def gen_recipients_from_cache(options, progress=False):
if not os.path.exists(cache_messages_path):
return {}
max_age = options['max_age']
dstore = shelve.open(cache_messages_path, flag='r', protocol=2)
recipients = {}
n_messages = len(dstore)
log.info('cache only, %d messages', n_messages)
if progress:
pstatus = log.PercentStatus(n_messages, prefix=' ')
for identifier in dstore:
if progress:
pstatus.inc()
pstatus.output()
msg = scan.Message()
msg.from_dict_only(dstore[identifier])
if options['skip_multiple_recipients'] and len(msg.to_emails) > 1:
continue
if options['exclude_mails_to_me'] and filter_any(config.is_this_me, msg.to_emails):
continue
if options['only_include_mails_from_me'] and not config.is_this_me(msg.from_email):
continue
if max_age >= 0 and msg.age > max_age:
continue
if msg.to_emails_str not in recipients:
r = scan.Recipient(msg.to_emails, msg)
recipients[msg.to_emails_str] = r
else:
recipients[msg.to_emails_str].add(msg)
if progress:
pstatus.finish()
dstore.close()
return recipients
def gen_recipients(mailboxes, options, use_cache=True, clean_cache=False, progress=False):
base = os.path.dirname(cache_messages_path)
if not os.path.exists(base):
os.makedirs(base)
if not os.path.exists(cache_messages_path):
# if there is no cache, it doesn't need to be cleaned
clean_cache = False
max_age = options['max_age']
flag = 'r' if clean_cache else 'c'
dstore = shelve.open(cache_messages_path, flag=flag, protocol=2)
dstore_write = shelve.open(cache_messages_tmp_path, flag='n', protocol=2) if clean_cache else dstore
recipients = {}
n_mailboxes = len(mailboxes)
for i, mb in enumerate(mailboxes):
if progress:
n_messages = len(mb)
log.info('[%d/%d] %s: %d messages', i+1, n_mailboxes, mb.path, n_messages)
pstatus = log.PercentStatus(n_messages, prefix=' ')
for msg in mb.messages():
if progress:
pstatus.inc()
pstatus.output()
d = dstore.get(msg.identifier)
if use_cache and d and not msg.has_changed(d):
msg.from_dict(d)
if clean_cache:
dstore_write[msg.identifier] = d
else:
if not msg.parse_header():
continue
d = {}
msg.parse_body()
if msg.identifier:
msg.to_dict(d)
dstore_write[msg.identifier] = d
if options['skip_multiple_recipients'] and len(msg.to_emails) > 1:
continue
if options['exclude_mails_to_me'] and filter_any(config.is_this_me, msg.to_emails):
continue
if options['only_include_mails_from_me'] and not config.is_this_me(msg.from_email):
continue
if max_age >= 0 and msg.age > max_age:
continue
if msg.to_emails_str not in recipients:
r = scan.Recipient(msg.to_emails, msg)
recipients[msg.to_emails_str] = r
else:
recipients[msg.to_emails_str].add(msg)
if progress:
pstatus.finish()
dstore.close()
if clean_cache:
dstore_write.close()
os.rename(cache_messages_tmp_path, cache_messages_path)
return recipients
body = self.body
if self.greeting:
body = self._re_greeting.sub(u'', body, 1).lstrip('\n')
mb = MessageBody(body, self._re_quote, self._re_smileys)
if len(mb.interleaved) + len(mb.bottom) > len(mb.top):
self.posting_style = u'inline'
init_guess_language()
words_to_guess = u' '.join(re.split(r'\W+', u'%s %s' % (mb.unquoted, mb.quoted))[:20])
guessed_language = guessLanguage(words_to_guess)
self.language = guessed_language if guessed_language != 'UNKNOWN' else ''
if not config.get('personalize_mailinglists') and filter_any(config.is_mailinglist, self.to_emails):
self.greeting = u''
return True
if not mb.unquoted:
self.greeting = u''
match = self._re_goodbye.search(mb.top.rstrip('\n'))
if not match:
match = self._re_goodbye.search(mb.bottom.rstrip('\n'))
if match:
body = self._re_goodbye.sub(u'', body, 1).strip('\n')
if body:
self.goodbye = match.group(1)
return True
