Skip to content

P
physical
Commit 13570b11 authored by 段英荣's avatar 段英荣
Browse files
Options

Merge branch 'master' of git.wanmeizhensuo.com:alpha/physical

parents dfc7b221 70876205
Hide whitespace changes
Inline Side-by-side
Showing with 231 additions and 23 deletions
search/utils/topic.py
View file @ 13570b11
...@@ -122,7 +122,7 @@ class TopicUtils(object): ...@@ -122,7 +122,7 @@ class TopicUtils(object):
@classmethod @classmethod
def get_recommend_topic_ids(cls,user_id,tag_id,offset,size,single_size,query=None,query_type=TopicPageType.FIND_PAGE, def get_recommend_topic_ids(cls,user_id,tag_id,offset,size,single_size,query=None,query_type=TopicPageType.FIND_PAGE,
filter_topic_id_list=[],test_score=False,must_topic_id_list=[],recommend_tag_list=[], filter_topic_id_list=[],test_score=False,must_topic_id_list=[],recommend_tag_list=[],
user_similar_score_list=[],index_type="topic",routing=None,attention_tag_list=[]): user_similar_score_list=[],index_type="topic",routing=None,attention_tag_list=[],current_topic_id =-1,topic_tag_list =[],topic_user_id =-1):
""" """
:remark:获取首页推荐帖子列表 :remark:获取首页推荐帖子列表
:param user_id: :param user_id:
...@@ -204,6 +204,25 @@ class TopicUtils(object): ...@@ -204,6 +204,25 @@ class TopicUtils(object):
"weight": 100 "weight": 100
} }
) )
if current_topic_id != -1:
if len(topic_tag_list) > 0:
functions_list.append(
{
"filter": {"bool": {
"should": {"terms": {"tag_list": topic_tag_list}}}},
"weight": 10000
}
)
if topic_user_id != -1:
functions_list.append(
{
"filter": {"bool": {
"should": {"term": {"user_id": topic_user_id}}}},
"weight": 5000
}
)
query_function_score = { query_function_score = {
"query": { "query": {
"bool": { "bool": {
...@@ -246,7 +265,12 @@ class TopicUtils(object): ...@@ -246,7 +265,12 @@ class TopicUtils(object):
query_function_score["query"]["bool"]["must_not"] = [ query_function_score["query"]["bool"]["must_not"] = [
{"terms":{"id":filter_topic_id_list}} {"terms":{"id":filter_topic_id_list}}
] ]
if current_topic_id != -1:
query_function_score["query"]["bool"]["must_not"] = [{
"term": {
"id": current_topic_id
}
}]
if query is not None: # 搜索帖子 if query is not None: # 搜索帖子
multi_fields = { multi_fields = {
'description': 200, 'description': 200,
...@@ -401,6 +425,131 @@ class TopicUtils(object): ...@@ -401,6 +425,131 @@ class TopicUtils(object):
logging.error("catch exception,err_msg:%s" % traceback.format_exc()) logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return [] return []
@classmethod
def top_get_topic_detail_recommend_list(cls, user_id, topic_id,have_read_topic_list, size, es_cli_obj=None,
index_type="topic", routing=None,collection_topic_tag_list = [],topic_tag_list = [],
topic_user_id =-1):
"""
:remark 帖子详情页推荐列表,缺少按时间衰减
:param user_id:
:param topic_tag_list:
:param topic_group_id:
:param topic_user_id:
:param offset:
:param size:
:return:
"""
try:
if not es_cli_obj:
es_cli_obj = ESPerform.get_cli()
q = dict()
q["query"] = dict()
# logging.warning("topic_tag_list:%s"%str(topic_tag_list))
functions_list = [
{
"linear": {
"create_time": {
"scale": "1d",
"decay": 0.5
}
}
}
]
# if len(topic_tag_list) > 0:
# functions_list.append(
# {
# "filter": {"bool": {
# "should": {"terms": {"tag_list": topic_tag_list}}}},
# "weight": 5000
# }
# )
# if topic_user_id != -1:
# functions_list.append(
# {
# "filter": {"bool": {
# "should": {"term": {"user_id": topic_user_id}}}},
# "weight": 5000
# }
# )
if len(topic_tag_list) != 0 or topic_user_id!= -1:
query_function_score = {
"query": {
"bool": {
"must": [
{"range": {"content_level": {"gte": 3, "lte": 6}}},
{"term": {"is_online": True}},
{"term": {"is_deleted": False}}
],
"must_not": {
"terms": {
"id": have_read_topic_list
}
}
}
},
"score_mode": "sum",
"boost_mode": "sum",
"functions": functions_list
}
else:
query_function_score = {
"query": {
"bool": {
"must": [
{"range": {"content_level": {"gte": 4, "lte": 6}}},
{"term": {"is_online": True}},
{"term": {"is_deleted": False}}
],
"must_not": {
"terms": {
"id": have_read_topic_list
}
}
}
},
"score_mode": "sum",
"boost_mode": "sum",
"functions": functions_list
}
if len(topic_tag_list) > 0:
query_function_score["query"]["bool"]["filter"] = {
"terms": {
"tag_list": topic_tag_list
}
}
if topic_user_id != -1:
query_function_score["query"]["bool"]["filter"] = {
"term": {
"user_id": topic_user_id
}
}
q["query"]["function_score"] = query_function_score
if topic_user_id == -1:
q["collapse"] = {
"field": "user_id"
}
q["_source"] = {
"includes": ["id", "pictorial_id", "user_id", "_score"]
}
# "includes": ["id", "pictorial_id", "user_id", "_score", "create_time", "content_level"]
# q['sort'] = [
# {"content_level": {"order": "desc"}},
# {"create_time": {"order": "desc"}}
# ]
result_dict = ESPerform.get_search_results(es_cli_obj, sub_index_name=index_type, query_body=q, size=size, routing=routing)
topic_id_list = list()
for item in result_dict["hits"]:
topic_id_list.append(item["_source"]["id"])
return topic_id_list
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return []
@classmethod @classmethod
def get_topic_tag_id_list(cls, topic_id, es_cli_obj=None): def get_topic_tag_id_list(cls, topic_id, es_cli_obj=None):
""" """
......
search/views/topic.py
View file @ 13570b11
...@@ -253,7 +253,7 @@ def home_query(device_id="", tag_id=-1, user_id=-1, query="", offset=0, size=10) ...@@ -253,7 +253,7 @@ def home_query(device_id="", tag_id=-1, user_id=-1, query="", offset=0, size=10)
@bind("physical/search/topic_detail_page_recommend") @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, 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): 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 :remark:帖子详情页推荐策略，缺少第一个卡片策略
:param user_id: :param user_id:
...@@ -265,31 +265,90 @@ def topic_detail_page_recommend(device_id="", user_id=-1, topic_id=-1, topic_pic ...@@ -265,31 +265,90 @@ def topic_detail_page_recommend(device_id="", user_id=-1, topic_id=-1, topic_pic
try: try:
if not isinstance(user_id, int): if not isinstance(user_id, int):
user_id = -1 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)
redis_key = "physical:topic_detail_page_recommend" + ":user_id:" + str(user_id) + ":device_id:" + str(device_id) else:
have_read_topic_redis_data = redis_client.get(redis_key) have_read_topic_list = list()
have_read_topic_list = json.loads(have_read_topic_redis_data) if have_read_topic_redis_data else [] redis_field_list = [b'have_read_topic_list']
# 获取es链接对象 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() 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)
# 获取帖子标签列表 result_list = []
topic_tag_list = TopicUtils.get_topic_tag_id_list(topic_id, es_cli_obj) # logging.warning("topic_tag_result:%s" % str(topic_tag_result))
# logging.warning("topic_user_result:%s" % str(topic_user_result))
result_list = TopicUtils.get_topic_detail_recommend_list(user_id, topic_id, topic_tag_list, topic_pictorial_id, # logging.warning("result:%s" % str(result))
topic_user_id, filter_topic_user_id,
have_read_topic_list, offset, size, es_cli_obj,index_type="topic",routing="4,5,6")
recommend_topic_ids_list = list() recommend_topic_ids_list = list()
if len(result_list) > 0: recommend_topic_ids_list.extend(topic_tag_result)
recommend_topic_ids_list = [item["_source"]["id"] for item in result_list] recommend_topic_ids_list.extend(topic_user_result)
recommend_topic_ids_list.extend(result)
have_read_topic_list.extend(recommend_topic_ids_list)
have_read_topic_len = len(have_read_topic_list)
if have_read_topic_len > 5000:
have_read_topic_list = have_read_topic_list[(have_read_topic_len - 5000):]
redis_client.set(redis_key, json.dumps(have_read_topic_list))
return {"recommend_topic_ids": recommend_topic_ids_list} return {"recommend_topic_ids": recommend_topic_ids_list}
except: except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc()) logging.error("catch exception,err_msg:%s" % traceback.format_exc())
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment