def test_lexicon(self):
# Assert lexicon from file (or file-like string).
f1 = u";;; Comments. \n schrödinger NNP \n cat NN"
f2 = StringIO.StringIO(u";;; Comments. \n schrödinger NNP \n cat NN")
v1 = text.Lexicon(path=f1)
v2 = text.Lexicon(path=f2)
self.assertEqual(v1[u"schrödinger"], "NNP")
self.assertEqual(v2[u"schrödinger"], "NNP")
print "pattern.text.Lexicon"
def test_lexicon(self):
# Assert loading and applying Brill lexicon and rules.
f1 = u";;; Comments. \n schrödinger NNP \n cat NN"
f2 = StringIO.StringIO(u"NN s fhassuf 1 NNS x")
f3 = StringIO.StringIO(u"VBD VB PREVTAG TO")
f4 = StringIO.StringIO(u"Schrödinger's cat PERS")
v = text.Lexicon(path=f1, morphology=f2, context=f3, entities=f4)
self.assertEqual(v[u"schrödinger"], "NNP")
self.assertEqual(v.morphology.apply(["cats", "NN"]), ["cats", "NNS"])
self.assertEqual(v.context.apply([["to", "TO"], ["be", "VBD"]]),
[["to", "TO"], ["be", "VB"]])
self.assertEqual(
v.entities.apply([[u"Schrödinger's", "NNP"], ["cat", "NN"]]),
[[u"Schrödinger's", "NNP-PERS"], ["cat", "NNP-PERS"]])
print "pattern.text.Lexicon"
print "pattern.text.Morphology"
print "pattern.text.Context"
print "pattern.text.Entities"