def test_output():
post_URL = request.args.get("postURL")
eg_subreddit = fu.subreddit(post_URL)
postAtt = fu.rcommentLink(post_URL)
eg_postTitle = postAtt.title
eg_comment = [comment.body for comment in postAtt.comments][0]
commentCreated = int([comment.created_utc for comment in postAtt.comments][0])
postCreated = postAtt.created_utc
timeDiffC = fu.timePresent(round(commentCreated - postCreated))
timeDiffCur = fu.timePresent(round(time.time() - commentCreated))
comLength = fu.cLength(fu.comLength(eg_comment))
sentiment = ""
sentiment = fu.sentiment(eg_comment)
features = fu.features(eg_subreddit, post_URL, eg_comment, commentCreated)
eg_prob = ""
eg_prob = fu.probability(features)
eg_score = [comment.score for comment in postAtt.comments][0]
eg_commentForm = fu.commentSent(eg_comment)
return render_template(
"testOutput.html",
eg_postTitle=eg_postTitle,
eg_commentForm=eg_commentForm,
eg_score=eg_score,
eg_prob=eg_prob,
sentiment=sentiment,
comLength=comLength,
timeDiffCur=timeDiffCur,
timeDiffC=timeDiffC,
eg_comment=eg_comment,
post_URL=post_URL,
eg_subreddit=eg_subreddit,
)
def output():
post = request.args.get("post")
comment = request.args.get("comment")
postAtt = fu.rcommentLink(post)
postTitle = postAtt.title
the_subreddit = ""
the_subreddit = fu.subreddit(post)
the_post = ""
the_post = post
your_comment = ""
your_comment = comment
features = fu.features(the_subreddit, post, comment, time.time())
timePost = ""
timePost = fu.timePresent(features[4])
comLength = ""
comLength = fu.cLength(features[5])
sentiment = ""
sentiment = fu.sentiment(your_comment)
prob = ""
prob = fu.probability(features)
commentForm = fu.commentSent(your_comment)
return render_template(
"output.html",
postTitle=postTitle,
commentForm=commentForm,
prob=prob,
timePost=timePost,
comLength=comLength,
sentiment=sentiment,
the_subreddit=the_subreddit,
the_post=the_post,
your_comment=your_comment,
)
model_lstm = tf.keras.models.load_model('model_lstm.h5')
with open('full_word.dat','rb') as fp:
full_word = pickle.load(fp)
with open('naver_count.dat', 'rb') as fp:
naver_count = pickle.load(fp)
with open('sentiment_pipe.dat', 'rb') as fp:
pipe = pickle.load(fp)
with open('sentiment_word2vec.dat', 'rb') as fp:
model = pickle.load(fp)
from konlpy.tag import Okt
import numpy as np
okt =Okt()
mapped_review, review_p = review_process(review)
r = model_lstm.predict(mapped_review)
pos=0
neg=0
for _ in range(len(review_p)):
try:
w2v_proba = sentiment(review_p[_])
if w2v_proba >= 0.6 and r[_][0] ==0:
neg+=1
else:
pos+=1
except:
pass
print(f'긍정적 리뷰 비율 : {pos/(pos+neg)*100}, 부정적 리뷰 비율: {neg/(pos+neg)*100}')
def get_sentiment(chat_id):
return fnc.sentiment(get_chat(chat_id))
