#!/usr/bin/env python # -*- coding: utf-8 -*- import os import time import ffmpy import hashlib import json import requests import random import datetime from django.conf import settings from django.db.models import F from celery import shared_task # from celery_once import QueueOnce from urllib.parse import urljoin from gm_types.push import PUSH_INFO_TYPE, AUTOMATED_PUSH from gm_types.mimas import TRACTATE_PLATFORM, TRACTATE_DATA_TYPE, TRACTATE_CONTENT_LEVEL, APPLET_PAGE_FROM, \ APPLET_SUBSCRIBE_MSG_TYPE, TRACATE_VIDEO_URL_SOURCE, PGC_TYPE, TRACTATE_STATUS from gm_types.gaia import ( VIDEO_CODE_STATUS, QINIU_VIDEO_HANDLE_STATUS_CODE, LIST_TRACTATE_FROM, ) from gm_upload import ( set_video_watermark, fetch_picture_and_save_to_qiniu, video_clipping, upload_file, video_delete, ) from gm_upload.utils.qiniu_tool import QiniuTool from gm_upload.utils.image_utils import Picture from talos.libs.image_utils import fetch_picture_and_save_to_qiniu_v2 from talos.logger import info_logger from talos.models.soft_article.soft_article import SoftArticleVideo from talos.models.soft_article.reply import SoftArticleReply from talos.models.tractate.vote import TractateReplyVote from talos.models.tractate.tractate import TractateTag from talos.models.tractate import TractateVideo, TractateExtra, TractateVote, Tractate, TractateReply, \ TractateReplyImages from talos.rpc import logging_exception from utils.common import get_new_video_name, replace_video_url_for_rich_text from utils.push import push_task_to_user_multi, vote_push, send_applet_subscribe_msg from utils.protocol import gm_protocol from talos.cache.base import ( tractate_pv_cache, tractate_vote_count_cache, vote_cache, pgc_tractate_cache ) from talos.services import UserService from talos.tasks import get_sleep_user from talos.tools.vote_tool import VoteTool from talos.cache.base import fake_vote_cache from communal.normal_manager import ( tag_manager, ) from talos.rpc import get_current_rpc_invoker from talos.models.tractate import TractateScore from communal.tasks import intelligent_push_task def fake_view_num(start, end): """ 帖子的浏览量灌水 :param start: :param end: :return: """ return random.randint(start, end) def upload_new_image_and_get_base_info(video_url): result = { "video_cover_url": "", "width": 0, "height": 0, } _video_cover_url = "{url}{params}".format( url=urljoin(settings.VIDEO_HOST, video_url), params=settings.VIDEO_PIC_URL) try: new_video_caver_url = fetch_picture_and_save_to_qiniu_v2(_video_cover_url) _base_data = Picture.get_image_base_info(new_video_caver_url) if not all(_base_data.values()): _base_data = { "width": 0, "height": 0, } result["video_cover_url"] = new_video_caver_url result.update(_base_data) return result except: # 上传失败则返回默认值 return result @shared_task def set_tractate_rich_text_water_video_url(tractate_id, raw_video_url, water_video_url): try: tractate_obj = Tractate.objects.get(id=tractate_id) except: tractate_obj = None if not tractate_obj: return rich_text, _ = replace_video_url_for_rich_text(tractate_obj.content, {raw_video_url: water_video_url}) tractate_obj.content = rich_text tractate_obj.save(update_fields=["content"]) @shared_task def set_tractate_video_water_mark_url(video_id): video = TractateVideo.objects.get(id=video_id) raw_video_url = video.raw_video_url pid = set_video_watermark( raw_video_url, get_new_video_name(raw_video_url), water_mark_url=settings.WATER_MARK_URL_FOR_VIDEO ) video.persistent_id = pid video.persistent_status = VIDEO_CODE_STATUS.WAITING video.save(update_fields=["persistent_id", "persistent_status"]) @shared_task def check_tractate_video_water_mark_url_is_finish(): videos = TractateVideo.objects.filter(persistent_status=VIDEO_CODE_STATUS.WAITING) for video in videos: _pid = video.persistent_id try: if _pid: result = json.loads(requests.get('http://api.qiniu.com/status/get/prefop?id=' + _pid).text) # 状态码0成功，1等待处理，2正在处理，3处理失败，4通知提交失败。 # 如果请求失败，返回包含如下内容的JSON字符串{"error": ""} code = result.get('code') error = result.get('error', '') if error: continue if code != QINIU_VIDEO_HANDLE_STATUS_CODE.SUCCESS: continue else: if result['items'][0]['code'] == QINIU_VIDEO_HANDLE_STATUS_CODE.SUCCESS: water_video_url = result['items'][0]['key'] # 刷新cdn缓存 QiniuTool.refresh_qiniu_resource_cache([urljoin(settings.VIDEO_HOST, water_video_url)]) # 上传一下有水印的视频第一帧图片 others = upload_new_image_and_get_base_info(water_video_url) if others.get("video_cover_url", ""): for k, v in others.items(): setattr(video, k, v) video.water_video_url = water_video_url video.persistent_status = VIDEO_CODE_STATUS.SUCCESS video.save() # 如果视频来源于帖子富文本中，则更新帖子富文本中的视频地址 if video.video_url_source == TRACATE_VIDEO_URL_SOURCE.RICH_TEXT: set_tractate_rich_text_water_video_url.delay(video.tractate_id, video.raw_video_url, water_video_url) else: continue except: logging_exception() @shared_task def tractate_fake_vote( repeat_times, tractate_id, need_view_increase=True, incr_range=[1, 1], alert='', force_push=False ): """ 异步增加虚拟点赞量、浏览量 :param repeat_times: :param tractate_id: :param incr_range: :param need_view_increase: :param alert: :return: """ now_date = datetime.datetime.now() if not 9 <= now_date.hour < 22: # 超过晚十点灌水任务不执行 return tractate_fake_vote_key = "tractate_fake_vote_{id}".format(id=tractate_id) is_set = fake_vote_cache.set(tractate_fake_vote_key, 1, nx=True, ex=30) if not is_set: return tractate = Tractate.objects.get(id=tractate_id) if tractate.user_del or not tractate.is_online or tractate.content_level == TRACTATE_CONTENT_LEVEL.BAD: info_logger.info('帖子:{}已下线, 不执行点赞灌水.'.format(tractate.id)) return users = get_sleep_user(repeat_times) cnt = 0 for u in users: u_nick_name = '' tractate_vote, created = TractateVote.objects.get_or_create(user_id=u, tractate_id=tractate.id, is_fake=True) if created: # 对作者增加点赞数 author = UserService.get_user_by_user_id(tractate.user_id) author.incr_vote_count() cnt += 1 if not u_nick_name: vote_user = UserService.get_user_by_user_id(u) u_nick_name = vote_user.nickname or u'更美用户' vote_tool = VoteTool(redis_c=vote_cache, user_id=tractate.user_id, new_version=True) vote_tool.receive_tractate_vote(tractate_vote.id) # 点赞推送 push_url = gm_protocol.get_tractate_detail( tractate_id=tractate.id, tractate_detail_from=LIST_TRACTATE_FROM.NOTICE_VOTE ) if not alert: alert = u'{user_name}赞了你的帖子{content}'.format( user_name=str(u_nick_name), content=tractate.content[:10]) if force_push: vote_push(user_id=tractate.user_id, alert=alert, push_type=AUTOMATED_PUSH.TRACTATE_GET_VOTED, push_url=push_url) if not need_view_increase: return # 增加可视点赞数 te, _ = TractateExtra.objects.get_or_create(tractate_id=tractate.id) te.favor_count = F('vote_count') + cnt te.save(update_fields=['vote_count']) tractate_vote_count_cache.incrby(str(tractate.id), cnt) # 每个点赞增加随机浏览量 view_num = 0 for i in range(cnt): view_num += random.randint(*incr_range) # 增加可视浏览数 tractate_pv_cache.incrby(str(tractate.id), view_num) @shared_task def reply_push(user_id, tractate_id, reply_id, content=None): """ :param user_id: 回复人的user_id :param tractate_id: :param reply_id: 发出的那条回复ID :param content: 评论的内容 :return: """ if not tractate_id and not reply_id: return try: reply_info = TractateReply.objects.get(id=reply_id) except TractateReply.DoesNotExist: return if reply_info.replied_id: try: replied_info = TractateReply.objects.get(id=reply_info.replied_id) except TractateReply.DoesNotExist: return push_user_id = replied_info.user_id push_type = AUTOMATED_PUSH.TRACTATE_REPLY_GET_REPLY content_id = replied_info.id # 二级评论传给demeter的id是二级评论id push_image = '' # 帖子评论暂时没有图片 else: try: tractate = Tractate.objects.get(id=tractate_id) except Tractate.DoesNotExist: return push_user_id = tractate.user_id images = list(TractateReplyImages.objects.using(settings.SLAVE_DB_NAME).filter( reply_id=reply_id, ).values_list('image_url', flat=True)) push_image = images and Picture(images[0]).watermarked or '' content_id = tractate.id push_type = AUTOMATED_PUSH.TRACTATE_GET_REPLY push_url = gm_protocol.get_user_answer_list() user = UserService.get_user_by_user_id(user_id=user_id) alert = u'@{}：{}...'.format(user.nickname if user else '更美用户', content[:25]) intelligent_push_task.apply_async( args=( content_id, [push_user_id], push_type, { 'type': PUSH_INFO_TYPE.GM_PROTOCOL, 'pushUrl': push_url, 'push_url': push_url, 'image': push_image, 'push_image': push_image, } ), kwargs={ "platform": None, "alert": alert, "others": { "title": "你收到了一条新评论", "alert": alert, }, "labels": { 'event_type': 'push', 'event': 'tractate_received_reply' }, }, ) @shared_task def set_softarticle_video_water_mark_url(video_id): # 医生后台帖子封面图修改 video = SoftArticleVideo.objects.get(id=video_id) raw_video_url = video.raw_video_url pid = set_video_watermark( raw_video_url, get_new_video_name(raw_video_url), water_mark_url=settings.WATER_MARK_URL_FOR_VIDEO ) video.persistent_id = pid video.persistent_status = VIDEO_CODE_STATUS.WAITING video.save(update_fields=["persistent_id", "persistent_status"]) @shared_task def check_soft_article_video_water_mark_url_is_finish(): videos = SoftArticleVideo.objects.filter(persistent_status=VIDEO_CODE_STATUS.WAITING) for video in videos: _pid = video.persistent_id try: if _pid: result = json.loads(requests.get('http://api.qiniu.com/status/get/prefop?id=' + _pid).text) # 状态码0成功，1等待处理，2正在处理，3处理失败，4通知提交失败。 # 如果请求失败，返回包含如下内容的JSON字符串{"error": ""} code = result.get('code') error = result.get('error', '') if error: continue if code != QINIU_VIDEO_HANDLE_STATUS_CODE.SUCCESS: continue else: if result['items'][0]['code'] == QINIU_VIDEO_HANDLE_STATUS_CODE.SUCCESS: water_video_url = result['items'][0]['key'] # 刷新cdn缓存 QiniuTool.refresh_qiniu_resource_cache([urljoin(settings.VIDEO_HOST, water_video_url)]) # 上传一下有水印的视频第一帧图片 others = upload_new_image_and_get_base_info(water_video_url) if others.get("video_cover_url", ""): for k, v in others.items(): setattr(video, k, v) video.water_video_url = water_video_url video.persistent_status = VIDEO_CODE_STATUS.SUCCESS video.save() else: continue except: logging_exception() @shared_task def doctor_reply_push(soft_article_id, reply_id): """ :param soft_article_id: :param reply_id: 发出的那条回复ID :return: """ if not soft_article_id and not reply_id: return try: reply_info = SoftArticleReply.objects.get(id=reply_id) except SoftArticleReply.DoesNotExist: return if reply_info.replied_id: try: replied_info = SoftArticleReply.objects.get(id=reply_info.replied_id, source_id=TRACTATE_PLATFORM.GM) except SoftArticleReply.DoesNotExist: return user = UserService.get_user_by_user_id(user_id=reply_info.user_id) if not user: return push_user_id = replied_info.user_id content = reply_info.content if len(reply_info.content) >= 10 else reply_info.content[:10] push_msg = "{user_name}回复了你的回复{content}".format(user_name=user.nickname, content=content) push_type = AUTOMATED_PUSH.TRACTATE_REPLY_GET_REPLY else: return push_url = gm_protocol.get_tractate_detail( comment_id=reply_id, tractate_id=replied_info.softarticle_id, data_type=TRACTATE_DATA_TYPE.DOCTOR, tractate_detail_from=LIST_TRACTATE_FROM.NOTICE_REPLY ) if reply_id else gm_protocol.get_comment_detail(tractate_id=replied_info.softarticle_id) extra = { 'type': PUSH_INFO_TYPE.GM_PROTOCOL, 'msgType': 4, 'pushUrl': push_url, 'push_url': push_url, } push_task_to_user_multi( user_ids=[push_user_id], push_type=push_type, extra=extra, labels={'event_type': 'push', 'event': 'question_received_answer'}, alert=push_msg, ) @shared_task def tractate_video_fsm_runner(): """ 状态机 考虑封装成类，根据状态给对应worker添加任务 :return: """ # 截取视频前两秒 clipping_id_list = list(TractateVideo.objects.filter( persistent_clip_status=VIDEO_CODE_STATUS.NOSTART ).values_list("id", flat=True)) for video_id in clipping_id_list: set_tractate_video_clipping_to_video.delay(video_id) # 检查七牛云截取状态 check_id_list = list(TractateVideo.objects.filter( persistent_clip_status=VIDEO_CODE_STATUS.WAITING ).values_list("id", flat=True)) for video_id in check_id_list: check_tractate_video_clipping_is_finish.delay(video_id) # 将七牛云截取成功的 mp4 转换为 webP 动图并上传 set_id_list = list(TractateVideo.objects.filter( persistent_clip_status=VIDEO_CODE_STATUS.OPERATING_LOCAL ).values_list("id", flat=True)) for video_id in set_id_list: set_tractate_video_webp_pic.delay(video_id) @shared_task def set_tractate_video_clipping_to_video(video_id): video = TractateVideo.objects.get(id=video_id) if video.persistent_clip_status != VIDEO_CODE_STATUS.NOSTART: return hash_key = hashlib.md5(str(time.time()).encode("utf8")).hexdigest() pid = video_clipping( video.raw_video_url, new_filename='{}_clip.mp4'.format(hash_key), video_type='mp4', water_mark_url=None, start_time=3, # water_mark_url=settings.WATER_MARK_URL_FOR_VIDEO ) video.persistent_clip_status = VIDEO_CODE_STATUS.WAITING video.persistent_clip_id = pid video.save(update_fields=['persistent_clip_id', 'persistent_clip_status']) @shared_task def check_tractate_video_clipping_is_finish(video_id): try: video = TractateVideo.objects.get(id=video_id) if video.persistent_clip_status != VIDEO_CODE_STATUS.WAITING: return if video.persistent_clip_id: result = json.loads(requests.get('{}{}'.format( settings.QINIU_VIDEO_INQUIRE_HOST, video.persistent_clip_id )).text) # 状态码0成功，1等待处理，2正在处理，3处理失败，4通知提交失败。 # 如果请求失败，返回包含如下内容的JSON字符串{"error": ""} code = result.get('code') error = result.get('error', '') if error: return if code not in ( QINIU_VIDEO_HANDLE_STATUS_CODE.SUCCESS, QINIU_VIDEO_HANDLE_STATUS_CODE.PROCESSING_FAIL ): return elif code == QINIU_VIDEO_HANDLE_STATUS_CODE.PROCESSING_FAIL: video.persistent_clip_status = VIDEO_CODE_STATUS.FAIL video.save(update_fields=['persistent_clip_status']) else: if result['items'][0]['code'] == QINIU_VIDEO_HANDLE_STATUS_CODE.SUCCESS: mp4_key = result['items'][0]['key'] # 刷新cdn缓存 QiniuTool.refresh_qiniu_resource_cache([urljoin(settings.VIDEO_HOST, mp4_key)]) video.intercept_video_url = mp4_key video.persistent_clip_status = VIDEO_CODE_STATUS.OPERATING_LOCAL video.save(update_fields=['persistent_clip_status', 'intercept_video_url']) # set_tractate_video_webp_pic(video.id) else: video.persistent_status = result['items'][0]['code'] video.save(update_fields=['persistent_clip_status']) except: logging_exception() @shared_task def set_tractate_video_webp_pic(video_id): video = TractateVideo.objects.get(id=video_id) if video.persistent_clip_status != VIDEO_CODE_STATUS.OPERATING_LOCAL: return mp4_key = video.intercept_video_url hash_key = hashlib.md5(str(time.time()).encode("utf8")).hexdigest() input_path = os.path.join(settings.VIDEO_CONVERT_PATH, "{}.mp4".format(hash_key)) output_path = os.path.join(settings.VIDEO_CONVERT_PATH, '{}.webp'.format(hash_key)) try: clipping_video_url = settings.VIDEO_HOST + mp4_key res = requests.get(clipping_video_url) # 文件存储到本地 with open(input_path, 'wb') as f: f.write(res.content) # 使用底层 FFmpeg 库进行转码 ff = ffmpy.FFmpeg( inputs={input_path: None}, outputs={output_path: settings.VIDEO_FF_ARGS} ) ff.run() # 上传webP图片 video.webp_url = upload_file(output_path) video.persistent_clip_status = VIDEO_CODE_STATUS.SUCCESS video.save(update_fields=['webp_url', 'persistent_clip_status']) # 删除中间数据　 video_delete(mp4_key) # 刷新缓存 QiniuTool.refresh_qiniu_resource_cache( [urljoin(settings.VIDEO_HOST, mp4_key)] ) except: logging_exception() video.persistent_clip_status = VIDEO_CODE_STATUS.FAIL video.save(update_fields=['persistent_clip_status']) finally: if os.path.exists(input_path): os.remove(input_path) if os.path.exists(output_path): os.remove(output_path) @shared_task def applet_replied_push(reply_id): """ 帖子的评论被评论，给帖子评论的用户发送小程序推送 自己给自己评论无效 :param reply_id: 发出的那条回复ID :return: """ if not reply_id: return try: reply_info = TractateReply.objects.get(id=reply_id) except TractateReply.DoesNotExist: return if not reply_info.replied_id: return try: replied_info = TractateReply.objects.get(id=reply_info.replied_id) except TractateReply.DoesNotExist: return reply_content = reply_info.content[:20] nickname = UserService.get_user_by_user_id(reply_info.user_id).nickname[:10] data = { "name1": { "value": nickname }, "thing2": { "value": reply_content }, "date3": { "value": "{date}".format(date=datetime.datetime.now().strftime('%Y年%m月%d日 %H:%M')) }, "thing4": { "value": "点击快速查看>>" } } # 模板id template_id = settings.APPLET_SUBSCRIBE_MSG.get("comment", "") # 跳转页面 page = '/packageBBS/pages/posts/main?tractate_id={tractate_id}&top_id={top_id}&reply_id=' \ '{reply_id}&from={from_page}&from_action={from_action}'.format( tractate_id=reply_info.tractate_id, top_id=reply_info.top_id, reply_id=reply_info.id, from_page=APPLET_PAGE_FROM.CARD, from_action=APPLET_SUBSCRIBE_MSG_TYPE.COMMENT ) # 给一级评论用户发 if reply_info.replied_id != reply_info.top_id: try: level_1_reply = TractateReply.objects.get(id=reply_info.top_id) except: level_1_reply = None if level_1_reply: # 自己给自己评论无效 if reply_info.user_id != level_1_reply.user_id: # 发送小程序推送 send_applet_subscribe_msg(level_1_reply.user_id, template_id, data=data, page=page) # 给被评论用户发，自己给自己评论无效 if reply_info.user_id == replied_info.user_id: return # 发送小程序推送 send_applet_subscribe_msg(replied_info.user_id, template_id, data=data, page=page) @shared_task def applet_reply_summary_push(reply_id): """ 用户评论完24小时后， 如果没有收到评论或点赞，则给用户推送帖子新增评论总数，或帖子的相关内容 如果被赞，且被赞数大于1，则推送新增被赞数 :param reply_id: 当前评论id :return: """ if not reply_id: return try: reply_info = TractateReply.objects.get(id=reply_id) except TractateReply.DoesNotExist: return user_id = reply_info.user_id # 查询当前用户有没有新的对帖子的评论 new_tractate_reply = TractateReply.objects.using(settings.SLAVE_DB_NAME).filter(id__gt=reply_id, user_id=user_id, is_online=True).exists() if new_tractate_reply: return replied_count = TractateReply.objects.using(settings.SLAVE_DB_NAME).\ filter(replied_id=reply_id, is_online=True).exclude(user_id=user_id).count() voted_count = TractateReplyVote.objects.using(settings.SLAVE_DB_NAME).\ filter(tractate_id=reply_info.tractate_id, reply_id=reply_id, is_online=True).exclude(user_id=user_id).count() tractate_id = reply_info.tractate_id tractate_content = Tractate.objects.filter(id=tractate_id).first().content[:10] # 当前评论有被评论或被赞 if replied_count or voted_count: return # 帖子新增的一级评论数（不包含自己的） additional_reply_count = TractateReply.objects.using(settings.SLAVE_DB_NAME).\ filter(id__gt=reply_id, tractate_id=tractate_id, top_id=0, is_online=True).exclude(user_id=user_id).count() # 有新增评论，发送24小内新增评论总数 if additional_reply_count: data = { "thing1": { "value": tractate_content }, "thing2": { "value": "你评论的帖子，新增{reply_count}条吐槽，立即查看".format(reply_count=additional_reply_count)[:20] } } # 模板id template_id = settings.APPLET_SUBSCRIBE_MSG.get("vote", "") # 跳转页面 page = '/packageBBS/pages/posts/main?tractate_id={tractate_id}&data_type=user_post&from={from_page}' \ '&from_action={from_action}'\ .format( tractate_id=reply_info.tractate_id, from_page=APPLET_PAGE_FROM.CARD, from_action=APPLET_SUBSCRIBE_MSG_TYPE.NEW_COMMENT ) # 无新增评论，推送评论的帖子的相同标签下的其它内容详情页 else: data = { "thing2": { "value": tractate_content }, "thing4": { "value": "亲！你关注过的话题又有新内容啦>>" } } # 模板id template_id = settings.APPLET_SUBSCRIBE_MSG.get("recommend", "") # 获取帖子标签 tag_id_list = TractateTag.objects.filter(tractate_id=tractate_id).values_list('tag_id', flat=True) tag_info = tag_manager.get_tags_info_by_ids(tag_id_list) tag_info_list = list(tag_info.values()) tag_name = tag_info_list and tag_info_list[0] and tag_info_list[0].get("tag_name") if not tag_name: return # 调策略 获取相关内容 def get_new_tractate_id(rpc_client, offset): res = rpc_client['doris/search/query_filter_tractate'](query=tag_name, size=1, offset=offset).unwrap() new_tractate_id = res and res.get("tractate_id_list") and res.get("tractate_id_list")[0] return new_tractate_id rpc_client = get_current_rpc_invoker() offset = random.randint(0, 200) origin_offset = offset new_tractate_id = None num = 6 while not new_tractate_id and num : try: new_tractate_id = get_new_tractate_id(rpc_client, offset) except: new_tractate_id = None offset = origin_offset % num if num > 1: offset += random.randint(0, 10) num -= 1 if not new_tractate_id: return # 跳转页面 page = '/packageBBS/pages/posts/main?tractate_id={tractate_id}&data_type=user_post&from={from_page}' \ '&from_action={from_action}' \ .format( tractate_id=new_tractate_id, from_page=APPLET_PAGE_FROM.CARD, from_action=APPLET_SUBSCRIBE_MSG_TYPE.RELATED_CONTENT ) # 发送小程序推送 send_applet_subscribe_msg(user_id, template_id, data=data, page=page) @shared_task def refresh_pgc_info(): """ 定时更新有pgc属性的帖子状态：1. 更新新加入的帖子；2. 更新 smart_rank 值 直接做好排序，写入redis :return: """ tractate_list = list(Tractate.objects.using(settings.SLAVE_DB_NAME).filter( pgc_type=PGC_TYPE.COMMUNITY, is_online=True, status=TRACTATE_STATUS.AUDIT_SUCCESS ).values_list( 'id', 'create_time', 'pgc_type' )) tractate_info = { i[0]: { 'tractate_id': i[0], 'tractate_create_time': i[1], 'pgc_type': i[2], 'tractate_score': 0, 'is_online': 1 } for i in tractate_list } tractate_ids = [i[0] for i in tractate_list] start_pos, end_pos = 0, 0 offset = 100 smart_rank_info = [] tractate_num = len(tractate_ids) while end_pos < tractate_num: end_pos = start_pos + offset if end_pos > tractate_num: end_pos = tractate_num smart_rank_info.extend(list(TractateScore.objects.filter( tractate_id__in=tractate_ids[start_pos:end_pos] ).values_list('tractate_id', 'tractate_score'))) start_pos = end_pos for item in smart_rank_info: tractate_info[item[0]]['tractate_score'] = item[1] # 先按发布时间倒叙排列，发布时间相同再按smr分数从高到低排列 scored_list = list(tractate_info.values()) scored_list.sort(key=lambda x: (x['tractate_create_time'], x['tractate_score']), reverse=True) sorted_ids = json.dumps([i['tractate_id'] for i in scored_list]) info_logger.info("更新pgc帖子缓存：{}".format(sorted_ids)) pgc_tractate_cache.set('pgc_tractate_sorted_ids', sorted_ids, 86400) pgc_tractate_cache.set('pgc_tractate_update_time', int(time.time()), 86400)