def postprocess(sent):
morphs = api.morph(sent)['morphs']
query = list()
for morph in morphs:
query.append(u'{}:{}'.format(morph['norm_surface'], morph['pos']))
morphs = api.rewrite_morph(rewrite_rule, query)['morphs']
sent = u''
for morph in morphs[1:-1]:
sent += u':'.join(morph.split(u':')[:-1])
return sent
def preprocess(text):
#TODO とりあえず最初の文の最初の名詞を利用, なければ最初の形態素
sentences = api.sentences(text)['sentences']
morphs = api.morph(sentences[0])['morphs']
for morph in morphs:
if '名詞' in morph['pos']:
seed = morph
break
else:
seed = morphs[1]
return seed
def generate_seed(text):
max_length = 0
for sentence in api.sentences(text)['sentences']:
morphs = api.morph(sentence)['morphs']
for morph in morphs:
if u'固有' in morph['pos'] or u'名詞' in morph['pos']:
seed = morph
break
else:
seed = morphs[1]
return seed
def scenario_based(text):
sent = api.sentences(text)
text = []
for s in sent['sentences']:
morphs = api.morph(s)
query = list()
for morph in morphs['morphs']:
query.append(u'{}:{}'.format(morph['surface'], morph['pos']))
texts = api.trigger(scenario_file,query)
for t in texts['texts']:
text.append(t)
if len(text)>0:
r = random.randint(0,len(text)-1)
return text[r]
return None
print json.dumps(reps, ensure_ascii=False, indent=4)
for rep in reps["replies"]:
print "======================================================"
if reps["grade"] == 0:
scenario_file = "scenario_c09.txt"
elif reps["grade"] == 1:
scenario_file = "scenario_c09.txt"
else:
scenario_file = "scenario_c09.txt"
sent = api.sentences(rep["text"])
text = []
for s in sent["sentences"]:
print "-------------------------------------------------------"
print s
morphs = api.morph(s)
print json.dumps(morphs, ensure_ascii=False, indent=4)
query = list()
for morph in morphs["morphs"]:
query.append(u"{}:{}".format(morph["surface"], morph["pos"]))
print json.dumps(query, ensure_ascii=False, indent=4)
texts = api.trigger(scenario_file, query)
print json.dumps(texts, ensure_ascii=False, indent=4)
for t in texts["texts"]:
text.append(t)
print t
if len(text) > 0:
r = random.randint(0, len(text) - 1)
print "++++++++++++++++++++++++++++++++++++++"
print str(len(text))
print text[r]
reply_text = u''
for morph in api.markov_chain(seed)['morphs'][1:-1]:
reply_text += u':'.join(morph.split(u':')[:-1])#morph.split(':')[0]
return postprocess(reply_text)
def reply_one(mention_id, user_name, text):
reply_text = twitter_based(text)
api.send_reply(mention_id, user_name, reply_text)
if __name__ == '__main__':
#api = get_api()
api = api.API('https://52.68.75.108', 'secret', 'js2015cps')
#rewrite_rule = 'team2_rewrite_{}.txt'.format(current_state)
rewrite_rule = 'rule_test.txt'
text = "どうして人間は魂の在処にこだわるの？"
for morph in api.morph(text)['morphs']:
print morph['norm_surface'] + "\t" + morph['pos']
print markov_based(text)
print twitter_based(text)
#result = api.get_reply()
#current_state = result['grade']
#replies = result['replies']
#for sentence in api.sentences('おはよう。こんにちは。')['sentences']:
# print sentence
#for tweet in api.get_reply()['replies']:
# print tweet['mention_id']
# print tweet['user_name']
# print tweet['text']
