def train(self, corpus_dir):
'''Train method the filter uses to teach according to !truth.txt file.'''
self.train_corpus_dir = corpus_dir
# test if the !truth.txt exists, else exit (no error raised)
truth_file = os.path.join(corpus_dir, '!truth.txt')
if os.path.exists(truth_file)==False:
self.trained = False
return
# create TrainingCorpus object for better training handling
Corpus = TrainingCorpus(corpus_dir)
# get spams/hams senders, return_paths and subjects
for fname in Corpus.truth_dict:
(sender, subject) = Corpus.parse_email(fname)
if Corpus.truth_dict[fname] == self.neg_tag:
self.save_ham_header(sender, subject)
else:
self.save_spam_header(sender, subject)
