0ef24b1937 2011-04-06 c.kworr@d4daf: #!/usr/bin/env python
0ef24b1937 2011-04-06 c.kworr@d4daf:
0ef24b1937 2011-04-06 c.kworr@d4daf: from __future__ import division, print_function, unicode_literals
0ef24b1937 2011-04-06 c.kworr@d4daf:
0ef24b1937 2011-04-06 c.kworr@d4daf: import gevent.monkey
0ef24b1937 2011-04-06 c.kworr@d4daf: gevent.monkey.patch_all()
0ef24b1937 2011-04-06 c.kworr@d4daf:
0ef24b1937 2011-04-06 c.kworr@d4daf: import fcntl, gevent.core, gevent.pool, gevent.queue, gevent.socket, os, psycopg2, re, sys
0ef24b1937 2011-04-06 c.kworr@d4daf:
0ef24b1937 2011-04-06 c.kworr@d4daf: # //inclusion start
0ef24b1937 2011-04-06 c.kworr@d4daf: # Copyright (C) 2010 Daniele Varrazzo <daniele.varrazzo@gmail.com>
0ef24b1937 2011-04-06 c.kworr@d4daf: # and licensed under the MIT license:
0ef24b1937 2011-04-06 c.kworr@d4daf:
0ef24b1937 2011-04-06 c.kworr@d4daf: def gevent_wait_callback(conn, timeout=None):
0ef24b1937 2011-04-06 c.kworr@d4daf: """A wait callback useful to allow gevent to work with Psycopg."""
0ef24b1937 2011-04-06 c.kworr@d4daf: while 1:
0ef24b1937 2011-04-06 c.kworr@d4daf: state = conn.poll()
0ef24b1937 2011-04-06 c.kworr@d4daf: if state == psycopg2.extensions.POLL_OK:
0ef24b1937 2011-04-06 c.kworr@d4daf: break
0ef24b1937 2011-04-06 c.kworr@d4daf: elif state == psycopg2.extensions.POLL_READ:
0ef24b1937 2011-04-06 c.kworr@d4daf: gevent.socket.wait_read(conn.fileno(), timeout=timeout)
0ef24b1937 2011-04-06 c.kworr@d4daf: elif state == psycopg2.extensions.POLL_WRITE:
0ef24b1937 2011-04-06 c.kworr@d4daf: gevent.socket.wait_write(conn.fileno(), timeout=timeout)
0ef24b1937 2011-04-06 c.kworr@d4daf: else:
0ef24b1937 2011-04-06 c.kworr@d4daf: raise psycopg2.OperationalError("Bad result from poll: %r" % state)
0ef24b1937 2011-04-06 c.kworr@d4daf:
0ef24b1937 2011-04-06 c.kworr@d4daf: if not hasattr(psycopg2.extensions, 'set_wait_callback'):
0ef24b1937 2011-04-06 c.kworr@d4daf: raise ImportError("support for coroutines not available in this Psycopg version (%s)" % psycopg2.__version__)
0ef24b1937 2011-04-06 c.kworr@d4daf: psycopg2.extensions.set_wait_callback(gevent_wait_callback)
0ef24b1937 2011-04-06 c.kworr@d4daf:
0ef24b1937 2011-04-06 c.kworr@d4daf: # //inclusion end
0ef24b1937 2011-04-06 c.kworr@d4daf:
39b97ced92 2011-06-05 c.kworr@d4daf: # this classes processes config file and substitutes default values
39b97ced92 2011-06-05 c.kworr@d4daf: class Config:
39b97ced92 2011-06-05 c.kworr@d4daf: __slots__ = frozenset(['_config', '_default', '_section', 'options'])
39b97ced92 2011-06-05 c.kworr@d4daf: _default = {
39b97ced92 2011-06-05 c.kworr@d4daf: 'log': {
39b97ced92 2011-06-05 c.kworr@d4daf: 'silent': 'no',
39b97ced92 2011-06-05 c.kworr@d4daf: },
39b97ced92 2011-06-05 c.kworr@d4daf: 'database': {
39b97ced92 2011-06-05 c.kworr@d4daf: 'host': 'localhost',
39b97ced92 2011-06-05 c.kworr@d4daf: 'database': 'squidTag',
39b97ced92 2011-06-05 c.kworr@d4daf: },}
39b97ced92 2011-06-05 c.kworr@d4daf:
39b97ced92 2011-06-05 c.kworr@d4daf: # function to read in config file
39b97ced92 2011-06-05 c.kworr@d4daf: def __init__(self):
39b97ced92 2011-06-05 c.kworr@d4daf: import ConfigParser, optparse, os
39b97ced92 2011-06-05 c.kworr@d4daf:
39b97ced92 2011-06-05 c.kworr@d4daf: parser = optparse.OptionParser()
39b97ced92 2011-06-05 c.kworr@d4daf: parser.add_option('-c', '--config', dest = 'config',
39b97ced92 2011-06-05 c.kworr@d4daf: help = 'config file location', metavar = 'FILE',
39b97ced92 2011-06-05 c.kworr@d4daf: default = '/usr/local/etc/squid-tagger.conf')
39b97ced92 2011-06-05 c.kworr@d4daf: parser.add_option('-d', '--dump', dest = 'dump',
39b97ced92 2011-06-05 c.kworr@d4daf: help = 'dump database', action = 'store_true', metavar = 'bool',
39b97ced92 2011-06-05 c.kworr@d4daf: default = False)
39b97ced92 2011-06-05 c.kworr@d4daf: parser.add_option('-f', '--flush-database', dest = 'flush_db',
39b97ced92 2011-06-05 c.kworr@d4daf: help = 'flush previous database on load', default = False,
39b97ced92 2011-06-05 c.kworr@d4daf: action = 'store_true', metavar = 'bool')
39b97ced92 2011-06-05 c.kworr@d4daf: parser.add_option('-l', '--load', dest = 'load',
39b97ced92 2011-06-05 c.kworr@d4daf: help = 'load database', action = 'store_true', metavar = 'bool',
39b97ced92 2011-06-05 c.kworr@d4daf: default = False)
39b97ced92 2011-06-05 c.kworr@d4daf: parser.add_option('-D', '--dump-conf', dest = 'dump_conf',
39b97ced92 2011-06-05 c.kworr@d4daf: help = 'dump filtering rules', default = False, metavar = 'bool',
39b97ced92 2011-06-05 c.kworr@d4daf: action = 'store_true')
39b97ced92 2011-06-05 c.kworr@d4daf: parser.add_option('-L', '--load-conf', dest = 'load_conf',
39b97ced92 2011-06-05 c.kworr@d4daf: help = 'load filtering rules', default = False, metavar = 'bool',
39b97ced92 2011-06-05 c.kworr@d4daf: action = 'store_true')
39b97ced92 2011-06-05 c.kworr@d4daf:
39b97ced92 2011-06-05 c.kworr@d4daf: (self.options, args) = parser.parse_args()
39b97ced92 2011-06-05 c.kworr@d4daf:
39b97ced92 2011-06-05 c.kworr@d4daf: assert os.access(self.options.config, os.R_OK), "Fatal error: can't read {}".format(self.options.config)
39b97ced92 2011-06-05 c.kworr@d4daf:
39b97ced92 2011-06-05 c.kworr@d4daf: self._config = ConfigParser.ConfigParser()
39b97ced92 2011-06-05 c.kworr@d4daf: self._config.readfp(open(self.options.config))
39b97ced92 2011-06-05 c.kworr@d4daf:
39b97ced92 2011-06-05 c.kworr@d4daf: # function to select config file section or create one
39b97ced92 2011-06-05 c.kworr@d4daf: def section(self, section):
39b97ced92 2011-06-05 c.kworr@d4daf: if not self._config.has_section(section):
39b97ced92 2011-06-05 c.kworr@d4daf: self._config.add_section(section)
39b97ced92 2011-06-05 c.kworr@d4daf: self._section = section
39b97ced92 2011-06-05 c.kworr@d4daf:
39b97ced92 2011-06-05 c.kworr@d4daf: # function to get config parameter, if parameter doesn't exists the default
39b97ced92 2011-06-05 c.kworr@d4daf: # value or None is substituted
39b97ced92 2011-06-05 c.kworr@d4daf: def __getitem__(self, name):
39b97ced92 2011-06-05 c.kworr@d4daf: if not self._config.has_option(self._section, name):
39b97ced92 2011-06-05 c.kworr@d4daf: if self._section in self._default:
39b97ced92 2011-06-05 c.kworr@d4daf: if name in self._default[self._section]:
39b97ced92 2011-06-05 c.kworr@d4daf: self._config.set(self._section, name, self._default[self._section][name])
39b97ced92 2011-06-05 c.kworr@d4daf: else:
39b97ced92 2011-06-05 c.kworr@d4daf: self._config.set(self._section, name, None)
39b97ced92 2011-06-05 c.kworr@d4daf: else:
39b97ced92 2011-06-05 c.kworr@d4daf: self._config.set(self._section, name, None)
39b97ced92 2011-06-05 c.kworr@d4daf: return(self._config.get(self._section, name))
39b97ced92 2011-06-05 c.kworr@d4daf:
39b97ced92 2011-06-05 c.kworr@d4daf: # initializing and reading in config file
39b97ced92 2011-06-05 c.kworr@d4daf: config = Config()
39b97ced92 2011-06-05 c.kworr@d4daf:
39b97ced92 2011-06-05 c.kworr@d4daf: # wrapper around syslog, can be muted
39b97ced92 2011-06-05 c.kworr@d4daf: class Logger(object):
39b97ced92 2011-06-05 c.kworr@d4daf: __slots__ = frozenset(['_syslog'])
39b97ced92 2011-06-05 c.kworr@d4daf:
39b97ced92 2011-06-05 c.kworr@d4daf: def __init__(self):
39b97ced92 2011-06-05 c.kworr@d4daf: config.section('log')
39b97ced92 2011-06-05 c.kworr@d4daf: if config['silent'] == 'yes':
39b97ced92 2011-06-05 c.kworr@d4daf: self._syslog = None
39b97ced92 2011-06-05 c.kworr@d4daf: else:
39b97ced92 2011-06-05 c.kworr@d4daf: import syslog
39b97ced92 2011-06-05 c.kworr@d4daf: self._syslog = syslog
39b97ced92 2011-06-05 c.kworr@d4daf: self._syslog.openlog(str('squidTag'))
39b97ced92 2011-06-05 c.kworr@d4daf:
39b97ced92 2011-06-05 c.kworr@d4daf: def info(self, message):
39b97ced92 2011-06-05 c.kworr@d4daf: if self._syslog != None:
39b97ced92 2011-06-05 c.kworr@d4daf: self._syslog.syslog(self._syslog.LOG_INFO, message)
39b97ced92 2011-06-05 c.kworr@d4daf:
39b97ced92 2011-06-05 c.kworr@d4daf: def notice(self, message):
39b97ced92 2011-06-05 c.kworr@d4daf: if self._syslog != None:
39b97ced92 2011-06-05 c.kworr@d4daf: self._syslog.syslog(self._syslog.LOG_NOTICE, message)
39b97ced92 2011-06-05 c.kworr@d4daf:
39b97ced92 2011-06-05 c.kworr@d4daf: logger = Logger()
39b97ced92 2011-06-05 c.kworr@d4daf:
0ef24b1937 2011-04-06 c.kworr@d4daf: # tiny wrapper around a file to make reads from it geventable
0ef24b1937 2011-04-06 c.kworr@d4daf: # or should i move this somewhere?
0ef24b1937 2011-04-06 c.kworr@d4daf:
0ef24b1937 2011-04-06 c.kworr@d4daf: class FReadlineQueue(gevent.queue.Queue):
0ef24b1937 2011-04-06 c.kworr@d4daf: # storing file descriptor, leftover
0ef24b1937 2011-04-06 c.kworr@d4daf: __slots__ = frozenset(['_fd', '_tail'])
0ef24b1937 2011-04-06 c.kworr@d4daf:
0ef24b1937 2011-04-06 c.kworr@d4daf: def __init__(self, fd):
0ef24b1937 2011-04-06 c.kworr@d4daf: # initialising class
0ef24b1937 2011-04-06 c.kworr@d4daf: gevent.queue.Queue.__init__(self)
0ef24b1937 2011-04-06 c.kworr@d4daf: # storing file descriptor
0ef24b1937 2011-04-06 c.kworr@d4daf: self._fd = fd
0ef24b1937 2011-04-06 c.kworr@d4daf: # using empty tail
0ef24b1937 2011-04-06 c.kworr@d4daf: self._tail = ''
0ef24b1937 2011-04-06 c.kworr@d4daf: # setting up event
0ef24b1937 2011-04-06 c.kworr@d4daf: self._install_wait()
0ef24b1937 2011-04-06 c.kworr@d4daf:
0ef24b1937 2011-04-06 c.kworr@d4daf: def _install_wait(self):
0ef24b1937 2011-04-06 c.kworr@d4daf: fileno = self._fd.fileno()
0ef24b1937 2011-04-06 c.kworr@d4daf: # putting file to nonblocking mode
0ef24b1937 2011-04-06 c.kworr@d4daf: fcntl.fcntl(fileno, fcntl.F_SETFL, fcntl.fcntl(fileno, fcntl.F_GETFL) | os.O_NONBLOCK)
0ef24b1937 2011-04-06 c.kworr@d4daf: # installing event handler
0ef24b1937 2011-04-06 c.kworr@d4daf: gevent.core.read_event(fileno, self._wait_helper)
0ef24b1937 2011-04-06 c.kworr@d4daf:
0ef24b1937 2011-04-06 c.kworr@d4daf: def _wait_helper(self, ev, evtype):
0ef24b1937 2011-04-06 c.kworr@d4daf: # reading one buffer from stream
0ef24b1937 2011-04-06 c.kworr@d4daf: buf = self._fd.read(4096)
0ef24b1937 2011-04-06 c.kworr@d4daf: # splitting stream by line ends
39b97ced92 2011-06-05 c.kworr@d4daf: rows = buf.decode('l1').split('\n')
0ef24b1937 2011-04-06 c.kworr@d4daf: # adding tail to the first element if there is some tail
0ef24b1937 2011-04-06 c.kworr@d4daf: if len(self._tail) > 0:
0ef24b1937 2011-04-06 c.kworr@d4daf: rows[0] = self._tail + rows[0]
0ef24b1937 2011-04-06 c.kworr@d4daf: # popping out last (incomplete) element
0ef24b1937 2011-04-06 c.kworr@d4daf: self._tail = rows.pop(-1)
0ef24b1937 2011-04-06 c.kworr@d4daf: # dropping all complete elements to the queue
0ef24b1937 2011-04-06 c.kworr@d4daf: for row in rows:
0ef24b1937 2011-04-06 c.kworr@d4daf: self.put_nowait(row)
39b97ced92 2011-06-05 c.kworr@d4daf: logger.info('request: ' + row)
0ef24b1937 2011-04-06 c.kworr@d4daf: if len(buf) > 0:
0ef24b1937 2011-04-06 c.kworr@d4daf: # no EOF, reinstalling event handler
0ef24b1937 2011-04-06 c.kworr@d4daf: gevent.core.read_event(self._fd.fileno(), self._wait_helper)
0ef24b1937 2011-04-06 c.kworr@d4daf: else:
0ef24b1937 2011-04-06 c.kworr@d4daf: # EOF found, sending EOF to queue
0ef24b1937 2011-04-06 c.kworr@d4daf: self.put_nowait(None)
0ef24b1937 2011-04-06 c.kworr@d4daf:
0ef24b1937 2011-04-06 c.kworr@d4daf: stdin = FReadlineQueue(sys.stdin)
0ef24b1937 2011-04-06 c.kworr@d4daf:
ddbf5288b9 2010-11-03 c.kworr@d4daf: # wrapper around database
0ef24b1937 2011-04-06 c.kworr@d4daf: class tagDB(object):
0ef24b1937 2011-04-06 c.kworr@d4daf: __slots__ = frozenset(['_cursor', '_db'])
ddbf5288b9 2010-11-03 c.kworr@d4daf:
ddbf5288b9 2010-11-03 c.kworr@d4daf: def __init__(self):
ddbf5288b9 2010-11-03 c.kworr@d4daf: config.section('database')
d2c7ba18a4 2011-09-14 c.kworr@d4daf: if config['host'] == None:
d2c7ba18a4 2011-09-14 c.kworr@d4daf: self._db = psycopg2.connect(
d2c7ba18a4 2011-09-14 c.kworr@d4daf: database = config['database'],
d2c7ba18a4 2011-09-14 c.kworr@d4daf: user = config['user'],
d2c7ba18a4 2011-09-14 c.kworr@d4daf: password = config['password']
d2c7ba18a4 2011-09-14 c.kworr@d4daf: )
d2c7ba18a4 2011-09-14 c.kworr@d4daf: else:
d2c7ba18a4 2011-09-14 c.kworr@d4daf: self._db = psycopg2.connect(
d2c7ba18a4 2011-09-14 c.kworr@d4daf: database = config['database'],
d2c7ba18a4 2011-09-14 c.kworr@d4daf: host = config['host'],
d2c7ba18a4 2011-09-14 c.kworr@d4daf: user = config['user'],
d2c7ba18a4 2011-09-14 c.kworr@d4daf: password = config['password']
d2c7ba18a4 2011-09-14 c.kworr@d4daf: )
0ef24b1937 2011-04-06 c.kworr@d4daf: self._cursor = self._db.cursor()
0ef24b1937 2011-04-06 c.kworr@d4daf:
0ef24b1937 2011-04-06 c.kworr@d4daf: def _field_names(self):
0ef24b1937 2011-04-06 c.kworr@d4daf: names = []
0ef24b1937 2011-04-06 c.kworr@d4daf: for record in self._cursor.description:
0ef24b1937 2011-04-06 c.kworr@d4daf: names.append(record.name)
0ef24b1937 2011-04-06 c.kworr@d4daf: return(names)
ddbf5288b9 2010-11-03 c.kworr@d4daf:
ddbf5288b9 2010-11-03 c.kworr@d4daf: def check(self, site, ip_address):
39b97ced92 2011-06-05 c.kworr@d4daf: self._cursor.execute("select * from (select redirect_url, regexp from site_rule where site <@ tripdomain(%s) and netmask >>= %s order by array_length(site, 1) desc) a group by redirect_url, regexp", [site, ip_address])
0ef24b1937 2011-04-06 c.kworr@d4daf: return(self._cursor.fetchall())
ddbf5288b9 2010-11-03 c.kworr@d4daf:
ddbf5288b9 2010-11-03 c.kworr@d4daf: def dump(self):
0ef24b1937 2011-04-06 c.kworr@d4daf: self._cursor.execute("select untrip(site) as site, tag::text, regexp from urls order by site, tag")
0ef24b1937 2011-04-06 c.kworr@d4daf: return(self._field_names(), self._cursor.fetchall())
ddbf5288b9 2010-11-03 c.kworr@d4daf:
ddbf5288b9 2010-11-03 c.kworr@d4daf: def load(self, data):
0ef24b1937 2011-04-06 c.kworr@d4daf: if config.options.flush_db:
0ef24b1937 2011-04-06 c.kworr@d4daf: self._cursor.execute('delete from urls;')
0ef24b1937 2011-04-06 c.kworr@d4daf: bundle = []
0ef24b1937 2011-04-06 c.kworr@d4daf: for row in data:
0ef24b1937 2011-04-06 c.kworr@d4daf: if len(row) == 2:
0ef24b1937 2011-04-06 c.kworr@d4daf: bundle.append([row[0], row[1], None])
0ef24b1937 2011-04-06 c.kworr@d4daf: else:
0ef24b1937 2011-04-06 c.kworr@d4daf: bundle.append([row[0], row[1], row[2]])
0ef24b1937 2011-04-06 c.kworr@d4daf: self._cursor.executemany("insert into urls (site, tag, regexp) values (tripdomain(%s), %s, %s)", bundle)
0ef24b1937 2011-04-06 c.kworr@d4daf: self._cursor.execute("update urls set regexp = NULL where regexp = ''")
0ef24b1937 2011-04-06 c.kworr@d4daf: self._db.commit()
ddbf5288b9 2010-11-03 c.kworr@d4daf:
ddbf5288b9 2010-11-03 c.kworr@d4daf: def load_conf(self, csv_data):
0ef24b1937 2011-04-06 c.kworr@d4daf: self._cursor.execute('delete from rules;')
0ef24b1937 2011-04-06 c.kworr@d4daf: bundle = []
0ef24b1937 2011-04-06 c.kworr@d4daf: for row in csv_data:
0ef24b1937 2011-04-06 c.kworr@d4daf: bundle.append([row[0], row[1], int(row[2]), int(row[3]), row[4], row[5], row[6]])
0ef24b1937 2011-04-06 c.kworr@d4daf: self._cursor.executemany("insert into rules (netmask, redirect_url, from_weekday, to_weekday, from_time, to_time, tag) values (%s::text::cidr, %s, %s, %s, %s::text::time, %s::text::time, %s::text::text[])", bundle)
0ef24b1937 2011-04-06 c.kworr@d4daf: self._db.commit()
ddbf5288b9 2010-11-03 c.kworr@d4daf:
ddbf5288b9 2010-11-03 c.kworr@d4daf: def dump_conf(self):
0ef24b1937 2011-04-06 c.kworr@d4daf: self._cursor.execute("select netmask, redirect_url, from_weekday, to_weekday, from_time, to_time, tag::text from rules")
0ef24b1937 2011-04-06 c.kworr@d4daf: return(self._field_names(), self._cursor.fetchall())
ddbf5288b9 2010-11-03 c.kworr@d4daf:
ddbf5288b9 2010-11-03 c.kworr@d4daf: # abstract class with basic checking functionality
0ef24b1937 2011-04-06 c.kworr@d4daf: class Checker(object):
0ef24b1937 2011-04-06 c.kworr@d4daf: __slots__ = frozenset(['_db', '_log', '_queue', '_request'])
ddbf5288b9 2010-11-03 c.kworr@d4daf:
39b97ced92 2011-06-05 c.kworr@d4daf: def __init__(self, queue, logger):
ddbf5288b9 2010-11-03 c.kworr@d4daf: self._db = tagDB()
39b97ced92 2011-06-05 c.kworr@d4daf: self._log = logger
ddbf5288b9 2010-11-03 c.kworr@d4daf: self._log.info('started\n')
ddbf5288b9 2010-11-03 c.kworr@d4daf: self._request = re.compile('^([0-9]+)\ (http|ftp):\/\/([-\w.:]+)\/([^ ]*)\ ([0-9.]+)\/(-|[\w\.]+)\ (-|\w+)\ (-|GET|HEAD|POST).*$')
0ef24b1937 2011-04-06 c.kworr@d4daf: self._queue = queue
ddbf5288b9 2010-11-03 c.kworr@d4daf:
ddbf5288b9 2010-11-03 c.kworr@d4daf: def process(self, id, site, ip_address, url_path, line = None):
0ef24b1937 2011-04-06 c.kworr@d4daf: #self._log.info('trying {}\n'.format(site))
ddbf5288b9 2010-11-03 c.kworr@d4daf: result = self._db.check(site, ip_address)
ddbf5288b9 2010-11-03 c.kworr@d4daf: reply = None
39b97ced92 2011-06-05 c.kworr@d4daf: #self._log.info('got {} lines from database'.format(len(result)))
ddbf5288b9 2010-11-03 c.kworr@d4daf: for row in result:
ddbf5288b9 2010-11-03 c.kworr@d4daf: if row != None and row[0] != None:
ddbf5288b9 2010-11-03 c.kworr@d4daf: if row[1] != None:
ddbf5288b9 2010-11-03 c.kworr@d4daf: self._log.info('trying regexp "{}" versus "{}"\n'.format(row[1], url_path))
ddbf5288b9 2010-11-03 c.kworr@d4daf: try:
ddbf5288b9 2010-11-03 c.kworr@d4daf: if re.compile(row[1]).match(url_path):
ddbf5288b9 2010-11-03 c.kworr@d4daf: reply = row[0].format(url_path)
ddbf5288b9 2010-11-03 c.kworr@d4daf: else:
ddbf5288b9 2010-11-03 c.kworr@d4daf: continue
ddbf5288b9 2010-11-03 c.kworr@d4daf: except:
ddbf5288b9 2010-11-03 c.kworr@d4daf: self._log.info("can't compile regexp")
ddbf5288b9 2010-11-03 c.kworr@d4daf: else:
ddbf5288b9 2010-11-03 c.kworr@d4daf: reply = row[0].format(url_path)
ddbf5288b9 2010-11-03 c.kworr@d4daf: if reply != None:
ddbf5288b9 2010-11-03 c.kworr@d4daf: self.writeline('{} {}\n'.format(id, reply))
ddbf5288b9 2010-11-03 c.kworr@d4daf: return(True)
ddbf5288b9 2010-11-03 c.kworr@d4daf: self.writeline('{}\n'.format(id))
ddbf5288b9 2010-11-03 c.kworr@d4daf:
0ef24b1937 2011-04-06 c.kworr@d4daf: def check(self):
0ef24b1937 2011-04-06 c.kworr@d4daf: while True:
0ef24b1937 2011-04-06 c.kworr@d4daf: line = self._queue.get()
0ef24b1937 2011-04-06 c.kworr@d4daf: if line == None:
0ef24b1937 2011-04-06 c.kworr@d4daf: break
39b97ced92 2011-06-05 c.kworr@d4daf: #self._log.info('request: ' + line)
0ef24b1937 2011-04-06 c.kworr@d4daf: request = self._request.match(line)
0ef24b1937 2011-04-06 c.kworr@d4daf: if request:
0ef24b1937 2011-04-06 c.kworr@d4daf: id = request.group(1)
0ef24b1937 2011-04-06 c.kworr@d4daf: #proto = request.group(2)
0ef24b1937 2011-04-06 c.kworr@d4daf: site = request.group(3)
0ef24b1937 2011-04-06 c.kworr@d4daf: url_path = request.group(4)
0ef24b1937 2011-04-06 c.kworr@d4daf: ip_address = request.group(5)
0ef24b1937 2011-04-06 c.kworr@d4daf: self.process(id, site, ip_address, url_path, line)
0ef24b1937 2011-04-06 c.kworr@d4daf: else:
0ef24b1937 2011-04-06 c.kworr@d4daf: self._log.info('bad request\n')
0ef24b1937 2011-04-06 c.kworr@d4daf: self.writeline(line + '\n')
ddbf5288b9 2010-11-03 c.kworr@d4daf:
ddbf5288b9 2010-11-03 c.kworr@d4daf: def writeline(self, string):
ddbf5288b9 2010-11-03 c.kworr@d4daf: self._log.info('sending: ' + string)
ddbf5288b9 2010-11-03 c.kworr@d4daf: sys.stdout.write(string)
ddbf5288b9 2010-11-03 c.kworr@d4daf: sys.stdout.flush()
ddbf5288b9 2010-11-03 c.kworr@d4daf:
ddbf5288b9 2010-11-03 c.kworr@d4daf: def loop(self):
0ef24b1937 2011-04-06 c.kworr@d4daf: pool = gevent.pool.Pool()
0ef24b1937 2011-04-06 c.kworr@d4daf: pool.spawn(self.check)
0ef24b1937 2011-04-06 c.kworr@d4daf: pool.join()
d2c7ba18a4 2011-09-14 c.kworr@d4daf:
d2c7ba18a4 2011-09-14 c.kworr@d4daf: # this classes processes config file and substitutes default values
d2c7ba18a4 2011-09-14 c.kworr@d4daf: class Config:
d2c7ba18a4 2011-09-14 c.kworr@d4daf: __slots__ = frozenset(['_config', '_default', '_section', 'options'])
d2c7ba18a4 2011-09-14 c.kworr@d4daf: _default = {
d2c7ba18a4 2011-09-14 c.kworr@d4daf: 'log': {
d2c7ba18a4 2011-09-14 c.kworr@d4daf: 'silent': 'no',
d2c7ba18a4 2011-09-14 c.kworr@d4daf: },
d2c7ba18a4 2011-09-14 c.kworr@d4daf: 'database': {
d2c7ba18a4 2011-09-14 c.kworr@d4daf: 'host': None,
d2c7ba18a4 2011-09-14 c.kworr@d4daf: 'database': 'squidTag',
d2c7ba18a4 2011-09-14 c.kworr@d4daf: },}
d2c7ba18a4 2011-09-14 c.kworr@d4daf:
d2c7ba18a4 2011-09-14 c.kworr@d4daf: # function to read in config file
d2c7ba18a4 2011-09-14 c.kworr@d4daf: def __init__(self):
d2c7ba18a4 2011-09-14 c.kworr@d4daf: import ConfigParser, optparse, os
d2c7ba18a4 2011-09-14 c.kworr@d4daf:
d2c7ba18a4 2011-09-14 c.kworr@d4daf: parser = optparse.OptionParser()
d2c7ba18a4 2011-09-14 c.kworr@d4daf: parser.add_option('-c', '--config', dest = 'config',
d2c7ba18a4 2011-09-14 c.kworr@d4daf: help = 'config file location', metavar = 'FILE',
d2c7ba18a4 2011-09-14 c.kworr@d4daf: default = '/usr/local/etc/squid-tagger.conf')
d2c7ba18a4 2011-09-14 c.kworr@d4daf: parser.add_option('-d', '--dump', dest = 'dump',
d2c7ba18a4 2011-09-14 c.kworr@d4daf: help = 'dump database', action = 'store_true', metavar = 'bool',
d2c7ba18a4 2011-09-14 c.kworr@d4daf: default = False)
d2c7ba18a4 2011-09-14 c.kworr@d4daf: parser.add_option('-f', '--flush-database', dest = 'flush_db',
d2c7ba18a4 2011-09-14 c.kworr@d4daf: help = 'flush previous database on load', default = False,
d2c7ba18a4 2011-09-14 c.kworr@d4daf: action = 'store_true', metavar = 'bool')
d2c7ba18a4 2011-09-14 c.kworr@d4daf: parser.add_option('-l', '--load', dest = 'load',
d2c7ba18a4 2011-09-14 c.kworr@d4daf: help = 'load database', action = 'store_true', metavar = 'bool',
d2c7ba18a4 2011-09-14 c.kworr@d4daf: default = False)
d2c7ba18a4 2011-09-14 c.kworr@d4daf: parser.add_option('-D', '--dump-conf', dest = 'dump_conf',
d2c7ba18a4 2011-09-14 c.kworr@d4daf: help = 'dump filtering rules', default = False, metavar = 'bool',
d2c7ba18a4 2011-09-14 c.kworr@d4daf: action = 'store_true')
d2c7ba18a4 2011-09-14 c.kworr@d4daf: parser.add_option('-L', '--load-conf', dest = 'load_conf',
d2c7ba18a4 2011-09-14 c.kworr@d4daf: help = 'load filtering rules', default = False, metavar = 'bool',
d2c7ba18a4 2011-09-14 c.kworr@d4daf: action = 'store_true')
d2c7ba18a4 2011-09-14 c.kworr@d4daf:
d2c7ba18a4 2011-09-14 c.kworr@d4daf: (self.options, args) = parser.parse_args()
d2c7ba18a4 2011-09-14 c.kworr@d4daf:
d2c7ba18a4 2011-09-14 c.kworr@d4daf: assert os.access(self.options.config, os.R_OK), "Fatal error: can't read {}".format(self.options.config)
d2c7ba18a4 2011-09-14 c.kworr@d4daf:
d2c7ba18a4 2011-09-14 c.kworr@d4daf: self._config = ConfigParser.ConfigParser()
d2c7ba18a4 2011-09-14 c.kworr@d4daf: self._config.readfp(open(self.options.config))
d2c7ba18a4 2011-09-14 c.kworr@d4daf:
d2c7ba18a4 2011-09-14 c.kworr@d4daf: # function to select config file section or create one
d2c7ba18a4 2011-09-14 c.kworr@d4daf: def section(self, section):
d2c7ba18a4 2011-09-14 c.kworr@d4daf: if not self._config.has_section(section):
d2c7ba18a4 2011-09-14 c.kworr@d4daf: self._config.add_section(section)
d2c7ba18a4 2011-09-14 c.kworr@d4daf: self._section = section
d2c7ba18a4 2011-09-14 c.kworr@d4daf:
d2c7ba18a4 2011-09-14 c.kworr@d4daf: # function to get config parameter, if parameter doesn't exists the default
d2c7ba18a4 2011-09-14 c.kworr@d4daf: # value or None is substituted
d2c7ba18a4 2011-09-14 c.kworr@d4daf: def __getitem__(self, name):
d2c7ba18a4 2011-09-14 c.kworr@d4daf: if not self._config.has_option(self._section, name):
d2c7ba18a4 2011-09-14 c.kworr@d4daf: if self._section in self._default:
d2c7ba18a4 2011-09-14 c.kworr@d4daf: if name in self._default[self._section]:
d2c7ba18a4 2011-09-14 c.kworr@d4daf: self._config.set(self._section, name, self._default[self._section][name])
d2c7ba18a4 2011-09-14 c.kworr@d4daf: else:
d2c7ba18a4 2011-09-14 c.kworr@d4daf: self._config.set(self._section, name, None)
d2c7ba18a4 2011-09-14 c.kworr@d4daf: else:
d2c7ba18a4 2011-09-14 c.kworr@d4daf: self._config.set(self._section, name, None)
d2c7ba18a4 2011-09-14 c.kworr@d4daf: return(self._config.get(self._section, name))
d2c7ba18a4 2011-09-14 c.kworr@d4daf:
d2c7ba18a4 2011-09-14 c.kworr@d4daf: # initializing and reading in config file
d2c7ba18a4 2011-09-14 c.kworr@d4daf: config = Config()
d500448801 2009-10-05 c.kworr@d4daf:
d301d9adc6 2010-08-13 c.kworr@d4daf: if config.options.dump or config.options.load or config.options.dump_conf or config.options.load_conf:
d301d9adc6 2010-08-13 c.kworr@d4daf: import csv
d301d9adc6 2010-08-13 c.kworr@d4daf:
d301d9adc6 2010-08-13 c.kworr@d4daf: tagdb = tagDB()
bde51dc0c7 2010-08-26 c.kworr@d4daf: data_fields = ['site', 'tag', 'regexp']
d301d9adc6 2010-08-13 c.kworr@d4daf: conf_fields = ['netmask', 'redirect_url', 'from_weekday', 'to_weekday', 'from_time', 'to_time', 'tag']
d301d9adc6 2010-08-13 c.kworr@d4daf:
d301d9adc6 2010-08-13 c.kworr@d4daf: if config.options.dump or config.options.dump_conf:
0ef24b1937 2011-04-06 c.kworr@d4daf: csv_writer = csv.writer(sys.stdout)
d301d9adc6 2010-08-13 c.kworr@d4daf: if config.options.dump:
bde51dc0c7 2010-08-26 c.kworr@d4daf: dump = tagdb.dump()
bde51dc0c7 2010-08-26 c.kworr@d4daf: elif config.options.dump_conf:
bde51dc0c7 2010-08-26 c.kworr@d4daf: dump = tagdb.dump_conf()
bde51dc0c7 2010-08-26 c.kworr@d4daf:
0ef24b1937 2011-04-06 c.kworr@d4daf: csv_writer.writerow(dump[0])
0ef24b1937 2011-04-06 c.kworr@d4daf: for line in dump[1]:
0ef24b1937 2011-04-06 c.kworr@d4daf: csv_writer.writerow(line)
d301d9adc6 2010-08-13 c.kworr@d4daf:
d301d9adc6 2010-08-13 c.kworr@d4daf: elif config.options.load or config.options.load_conf:
d301d9adc6 2010-08-13 c.kworr@d4daf: csv_reader = csv.reader(sys.stdin)
d301d9adc6 2010-08-13 c.kworr@d4daf: first_row = next(csv_reader)
d301d9adc6 2010-08-13 c.kworr@d4daf:
d301d9adc6 2010-08-13 c.kworr@d4daf: if config.options.load:
bde51dc0c7 2010-08-26 c.kworr@d4daf: fields = data_fields
bde51dc0c7 2010-08-26 c.kworr@d4daf: load = tagdb.load
bde51dc0c7 2010-08-26 c.kworr@d4daf: elif config.options.load_conf:
bde51dc0c7 2010-08-26 c.kworr@d4daf: fields = conf_fields
bde51dc0c7 2010-08-26 c.kworr@d4daf: load = tagdb.load_conf
bde51dc0c7 2010-08-26 c.kworr@d4daf:
bde51dc0c7 2010-08-26 c.kworr@d4daf: assert first_row == fields, 'File must contain csv data with theese columns: ' + repr(fields)
bde51dc0c7 2010-08-26 c.kworr@d4daf: load(csv_reader)
d301d9adc6 2010-08-13 c.kworr@d4daf:
d301d9adc6 2010-08-13 c.kworr@d4daf: else:
d301d9adc6 2010-08-13 c.kworr@d4daf: # main loop
39b97ced92 2011-06-05 c.kworr@d4daf: Checker(stdin, logger).loop()