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
  27         self.created_mailboxes = []
  28         self.acl_mailboxes = []
  29
  30         return
  31
  32     @classmethod
  33     def _message_read(cls, filename):
  34         """Retrieves a message from the message file."""
  35
  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
  43         return message
  44
  45     @classmethod
  46     def load_mailbox_list(cls, mboxlistfile = ""):
  47         """Load the list of mailboxes and acl rights for each from file."""
  48
  49         mboxdb = {}
  50         if mboxlistfile != "":
  51             try:
  52                 with open(mboxlistfile, 'r') as acl_file:
  53                     for line in acl_file:
  54                         acls = {}
  55                         linedata = MBOXFILE_LINE.match(line).groups()
  56                         if len(linedata) == 0:
  57                             logging.warning("Illegal line in mailbox list dump: %s", line)
  58                             continue
  59                         key = linedata[0]
  60                         aclstr = linedata[1]
  61
  62                         # loop through acl rights string and build dictionary of users and rights
  63                         while(aclstr != ""):
  64                             acldata = ACL_STRING.match(aclstr).groups()
  65                             if len(acldata) == 0:
  66                                 logging.error("Illegal acl string in mailbox list dump: %s", line)
  67                                 continue
  68                             aclstr = acldata[2]
  69                             acls[acldata[0]] = acldata[1]
  70
  71                         mboxdb[key] = acls
  72             except IOError:
  73                 logging.warning("Could not open mboxlist file %s", mboxlistfile)
  74
  75         return mboxdb
  76
  77     def load_mails(self, filepath, mailpath):
  78         """Loads all e-mails from file hierarchy.
  79         This recursive generator always returns a tuple of
  80         the next found (e-mail, mailbox to store, internaldate)."""
  81
  82         logging.debug("Entered directory %s -> %s", filepath, mailpath)
  83         try:
  84             filepath = os.path.abspath(filepath)
  85             os.chdir(filepath)
  86         except OSError:
  87             logging.warning("Can't open the directory %s", filepath)
  88             return
  89         # mark mailboxes that should be created
  90         self.created_mailboxes.append(mailpath)
  91         subpaths = os.listdir(filepath)
  92         for subpath in subpaths:
  93             if subpath == "." or subpath == "..":
  94                 continue
  95             new_filepath = filepath + "/" + subpath
  96             if (os.path.isfile(new_filepath)):
  97                 if os.path.getsize(new_filepath) == 0:
  98                     logging.info("Skipping empty file %s", subpath)
  99                 else:
 100                     if MAIL_FILENAME.match(subpath):
 101                         logging.info("Injecting file %s", subpath)
 102                         try:
 103                             message = self._message_read(new_filepath)
 104                             # suggest file modification date for internaldate
 105                             yield (message, mailpath, os.path.getmtime(new_filepath))
 106                         except IOError:
 107                             logging.warning("Could not retrieve mail from the file %s", new_filepath)
 108             else:
 109                 if os.path.isdir(new_filepath):
 110                     # cyrus ^ vs . storage replacement
 111                     subpath = subpath.replace("^", ".")
 112                     new_mailpath = mailpath + "/" + subpath
 113                     logging.debug("Inserting mails from directory %s into mailbox %s", new_filepath, new_mailpath)
 114                     # load_mails($mboxdbref, $origuser, $targetuser)
 115                     rcrs_generator = self.load_mails(new_filepath, new_mailpath)
 116                     # you enter the generator in the for loop
 117                     for rcr in rcrs_generator:
 118                         yield rcr
 119                     logging.debug("Done with directory %s and mailbox %s", new_filepath, new_mailpath)
 120         # mark mailboxes that need acl update
 121         self.acl_mailboxes.append(mailpath)
 122
 123         return