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 39b97ced92:

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
fc934cead1 2009-10-13   32: 
fc934cead1 2009-10-13   33: # this classes processes config file and substitutes default values
fc934cead1 2009-10-13   34: class Config:
ae30851739 2010-08-12   35: 	__slots__ = frozenset(['_config', '_default', '_section', 'options'])
b93dc49210 2009-10-13   36: 	_default = {
fc934cead1 2009-10-13   37: 		'log': {
fc934cead1 2009-10-13   38: 			'silent': 'no',
fc934cead1 2009-10-13   39: 		},
fc934cead1 2009-10-13   40: 		'database': {
fc934cead1 2009-10-13   41: 			'host': 'localhost',
fc934cead1 2009-10-13   42: 			'database': 'squidTag',
fc934cead1 2009-10-13   43: 	},}
fc934cead1 2009-10-13   44: 
fc934cead1 2009-10-13   45: 	# function to read in config file
fc934cead1 2009-10-13   46: 	def __init__(self):
0ef24b1937 2011-04-06   47: 		import ConfigParser, optparse, os
ae30851739 2010-08-12   48: 
fc934cead1 2009-10-13   49: 		parser = optparse.OptionParser()
fc934cead1 2009-10-13   50: 		parser.add_option('-c', '--config', dest = 'config',
fc934cead1 2009-10-13   51: 			help = 'config file location', metavar = 'FILE',
fc934cead1 2009-10-13   52: 			default = '/usr/local/etc/squid-tagger.conf')
ae30851739 2010-08-12   53: 		parser.add_option('-d', '--dump', dest = 'dump',
ae30851739 2010-08-12   54: 			help = 'dump database', action = 'store_true', metavar = 'bool',
ae30851739 2010-08-12   55: 			default = False)
31e69c4237 2010-08-12   56: 		parser.add_option('-f', '--flush-database', dest = 'flush_db',
31e69c4237 2010-08-12   57: 			help = 'flush previous database on load', default = False,
31e69c4237 2010-08-12   58: 			action = 'store_true', metavar = 'bool')
31e69c4237 2010-08-12   59: 		parser.add_option('-l', '--load', dest = 'load',
31e69c4237 2010-08-12   60: 			help = 'load database', action = 'store_true', metavar = 'bool',
31e69c4237 2010-08-12   61: 			default = False)
d301d9adc6 2010-08-13   62: 		parser.add_option('-D', '--dump-conf', dest = 'dump_conf',
d301d9adc6 2010-08-13   63: 			help = 'dump filtering rules', default = False, metavar = 'bool',
d301d9adc6 2010-08-13   64: 			action = 'store_true')
d301d9adc6 2010-08-13   65: 		parser.add_option('-L', '--load-conf', dest = 'load_conf',
d301d9adc6 2010-08-13   66: 			help = 'load filtering rules', default = False, metavar = 'bool',
d301d9adc6 2010-08-13   67: 			action = 'store_true')
7c13294e9f 2010-08-07   68: 
ae30851739 2010-08-12   69: 		(self.options, args) = parser.parse_args()
7c13294e9f 2010-08-07   70: 
ae30851739 2010-08-12   71: 		assert os.access(self.options.config, os.R_OK), "Fatal error: can't read {}".format(self.options.config)
7c13294e9f 2010-08-07   72: 
0ef24b1937 2011-04-06   73: 		self._config = ConfigParser.ConfigParser()
ae30851739 2010-08-12   74: 		self._config.readfp(open(self.options.config))
fc934cead1 2009-10-13   75: 
fc934cead1 2009-10-13   76: 	# function to select config file section or create one
fc934cead1 2009-10-13   77: 	def section(self, section):
fc934cead1 2009-10-13   78: 		if not self._config.has_section(section):
fc934cead1 2009-10-13   79: 			self._config.add_section(section)
fc934cead1 2009-10-13   80: 		self._section = section
fc934cead1 2009-10-13   81: 
fc934cead1 2009-10-13   82: 	# function to get config parameter, if parameter doesn't exists the default
fc934cead1 2009-10-13   83: 	# value or None is substituted
fc934cead1 2009-10-13   84: 	def __getitem__(self, name):
fc934cead1 2009-10-13   85: 		if not self._config.has_option(self._section, name):
b93dc49210 2009-10-13   86: 			if self._section in self._default:
b93dc49210 2009-10-13   87: 				if name in self._default[self._section]:
fc934cead1 2009-10-13   88: 					self._config.set(self._section, name, self._default[self._section][name])
fc934cead1 2009-10-13   89: 				else:
fc934cead1 2009-10-13   90: 					self._config.set(self._section, name, None)
fc934cead1 2009-10-13   91: 			else:
fc934cead1 2009-10-13   92: 				self._config.set(self._section, name, None)
b93dc49210 2009-10-13   93: 		return(self._config.get(self._section, name))
d500448801 2009-10-05   94: 
fc934cead1 2009-10-13   95: # initializing and reading in config file
fc934cead1 2009-10-13   96: config = Config()
d500448801 2009-10-05   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: 
39b97ced92 2011-06-05  121: # tiny wrapper around a file to make reads from it geventable
39b97ced92 2011-06-05  122: # or should i move this somewhere?
39b97ced92 2011-06-05  123: 
39b97ced92 2011-06-05  124: class FReadlineQueue(gevent.queue.Queue):
39b97ced92 2011-06-05  125: 	# storing file descriptor, leftover
39b97ced92 2011-06-05  126: 	__slots__ = frozenset(['_fd', '_tail'])
39b97ced92 2011-06-05  127: 
39b97ced92 2011-06-05  128: 	def __init__(self, fd):
39b97ced92 2011-06-05  129: 		# initialising class
39b97ced92 2011-06-05  130: 		gevent.queue.Queue.__init__(self)
39b97ced92 2011-06-05  131: 		# storing file descriptor
39b97ced92 2011-06-05  132: 		self._fd = fd
39b97ced92 2011-06-05  133: 		# using empty tail
39b97ced92 2011-06-05  134: 		self._tail = ''
39b97ced92 2011-06-05  135: 		# setting up event
39b97ced92 2011-06-05  136: 		self._install_wait()
39b97ced92 2011-06-05  137: 
39b97ced92 2011-06-05  138: 	def _install_wait(self):
39b97ced92 2011-06-05  139: 		fileno = self._fd.fileno()
39b97ced92 2011-06-05  140: 		# putting file to nonblocking mode
39b97ced92 2011-06-05  141: 		fcntl.fcntl(fileno, fcntl.F_SETFL, fcntl.fcntl(fileno, fcntl.F_GETFL)  | os.O_NONBLOCK)
39b97ced92 2011-06-05  142: 		# installing event handler
39b97ced92 2011-06-05  143: 		gevent.core.read_event(fileno, self._wait_helper)
39b97ced92 2011-06-05  144: 
39b97ced92 2011-06-05  145: 	def _wait_helper(self, ev, evtype):
39b97ced92 2011-06-05  146: 		# reading one buffer from stream
39b97ced92 2011-06-05  147: 		buf = self._fd.read(4096)
39b97ced92 2011-06-05  148: 		# splitting stream by line ends
39b97ced92 2011-06-05  149: 		rows = buf.decode('l1').split('\n')
39b97ced92 2011-06-05  150: 		# adding tail to the first element if there is some tail
39b97ced92 2011-06-05  151: 		if len(self._tail) > 0:
39b97ced92 2011-06-05  152: 			rows[0] = self._tail + rows[0]
39b97ced92 2011-06-05  153: 		# popping out last (incomplete) element
39b97ced92 2011-06-05  154: 		self._tail = rows.pop(-1)
39b97ced92 2011-06-05  155: 		# dropping all complete elements to the queue
39b97ced92 2011-06-05  156: 		for row in rows:
39b97ced92 2011-06-05  157: 			self.put_nowait(row)
39b97ced92 2011-06-05  158: 			logger.info('request: ' + row)
39b97ced92 2011-06-05  159: 		if len(buf) > 0:
39b97ced92 2011-06-05  160: 			# no EOF, reinstalling event handler
39b97ced92 2011-06-05  161: 			gevent.core.read_event(self._fd.fileno(), self._wait_helper)
39b97ced92 2011-06-05  162: 		else:
39b97ced92 2011-06-05  163: 			# EOF found, sending EOF to queue
39b97ced92 2011-06-05  164: 			self.put_nowait(None)
39b97ced92 2011-06-05  165: 
39b97ced92 2011-06-05  166: stdin = FReadlineQueue(sys.stdin)
39b97ced92 2011-06-05  167: 
39b97ced92 2011-06-05  168: # wrapper around database
39b97ced92 2011-06-05  169: class tagDB(object):
39b97ced92 2011-06-05  170: 	__slots__ = frozenset(['_cursor', '_db'])
39b97ced92 2011-06-05  171: 
39b97ced92 2011-06-05  172: 	def __init__(self):
39b97ced92 2011-06-05  173: 		config.section('database')
39b97ced92 2011-06-05  174: 		self._db = psycopg2.connect(
39b97ced92 2011-06-05  175: 			database = config['database'],
39b97ced92 2011-06-05  176: 			host = config['host'],
39b97ced92 2011-06-05  177: 			user = config['user'],
39b97ced92 2011-06-05  178: 			password = config['password'],
39b97ced92 2011-06-05  179: 		)
39b97ced92 2011-06-05  180: 		self._cursor = self._db.cursor()
39b97ced92 2011-06-05  181: 
39b97ced92 2011-06-05  182: 	def _field_names(self):
39b97ced92 2011-06-05  183: 		names = []
39b97ced92 2011-06-05  184: 		for record in self._cursor.description:
39b97ced92 2011-06-05  185: 			names.append(record.name)
39b97ced92 2011-06-05  186: 		return(names)
39b97ced92 2011-06-05  187: 
39b97ced92 2011-06-05  188: 	def check(self, site, ip_address):
39b97ced92 2011-06-05  189: 		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])
39b97ced92 2011-06-05  190: 		return(self._cursor.fetchall())
39b97ced92 2011-06-05  191: 
39b97ced92 2011-06-05  192: 	def dump(self):
39b97ced92 2011-06-05  193: 		self._cursor.execute("select untrip(site) as site, tag::text, regexp from urls order by site, tag")
39b97ced92 2011-06-05  194: 		return(self._field_names(), self._cursor.fetchall())
39b97ced92 2011-06-05  195: 
39b97ced92 2011-06-05  196: 	def load(self, data):
39b97ced92 2011-06-05  197: 		if config.options.flush_db:
39b97ced92 2011-06-05  198: 			self._cursor.execute('delete from urls;')
39b97ced92 2011-06-05  199: 		bundle = []
39b97ced92 2011-06-05  200: 		for row in data:
39b97ced92 2011-06-05  201: 			if len(row) == 2:
39b97ced92 2011-06-05  202: 				bundle.append([row[0], row[1], None])
39b97ced92 2011-06-05  203: 			else:
39b97ced92 2011-06-05  204: 				bundle.append([row[0], row[1], row[2]])
39b97ced92 2011-06-05  205: 		self._cursor.executemany("insert into urls (site, tag, regexp) values (tripdomain(%s), %s, %s)", bundle)
39b97ced92 2011-06-05  206: 		self._cursor.execute("update urls set regexp = NULL where regexp = ''")
39b97ced92 2011-06-05  207: 		self._db.commit()
39b97ced92 2011-06-05  208: 
39b97ced92 2011-06-05  209: 	def load_conf(self, csv_data):
39b97ced92 2011-06-05  210: 		self._cursor.execute('delete from rules;')
39b97ced92 2011-06-05  211: 		bundle = []
39b97ced92 2011-06-05  212: 		for row in csv_data:
39b97ced92 2011-06-05  213: 			bundle.append([row[0], row[1], int(row[2]), int(row[3]), row[4], row[5], row[6]])
39b97ced92 2011-06-05  214: 		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)
39b97ced92 2011-06-05  215: 		self._db.commit()
39b97ced92 2011-06-05  216: 
39b97ced92 2011-06-05  217: 	def dump_conf(self):
39b97ced92 2011-06-05  218: 		self._cursor.execute("select netmask, redirect_url, from_weekday, to_weekday, from_time, to_time, tag::text from rules")
39b97ced92 2011-06-05  219: 		return(self._field_names(), self._cursor.fetchall())
39b97ced92 2011-06-05  220: 
39b97ced92 2011-06-05  221: # abstract class with basic checking functionality
39b97ced92 2011-06-05  222: class Checker(object):
39b97ced92 2011-06-05  223: 	__slots__ = frozenset(['_db', '_log', '_queue', '_request'])
39b97ced92 2011-06-05  224: 
39b97ced92 2011-06-05  225: 	def __init__(self, queue, logger):
39b97ced92 2011-06-05  226: 		self._db = tagDB()
39b97ced92 2011-06-05  227: 		self._log = logger
39b97ced92 2011-06-05  228: 		self._log.info('started\n')
39b97ced92 2011-06-05  229: 		self._request = re.compile('^([0-9]+)\ (http|ftp):\/\/([-\w.:]+)\/([^ ]*)\ ([0-9.]+)\/(-|[\w\.]+)\ (-|\w+)\ (-|GET|HEAD|POST).*$')
39b97ced92 2011-06-05  230: 		self._queue = queue
39b97ced92 2011-06-05  231: 
39b97ced92 2011-06-05  232: 	def process(self, id, site, ip_address, url_path, line = None):
39b97ced92 2011-06-05  233: 		#self._log.info('trying {}\n'.format(site))
39b97ced92 2011-06-05  234: 		result = self._db.check(site, ip_address)
39b97ced92 2011-06-05  235: 		reply = None
39b97ced92 2011-06-05  236: 		#self._log.info('got {} lines from database'.format(len(result)))
39b97ced92 2011-06-05  237: 		for row in result:
39b97ced92 2011-06-05  238: 			if row != None and row[0] != None:
39b97ced92 2011-06-05  239: 				if row[1] != None:
39b97ced92 2011-06-05  240: 					self._log.info('trying regexp "{}" versus "{}"\n'.format(row[1], url_path))
39b97ced92 2011-06-05  241: 					try:
39b97ced92 2011-06-05  242: 						if re.compile(row[1]).match(url_path):
39b97ced92 2011-06-05  243: 							reply = row[0].format(url_path)
39b97ced92 2011-06-05  244: 						else:
39b97ced92 2011-06-05  245: 							continue
39b97ced92 2011-06-05  246: 					except:
39b97ced92 2011-06-05  247: 						self._log.info("can't compile regexp")
39b97ced92 2011-06-05  248: 				else:
39b97ced92 2011-06-05  249: 					reply = row[0].format(url_path)
39b97ced92 2011-06-05  250: 			if reply != None:
39b97ced92 2011-06-05  251: 				self.writeline('{} {}\n'.format(id, reply))
39b97ced92 2011-06-05  252: 				return(True)
39b97ced92 2011-06-05  253: 		self.writeline('{}\n'.format(id))
39b97ced92 2011-06-05  254: 
39b97ced92 2011-06-05  255: 	def check(self):
39b97ced92 2011-06-05  256: 		while True:
39b97ced92 2011-06-05  257: 			line = self._queue.get()
39b97ced92 2011-06-05  258: 			if line == None:
39b97ced92 2011-06-05  259: 				break
39b97ced92 2011-06-05  260: 			#self._log.info('request: ' + line)
39b97ced92 2011-06-05  261: 			request = self._request.match(line)
39b97ced92 2011-06-05  262: 			if request:
39b97ced92 2011-06-05  263: 				id = request.group(1)
39b97ced92 2011-06-05  264: 				#proto = request.group(2)
39b97ced92 2011-06-05  265: 				site = request.group(3)
39b97ced92 2011-06-05  266: 				url_path = request.group(4)
39b97ced92 2011-06-05  267: 				ip_address = request.group(5)
39b97ced92 2011-06-05  268: 				self.process(id, site, ip_address, url_path, line)
39b97ced92 2011-06-05  269: 			else:
39b97ced92 2011-06-05  270: 				self._log.info('bad request\n')
39b97ced92 2011-06-05  271: 				self.writeline(line + '\n')
39b97ced92 2011-06-05  272: 
39b97ced92 2011-06-05  273: 	def writeline(self, string):
39b97ced92 2011-06-05  274: 		self._log.info('sending: ' + string)
39b97ced92 2011-06-05  275: 		sys.stdout.write(string)
39b97ced92 2011-06-05  276: 		sys.stdout.flush()
39b97ced92 2011-06-05  277: 
39b97ced92 2011-06-05  278: 	def loop(self):
39b97ced92 2011-06-05  279: 		pool = gevent.pool.Pool()
39b97ced92 2011-06-05  280: 		pool.spawn(self.check)
39b97ced92 2011-06-05  281: 		pool.join()
39b97ced92 2011-06-05  282: 
d301d9adc6 2010-08-13  283: if config.options.dump or config.options.load or config.options.dump_conf or config.options.load_conf:
d301d9adc6 2010-08-13  284: 	import csv
d301d9adc6 2010-08-13  285: 
d301d9adc6 2010-08-13  286: 	tagdb = tagDB()
bde51dc0c7 2010-08-26  287: 	data_fields = ['site', 'tag', 'regexp']
d301d9adc6 2010-08-13  288: 	conf_fields = ['netmask', 'redirect_url', 'from_weekday', 'to_weekday', 'from_time', 'to_time', 'tag']
d301d9adc6 2010-08-13  289: 
d301d9adc6 2010-08-13  290: 	if config.options.dump or config.options.dump_conf:
0ef24b1937 2011-04-06  291: 		csv_writer = csv.writer(sys.stdout)
d301d9adc6 2010-08-13  292: 		if config.options.dump:
bde51dc0c7 2010-08-26  293: 			dump = tagdb.dump()
bde51dc0c7 2010-08-26  294: 		elif config.options.dump_conf:
bde51dc0c7 2010-08-26  295: 			dump = tagdb.dump_conf()
bde51dc0c7 2010-08-26  296: 
0ef24b1937 2011-04-06  297: 		csv_writer.writerow(dump[0])
0ef24b1937 2011-04-06  298: 		for line in dump[1]:
0ef24b1937 2011-04-06  299: 			csv_writer.writerow(line)
d301d9adc6 2010-08-13  300: 
d301d9adc6 2010-08-13  301: 	elif config.options.load or config.options.load_conf:
d301d9adc6 2010-08-13  302: 		csv_reader = csv.reader(sys.stdin)
d301d9adc6 2010-08-13  303: 		first_row = next(csv_reader)
d301d9adc6 2010-08-13  304: 
d301d9adc6 2010-08-13  305: 		if config.options.load:
bde51dc0c7 2010-08-26  306: 			fields = data_fields
bde51dc0c7 2010-08-26  307: 			load = tagdb.load
bde51dc0c7 2010-08-26  308: 		elif config.options.load_conf:
bde51dc0c7 2010-08-26  309: 			fields = conf_fields
bde51dc0c7 2010-08-26  310: 			load = tagdb.load_conf
bde51dc0c7 2010-08-26  311: 
bde51dc0c7 2010-08-26  312: 		assert first_row == fields, 'File must contain csv data with theese columns: ' + repr(fields)
bde51dc0c7 2010-08-26  313: 		load(csv_reader)
d301d9adc6 2010-08-13  314: 
d301d9adc6 2010-08-13  315: else:
d301d9adc6 2010-08-13  316: 	# main loop
39b97ced92 2011-06-05  317: 	Checker(stdin, logger).loop()