Basic XML Output Parsing

.gitmodules

@@ -1,3 +1,6 @@
 [submodule "ghidra/ghidra"]
 	path = ghidra/ghidra
 	url = https://github.com/thestr4ng3r/ghidra.git
+[submodule "third-party/pugixml"]
+	path = third-party/pugixml
+	url = https://github.com/zeux/pugixml.git

CMakeLists.txt

@@ -9,6 +9,7 @@ list(APPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_SOURCE_DIR}/cmake")
 set(CMAKE_CXX_STANDARD 11)
 
 add_subdirectory(ghidra)
+add_subdirectory(third-party)
 
 set(SOURCE
 		src/core_ghidra.cpp
@@ -27,7 +28,9 @@ set(SOURCE
 		src/R2Emit.cpp
 		src/R2Emit.h
 		src/AnnotatedCode.h
-		src/AnnotatedCode.c)
+		src/AnnotatedCode.c
+		src/CodeXMLParse.h
+		src/CodeXMLParse.cpp)
 
 find_package(Radare2 REQUIRED)
 
@@ -37,7 +40,8 @@ endif()
 
 add_library(core_ghidra SHARED ${SOURCE})
 target_link_libraries(core_ghidra ghidra_decompiler_base ghidra_libdecomp ghidra_decompiler_sleigh)
+target_link_libraries(core_ghidra pugixml)
+target_link_libraries(core_ghidra Radare2::libr)
 set_target_properties(core_ghidra PROPERTIES
 		OUTPUT_NAME core_ghidra
 		PREFIX "")
-target_link_libraries(core_ghidra Radare2::libr)

src/CodeXMLParse.cpp

@@ -0,0 +1,50 @@
+
+#include "CodeXMLParse.h"
+#include <funcdata.hh>
+#include <r_util.h>
+#include <pugixml.hpp>
+#include <sstream>
+#include <string>
+
+static void ParseNode(pugi::xml_node node, Funcdata *func, std::ostream &stream, RAnnotatedCode *code)
+{
+	if(node.type() == pugi::xml_node_type::node_pcdata)
+	{
+		stream << node.value();
+		return;
+	}
+
+	if(strcmp(node.name(), "break") == 0)
+	{
+		stream << "\n";
+		stream << std::string(node.attribute("indent").as_uint(0), ' ');
+	}
+
+	for(pugi::xml_node child : node)
+		ParseNode(child, func, stream, code);
+}
+
+R_API RAnnotatedCode *ParseCodeXML(Funcdata *func, const char *xml)
+{
+	pugi::xml_document doc;
+	if(!doc.load_string(xml, pugi::parse_default | pugi::parse_ws_pcdata))
+		return nullptr;
+
+	std::stringstream ss;
+	RAnnotatedCode *code = r_annotated_code_new(nullptr);
+	if(!code)
+		return nullptr;
+
+	ParseNode(doc.child("function"), func, ss, code);
+
+	std::string str = ss.str();
+	code->code = reinterpret_cast<char *>(r_malloc(str.length() + 1));
+	if(!code->code)
+	{
+		r_annotated_code_free(code);
+		return nullptr;
+	}
+	memcpy(code->code, str.c_str(), str.length());
+	code->code[str.length()] = '\0';
+	return code;
+}

src/CodeXMLParse.h

@@ -0,0 +1,11 @@
+
+#ifndef R2GHIDRA_CODEXMLPARSE_H
+#define R2GHIDRA_CODEXMLPARSE_H
+
+#include "AnnotatedCode.h"
+
+class Funcdata;
+
+R_API RAnnotatedCode *ParseCodeXML(Funcdata *func, const char *xml);
+
+#endif //R2GHIDRA_CODEXMLPARSE_H

src/core_ghidra.cpp

@@ -1,6 +1,7 @@
 /* radare - LGPL - Copyright 2019 - thestr4ng3r */
 
 #include "R2Architecture.h"
+#include "CodeXMLParse.h"
 
 #include <libdecomp.hh>
 #include <printc.hh>
@@ -150,15 +151,19 @@ static void Decompile(RCore *core, DecompileMode mode)
 			func->saveXml(out_stream, true);
 			out_stream << "</function><code>";
 		}
+		else if(mode == DecompileMode::OFFSET)
+		{
+			arch.print->setXML(true);
+		}
 
 		switch(mode)
 		{
 			case DecompileMode::XML:
 			case DecompileMode::DEFAULT:
+			case DecompileMode::OFFSET:
 				arch.print->docFunction(func);
 				break;
 			case DecompileMode::STATEMENTS:
-			case DecompileMode::OFFSET:
 				arch.print_with_offsets->docFunction(func);
 				break;
 			case DecompileMode::DEBUG_XML:
@@ -168,6 +173,16 @@ static void Decompile(RCore *core, DecompileMode mode)
 
 		if(mode == DecompileMode::OFFSET)
 		{
+			RAnnotatedCode *code = ParseCodeXML(func, out_stream.str().c_str());
+			if(!code)
+			{
+				eprintf("Failed to parse XML code\n");
+				return;
+			}
+			// TODO: print code with offsets from RAnnotatedCode here
+			r_cons_print(code->code);
+			r_annotated_code_free(code);
+			return;
 			ut64 offset;
 			string line;
 			std::stringstream line_stream;

third-party/CMakeLists.txt

@@ -0,0 +1,8 @@
+
+add_library(pugixml STATIC
+		pugixml/src/pugiconfig.hpp
+		pugixml/src/pugixml.cpp
+		pugixml/src/pugixml.hpp)
+target_include_directories(pugixml PUBLIC pugixml/src)
+set_target_properties(pugixml PROPERTIES POSITION_INDEPENDENT_CODE ON)
+