class TestTyper(unittest.TestCase):
def setUp(self):
self.t = Typer()
def tearDown(self):
self.t = None
def test(self):
self.assertEqual(self.t.type("tha", "V-p", "BIPP")[1], "s[dcl pres cons]/pp/n")
self.assertEqual(self.t.type("bha", "V-s", "BIPP")[1], "s[dcl past cons]/pp/n")
def setUp(self):
self.t = Typer()
return s.replace("&", "&")
brownfile = open(sys.argv[1], 'rb')
corpus = pickle.load(brownfile)
brownfile.close()
output = open(sys.argv[2], 'w')
# features
with open("resources/features.txt") as f:
for line in f:
output.write(line)
# type-changing and type-raising rules
with open("resources/rules.txt") as r:
for line in r:
output.write(line)
retagger = Retagger()
typer = Typer()
families = set()
words = set()
# assumes a single list rather than a list of lists (need to think about this)
for surface, pos in corpus:
if pos != "":
tags = retagger.retag(surface, pos)
for tag in tags:
newtagtype = typer.type(surface, pos, tag)
newtag = newtagtype[0]
type = newtagtype[1]
families.add("family %s { entry: %s; }" % (newtag, type))
words.add('word "%s_%s":%s; # %s' % (tidyword(surface), newtag, newtag, pos))
for family in sorted(families):
output.write(family + '\n')
