app.py 21 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480
  1. import random
  2. import os
  3. import logging
  4. import json
  5. import time
  6. import traceback
  7. import ast
  8. from gevent import monkey
  9. monkey.patch_all()
  10. from flask import Flask, request
  11. from log import Log
  12. from config import set_config
  13. from recommend import video_homepage_recommend, video_relevant_recommend
  14. from category import get_category_videos
  15. from video_recall import PoolRecall
  16. from db_helper import RedisHelper
  17. from gevent.pywsgi import WSGIServer
  18. from multiprocessing import cpu_count, Process
  19. from utils import update_video_w_h_rate
  20. from user2new import user2new
  21. from params_helper import Params
  22. from manager_op import get_video_list, search_video
  23. from ad_recommend import ad_recommend_predict
  24. # from werkzeug.middleware.profiler import ProfilerMiddleware
  25. # from geventwebsocket.handler import WebSocketHandler
  26. import xgboost as xgb
  27. from xgboost.sklearn import XGBClassifier
  28. app = Flask(__name__)
  29. log_ = Log()
  30. config_ = set_config()
  31. # 模型加载
  32. model = XGBClassifier()
  33. booster = xgb.Booster()
  34. booster.load_model('./data/ad_xgb.model')
  35. model._Booster = booster
  36. @app.route('/healthcheck')
  37. def health_check():
  38. return 'ok!'
  39. # 首页推荐及tab分类
  40. @app.route('/applet/video/homepage/recommend', methods=['GET', 'POST'])
  41. def homepage_recommend():
  42. start_time = time.time()
  43. # in_homepage = start_time * 1000 + random.randint(0, 100)
  44. # log_.info({'type': 'homepage', 'in_homepage': in_homepage})
  45. try:
  46. # log_.info({'request_headers': request.headers})
  47. request_data = json.loads(request.get_data())
  48. request_id = request_data.get('requestId')
  49. mid = request_data.get('mid')
  50. uid = request_data.get('uid')
  51. category_id = request_data.get('categoryId')
  52. size = request_data.get('size', 4)
  53. app_type = request_data.get('appType', 4)
  54. algo_type = request_data.get('algoType')
  55. client_info = request_data.get('clientInfo')
  56. ab_exp_info = request_data.get('abExpInfo', None)
  57. ab_info_data = request_data.get('abInfoData', None)
  58. version_audit_status = request_data.get('versionAuditStatus', 2) # 小程序版本审核参数:1-审核中,2-审核通过,默认:2
  59. machineinfoBrand = request_data.get('machineinfoBrand', '')
  60. machineinfoModel = request_data.get('machineinfoModel', '')
  61. machineinfoPlatform = request_data.get('machineinfoPlatform', '')
  62. pagesource = request_data.get('pageSource', '')
  63. versioncode = request_data.get('versionCode', 0)
  64. recommendsource = request_data.get('recommendSource', '0')
  65. sencetype = request_data.get('senceType', 0)
  66. recomTraceId=request_data.get('recomTraceId', "")
  67. env_dict = {}
  68. try:
  69. env_dict['app_type'] = int(app_type)
  70. env_dict['pagesource'] = str(pagesource)
  71. env_dict['versioncode'] = int(versioncode)
  72. env_dict['machineinfo_brand'] = str(machineinfoBrand)
  73. env_dict['machineinfo_model'] = str(machineinfoModel)
  74. env_dict['machineinfo_platform'] = str(machineinfoPlatform)
  75. env_dict['recommendsource'] = str(recommendsource)
  76. env_dict['sencetype'] = int(sencetype)
  77. env_dict['recomTraceId'] = recomTraceId
  78. env_dict['recomInterface'] = "homepage"
  79. except:
  80. env_dict['app_type'] = 4
  81. env_dict['pagesource'] = str(pagesource)
  82. env_dict['versioncode'] = 0
  83. env_dict['machineinfo_brand'] = str(machineinfoBrand)
  84. env_dict['machineinfo_model'] = str(machineinfoModel)
  85. env_dict['machineinfo_platform'] = str(machineinfoPlatform)
  86. env_dict['recommendsource'] = str(recommendsource)
  87. env_dict['sencetype'] = sencetype
  88. env_dict['recomTraceId'] = recomTraceId
  89. env_dict['recomInterface'] = "homepage"
  90. log_.error("feature error",env_dict)
  91. params = Params(request_id=request_id)
  92. # size默认为10
  93. if not size:
  94. size = 10
  95. if category_id in config_.CATEGORY['recommend']:
  96. # 推荐
  97. recommend_result = video_homepage_recommend(
  98. request_id=request_id,
  99. mid=mid,
  100. uid=uid,
  101. size=size,
  102. app_type=app_type,
  103. algo_type=algo_type,
  104. client_info=client_info,
  105. ab_exp_info=ab_exp_info,
  106. params=params,
  107. ab_info_data=ab_info_data,
  108. version_audit_status=version_audit_status,
  109. env_dict = env_dict
  110. )
  111. result = {'code': 200, 'message': 'success', 'data': {'videos': recommend_result['videos']}}
  112. log_message = {
  113. 'requestUri': '/applet/video/homepage/recommend',
  114. 'logTimestamp': int(time.time() * 1000),
  115. 'request_id': request_id,
  116. 'app_type': app_type,
  117. 'client_info': client_info,
  118. 'ab_exp_info': ab_exp_info,
  119. 'ab_info_data': ab_info_data,
  120. 'version_audit_status': version_audit_status,
  121. 'category_id': category_id,
  122. 'mid': mid,
  123. 'uid': uid,
  124. 'getRecommendParamsTime': recommend_result.get('getRecommendParamsTime', ''),
  125. 'getRecommendResultTime': recommend_result.get('getRecommendResultTime', ''),
  126. 'updateRedisDataTime': recommend_result.get('updateRedisDataTime', ''),
  127. 'recommendOperation': recommend_result.get('recommendOperation', ''),
  128. 'result': result,
  129. 'executeTime': (time.time() - start_time) * 1000,
  130. 'fea_info': recommend_result.get('fea_info', {})
  131. }
  132. log_.info(log_message)
  133. # log_.info('category_id: {}, mid: {}, uid: {}, result: {}, execute time = {}ms'.format(
  134. # category_id, mid, uid, result, (time.time() - start_time)*1000))
  135. return json.dumps(result)
  136. elif category_id in config_.CATEGORY['other']:
  137. # 其他类别
  138. videos = get_category_videos()
  139. result = {'code': 200, 'message': 'success', 'data': {'videos': videos}}
  140. log_.info('category_id: {}, mid: {}, uid: {}, result: {}, execute time = {}ms'.format(
  141. category_id, mid, uid, result, (time.time() - start_time) * 1000))
  142. return json.dumps(result)
  143. else:
  144. log_.error('categoryId error, categoryId = {}'.format(category_id))
  145. result = {'code': -1, 'message': 'categoryId error'}
  146. return json.dumps(result)
  147. except Exception as e:
  148. log_.error(traceback.format_exc())
  149. result = {'code': -1, 'message': 'fail'}
  150. return json.dumps(result)
  151. # 相关推荐
  152. @app.route('/applet/video/relevant/recommend', methods=['GET', 'POST'])
  153. def relevant_recommend():
  154. start_time = time.time()
  155. # in_relevant = start_time * 1000 + random.randint(0, 100)
  156. # log_.info({"type": "relevant", "in_relevant": in_relevant})
  157. try:
  158. request_data = json.loads(request.get_data())
  159. request_id = request_data.get('requestId')
  160. # log_.info({
  161. # 'logTimestamp': int(time.time() * 1000),
  162. # 'request_id': request_id,
  163. # 'in_relevant': in_relevant,
  164. # 'type': "relevant_recommend",
  165. # 'text': 'in relevant_recommend',
  166. # 'executeTime': (time.time() - start_time) * 1000
  167. # })
  168. mid = request_data.get('mid')
  169. uid = request_data.get('uid')
  170. video_id = request_data.get('videoId')
  171. # up_uid = request_data.get('upUid')
  172. # share_mid = request_data.get('shareMid')
  173. # share_uid = request_data.get('shareUid')
  174. # page_num = request_data.get('pageNum', 1)
  175. page_size = request_data.get('pageSize', 4)
  176. app_type = request_data.get('appType')
  177. client_info = request_data.get('clientInfo')
  178. ab_exp_info = request_data.get('abExpInfo', None)
  179. page_type = request_data.get('pageType') # 1:详情页;2:分享页
  180. ab_info_data = request_data.get('abInfoData', None)
  181. version_audit_status = request_data.get('versionAuditStatus', 2) # 小程序版本审核参数:1-审核中,2-审核通过,默认:2
  182. machineinfoBrand = request_data.get('machineinfoBrand', '')
  183. machineinfoModel = request_data.get('machineinfoModel', '')
  184. machineinfoPlatform = request_data.get('machineinfoPlatform', '')
  185. pagesource = request_data.get('pageSource', '')
  186. versioncode = request_data.get('versionCode', 0)
  187. recommendsource = request_data.get('recommendSource', '0')
  188. sencetype = request_data.get('senceType', 0)
  189. recomTraceId = request_data.get('recomTraceId', "")
  190. env_dict = {}
  191. try:
  192. env_dict['app_type'] = int(app_type)
  193. env_dict['pagesource'] = str(pagesource)
  194. env_dict['versioncode'] = int(versioncode)
  195. env_dict['machineinfo_brand'] = str(machineinfoBrand)
  196. env_dict['machineinfo_model'] = str(machineinfoModel)
  197. env_dict['machineinfo_platform'] = str(machineinfoPlatform)
  198. env_dict['recommendsource'] = str(recommendsource)
  199. env_dict['sencetype'] = int(sencetype)
  200. env_dict['recomInterface'] = "relevant_recommend"
  201. env_dict['recomTraceId'] = recomTraceId
  202. env_dict['relevant_video_id'] = video_id
  203. except:
  204. env_dict['app_type'] = 4
  205. env_dict['pagesource'] = str(pagesource)
  206. env_dict['versioncode'] = 0
  207. env_dict['machineinfo_brand'] = str(machineinfoBrand)
  208. env_dict['machineinfo_model'] = str(machineinfoModel)
  209. env_dict['machineinfo_platform'] = str(machineinfoPlatform)
  210. env_dict['recommendsource'] = str(recommendsource)
  211. env_dict['sencetype'] = sencetype
  212. env_dict['recomInterface'] = "relevant_recommend"
  213. env_dict['relevant_video_id'] = video_id
  214. env_dict['recomTraceId'] = recomTraceId
  215. log_.error("feature error", env_dict)
  216. params = Params(request_id=request_id)
  217. recommend_result = video_relevant_recommend(
  218. request_id=request_id,
  219. video_id=video_id,
  220. mid=mid,
  221. uid=uid,
  222. size=page_size,
  223. app_type=app_type,
  224. ab_exp_info=ab_exp_info,
  225. client_info=client_info,
  226. page_type=page_type,
  227. params=params,
  228. ab_info_data=ab_info_data,
  229. version_audit_status=version_audit_status,
  230. env_dict = env_dict
  231. )
  232. result = {'code': 200, 'message': 'success', 'data': {'videos': recommend_result['videos']}}
  233. log_message = {
  234. 'requestUri': '/applet/video/relevant/recommend',
  235. 'logTimestamp': int(time.time() * 1000),
  236. 'request_id': request_id,
  237. 'app_type': app_type,
  238. 'client_info': client_info,
  239. 'ab_exp_info': ab_exp_info,
  240. 'ab_info_data': ab_info_data,
  241. 'version_audit_status': version_audit_status,
  242. 'mid': mid,
  243. 'uid': uid,
  244. 'getRecommendParamsTime': recommend_result.get('getRecommendParamsTime', ''),
  245. 'getRecommendResultTime': recommend_result.get('getRecommendResultTime', ''),
  246. 'updateRedisDataTime': recommend_result.get('updateRedisDataTime', ''),
  247. 'recommendOperation': recommend_result.get('recommendOperation', ''),
  248. 'result': result,
  249. 'executeTime': (time.time() - start_time) * 1000,
  250. 'fea_info': recommend_result.get('fea_info', {})
  251. }
  252. log_.info(log_message)
  253. # log_.info('app_type: {}, mid: {}, uid: {}, relevant-result: {}, execute time = {}ms'.format(
  254. # app_type, mid, uid, result, (time.time() - start_time) * 1000))
  255. return json.dumps(result)
  256. except Exception as e:
  257. log_.error(traceback.format_exc())
  258. result = {'code': -1, 'message': 'fail'}
  259. return json.dumps(result)
  260. # 管理后台实时修改rov
  261. @app.route('/applet/video/update/rov', methods=['GET', 'POST'])
  262. def update_rov():
  263. try:
  264. # log_.info({'requestUri': '/applet/video/update/rov', 'request_initial': request.get_data()})
  265. request_data = json.loads(request.get_data())
  266. log_.info({'requestUri': '/applet/video/update/rov',
  267. 'logTimestamp': int(time.time() * 1000),
  268. 'requestData': request_data})
  269. # log_.info('update_rov request data: {}'.format(request_data))
  270. # app_type = request_data.get('appType')
  271. video_id = request_data.get('videoId')
  272. rov_score = request_data.get('rovScore')
  273. redis_helper = RedisHelper()
  274. # 将修改ROV值视频的 videoId 和 rovScore 存入对应的redis中
  275. redis_helper.update_score_with_value(key_name=config_.UPDATE_ROV_KEY_NAME, value=video_id, score=rov_score)
  276. redis_helper.update_score_with_value(key_name=config_.UPDATE_ROV_KEY_NAME_APP, value=video_id, score=rov_score)
  277. # ###### 下线 横屏实验
  278. # # 判断该视频是否为 横屏视频,如果是则 存入rov召回池横屏视频 redis 中
  279. # update_video_w_h_rate(video_id=int(video_id), key_name=config_.W_H_RATE_UP_1_VIDEO_LIST_KEY_NAME['rov_recall'])
  280. result = {'code': 200, 'message': 'update rov success'}
  281. log_.info({'requestUri': '/applet/video/update/rov', 'logTimestamp': int(time.time() * 1000), 'result': result})
  282. # log_.info('result: {}'.format(result))
  283. return json.dumps(result)
  284. except Exception as e:
  285. log_.error(traceback.format_exc())
  286. result = {'code': -1, 'message': 'update rov fail'}
  287. return json.dumps(result)
  288. # 管理后台指定用户恢复成新用户
  289. @app.route('/applet/user/to_new', methods=['GET', 'POST'])
  290. def user_to_new():
  291. try:
  292. request_data = json.loads(request.get_data())
  293. log_.info({'requestUri': '/applet/user/to_new', 'requestData': request_data})
  294. app_type = request_data.get('appType', None)
  295. mid = request_data.get('mid')
  296. uid = request_data.get('uid')
  297. user2new(app_type=app_type, mid=mid, uid=uid)
  298. result = {'code': 200, 'message': 'success'}
  299. log_.info({'requestUri': '/applet/user/to_new', 'result': result})
  300. return json.dumps(result)
  301. except Exception as e:
  302. log_.error(traceback.format_exc())
  303. result = {'code': -1, 'message': 'fail'}
  304. return json.dumps(result)
  305. # 管理后台算法视频列表可视化 - 视频数据表类型获取
  306. @app.route('/applet/video/get_video_type_list', methods=['GET', 'POST'])
  307. def get_video_type_list():
  308. try:
  309. data = [
  310. {'dataListDesc': val.get('dataListDesc'), 'dataListCode': val.get('dataListCode')}
  311. for key, val in config_.VIDEO_DATA_LIST_MAPPING.items()
  312. ]
  313. data.sort(key=lambda x: x['dataListCode'], reverse=False)
  314. result = {'code': 200, 'message': 'success', 'data': data}
  315. return json.dumps(result)
  316. except Exception as e:
  317. log_.error(traceback.format_exc())
  318. result = {'code': -1, 'message': 'fail'}
  319. return json.dumps(result)
  320. # 管理后台算法视频列表可视化 - 获取视频列表
  321. @app.route('/applet/video/get_online_list', methods=['GET', 'POST'])
  322. def get_video_online_list():
  323. try:
  324. request_data = json.loads(request.get_data())
  325. ab_exp_code = request_data.get('abExpCode', None)
  326. search_time = request_data.get('searchTime', None)
  327. data_list_type = request_data.get('dataListType', None)
  328. region_code = request_data.get('regionCode', None)
  329. video_id = request_data.get('videoId', None)
  330. page_num = request_data.get('pageNum', 1)
  331. page_size = request_data.get('pageSize', 100)
  332. if video_id is None:
  333. result = get_video_list(ab_exp_code=ab_exp_code, search_time=search_time, data_list_type=data_list_type,
  334. region_code=region_code, page_num=page_num, page_size=page_size)
  335. else:
  336. result = search_video(ab_exp_code=ab_exp_code, search_time=search_time, data_list_type=data_list_type,
  337. region_code=region_code, video_id=video_id, page_num=page_num, page_size=page_size)
  338. return json.dumps(result)
  339. except Exception as e:
  340. log_.error(traceback.format_exc())
  341. result = {'code': -1, 'message': 'fail'}
  342. return json.dumps(result)
  343. # 广告推荐
  344. @app.route('/applet/ad/predict', methods=['GET', 'POST'])
  345. def ad_predict():
  346. start_time = time.time()
  347. try:
  348. request_data = json.loads(request.get_data())
  349. mid = request_data.get('mid')
  350. video_id = request_data.get('videoId')
  351. app_type = request_data.get('appType')
  352. ab_exp_info = request_data.get('abExpInfo')
  353. ab_test_code = request_data.get('abTestCode')
  354. care_model_status = request_data.get('careModelStatus', 1) # 用户关怀模式状态 1: 未开启,2: 开启, 默认: 1
  355. predict_result = ad_recommend_predict(model=model,
  356. app_type=app_type,
  357. mid=mid,
  358. video_id=video_id,
  359. ab_exp_info=ab_exp_info,
  360. ab_test_code=ab_test_code,
  361. care_model_status=care_model_status)
  362. if predict_result is None:
  363. result = {'code': -1, 'message': 'fail'}
  364. else:
  365. result = {'code': 200, 'message': 'success', 'data': predict_result.get('ad_predict')}
  366. log_message = {
  367. 'requestUri': '/applet/ad/predict',
  368. 'request_data': request_data,
  369. 'logTimestamp': int(time.time() * 1000),
  370. 'app_type': app_type,
  371. 'mid': mid,
  372. 'video_id': video_id,
  373. 'predict_result': predict_result,
  374. 'result': result,
  375. 'executeTime': (time.time() - start_time) * 1000
  376. }
  377. log_.info(log_message)
  378. return json.dumps(result)
  379. except Exception as e:
  380. log_.error(traceback.format_exc())
  381. result = {'code': -1, 'message': 'fail'}
  382. return json.dumps(result)
  383. # app热榜
  384. @app.route('/app/video/hot_list', methods=['GET', 'POST'])
  385. def app_video_hot_list():
  386. try:
  387. page_size = 10
  388. request_data = request.get_data()
  389. request_data = json.loads(request_data)
  390. page = request_data.get('page', 0)
  391. log_.info({'requestUri': '/app/video/hot_lis', 'requestData': request_data})
  392. # log_.info('app_video_hot_list request data: {}'.format(request_data))
  393. redis_helper = RedisHelper()
  394. datas = redis_helper.get_data_from_redis('app_video_hot_list')
  395. if datas is None or len(datas) == 0:
  396. result = {'code': -1, 'message': 'no data'}
  397. log_.info({'requestUri': '/app/video/hot_lis', 'result': result})
  398. # log_.info('result: {}'.format(result))
  399. return json.dumps(result)
  400. datas = ast.literal_eval(datas)
  401. total_page = int(len(datas)/page_size)
  402. if len(datas)%page_size > 0:
  403. total_page += 1
  404. if page > total_page -1 :
  405. result = {'code': -1, 'message': 'page exceed max'}
  406. log_.info({'requestUri': '/app/video/hot_lis', 'result': result})
  407. # log_.info('result: {}'.format(result))
  408. return json.dumps(result)
  409. result = {'code': 200, 'message': '', 'data': {'total_page': total_page,
  410. 'hot_list': datas[page*page_size:page*page_size+page_size]}}
  411. log_.info({'requestUri': '/app/video/hot_lis', 'result': result})
  412. # log_.info('result: {}'.format(result))
  413. return json.dumps(result)
  414. except Exception as e:
  415. log_.error(e)
  416. # print(traceback.format_exc())
  417. result = {'code': -1, 'message': 'fail'}
  418. return json.dumps(result)
  419. def serve_forever(ip='0.0.0.0', port=5001):
  420. pywsgi.WSGIServer((ip, port), app).serve_forever()
  421. def apprun(MULTI_PROCESS=True, ip='0.0.0.0', port=5001):
  422. if MULTI_PROCESS == False:
  423. WSGIServer((ip, port), app).serve_forever()
  424. else:
  425. # mulserver = WSGIServer((ip, port), app, handler_class=WebSocketHandler)
  426. mulserver = WSGIServer((ip, port), app)
  427. mulserver.start()
  428. def server_forever():
  429. mulserver.start_accepting()
  430. mulserver._stop_event.wait()
  431. #for i in range(cpu_count()):
  432. for i in range(20):
  433. p = Process(target=server_forever)
  434. p.start()
  435. if __name__ == '__main__':
  436. app.run()
  437. #server = pywsgi.WSGIServer(('0.0.0.0', 5000), app)
  438. #server.serve_forever()
  439. # apprun()