def print_template(synset):
value = {}
verb = synset_name(synset)
for tense in verbs.TENSES:
con = verbs.conjugate(verb, tense)
if con:
value['actual'] = con
value['time'] = tense[0]
value['pov'] = tense[1]
value['num'] = tense[2]
value['dir'] = tense[3]
value['prog'] = tense[4]
print template.render({'value': value})
def rephrase_question(q_tagged, debug=False):
score_mod = 1.
q_tokens = [t[0] for t in q_tagged]
if q_tagged and q_tagged[0][1] == config.unknown_tag:
return q_tokens, score_mod
qws = [
i for i in range(len(q_tokens))
if q_tokens[i].lower() in question_words_single
]
if not qws:
if q_tokens[-1] in '!?.':
q_tokens = q_tokens[:-1]
return q_tokens, score_mod
qw = min(qws)
if q_tokens[-1] not in '!?.':
q_tokens = q_tokens + ['?']
q_tagged.append((u'?', u'.'))
query_beg, query = q_tokens[:qw + 1], q_tokens[qw + 1:]
pos = q_tagged[qw + 1:]
verbs = [i for i in range(len(query)) if pos[i][1][:2] in ['VB', 'MD']]
if not verbs:
return query_beg + query[:-1], score_mod
verb_idx = min(verbs)
nouns = [
i for i in range(verb_idx + 1, len(query))
if pos[i][1].startswith('NN')
]
if not nouns:
return query_beg + query[:-1], score_mod
tree = parse_tree(pos[verb_idx:])
noun_phrase = list(zip(*get_noun_phrases(tree).next())[0])
verb_insert_idx = min(i for i in range(len(query))
if query[i:i + len(noun_phrase)] == noun_phrase)
verb_insert_idx += len(noun_phrase)
verb = query[verb_idx]
if pos[verb_idx][1] != 'MD' and vb.conjugate(verb) not in aux_vbs:
return query_beg + query[:-1], score_mod
if verb == 'do':
query_new = query[:verb_idx] + query[verb_idx + 1:-1]
elif verb in ['does', 'did']:
verb_to_correct_idx = min([i for i in verbs if i >= verb_insert_idx]
or [None])
if verb_to_correct_idx is None:
query_new = query[:-1]
else:
v = query[verb_to_correct_idx]
corrected = vb.conjugate(
v, '3sg') if verb == 'does' else vb.conjugate(v, 'ppl')
query_new = query[:verb_idx] + \
query[verb_idx+1:verb_to_correct_idx] + \
[corrected] + query[verb_to_correct_idx+1:-1]
else:
query_new = query[:verb_idx] + \
query[verb_idx+1:verb_insert_idx] + \
[verb] + query[verb_insert_idx:-1]
if debug:
print(q_tokens, qw, verb_insert_idx, query)
# penalty for questions better suited for SimpleWikiTalker
if ' '.join(q_tokens[qw:qw+2]).lower() in swt_starts and \
verb_insert_idx == len(query) - 1:
print("SQUAD: penalizing swt start")
score_mod = SWT_QUESTION_PENALTY
return query_beg + query_new, score_mod
def makeSameTense(self, w1, w2):
tense = count([i[0] for i in tenses(w2)], stopwords=True)
tense = sorted(tense, key=operator.itemgetter(2))
return verbs.conjugate(w1, tense[0])
def makeSameTense(self, w1, w2):
tense = count([i[0] for i in tenses(w2)], stopwords=True)
tense = sorted(tense, key=operator.itemgetter(2))
return verbs.conjugate(w1, tense[0])
def is_verb(w, tag):
return tag.startswith('VB') and vb.conjugate(w) not in banned_verbs
