Squid url redirector

Annotation For squid-tagger.py
anonymous

Annotation For squid-tagger.py

Origin for each line in squid-tagger.py from check-in d2c7ba18a4:

0ef24b1937 2011-04-06    1: #!/usr/bin/env python
0ef24b1937 2011-04-06    2: 
0ef24b1937 2011-04-06    3: from __future__ import division, print_function, unicode_literals
0ef24b1937 2011-04-06    4: 
0ef24b1937 2011-04-06    5: import gevent.monkey
0ef24b1937 2011-04-06    6: gevent.monkey.patch_all()
0ef24b1937 2011-04-06    7: 
0ef24b1937 2011-04-06    8: import fcntl, gevent.core, gevent.pool, gevent.queue, gevent.socket, os, psycopg2, re, sys
0ef24b1937 2011-04-06    9: 
0ef24b1937 2011-04-06   10: # //inclusion start
0ef24b1937 2011-04-06   11: # Copyright (C) 2010 Daniele Varrazzo <daniele.varrazzo@gmail.com>
0ef24b1937 2011-04-06   12: # and licensed under the MIT license:
0ef24b1937 2011-04-06   13: 
0ef24b1937 2011-04-06   14: def gevent_wait_callback(conn, timeout=None):
0ef24b1937 2011-04-06   15: 	"""A wait callback useful to allow gevent to work with Psycopg."""
0ef24b1937 2011-04-06   16: 	while 1:
0ef24b1937 2011-04-06   17: 		state = conn.poll()
0ef24b1937 2011-04-06   18: 		if state == psycopg2.extensions.POLL_OK:
0ef24b1937 2011-04-06   19: 			break
0ef24b1937 2011-04-06   20: 		elif state == psycopg2.extensions.POLL_READ:
0ef24b1937 2011-04-06   21: 			gevent.socket.wait_read(conn.fileno(), timeout=timeout)
0ef24b1937 2011-04-06   22: 		elif state == psycopg2.extensions.POLL_WRITE:
0ef24b1937 2011-04-06   23: 			gevent.socket.wait_write(conn.fileno(), timeout=timeout)
0ef24b1937 2011-04-06   24: 		else:
0ef24b1937 2011-04-06   25: 			raise psycopg2.OperationalError("Bad result from poll: %r" % state)
0ef24b1937 2011-04-06   26: 
0ef24b1937 2011-04-06   27: if not hasattr(psycopg2.extensions, 'set_wait_callback'):
0ef24b1937 2011-04-06   28: 	raise ImportError("support for coroutines not available in this Psycopg version (%s)" % psycopg2.__version__)
0ef24b1937 2011-04-06   29: 	psycopg2.extensions.set_wait_callback(gevent_wait_callback)
0ef24b1937 2011-04-06   30: 
0ef24b1937 2011-04-06   31: # //inclusion end
0ef24b1937 2011-04-06   32: 
39b97ced92 2011-06-05   33: # this classes processes config file and substitutes default values
39b97ced92 2011-06-05   34: class Config:
39b97ced92 2011-06-05   35: 	__slots__ = frozenset(['_config', '_default', '_section', 'options'])
39b97ced92 2011-06-05   36: 	_default = {
39b97ced92 2011-06-05   37: 		'log': {
39b97ced92 2011-06-05   38: 			'silent': 'no',
39b97ced92 2011-06-05   39: 		},
39b97ced92 2011-06-05   40: 		'database': {
39b97ced92 2011-06-05   41: 			'host': 'localhost',
39b97ced92 2011-06-05   42: 			'database': 'squidTag',
39b97ced92 2011-06-05   43: 	},}
39b97ced92 2011-06-05   44: 
39b97ced92 2011-06-05   45: 	# function to read in config file
39b97ced92 2011-06-05   46: 	def __init__(self):
39b97ced92 2011-06-05   47: 		import ConfigParser, optparse, os
39b97ced92 2011-06-05   48: 
39b97ced92 2011-06-05   49: 		parser = optparse.OptionParser()
39b97ced92 2011-06-05   50: 		parser.add_option('-c', '--config', dest = 'config',
39b97ced92 2011-06-05   51: 			help = 'config file location', metavar = 'FILE',
39b97ced92 2011-06-05   52: 			default = '/usr/local/etc/squid-tagger.conf')
39b97ced92 2011-06-05   53: 		parser.add_option('-d', '--dump', dest = 'dump',
39b97ced92 2011-06-05   54: 			help = 'dump database', action = 'store_true', metavar = 'bool',
39b97ced92 2011-06-05   55: 			default = False)
39b97ced92 2011-06-05   56: 		parser.add_option('-f', '--flush-database', dest = 'flush_db',
39b97ced92 2011-06-05   57: 			help = 'flush previous database on load', default = False,
39b97ced92 2011-06-05   58: 			action = 'store_true', metavar = 'bool')
39b97ced92 2011-06-05   59: 		parser.add_option('-l', '--load', dest = 'load',
39b97ced92 2011-06-05   60: 			help = 'load database', action = 'store_true', metavar = 'bool',
39b97ced92 2011-06-05   61: 			default = False)
39b97ced92 2011-06-05   62: 		parser.add_option('-D', '--dump-conf', dest = 'dump_conf',
39b97ced92 2011-06-05   63: 			help = 'dump filtering rules', default = False, metavar = 'bool',
39b97ced92 2011-06-05   64: 			action = 'store_true')
39b97ced92 2011-06-05   65: 		parser.add_option('-L', '--load-conf', dest = 'load_conf',
39b97ced92 2011-06-05   66: 			help = 'load filtering rules', default = False, metavar = 'bool',
39b97ced92 2011-06-05   67: 			action = 'store_true')
39b97ced92 2011-06-05   68: 
39b97ced92 2011-06-05   69: 		(self.options, args) = parser.parse_args()
39b97ced92 2011-06-05   70: 
39b97ced92 2011-06-05   71: 		assert os.access(self.options.config, os.R_OK), "Fatal error: can't read {}".format(self.options.config)
39b97ced92 2011-06-05   72: 
39b97ced92 2011-06-05   73: 		self._config = ConfigParser.ConfigParser()
39b97ced92 2011-06-05   74: 		self._config.readfp(open(self.options.config))
39b97ced92 2011-06-05   75: 
39b97ced92 2011-06-05   76: 	# function to select config file section or create one
39b97ced92 2011-06-05   77: 	def section(self, section):
39b97ced92 2011-06-05   78: 		if not self._config.has_section(section):
39b97ced92 2011-06-05   79: 			self._config.add_section(section)
39b97ced92 2011-06-05   80: 		self._section = section
39b97ced92 2011-06-05   81: 
39b97ced92 2011-06-05   82: 	# function to get config parameter, if parameter doesn't exists the default
39b97ced92 2011-06-05   83: 	# value or None is substituted
39b97ced92 2011-06-05   84: 	def __getitem__(self, name):
39b97ced92 2011-06-05   85: 		if not self._config.has_option(self._section, name):
39b97ced92 2011-06-05   86: 			if self._section in self._default:
39b97ced92 2011-06-05   87: 				if name in self._default[self._section]:
39b97ced92 2011-06-05   88: 					self._config.set(self._section, name, self._default[self._section][name])
39b97ced92 2011-06-05   89: 				else:
39b97ced92 2011-06-05   90: 					self._config.set(self._section, name, None)
39b97ced92 2011-06-05   91: 			else:
39b97ced92 2011-06-05   92: 				self._config.set(self._section, name, None)
39b97ced92 2011-06-05   93: 		return(self._config.get(self._section, name))
39b97ced92 2011-06-05   94: 
39b97ced92 2011-06-05   95: # initializing and reading in config file
39b97ced92 2011-06-05   96: config = Config()
39b97ced92 2011-06-05   97: 
39b97ced92 2011-06-05   98: # wrapper around syslog, can be muted
39b97ced92 2011-06-05   99: class Logger(object):
39b97ced92 2011-06-05  100: 	__slots__ = frozenset(['_syslog'])
39b97ced92 2011-06-05  101: 
39b97ced92 2011-06-05  102: 	def __init__(self):
39b97ced92 2011-06-05  103: 		config.section('log')
39b97ced92 2011-06-05  104: 		if config['silent'] == 'yes':
39b97ced92 2011-06-05  105: 			self._syslog = None
39b97ced92 2011-06-05  106: 		else:
39b97ced92 2011-06-05  107: 			import syslog
39b97ced92 2011-06-05  108: 			self._syslog = syslog
39b97ced92 2011-06-05  109: 			self._syslog.openlog(str('squidTag'))
39b97ced92 2011-06-05  110: 
39b97ced92 2011-06-05  111: 	def info(self, message):
39b97ced92 2011-06-05  112: 		if self._syslog != None:
39b97ced92 2011-06-05  113: 			self._syslog.syslog(self._syslog.LOG_INFO, message)
39b97ced92 2011-06-05  114: 
39b97ced92 2011-06-05  115: 	def notice(self, message):
39b97ced92 2011-06-05  116: 		if self._syslog != None:
39b97ced92 2011-06-05  117: 			self._syslog.syslog(self._syslog.LOG_NOTICE, message)
39b97ced92 2011-06-05  118: 
39b97ced92 2011-06-05  119: logger = Logger()
39b97ced92 2011-06-05  120: 
0ef24b1937 2011-04-06  121: # tiny wrapper around a file to make reads from it geventable
0ef24b1937 2011-04-06  122: # or should i move this somewhere?
0ef24b1937 2011-04-06  123: 
0ef24b1937 2011-04-06  124: class FReadlineQueue(gevent.queue.Queue):
0ef24b1937 2011-04-06  125: 	# storing file descriptor, leftover
0ef24b1937 2011-04-06  126: 	__slots__ = frozenset(['_fd', '_tail'])
0ef24b1937 2011-04-06  127: 
0ef24b1937 2011-04-06  128: 	def __init__(self, fd):
0ef24b1937 2011-04-06  129: 		# initialising class
0ef24b1937 2011-04-06  130: 		gevent.queue.Queue.__init__(self)
0ef24b1937 2011-04-06  131: 		# storing file descriptor
0ef24b1937 2011-04-06  132: 		self._fd = fd
0ef24b1937 2011-04-06  133: 		# using empty tail
0ef24b1937 2011-04-06  134: 		self._tail = ''
0ef24b1937 2011-04-06  135: 		# setting up event
0ef24b1937 2011-04-06  136: 		self._install_wait()
0ef24b1937 2011-04-06  137: 
0ef24b1937 2011-04-06  138: 	def _install_wait(self):
0ef24b1937 2011-04-06  139: 		fileno = self._fd.fileno()
0ef24b1937 2011-04-06  140: 		# putting file to nonblocking mode
0ef24b1937 2011-04-06  141: 		fcntl.fcntl(fileno, fcntl.F_SETFL, fcntl.fcntl(fileno, fcntl.F_GETFL)  | os.O_NONBLOCK)
0ef24b1937 2011-04-06  142: 		# installing event handler
0ef24b1937 2011-04-06  143: 		gevent.core.read_event(fileno, self._wait_helper)
0ef24b1937 2011-04-06  144: 
0ef24b1937 2011-04-06  145: 	def _wait_helper(self, ev, evtype):
0ef24b1937 2011-04-06  146: 		# reading one buffer from stream
0ef24b1937 2011-04-06  147: 		buf = self._fd.read(4096)
0ef24b1937 2011-04-06  148: 		# splitting stream by line ends
39b97ced92 2011-06-05  149: 		rows = buf.decode('l1').split('\n')
0ef24b1937 2011-04-06  150: 		# adding tail to the first element if there is some tail
0ef24b1937 2011-04-06  151: 		if len(self._tail) > 0:
0ef24b1937 2011-04-06  152: 			rows[0] = self._tail + rows[0]
0ef24b1937 2011-04-06  153: 		# popping out last (incomplete) element
0ef24b1937 2011-04-06  154: 		self._tail = rows.pop(-1)
0ef24b1937 2011-04-06  155: 		# dropping all complete elements to the queue
0ef24b1937 2011-04-06  156: 		for row in rows:
0ef24b1937 2011-04-06  157: 			self.put_nowait(row)
39b97ced92 2011-06-05  158: 			logger.info('request: ' + row)
0ef24b1937 2011-04-06  159: 		if len(buf) > 0:
0ef24b1937 2011-04-06  160: 			# no EOF, reinstalling event handler
0ef24b1937 2011-04-06  161: 			gevent.core.read_event(self._fd.fileno(), self._wait_helper)
0ef24b1937 2011-04-06  162: 		else:
0ef24b1937 2011-04-06  163: 			# EOF found, sending EOF to queue
0ef24b1937 2011-04-06  164: 			self.put_nowait(None)
0ef24b1937 2011-04-06  165: 
0ef24b1937 2011-04-06  166: stdin = FReadlineQueue(sys.stdin)
0ef24b1937 2011-04-06  167: 
ddbf5288b9 2010-11-03  168: # wrapper around database
0ef24b1937 2011-04-06  169: class tagDB(object):
0ef24b1937 2011-04-06  170: 	__slots__ = frozenset(['_cursor', '_db'])
ddbf5288b9 2010-11-03  171: 
ddbf5288b9 2010-11-03  172: 	def __init__(self):
ddbf5288b9 2010-11-03  173: 		config.section('database')
d2c7ba18a4 2011-09-14  174: 		if config['host'] == None:
d2c7ba18a4 2011-09-14  175: 			self._db = psycopg2.connect(
d2c7ba18a4 2011-09-14  176: 				database = config['database'],
d2c7ba18a4 2011-09-14  177: 				user = config['user'],
d2c7ba18a4 2011-09-14  178: 				password = config['password']
d2c7ba18a4 2011-09-14  179: 			)
d2c7ba18a4 2011-09-14  180: 		else:
d2c7ba18a4 2011-09-14  181: 			self._db = psycopg2.connect(
d2c7ba18a4 2011-09-14  182: 				database = config['database'],
d2c7ba18a4 2011-09-14  183: 				host = config['host'],
d2c7ba18a4 2011-09-14  184: 				user = config['user'],
d2c7ba18a4 2011-09-14  185: 				password = config['password']
d2c7ba18a4 2011-09-14  186: 			)
0ef24b1937 2011-04-06  187: 		self._cursor = self._db.cursor()
0ef24b1937 2011-04-06  188: 
0ef24b1937 2011-04-06  189: 	def _field_names(self):
0ef24b1937 2011-04-06  190: 		names = []
0ef24b1937 2011-04-06  191: 		for record in self._cursor.description:
0ef24b1937 2011-04-06  192: 			names.append(record.name)
0ef24b1937 2011-04-06  193: 		return(names)
ddbf5288b9 2010-11-03  194: 
ddbf5288b9 2010-11-03  195: 	def check(self, site, ip_address):
39b97ced92 2011-06-05  196: 		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  197: 		return(self._cursor.fetchall())
ddbf5288b9 2010-11-03  198: 
ddbf5288b9 2010-11-03  199: 	def dump(self):
0ef24b1937 2011-04-06  200: 		self._cursor.execute("select untrip(site) as site, tag::text, regexp from urls order by site, tag")
0ef24b1937 2011-04-06  201: 		return(self._field_names(), self._cursor.fetchall())
ddbf5288b9 2010-11-03  202: 
ddbf5288b9 2010-11-03  203: 	def load(self, data):
0ef24b1937 2011-04-06  204: 		if config.options.flush_db:
0ef24b1937 2011-04-06  205: 			self._cursor.execute('delete from urls;')
0ef24b1937 2011-04-06  206: 		bundle = []
0ef24b1937 2011-04-06  207: 		for row in data:
0ef24b1937 2011-04-06  208: 			if len(row) == 2:
0ef24b1937 2011-04-06  209: 				bundle.append([row[0], row[1], None])
0ef24b1937 2011-04-06  210: 			else:
0ef24b1937 2011-04-06  211: 				bundle.append([row[0], row[1], row[2]])
0ef24b1937 2011-04-06  212: 		self._cursor.executemany("insert into urls (site, tag, regexp) values (tripdomain(%s), %s, %s)", bundle)
0ef24b1937 2011-04-06  213: 		self._cursor.execute("update urls set regexp = NULL where regexp = ''")
0ef24b1937 2011-04-06  214: 		self._db.commit()
ddbf5288b9 2010-11-03  215: 
ddbf5288b9 2010-11-03  216: 	def load_conf(self, csv_data):
0ef24b1937 2011-04-06  217: 		self._cursor.execute('delete from rules;')
0ef24b1937 2011-04-06  218: 		bundle = []
0ef24b1937 2011-04-06  219: 		for row in csv_data:
0ef24b1937 2011-04-06  220: 			bundle.append([row[0], row[1], int(row[2]), int(row[3]), row[4], row[5], row[6]])
0ef24b1937 2011-04-06  221: 		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  222: 		self._db.commit()
ddbf5288b9 2010-11-03  223: 
ddbf5288b9 2010-11-03  224: 	def dump_conf(self):
0ef24b1937 2011-04-06  225: 		self._cursor.execute("select netmask, redirect_url, from_weekday, to_weekday, from_time, to_time, tag::text from rules")
0ef24b1937 2011-04-06  226: 		return(self._field_names(), self._cursor.fetchall())
ddbf5288b9 2010-11-03  227: 
ddbf5288b9 2010-11-03  228: # abstract class with basic checking functionality
0ef24b1937 2011-04-06  229: class Checker(object):
0ef24b1937 2011-04-06  230: 	__slots__ = frozenset(['_db', '_log', '_queue', '_request'])
ddbf5288b9 2010-11-03  231: 
39b97ced92 2011-06-05  232: 	def __init__(self, queue, logger):
ddbf5288b9 2010-11-03  233: 		self._db = tagDB()
39b97ced92 2011-06-05  234: 		self._log = logger
ddbf5288b9 2010-11-03  235: 		self._log.info('started\n')
ddbf5288b9 2010-11-03  236: 		self._request = re.compile('^([0-9]+)\ (http|ftp):\/\/([-\w.:]+)\/([^ ]*)\ ([0-9.]+)\/(-|[\w\.]+)\ (-|\w+)\ (-|GET|HEAD|POST).*$')
0ef24b1937 2011-04-06  237: 		self._queue = queue
ddbf5288b9 2010-11-03  238: 
ddbf5288b9 2010-11-03  239: 	def process(self, id, site, ip_address, url_path, line = None):
0ef24b1937 2011-04-06  240: 		#self._log.info('trying {}\n'.format(site))
ddbf5288b9 2010-11-03  241: 		result = self._db.check(site, ip_address)
ddbf5288b9 2010-11-03  242: 		reply = None
39b97ced92 2011-06-05  243: 		#self._log.info('got {} lines from database'.format(len(result)))
ddbf5288b9 2010-11-03  244: 		for row in result:
ddbf5288b9 2010-11-03  245: 			if row != None and row[0] != None:
ddbf5288b9 2010-11-03  246: 				if row[1] != None:
ddbf5288b9 2010-11-03  247: 					self._log.info('trying regexp "{}" versus "{}"\n'.format(row[1], url_path))
ddbf5288b9 2010-11-03  248: 					try:
ddbf5288b9 2010-11-03  249: 						if re.compile(row[1]).match(url_path):
ddbf5288b9 2010-11-03  250: 							reply = row[0].format(url_path)
ddbf5288b9 2010-11-03  251: 						else:
ddbf5288b9 2010-11-03  252: 							continue
ddbf5288b9 2010-11-03  253: 					except:
ddbf5288b9 2010-11-03  254: 						self._log.info("can't compile regexp")
ddbf5288b9 2010-11-03  255: 				else:
ddbf5288b9 2010-11-03  256: 					reply = row[0].format(url_path)
ddbf5288b9 2010-11-03  257: 			if reply != None:
ddbf5288b9 2010-11-03  258: 				self.writeline('{} {}\n'.format(id, reply))
ddbf5288b9 2010-11-03  259: 				return(True)
ddbf5288b9 2010-11-03  260: 		self.writeline('{}\n'.format(id))
ddbf5288b9 2010-11-03  261: 
0ef24b1937 2011-04-06  262: 	def check(self):
0ef24b1937 2011-04-06  263: 		while True:
0ef24b1937 2011-04-06  264: 			line = self._queue.get()
0ef24b1937 2011-04-06  265: 			if line == None:
0ef24b1937 2011-04-06  266: 				break
39b97ced92 2011-06-05  267: 			#self._log.info('request: ' + line)
0ef24b1937 2011-04-06  268: 			request = self._request.match(line)
0ef24b1937 2011-04-06  269: 			if request:
0ef24b1937 2011-04-06  270: 				id = request.group(1)
0ef24b1937 2011-04-06  271: 				#proto = request.group(2)
0ef24b1937 2011-04-06  272: 				site = request.group(3)
0ef24b1937 2011-04-06  273: 				url_path = request.group(4)
0ef24b1937 2011-04-06  274: 				ip_address = request.group(5)
0ef24b1937 2011-04-06  275: 				self.process(id, site, ip_address, url_path, line)
0ef24b1937 2011-04-06  276: 			else:
0ef24b1937 2011-04-06  277: 				self._log.info('bad request\n')
0ef24b1937 2011-04-06  278: 				self.writeline(line + '\n')
ddbf5288b9 2010-11-03  279: 
ddbf5288b9 2010-11-03  280: 	def writeline(self, string):
ddbf5288b9 2010-11-03  281: 		self._log.info('sending: ' + string)
ddbf5288b9 2010-11-03  282: 		sys.stdout.write(string)
ddbf5288b9 2010-11-03  283: 		sys.stdout.flush()
ddbf5288b9 2010-11-03  284: 
ddbf5288b9 2010-11-03  285: 	def loop(self):
0ef24b1937 2011-04-06  286: 		pool = gevent.pool.Pool()
0ef24b1937 2011-04-06  287: 		pool.spawn(self.check)
0ef24b1937 2011-04-06  288: 		pool.join()
d2c7ba18a4 2011-09-14  289: 
d2c7ba18a4 2011-09-14  290: # this classes processes config file and substitutes default values
d2c7ba18a4 2011-09-14  291: class Config:
d2c7ba18a4 2011-09-14  292: 	__slots__ = frozenset(['_config', '_default', '_section', 'options'])
d2c7ba18a4 2011-09-14  293: 	_default = {
d2c7ba18a4 2011-09-14  294: 		'log': {
d2c7ba18a4 2011-09-14  295: 			'silent': 'no',
d2c7ba18a4 2011-09-14  296: 		},
d2c7ba18a4 2011-09-14  297: 		'database': {
d2c7ba18a4 2011-09-14  298: 			'host': None,
d2c7ba18a4 2011-09-14  299: 			'database': 'squidTag',
d2c7ba18a4 2011-09-14  300: 	},}
d2c7ba18a4 2011-09-14  301: 
d2c7ba18a4 2011-09-14  302: 	# function to read in config file
d2c7ba18a4 2011-09-14  303: 	def __init__(self):
d2c7ba18a4 2011-09-14  304: 		import ConfigParser, optparse, os
d2c7ba18a4 2011-09-14  305: 
d2c7ba18a4 2011-09-14  306: 		parser = optparse.OptionParser()
d2c7ba18a4 2011-09-14  307: 		parser.add_option('-c', '--config', dest = 'config',
d2c7ba18a4 2011-09-14  308: 			help = 'config file location', metavar = 'FILE',
d2c7ba18a4 2011-09-14  309: 			default = '/usr/local/etc/squid-tagger.conf')
d2c7ba18a4 2011-09-14  310: 		parser.add_option('-d', '--dump', dest = 'dump',
d2c7ba18a4 2011-09-14  311: 			help = 'dump database', action = 'store_true', metavar = 'bool',
d2c7ba18a4 2011-09-14  312: 			default = False)
d2c7ba18a4 2011-09-14  313: 		parser.add_option('-f', '--flush-database', dest = 'flush_db',
d2c7ba18a4 2011-09-14  314: 			help = 'flush previous database on load', default = False,
d2c7ba18a4 2011-09-14  315: 			action = 'store_true', metavar = 'bool')
d2c7ba18a4 2011-09-14  316: 		parser.add_option('-l', '--load', dest = 'load',
d2c7ba18a4 2011-09-14  317: 			help = 'load database', action = 'store_true', metavar = 'bool',
d2c7ba18a4 2011-09-14  318: 			default = False)
d2c7ba18a4 2011-09-14  319: 		parser.add_option('-D', '--dump-conf', dest = 'dump_conf',
d2c7ba18a4 2011-09-14  320: 			help = 'dump filtering rules', default = False, metavar = 'bool',
d2c7ba18a4 2011-09-14  321: 			action = 'store_true')
d2c7ba18a4 2011-09-14  322: 		parser.add_option('-L', '--load-conf', dest = 'load_conf',
d2c7ba18a4 2011-09-14  323: 			help = 'load filtering rules', default = False, metavar = 'bool',
d2c7ba18a4 2011-09-14  324: 			action = 'store_true')
d2c7ba18a4 2011-09-14  325: 
d2c7ba18a4 2011-09-14  326: 		(self.options, args) = parser.parse_args()
d2c7ba18a4 2011-09-14  327: 
d2c7ba18a4 2011-09-14  328: 		assert os.access(self.options.config, os.R_OK), "Fatal error: can't read {}".format(self.options.config)
d2c7ba18a4 2011-09-14  329: 
d2c7ba18a4 2011-09-14  330: 		self._config = ConfigParser.ConfigParser()
d2c7ba18a4 2011-09-14  331: 		self._config.readfp(open(self.options.config))
d2c7ba18a4 2011-09-14  332: 
d2c7ba18a4 2011-09-14  333: 	# function to select config file section or create one
d2c7ba18a4 2011-09-14  334: 	def section(self, section):
d2c7ba18a4 2011-09-14  335: 		if not self._config.has_section(section):
d2c7ba18a4 2011-09-14  336: 			self._config.add_section(section)
d2c7ba18a4 2011-09-14  337: 		self._section = section
d2c7ba18a4 2011-09-14  338: 
d2c7ba18a4 2011-09-14  339: 	# function to get config parameter, if parameter doesn't exists the default
d2c7ba18a4 2011-09-14  340: 	# value or None is substituted
d2c7ba18a4 2011-09-14  341: 	def __getitem__(self, name):
d2c7ba18a4 2011-09-14  342: 		if not self._config.has_option(self._section, name):
d2c7ba18a4 2011-09-14  343: 			if self._section in self._default:
d2c7ba18a4 2011-09-14  344: 				if name in self._default[self._section]:
d2c7ba18a4 2011-09-14  345: 					self._config.set(self._section, name, self._default[self._section][name])
d2c7ba18a4 2011-09-14  346: 				else:
d2c7ba18a4 2011-09-14  347: 					self._config.set(self._section, name, None)
d2c7ba18a4 2011-09-14  348: 			else:
d2c7ba18a4 2011-09-14  349: 				self._config.set(self._section, name, None)
d2c7ba18a4 2011-09-14  350: 		return(self._config.get(self._section, name))
d2c7ba18a4 2011-09-14  351: 
d2c7ba18a4 2011-09-14  352: # initializing and reading in config file
d2c7ba18a4 2011-09-14  353: config = Config()
d500448801 2009-10-05  354: 
d301d9adc6 2010-08-13  355: if config.options.dump or config.options.load or config.options.dump_conf or config.options.load_conf:
d301d9adc6 2010-08-13  356: 	import csv
d301d9adc6 2010-08-13  357: 
d301d9adc6 2010-08-13  358: 	tagdb = tagDB()
bde51dc0c7 2010-08-26  359: 	data_fields = ['site', 'tag', 'regexp']
d301d9adc6 2010-08-13  360: 	conf_fields = ['netmask', 'redirect_url', 'from_weekday', 'to_weekday', 'from_time', 'to_time', 'tag']
d301d9adc6 2010-08-13  361: 
d301d9adc6 2010-08-13  362: 	if config.options.dump or config.options.dump_conf:
0ef24b1937 2011-04-06  363: 		csv_writer = csv.writer(sys.stdout)
d301d9adc6 2010-08-13  364: 		if config.options.dump:
bde51dc0c7 2010-08-26  365: 			dump = tagdb.dump()
bde51dc0c7 2010-08-26  366: 		elif config.options.dump_conf:
bde51dc0c7 2010-08-26  367: 			dump = tagdb.dump_conf()
bde51dc0c7 2010-08-26  368: 
0ef24b1937 2011-04-06  369: 		csv_writer.writerow(dump[0])
0ef24b1937 2011-04-06  370: 		for line in dump[1]:
0ef24b1937 2011-04-06  371: 			csv_writer.writerow(line)
d301d9adc6 2010-08-13  372: 
d301d9adc6 2010-08-13  373: 	elif config.options.load or config.options.load_conf:
d301d9adc6 2010-08-13  374: 		csv_reader = csv.reader(sys.stdin)
d301d9adc6 2010-08-13  375: 		first_row = next(csv_reader)
d301d9adc6 2010-08-13  376: 
d301d9adc6 2010-08-13  377: 		if config.options.load:
bde51dc0c7 2010-08-26  378: 			fields = data_fields
bde51dc0c7 2010-08-26  379: 			load = tagdb.load
bde51dc0c7 2010-08-26  380: 		elif config.options.load_conf:
bde51dc0c7 2010-08-26  381: 			fields = conf_fields
bde51dc0c7 2010-08-26  382: 			load = tagdb.load_conf
bde51dc0c7 2010-08-26  383: 
bde51dc0c7 2010-08-26  384: 		assert first_row == fields, 'File must contain csv data with theese columns: ' + repr(fields)
bde51dc0c7 2010-08-26  385: 		load(csv_reader)
d301d9adc6 2010-08-13  386: 
d301d9adc6 2010-08-13  387: else:
d301d9adc6 2010-08-13  388: 	# main loop
39b97ced92 2011-06-05  389: 	Checker(stdin, logger).loop()