Script for converting udis86 optable.xml to YAML

xml2yaml

import sys

from lxml import etree

parser = etree.XMLParser(remove_comments=False)
tree = etree.parse('docs/x86/optable.xml', parser=parser)
root = tree.getroot()

for e in root:
	if e.tag == 'instruction':
		mnemonics = [m for m in e.iterchildren(tag='mnemonic')]
		assert len(mnemonics) == 1
		mnemonic = mnemonics[0].text

		print "%s:" % mnemonic

		for d in e.iterchildren(tag='def'):
			start = "-"

			vendor = [x for x in d.iterchildren(tag='vendor')] or [x for x in e.iterchildren(tag='vendor')]
			assert len(vendor) <= 1
			if vendor:
				print " %s vendor: %s" % (start, vendor[0].text)
				start = " "

			pfxs = [x for x in d.iterchildren(tag='pfx')]
			assert len(pfxs) <= 1
			if len(pfxs) == 1:
				print " %s pfx: %s" % (start, pfxs[0].text)
				start = " "

			opc = [x for x in d.iterchildren(tag='opc')]
			assert len(opc) == 1
			#print " %s opc: \"%s\"" % (start, opc[0].text)
			print " %s opc: %s" % (start, opc[0].text)
			start = " "

			oprs = [x for x in d.iterchildren(tag='opr')]
			if len(oprs) >= 2:
				print >>sys.stderr, "warning: opcode %s has more than one set of operands" % mnemonic
			if len(oprs) >= 1:
				print " %s opr: %s" % (start, oprs[-1].text)
				start = " "

			modes = [x for x in d.iterchildren(tag='mode')]
			assert len(modes) <= 1
			if len(modes) == 1:
				print " %s mode: %s" % (start, modes[0].text)
				start = " "

			syns = [x for x in d.iterchildren(tag='syn')]
			assert len(syns) <= 1
			if len(syns) == 1:
				print " %s syn: %s" % (start, syns[0].text)
				start = " "

			cpuid = [x for x in d.iterchildren(tag='cpuid')] or [x for x in e.iterchildren(tag='cpuid')]
			assert len(cpuid) <= 1, "%s cpuids: %s" % (mnemonic, ", ".join([x.text for x in cpuid]))
			if cpuid:
				print "   cpuid: %s" % cpuid[0].text

	elif e.tag is etree.Comment:
		for line in e.text.splitlines():
			print "# %s" % line
	else:
		raise "Unhandled tag"

	print