def test_getTextLabelTuplesInSentences(self): info = self.info tuples = TextMarker.getTextLabelTuples(info) tupless = TextMarker.getTextLabelTuplesInSentences(info) # for tuples in tupless: # for tup in tuples: # print tup # print '' assert type(tupless) is list assert type(tupless[0]) is list assert type(tupless[0][0]) is tuple assert len(tuples) == len([item for tuples in tupless for item in tuples])
def test_getTextLabelTuplesInSentences(self):
info = self.info
tuples = TextMarker.getTextLabelTuples(info)
tupless = TextMarker.getTextLabelTuplesInSentences(info)
# for tuples in tupless:
# for tup in tuples:
# print tup
# print ''
assert type(tupless) is list
assert type(tupless[0]) is list
assert type(tupless[0][0]) is tuple
assert len(tuples) == len(
[item for tuples in tupless for item in tuples])
def test_getFeaturesInSentence(self):
lts = TextMarker.getTextLabelTuplesInSentences(self.info)
fword = FeatureExtractor.getFeaturesInSentence(lts[0])[0][0]
# for lt in lts:
# featuress = FeatureExtractor.getFeaturesInSentence(lt)
# print ''
# for x in featuress:
# print x
assert "tok" in fword
assert "tokkind" in fword
assert "ne" in fword
assert "contextfe" in fword
assert "morphfe" in fword
assert "posfe" in fword
assert "bef1tok" in fword
assert "bef1tokkind" in fword
assert "bef1ne" in fword
assert "bef1contextfe" in fword
assert "bef1morphfe" in fword
assert "bef1posfe" in fword
assert "bef2tok" in fword
assert "bef2tokkind" in fword
assert "bef2ne" in fword
assert "bef2contextfe" in fword
assert "bef2morphfe" in fword
assert "bef2posfe" in fword
def test_getMarkedText(self):
info = self.info
mtext = TextMarker.getMarkedText(info)
searchObj = re.findall(r'\[(.+?)\](.+?)\[.+?\]', mtext)
# for x in searchObj:
# print x
# print mtext
assert len(searchObj) > 1
def test_getMarkedText(self): info = self.info mtext = TextMarker.getMarkedText(info) searchObj = re.findall( r'\[(.+?)\](.+?)\[.+?\]', mtext) # for x in searchObj: # print x # print mtext assert len(searchObj)>1
def test_getTaggedText(self):
info = self.info
# print json.dumps(info, default=lambda o: o.__dict__, indent=2)
# assert 1==2
res = TextMarker.getTaggedText(info)
assert "[bwhat]" in res
assert "[bwho]" in res
assert "[bwhen]" in res
assert "[bwhere]" in res
assert "[bwhy]" in res
assert "[bhow]" in res
def test_getTaggedText(self): info = self.info # print json.dumps(info, default=lambda o: o.__dict__, indent=2) # assert 1==2 res = TextMarker.getTaggedText(info) assert "[bwhat]" in res assert "[bwho]" in res assert "[bwhen]" in res assert "[bwhere]" in res assert "[bwhy]" in res assert "[bhow]" in res
def getFitursFromInfo(info5w1hs):
fiturs = []
for info in info5w1hs:
# print info.text
# print ''
info = FeatureExtractor._INANLP(info)
# print info.text
# print ''
for idxsentence, tupls in enumerate(TextMarker.getTextLabelTuplesInSentences(info)):
# print tupls
featuress = FeatureExtractor.getFeaturesInSentence("%d" % (idxsentence), tupls)
for features in featuress:
fiturs.append(features)
return fiturs
def test_getOtherTaggedText(self):
info = self.info
res = TextMarker.getOtherTaggedText(info)
# print res
# print ''
searchObj = re.findall(r'(\[b.+?\].+?\[e.+?\])', res)
# for x in searchObj:
# print x
assert re.search(r'(\[bwhat\].+?\[ewhat\])', res)
assert re.search(r'(\[bwho\].+?\[ewho\])', res)
assert re.search(r'(\[bwhen\].+?\[ewhen\])', res)
assert re.search(r'(\[bwhere\].+?\[ewhere\])', res)
assert re.search(r'(\[bwhy\].+?\[ewhy\])', res)
assert re.search(r'(\[bhow\].+?\[ehow\])', res)
assert re.search(r'(\[bother\].+?\[eother\])', res)
def test_getOtherTaggedText(self): info = self.info res = TextMarker.getOtherTaggedText(info) # print res # print '' searchObj = re.findall( r'(\[b.+?\].+?\[e.+?\])', res) # for x in searchObj: # print x assert re.search(r'(\[bwhat\].+?\[ewhat\])', res) assert re.search(r'(\[bwho\].+?\[ewho\])', res) assert re.search(r'(\[bwhen\].+?\[ewhen\])', res) assert re.search(r'(\[bwhere\].+?\[ewhere\])', res) assert re.search(r'(\[bwhy\].+?\[ewhy\])', res) assert re.search(r'(\[bhow\].+?\[ehow\])', res) assert re.search(r'(\[bother\].+?\[eother\])', res)
def test_getTextLabelTuples(self):
info = self.info
tuples = TextMarker.getTextLabelTuples(info)
assert len(tuples) > 0
assert tuples[0]
assert tuples[1]
def test_getTaggedTexts(self):
infos = MDP.get5w1h([6])
res = TextMarker.getTaggedTexts(infos)
assert len(res) > 1
def test_getTextLabelTuples(self): info = self.info tuples = TextMarker.getTextLabelTuples(info) assert len(tuples)>0 assert tuples[0] assert tuples[1]
def test_getTaggedTexts(self): infos = MDP.get5w1h([6]) res = TextMarker.getTaggedTexts(infos) assert len(res)>1