aboutsummaryrefslogblamecommitdiffstats
path: root/archive_folder.py
blob: 4e0f8eb8c806a5fb1a73791053d730f7b9eec280 (plain) (tree)
1
2
3
4
5
6
7
8
9
10
                      

                                                                        
               
                  

                   


                   
              
              
         
         
          
                                    

                                                                    
                                                          

                                   
                           

        

                           
 

                            
 
                                                            
                                      
                                       
 
 

                                             


                                    

                                    





                                                                      


                                           
                                                                   







                                                                














                                                                           


                      

                                                                     


                     


                                                       
                                
                             
                                      
                                                
 


                                                                                           
                                  

                     


                                                                        

                            
 
                              
                             
                                             
                                                

                               
                                        
                                                          
                                               


                                                    
                                    
                                                   
                                       
 

                                                      
                                              


                                                                      
                                    









                                                                        


                                                 


                                                 


                                                                            




                                                                                   
 
                 

                                      
                                           
                   
 

                                                         
 
                                             


                                                                        



                                                                              





                                                                          




                                                                       
                                                                    




                                                                       
                                                                          







                                                             
                                                                    




                                                                       
                                                                          


                                                            
 
 
                          
                                                                       
                                  

                                                       
                                               
                                        
                                                             

                              


                                                                                
                                                
                      


                                                                  
                                   
                                
                                      
                                                                                 







                                                                
 
                 






                                                                            







                                                              
                                                 
                       
                                                

                                                     







                                                                            

                              
                                       
                                       
                                                
 




                                                                     
                                                     
                                      
 
                           
                                                      
 


                        


                                          
                        


                          
                                            
                                    
                                           

                                                                                           


                                                                    
                                           


                                                                                    

                                                                                 
 
                            



                                   
 

                                         

                                                           
                            
                                                       

                                        

                                                         
                                                                                         
                                                       
#!/usr/bin/env python3
# note http://docs.python.org/lib/module-doctest.html
# resp. file:///usr/share/doc/python-docs-*/html/lib/module-doctest.html
import argparse
import collections
import configparser
import email
import email.header
import email.utils
import locale
import logging
import imaplib
import os
import re
import sys
from datetime import date, timedelta

logging.basicConfig(format='%(levelname)s:%(funcName)s:%(message)s',
                    stream=sys.stdout, level=logging.INFO)
log = logging.getLogger('imapArch')

class ServerError(IOError):
    pass

class FolderError(IOError):
    pass

class MessageError(IOError):
    pass

Capas = collections.namedtuple('Capas', ['MOVE', 'UIDPLUS'])
SEP_RE = re.compile(r'\s+"([/.])"\s+')
FOLDER_RE = re.compile(r'\s+"[/.]"\s+')


class Message(object):
    """Abstraction over one email message."""
    def __init__(self, folder, uid):
        self.folder = folder
        self.box = self.folder.box
        self.uid = uid
        self.msg = self.__get_body()
        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):
        typ, data = self.box.uid('FETCH', '%s (RFC822)' % self.uid)
        if typ == 'OK':
            return email.message_from_bytes(data[0][1])
        else:
            raise MessageError('Cannot parse message %s:\n%s!' %
                               (self.uid, data[0][1]))

    def __get_subject(self):
        out_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):
        return "%s in %s: %s (%s)" % (self.uid, self.folder.fld_name,
                                      self.subject, self.date)


class Folder(object):
    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, 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):
        ok, _ = self.box.select(mailbox=self.fld_name)
        if ok != 'OK':
            raise FolderError('Cannot select folder %s' % self.fld_name)
        self.selected = True
        return self

    def __create_folder(self):
        sep = self.folder_sep
        split_name = self.fld_name.split(sep)
        log.debug('split_name = %s', split_name)
        target = ''
        for part in split_name:
            log.debug('part = %s', part)
            target += '{}{}'.format(part, self.folder_sep)
            fld = target.strip(self.folder_sep)
            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 = '{}/'.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):
        # type: () -> None
        """
        Find current folder hierarchy separator, either from the current
        folder, or if it doesn't exist from INBOX.
        """
        data = self.__list_folder()

        if data is None:
            data = self.__list_folder('""', '""')

        parse_data = SEP_RE.search(data.decode())
        if parse_data is not None:
            return parse_data.group(1)
        else:
            raise ServerError('Cannot find folder separator from %s' % data)

    def __emails_search(self, search_type, date_str):
        ok, res = self.box.uid('SEARCH', '%s %s' % (search_type.upper(), date_str))
        if ok != 'OK':
            raise MessageError('SEARCH %s %s failed!' %
                               (search_type.lower(), date_str))

        msgs = []
        uids = res[0].decode().split()
        for uid in uids:
            msgs.append(Message(self, uid))
        return msgs

    def emails_before(self, before_str):
        return self.__emails_search('BEFORE', before_str)

    def get_archive_folder(self, msg, aroot):
        raw_name = self.folder_sep.join((aroot, msg.date.strftime("%Y"),
                                         self.fld_name))
        return raw_name.strip('/')

    def move_messages(self, messages):
        assert self.selected == False, 'Target folder should not be selected.'
        if self.box.features_present.MOVE:
            # We cannot move self.box.uid('MOVE', messages, self.fld_name)
            # because IMAP4_SSL.uid has a protection against unknown
            # commands. https://bugs.python.org/issue33336
            typ, dat = self.box._simple_command('UID', 'MOVE',
                                                 messages, self.fld_name)
            ok, data = self.box._untagged_response(typ, dat, 'MOVE')
            log.debug('MOVE ok = %s, data = %s', ok, data)
            if ok != 'OK':
                raise FolderError('Cannot move messages to folder %s' %
                                  self.fld_name)
        elif self.box.features_present.UIDPLUS:
            ok, data = self.box.uid('COPY', messages, self.fld_name)
            log.debug('COPY ok = %s, data = %s', ok, data)
            if ok != 'OK':
                raise FolderError('Cannot copy messages to folder %s' %
                                  self.fld_name)
            ok, data = self.box.uid('STORE',
                                    r'+FLAGS.SILENT (\DELETED)', messages)
            log.debug('STORE ok = %s, data = %s', ok, data)
            if ok != 'OK':
                raise FolderError('Cannot delete messages-')
            ok, data = self.box.uid('EXPUNGE', messages)
            log.debug('EXPUNGE ok = %s, data = %s', ok, data)
            if ok != 'OK':
                raise FolderError('Cannot expunge messages.')
        else:
            ok, data = self.box.uid('COPY', messages, self.fld_name)
            log.debug('COPY ok = %s, data = %s', ok, data)
            if ok != 'OK':
                raise FolderError('Cannot copy messages to folder %s' %
                                  self.fld_name)
            ok, data = self.box.uid('STORE',
                                    r'+FLAGS.SILENT (\DELETED)', messages)
            log.debug('STORE ok = %s, data = %s', ok, data)
            if ok != 'OK':
                raise FolderError('Cannot delete messages-')


class EmailServer(object):
    def __init__(self, serverKey=None, archive_root=None, verbosity=0):
        config = self.get_config()
        acc_name = serverKey if serverKey is not None \
            else config['general']['account']
        self.cfg = dict(config.items(acc_name))
        self.archive_root = archive_root
        self.box, self.all_folders = self.__login(**self.cfg)
        if verbosity > 1:
            self.box.debug = 4

    def __login(self, host='localhost', username=None, password=None, ssl=None):
        box = imaplib.IMAP4_SSL(host=host)
        ok, data = box.login(username, password)
        if ok != 'OK':
            raise ServerError('Cannot login with credentials %s' %
                              str((host, username, password,)))

        ok, data = box.capability()
        capas = data[0].decode()
        log.debug('capas = %s', capas)
        box.features_present = Capas._make(['MOVE' in capas, 'UIDPLUS' in capas])

        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():
        # In case the configuration file is missing, only empty list will be
        # returned
        config = configparser.ConfigParser()
        config.read(os.path.expanduser('~/.config/imap_archiver.cfg'))
        return config

    def archive_folder(self, folder_name, before_date):
        # type: (str, datetime.date) -> None
        """
        Archive one folder to the proper archiving positioins.

        :param: folder_name
        :param: before_date
        """
        print('Archiving {}'.format(folder_name))
        copy_cache = {}
        fld = Folder(self, folder_name).select()
        before_str = before_date.strftime('%d-%b-%Y')
        for msg in fld.emails_before(before_str):
            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.debug('***** %s:', key)
            for msg in copy_cache[key]:
                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, arch_dir, create=True)
            dir.move_messages(msg_ids)

    def list_folders(self):
        return sorted(self.all_folders, key=str.lower)

    def __enter__(self):
        return self

    def __exit__(self, *args):
        if args != (None, None, None):
            log.warning('args = %s', args)
        self.box.close()


if __name__ == '__main__':
    locale.setlocale(locale.LC_ALL, 'en_US')
    argp = argparse.ArgumentParser()
    subps = argp.add_subparsers(dest='cmd')
    subp_servers = subps.add_parser('servers', help='List available servers (by keywords)')
    subp_list = subps.add_parser('list', help='List all folders')
    subp_arc = subps.add_parser('archive', help='Archive folder[s]')
    subp_arc.add_argument('-d', type=int, default=14, dest='days',
                          help='How old messages we should keep')
    subp_arc.add_argument('-s', '--server',
                          help='Symbolic name of the server to be used')
    subp_arc.add_argument('folder', help='Folder which should be archived')
    subp_arc.add_argument('archive', help='Root folder to store annual archives to')
    argp.add_argument('-v', action='count', dest='verbosity', default=0,
                      help='Verbosity of the operation (-vv shows IMAP chatter)')

    args = argp.parse_args()
    log.debug('args = %s', args)

    if args.verbosity > 0:
        log.setLevel(logging.DEBUG)

    if args.cmd == 'servers':
        config = EmailServer.get_config()
        sects = set(config.keys()) - {'DEFAULT', 'general'}
        print('Available servers:\n%s' % tuple(sects))
    elif args.cmd == 'list':
        myStore = EmailServer(verbosity=args.verbosity)
        folders = myStore.list_folders()
        print('\n'.join(folders))
    else:
        before = date.today() - timedelta(days=args.days)
        with EmailServer(args.server, args.archive, verbosity=args.verbosity) as myStore:
            myStore.archive_folder(args.folder, before)