def __str__(self):
words_with_plural = list()
for word in self.words.split(' '):
plural = plural_singular.get_plurals(word, cbow=self.wordembedding)
if plural:
words_with_plural.append('[{} {}]'.format(word, plural))
else:
words_with_plural.append(word)
words_with_plural = ' '.join(words_with_plural)
if len(self.group) > 1:
gen_rule = ('u: {label} (<<{words}>>)\n\t'
'$quest = %originalsentence\n\t'
'^pick(~not_well_understood), %user, '
'mas ^pick(~search_options):\n\t - {questions}\n'
'{group_rejoinders}').format(
label=self.label,
words=words_with_plural,
questions='\n\t - '.join(self.questions) + ' -',
group_rejoinders=self.rejoinders_text())
else:
gen_rule = ('u: {label} (<<{words}>>)\n\t'
'$quest = %originalsentence\n\t'
'^pick(~not_well_understood), %user, '
'^pick(~you_mean) "{sugestion}"?\n'
'{group_rejoinders}').format(
label=self.label,
words=words_with_plural,
sugestion=self.group[0].title,
group_rejoinders=self.rejoinders_text())
return gen_rule
def generate_keywords_plurals(self):
plurals = list()
if self.name.endswith('_gen'):
return plurals
for kw in self.keywords:
plural = ''
for subkw in kw.split(' '):
plural_sub = plural_singular.get_plurals(
subkw, cbow=self.wordembedding)
if plural_sub:
plural = plural + ' ' + plural_sub
plurals.append(plural)
return plurals
def add_plurals(self, embedding_model):
question = self.add_syns_question
# if self.splited_entities:
# entities = self.splited_entities
# else:
entities = list()
tg_words = find_keywords.tag_text(self.add_syns_question)
for word in tg_words.split(' '):
if (
not word.startswith('*') and
find_keywords.has_tags(word, ['N', 'NPROP', 'ADJ', 'PCP'])
):
entities.append(word.split('/')[0])
for ent in entities:
ent_plural = plural_singular.get_plurals(ent, embedding_model)
if ent_plural:
question = question.replace(
ent, '[{} {}]'.format(ent, ent_plural)
)
self.add_syns_question = question
def generate_topics_rule(rejoinder, cbow, syns):
result = re.match(
r'a: (?P<label>M\d+) \((?P<words>.*)\).*?'
r'(?P<options>\^pick.*?)(?P<patterns>b:.*)',
rejoinder,
flags=re.DOTALL).groupdict()
rule_words = get_splited_words(result['words'])
rejoinders = list()
patterns = re.finditer(r'b: \((?P<words>.*?)\) (?P<action>.*?)\n',
result['patterns'])
for pat in patterns:
pat_dict = pat.groupdict()
pat_words = get_splited_words(pat_dict['words'])
# Remove stopwords words in rule keywords
words = [
w for w in pat_words if w not in rule_words and w not in stopwords
]
words = get_words_syns(words, syns)
plurals = list()
for w in words:
plural = plural_singular.get_plurals(w, cbow=cbow)
if plural:
plurals.append(plural)
rejoinders.append('a: ({words}) {action} ^end(RULE)'.format(
words='[{}]'.format(' '.join(words + plurals)),
action=pat_dict['action']))
rule = ('u: ({words}) ^refine()\n\t{rejoinders}\n\t'
'a: ()\n'
'\t\tif($go_to_menu){{^reuse(~menu.{label})}}\n'
'\t\telse {{$go_to_menu = true}}').format(
words=result['words'],
rejoinders='\n\t'.join(rejoinders),
options=result['options'],
label=result['label'])
return rule
def add_plurals_and_nouns_syns(self, embedding_model):
question = self.add_syns_question
entities = list()
tg_words = find_keywords.tag_text(self.add_syns_question)
for word in tg_words.split(' '):
if (not word.startswith('*') and find_keywords.has_tags(
word, ['N', 'NPROP', 'ADJ', 'PCP'])):
entities.append(word.split('/')[0])
for ent in entities:
syns = self.get_syns(ent)
plurals = list()
for syn in syns:
syn_plural = plural_singular.get_plurals(syn, embedding_model)
if syn_plural:
plurals.append(syn_plural)
plurals_text = ' {}'.format(' '.join(plurals)) if plurals else ''
question = question.replace(
ent, '[{}{}]'.format(' '.join(syns), plurals_text))
self.add_syns_question = question
def test_plural_singular(self):
assert 'pedidos' == plural_singular.get_plurals('pedido')
def generate_topic_menu(topics, cbow, syns):
top_names = list()
rejoinders = list()
index = 0
for topic in topics:
if topic.beauty_name:
top_names.append(topic.beauty_name)
options = list()
options_rule = list()
for rule in topic.rules:
if rule.title:
options.append(rule.title)
output = '^reuse(~{}.{})'.format(topic.name, rule.label)
options_rule.append('\n\t\tb: ({pattern}) {output}'.format(
pattern=rule.title, output=output))
# Join options
options = '\n\t\t- '.join(options)
options_text = (
'\n\t\t^pick(~all_right), aqui estão opções relacionadas a '
'"{entity}":\n\t\t- {options} \n\t\t- Menu completo - '
).format(entity=topic.beauty_name, options=options)
options_rule = ''.join(options_rule)
name = topic.beauty_name.lower()
plural = ''
# Add syns to topic name
names = list()
for subname in name.split(' '):
names.extend(get_syns(subname, syns))
for subname in names:
plural_sub = plural_singular.get_plurals(subname, cbow=cbow)
if plural_sub:
plural = plural + ' ' + plural_sub
pattern = '[{} {}]'.format(' '.join(names),
plural) if plural else name
# Remove extra spaces
pattern = pattern.replace(' ', ' ')
rej = 'a: M{num} ({pattern}){options_text}{options_rule}'.format(
num=index,
pattern=pattern,
options_text=options_text,
options_rule=options_rule,
)
rejoinders.append(rej)
index = index + 1
topics_names = '- {}'.format('\n\t- '.join(top_names))
menu = TopicMenu(topics_names,
'\n\t'.join(rejoinders),
'menu',
words='menu opções ajuda',
extra='nostay ')
rules = [generate_topics_rule(rej, cbow, syns) for rej in rejoinders]
topics = TopicTopics('', rules, 'topics')
return menu, topics