extract-essays.py

# coding: utf-8
import email, imaplib, chardet, pickle

m = imaplib.IMAP4_SSL('imap.gmail.com', 993)
m.login('nb@nikolay.bg', 'тринайсетзелениголиорангутанагонятгоркатапеперудка')
m.select('&BD8ENQRCBDA- &BDcEMAQ0BDAERwQw-') # пета задача
typ, data = m.search(None, 'ALL')
essays = []
for num in data[0].split():
    print 'Getting %s' % num
    typ, data = m.fetch(num, '(RFC822)')
    raw_text = data[0][1]
    multipart = email.message_from_string(raw_text)
    message = multipart if not multipart.is_multipart() else multipart.get_payload()[0]
    text = message.get_payload(decode=True)
    if chardet.detect(text)['encoding'] == 'windows-1251':        
        text = unicode(text, 'windows-1251').encode('utf-8')
    essays.append({'from': multipart.get('From'), 'text': text})
    print text
pickle.dump(essays, open('essays', 'wb'))
m.close()
m.logout()