translate_sentistrength_output_to_binary.py

# 2018 - Isaias S. Silva
# This script converts the sentistrength output to a CSV one with zeros and ones, so we can use in a two class classifier.

import requests
import time
import datetime
import sys
import json

#Attention to the file names bellow
output_file = open('output_results_processed.txt', 'w', encoding='utf-8') 
output_file.close()
output_file = open('output_results_processed.txt', 'a', encoding='utf-8') 

with open('output+results.txt','r', encoding='utf-8') as myFile:
    cont = 1
    line = myFile.readline()
    while line:
        phrase = line.split('	')
        #print(phrase[0]) ## frase
        #print(phrase[1]) ## positivo
        #print(phrase[2]) ## negativo
        n1 = abs(int(phrase[1]))
        n2 = abs(int(phrase[2]))
        #ignore empty lines
        if(phrase[0].strip()==""):
            print(str(cont)+ " Vazio: "+ phrase[0])
            line = myFile.readline()  
            cont+=1
            time.sleep(0.1)
            continue
        if n1 == n2:
            print('neutro')
            output_file.write(phrase[0]+","+"0\n")
        elif n1 > n2:
            print('positivo!')
            output_file.write(phrase[0]+","+"0\n")
        else:
            print('negativo!')
            output_file.write(phrase[0]+","+"1\n")
            
        line = myFile.readline()  
        cont+=1      
        #time.sleep(0.2)

output_file.close()
print('Completed!')



#quando meu primo fica aqui ele só assisti parabéns galinha      1       -1