developer.intra2net.com Git - imap-restore-mail/blob - file_iterator.py

   1 '''
   2 restore-mail-inject.py - Tool to inject mails via IMAP
   3
   4 Copyright (c) 2012 Intra2net AG
   5 '''
   6
   7 import os
   8 import re
   9 import logging
  10
  11 MAIL_FILENAME = re.compile("^[0-9]+\.$")
  12 MBOXFILE_LINE = re.compile("^(.*?)\t(?:\d )?default[\t ](.*)$")
  13 ACL_STRING = re.compile("^(.*?)[\t ](.*?)\t(.*)$")
  14
  15 class FileIterator:
  16     """This class iterates through the e-mail files."""
  17
  18     # class attributes
  19     # mailboxes created during file traversal
  20     created_mailboxes = None
  21     # mailboxes to update during file traversal
  22     acl_mailboxes = None
  23
  24     def __init__(self):
  25         """Creates a connection and a user session."""
  26         self.created_mailboxes = []
  27         self.acl_mailboxes = []
  28         return
  29
  30     def __del__(self):
  31         """Closes the connection and the user session."""
  32         return
  33
  34     def _message_read(self, filename):
  35         """Retrieves a message from the message file."""
  36         try:
  37             with open(filename, "r") as msgfile:
  38                 message = msgfile.read()
  39         except IOError:
  40             logging.warning("Could not open the e-mail file %s", filename)
  41             raise
  42         return message
  43
  44     def load_mailbox_list(self, mboxlistfile = ""):
  45         mboxdb = {}
  46         if mboxlistfile != "":
  47             try:
  48                 with open(mboxlistfile, 'r') as acl_file:
  49                     for line in acl_file:
  50                         acls = {}
  51                         linedata = MBOXFILE_LINE.match(line).groups()
  52                         #!!! test this condition
  53                         if len(linedata) == 0:
  54                             logging.warning("Illegal line in mailbox list dump: %s" % line)
  55                             continue
  56                         key = linedata[0]
  57                         aclstr = linedata[1]
  58
  59                         # loop through acl rights string and build dictionary of users and rights
  60                         while(aclstr != ""):
  61                             acldata = ACL_STRING.match(aclstr).groups()
  62                             if len(acldata) == 0:
  63                                 logging.error("Illegal acl string in mailbox list dump: %s" % line)
  64                                 continue
  65                             aclstr = acldata[2]
  66                             acls[acldata[0]]=acldata[1]
  67
  68                         mboxdb[key] = acls
  69             except IOError:
  70                 logging.warning("Could not open mboxlist file %s" % mboxlistfile)
  71         return mboxdb
  72
  73     def load_mails(self, filepath, mailpath):
  74         """Loads all e-mails from file hierarchy.
  75         This recursive generator always returns a tuple of
  76         the next found (e-mail, mailbox to store, internaldate)."""
  77         logging.debug("Entered directory %s -> %s" % (filepath, mailpath))
  78         try:
  79             filepath = os.path.abspath(filepath)
  80             os.chdir(filepath)
  81         except OSError:
  82             logging.warning("Can't open the directory %s" % filepath)
  83             return
  84         # mark mailboxes that should be created
  85         self.created_mailboxes.append(mailpath)
  86         subpaths = os.listdir(filepath)
  87         for subpath in subpaths:
  88             if subpath == "." or subpath == "..":
  89                 continue
  90             new_filepath = filepath + "/" + subpath
  91             if (os.path.isfile(new_filepath)):
  92                 if os.path.getsize(new_filepath) == 0:
  93                     logging.info("Skipping empty file %s" % subpath)
  94                 else:
  95                     if MAIL_FILENAME.match(subpath):
  96                         logging.info("Injecting file %s" % subpath)
  97                         try:
  98                             message = self._message_read(new_filepath)
  99                             # suggest file modification date for internaldate
 100                             yield (message, mailpath, os.path.getmtime(new_filepath))
 101                         except IOError:
 102                             logging.warning("Could not retrieve mail from the file %s" % new_filepath)
 103             else:
 104                 if os.path.isdir(new_filepath):
 105                     # cyrus ^ vs . storage replacement
 106                     subpath = subpath.replace("^", ".")
 107                     new_mailpath = mailpath + "/" + subpath
 108                     logging.debug("Inserting mails from directory %s into mailbox %s" % (new_filepath, new_mailpath))
 109                     # load_mails($mboxdbref, $origuser, $targetuser)
 110                     rcrs_generator = self.load_mails(new_filepath, new_mailpath)
 111                     # you enter the generator in the for loop
 112                     for rcr in rcrs_generator:
 113                         yield rcr
 114                     logging.debug("Done with directory %s and mailbox %s" % (new_filepath, new_mailpath))
 115         # mark mailboxes that need acl update
 116         self.acl_mailboxes.append(mailpath)
 117         return