class TestRetagger(unittest.TestCase):
def setUp(self):
self.r = Retagger()
def tearDown(self):
self.r = None
def test(self):
ag = self.r.retag("ag", "Sa")
self.assertTrue("ASP" in ag)
self.assertEqual(self.r.retag("a'", "Sa"), ["ASP"])
agus = self.r.retag("agus", "Cc")
self.assertTrue("CONJ" in agus)
air = self.r.retag("air", "Sp")
self.assertTrue("PP" in air)
self.assertTrue("P" in air)
self.assertEqual(self.r.retag("droch", "Ar"), ["DET"]) # not really
comma = self.r.retag(",", "Fi")
self.assertTrue("PUNC" in comma)
fullstop = self.r.retag(".", "Fe")
self.assertTrue("PUNC" in fullstop)
self.assertEqual(self.r.retag("le", "Sp"), ["P"])
self.assertEqual(self.r.retag("gun", "Qa"), ["GU"])
self.assertEqual(self.r.retag("dìreach", "Rg"), ["ADV"])
self.assertEqual(self.r.retag("Comhairle", "Ncsdf"), ["N"])
self.assertEqual(self.r.retag("galain", "Ncsfn"), ["N"])
self.assertEqual(self.r.retag("an", "Tdsf"), ["DET"])
self.assertEqual(self.r.retag("na", "Tdsfg"), ["DETNMOD"])
self.assertEqual(self.r.retag("[1]", "Xsc"), ["ADVPRE"])
radh = self.r.retag("ràdh", "Nv")
self.assertTrue("VPROP" in radh)
rinn = self.r.retag("rinn", "V-s")
self.assertTrue("TRANS" in rinn)
tha = self.r.retag("tha", "V-p")
self.assertTrue("BIPP" in tha)
self.assertTrue("BIPROG" in tha)
def setUp(self):
self.r = Retagger()
else:
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):
