wxbot.py 20 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543
  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. def get_uuid(self):
  43. url = 'https://login.weixin.qq.com/jslogin'
  44. params = {
  45. 'appid': 'wx782c26e4c19acffb',
  46. 'fun': 'new',
  47. 'lang': 'zh_CN',
  48. '_': int(time.time())*1000 + random.randint(1,999),
  49. }
  50. r = self.session.get(url, params=params)
  51. r.encoding = 'utf-8'
  52. data = r.text
  53. regx = r'window.QRLogin.code = (\d+); window.QRLogin.uuid = "(\S+?)"'
  54. pm = re.search(regx, data)
  55. if pm:
  56. code = pm.group(1)
  57. self.uuid = pm.group(2)
  58. return code == '200'
  59. return False
  60. def gen_qr_code(self, qr_file_path):
  61. string = 'https://login.weixin.qq.com/l/' + self.uuid
  62. qr = qrcode.QRCode()
  63. qr.border = 1
  64. qr.add_data(string)
  65. qr.make(fit=True)
  66. img = qr.make_image()
  67. img.save(qr_file_path)
  68. def wait4login(self, tip):
  69. time.sleep(tip)
  70. url = 'https://login.weixin.qq.com/cgi-bin/mmwebwx-bin/login?tip=%s&uuid=%s&_=%s' % (tip, self.uuid, int(time.time()))
  71. r = self.session.get(url)
  72. r.encoding = 'utf-8'
  73. data = r.text
  74. param = re.search(r'window.code=(\d+);', data)
  75. code = param.group(1)
  76. if code == '201':
  77. return True
  78. elif code == '200':
  79. param = re.search(r'window.redirect_uri="(\S+?)";', data)
  80. redirect_uri = param.group(1) + '&fun=new'
  81. self.redirect_uri = redirect_uri
  82. self.base_uri = redirect_uri[:redirect_uri.rfind('/')]
  83. return True
  84. elif code == '408':
  85. print '[ERROR] WeChat login timeout .'
  86. else:
  87. print '[ERROR] WeChat login exception .'
  88. return False
  89. def login(self):
  90. r = self.session.get(self.redirect_uri)
  91. r.encoding = 'utf-8'
  92. data = r.text
  93. doc = xml.dom.minidom.parseString(data)
  94. root = doc.documentElement
  95. for node in root.childNodes:
  96. if node.nodeName == 'skey':
  97. self.skey = node.childNodes[0].data
  98. elif node.nodeName == 'wxsid':
  99. self.sid = node.childNodes[0].data
  100. elif node.nodeName == 'wxuin':
  101. self.uin = node.childNodes[0].data
  102. elif node.nodeName == 'pass_ticket':
  103. self.pass_ticket = node.childNodes[0].data
  104. if '' in (self.skey, self.sid, self.uin, self.pass_ticket):
  105. return False
  106. self.base_request = {
  107. 'Uin': self.uin,
  108. 'Sid': self.sid,
  109. 'Skey': self.skey,
  110. 'DeviceID': self.device_id,
  111. }
  112. return True
  113. def init(self):
  114. url = self.base_uri + '/webwxinit?r=%i&lang=en_US&pass_ticket=%s' % (int(time.time()), self.pass_ticket)
  115. params = {
  116. 'BaseRequest': self.base_request
  117. }
  118. r = self.session.post(url, json=params)
  119. r.encoding = 'utf-8'
  120. dic = json.loads(r.text)
  121. self.sync_key = dic['SyncKey']
  122. self.user = dic['User']
  123. self.sync_key_str = '|'.join([ str(keyVal['Key']) + '_' + str(keyVal['Val']) for keyVal in self.sync_key['List'] ])
  124. return dic['BaseResponse']['Ret'] == 0
  125. def status_notify(self):
  126. url = self.base_uri + '/webwxstatusnotify?lang=zh_CN&pass_ticket=%s' % (self.pass_ticket)
  127. self.base_request['Uin'] = int(self.base_request['Uin'])
  128. params = {
  129. 'BaseRequest': self.base_request,
  130. "Code": 3,
  131. "FromUserName": self.user['UserName'],
  132. "ToUserName": self.user['UserName'],
  133. "ClientMsgId": int(time.time())
  134. }
  135. r = self.session.post(url, json=params)
  136. r.encoding = 'utf-8'
  137. dic = json.loads(r.text)
  138. return dic['BaseResponse']['Ret'] == 0
  139. def get_contact(self):
  140. url = self.base_uri + '/webwxgetcontact?pass_ticket=%s&skey=%s&r=%s' % (self.pass_ticket, self.skey, int(time.time()))
  141. r = self.session.post(url, json={})
  142. r.encoding = 'utf-8'
  143. if self.DEBUG:
  144. with open('contacts.json', 'w') as f:
  145. f.write(r.text.encode('utf-8'))
  146. dic = json.loads(r.text)
  147. self.member_list = dic['MemberList']
  148. contact_list = self.member_list[:]
  149. SpecialUsers = ['newsapp','fmessage','filehelper','weibo','qqmail','fmessage','tmessage','qmessage','qqsync','floatbottle','lbsapp','shakeapp','medianote',
  150. 'qqfriend','readerapp','blogapp','facebookapp','masssendapp','meishiapp','feedsapp','voip','blogappweixin','weixin','brandsessionholder','weixinreminder','wxid_novlwrv3lqwv11',
  151. 'gh_22b87fa7cb3c','officialaccounts','notification_messages','wxid_novlwrv3lqwv11','gh_22b87fa7cb3c','wxitil','userexperience_alarm','notification_messages']
  152. for contact in contact_list:
  153. if contact['VerifyFlag'] & 8 != 0: # public account
  154. contact_list.remove(contact)
  155. elif contact['UserName'] in SpecialUsers: # special account
  156. contact_list.remove(contact)
  157. elif contact['UserName'].find('@@') != -1: # group
  158. self.group_list.append(contact)
  159. contact_list.remove(contact)
  160. elif contact['UserName'] == self.user['UserName']: # self
  161. contact_list.remove(contact)
  162. self.contact_list = contact_list
  163. if self.DEBUG:
  164. with open('contactlist.json', 'w') as f:
  165. f.write(json.dumps(self.contact_list))
  166. return True
  167. def batch_get_contact(self):
  168. url = self.base_uri + '/webwxbatchgetcontact?type=ex&r=%s&pass_ticket=%s' % (int(time.time()), self.pass_ticket)
  169. params = {
  170. 'BaseRequest': self.base_request,
  171. "Count": len(self.group_list),
  172. "List": [ {"UserName": g['UserName'], "EncryChatRoomId":""} for g in self.group_list ]
  173. }
  174. r = self.session.post(url, data=params)
  175. r.encoding = 'utf-8'
  176. dic = json.loads(r.text)
  177. return True
  178. def test_sync_check(self):
  179. for host in ['webpush', 'webpush2']:
  180. self.sync_host = host
  181. [retcode, selector] = self.sync_check()
  182. if retcode == '0':
  183. return True
  184. return False
  185. def sync_check(self):
  186. params = {
  187. 'r': int(time.time()),
  188. 'sid': self.sid,
  189. 'uin': self.uin,
  190. 'skey': self.skey,
  191. 'deviceid': self.device_id,
  192. 'synckey': self.sync_key_str,
  193. '_': int(time.time()),
  194. }
  195. url = 'https://' + self.sync_host + '.weixin.qq.com/cgi-bin/mmwebwx-bin/synccheck?' + urllib.urlencode(params)
  196. r = self.session.get(url)
  197. r.encoding = 'utf-8'
  198. data = r.text
  199. pm = re.search(r'window.synccheck={retcode:"(\d+)",selector:"(\d+)"}', data)
  200. retcode = pm.group(1)
  201. selector = pm.group(2)
  202. return [retcode, selector]
  203. def sync(self):
  204. url = self.base_uri + '/webwxsync?sid=%s&skey=%s&lang=en_US&pass_ticket=%s' % (self.sid, self.skey, self.pass_ticket)
  205. params = {
  206. 'BaseRequest': self.base_request,
  207. 'SyncKey': self.sync_key,
  208. 'rr': ~int(time.time())
  209. }
  210. r = self.session.post(url, json=params)
  211. r.encoding = 'utf-8'
  212. dic = json.loads(r.text)
  213. if dic['BaseResponse']['Ret'] == 0:
  214. self.sync_key = dic['SyncKey']
  215. self.sync_key_str = '|'.join([ str(keyVal['Key']) + '_' + str(keyVal['Val']) for keyVal in self.sync_key['List'] ])
  216. return dic
  217. def get_icon(self, id):
  218. url = self.base_uri + '/webwxgeticon?username=%s&skey=%s' % (id, self.skey)
  219. r = self.session.get(url)
  220. data = r.content
  221. fn = 'img_'+id+'.jpg'
  222. with open(fn, 'wb') as f:
  223. f.write(data)
  224. return fn
  225. def get_head_img(self, id):
  226. url = self.base_uri + '/webwxgetheadimg?username=%s&skey=%s' % (id, self.skey)
  227. r = self.session.get(url)
  228. data = r.content
  229. fn = 'img_'+id+'.jpg'
  230. with open(fn, 'wb') as f:
  231. f.write(data)
  232. return fn
  233. def get_msg_img_url(self, msgid):
  234. return self.base_uri + '/webwxgetmsgimg?MsgID=%s&skey=%s' % (msgid, self.skey)
  235. def get_msg_img(self, msgid):
  236. url = self.base_uri + '/webwxgetmsgimg?MsgID=%s&skey=%s' % (msgid, self.skey)
  237. r = self.session.get(url)
  238. data = r.content
  239. fn = 'img_'+msgid+'.jpg'
  240. with open(fn, 'wb') as f:
  241. f.write(data)
  242. return fn
  243. def get_voice_url(self, msgid):
  244. return self.base_uri + '/webwxgetvoice?msgid=%s&skey=%s' % (msgid, self.skey)
  245. def get_voice(self, msgid):
  246. url = self.base_uri + '/webwxgetvoice?msgid=%s&skey=%s' % (msgid, self.skey)
  247. r = self.session.get(url)
  248. data = r.content
  249. fn = 'voice_'+msgid+'.mp3'
  250. with open(fn, 'wb') as f:
  251. f.write(data)
  252. return fn
  253. #Get the NickName or RemarkName of an user by user id
  254. def get_user_remark_name(self, uid):
  255. name = 'unknown group' if uid[:2] == '@@' else 'stranger'
  256. for member in self.member_list:
  257. if member['UserName'] == uid:
  258. name = member['RemarkName'] if member['RemarkName'] else member['NickName']
  259. return name
  260. #Get user id of an user
  261. def get_user_id(self, name):
  262. for member in self.member_list:
  263. if name == member['RemarkName'] or name == member['NickName'] or name == member['UserName']:
  264. return member['UserName']
  265. return None
  266. '''
  267. msg:
  268. msg_id
  269. msg_type_id
  270. user_id
  271. user_name
  272. content
  273. '''
  274. def handle_msg_all(self, msg):
  275. pass
  276. '''
  277. msg_type_id:
  278. 1 -> Location
  279. 2 -> FileHelper
  280. 3 -> Self
  281. 4 -> Group
  282. 5 -> User Text Message
  283. 6 -> Image
  284. 7 -> Voice
  285. 8 -> Recommend
  286. 9 -> Animation
  287. 10 -> Share
  288. 11 -> Video
  289. 12 -> Video Call
  290. 13 -> Redraw
  291. 14 -> Init Message
  292. 99 -> Unknown
  293. '''
  294. def handle_msg(self, r):
  295. for msg in r['AddMsgList']:
  296. mtype = msg['MsgType']
  297. name = self.get_user_remark_name(msg['FromUserName'])
  298. content = msg['Content'].replace('&lt;','<').replace('&gt;','>')
  299. msg_id = msg['MsgId']
  300. msg_type_id = 99
  301. if mtype == 51: #init message
  302. msg_type_id = 14
  303. elif mtype == 1:
  304. if content.find('http://weixin.qq.com/cgi-bin/redirectforward?args=') != -1:
  305. r = self.session.get(content)
  306. r.encoding = 'gbk'
  307. data = r.text
  308. pos = self.search_content('title', data, 'xml')
  309. msg_type_id = 1
  310. content = {'location': pos, 'xml': data}
  311. if self.DEBUG:
  312. print '[Location] %s : I am at %s ' % (name, pos)
  313. elif msg['ToUserName'] == 'filehelper':
  314. msg_type_id = 2
  315. content = content.replace('<br/>','\n')
  316. if self.DEBUG:
  317. print '[File] %s : %s' % (name, )
  318. elif msg['FromUserName'] == self.user['UserName']: #self
  319. msg_type_id = 3
  320. elif msg['FromUserName'][:2] == '@@':
  321. [people, content] = content.split(':<br/>')
  322. group = self.get_user_remark_name(msg['FromUserName'])
  323. name = self.get_user_remark_name(people)
  324. msg_type_id = 4
  325. content = {'group_id': msg['FromUserName'], 'group_name': group, 'user': people, 'user_name': name, 'msg': content}
  326. if self.DEBUG:
  327. print '[Group] |%s| %s: %s' % (group, name, content.replace('<br/>','\n'))
  328. else:
  329. msg_type_id = 5
  330. if self.DEBUG:
  331. print '[Text] ', name, ' : ', content
  332. elif mtype == 3:
  333. msg_type_id = 6
  334. content = self.get_msg_img_url(msg_id)
  335. if self.DEBUG:
  336. image = self.get_msg_img(msg_id)
  337. print '[Image] %s : %s' % (name, image)
  338. elif mtype == 34:
  339. msg_type_id = 7
  340. content = self.get_voice_url(msg_id)
  341. if self.DEBUG:
  342. voice = self.get_voice(msg_id)
  343. print '[Voice] %s : %s' % (name, voice)
  344. elif mtype == 42:
  345. msg_type_id = 8
  346. info = msg['RecommendInfo']
  347. content = {}
  348. content['nickname'] = info['NickName']
  349. content['alias'] = info['Alias']
  350. content['province'] = info['Province']
  351. content['city'] = info['City']
  352. content['gender'] = ['unknown', 'male', 'female'][info['Sex']]
  353. if self.DEBUG:
  354. print '[Recommend] %s : ' % name
  355. print '========================='
  356. print '= NickName: %s' % info['NickName']
  357. print '= Alias: %s' % info['Alias']
  358. print '= Local: %s %s' % (info['Province'], info['City'])
  359. print '= Gender: %s' % ['unknown', 'male', 'female'][info['Sex']]
  360. print '========================='
  361. elif mtype == 47:
  362. msg_type_id = 9
  363. url = self.search_content('cdnurl', content)
  364. content = url
  365. if self.DEBUG:
  366. print '[Animation] %s : %s' % (name, url)
  367. elif mtype == 49:
  368. msg_type_id = 10
  369. appMsgType = defaultdict(lambda : "")
  370. appMsgType.update({5:'link', 3:'music', 7:'weibo'})
  371. 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')}
  372. if self.DEBUG:
  373. print '[Share] %s : %s' % (name, appMsgType[msg['AppMsgType']])
  374. print '========================='
  375. print '= title: %s' % msg['FileName']
  376. print '= desc: %s' % self.search_content('des', content, 'xml')
  377. print '= link: %s' % msg['Url']
  378. print '= from: %s' % self.search_content('appname', content, 'xml')
  379. print '========================='
  380. elif mtype == 62:
  381. msg_type_id = 11
  382. if self.DEBUG:
  383. print '[Video] ', name, ' sent you a video, please check on mobiles'
  384. elif mtype == 53:
  385. msg_type_id = 12
  386. if self.DEBUG:
  387. print '[Video Call] ', name, ' call you'
  388. elif mtype == 10002:
  389. msg_type_id = 13
  390. if self.DEBUG:
  391. print '[Redraw] ', name, ' redraw back a message'
  392. else:
  393. msg_type_id = 99
  394. if self.DEBUG:
  395. print '[Unknown] : %s' % str(mtype)
  396. print msg
  397. message = {'msg_id':msg_id, 'msg_type_id': msg_type_id, 'content': content, 'user_id': msg['FromUserName'], 'user_name': name}
  398. self.handle_msg_all(message)
  399. def schedule(self):
  400. pass
  401. def proc_msg(self):
  402. self.test_sync_check()
  403. while True:
  404. [retcode, selector] = self.sync_check()
  405. if retcode == '1100': # User have login on mobile
  406. pass
  407. elif retcode == '0':
  408. if selector == '2':
  409. r = self.sync()
  410. if r is not None:
  411. self.handle_msg(r)
  412. elif selector == '7': # Play WeChat on mobile
  413. r = self.sync()
  414. if r is not None:
  415. self.handle_msg(r)
  416. elif selector == '0':
  417. time.sleep(1)
  418. self.schedule()
  419. def send_msg_by_uid(self, word, dst = 'filehelper'):
  420. url = self.base_uri + '/webwxsendmsg?pass_ticket=%s' % (self.pass_ticket)
  421. msg_id = str(int(time.time()*1000)) + str(random.random())[:5].replace('.','')
  422. params = {
  423. 'BaseRequest': self.base_request,
  424. 'Msg': {
  425. "Type": 1,
  426. "Content": make_unicode(word),
  427. "FromUserName": self.user['UserName'],
  428. "ToUserName": dst,
  429. "LocalID": msg_id,
  430. "ClientMsgId": msg_id
  431. }
  432. }
  433. headers = {'content-type': 'application/json; charset=UTF-8'}
  434. data = json.dumps(params, ensure_ascii=False).encode('utf8')
  435. r = self.session.post(url, data = data, headers = headers)
  436. dic = r.json()
  437. return dic['BaseResponse']['Ret'] == 0
  438. def send_msg(self, name, word, isfile = False):
  439. uid = self.get_user_id(name)
  440. if uid:
  441. if isfile:
  442. with open(word, 'r') as f:
  443. result = True
  444. for line in f.readlines():
  445. line = line.replace('\n','')
  446. print '-> '+name+': '+line
  447. if self.send_msg_by_uid(line, uid):
  448. pass
  449. else:
  450. result = False
  451. time.sleep(1)
  452. return result
  453. else:
  454. if self.send_msg_by_uid(word, uid):
  455. return True
  456. else:
  457. return False
  458. else:
  459. if self.DEBUG:
  460. print '[ERROR] This user does not exist .'
  461. return False
  462. def search_content(self, key, content, fmat = 'attr'):
  463. if fmat == 'attr':
  464. pm = re.search(key+'\s?=\s?"([^"<]+)"', content)
  465. if pm: return pm.group(1)
  466. elif fmat == 'xml':
  467. pm=re.search('<{0}>([^<]+)</{0}>'.format(key),content)
  468. if pm: return pm.group(1)
  469. return 'unknown'
  470. def run(self):
  471. self.get_uuid()
  472. self.gen_qr_code('qr.jpg')
  473. print '[INFO] Please use WeCaht to scan QR code in qr.jpg .'
  474. self.wait4login(1)
  475. print '[INFO] Please confirm to login .'
  476. self.wait4login(0)
  477. if self.login():
  478. print '[INFO] Web WeChat login succeed .'
  479. else:
  480. print '[ERROR] Web WeChat login failed .'
  481. return
  482. if self.init():
  483. print '[INFO] Web WeChat init succeed .'
  484. else:
  485. print '[INFO] Web WeChat init failed'
  486. return
  487. self.status_notify()
  488. self.get_contact()
  489. print '[INFO] Get %d contacts' % len(self.contact_list)
  490. print '[INFO] Start to process messages .'
  491. self.proc_msg()