Add line formatter API and a generic line formatter plugin

Author: D0ntPanic

basicblock.cpp

@@ -36,6 +36,24 @@ DisassemblySettings::DisassemblySettings(BNDisassemblySettings* settings)
 }
 
 
+Ref<DisassemblySettings> DisassemblySettings::GetDefaultSettings()
+{
+	return new DisassemblySettings(BNDefaultDisassemblySettings());
+}
+
+
+Ref<DisassemblySettings> DisassemblySettings::GetDefaultGraphSettings()
+{
+	return new DisassemblySettings(BNDefaultGraphDisassemblySettings());
+}
+
+
+Ref<DisassemblySettings> DisassemblySettings::GetDefaultLinearSettings()
+{
+	return new DisassemblySettings(BNDefaultLinearDisassemblySettings());
+}
+
+
 DisassemblySettings* DisassemblySettings::Duplicate()
 {
 	return new DisassemblySettings(BNDuplicateDisassemblySettings(m_object));
@@ -144,6 +162,73 @@ DisassemblyTextLine::DisassemblyTextLine()
 }
 
 
+size_t DisassemblyTextLine::GetTotalWidth() const
+{
+	size_t result = 0;
+	for (auto& i : tokens)
+		result += i.width;
+	return result;
+}
+
+
+static void FindAddressAndIndentationTokens(
+	const vector<InstructionTextToken>& tokens, const std::function<void(const InstructionTextToken&)>& callback)
+{
+	size_t startToken = 0;
+	for (size_t i = 0; i < tokens.size(); i++)
+	{
+		if (tokens[i].type == AddressSeparatorToken)
+		{
+			startToken = i + 1;
+			break;
+		}
+	}
+
+	for (size_t i = 0; i < startToken; i++)
+		callback(tokens[i]);
+	for (size_t i = startToken; i < tokens.size(); i++)
+	{
+		if (tokens[i].type == AddressDisplayToken || tokens[i].type == AddressSeparatorToken
+			|| tokens[i].type == CollapseStateIndicatorToken)
+		{
+			callback(tokens[i]);
+			continue;
+		}
+
+		bool whitespace = true;
+		for (auto ch : tokens[i].text)
+		{
+			if (!isspace(ch))
+			{
+				whitespace = false;
+				break;
+			}
+		}
+
+		if (!whitespace)
+			break;
+
+		callback(tokens[i]);
+	}
+}
+
+
+size_t DisassemblyTextLine::GetAddressAndIndentationWidth() const
+{
+	size_t result = 0;
+	FindAddressAndIndentationTokens(tokens, [&](const InstructionTextToken& token) { result += token.width; });
+	return result;
+}
+
+
+vector<InstructionTextToken> DisassemblyTextLine::GetAddressAndIndentationTokens() const
+{
+	vector<InstructionTextToken> result;
+	FindAddressAndIndentationTokens(tokens, [&](const InstructionTextToken& token) { result.push_back(token); });
+	return result;
+}
+
+
 BasicBlock::BasicBlock(BNBasicBlock* block)
 {
 	m_object = block;

binaryninjaapi.h

@@ -4057,6 +4057,10 @@ namespace BinaryNinja {
 		DisassemblyTextLineTypeInfo typeInfo;
 
 		DisassemblyTextLine();
+
+		size_t GetTotalWidth() const;
+		size_t GetAddressAndIndentationWidth() const;
+		std::vector<InstructionTextToken> GetAddressAndIndentationTokens() const;
 	};
 
 	/*!
@@ -10175,6 +10179,10 @@ namespace BinaryNinja {
 		DisassemblySettings(BNDisassemblySettings* settings);
 		DisassemblySettings* Duplicate();
 
+		static Ref<DisassemblySettings> GetDefaultSettings();
+		static Ref<DisassemblySettings> GetDefaultGraphSettings();
+		static Ref<DisassemblySettings> GetDefaultLinearSettings();
+
 		bool IsOptionSet(BNDisassemblyOption option) const;
 		void SetOption(BNDisassemblyOption option, bool state = true);
 
@@ -13698,6 +13706,82 @@ namespace BinaryNinja {
 		std::set<SSAVariable> GetSSAVariables();
 	};
 
+	struct LineFormatterSettings
+	{
+		Ref<HighLevelILFunction> highLevelIL;
+		size_t desiredLineLength;
+		size_t minimumContentLength;
+		size_t tabWidth;
+		std::string languageName;
+		std::string commentStartString;
+		std::string commentEndString;
+		std::string annotationStartString;
+		std::string annotationEndString;
+
+		/*! Gets the default line formatter settings for High Level IL code.
+
+		    \param settings The settings for reformatting.
+		    \param func High Level IL function to be reformatted.
+		    \return Settings for reformatting.
+		*/
+		static LineFormatterSettings GetDefault(DisassemblySettings* settings, HighLevelILFunction* func);
+
+		/*! Gets the default line formatter settings for a language representation function.
+
+		    \param settings The settings for reformatting.
+		    \param func Language representation function to be reformatted.
+		    \return Settings for reformatting.
+		*/
+		static LineFormatterSettings GetLanguageRepresentationSettings(
+			DisassemblySettings* settings, LanguageRepresentationFunction* func);
+
+		static LineFormatterSettings FromAPIObject(const BNLineFormatterSettings* settings);
+		BNLineFormatterSettings ToAPIObject() const;
+	};
+
+	class LineFormatter : public StaticCoreRefCountObject<BNLineFormatter>
+	{
+		std::string m_nameForRegister;
+
+		static BNDisassemblyTextLine* FormatLinesCallback(void* ctxt, BNDisassemblyTextLine* inLines, size_t inCount,
+			const BNLineFormatterSettings* settings, size_t* outCount);
+		static void FreeLinesCallback(void* ctxt, BNDisassemblyTextLine* lines, size_t count);
+
+	public:
+		LineFormatter(const std::string& name);
+		LineFormatter(BNLineFormatter* formatter);
+
+		/*! Registers the line formatter.
+
+		    \param formatter The line formatter to register.
+		*/
+		static void Register(LineFormatter* formatter);
+
+		static std::vector<Ref<LineFormatter>> GetList();
+		static Ref<LineFormatter> GetByName(const std::string& name);
+		static Ref<LineFormatter> GetDefault();
+
+		/*! Reformats the given list of lines. Returns a new list of lines containing the reformatted code.
+
+		    \param lines The lines to reformat.
+		    \param settings The settings for reformatting.
+		    \return A new list of reformatted lines.
+		*/
+		virtual std::vector<DisassemblyTextLine> FormatLines(
+			const std::vector<DisassemblyTextLine>& lines, const LineFormatterSettings& settings) = 0;
+	};
+
+	class CoreLineFormatter : public LineFormatter
+	{
+	public:
+		CoreLineFormatter(BNLineFormatter* formatter);
+
+		std::vector<DisassemblyTextLine> FormatLines(
+			const std::vector<DisassemblyTextLine>& lines, const LineFormatterSettings& settings) override;
+	};
+
+	class LanguageRepresentationFunctionType;
+
 	/*! LanguageRepresentationFunction represents a single function in a registered high level language.
 
 	    \ingroup highlevelil
@@ -13707,7 +13791,8 @@ namespace BinaryNinja {
 	        BNFreeLanguageRepresentationFunction>
 	{
 	public:
-		LanguageRepresentationFunction(Architecture* arch, Function* func, HighLevelILFunction* highLevelIL);
+		LanguageRepresentationFunction(LanguageRepresentationFunctionType* type, Architecture* arch, Function* func,
+			HighLevelILFunction* highLevelIL);
 		LanguageRepresentationFunction(BNLanguageRepresentationFunction* func);
 
 		/*! Gets the lines of tokens for a given High Level IL instruction.
@@ -13746,6 +13831,7 @@ namespace BinaryNinja {
 		*/
 		BNHighlightColor GetHighlight(BasicBlock* block);
 
+		Ref<LanguageRepresentationFunctionType> GetLanguage() const;
 		Ref<Architecture> GetArchitecture() const;
 		Ref<Function> GetFunction() const;
 		Ref<HighLevelILFunction> GetHighLevelILFunction() const;
@@ -13894,6 +13980,13 @@ namespace BinaryNinja {
 		*/
 		virtual Ref<TypeParser> GetTypeParser() { return nullptr; }
 
+		/*! Returns the line formatter for formatting code in this language. If NULL is returned, the default
+		    formatter will be used.
+
+		    \return The optional formatter for formatting code in this language.
+		*/
+		virtual Ref<LineFormatter> GetLineFormatter() { return nullptr; }
+
 		/*! Returns a list of lines representing a function prototype in this language. If no lines are returned, the
 		    default C-style prototype will be used.
 
@@ -13920,6 +14013,7 @@ namespace BinaryNinja {
 		static bool IsValidCallback(void* ctxt, BNBinaryView* view);
 		static BNTypePrinter* GetTypePrinterCallback(void* ctxt);
 		static BNTypeParser* GetTypeParserCallback(void* ctxt);
+		static BNLineFormatter* GetLineFormatterCallback(void* ctxt);
 		static BNDisassemblyTextLine* GetFunctionTypeTokensCallback(
 			void* ctxt, BNFunction* func, BNDisassemblySettings* settings, size_t* count);
 		static void FreeLinesCallback(void* ctxt, BNDisassemblyTextLine* lines, size_t count);
@@ -13934,6 +14028,7 @@ namespace BinaryNinja {
 		bool IsValid(BinaryView* view) override;
 		Ref<TypePrinter> GetTypePrinter() override;
 		Ref<TypeParser> GetTypeParser() override;
+		Ref<LineFormatter> GetLineFormatter() override;
 		std::vector<DisassemblyTextLine> GetFunctionTypeTokens(
 			Function* func, DisassemblySettings* settings = nullptr) override;
 	};

binaryninjacore.h

@@ -44,7 +44,7 @@
 // will require rebuilding. The minimum version is increased when there are
 // incompatible changes that break binary compatibility, such as changes to
 // existing types or functions.
-#define BN_MINIMUM_CORE_ABI_VERSION 86
+#define BN_MINIMUM_CORE_ABI_VERSION 89
 
 #ifdef __GNUC__
 	#ifdef BINARYNINJACORE_LIBRARY
@@ -302,6 +302,7 @@ extern "C"
 	typedef struct BNDemangler BNDemangler;
 	typedef struct BNFirmwareNinja BNFirmwareNinja;
 	typedef struct BNFirmwareNinjaReferenceNode BNFirmwareNinjaReferenceNode;
+	typedef struct BNLineFormatter BNLineFormatter;
 
 	//! Console log levels
 	typedef enum BNLogLevel
@@ -727,6 +728,7 @@ extern "C"
 		HighLevelILLinearDisassembly = 65,
 		WaitForIL = 66,
 		IndentHLILBody = 67,
+		DisableLineFormatting = 68,
 
 		// Debugging options
 		ShowFlagUsage = 128,
@@ -3439,6 +3441,7 @@ extern "C"
 		bool (*isValid)(void* ctxt, BNBinaryView* view);
 		BNTypePrinter* (*getTypePrinter)(void* ctxt);
 		BNTypeParser* (*getTypeParser)(void* ctxt);
+		BNLineFormatter* (*getLineFormatter)(void* ctxt);
 		BNDisassemblyTextLine* (*getFunctionTypeTokens)(
 			void* ctxt, BNFunction* func, BNDisassemblySettings* settings, size_t* count);
 		void (*freeLines)(void* ctxt, BNDisassemblyTextLine* lines, size_t count);
@@ -3538,6 +3541,28 @@ extern "C"
 		size_t unique;
 	} BNFirmwareNinjaDeviceAccesses;
 
+	typedef struct BNLineFormatterSettings
+	{
+		BNHighLevelILFunction* highLevelIL;
+		size_t desiredLineLength;
+		size_t minimumContentLength;
+		size_t tabWidth;
+		char* languageName;
+		char* commentStartString;
+		char* commentEndString;
+		char* annotationStartString;
+		char* annotationEndString;
+	} BNLineFormatterSettings;
+
+	typedef struct BNCustomLineFormatter
+	{
+		void* context;
+		BNDisassemblyTextLine* (*formatLines)(void* ctxt, BNDisassemblyTextLine* inLines, size_t inCount,
+			const BNLineFormatterSettings* settings, size_t* outCount);
+		void (*freeLines)(void* ctxt, BNDisassemblyTextLine* lines, size_t count);
+	} BNCustomLineFormatter;
+
+
 	BINARYNINJACOREAPI char* BNAllocString(const char* contents);
 	BINARYNINJACOREAPI char* BNAllocStringWithLength(const char* contents, size_t len);
 	BINARYNINJACOREAPI void BNFreeString(char* str);
@@ -5454,6 +5479,9 @@ extern "C"
 
 	// Disassembly settings
 	BINARYNINJACOREAPI BNDisassemblySettings* BNCreateDisassemblySettings(void);
+	BINARYNINJACOREAPI BNDisassemblySettings* BNDefaultDisassemblySettings(void);
+	BINARYNINJACOREAPI BNDisassemblySettings* BNDefaultGraphDisassemblySettings(void);
+	BINARYNINJACOREAPI BNDisassemblySettings* BNDefaultLinearDisassemblySettings(void);
 	BINARYNINJACOREAPI BNDisassemblySettings* BNNewDisassemblySettingsReference(BNDisassemblySettings* settings);
 	BINARYNINJACOREAPI BNDisassemblySettings* BNDuplicateDisassemblySettings(BNDisassemblySettings* settings);
 	BINARYNINJACOREAPI void BNFreeDisassemblySettings(BNDisassemblySettings* settings);
@@ -6144,15 +6172,19 @@ extern "C"
 		BNLanguageRepresentationFunctionType* type, BNBinaryView* view);
 	BINARYNINJACOREAPI BNTypePrinter* BNGetLanguageRepresentationFunctionTypePrinter(BNLanguageRepresentationFunctionType* type);
 	BINARYNINJACOREAPI BNTypeParser* BNGetLanguageRepresentationFunctionTypeParser(BNLanguageRepresentationFunctionType* type);
+	BINARYNINJACOREAPI BNLineFormatter* BNGetLanguageRepresentationFunctionTypeLineFormatter(
+		BNLanguageRepresentationFunctionType* type);
 	BINARYNINJACOREAPI BNDisassemblyTextLine* BNGetLanguageRepresentationFunctionTypeFunctionTypeTokens(
 		BNLanguageRepresentationFunctionType* type, BNFunction* func, BNDisassemblySettings* settings, size_t* count);
 
 	BINARYNINJACOREAPI BNLanguageRepresentationFunction* BNCreateCustomLanguageRepresentationFunction(
-		BNArchitecture* arch, BNFunction* func, BNHighLevelILFunction* highLevelIL,
-		BNCustomLanguageRepresentationFunction* callbacks);
+		BNLanguageRepresentationFunctionType* type, BNArchitecture* arch, BNFunction* func,
+		BNHighLevelILFunction* highLevelIL, BNCustomLanguageRepresentationFunction* callbacks);
 	BINARYNINJACOREAPI BNLanguageRepresentationFunction* BNNewLanguageRepresentationFunctionReference(
 		BNLanguageRepresentationFunction* func);
 	BINARYNINJACOREAPI void BNFreeLanguageRepresentationFunction(BNLanguageRepresentationFunction* func);
+	BINARYNINJACOREAPI BNLanguageRepresentationFunctionType* BNGetLanguageRepresentationType(
+		BNLanguageRepresentationFunction* func);
 	BINARYNINJACOREAPI BNArchitecture* BNGetLanguageRepresentationArchitecture(BNLanguageRepresentationFunction* func);
 	BINARYNINJACOREAPI BNFunction* BNGetLanguageRepresentationOwnerFunction(BNLanguageRepresentationFunction* func);
 	BINARYNINJACOREAPI BNHighLevelILFunction* BNGetLanguageRepresentationILFunction(BNLanguageRepresentationFunction* func);
@@ -8057,6 +8089,25 @@ extern "C"
 	BINARYNINJACOREAPI void BNFreeFirmwareNinjaReferenceNode(BNFirmwareNinjaReferenceNode* node);
 	BINARYNINJACOREAPI BNFirmwareNinjaReferenceNode* BNNewFirmwareNinjaReferenceNodeReference(BNFirmwareNinjaReferenceNode* node);
 	BINARYNINJACOREAPI void BNFreeFirmwareNinjaReferenceNodes(BNFirmwareNinjaReferenceNode** nodes, size_t count);
+
+	// Line formatters
+	BINARYNINJACOREAPI BNLineFormatter* BNRegisterLineFormatter(const char* name, BNCustomLineFormatter* callbacks);
+	BINARYNINJACOREAPI BNLineFormatter** BNGetLineFormatterList(size_t* count);
+	BINARYNINJACOREAPI void BNFreeLineFormatterList(BNLineFormatter** formatters);
+	BINARYNINJACOREAPI BNLineFormatter* BNGetLineFormatterByName(const char* name);
+	BINARYNINJACOREAPI BNLineFormatter* BNGetDefaultLineFormatter();
+
+	BINARYNINJACOREAPI char* BNGetLineFormatterName(BNLineFormatter* formatter);
+
+	BINARYNINJACOREAPI BNDisassemblyTextLine* BNFormatLines(BNLineFormatter* formatter, BNDisassemblyTextLine* inLines,
+		size_t inCount, const BNLineFormatterSettings* settings, size_t* outCount);
+
+	BINARYNINJACOREAPI BNLineFormatterSettings* BNGetDefaultLineFormatterSettings(
+		BNDisassemblySettings* settings, BNHighLevelILFunction* func);
+	BINARYNINJACOREAPI BNLineFormatterSettings* BNGetLanguageRepresentationLineFormatterSettings(
+		BNDisassemblySettings* settings, BNLanguageRepresentationFunction* func);
+	BINARYNINJACOREAPI void BNFreeLineFormatterSettings(BNLineFormatterSettings* settings);
+
 #ifdef __cplusplus
 }
 #endif