diff options
Diffstat (limited to 'archive_folder.py')
| -rwxr-xr-x | archive_folder.py | 91 |
1 files changed, 59 insertions, 32 deletions
diff --git a/archive_folder.py b/archive_folder.py index 86afaa0..7c1463f 100755 --- a/archive_folder.py +++ b/archive_folder.py @@ -16,7 +16,7 @@ import sys from datetime import date, timedelta logging.basicConfig(format='%(levelname)s:%(funcName)s:%(message)s', - stream=sys.stdout, level=logging.DEBUG) + stream=sys.stdout, level=logging.ERROR) log = logging.getLogger('imapArch') class ServerError(IOError): @@ -40,7 +40,12 @@ class Message(object): self.box = self.folder.box self.uid = uid self.msg = self.__get_body() - self.date = email.utils.parsedate_to_datetime(self.msg['Date']) + date_header = self.msg['Date'] + try: + self.date = email.utils.parsedate_to_datetime(date_header) + except TypeError: + log.warning('Cannot parse date string %s', date_header) + self.date = None self.subject = self.__get_subject() def __get_body(self): @@ -53,11 +58,21 @@ class Message(object): def __get_subject(self): out_str = '' - for raw_str, enc in email.header.decode_header(self.msg['Subject']): - if isinstance(raw_str, bytes): - out_str += raw_str.decode(enc or 'utf8') - else: - out_str += raw_str + subj_str = self.msg['Subject'] + if subj_str is not None: + for raw_str, enc in email.header.decode_header(subj_str): + if enc == 'windows-874': + enc = 'iso8859_11' + elif enc is None or ('unknown' in enc): + enc = None + if isinstance(raw_str, bytes): + try: + out_str += raw_str.decode(enc or 'utf8', 'replace') + except LookupError: + log.error('Cannot find encoding %s', enc) + raise + else: + out_str += raw_str return out_str def __str__(self): @@ -66,16 +81,17 @@ class Message(object): class Folder(object): - def __init__(self, box, fld_name, create=False): - self.box = box + def __init__(self, server, fld_name, create=False): + self.server = server + self.box = server.box self.fld_name = fld_name self.selected = False self.__create_missing = create self.folder_sep = self.__get_separator() - log.debug('self.__create_missing = %s, self.__list_folder = %s', - self.__create_missing, self.__list_folder()) - if self.__create_missing and self.__list_folder() is None: + log.debug('self.__create_missing = %s, fld_name = %s, exists %s', + self.__create_missing, fld_name, fld_name not in self.server.all_folders) + if self.__create_missing and (fld_name not in self.server.all_folders): self.__create_folder() def select(self): @@ -94,16 +110,20 @@ class Folder(object): log.debug('part = %s', part) target += '{}{}'.format(part, self.folder_sep) fld = target.strip(self.folder_sep) - if self.__list_folder(fld) is None: + log.debug('fld = %s', fld) + if fld not in self.server.all_folders: + log.debug('Creating folder %s', fld) self.box.create(fld) + self.server.all_folders.append(fld) self.box.subscribe(fld) def __list_folder(self, name=None, wildcards='*'): if name is None: - name = self.fld_name + name = '{}/'.format(self.fld_name) ok, data = self.box.list(name, wildcards) if ok != 'OK': raise ServerError('Cannot list folder %s' % self.fld_name) + log.debug('data = %s', data) return data[0] def __get_separator(self): @@ -130,7 +150,8 @@ class Folder(object): (search_type.lower(), date_str)) msgs = [] - for uid in res[0].decode().split(' '): + uids = res[0].decode().split() + for uid in uids: msgs.append(Message(self, uid)) return msgs @@ -190,7 +211,7 @@ class EmailServer(object): else config['general']['account'] self.cfg = dict(config.items(acc_name)) self.archive_root = archive_root - self.box = self.__login(**self.cfg) + self.box, self.all_folders = self.__login(**self.cfg) # self.box.debug = 4 def __login(self, host='localhost', username=None, password=None, ssl=None): @@ -204,7 +225,14 @@ class EmailServer(object): capas = data[0].decode() log.debug('capas = %s', capas) box.features_present = Capas._make(['MOVE' in capas, 'UIDPLUS' in capas]) - return box + + ok, data = box.list() + if ok != 'OK': + raise IOError('Cannot list folders!') + + folders = [FOLDER_RE.split(x.decode())[1] for x in data] + + return box, folders @staticmethod def get_config(): @@ -222,36 +250,35 @@ class EmailServer(object): :param: folder_name :param: before_date """ + print('Archiving {}'.format(folder_name)) copy_cache = {} - fld = Folder(self.box, folder_name).select() + fld = Folder(self, folder_name).select() before_str = before_date.strftime('%d-%b-%Y') for msg in fld.emails_before(before_str): - arch_folder = fld.get_archive_folder(msg, self.archive_root) - if arch_folder in copy_cache: - copy_cache[arch_folder].append(msg) - else: - copy_cache[arch_folder] = [msg] + sys.stdout.write('.') + if msg.date is not None: + arch_folder = fld.get_archive_folder(msg, self.archive_root) + if arch_folder in copy_cache: + copy_cache[arch_folder].append(msg) + else: + copy_cache[arch_folder] = [msg] + sys.stdout.write('\n') for key in copy_cache: - log.info('***** %s:', key) + log.debug('***** %s:', key) for msg in copy_cache[key]: - log.info('\tmsg: %s', str(msg)) + log.debug('\tmsg: %s', str(msg)) # Go through the cache and make moves. for arch_dir in copy_cache: msg_ids = ','.join([x.uid for x in copy_cache[arch_dir]]) log.debug('arch_dir = %s, msgs = %s', arch_dir, msg_ids) - dir = Folder(self.box, arch_dir, create=True) + dir = Folder(self, arch_dir, create=True) dir.move_messages(msg_ids) def list_folders(self): - ok, data = self.box.list() - if ok != 'OK': - raise IOError('Cannot list folders!') - - return sorted([FOLDER_RE.split(x.decode())[1] for x in data], - key=str.lower) + return sorted(self.all_folders, key=str.lower) def __enter__(self): return self |