#!/usr/bin/env python
# -*- coding: utf-8 -*-
from __future__ import unicode_literals, absolute_import, print_function
from gm_rpcd.all import bind
import logging
import traceback
import json
from search.utils.topic import TopicUtils
from trans2es.models.topic import TopicHomeRecommend
from libs.es import ESPerform
from libs.cache import redis_client
from search.utils.common import *
from libs.es import ESPerform
from django.conf import settings
def get_discover_page_topic_ids(user_id, device_id, size, query_type=TopicPageType.FIND_PAGE):
try:
if user_id == -1:
redis_key = "physical:discover_page" + ":user_id:" + str(user_id) + ":device_id:" + device_id
else:
redis_key = "physical:discover_page" + ":user_id:" + str(user_id)
redis_field_list = [b'have_read_topic_id']
redis_field_val_list = redis_client.hmget(redis_key, redis_field_list)
have_read_topic_id_list = json.loads(redis_field_val_list[0]) if redis_field_val_list[0] else []
if have_read_topic_id_list == None:
have_read_topic_id_list = list()
recommend_topic_ids = TopicUtils.get_recommend_topic_ids(user_id=user_id, tag_id=0, offset=0, size=size,single_size=size,
query_type=query_type,
filter_topic_id_list=have_read_topic_id_list,index_type="topic",routing="4,5,6")
have_read_topic_id_list.extend(recommend_topic_ids)
redis_dict = {
"have_read_topic_id": json.dumps(have_read_topic_id_list)
}
redis_client.hmset(redis_key, redis_dict)
return recommend_topic_ids
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return []
def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query=None,
query_type=TopicPageType.HOME_RECOMMEND,promote_topic_list = [],disable_collpase=False,usefulrecall = -1,useful_tag_list =[]):
try:
topic_star_routing = "6"
index_type = "topic-high-star"
device_redis_key = ""
if query is None:
if user_id>0:
redis_key = "physical:home_recommend" + ":user_id:" + str(user_id) + ":query_type:" + str(query_type)
device_redis_key = "physical:home_recommend" + ":device_id:" + device_id + ":query_type:" + str(query_type)
else:
redis_key = "physical:home_recommend" + ":device_id:" + device_id + ":query_type:" + str(query_type)
else:
topic_star_routing = "3,4,5,6"
index_type = "topic"
if user_id>0:
redis_key = "physical:home_query" + ":user_id:" + str(user_id) + ":query:" + str(query) + ":query_type:" + str(query_type)
device_redis_key = "physical:home_query" + ":device_id:" + device_id + ":query:" + str(query) + ":query_type:" + str(query_type)
else:
redis_key = "physical:home_query" + ":device_id:" + device_id + ":query:" + str(query) + ":query_type:" + str(query_type)
redis_field_list = [b'have_read_topic_list']
redis_field_val_list = redis_client.hmget(redis_key, redis_field_list)
# 获取已读帖子
have_read_topic_id_list = list()
if redis_field_val_list[0]:
if query is None:
have_read_topic_id_list = list(json.loads(redis_field_val_list[0]))
else:
if offset>0: # 首次搜索时不需要过滤已读
have_read_topic_id_list = list(json.loads(redis_field_val_list[0]))
elif len(device_redis_key)>0:
redis_field_val_list = redis_client.hmget(device_redis_key, redis_field_list)
if redis_field_val_list[0]:
if query is None:
have_read_topic_id_list = list(json.loads(redis_field_val_list[0]))
else:
if offset>0: # 首次搜索时不需要过滤已读
have_read_topic_id_list = list(json.loads(redis_field_val_list[0]))
if have_read_topic_id_list == None:
have_read_topic_id_list = list()
have_read_topic_id_list.extend(promote_topic_list)
useful_topic_id_list = list()
recommend_topic_user_list = list()
attention_tag_list = list()
recommend_topic_list = list()
if query is None:
if user_id != -1:
# 有用标签召回
if usefulrecall != -1:
useful_topic_id_list = TopicUtils.userful_tag_topic_list(user_id, have_read_topic_id_list, 4,
"topic-high-star", "6",useful_tag_list=useful_tag_list)
# user_similar_score_redis_key = "physical:user_similar_score:user_id:" + str(user_id)
# redis_user_similar_score_redis_val = redis_client.get(user_similar_score_redis_key)
# user_similar_score_redis_list = json.loads(
# redis_user_similar_score_redis_val) if redis_user_similar_score_redis_val else []
size = size - len(useful_topic_id_list)
have_read_topic_id_list.extend(useful_topic_id_list)
# linucb 推荐帖子
topic_recommend_redis_key = "physical:linucb:topic_recommend:device_id:" + str(device_id)
recommend_topic_dict = redis_client.hgetall(topic_recommend_redis_key)
linucb_recommend_topic_id_list = list()
recommend_topic_list = list()
if b"data" in recommend_topic_dict:
linucb_recommend_topic_id_list = json.loads(recommend_topic_dict[b"data"])
if linucb_recommend_topic_id_list == None:
linucb_recommend_topic_id_list = list()
# 推荐帖子是强插的，要保证推荐帖子不在已读里
logging.warning("type1:%s,type2:%s"%(type(linucb_recommend_topic_id_list),type(have_read_topic_id_list)))
recommend_topic_id_list = list(set(linucb_recommend_topic_id_list) - set(have_read_topic_id_list))
recommend_topic_id_list.sort(key=linucb_recommend_topic_id_list.index)
# cursor = int(str(recommend_topic_dict[b"cursor"], encoding="utf-8"))
# newcursor = cursor + 6
if len(recommend_topic_id_list) > 0:
recommend_topic_list = recommend_topic_id_list[0:size]
# redis_client.hset(topic_recommend_redis_key, "cursor", newcursor)
if b"datadict" in recommend_topic_dict:
linucb_recommend_topic_id_dict = json.loads(recommend_topic_dict[b"datadict"])
if linucb_recommend_topic_id_dict is not None and len(recommend_topic_list) >0:
for i in recommend_topic_list:
recommend_topic_user_list.append(linucb_recommend_topic_id_dict[str(i)])
# if have_read_topic_id_list == None:
# have_read_topic_id_list = list()
# 用户关注标签
redis_tag_data = redis_client.hget("physical:linucb:register_user_tag_info", user_id)
attention_tag_list = json.loads(redis_tag_data) if redis_tag_data else []
if len(recommend_topic_list)>0:
size = size-len(recommend_topic_list)
have_read_topic_id_list.extend(recommend_topic_list)
# have_read_topic_id_list_add_promote = list()
# have_read_topic_id_list_add_promote.extend(have_read_topic_id_list)
# promote_recommend_topic_id_list = TopicHomeRecommend.objects.using(settings.SLAVE_DB_NAME).filter(
# is_online=1).values_list("topic_id",flat=True)
#
# for topic_id in promote_recommend_topic_id_list:
# have_read_topic_id_list_add_promote.append(topic_id)
rank_topic_id_list = list()
if size>0:
rank_topic_id_list = TopicUtils.get_recommend_topic_ids(user_id=user_id, tag_id=tag_id, offset=0, size=size,
single_size=size,query=query, query_type=query_type,
filter_topic_id_list=have_read_topic_id_list,
index_type=index_type,routing=topic_star_routing,attention_tag_list=attention_tag_list,linucb_user_id_list=recommend_topic_user_list,disable_collpase=disable_collpase)
# if len(recommend_topic_list) == 6 and query is None:
# if (size < 11):
# topic_id_list.extend(rank_topic_id_list[0:3])
# topic_id_list.extend(recommend_topic_list[0:3])
# topic_id_list.extend(rank_topic_id_list[3:size])
# topic_id_list.extend(recommend_topic_list[3:6])
# else:
# topic_id_list.extend(rank_topic_id_list[0:size - 7])
# topic_id_list.extend(recommend_topic_list[0:3])
# topic_id_list.extend(rank_topic_id_list[size - 7:size])
# topic_id_list.extend(recommend_topic_list[3:6])
# else:
# topic_id_list.extend(rank_topic_id_list)
have_read_topic_id_list.extend(rank_topic_id_list)
if len(have_read_topic_id_list) > 30000:
cut_len = len(have_read_topic_id_list)-30000
have_read_topic_id_list = have_read_topic_id_list[cut_len:]
redis_dict = {
"have_read_topic_list": json.dumps(have_read_topic_id_list),
}
redis_client.hmset(redis_key, redis_dict)
# 每个session key保存60天
redis_client.expire(redis_key, 60 * 60 * 24 * 60)
if usefulrecall != -1:
return recommend_topic_list,rank_topic_id_list,useful_topic_id_list
else:
return recommend_topic_list, rank_topic_id_list
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
if usefulrecall != -1:
return [],[],[]
else:
return [], []
@bind("physical/search/query_tag_id_by_topic")
def query_tag_id_by_topic(offset=0, size=10, topic_id_list=[], user_id=-1):
try:
return TopicUtils.get_topic_tag_info(offset, size, topic_id_list, user_id)
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return {}
@bind("physical/search/home_recommend")
def home_recommend(device_id="", user_id=-1, offset=0, size=10, query_type=TopicPageType.HOME_RECOMMEND,promote_topic_list=[],usefulrecall = -1,useful_tag_list=[]):
"""
:remark:首页推荐，目前只推荐日记
:param session_id:
:param user_id:
:param offset:
:param size:
:return:
"""
try:
if not user_id:
user_id = -1
if not isinstance(device_id, str):
device_id = ""
recommend_topic_ids = list()
rank_topic_ids = list()
useful_topic_ids = list()
es_node_load_high_flag = False
# try:
# es_node_load_high_flag = ESPerform.if_es_node_load_high(ESPerform.get_cli())
# except:
# logging.error("catch exception,err_msg:%s" % traceback.format_exc())
# es_node_load_high_flag = True
if es_node_load_high_flag:
temp_downgrading_key = "physical:home_recommend:user_id:241407656:query_type:1"
redis_field_list = [b'have_read_topic_list']
redis_field_val_list = redis_client.hmget(temp_downgrading_key, redis_field_list)
if redis_field_val_list[0]:
have_read_topic_id_list = list(json.loads(redis_field_val_list[0]))
if len(have_read_topic_id_list) > offset:
recommend_topic_ids = have_read_topic_id_list[offset:offset+size]
else:
recommend_topic_ids = have_read_topic_id_list[0:size]
else:
if usefulrecall != -1 and len(useful_tag_list)>0:
recommend_topic_ids,rank_topic_ids,useful_topic_ids = get_home_recommend_topic_ids(user_id, device_id, tag_id=0, offset=0, size=size,
query_type=query_type,promote_topic_list=promote_topic_list,usefulrecall=usefulrecall,useful_tag_list=useful_tag_list)
return {"linucb_topic_ids": recommend_topic_ids,"rank_topic_ids":rank_topic_ids,"useful_topic_ids":useful_topic_ids}
else:
recommend_topic_ids,rank_topic_ids = get_home_recommend_topic_ids(user_id, device_id, tag_id=0, offset=0, size=size,
query_type=query_type,promote_topic_list=promote_topic_list)
return {"linucb_topic_ids": recommend_topic_ids,"rank_topic_ids":rank_topic_ids}
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
if usefulrecall != -1:
return {"linucb_topic_ids": [],"rank_topic_ids":[],"useful_topic_ids":[]}
else:
return {"linucb_topic_ids": [], "rank_topic_ids": []}
@bind("physical/search/discover_page")
def discover_page(device_id="", user_id=-1, size=10):
"""
:remark:首页推荐，目前只推荐日记
:param session_id:
:param user_id:
:param offset:
:param size:
:return:
"""
try:
if not user_id:
user_id = -1
if not isinstance(device_id, str):
device_id = ""
recommend_topic_ids = get_discover_page_topic_ids(user_id, device_id, size, query_type=TopicPageType.FIND_PAGE)
return {"recommend_topic_ids": recommend_topic_ids}
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return {"recommend_topic_ids": []}
@bind("physical/search/home_query")
def home_query(device_id="", tag_id=-1, user_id=-1, query="", offset=0, size=10):
"""
:remark:首页搜索，目前只推荐日记
:param session_id:
:param user_id:
:param query:
:param offset:
:param size:
:return:
"""
try:
if not user_id:
user_id = -1
if not isinstance(device_id, str):
device_id = ""
recommend_topic_list, rank_topic_id_list = get_home_recommend_topic_ids(user_id, device_id, tag_id, offset=offset, size=size, query=query)
if len(rank_topic_id_list)>0 and len(rank_topic_id_list)<size:
recommend_topic_list, rank_topic_id_list = get_home_recommend_topic_ids(user_id, device_id, tag_id,
offset=offset, size=size,
query=query,disable_collpase=True)
return {"recommend_topic_ids": rank_topic_id_list}
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return {"recommend_topic_ids": []}
@bind("physical/search/topic_detail_page_recommend")
def topic_detail_page_recommend(device_id="", user_id=-1, topic_id=-1, topic_pictorial_id=-1, topic_user_id=-1,
filter_topic_user_id=False, offset=0, size=10, topic_tag_list=[]):
"""
:remark:帖子详情页推荐策略，缺少第一个卡片策略
:param user_id:
:param topic_id:
:param topic_group_id:
:param topic_user_id:
:return:
"""
try:
if not isinstance(user_id, int):
user_id = -1
if user_id > 0:
redis_key = "physical:topic_detail_recommend" + ":user_id:" + str(user_id) + "topic_id:"+str(topic_id)
else:
redis_key = "physical:topic_detail_recommend" + ":device_id:" + device_id + "topic_id:"+str(topic_id)
if int(offset) == 0:
have_read_topic_list = list()
# redis_dict = {
# "have_read_topic_id": json.dumps(have_read_topic_list)
# }
redis_client.delete(redis_key)
# redis_client.expire(redis_key, 60 * 60 * 24)
else:
have_read_topic_list = list()
redis_field_list = [b'have_read_topic_list']
have_read_topic_redis_data = redis_client.hmget(redis_key,redis_field_list)
have_read_topic_list = json.loads(have_read_topic_redis_data[0]) if have_read_topic_redis_data[0] else []
es_cli_obj = ESPerform.get_cli()
have_read_topic_list.append(topic_id)
topic_user_result = list()
topic_tag_result = list()
result = list()
if len(topic_tag_list) != 0:
topic_tag_result = TopicUtils.top_get_topic_detail_recommend_list(user_id,topic_id,have_read_topic_list,size,es_cli_obj,
index_type="topic",routing="3,4,5,6",topic_tag_list = topic_tag_list)
topic_tag_size = len(topic_tag_result)
have_read_topic_list.extend(topic_tag_result)
else:
topic_tag_size = 0
if topic_tag_size <size:
size = size - topic_tag_size
if topic_user_id != -1:
topic_user_result = TopicUtils.top_get_topic_detail_recommend_list(user_id, topic_id, have_read_topic_list,
size, es_cli_obj,
index_type="topic", routing="3,4,5,6",topic_user_id = topic_user_id
)
topic_user_size = len(topic_user_result)
have_read_topic_list.extend(topic_user_result)
if topic_user_size < size:
size = size - topic_user_size
result = TopicUtils.top_get_topic_detail_recommend_list(user_id, topic_id,
have_read_topic_list,
size, es_cli_obj,
index_type="topic", routing="4,5,6")
have_read_topic_list.extend(result)
# have_read_topic_redis_data = redis_client.get(redis_key)
# have_read_topic_list = json.loads(have_read_topic_redis_data) if have_read_topic_redis_data else []
#
# redis_key = "physical:topic_detail_page_recommend" + ":user_id:" + str(user_id) + ":device_id:" + str(device_id)
# have_read_topic_redis_data = redis_client.get(redis_key)
# have_read_topic_list = json.loads(have_read_topic_redis_data) if have_read_topic_redis_data else []
#
# # 获取es链接对象
# es_cli_obj = ESPerform.get_cli()
#
# # 获取帖子标签列表
# topic_tag_list = TopicUtils.get_topic_tag_id_list(topic_id, es_cli_obj)
# result_list = TopicUtils.get_topic_detail_recommend_list(user_id, topic_id, topic_tag_list, topic_pictorial_id,
# topic_user_id, filter_topic_user_id,
# have_read_topic_list, offset, size, es_cli_obj,index_type="topic",routing="4,5,6")
# result_list = TopicUtils.get_recommend_topic_ids(user_id, tag_id=-1, single_size=size, offset=offset, size=size,
# topic_user_id=topic_user_id, current_topic_id=topic_id,
# topic_tag_list=topic_tag_list)
#
# user_recommend_list = TopicUtils.get_recommend_topic_ids(user_id, tag_id=-1, single_size=size, offset=offset, size=size,
# topic_user_id=topic_user_id, current_topic_id=topic_id,
# topic_tag_list=topic_tag_list)
redis_dict = {
"have_read_topic_list": json.dumps(have_read_topic_list),
}
redis_client.hmset(redis_key, redis_dict)
redis_client.expire(redis_key, 60 * 60 * 24)
result_list = []
logging.warning("topic_tag_result:%s" % str(topic_tag_result))
logging.warning("topic_user_result:%s" % str(topic_user_result))
logging.warning("result:%s" % str(result))
recommend_topic_ids_list = list()
recommend_topic_ids_list.extend(topic_tag_result)
recommend_topic_ids_list.extend(topic_user_result)
recommend_topic_ids_list.extend(result)
return {"recommend_topic_ids": recommend_topic_ids_list}
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return {"recommend_topic_ids": []}
@bind("physical/search/query_topic_by_tag_aggregation")
def query_topic_by_tag_aggregation(user_id, tag_id, offset, size):
"""
:remark 按标签聚合召回帖子
:param tag_id:
:param offset:
:param size:
:return:
"""
try:
if not user_id:
user_id = -1
if not tag_id:
tag_id = -1
result_list = TopicUtils.get_tag_aggregation_topic_id_list(user_id, tag_id, offset, size)
recommend_topic_ids_list = list()
if len(result_list) > 0:
recommend_topic_ids_list = [item["_source"]["id"] for item in result_list]
return {"recommend_topic_id": recommend_topic_ids_list}
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return {"recommend_topic_id": []}
@bind("physical/search/topic")
def topic_search(filters, nfilters=None, sorts_by=None, offset=0, size=10):
"""帖子搜索。"""
try:
(topic_id_list,total_count) = TopicUtils.list_topic_ids(filters=filters, nfilters=nfilters,
sorts_by=sorts_by, offset=offset, size=size)
return {
"topic_ids": topic_id_list,
"total_count": total_count
}
except:
logging.error("catch exception, err_msg:%s" % traceback.format_exc())
return {"topic_ids": [], "total_count": 0}
@bind("physical/search/query_topic_by_user_similarity")
def query_topic_by_user_similarity(topic_similarity_score_dict, offset=0, size=10):
"""
:remark 按帖子所属用户相似度召回帖子
:param tag_id:
:param offset:
:param size:
:return:
"""
try:
must_topic_id_list = list(topic_similarity_score_dict.keys())
topic_id_list = TopicUtils.get_recommend_topic_ids(tag_id=0, user_id=-1, offset=offset, size=size,single_size=size,
must_topic_id_list=must_topic_id_list,index_type="topic",routing="4,5,6")
return {"recommend_topic_ids": topic_id_list}
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return {"recommend_topic_id": []}