terom@259: #!/usr/bin/env python
terom@259: 
terom@259: """
terom@268:     Import hosts from existing BIND or dhcpd files.
terom@259: """
terom@259: 
terom@268: import pvl.args
terom@259: import pvl.dns.zone
terom@259: import pvl.dhcp.config
terom@261: import pvl.ldap.args
terom@259: 
terom@263: import ipaddr
terom@263: import optparse
terom@259: import collections
terom@259: import re
terom@266: import logging; log = logging.getLogger('pvl.hosts-import')
terom@259: 
terom@259: __version__ = '0.1'
terom@259: 
terom@259: def parse_options (argv) :
terom@259:     """
terom@259:         Parse command-line arguments.
terom@259:     """
terom@259: 
terom@259:     parser = optparse.OptionParser(
terom@259:             prog        = argv[0],
terom@259:             usage       = '%prog: [options]',
terom@259:             version     = __version__,
terom@259: 
terom@259:             # module docstring
terom@259:             description = __doc__,
terom@259:     )
terom@259: 
terom@259:     # logging
terom@259:     parser.add_option_group(pvl.args.parser(parser))
terom@261:     parser.add_option_group(pvl.ldap.args.parser(parser))
terom@259: 
terom@259:     parser.add_option('-c', '--input-charset',  metavar='CHARSET',  default='utf-8', 
terom@259:             help="Encoding used for input files")
terom@259: 
terom@259:     parser.add_option('--output-charset',       metavar='CHARSET',  default='utf-8', 
terom@259:             help="Encoding used for output files")
terom@259: 
terom@259:     # input
terom@259:     parser.add_option('--import-zone-hosts',    metavar='FILE',
terom@259:             help="Load hosts from DNS zone")
terom@259: 
terom@259:     parser.add_option('--import-dhcp-hosts',    metavar='FILE',
terom@259:             help="Load hosts from DHCP config")
terom@259: 
terom@275:     parser.add_option('--zone-comments-default-owner',  action='store_const',
terom@275:             dest='zone_comments_default', const='owner',
terom@275:             help="Import DNS zone comment as owner comment")
terom@275: 
terom@275:     parser.add_option('--zone-comments-default-host',   action='store_const',
terom@275:             dest='zone_comments_default', const='host',
terom@275:             help="Import DNS zone comment as host comment")
terom@275: 
terom@261:     parser.add_option('--dump-host-comments',   action='store_true',
terom@261:             help="Dump out info on imported host comments")
terom@261: 
terom@259:     # defaults
terom@259:     parser.add_option('--hosts-domain',         metavar='DOMAIN',
terom@259:             help="Default domain for hosts")
terom@259:     
terom@259:     parser.add_option('--zone-unused',          metavar='HOST',
terom@259:             help="DNS name for unallocated hosts")
terom@259: 
terom@259:     # output
terom@259:     parser.add_option('--output-hosts',         metavar='FILE',
terom@259:             help="Output hosts file")
terom@259: 
terom@265:     parser.add_option('--output-prefix',        metavar='PREFIX',
terom@265:             help="Select hosts by ip prefix")
terom@263: 
terom@259:     # defaults
terom@259:     parser.set_defaults(
terom@259: 
terom@259:     )
terom@259:     
terom@259:     # parse
terom@259:     options, args = parser.parse_args(argv[1:])
terom@259: 
terom@259:     # apply
terom@259:     pvl.args.apply(options, argv[0])
terom@259: 
terom@259:     return options, args
terom@259: 
terom@263: def import_zone_hosts (options, file) :
terom@259:     """
terom@259:         Yield host info from zonefile records.
terom@259:     """
terom@259: 
terom@279:     for rr in pvl.dns.zone.ZoneRecord.load(file,
terom@279:             # generated hosts need to imported by hand...
terom@279:             expand_generate = False,
terom@279:     ) :
terom@259:         if options.zone_unused and rr.name == options.zone_unused :
terom@259:             log.debug("%s: skip %s", rr.name, rr)
terom@259:             continue
terom@259: 
terom@275:         elif rr.type in ('A', 'AAAA') :
terom@259:             ip, = rr.data
terom@259: 
terom@265:             yield rr.name, 'ip', ipaddr.IPAddress(ip)
terom@259: 
terom@259:             if rr.comment :
terom@260:                 yield rr.name, 'comment', rr.comment
terom@259: 
terom@275:             if rr.origin :
terom@275:                 yield rr.name, 'domain', rr.origin
terom@275: 
terom@259:         elif rr.type == 'CNAME' :
terom@259:             host, = rr.data
terom@259: 
terom@259:             yield host, 'alias', rr.name
terom@259: 
terom@259:         else :
terom@259:             log.warn("%s: unknown rr: %s", rr.name, rr)
terom@259: 
terom@263: def import_dhcp_host (options, host, items) :
terom@259:     """
terom@259:         Yield host infos from a dhcp host ... { ... }
terom@259:     """
terom@259: 
terom@259:     hostname = None
terom@259:     ethernet = []
terom@259:     fixed_address = None
terom@259: 
terom@259:     for item in items :
terom@259:         item, args = item[0], item[1:]
terom@259: 
terom@259:         if item == 'hardware' :
terom@259:             _ethernet, ethernet = args
terom@259:             assert _ethernet == 'ethernet'
terom@259:         elif item == 'fixed-address' :
terom@259:             fixed_address, = args
terom@259:         elif item == 'option' :
terom@259:             option = args.pop(0)
terom@259: 
terom@259:             if option == 'host-name' :
terom@259:                 hostname, = args
terom@259:             else :
terom@259:                 log.warn("host %s: ignore unknown option: %s", host, option)
terom@259:         else :
terom@259:             log.warn("host %s: ignore unknown item: %s", host, item)
terom@259: 
terom@259:     # determine hostname
terom@264:     suffix = None
terom@264: 
terom@264:     if '-' in host :
terom@259:         hostname, suffix = host.rsplit('-', 1)
terom@259:     else :
terom@259:         hostname = host
terom@259: 
terom@264:     if fixed_address and not re.match(r'\d+\.\d+\.\d+.\d+', fixed_address) :
terom@264:         hostname, domain = fixed_address.split('.', 1)
terom@264: 
terom@264:     #if suffix :
terom@268:     #    yield hostname, ('ethernet', suffix), ethernet
terom@264:     if hostname and ethernet :
terom@259:         yield hostname, 'ethernet', ethernet
terom@264:     else :
terom@264:         log.warn("%s: no hostname/ethernet: %s/%s", host, hostname, ethernet)
terom@259: 
terom@263: def import_dhcp_hosts (options, blocks) :
terom@259:     """
terom@259:         Process hosts from a parsed block
terom@259:     """
terom@259: 
terom@259:     for block, items, blocks in blocks :
terom@259:         log.info("%s", block)
terom@259:         
terom@259:         block, args = block[0], block[1:]
terom@259: 
terom@259:         if block == 'group' :
terom@263:             for info in import_dhcp_hosts(options, blocks) :
terom@259:                 yield info
terom@259:         elif block == 'host' :
terom@259:             host, = args
terom@259: 
terom@259:             try :
terom@263:                 for info in import_dhcp_host(options, host, items) :
terom@259:                     yield info
terom@259:             except ValueError as error :
terom@259:                 log.warn("%s: invalid host: %s", host, error)
terom@259:         else:
terom@259:             log.warn("ignore unknown block: %s", block)
terom@259: 
terom@263: def import_dhcp_conf (options, file) :
terom@259:     items, blocks = pvl.dhcp.config.DHCPConfigParser().load(file)
terom@259: 
terom@259:     for item in items :
terom@259:         item, args = item[0], item[1:]
terom@259: 
terom@259:         if item == 'include' :
terom@259:             include, = args
terom@263:             for info in import_dhcp_conf(options, pvl.args.apply_file(include)) :
terom@259:                 yield info
terom@259:         else :
terom@259:             log.warn("ignore unknown item: %s", item)
terom@259:     
terom@263:     for info in import_dhcp_hosts(options, blocks) :
terom@259:         yield info
terom@259: 
terom@260: ZONE_COMMENTS = (
terom@260:         re.compile(r'(?P<owner>[^/]+)\s*-\s+(?P<host>.+)'),
terom@260:         re.compile(r'(?P<group>.+?)\s*/\s*(?P<owner>.+)\s+[/-]\s+(?P<host>.+)'),
terom@260:         re.compile(r'(?P<group>.+?)\s*/\s*(?P<owner>.+)\s+[(]\s*(?P<host>.+)[)]'),
terom@260:         re.compile(r'(?P<group>.+?)\s*/\s*(?P<owner>.+)'),
terom@260: )
terom@260: 
terom@260: ZONE_OWNER_MAIL = re.compile(r'(?P<owner>.*?)\s*<(?P<mail>.+?)>')
terom@260: 
terom@260: def process_zone_comment (options, hostname, comment) :
terom@260:     """
terom@260:         Attempt to parse a host comment field... :D
terom@260: 
terom@260:         Yields (field, value) bits
terom@260:     """
terom@260: 
terom@260:     for regex in ZONE_COMMENTS :
terom@260:         match = regex.match(comment)
terom@260: 
terom@260:         if match :
terom@275:             matches = match.groupdict()
terom@279: 
terom@279:             log.info("%s: matched comment: %s", hostname, comment)
terom@260:             break
terom@260:     else :
terom@275:         if options.zone_comments_default :
terom@279:             log.info("%s: default comment: %s", hostname, comment)
terom@275:             matches = { options.zone_comments_default: comment }
terom@275:         else :
terom@275:             log.warn("%s: unknown comment: %s", hostname, comment)
terom@275:             return
terom@260:     
terom@260:     owner = matches.pop('owner', None)
terom@260:     
terom@260:     if owner :
terom@260:         mail_match = ZONE_OWNER_MAIL.match(owner)
terom@260: 
terom@260:         if mail_match :
terom@260:             mail_matches = mail_match.groupdict()
terom@260:             
terom@260:             owner = mail_matches['owner']
terom@261:             yield 'mail', mail_matches['mail'].strip()
terom@275:     
terom@275:         yield 'owner', owner.strip()
terom@260: 
terom@260:     for field, value in matches.iteritems() :
terom@260:         if value :
terom@261:             yield field, value.strip()
terom@261: 
terom@263: NONE_OWNERS = set((
terom@263:     u'tech',
terom@263:     u'atk',
terom@263:     u'toimisto',
terom@263: ))
terom@261: 
terom@263: def process_host_owner_ldap (options, host, info) :
terom@261:     """
terom@261:         Yield guesses for user from LDAP.
terom@261:     """
terom@261: 
terom@261:     if info.get('mail') :
terom@261:         for user in options.ldap.users.filter(
terom@261:                 { 'mailLocalAddress': info['mail'] },
terom@261:                 { 'uid': info['mail'] },
terom@261:         ) :
terom@263:             yield user, None
terom@261: 
terom@261:     if info.get('group') and info.get('owner') :
terom@261:         groups = options.ldap.groups.filter(cn=info['group'])
terom@261: 
terom@261:         for group in groups :
terom@261:             for user in options.ldap.users.filter({
terom@261:                 'gidNumber': group['gidNumber'],
terom@261:                 'cn': info['owner'],
terom@261:             }) :
terom@263:                 yield user, group
terom@261: 
terom@261:     if info.get('owner') :
terom@261:             for user in options.ldap.users.filter({
terom@261:                 'cn': info['owner'],
terom@261:             }) :
terom@263:                 yield user, None
terom@263: 
terom@263: def process_host_owner (options, host, info) :
terom@263:     """
terom@263:         Return (owner, comment) for host based on info, or None.
terom@263:     """
terom@263: 
terom@275:     owner = info.get('owner')
terom@275: 
terom@275:     if owner and owner.lower() in NONE_OWNERS :
terom@264:         return False
terom@263:     
terom@263:     # from ldap?
terom@263:     for ldap in process_host_owner_ldap(options, host, info) :
terom@263:         user, group = ldap
terom@263:         
terom@263:         if not group :
terom@263:             # get group from ldap
terom@263:             group = options.ldap.users.group(user)
terom@263:         
terom@263:         return user['uid'], u"{group} / {user}".format(
terom@263:                 user    = user.getunicode('cn'),
terom@263:                 group   = group.getunicode('cn'),
terom@263:         )
terom@261: 
terom@261: def process_host_comments (options, host, info) :
terom@261:     """
terom@261:         Process host fields from comment.
terom@261: 
terom@261:         Attempts to find owner from LDAP..
terom@261:     """
terom@261: 
terom@261:     log.debug("%s: %s", host, info)
terom@260:     
terom@263:     owner = process_host_owner(options, host, info) 
terom@261: 
terom@264:     if owner is False :
terom@264:         # do not mark any owner
terom@264:         pass
terom@264: 
terom@264:     elif owner :
terom@263:         owner, comment = owner
terom@263:         
terom@263:         log.info("%s: %s (%s)", host, owner, comment)
terom@263:         
terom@268:         yield 'comment-owner', comment
terom@261:         yield 'owner', owner,
terom@264: 
terom@279:     elif 'group' in info or 'owner' in info :
terom@264:         log.warn("%s: unknown owner: %s", host, info)
terom@268:         yield 'comment-owner', "{group} / {owner}".format(
terom@264:                 group   = info.get('group', ''),
terom@264:                 owner   = info.get('owner', ''),
terom@264:         )
terom@262:     
terom@262:     if info.get('host') :
terom@268:         yield 'comment-host', info['host']
terom@260: 
terom@260: def process_hosts_comments (options, import_hosts) :
terom@260:     """
terom@260:         Parse out comments from host imports..
terom@260:     """
terom@260: 
terom@260:     for host, field, value in import_hosts :
terom@260:         if field != 'comment':
terom@260:             yield host, field, value
terom@260:             continue
terom@260: 
terom@260:         fields = dict(process_zone_comment(options, host, value))
terom@261:         
terom@261:         if options.dump_host_comments :
terom@261:             print u"{host:20} {comment:80} = {group:15} / {owner:20} <{mail:20}> / {hostinfo}".format(
terom@261:                     host        = host,
terom@261:                     comment     = value,
terom@261:                     group       = fields.get('group', ''),
terom@261:                     owner       = fields.get('owner', ''),
terom@261:                     mail        = fields.get('mail', ''),
terom@261:                     hostinfo    = fields.get('host', ''),
terom@261:             ).encode('utf-8')
terom@261:         
terom@260: 
terom@261:         for field, value in process_host_comments(options, host, fields) :
terom@261:             yield host, field, value
terom@263: 
terom@263: def apply_hosts_import (options) :
terom@259:     """
terom@263:         Import host infos from given files.
terom@259:     """
terom@259: 
terom@263:     if options.import_zone_hosts:
terom@263:         for info in import_zone_hosts(options,
terom@268:                 pvl.args.apply_file(options.import_zone_hosts, 'r', options.input_charset)) :
terom@263:             yield info
terom@263:     
terom@263:     if options.import_dhcp_hosts:
terom@263:         for info in import_dhcp_conf(options,
terom@268:                 pvl.args.apply_file(options.import_dhcp_hosts, 'r', options.input_charset)) :
terom@263:             yield info
terom@263:        
terom@263: def import_hosts (options) :
terom@263:     """
terom@263:         Import hosts from dns/dhcp.
terom@263:     """
terom@263: 
terom@263:     import_hosts = apply_hosts_import(options)
terom@263:     import_hosts = process_hosts_comments(options, import_hosts)
terom@263:     
terom@263:     # gather
terom@259:     hosts = collections.defaultdict(lambda: collections.defaultdict(list))
terom@259: 
terom@259:     for host, field, value in import_hosts :
terom@259:         hosts[host][field].append(value)
terom@259:     
terom@259:     return hosts.iteritems()
terom@259: 
terom@265: def process_export_hosts (options, hosts) :
terom@265:     if options.output_prefix :
terom@265:         prefix = ipaddr.IPNetwork(options.output_prefix)
terom@265:     else :
terom@265:         prefix = None
terom@265: 
terom@263:     for host, fields in hosts :
terom@263:         ip = fields.get('ip')
terom@265:         
terom@265:         # sort by IP
terom@265:         if ip :
terom@268:             sort = ip[0]
terom@265:         else :
terom@265:             # fake, to sort correctly
terom@265:             sort = ipaddr.IPAddress(0)
terom@265:         
terom@265:         # select
terom@265:         if prefix:
terom@265:             if not (ip and ip in prefix) :
terom@265:                 continue
terom@263: 
terom@265:         yield sort, host, fields
terom@263: 
terom@263: def export_hosts (options, hosts) :
terom@263:     """
terom@263:         Export hosts to file.
terom@263:     """
terom@263: 
terom@263:     file = pvl.args.apply_file(options.output_hosts, 'w', options.output_charset)
terom@263: 
terom@265:     # filter + sort
terom@265:     hosts = [(host, fields) for sort, host, fields in sorted(process_export_hosts(options, hosts))]
terom@263: 
terom@263:     for host, fields in hosts :
terom@268:         for comment in fields.get('comment-host', ()):
terom@263:             print >>file, u"# {comment}".format(comment=comment)
terom@263: 
terom@263:         print >>file, u"[{host}]".format(host=host)
terom@263:         
terom@263:         for field, fmt in (
terom@263:                 ('ip',              None),
terom@263:                 ('ethernet',        None),
terom@268:                 ('owner',           u"\t{field:15} = {value} # {fields[comment-owner][0]}"),
terom@268:                 ('alias',           None),
terom@263:         ) :
terom@263:             if not fmt :
terom@263:                 fmt = u"\t{field:15} = {value}"
terom@268:             
terom@268:             values = fields.get(field, ())
terom@263: 
terom@268:             if len(values) > 1 :
terom@268:                 for index, value in enumerate(values, 1) :
terom@268:                     print >>file, fmt.format(
terom@268:                             field   = "{field}.{index}".format(field=field, index=index),
terom@268:                             value   = value,
terom@268:                             fields  = fields
terom@268:                     )
terom@268:             elif len(values) > 0 :
terom@268:                 value, = values
terom@263:                 print >>file, fmt.format(field=field, value=value, fields=fields)
terom@263:         
terom@263:         print >>file
terom@259: 
terom@259: def main (argv) :
terom@259:     options, args = parse_options(argv)
terom@261: 
terom@261:     options.ldap = pvl.ldap.args.apply(options)
terom@259:     
terom@259:     if args :
terom@259:         # direct from file
terom@259:         hosts = pvl.args.apply_files(args, 'r', options.input_charset)
terom@259:     else :
terom@259:         # import
terom@263:         hosts = import_hosts(options)
terom@259:    
terom@259:     # output
terom@259:     if options.output_hosts :
terom@263:         export_hosts(options, hosts)
terom@259: 
terom@259: if __name__ == '__main__':
terom@259:     pvl.args.main(main)