def sentence_db(nlp, fh):
sentences = list()
clause_list = list()
for line in fh.readlines():
line = line.decode('utf8').strip()
src, text = line.split("\t")
span_obj = first_s(nlp, text)
ccs = clauses(span_obj)
pps = prep_phrases(span_obj.root)
agree = requires_past_tense_agreement(span_obj)
# keep sentences with no recognizable subject
try:
nsubj = get_nsubj(span_obj)
plural = nsubj_is_plural(nsubj)
except ValueError:
nsubj = None
plural = None
sentences.append({
'src': int(src),
'text': text,
'span': span_obj,
'nsubj': nsubj,
'agree': agree,
'plural': plural,
'pps': pps,
})
if len(ccs) > 1:
clause_list.extend([(src, span_obj, c) for c in ccs])
for src, span_obj, clause in clause_list:
pps = prep_phrases(clause.root)
agree = requires_past_tense_agreement(clause)
try:
nsubj = get_nsubj(clause)
plural = nsubj_is_plural(nsubj)
except ValueError:
continue
sentences.append({
'src': int(src),
'text': clause.text,
'span': clause,
'nsubj': nsubj,
'agree': agree,
'plural': plural,
'pps': [],
})
return sentences
def sentence_db(nlp, fh):
sentences = list()
clause_list = list()
for line in fh.readlines():
line = line.decode('utf8').strip()
src, text = line.split("\t")
span_obj = first_s(nlp, text)
ccs = clauses(span_obj)
pps = prep_phrases(span_obj.root)
agree = requires_past_tense_agreement(span_obj)
# keep sentences with no recognizable subject
try:
nsubj = get_nsubj(span_obj)
plural = nsubj_is_plural(nsubj)
except ValueError:
nsubj = None
plural = None
sentences.append({
'src': int(src),
'text': text,
'span': span_obj,
'nsubj': nsubj,
'agree': agree,
'plural': plural,
'pps': pps,
})
if len(ccs) > 1:
clause_list.extend([(src, span_obj, c) for c in ccs])
for src, span_obj, clause in clause_list:
pps = prep_phrases(clause.root)
agree = requires_past_tense_agreement(clause)
try:
nsubj = get_nsubj(clause)
plural = nsubj_is_plural(nsubj)
except ValueError:
continue
sentences.append({
'src': int(src),
'text': clause.text,
'span': clause,
'nsubj': nsubj,
'agree': agree,
'plural': plural,
'pps': [],
})
return sentences
def random_sentence_for_nsubj(sdb, nsubj):
no_agreement_sentences = [x for x in sdb if x['nsubj'] is not None \
and not(x['agree'])]
agree_sg = [x for x in sdb if x['nsubj'] is not None and \
(not(x['plural']) and x['agree'])]
agree_pl = [x for x in sdb if x['nsubj'] is not None and \
(x['plural'] and x['agree'])]
if nsubj is not None:
if nsubj_is_plural(nsubj):
d = random.choice(no_agreement_sentences + agree_pl)
pronoun = "they"
else:
d = random.choice(no_agreement_sentences + agree_sg)
pronoun = "it"
else:
d = random.choice(no_agreement_sentences + agree_sg)
pronoun = "it"
return pronoun, d
def random_sentence_for_nsubj(sdb, nsubj):
no_agreement_sentences = [x for x in sdb if x['nsubj'] is not None \
and not(x['agree'])]
agree_sg = [x for x in sdb if x['nsubj'] is not None and \
(not(x['plural']) and x['agree'])]
agree_pl = [x for x in sdb if x['nsubj'] is not None and \
(x['plural'] and x['agree'])]
if nsubj is not None:
if nsubj_is_plural(nsubj):
d = random.choice(no_agreement_sentences + agree_pl)
pronoun = "they"
else:
d = random.choice(no_agreement_sentences + agree_sg)
pronoun = "it"
else:
d = random.choice(no_agreement_sentences + agree_sg)
pronoun = "it"
return pronoun, d
def reminded(sdb, state):
verbs = ['reminded me of', 'reminded you of', 'reminded us of',
'recalled', 'brought to mind', 'evoked', 'suggested',
'seemed like', 'resembled', 'had the quality of']
adverbs = ['somehow', 'at the time', 'sometimes', 'at first',
'maybe']
if len(state.topics) > 0 and state.topics[-1] is not None and \
nsubj_is_plural(state.topics[-1]):
subj = 'they'
else:
subj = 'it'
if len(state.subj_orth) > 0 \
and state.subj_orth[-1].lower() in ('it', 'they') \
and len(state.topics) > 0 \
and state.topics[-1] is not None:
subj = "the " + state.topics[-1].root.orth_
state.subj_orth.append(subj)
nps = [x['nsubj'] for x in sdb if x['nsubj'] is not None]
s = subj + " " + random.choice(verbs) + " " + random.choice(nps).text
if random.randrange(6) == 0:
s = random.choice(adverbs) + " " + s
return s
def reminded(sdb, state):
verbs = [
'reminded me of', 'reminded you of', 'reminded us of', 'recalled',
'brought to mind', 'evoked', 'suggested', 'seemed like', 'resembled',
'had the quality of'
]
adverbs = ['somehow', 'at the time', 'sometimes', 'at first', 'maybe']
if len(state.topics) > 0 and state.topics[-1] is not None and \
nsubj_is_plural(state.topics[-1]):
subj = 'they'
else:
subj = 'it'
if len(state.subj_orth) > 0 \
and state.subj_orth[-1].lower() in ('it', 'they') \
and len(state.topics) > 0 \
and state.topics[-1] is not None:
subj = "the " + state.topics[-1].root.orth_
state.subj_orth.append(subj)
nps = [x['nsubj'] for x in sdb if x['nsubj'] is not None]
s = subj + " " + random.choice(verbs) + " " + random.choice(nps).text
if random.randrange(6) == 0:
s = random.choice(adverbs) + " " + s
return s