def get_emails(verbose=True):
picklefile = 'pickled_emails.pickle'
if os.path.isfile(picklefile):
with open(picklefile,'r') as load_from:
emails = cPickle.load(load_from)
else:
emails = parseEmails('.',printInfo=verbose)
with open(picklefile,'w') as store_to:
cPickle.dump(emails,store_to)
return emails
def get_sequence_data():
txtfile = 'sequence_data.txt'
infofile = 'data_info.txt'
if os.path.isfile(txtfile):
features,labels = read_sequences(txtfile)
label_names = read_info(infofile)
else:
emails = parseEmails('.')
features,labels,words,labelVals = get_keras_features(emails,as_matrix=False)
write_sequences(txtfile,features,labels)
write_info(infofile,labelVals)
num_labels = max(labels)+1
return features,labels,label_names
def get_sequence_data():
txtfile = 'sequence_data.txt'
infofile = 'data_info.txt'
if os.path.isfile(txtfile):
features,labels = read_sequences(txtfile)
label_names = read_info(infofile)
else:
emails = parseEmails('.')
features,labels,words,labelVals = get_keras_features(emails,as_matrix=False)
write_sequences(txtfile,features,labels)
write_info(infofile,labelVals)
num_labels = max(labels)+1
return features,labels,label_names