d500448801 2009-10-05 1: #!/usr/bin/env python3.1
d500448801 2009-10-05 2:
d500448801 2009-10-05 3: import configparser, optparse, os, postgresql.api, re, sys, _thread
d500448801 2009-10-05 4:
b93dc49210 2009-10-13 5: # wrapper around syslog, can be muted
d500448801 2009-10-05 6: class Logger:
d500448801 2009-10-05 7: __slots__ = frozenset(['_syslog'])
d500448801 2009-10-05 8:
d500448801 2009-10-05 9: def __init__(self):
d500448801 2009-10-05 10: config.section('log')
d500448801 2009-10-05 11: if config['silent'] == 'yes':
d500448801 2009-10-05 12: self._syslog = None
d500448801 2009-10-05 13: else:
d500448801 2009-10-05 14: import syslog
d500448801 2009-10-05 15: self._syslog = syslog
d500448801 2009-10-05 16: self._syslog.openlog('squidTag')
d500448801 2009-10-05 17:
d500448801 2009-10-05 18: def info(self, message):
4b22e25f24 2009-10-07 19: if self._syslog:
d500448801 2009-10-05 20: self._syslog.syslog(self._syslog.LOG_INFO, message)
d500448801 2009-10-05 21:
d500448801 2009-10-05 22: def notice(self, message):
4b22e25f24 2009-10-07 23: if self._syslog:
d500448801 2009-10-05 24: self._syslog.syslog(self._syslog.LOG_NOTICE, message)
d500448801 2009-10-05 25:
b93dc49210 2009-10-13 26: # wrapper around database
d500448801 2009-10-05 27: class tagDB:
9450c03d41 2010-08-07 28: __slots__ = frozenset(('_check_stmt', '_db'))
b93dc49210 2009-10-13 29:
b93dc49210 2009-10-13 30: def __init__(self):
9450c03d41 2010-08-07 31: config.section('database')
9450c03d41 2010-08-07 32: self._db = postgresql.open(
9450c03d41 2010-08-07 33: 'pq://{}:{}@{}/{}'.format(
9450c03d41 2010-08-07 34: config['user'],
9450c03d41 2010-08-07 35: config['password'],
9450c03d41 2010-08-07 36: config['host'],
9450c03d41 2010-08-07 37: config['database'],
9450c03d41 2010-08-07 38: ) )
9450c03d41 2010-08-07 39: self._check_stmt = self._db.prepare("select redirect_url, regexp from site_rule where site <@ tripdomain($1) and netmask >> $2::text::inet order by array_length(site, 1) desc")
b93dc49210 2009-10-13 40:
b93dc49210 2009-10-13 41: def check(self, site, ip_address):
b93dc49210 2009-10-13 42: return(self._check_stmt(site, ip_address))
b93dc49210 2009-10-13 43:
b93dc49210 2009-10-13 44: # abstract class with basic checking functionality
b93dc49210 2009-10-13 45: class Checker:
ed7808827d 2009-10-14 46: __slots__ = frozenset(['_db', '_log'])
7e3418d94f 2009-10-12 47:
7e3418d94f 2009-10-12 48: def __init__(self):
b93dc49210 2009-10-13 49: self._db = tagDB()
b93dc49210 2009-10-13 50: self._log = Logger()
b93dc49210 2009-10-13 51:
ed7808827d 2009-10-14 52: def process(self, id, site, ip_address, url_path, line = None):
b93dc49210 2009-10-13 53: self._log.info('trying {}\n'.format(site))
b93dc49210 2009-10-13 54: result = self._db.check(site, ip_address)
b93dc49210 2009-10-13 55: #reply = '{}://{}/{}'.format(req[4], req[1], req[3])
b93dc49210 2009-10-13 56: reply = '-'
b93dc49210 2009-10-13 57: for row in result:
b93dc49210 2009-10-13 58: if row != None and row[0] != None:
b93dc49210 2009-10-13 59: if row[1] != None:
b93dc49210 2009-10-13 60: self._log.info('trying regexp "{}" versus "{}"\n'.format(row[1], url_path))
d2c54d0451 2010-03-01 61: try:
d2c54d0451 2010-03-01 62: if re.compile(row[1]).match(url_path):
1fa8a88371 2010-07-14 63: reply = row[0].format(url_path)
d2c54d0451 2010-03-01 64: break
d2c54d0451 2010-03-01 65: else:
d2c54d0451 2010-03-01 66: continue
d2c54d0451 2010-03-01 67: except:
d2c54d0451 2010-03-01 68: self._log.info("can't compile regexp")
b93dc49210 2009-10-13 69: else:
1fa8a88371 2010-07-14 70: reply = row[0].format(url_path)
b93dc49210 2009-10-13 71: break
b93dc49210 2009-10-13 72: self.writeline('{} {}\n'.format(id, reply))
7e3418d94f 2009-10-12 73:
7e3418d94f 2009-10-12 74: def check(self, line):
7e3418d94f 2009-10-12 75: request = re.compile('^([0-9]+)\ (http|ftp):\/\/([-\w.:]+)\/([^ ]*)\ ([0-9.]+)\/(-|[\w\.]+)\ (-|\w+)\ (-|GET|HEAD|POST).*$').match(line)
7e3418d94f 2009-10-12 76: if request:
7e3418d94f 2009-10-12 77: id = request.group(1)
7e3418d94f 2009-10-12 78: #proto = request.group(2)
7e3418d94f 2009-10-12 79: site = request.group(3)
7e3418d94f 2009-10-12 80: url_path = request.group(4)
7e3418d94f 2009-10-12 81: ip_address = request.group(5)
ed7808827d 2009-10-14 82: self.process(id, site, ip_address, url_path, line)
26fc9b34d9 2010-08-07 83: return(True)
7e3418d94f 2009-10-12 84: else:
7e3418d94f 2009-10-12 85: self._log.info('bad request\n')
b93dc49210 2009-10-13 86: self.writeline(line)
26fc9b34d9 2010-08-07 87: return(False)
b93dc49210 2009-10-13 88:
b93dc49210 2009-10-13 89: def writeline(self, string):
b93dc49210 2009-10-13 90: self._log.info('sending: ' + string)
b93dc49210 2009-10-13 91: sys.stdout.write(string)
b93dc49210 2009-10-13 92: sys.stdout.flush()
b93dc49210 2009-10-13 93:
ed7808827d 2009-10-14 94: def loop(self):
ed7808827d 2009-10-14 95: while True:
ed7808827d 2009-10-14 96: line = sys.stdin.readline()
ed7808827d 2009-10-14 97: if len(line) == 0:
ed7808827d 2009-10-14 98: break
ed7808827d 2009-10-14 99: self.check(line)
ed7808827d 2009-10-14 100:
b93dc49210 2009-10-13 101: # threaded checking facility
b93dc49210 2009-10-13 102: class CheckerThread(Checker):
ed7808827d 2009-10-14 103: __slots__ = frozenset(['_lock', '_lock_exit', '_lock_queue', '_queue'])
b93dc49210 2009-10-13 104:
b93dc49210 2009-10-13 105: def __init__(self):
ed7808827d 2009-10-14 106: # basic initialisation
b93dc49210 2009-10-13 107: Checker.__init__(self)
ed7808827d 2009-10-14 108:
b93dc49210 2009-10-13 109: # Spin lock. Loop acquires it on start then releases it when holding queue
b93dc49210 2009-10-13 110: # lock. This way the thread proceeds without stops while queue has data and
b93dc49210 2009-10-13 111: # gets stalled when no data present. The lock is released by queue writer
b93dc49210 2009-10-13 112: # after storing something into the queue
b93dc49210 2009-10-13 113: self._lock = _thread.allocate_lock()
ed7808827d 2009-10-14 114: self._lock_exit = _thread.allocate_lock()
b93dc49210 2009-10-13 115: self._lock_queue = _thread.allocate_lock()
b93dc49210 2009-10-13 116: self._lock.acquire()
b93dc49210 2009-10-13 117: self._queue = []
b93dc49210 2009-10-13 118: _thread.start_new_thread(self._start, ())
b93dc49210 2009-10-13 119:
b93dc49210 2009-10-13 120: def _start(self):
b93dc49210 2009-10-13 121: while True:
b93dc49210 2009-10-13 122: self._lock.acquire()
ed7808827d 2009-10-14 123: with self._lock_queue:
ed7808827d 2009-10-14 124: # yes this should be written this way, and yes, this is why I hate threading
ed7808827d 2009-10-14 125: if len(self._queue) > 1:
ed7808827d 2009-10-14 126: if self._lock.locked():
ed7808827d 2009-10-14 127: self._lock.release()
ed7808827d 2009-10-14 128: req = self._queue.pop(0)
ed7808827d 2009-10-14 129: Checker.process(self, req[0], req[1], req[2], req[3])
ed7808827d 2009-10-14 130: with self._lock_queue:
ed7808827d 2009-10-14 131: if len(self._queue) == 0:
ed7808827d 2009-10-14 132: if self._lock_exit.locked():
ed7808827d 2009-10-14 133: self._lock_exit.release()
ed7808827d 2009-10-14 134:
ed7808827d 2009-10-14 135: def process(self, id, site, ip_address, url_path, line):
ed7808827d 2009-10-14 136: with self._lock_queue:
ed7808827d 2009-10-14 137: self._queue.append((id, site, ip_address, url_path))
ed7808827d 2009-10-14 138: self._log.info('request {} queued ({})\n'.format(id, line))
ed7808827d 2009-10-14 139: if not self._lock_exit.locked():
ed7808827d 2009-10-14 140: self._lock_exit.acquire()
ed7808827d 2009-10-14 141: if self._lock.locked():
ed7808827d 2009-10-14 142: self._lock.release()
ed7808827d 2009-10-14 143:
ed7808827d 2009-10-14 144: def loop(self):
ed7808827d 2009-10-14 145: while True:
ed7808827d 2009-10-14 146: line = sys.stdin.readline()
ed7808827d 2009-10-14 147: if len(line) == 0:
ed7808827d 2009-10-14 148: break
ed7808827d 2009-10-14 149: self.check(line)
ed7808827d 2009-10-14 150: self._lock_exit.acquire()
ed7808827d 2009-10-14 151:
26fc9b34d9 2010-08-07 152: # kqueue enabled class for BSD's
ed7808827d 2009-10-14 153: class CheckerKqueue(Checker):
ed7808827d 2009-10-14 154: __slots__ = frozenset(['_kq', '_select', '_queue'])
ed7808827d 2009-10-14 155:
ed7808827d 2009-10-14 156: def __init__(self):
ed7808827d 2009-10-14 157: # basic initialisation
ed7808827d 2009-10-14 158: Checker.__init__(self)
ed7808827d 2009-10-14 159:
ed7808827d 2009-10-14 160: # importing select module
ed7808827d 2009-10-14 161: import select
ed7808827d 2009-10-14 162: self._select = select
ed7808827d 2009-10-14 163:
ed7808827d 2009-10-14 164: # kreating kqueue
ed7808827d 2009-10-14 165: self._kq = self._select.kqueue()
ed7808827d 2009-10-14 166: assert (self._kq.fileno() != -1)
ed7808827d 2009-10-14 167:
ed7808827d 2009-10-14 168: # watching sys.stdin for data
ed7808827d 2009-10-14 169: self._kq.control([self._select.kevent(sys.stdin, self._select.KQ_FILTER_READ, self._select.KQ_EV_ADD)], 0)
ed7808827d 2009-10-14 170:
ed7808827d 2009-10-14 171: # creating data queue
ed7808827d 2009-10-14 172: self._queue = []
ed7808827d 2009-10-14 173:
ed7808827d 2009-10-14 174: def loop(self):
ed7808827d 2009-10-14 175: # Wait for data by default
ed7808827d 2009-10-14 176: timeout = None
26fc9b34d9 2010-08-07 177: eof = False
26fc9b34d9 2010-08-07 178: buffer = ''
ed7808827d 2009-10-14 179: while True:
26fc9b34d9 2010-08-07 180: # checking if there is any data or witing for data to arrive
ed7808827d 2009-10-14 181: kevs = self._kq.control(None, 1, timeout)
26fc9b34d9 2010-08-07 182: if len(kevs) > 0 and kevs[0].filter == self._select.KQ_FILTER_READ and kevs[0].data > 0:
6c8b368359 2010-08-07 183: # reading data in
6c8b368359 2010-08-07 184: new_buffer = sys.stdin.read(kevs[0].data)
6c8b368359 2010-08-07 185: # if no data was sent - we have reached and of file
6c8b368359 2010-08-07 186: if len(new_buffer) == 0:
6c8b368359 2010-08-07 187: eof = True
6c8b368359 2010-08-07 188: else:
6c8b368359 2010-08-07 189: # adding current buffer to old buffer remains
6c8b368359 2010-08-07 190: buffer += new_buffer
6c8b368359 2010-08-07 191: # splitting to lines
6c8b368359 2010-08-07 192: lines = buffer.split('\n')
6c8b368359 2010-08-07 193: # last line that was not terminate by newline returns to buffer
6c8b368359 2010-08-07 194: buffer = lines[-1]
6c8b368359 2010-08-07 195: # an only if there was at least one newline
6c8b368359 2010-08-07 196: if len(lines) > 1:
6c8b368359 2010-08-07 197: for line in lines[:-1]:
6c8b368359 2010-08-07 198: # add data to the queue
6c8b368359 2010-08-07 199: if self.check(line + '\n'):
6c8b368359 2010-08-07 200: # don't wait for more data, start processing
6c8b368359 2010-08-07 201: timeout = 0
26fc9b34d9 2010-08-07 202: # if kq reports en of stream refuse to read more of it
26fc9b34d9 2010-08-07 203: if kevs[0].flags >> 15 == 1:
26fc9b34d9 2010-08-07 204: eof = True
ed7808827d 2009-10-14 205: else:
26fc9b34d9 2010-08-07 206: if len(self._queue) > 0:
26fc9b34d9 2010-08-07 207: req = self._queue.pop(0)
26fc9b34d9 2010-08-07 208: Checker.process(self, req[0], req[1], req[2], req[3])
26fc9b34d9 2010-08-07 209: if len(self._queue) == 0:
26fc9b34d9 2010-08-07 210: # wait for data - we have nothing to process
26fc9b34d9 2010-08-07 211: timeout = None
26fc9b34d9 2010-08-07 212: elif eof:
26fc9b34d9 2010-08-07 213: break
ed7808827d 2009-10-14 214:
ed7808827d 2009-10-14 215: def process(self, id, site, ip_address, url_path, line):
26fc9b34d9 2010-08-07 216: # simply adding data to the queue
ed7808827d 2009-10-14 217: self._queue.append((id, site, ip_address, url_path))
ed7808827d 2009-10-14 218: self._log.info('request {} queued ({})\n'.format(id, line))
7e3418d94f 2009-10-12 219:
fc934cead1 2009-10-13 220: # this classes processes config file and substitutes default values
d500448801 2009-10-05 221: class Config:
b93dc49210 2009-10-13 222: __slots__ = frozenset(['_config', '_default', '_section'])
b93dc49210 2009-10-13 223: _default = {
b93dc49210 2009-10-13 224: 'reactor': {
b93dc49210 2009-10-13 225: 'reactor': 'thread',
b93dc49210 2009-10-13 226: },
fc934cead1 2009-10-13 227: 'log': {
fc934cead1 2009-10-13 228: 'silent': 'no',
fc934cead1 2009-10-13 229: },
fc934cead1 2009-10-13 230: 'database': {
fc934cead1 2009-10-13 231: 'host': 'localhost',
fc934cead1 2009-10-13 232: 'database': 'squidTag',
fc934cead1 2009-10-13 233: },}
d500448801 2009-10-05 234:
fc934cead1 2009-10-13 235: # function to read in config file
d500448801 2009-10-05 236: def __init__(self):
d500448801 2009-10-05 237: parser = optparse.OptionParser()
d500448801 2009-10-05 238: parser.add_option('-c', '--config', dest = 'config',
d500448801 2009-10-05 239: help = 'config file location', metavar = 'FILE',
d500448801 2009-10-05 240: default = '/usr/local/etc/squid-tagger.conf')
d500448801 2009-10-05 241:
d500448801 2009-10-05 242: (options, args) = parser.parse_args()
d500448801 2009-10-05 243:
d500448801 2009-10-05 244: if not os.access(options.config, os.R_OK):
d500448801 2009-10-05 245: print("Can't read {}: exitting".format(options.config))
d500448801 2009-10-05 246: sys.exit(2)
d500448801 2009-10-05 247:
d500448801 2009-10-05 248: self._config = configparser.ConfigParser()
d500448801 2009-10-05 249: self._config.readfp(open(options.config))
d500448801 2009-10-05 250:
fc934cead1 2009-10-13 251: # function to select config file section or create one
d500448801 2009-10-05 252: def section(self, section):
fc934cead1 2009-10-13 253: if not self._config.has_section(section):
fc934cead1 2009-10-13 254: self._config.add_section(section)
d500448801 2009-10-05 255: self._section = section
d500448801 2009-10-05 256:
fc934cead1 2009-10-13 257: # function to get config parameter, if parameter doesn't exists the default
fc934cead1 2009-10-13 258: # value or None is substituted
d500448801 2009-10-05 259: def __getitem__(self, name):
fc934cead1 2009-10-13 260: if not self._config.has_option(self._section, name):
b93dc49210 2009-10-13 261: if self._section in self._default:
b93dc49210 2009-10-13 262: if name in self._default[self._section]:
fc934cead1 2009-10-13 263: self._config.set(self._section, name, self._default[self._section][name])
fc934cead1 2009-10-13 264: else:
fc934cead1 2009-10-13 265: self._config.set(self._section, name, None)
fc934cead1 2009-10-13 266: else:
fc934cead1 2009-10-13 267: self._config.set(self._section, name, None)
b93dc49210 2009-10-13 268: return(self._config.get(self._section, name))
d500448801 2009-10-05 269:
fc934cead1 2009-10-13 270: # initializing and reading in config file
d500448801 2009-10-05 271: config = Config()
d500448801 2009-10-05 272:
b93dc49210 2009-10-13 273: config.section('reactor')
b93dc49210 2009-10-13 274: if config['reactor'] == 'thread':
b93dc49210 2009-10-13 275: checker = CheckerThread()
ed7808827d 2009-10-14 276: elif config['reactor'] == 'plain':
ed7808827d 2009-10-14 277: checker = Checker()
ed7808827d 2009-10-14 278: elif config['reactor'] == 'kqueue':
ed7808827d 2009-10-14 279: checker = CheckerKqueue()
b93dc49210 2009-10-13 280:
ed7808827d 2009-10-14 281: checker.loop()