cut_word.py

#!/usr/bin/python
from posixpath import split
from konlpy.tag import Mecab, Kkma, Hannanum, Okt, _komoran
from konlpy.utils import pprint
import pandas as pd
from flask import Flask,render_template,redirect,url_for
import sys
import requests
import json

#comments = ['나는 학생이다', '진짜 별로다', '아 에어컨 춥다...ㅠㅠ', '행복행', 'hello']  # 다운에게 받는 댓글 리스트
split_list = []  # 유나에게 넘기는 한 댓글 분석한 리스트
preference = {"like": 0, "neutral": 0, "dislike": 0}
word = -2  # like=1 , neutral=0 dislike=-1
global li
li= 0
global ne
ne= 0
global di
di= 0
text=[]
count={}
preferenceCounts=[]
client_id = "nwnijty7r7"
client_secret = "QxLuVWoGt2eYkSnfGjMFd0bgkwfYYX6C2Fk9dbmL"
url="https://naveropenapi.apigw.ntruss.com/sentiment-analysis/v1/analyze"
headers = {
    "X-NCP-APIGW-API-KEY-ID": client_id,
    "X-NCP-APIGW-API-KEY": client_secret,
    "Content-Type": "application/json"
}
#-*- codig: utf-8 -*-
def analysis(comment):      
    data = {
    "content": comment
    }
    print(json.dumps(data, indent=4, sort_keys=True))
    response = requests.post(url, data=json.dumps(data), headers=headers)
    rescode = response.status_code
    if(rescode == 200):
        preference_check(json.loads(response.text)['document']['sentiment'])
        print (response.text)
    else:
        print("Error : "+response.text)

def korean_split(comment):  # 댓글 하나씩 분석해서 어간 list 만들어서 리턴
    tokenizer = Okt()
    split_list = tokenizer.nouns(comment)
    return split_list


def word_count(comments):  # 댓글리스트 모든 댓글 카운트 & top 5개 뽑기        


    for comment in comments:  # 댓글 하나씩 넘, 모든 댓글 다 넘길 때까지 반복
        tmp=korean_split(comment)
        text.append(tmp)        

    for x in text:
        for y in x:
            count[y]=count.get(y,0)+1  

    keyword = pd.Series(count)
    sortings=keyword.sort_values(ascending=False)
    keywords = list(sortings.index)
    return keywords

def preference_check(word):
    if (word == 'positive'):
        global li
        li+=1
    elif (word == 'neutral'):
        global ne
        ne+=1
    else:
        global di
        di+=1


def cut_word(comments):
    word_count(comments)  # 키워드 총 분석
    for comment in comments:  # 댓글 하나씩 넘, 모든 댓글 다 넘길 때까지 반복
        analysis(comment)

    keyword=word_count(comments)
    preference = {}
    preference["like"] = li
    preference["neutral"] = ne
    preference["dislike"] = di
    print(count)
    print(keyword)
    print(preference)
    preferenceCounts=list(preference.values())
    return keyword, preferenceCounts, preference
    # 효정에게 keyword, preference 전달