def main(imap_folder_name):
"Mostly copy-pasted form drymail-train; probably need to be cleaned up"
with config.IMAP_connection() as im:
for msg_num, msg in each_imap_message(im,imap_folder_name):
cat, prob = Category.classify(str(msg))
print "%s(%f): %s" % (cat.name, prob,
msg.get('Subject', '(none)').replace('\n',' '))
def do_supertrain(self, cat):
"""train the current message on the given category until the
classifier says that's what it is"""
train_count = 0
while True:
train_count += 1
(newcat, prob) = Category.classify(str(self.current_message))
if newcat == cat:
break
print "Training %d... (%s %.2f%%)" % (train_count, newcat.name, prob*100)
cat.train(str(self.current_message))
self._classify()
def _classify(self):
self.current_category = self.current_probability = None
if not Category.all():
print "Can't classify because I have no categories"
return
if not self.current_message:
return
cat, prob = Category.classify(str(self.current_message))
self.current_category = cat
self.current_probability = prob
return self.current_category
