Mercurial > code > home > repos > homeauto
comparison service/wifi/scrape.py @ 421:47d7dd31bb2c
port to py3
Ignore-this: e1a2e6bb730111e76f5a5dd2366d498a
author | drewp@bigasterisk.com |
---|---|
date | Sat, 30 Mar 2019 18:27:17 -0700 |
parents | service/wifi/wifi.py@a530d9c5b280 |
children | 19460b3f7baf |
comparison
equal
deleted
inserted
replaced
420:a530d9c5b280 | 421:47d7dd31bb2c |
---|---|
1 import re, ast, logging, socket, json, base64 | |
2 from twisted.internet.defer import inlineCallbacks, returnValue | |
3 from cyclone.httpclient import fetch | |
4 from rdflib import Literal, Graph, RDFS, URIRef | |
5 | |
6 log = logging.getLogger() | |
7 | |
8 class Router(object): | |
9 def __repr__(self): | |
10 return repr(self.__dict__) | |
11 | |
12 class Wifi(object): | |
13 """ | |
14 gather the users of wifi from the tomato routers | |
15 """ | |
16 def __init__(self, accessN3="/my/proj/openid_proxy/access.n3"): | |
17 self.rereadConfig() | |
18 #self._loadRouters(accessN3, tomatoUrl) | |
19 | |
20 def rereadConfig(self): | |
21 self.graph = Graph() | |
22 self.graph.parse('config.n3', format='n3') | |
23 | |
24 | |
25 def _loadRouters(self, accessN3, tomatoUrl): | |
26 g = Graph() | |
27 g.parse(accessN3, format="n3") | |
28 repl = { | |
29 '/wifiRouter1/' : None, | |
30 #'/tomato2/' : None | |
31 } | |
32 for k in repl: | |
33 rows = list(g.query(''' | |
34 PREFIX p: <http://bigasterisk.com/openid_proxy#> | |
35 SELECT ?prefix WHERE { | |
36 ?site | |
37 p:requestPrefix ?public; | |
38 p:proxyUrlPrefix ?prefix | |
39 . | |
40 }''', initBindings={"public" : Literal(k)})) | |
41 repl[k] = str(rows[0][0]) | |
42 log.debug('repl %r', repl) | |
43 | |
44 self.routers = [] | |
45 for url in tomatoUrl: | |
46 name = url | |
47 for k, v in repl.items(): | |
48 url = url.replace(k, v) | |
49 | |
50 r = Router() | |
51 http, tail = url.split('//', 1) | |
52 userPass, tail = tail.split("@", 1) | |
53 r.url = http + '//' + tail | |
54 r.headers = {'Authorization': ['Basic %s' % userPass.encode('base64').strip()]} | |
55 r.name = {'wifiRouter1' : 'bigasterisk5', | |
56 'tomato2' : 'bigasterisk4'}[name.split('/')[1]] | |
57 self.routers.append(r) | |
58 | |
59 @inlineCallbacks | |
60 def getPresentMacAddrs(self): | |
61 self.rereadConfig() | |
62 rows = yield loadOrbiData() | |
63 for row in rows: | |
64 if 'clientHostname' in row: | |
65 row['name'] = row['clientHostname'] | |
66 mac = URIRef('http://bigasterisk.com/mac/%s' % row['mac'].lower()) | |
67 label = self.graph.value(mac, RDFS.label) | |
68 if label: | |
69 row['name'] = label | |
70 returnValue(rows) | |
71 | |
72 @inlineCallbacks | |
73 def getPresentMacAddrs_multirouter(self): | |
74 rows = [] | |
75 | |
76 for router in self.routers: | |
77 log.debug("GET %s", router) | |
78 try: | |
79 resp = yield fetch(router.url, headers=router.headers, | |
80 timeout=2) | |
81 except socket.error: | |
82 log.warn("get on %s failed" % router) | |
83 continue | |
84 data = resp.body | |
85 if 'Wireless -- Authenticated Stations' in data: | |
86 # zyxel 'Station Info' page | |
87 rows.extend(self._parseZyxel(data, router.name)) | |
88 else: | |
89 # tomato page | |
90 rows.extend(self._parseTomato(data, router.name)) | |
91 | |
92 for r in rows: | |
93 try: | |
94 r['name'] = self.knownMacAddr[r['mac']] | |
95 except KeyError: | |
96 pass | |
97 | |
98 returnValue(rows) | |
99 | |
100 def _parseZyxel(self, data, routerName): | |
101 import lxml.html.soupparser | |
102 | |
103 root = lxml.html.soupparser.fromstring(data) | |
104 for tr in root.cssselect('tr'): | |
105 mac, assoc, uth, ssid, iface = [td.text_content().strip() for td in tr.getchildren()] | |
106 if mac == "MAC": | |
107 continue | |
108 assoc = assoc.lower() == 'yes' | |
109 yield dict(router=routerName, mac=mac, assoc=assoc, connected=assoc) | |
110 | |
111 def _parseTomato(self, data, routerName): | |
112 for iface, mac, signal in jsValue(data, 'wldev'): | |
113 yield dict(router=routerName, mac=mac, signal=signal, connected=bool(signal)) | |
114 | |
115 | |
116 @inlineCallbacks | |
117 def loadUvaData(): | |
118 import lxml.html.soupparser | |
119 | |
120 config = json.load(open("priv-uva.json")) | |
121 headers = {'Authorization': ['Basic %s' % config['userPass'].encode('base64').strip()]} | |
122 resp = yield fetch('http://10.2.0.2/wlstationlist.cmd', headers=headers) | |
123 root = lxml.html.soupparser.fromstring(resp.body) | |
124 byMac = {} | |
125 for tr in root.cssselect('tr'): | |
126 mac, connected, auth, ssid, iface = [td.text_content().strip() for td in tr.getchildren()] | |
127 if mac == "MAC": | |
128 continue | |
129 connected = connected.lower() == 'yes' | |
130 byMac[mac] = dict(mac=mac, connected=connected, auth=auth == 'Yes', ssid=ssid, iface=iface) | |
131 | |
132 resp = yield fetch('http://10.2.0.2/DHCPTable.asp', headers=headers) | |
133 for row in re.findall(r'new AAA\((.*)\)', resp.body): | |
134 clientHostname, ipaddr, mac, expires, iface = [s.strip("'") for s in row.rsplit(',', 4)] | |
135 if clientHostname == 'wlanadv.none': | |
136 continue | |
137 byMac.setdefault(mac, {}).update(dict( | |
138 clientHostname=clientHostname, connection=iface, ipaddr=ipaddr, dhcpExpires=expires)) | |
139 | |
140 returnValue(sorted(byMac.values())) | |
141 | |
142 @inlineCallbacks | |
143 def loadCiscoData(): | |
144 config = json.load(open("priv-uva.json")) | |
145 headers = {'Authorization': ['Basic %s' % config['userPass'].encode('base64').strip()]} | |
146 print(headers) | |
147 resp = yield fetch('http://10.2.0.2/', headers=headers) | |
148 print(resp.body) | |
149 returnValue([]) | |
150 | |
151 @inlineCallbacks | |
152 def loadOrbiData(): | |
153 config = json.load(open("priv-uva.json")) | |
154 headers = {b'Authorization': [ | |
155 b'Basic %s' % base64.encodebytes(config['userPass'].encode('utf8')).strip()]} | |
156 resp = yield fetch(b'http://orbi.bigasterisk.com/DEV_device_info.htm', method=b'GET', headers=headers) | |
157 print('back from fetch') | |
158 | |
159 if not resp.body.startswith((b'device=', b'device_changed=0\ndevice=', b'device_changed=1\ndevice=')): | |
160 raise ValueError(resp.body) | |
161 | |
162 ret = [] | |
163 for row in json.loads(resp.body.split(b'device=', 1)[-1]): | |
164 ret.append(dict( | |
165 connected=True, | |
166 ipaddr=row['ip'], | |
167 mac=row['mac'].lower(), | |
168 contype=row['contype'], | |
169 model=row['model'], | |
170 clientHostname=row['name'] if row['name'] != 'Unknown' else None)) | |
171 returnValue(ret) | |
172 | |
173 | |
174 def jsValue(js, variableName): | |
175 # using literal_eval instead of json parser to handle the trailing commas | |
176 val = re.search(variableName + r'\s*=\s*(.*?);', js, re.DOTALL).group(1) | |
177 return ast.literal_eval(val) |