first version of jos2wccl rule conv

tools/ruljos2wccl.py

+#!/usr/bin/python
+# -*- coding: utf-8 -*-
+import os, sys, codecs, re
+from optparse import OptionParser
+from StringIO import StringIO
+
+descr = """%prog [options] IN OUT
+
+Attempts to convert JOSKIPI rules to WCCL rules.
+NOTE: this is based on very naive heuristics.
+"""
+
+# TODO agr bits
+# TODO isbig -> regex
+
+p_brackets = re.compile(u'{(\\s*"[^"]*"(\\s*,\\s*"[^"]*")*\\s*)}', re.U)
+
+def jos2ccl(what):
+	what = what.replace('flex[', 'class[')
+	return p_brackets.sub(u'[\\1]', what)
+	
+class IndentWriter:
+	def __init__(self, out, baseindent = 0):
+		self.out = out
+		self.indent = baseindent
+	
+	def write(self, text):
+		for line in text.split('\n'):
+			line = line.strip()
+			if line:
+				meat = line.split('//', 1)[0]
+				idelta = meat.count('(') - meat.count(')')
+				nowindent = (self.indent + idelta) if meat.startswith(')') else self.indent
+				self.out.write('\t' * nowindent)
+				self.out.write(line)
+				self.out.write('\n')
+				self.indent += idelta
+			else:
+				self.out.write('\n')
+
+class Rule:
+	def __init__(self, text):
+		pre, cond = text.split(':-')
+		act, name = pre.split('#')
+		self.cond = jos2ccl(cond.strip())
+		self.act = jos2ccl(act.strip())
+		self.name = name.strip()
+	
+	def write(self, out, comma = False):
+		out.write('rule("%s",\n' % self.name)
+		out.write('%s,\n' % self.cond)
+		out.write('%s\n' % self.act)
+		out.write(')%s\n' % (',' if comma else ''))
+
+def rule_texts(infile):
+	buf = StringIO()
+	for line in infile:
+		if 'delete' in line: # TODO: regex with other actions
+			yield buf.getvalue()
+			buf = StringIO()
+		buf.write(line)
+	yield buf.getvalue()
+
+def rules(infile):
+	for rt in rule_texts(infile):
+		rt = '\n'.join(line.split('//', 1)[0] for line in rt.split('\n'))
+		if rt.strip():
+			yield Rule(rt)
+
+def go():
+	parser = OptionParser(usage=descr)
+	parser.add_option('-v', '--verbose', action='store_true', dest='verbose', default=False, help='set the verbose mode')
+	
+	(options, args) = parser.parse_args()
+	
+	if len(args) != 2:
+		print 'You need to provide input and output paths'
+		print 'See %s --help' % os.path.basename(sys.argv[0])
+		sys.exit(1)
+	
+	inpath, outpath = args
+	
+	inf = codecs.open(inpath, 'rb', 'utf-8')
+	outf = codecs.open(outpath, 'wb', 'utf-8')
+	w = IndentWriter(outf)
+	
+	indent = 0
+	w.write('rules(\n')
+	# quick and dirty: to get all but last
+	allrules = [r for r in rules(inf)]
+	for rule in allrules[:-1]:
+		rule.write(w, True)
+	allrules[-1].write(w, False) # no comma
+	w.write(')\n')
+	
+	outf.close()
+	inf.close()
+
+if __name__ == '__main__':
+	go()