wxbot.py 21 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548
  1. #!/usr/bin/env python
  2. # coding: utf-8
  3. import qrcode
  4. import requests
  5. import json
  6. import xml.dom.minidom
  7. import multiprocessing
  8. import urllib
  9. import time, re, sys, os, random
  10. def utf82gbk(string):
  11. return string.decode('utf8').encode('gbk')
  12. def make_unicode(data):
  13. if not data:
  14. return data
  15. result = None
  16. if type(data) == unicode:
  17. result = data
  18. elif type(data) == str:
  19. result = data.decode('utf-8')
  20. return result
  21. class WXBot:
  22. def __init__(self):
  23. self.DEBUG = False
  24. self.uuid = ''
  25. self.base_uri = ''
  26. self.redirect_uri= ''
  27. self.uin = ''
  28. self.sid = ''
  29. self.skey = ''
  30. self.pass_ticket = ''
  31. self.device_id = 'e' + repr(random.random())[2:17]
  32. self.base_request = {}
  33. self.sync_key_str = ''
  34. self.sync_key = []
  35. self.user = []
  36. self.member_list = []
  37. self.contact_list = []
  38. self.group_list = []
  39. self.sync_host = ''
  40. self.session = requests.Session()
  41. self.session.headers.update({'User-Agent': 'Mozilla/5.0 (X11; Linux i686; U;) Gecko/20070322 Kazehakase/0.4.5'})
  42. self.conf = {'qr': 'tty',}
  43. def get_uuid(self):
  44. url = 'https://login.weixin.qq.com/jslogin'
  45. params = {
  46. 'appid': 'wx782c26e4c19acffb',
  47. 'fun': 'new',
  48. 'lang': 'zh_CN',
  49. '_': int(time.time())*1000 + random.randint(1,999),
  50. }
  51. r = self.session.get(url, params=params)
  52. r.encoding = 'utf-8'
  53. data = r.text
  54. regx = r'window.QRLogin.code = (\d+); window.QRLogin.uuid = "(\S+?)"'
  55. pm = re.search(regx, data)
  56. if pm:
  57. code = pm.group(1)
  58. self.uuid = pm.group(2)
  59. return code == '200'
  60. return False
  61. def gen_qr_code(self, qr_file_path):
  62. string = 'https://login.weixin.qq.com/l/' + self.uuid
  63. qr = qrcode.QRCode()
  64. qr.border = 1
  65. qr.add_data(string)
  66. qr.make(fit=True)
  67. if self.conf['qr'] == 'png':
  68. img = qr.make_image()
  69. img.save(qr_file_path)
  70. elif self.conf['qr'] == 'tty':
  71. qr.print_tty()
  72. def wait4login(self, tip):
  73. time.sleep(tip)
  74. url = 'https://login.weixin.qq.com/cgi-bin/mmwebwx-bin/login?tip=%s&uuid=%s&_=%s' % (tip, self.uuid, int(time.time()))
  75. r = self.session.get(url)
  76. r.encoding = 'utf-8'
  77. data = r.text
  78. param = re.search(r'window.code=(\d+);', data)
  79. code = param.group(1)
  80. if code == '201':
  81. return True
  82. elif code == '200':
  83. param = re.search(r'window.redirect_uri="(\S+?)";', data)
  84. redirect_uri = param.group(1) + '&fun=new'
  85. self.redirect_uri = redirect_uri
  86. self.base_uri = redirect_uri[:redirect_uri.rfind('/')]
  87. return True
  88. elif code == '408':
  89. print '[ERROR] WeChat login timeout .'
  90. else:
  91. print '[ERROR] WeChat login exception .'
  92. return False
  93. def login(self):
  94. r = self.session.get(self.redirect_uri)
  95. r.encoding = 'utf-8'
  96. data = r.text
  97. doc = xml.dom.minidom.parseString(data)
  98. root = doc.documentElement
  99. for node in root.childNodes:
  100. if node.nodeName == 'skey':
  101. self.skey = node.childNodes[0].data
  102. elif node.nodeName == 'wxsid':
  103. self.sid = node.childNodes[0].data
  104. elif node.nodeName == 'wxuin':
  105. self.uin = node.childNodes[0].data
  106. elif node.nodeName == 'pass_ticket':
  107. self.pass_ticket = node.childNodes[0].data
  108. if '' in (self.skey, self.sid, self.uin, self.pass_ticket):
  109. return False
  110. self.base_request = {
  111. 'Uin': self.uin,
  112. 'Sid': self.sid,
  113. 'Skey': self.skey,
  114. 'DeviceID': self.device_id,
  115. }
  116. return True
  117. def init(self):
  118. url = self.base_uri + '/webwxinit?r=%i&lang=en_US&pass_ticket=%s' % (int(time.time()), self.pass_ticket)
  119. params = {
  120. 'BaseRequest': self.base_request
  121. }
  122. r = self.session.post(url, data=json.dumps(params))
  123. r.encoding = 'utf-8'
  124. dic = json.loads(r.text)
  125. self.sync_key = dic['SyncKey']
  126. self.user = dic['User']
  127. self.sync_key_str = '|'.join([ str(keyVal['Key']) + '_' + str(keyVal['Val']) for keyVal in self.sync_key['List'] ])
  128. return dic['BaseResponse']['Ret'] == 0
  129. def status_notify(self):
  130. url = self.base_uri + '/webwxstatusnotify?lang=zh_CN&pass_ticket=%s' % (self.pass_ticket)
  131. self.base_request['Uin'] = int(self.base_request['Uin'])
  132. params = {
  133. 'BaseRequest': self.base_request,
  134. "Code": 3,
  135. "FromUserName": self.user['UserName'],
  136. "ToUserName": self.user['UserName'],
  137. "ClientMsgId": int(time.time())
  138. }
  139. r = self.session.post(url, data=json.dumps(params))
  140. r.encoding = 'utf-8'
  141. dic = json.loads(r.text)
  142. return dic['BaseResponse']['Ret'] == 0
  143. def get_contact(self):
  144. url = self.base_uri + '/webwxgetcontact?pass_ticket=%s&skey=%s&r=%s' % (self.pass_ticket, self.skey, int(time.time()))
  145. r = self.session.post(url, data='{}')
  146. r.encoding = 'utf-8'
  147. if self.DEBUG:
  148. with open('contacts.json', 'w') as f:
  149. f.write(r.text.encode('utf-8'))
  150. dic = json.loads(r.text)
  151. self.member_list = dic['MemberList']
  152. contact_list = self.member_list[:]
  153. SpecialUsers = ['newsapp','fmessage','filehelper','weibo','qqmail','fmessage','tmessage','qmessage','qqsync','floatbottle','lbsapp','shakeapp','medianote',
  154. 'qqfriend','readerapp','blogapp','facebookapp','masssendapp','meishiapp','feedsapp','voip','blogappweixin','weixin','brandsessionholder','weixinreminder','wxid_novlwrv3lqwv11',
  155. 'gh_22b87fa7cb3c','officialaccounts','notification_messages','wxid_novlwrv3lqwv11','gh_22b87fa7cb3c','wxitil','userexperience_alarm','notification_messages']
  156. for contact in contact_list:
  157. if contact['VerifyFlag'] & 8 != 0: # public account
  158. contact_list.remove(contact)
  159. elif contact['UserName'] in SpecialUsers: # special account
  160. contact_list.remove(contact)
  161. elif contact['UserName'].find('@@') != -1: # group
  162. self.group_list.append(contact)
  163. contact_list.remove(contact)
  164. elif contact['UserName'] == self.user['UserName']: # self
  165. contact_list.remove(contact)
  166. self.contact_list = contact_list
  167. if self.DEBUG:
  168. with open('contactlist.json', 'w') as f:
  169. f.write(json.dumps(self.contact_list))
  170. return True
  171. def batch_get_contact(self):
  172. url = self.base_uri + '/webwxbatchgetcontact?type=ex&r=%s&pass_ticket=%s' % (int(time.time()), self.pass_ticket)
  173. params = {
  174. 'BaseRequest': self.base_request,
  175. "Count": len(self.group_list),
  176. "List": [ {"UserName": g['UserName'], "EncryChatRoomId":""} for g in self.group_list ]
  177. }
  178. r = self.session.post(url, data=params)
  179. r.encoding = 'utf-8'
  180. dic = json.loads(r.text)
  181. return True
  182. def test_sync_check(self):
  183. for host in ['webpush', 'webpush2']:
  184. self.sync_host = host
  185. [retcode, selector] = self.sync_check()
  186. if retcode == '0':
  187. return True
  188. return False
  189. def sync_check(self):
  190. params = {
  191. 'r': int(time.time()),
  192. 'sid': self.sid,
  193. 'uin': self.uin,
  194. 'skey': self.skey,
  195. 'deviceid': self.device_id,
  196. 'synckey': self.sync_key_str,
  197. '_': int(time.time()),
  198. }
  199. url = 'https://' + self.sync_host + '.weixin.qq.com/cgi-bin/mmwebwx-bin/synccheck?' + urllib.urlencode(params)
  200. r = self.session.get(url)
  201. r.encoding = 'utf-8'
  202. data = r.text
  203. pm = re.search(r'window.synccheck={retcode:"(\d+)",selector:"(\d+)"}', data)
  204. retcode = pm.group(1)
  205. selector = pm.group(2)
  206. return [retcode, selector]
  207. def sync(self):
  208. url = self.base_uri + '/webwxsync?sid=%s&skey=%s&lang=en_US&pass_ticket=%s' % (self.sid, self.skey, self.pass_ticket)
  209. params = {
  210. 'BaseRequest': self.base_request,
  211. 'SyncKey': self.sync_key,
  212. 'rr': ~int(time.time())
  213. }
  214. r = self.session.post(url, data=json.dumps(params))
  215. r.encoding = 'utf-8'
  216. dic = json.loads(r.text)
  217. if dic['BaseResponse']['Ret'] == 0:
  218. self.sync_key = dic['SyncKey']
  219. self.sync_key_str = '|'.join([ str(keyVal['Key']) + '_' + str(keyVal['Val']) for keyVal in self.sync_key['List'] ])
  220. return dic
  221. def get_icon(self, id):
  222. url = self.base_uri + '/webwxgeticon?username=%s&skey=%s' % (id, self.skey)
  223. r = self.session.get(url)
  224. data = r.content
  225. fn = 'img_'+id+'.jpg'
  226. with open(fn, 'wb') as f:
  227. f.write(data)
  228. return fn
  229. def get_head_img(self, id):
  230. url = self.base_uri + '/webwxgetheadimg?username=%s&skey=%s' % (id, self.skey)
  231. r = self.session.get(url)
  232. data = r.content
  233. fn = 'img_'+id+'.jpg'
  234. with open(fn, 'wb') as f:
  235. f.write(data)
  236. return fn
  237. def get_msg_img_url(self, msgid):
  238. return self.base_uri + '/webwxgetmsgimg?MsgID=%s&skey=%s' % (msgid, self.skey)
  239. def get_msg_img(self, msgid):
  240. url = self.base_uri + '/webwxgetmsgimg?MsgID=%s&skey=%s' % (msgid, self.skey)
  241. r = self.session.get(url)
  242. data = r.content
  243. fn = 'img_'+msgid+'.jpg'
  244. with open(fn, 'wb') as f:
  245. f.write(data)
  246. return fn
  247. def get_voice_url(self, msgid):
  248. return self.base_uri + '/webwxgetvoice?msgid=%s&skey=%s' % (msgid, self.skey)
  249. def get_voice(self, msgid):
  250. url = self.base_uri + '/webwxgetvoice?msgid=%s&skey=%s' % (msgid, self.skey)
  251. r = self.session.get(url)
  252. data = r.content
  253. fn = 'voice_'+msgid+'.mp3'
  254. with open(fn, 'wb') as f:
  255. f.write(data)
  256. return fn
  257. #Get the NickName or RemarkName of an user by user id
  258. def get_user_remark_name(self, uid):
  259. name = 'unknown group' if uid[:2] == '@@' else 'stranger'
  260. for member in self.member_list:
  261. if member['UserName'] == uid:
  262. name = member['RemarkName'] if member['RemarkName'] else member['NickName']
  263. return name
  264. #Get user id of an user
  265. def get_user_id(self, name):
  266. for member in self.member_list:
  267. if name == member['RemarkName'] or name == member['NickName'] or name == member['UserName']:
  268. return member['UserName']
  269. return None
  270. '''
  271. msg:
  272. msg_id
  273. msg_type_id
  274. user_id
  275. user_name
  276. content
  277. '''
  278. def handle_msg_all(self, msg):
  279. pass
  280. '''
  281. msg_type_id:
  282. 1 -> Location
  283. 2 -> FileHelper
  284. 3 -> Self
  285. 4 -> Group
  286. 5 -> User Text Message
  287. 6 -> Image
  288. 7 -> Voice
  289. 8 -> Recommend
  290. 9 -> Animation
  291. 10 -> Share
  292. 11 -> Video
  293. 12 -> Video Call
  294. 13 -> Redraw
  295. 14 -> Init Message
  296. 99 -> Unknown
  297. '''
  298. def handle_msg(self, r):
  299. for msg in r['AddMsgList']:
  300. mtype = msg['MsgType']
  301. name = self.get_user_remark_name(msg['FromUserName'])
  302. content = msg['Content'].replace('&lt;','<').replace('&gt;','>')
  303. msg_id = msg['MsgId']
  304. msg_type_id = 99
  305. if mtype == 51: #init message
  306. msg_type_id = 14
  307. elif mtype == 1:
  308. if content.find('http://weixin.qq.com/cgi-bin/redirectforward?args=') != -1:
  309. r = self.session.get(content)
  310. r.encoding = 'gbk'
  311. data = r.text
  312. pos = self.search_content('title', data, 'xml')
  313. msg_type_id = 1
  314. content = {'location': pos, 'xml': data}
  315. if self.DEBUG:
  316. print '[Location] %s : I am at %s ' % (name, pos)
  317. elif msg['ToUserName'] == 'filehelper':
  318. msg_type_id = 2
  319. content = content.replace('<br/>','\n')
  320. if self.DEBUG:
  321. print '[File] %s : %s' % (name, )
  322. elif msg['FromUserName'] == self.user['UserName']: #self
  323. msg_type_id = 3
  324. elif msg['FromUserName'][:2] == '@@':
  325. [people, content] = content.split(':<br/>')
  326. group = self.get_user_remark_name(msg['FromUserName'])
  327. name = self.get_user_remark_name(people)
  328. msg_type_id = 4
  329. content = {'group_id': msg['FromUserName'], 'group_name': group, 'user': people, 'user_name': name, 'msg': content}
  330. if self.DEBUG:
  331. print '[Group] |%s| %s: %s' % (group, name, content.replace('<br/>','\n'))
  332. else:
  333. msg_type_id = 5
  334. if self.DEBUG:
  335. print '[Text] ', name, ' : ', content
  336. elif mtype == 3:
  337. msg_type_id = 6
  338. content = self.get_msg_img_url(msg_id)
  339. if self.DEBUG:
  340. image = self.get_msg_img(msg_id)
  341. print '[Image] %s : %s' % (name, image)
  342. elif mtype == 34:
  343. msg_type_id = 7
  344. content = self.get_voice_url(msg_id)
  345. if self.DEBUG:
  346. voice = self.get_voice(msg_id)
  347. print '[Voice] %s : %s' % (name, voice)
  348. elif mtype == 42:
  349. msg_type_id = 8
  350. info = msg['RecommendInfo']
  351. content = {}
  352. content['nickname'] = info['NickName']
  353. content['alias'] = info['Alias']
  354. content['province'] = info['Province']
  355. content['city'] = info['City']
  356. content['gender'] = ['unknown', 'male', 'female'][info['Sex']]
  357. if self.DEBUG:
  358. print '[Recommend] %s : ' % name
  359. print '========================='
  360. print '= NickName: %s' % info['NickName']
  361. print '= Alias: %s' % info['Alias']
  362. print '= Local: %s %s' % (info['Province'], info['City'])
  363. print '= Gender: %s' % ['unknown', 'male', 'female'][info['Sex']]
  364. print '========================='
  365. elif mtype == 47:
  366. msg_type_id = 9
  367. url = self.search_content('cdnurl', content)
  368. content = url
  369. if self.DEBUG:
  370. print '[Animation] %s : %s' % (name, url)
  371. elif mtype == 49:
  372. msg_type_id = 10
  373. appMsgType = defaultdict(lambda : "")
  374. appMsgType.update({5:'link', 3:'music', 7:'weibo'})
  375. content = {'type': appMsgType[msg['AppMsgType']], 'title': msg['FileName'], 'desc': self.search_content('des', content, 'xml'), 'url': msg['Url'], 'from': self.search_content('appname', content, 'xml')}
  376. if self.DEBUG:
  377. print '[Share] %s : %s' % (name, appMsgType[msg['AppMsgType']])
  378. print '========================='
  379. print '= title: %s' % msg['FileName']
  380. print '= desc: %s' % self.search_content('des', content, 'xml')
  381. print '= link: %s' % msg['Url']
  382. print '= from: %s' % self.search_content('appname', content, 'xml')
  383. print '========================='
  384. elif mtype == 62:
  385. msg_type_id = 11
  386. if self.DEBUG:
  387. print '[Video] ', name, ' sent you a video, please check on mobiles'
  388. elif mtype == 53:
  389. msg_type_id = 12
  390. if self.DEBUG:
  391. print '[Video Call] ', name, ' call you'
  392. elif mtype == 10002:
  393. msg_type_id = 13
  394. if self.DEBUG:
  395. print '[Redraw] ', name, ' redraw back a message'
  396. else:
  397. msg_type_id = 99
  398. if self.DEBUG:
  399. print '[Unknown] : %s' % str(mtype)
  400. print msg
  401. message = {'msg_id':msg_id, 'msg_type_id': msg_type_id, 'content': content, 'user_id': msg['FromUserName'], 'user_name': name}
  402. self.handle_msg_all(message)
  403. def schedule(self):
  404. pass
  405. def proc_msg(self):
  406. self.test_sync_check()
  407. while True:
  408. [retcode, selector] = self.sync_check()
  409. if retcode == '1100': # User have login on mobile
  410. pass
  411. elif retcode == '0':
  412. if selector == '2':
  413. r = self.sync()
  414. if r is not None:
  415. self.handle_msg(r)
  416. elif selector == '7': # Play WeChat on mobile
  417. r = self.sync()
  418. if r is not None:
  419. self.handle_msg(r)
  420. elif selector == '0':
  421. time.sleep(1)
  422. self.schedule()
  423. def send_msg_by_uid(self, word, dst = 'filehelper'):
  424. url = self.base_uri + '/webwxsendmsg?pass_ticket=%s' % (self.pass_ticket)
  425. msg_id = str(int(time.time()*1000)) + str(random.random())[:5].replace('.','')
  426. params = {
  427. 'BaseRequest': self.base_request,
  428. 'Msg': {
  429. "Type": 1,
  430. "Content": make_unicode(word),
  431. "FromUserName": self.user['UserName'],
  432. "ToUserName": dst,
  433. "LocalID": msg_id,
  434. "ClientMsgId": msg_id
  435. }
  436. }
  437. headers = {'content-type': 'application/json; charset=UTF-8'}
  438. data = json.dumps(params, ensure_ascii=False).encode('utf8')
  439. r = self.session.post(url, data = data, headers = headers)
  440. dic = r.json()
  441. return dic['BaseResponse']['Ret'] == 0
  442. def send_msg(self, name, word, isfile = False):
  443. uid = self.get_user_id(name)
  444. if uid:
  445. if isfile:
  446. with open(word, 'r') as f:
  447. result = True
  448. for line in f.readlines():
  449. line = line.replace('\n','')
  450. print '-> '+name+': '+line
  451. if self.send_msg_by_uid(line, uid):
  452. pass
  453. else:
  454. result = False
  455. time.sleep(1)
  456. return result
  457. else:
  458. if self.send_msg_by_uid(word, uid):
  459. return True
  460. else:
  461. return False
  462. else:
  463. if self.DEBUG:
  464. print '[ERROR] This user does not exist .'
  465. return False
  466. def search_content(self, key, content, fmat = 'attr'):
  467. if fmat == 'attr':
  468. pm = re.search(key+'\s?=\s?"([^"<]+)"', content)
  469. if pm: return pm.group(1)
  470. elif fmat == 'xml':
  471. pm=re.search('<{0}>([^<]+)</{0}>'.format(key),content)
  472. if pm: return pm.group(1)
  473. return 'unknown'
  474. def run(self):
  475. self.get_uuid()
  476. self.gen_qr_code('qr.png')
  477. print '[INFO] Please use WeCaht to scan the QR code .'
  478. self.wait4login(1)
  479. print '[INFO] Please confirm to login .'
  480. self.wait4login(0)
  481. if self.login():
  482. print '[INFO] Web WeChat login succeed .'
  483. else:
  484. print '[ERROR] Web WeChat login failed .'
  485. return
  486. if self.init():
  487. print '[INFO] Web WeChat init succeed .'
  488. else:
  489. print '[INFO] Web WeChat init failed'
  490. return
  491. self.status_notify()
  492. self.get_contact()
  493. print '[INFO] Get %d contacts' % len(self.contact_list)
  494. print '[INFO] Start to process messages .'
  495. self.proc_msg()