from sklearn.feature_extraction.text import CountVectorizer
import json
import pandas
import numpy

corpus_path = 'data/training/training-data.csv'

# prepare training data for bow (corpus)
X_training = []
dataframe = pandas.read_csv(corpus_path, header=None)
for i in xrange(len(dataframe[0])):
    X_training.append(dataframe[0][i])
sentences = numpy.array(X_training)

# create bow vocabulary
vectorizer = CountVectorizer()
vectorizer.fit_transform(sentences).todense()

# save vocabulary to json file
with open ('vocabulary.json', 'w') as vocabFile:
    json.dump(vectorizer.vocabulary_ , vocabFile)

print "vocabulary is saved"