lab2 nbc import texts

lab2importtexts.py

import os
import numpy as np

imported_data = []

dir_paths = ['./data/movie_review/neg/', './data/movie_review/pos/']

for dir_path in dir_paths:
    for filename in os.listdir(dir_path):
        with open(dir_path + "/" + filename, 'r') as content_file:
            imported_data.append([content_file.read(), dir_paths.index(dir_path)])
    print "READ", len(os.listdir(dir_path)), "FILES from", dir_path


X = np.array(imported_data)
print X