server.py 18 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480
  1. #!/usr/bin/python
  2. # -*- coding: utf-8 -*-
  3. from __future__ import print_function, unicode_literals
  4. from BaseHTTPServer import BaseHTTPRequestHandler, HTTPServer
  5. import cgi
  6. import json
  7. import logging
  8. import re
  9. import socket
  10. from SocketServer import ThreadingMixIn
  11. import time
  12. import ffstatus
  13. # each match of these regex is removed to normalize an ISP's name
  14. ISP_NORMALIZATIONS = [
  15. # normalize name: strip company indication
  16. re.compile(r'(AG|UG|G?mbH( ?& ?Co\.? ?(OH|K)G)?)$', flags=re.IGNORECASE),
  17. # normalize name: strip "pool" suffixes
  18. re.compile(r'(dynamic )?(customer |subscriber )?(ip )?(pool|(address )?range|addresses)$', flags=re.IGNORECASE),
  19. # normalize name: strip "B2B" and aggregation suffixes
  20. re.compile(r'(aggregate|aggregation)?$', flags=re.IGNORECASE),
  21. re.compile(r'(B2B)?$', flags=re.IGNORECASE),
  22. # normalize name: strip country suffixes (in Germany)
  23. re.compile(r'(' +
  24. 'DE|Deutschland|Germany|' +
  25. 'Nordrhein[- ]Westfalen|NRW|' +
  26. 'Baden[- ]Wuerttemburg|BW|' +
  27. 'Hessen|' +
  28. 'Niedersachsen|' +
  29. 'Rheinland[- ]Pfalz|RLP' +
  30. ')$',
  31. flags=re.IGNORECASE),
  32. ]
  33. def normalize_ispname(isp):
  34. """Removes all matches on ISP_NORMALIZATIONS."""
  35. isp = isp.strip()
  36. for regex in ISP_NORMALIZATIONS:
  37. isp = regex.sub('', isp).strip()
  38. return isp
  39. class BatcaveHttpRequestHandler(BaseHTTPRequestHandler):
  40. def __init__(self, request, client_address, server):
  41. self.logger = logging.getLogger('API')
  42. BaseHTTPRequestHandler.__init__(self, request, client_address, server)
  43. def parse_url_pathquery(self):
  44. """Extracts the query parameters from the request path."""
  45. url = re.match(r'^/(?P<path>.*?)(\?(?P<query>.+))?$', self.path.strip())
  46. if url is None:
  47. logging.warn('Failed to parse URL \'' + str(self.path) + '\'.')
  48. return ( None, None )
  49. path = url.group('path')
  50. query = {}
  51. if not url.group('query') is None:
  52. for m in re.finditer(r'(?P<key>.+?)=(?P<value>.+?)(&|$)', url.group('query')):
  53. query[m.group('key')] = m.group('value')
  54. return ( path, query )
  55. def do_GET(self):
  56. """Handles all HTTP GET requests."""
  57. path, query = self.parse_url_pathquery()
  58. if path is None:
  59. self.send_error(400, 'Could not parse URL (' + str(self.path) + ')')
  60. return
  61. # / - index page, shows generic help
  62. if path == '':
  63. self.respond_index(query)
  64. return
  65. # /list - list stored nodes
  66. if path == 'list':
  67. self.respond_list(query)
  68. return
  69. # /vpn - notification endpoint for gateway's VPN connections
  70. if path == 'vpn':
  71. self.respond_vpn(query)
  72. return
  73. # /providers
  74. if path == 'providers':
  75. self.respond_providers(query)
  76. return
  77. # /node/<id>.json - node's data
  78. # /node/<id>/field - return specific field from node's data
  79. m = re.match(r'node/(?P<id>[a-fA-F0-9]{12})(?P<cmd>\.json|/[a-zA-Z0-9_\-\.]+)$', path)
  80. if m != None:
  81. cmd = m.group('cmd')
  82. nodeid = m.group('id').lower()
  83. if cmd == '.json':
  84. self.respond_node(nodeid)
  85. else:
  86. self.respond_nodedetail(nodeid, cmd[1:])
  87. return
  88. # /status/<id> - node's status
  89. m = re.match(r'status/([a-f0-9]{12})$', path)
  90. if m != None:
  91. self.respond_nodestatus(m.group(1))
  92. return
  93. # no match -> 404
  94. self.send_error(404, 'The URL \'{0}\' was not found here.'.format(path))
  95. def do_POST(self):
  96. """Handles all HTTP POST requests."""
  97. path, query = self.parse_url_pathquery()
  98. if path is None:
  99. self.send_error(400, 'Could not parse URL (' + str(self.path) + ')')
  100. return
  101. params = self.parse_post_params()
  102. # node id/mac to name mapping
  103. if path == 'idmac2name':
  104. self.respond_nodeidmac2name(params)
  105. return
  106. # no match -> 404
  107. self.send_error(404, 'The URL \'{0}\' was not found here.'.format(path))
  108. def send_nocache_headers(self):
  109. """Sets HTTP headers indicating that this response shall not be cached."""
  110. self.send_header('Cache-Control', 'no-cache, no-store, must-revalidate')
  111. self.send_header('Pragma', 'no-cache')
  112. self.send_header('Expires', '0')
  113. def send_headers(self, content_type='text/html; charset=utf-8', nocache=True):
  114. """Send HTTP 200 Response header with the given Content-Type.
  115. Optionally send no-caching headers, too."""
  116. self.send_response(200)
  117. self.send_header('Content-Type', content_type)
  118. if nocache:
  119. self.send_nocache_headers()
  120. self.end_headers()
  121. def parse_post_params(self):
  122. ctype, pdict = cgi.parse_header(self.headers.getheader('content-type'))
  123. if ctype == 'multipart/form-data':
  124. postvars = cgi.parse_multipart(self.rfile, pdict)
  125. elif ctype == 'application/x-www-form-urlencoded':
  126. length = int(self.headers.getheader('content-length'))
  127. postvars = cgi.parse_qs(self.rfile.read(length), keep_blank_values=1)
  128. else:
  129. postvars = {}
  130. return postvars
  131. def respond_index(self, query):
  132. """Display the index page."""
  133. storage = self.server.storage
  134. self.send_headers()
  135. self.wfile.write('<!DOCTYPE html><html><head><title>BATCAVE</title></head>\n')
  136. self.wfile.write('<body>\n')
  137. self.wfile.write('<H1 title="Batman/Alfred Transmission Collection, Aggregation & Value Engine">BATCAVE</H1>\n')
  138. self.wfile.write('<p>Dies ist ein interner Hintergrund-Dienst. Er wird nur von anderen Diensten\n')
  139. self.wfile.write('angesprochen und sollte aus einer Mehrzahl von Gr&uuml;nden nicht &ouml;ffentlich\n')
  140. self.wfile.write('zug&auml;nglich sein.</p>\n')
  141. self.wfile.write('<H2>Status</H2>\n')
  142. self.wfile.write('Daten: <span id="datacount" class="value">')
  143. self.wfile.write(len(storage.data))
  144. self.wfile.write('</span>\n')
  145. self.wfile.write('<H2>API</H2>\n')
  146. self.wfile.write('<p>Grundsätzlich ist das Antwort-Format JSON und alle Daten sind Live-Daten (kein Cache) die ggf. etwas Bearbeitungs-Zeit erfordern.</p>')
  147. self.wfile.write('<dl>\n')
  148. self.wfile.write('<dt><a href="/nodes.json">nodes.json</a></dt><dd>zur Verwendung mit ffmap (MACs anonymisiert)</dd>\n')
  149. self.wfile.write('<dt><a href="/node/ff00ff00ff00.json">/node/&lt;id&gt;.json</a></dt><dd><u>alle</u> vorhandenen Information zu der gewünschten Node</dd>\n')
  150. self.wfile.write('</dl>\n')
  151. self.wfile.write('</body></html>')
  152. def respond_list(self, query):
  153. """List stored data."""
  154. self.send_headers()
  155. self.wfile.write('<!DOCTYPE html><html>\n')
  156. self.wfile.write('<head><title>BATCAVE</title></head>\n')
  157. self.wfile.write('<body>\n')
  158. self.wfile.write('<H1>BATCAVE - LIST</H1>\n')
  159. self.wfile.write('<table>\n')
  160. self.wfile.write('<thead><tr><th>ID</th><th>Name</th></tr></thead>\n')
  161. self.wfile.write('<tbody>\n')
  162. sortkey = query['sort'] if 'sort' in query else None
  163. data = self.server.storage.get_nodes(sortby=sortkey)
  164. for node in data:
  165. nodeid = node['node_id']
  166. nodename = node['hostname'] if 'hostname' in node else '&lt;?&gt;'
  167. self.wfile.write('<tr><td><a href="/node/' + nodeid + '.json">' + nodeid + '</a></td><td>' + nodename + '</td></tr>')
  168. self.wfile.write('</tbody>\n')
  169. self.wfile.write('</table>\n')
  170. def respond_node(self, rawid):
  171. """Display node data."""
  172. # search node by the given id
  173. node = self.server.storage.find_node(rawid)
  174. # handle unknown nodes
  175. if node is None:
  176. self.send_error(404, 'No node with id \'' + rawid + '\' present.')
  177. return
  178. # dump node data as JSON
  179. self.send_headers('text/json')
  180. self.wfile.write(json.dumps(node))
  181. def respond_nodestatus(self, rawid):
  182. """Display node status."""
  183. status = self.server.storage.get_nodestatus(rawid)
  184. if status is None:
  185. self.send_error(404, 'No node with id \'' + rawid + '\' present.')
  186. self.send_headers('text/plain')
  187. self.wfile.write(status)
  188. def respond_nodeidmac2name(self, ids):
  189. """Return a mapping of the given IDs (or MACs) into their hostname."""
  190. self.send_headers('text/plain')
  191. for nodeid in ids:
  192. node = self.server.storage.find_node(nodeid) if not ':' in nodeid else self.server.storage.find_node_by_mac(nodeid)
  193. nodename = node['hostname'] if (not node is None) and 'hostname' in node else nodeid
  194. self.wfile.write('{0}={1}\n'.format(nodeid, nodename))
  195. def respond_nodedetail(self, nodeid, field):
  196. """Return a field from the given node - a string is returned as text, all other as JSON."""
  197. node = self.server.storage.find_node(nodeid)
  198. if node is None:
  199. self.send_error(404, 'No node with id \'' + nodeid + '\' present.')
  200. return
  201. return_count = False
  202. if field.endswith('.count'):
  203. return_count = True
  204. field = field[0:-6]
  205. if not field in node:
  206. self.send_error(404, 'The node \'' + nodeid + '\' does not have a field named \'' + str(field) + '\'.')
  207. return
  208. value = node[field]
  209. if return_count:
  210. value = len(value)
  211. self.send_headers('text/plain' if isinstance(value, basestring) or isinstance(value, int) else 'text/json')
  212. self.wfile.write(value if isinstance(value, basestring) else json.dumps(value))
  213. def respond_vpn(self, query):
  214. storage = self.server.storage
  215. peername = query['peer'] if 'peer' in query else None
  216. key = query['key'] if 'key' in query else None
  217. action = query['action'] if 'action' in query else None
  218. remote = query['remote'] if 'remote' in query else None
  219. gw = query['gw'] if 'gw' in query else None
  220. ts = query['ts'] if 'ts' in query else time.time()
  221. if action == 'list':
  222. self.respond_vpnlist()
  223. return
  224. if action != 'establish' and action != 'disestablish':
  225. self.logger.error('VPN: unknown action \'{0}\''.format(action))
  226. self.send_error(400, 'Invalid action.')
  227. return
  228. check = {'peername': peername, 'key': key, 'remote': remote, 'gw': gw}
  229. for k, val in check.items():
  230. if val is None or len(val.strip()) == 0:
  231. self.logger.error('VPN {0}: no or empty {1}'.format(action, k))
  232. self.send_error(400, 'Missing value for ' + str(k))
  233. return
  234. try:
  235. if action == 'establish':
  236. self.server.storage.log_vpn_connect(
  237. key, peername, remote, gw, ts)
  238. elif action == 'disestablish':
  239. self.server.storage.log_vpn_connect(key, gw, ts)
  240. else:
  241. self.logger.error('Unknown VPN action \'%s\' not filtered.',
  242. action)
  243. self.send_error(500)
  244. return
  245. except ffstatus.exceptions.VpnKeyFormatError:
  246. self.logger.error('VPN peer \'{0}\' {1}: bad key \'{2}\''.format(
  247. peername, action, key,
  248. ))
  249. self.send_error(400, 'Bad key.')
  250. return
  251. self.send_headers('text/plain')
  252. self.wfile.write('OK')
  253. storage.save()
  254. def respond_vpnlist(self):
  255. self.send_headers()
  256. self.wfile.write('<!DOCTYPE html>\n')
  257. self.wfile.write('<html><head><title>BATCAVE - VPN LIST</title></head>\n')
  258. self.wfile.write('<body>\n')
  259. self.wfile.write('<style type="text/css">\n')
  260. self.wfile.write('table { border: 2px solid #999; border-collapse: collapse; }\n')
  261. self.wfile.write('th, td { border: 1px solid #CCC; }\n')
  262. self.wfile.write('table tbody tr.online { background-color: #CFC; }\n')
  263. self.wfile.write('table tbody tr.offline { background-color: #FCC; }\n')
  264. self.wfile.write('</style>\n')
  265. self.wfile.write('<table>\n<thead>\n')
  266. gateways = self.server.storage.get_vpn_gateways()
  267. self.wfile.write('<tr><th rowspan="2">names (key)</th><th colspan="' + str(len(gateways)) + '">active</th><th colspan="' + str(len(gateways)) + '">last</th></tr>\n')
  268. self.wfile.write('<tr><th>' + '</th><th>'.join(gateways) + '</th><th>' + '</th><th>'.join(gateways) + '</th></tr>\n')
  269. self.wfile.write('</thead>\n')
  270. for item in self.server.storage.get_vpn_connections():
  271. self.wfile.write('<tr class="{0}">'.format('online' if item['online'] else 'offline'))
  272. self.wfile.write('<td title="{0}">{1}</td>'.format(
  273. item['key'],
  274. ' / '.join(item['names']) if len(item['names']) > 0 else '?',
  275. ))
  276. for conntype in ['active', 'last']:
  277. for gateway in gateways:
  278. remote = ''
  279. if conntype in item['remote'] and gateway in item['remote'][conntype]:
  280. remote = item['remote'][conntype][gateway]
  281. if isinstance(remote, dict):
  282. remote = remote['name']
  283. symbol = '&check;' if len(remote) > 0 else '&times;'
  284. self.wfile.write('<td title="{0}">{1}</td>'.format(
  285. remote, symbol))
  286. self.wfile.write('</tr>\n')
  287. self.wfile.write('</table>\n')
  288. self.wfile.write('</body>')
  289. self.wfile.write('</html>')
  290. def respond_providers(self, query):
  291. """Return a summary of providers."""
  292. outputformat = query['format'].lower() if 'format' in query else 'html'
  293. isps = {}
  294. ispblocks = {}
  295. for item in self.server.storage.get_vpn_connections():
  296. if item['count']['active'] == 0:
  297. continue
  298. remotes = []
  299. for gateway in item['remote']['active']:
  300. remote = item['remote']['active'][gateway]
  301. remotes.append(remote)
  302. if len(remotes) == 0:
  303. self.logger.warn(
  304. 'VPN key \'%s\' is marked with active remotes but 0 found?',
  305. item['key'])
  306. continue
  307. item_isps = set()
  308. for remote in remotes:
  309. isp = "UNKNOWN"
  310. ispblock = remote
  311. if isinstance(remote, dict):
  312. ispblock = remote['name']
  313. desc_lines = remote['description'].split('\n')
  314. isp = normalize_ispname(desc_lines[0])
  315. if not isp in ispblocks:
  316. ispblocks[isp] = set()
  317. ispblocks[isp].add(ispblock)
  318. item_isps.add(isp)
  319. if len(item_isps) == 0:
  320. item_isps.add('unknown')
  321. elif len(item_isps) > 1:
  322. self.logger.warn('VPN key \'{0}\' has {1} active IPs which resolved to {2} ISPs: \'{3}\''.format(key, len(ips), len(item_isps), '\', \''.join(item_isps)))
  323. for isp in item_isps:
  324. if not isp in isps:
  325. isps[isp] = 0
  326. isps[isp] += 1.0 / len(item_isps)
  327. isps_sum = sum([isps[x] for x in isps])
  328. if outputformat == 'csv':
  329. self.send_headers('text/csv')
  330. self.wfile.write('Count;Name\n')
  331. for isp in isps:
  332. self.wfile.write('{0};"{1}"\n'.format(isps[isp], isp))
  333. elif outputformat == 'json':
  334. self.send_headers('text/json')
  335. data = [
  336. {
  337. 'name': isp,
  338. 'count': isps[isp],
  339. 'percentage': isps[isp]*100.0/isps_sum,
  340. 'blocks': [block for block in ispblocks[isp]],
  341. } for isp in isps
  342. ]
  343. self.wfile.write(json.dumps(data))
  344. elif outputformat == 'html':
  345. self.send_headers()
  346. self.wfile.write('<!DOCTYPE html><html>\n')
  347. self.wfile.write('<head><title>BATCAVE - PROVIDERS</title></head>\n')
  348. self.wfile.write('<body>\n')
  349. self.wfile.write('<table border="2">\n')
  350. self.wfile.write('<thead><tr><th>Count</th><th>Percentage</th><th>Name</th><th>Blocks</th></tr></thead>\n')
  351. self.wfile.write('<tbody>\n')
  352. for isp in sorted(isps, key=lambda x: isps[x], reverse=True):
  353. self.wfile.write('<tr><td>{0}</td><td>{1:.1f}%</td><td>{2}</td><td>{3}</td></tr>\n'.format(
  354. isps[isp],
  355. isps[isp]*100.0/isps_sum,
  356. isp,
  357. ', '.join(sorted(ispblocks[isp])) if isp in ispblocks else '?',
  358. ))
  359. self.wfile.write('</tbody></table>\n')
  360. self.wfile.write('<p>Totals: {0} ISPs, {1} connections</p>\n'.format(len(isps), isps_sum))
  361. self.wfile.write('</body></html>')
  362. else:
  363. self.send_error(400, 'Unknown output format.')
  364. class ApiServer(ThreadingMixIn, HTTPServer):
  365. def __init__(self, endpoint, storage):
  366. if ':' in endpoint[0]:
  367. self.address_family = socket.AF_INET6
  368. HTTPServer.__init__(self, endpoint, BatcaveHttpRequestHandler)
  369. self.storage = storage
  370. def __str__(self):
  371. return 'ApiServer on {0}'.format(self.server_address)
  372. if __name__ == '__main__':
  373. dummystorage = ffstatus.basestorage.BaseStorage()
  374. server = ApiServer(('0.0.0.0', 8888), dummystorage)
  375. print("Server:", str(server))
  376. server.serve_forever()