A script to use with ghidra for importing DF structures into ghidra. Does not completely work yet; only imports global symbols at the moment.

genscript.py

#TODO write a description for this script
#@author 
#@category DwarfFortress
#@keybinding 
#@menupath 
#@toolbar 

#import xml.etree.ElementTree as ET

from ghidra.program.model.data import CategoryPath
from ghidra.program.model.data import StructureDataType
from ghidra.program.model.data import DataTypeConflictHandler
from ghidra.program.model.symbol import SourceType

from ghidra.util.xml import XmlUtilities
from java.io import File
from org.jdom.filter import ElementFilter

import re

importver = 'v0.44.12 SDL win64'

translate = {
	'padding': 'undefined',
	'bool': 'byte',
	's-float': 'float',
	'ptr-string': 'char*',
	'uint32_t': 'uint',
	'stl-fstream': 'stl::fstream',
	'int8_t': 'byte',
	'long': 'longlong',
	'uint16_t': 'ushort',
	'int16_t': 'short',
	'uint8_t': 'ubyte',
	'int64_t': 'longlong',
	'int32_t': 'int',
	'stl-string': 'std::string'
}

class DFField:
	def __init__(self, type, name, comment=None):
		self.name = name
		self.type = type
		self.comment = comment

class DFStruct:
	def __init__(self, type, name, comment=None):
		self.name = name
		self.type = type
		self.comment = comment
		self.fields = {}
		
	def addField(self, type, name, comment=None):
		self.fields[name] = DFField(type, name, comment)

def elementFilter(str):
	return ElementFilter(str)

def vec (c,typ):
	return c + '<' + typ + '>'

def ptr (typ):
	return typ + '*'
	
def arr (cnt,typ):
	return typ + '[' + cnt + ']'

def process(fld, typ, nam):
	ftype = typ
	fname = nam
	fcomm = fld.getAttributeValue('comment')
	if (ftype == 'bitfield' or ftype == 'compound'):
		ftype = fld.getAttributeValue('type-name')
	elif (ftype == 'stl-vector' or ftype == 'stl-deque' or 
		  ftype == 'stl-set' or ftype == 'stl-bit-vector' or
		  ftype == 'df-flagarray' or ftype == 'df-static-flagarray' or 
		  ftype == 'df-array' or ftype == 'df-linked-list'):
		vtype = fld.getAttributeValue('type-name')
		if (not vtype):
			ptype = fld.getAttributeValue('pointer-type')
			if (ptype):
				vtype = ptr(ptype)
			elif fld.getChild('pointer'):
				aname = 'T_' + fname
				tt = DFStruct(None, aname, fld.getAttributeValue('comment'))
				vtype = make_type(tt, fld.getChild('pointer'))
			else:
				vtype = vec(ftype,'undefined')
		ftype = vec(ftype,vtype)
	elif (ftype == 'static-array'):
		ptype = fld.getAttributeValue('type-name')
		count = fld.getAttributeValue('count')
		aname = 'T_' + fname
		if (not ptype):
			tt = DFStruct(None, aname, fld.getAttributeValue('comment'))
			ptype = make_type(tt, fld)
		ftype = arr(count,ptype)
	elif (ftype == 'pointer'):
		ptype = fld.getAttributeValue('type-name')
		aname = 'T_' + fname
		if (not ptype):
			tt = DFStruct(None, aname, fld.getAttributeValue('comment'))
			ptype = make_type(tt, fld)
		ftype = ptr(ptype)
	elif (ftype == 'compound'):
		ctype = fld.getAttributeValue('type-name')
		aname = 'T_' + fname
		if (not ctype):
			tt = DFStruct(None, aname, fld.getAttributeValue('comment'))
			ctype = make_type(tt, fld)
		ftype = ctype
	elif (ftype == 'enum'):
		aname = 'T_' + fname
		btype = fld.getAttributeValue('base-type') or 'int32_t'
		etype = DFStruct(btype, aname, fld.getAttributeValue('comment'))
		ftype = etype.name
	elif (ftype == 'static-string'):
		cnt = fld.getAttributeValue('size')
		ftype = arr(cnt, 'char')
	elif (ftype in translate):
		ftype = translate[ftype]

	return ftype

	
def make_type(typ, spec):
	anon = 0
	for fld in spec.getChildren():
		ftype = fld.getName()
		if (ftype == 'virtual-methods' or ftype == 'custom-methods'):
			continue
		if (ftype == 'code-helper' or ftype == 'extra-include' or ftype == 'comment'):
			continue
		fname = fld.getAttributeValue('name')
		if not fname:
			anon = anon + 1
			fname = 'anon_' + str(anon)
		ftype = process(fld, ftype, fname)
		fcomm = fld.getAttributeValue('comment')
		typ.addField(ftype, fname, fcomm)
	return typ.name
		
pgm = getCurrentProgram()
st = pgm.getSymbolTable()
ns = pgm.getGlobalNamespace()
af = pgm.getAddressFactory()

def addtype():
	pgm = getCurrentProgram()
	dtm = pgm.getDataTypeManager()
	cpr = dtm.getRootCategory().getCategoryPath()

	dfh_cp = CategoryPath(cpr,"dfhack")
	dfh_cat = dtm.createCategory(dfh_cp)

	newtype = StructureDataType(dfh_cp, "newtype", 0)

	fieldtype = dtm.getDataType(cpr, "word")
	newtype.add(fieldtype, 2, "field1", "")

	dtm.addDataType(newtype,DataTypeConflictHandler.REPLACE_HANDLER)

def main():
	dir = askDirectory("Directory", "Path to df-structures repo:")
	print dir

	getXML = lambda(f):	 XmlUtilities.readDocFromFile(File(dir, f))
		
	tree = getXML('symbols.xml')

	target = None
	ent = tree.getDescendants(elementFilter('symbol-table'))
	for e in ent:
		name = e.getAttributeValue('name')
		print name
		if name == importver: 
			target = e
			break
			
	for ga in target.getDescendants(elementFilter('global-address')):
		sym = ga.getAttributeValue('name')
		val = ga.getAttributeValue('value')
		addr = af.getAddress(val)
		l = st.createLabel(addr, sym, SourceType.IMPORTED)
		
	files = dir.listFiles()
	fmatcher = re.compile ( r'df\.(.*)\.xml' )

	types = {}

	for f in files:
		mat = fmatcher.match(f.getName())
		if mat:
			print 'Reading ' + f.getName()
			data = XmlUtilities.readDocFromFile(f).getRootElement()
			for e in data.getChildren():
				tag = e.getName()
#				print '>' + tag
				n = e.getAttributeValue('type-name')
				on = e.getAttributeValue('original-name')
				t = e.getAttributeValue('base-type') or 'int32_t'
				c = e.getAttributeValue('comment')
				typ = DFStruct(t,on or n,c)
				if tag == 'bitfield-type' or tag == 'enum-type':
					if t: typedef[n] = typ
				elif tag == 'class-type' or tag == 'struct-type':
					sup = e.getAttributeValue('inherits-from')
					if (sup):
						typ.super = sup
					types[n] = make_type(typ, e)
				elif tag == 'global-object':
					gln = e.getAttributeValue('name')
					if n:
						ftype = process(e, n, gln)
					else:
						tt = DFStruct(None, 'T_' + gln)
						ftype = make_type(tt, e)
					print 'global ' + gln + ' is type ' + ftype

	for t in types:
		print t, types[t]
		

main()