentLength;
 	result.tabWidth = tabWidth;
 	result.maximumAnnotationLength = maximumAnnotationLength;
+	result.stringWrappingWidth = stringWrappingWidth;
 	result.languageName = (char*)languageName.c_str();
 	result.commentStartString = (char*)commentStartString.c_str();
 	result.commentEndString = (char*)commentEndString.c_str();