Ghidra script to load in function symbols of Link32/Visual C++ .map files

VSMapLoader.java

// Ghidra script to load in symbols of Link32/Visual C++ .map files
// aiding reverse-engineering of Windows/XBOX executables with
// said map files somehow available. Maybe this could be made better?
//@author Lily
//@category Lily.Tools

import java.io.*;
import java.util.*;

import ghidra.app.script.GhidraScript;
import ghidra.program.model.symbol.*;
import ghidra.program.model.address.*;
import ghidra.program.model.mem.*;
import ghidra.program.model.listing.*;
import ghidra.program.flatapi.FlatProgramAPI;

public class VSMapLoader extends GhidraScript {
	
	// map entry
	public class MapEntry {
		// The code segment we are in.
		MemoryBlock cs;
		
		// The address of the function relative to CS start.
		Address address;
		
		// The name to override Ghidra's name with.
		String name;
	}
	
	// we cache here for speedy nyoomiation
	// probably doesn't matter cause java is a slow piece of shit language
	// but gotta make it go a little bit faster somehow
	private Vector<MapEntry> mapFileParsedEntries;
	private Vector<Function> allFunctions;
	
	private void InitalizeFunctions() {
		allFunctions = new Vector<Function>();
		
		Function f = getFirstFunction();
		while(f != null) {
			allFunctions.add(f);
			f = getFunctionAfter(f);
		}
	}
	
	// The VS map format is 
	// 0002:00179b70       _FILESYS_init              008a2870 f   realfilea:filesys.obj
	// csid:raddress       function_name              addr2    f   <libfile:>objfile
	// ^    ^              ^                          
	// we care about these only
	
	private void ParseMap(File file) throws IOException {
		mapFileParsedEntries = new Vector<MapEntry>();
		FileReader reader = new FileReader(file);
		BufferedReader breader = new BufferedReader(reader);
		String line;
		
		// first skip until the Address marking
		// if this fails, then the file probably wasn't a map file
		while((line = breader.readLine()) != null) {
			if(line.startsWith("  Address"))
				break;
		}
		breader.readLine(); // then another to avoid the empty line
		
		while((line = breader.readLine()) != null) {
			line = line.trim();
			line = line.replaceAll("\\s+", " ");
			
			String[] stuff = line.split(" ");
			
			// break out if we're about to parse the "entry point"
			if(stuff[0] == "entry")
				break;
			
			// at least 2 elements
			if(stuff.length < 2)
				break;
			
			// [0] == CS
			// [1] == Address (Relative to CS start)
			String[] addrPair = stuff[0].split(":");
			String funcName = stuff[1];
			
			MapEntry me = new MapEntry();
			
			me.cs = getCodeSegment(Integer.parseInt(addrPair[0], 16));
			me.address = getAddress(Long.parseLong(addrPair[1], 16));
			me.name = funcName;
			mapFileParsedEntries.add(me);
		}
		
	}
	
	private Address getAddress(long offset) {
		return currentProgram.getAddressFactory().getDefaultAddressSpace().getAddress(offset);		
	}
	
	// this allows us to directly use the code-segment ID
	// from the map file
	private MemoryBlock getCodeSegment(int index) {
		MemoryBlock[] blocks = getMemoryBlocks();
		return blocks[index];
	}
	
	public void run() throws Exception {
		InitalizeFunctions();
		
		File file = askFile("Please provide a Visual C++ .MAP file", "OK");
		
		ParseMap(file);
		
		FunctionManager functionMgr = currentProgram.getFunctionManager();
		
		for(MapEntry mapEntry : mapFileParsedEntries) {
			// Ghidra stores its function addresses
			// already in the proper code segments,
			// so we need to do that with our map entries.
			Address adjusted = mapEntry.cs.getStart().add(mapEntry.address.getOffset());
			Function func = functionMgr.getFunctionAt(adjusted);
			
			if(func != null) {
				String oldName = func.getName();
				println(String.format("Map file %1s (at %2s) matches %3s", mapEntry.name, mapEntry.address , oldName));
				
				func.setName(mapEntry.name, SourceType.ANALYSIS);
				println(String.format("Adjusted %1s to %2s!", oldName, mapEntry.name));		
			}
		}

	}
	

	
	
}