Commit 6dbaf23d authored by lixiaofang's avatar lixiaofang

topic

parent d147696f
...@@ -15,6 +15,7 @@ from libs.es import ESPerform ...@@ -15,6 +15,7 @@ from libs.es import ESPerform
from django.conf import settings from django.conf import settings
def get_discover_page_topic_ids(user_id, device_id, size, query_type=TopicPageType.FIND_PAGE): def get_discover_page_topic_ids(user_id, device_id, size, query_type=TopicPageType.FIND_PAGE):
try: try:
if user_id == -1: if user_id == -1:
...@@ -29,11 +30,10 @@ def get_discover_page_topic_ids(user_id, device_id, size, query_type=TopicPageTy ...@@ -29,11 +30,10 @@ def get_discover_page_topic_ids(user_id, device_id, size, query_type=TopicPageTy
if have_read_topic_id_list == None: if have_read_topic_id_list == None:
have_read_topic_id_list = list() 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, recommend_topic_ids = TopicUtils.get_recommend_topic_ids(user_id=user_id, tag_id=0, offset=0, size=size,single_size=size,
single_size=size,
query_type=query_type, query_type=query_type,
filter_topic_id_list=have_read_topic_id_list, filter_topic_id_list=have_read_topic_id_list,index_type="topic",routing="4,5,6")
index_type="topic", routing="4,5,6")
have_read_topic_id_list.extend(recommend_topic_ids) have_read_topic_id_list.extend(recommend_topic_ids)
redis_dict = { redis_dict = {
...@@ -48,25 +48,27 @@ def get_discover_page_topic_ids(user_id, device_id, size, query_type=TopicPageTy ...@@ -48,25 +48,27 @@ def get_discover_page_topic_ids(user_id, device_id, size, query_type=TopicPageTy
def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query=None, 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, query_type=TopicPageType.HOME_RECOMMEND,promote_topic_list = [],disable_collpase=False,usefulrecall = -1,useful_tag_list =[]):
usefulrecall=-1):
try: try:
topic_star_routing = "6" topic_star_routing = "6"
index_type = "topic-high-star" index_type = "topic-high-star"
device_redis_key = ""
if query is None: if query is None:
if user_id > 0: if user_id>0:
redis_key = "physical:home_recommend" + ":user_id:" + str(user_id) + ":query_type:" + str(query_type) 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: else:
redis_key = "physical:home_recommend" + ":device_id:" + device_id + ":query_type:" + str(query_type) redis_key = "physical:home_recommend" + ":device_id:" + device_id + ":query_type:" + str(query_type)
else: else:
topic_star_routing = "3,4,5,6" topic_star_routing = "3,4,5,6"
index_type = "topic" index_type = "topic"
if user_id > 0: if user_id>0:
redis_key = "physical:home_query" + ":user_id:" + str(user_id) + ":query:" + str( redis_key = "physical:home_query" + ":user_id:" + str(user_id) + ":query:" + str(query) + ":query_type:" + str(query_type)
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: else:
redis_key = "physical:home_query" + ":device_id:" + device_id + ":query:" + str( redis_key = "physical:home_query" + ":device_id:" + device_id + ":query:" + str(query) + ":query_type:" + str(query_type)
query) + ":query_type:" + str(query_type)
redis_field_list = [b'have_read_topic_list'] redis_field_list = [b'have_read_topic_list']
redis_field_val_list = redis_client.hmget(redis_key, redis_field_list) redis_field_val_list = redis_client.hmget(redis_key, redis_field_list)
...@@ -78,8 +80,18 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query ...@@ -78,8 +80,18 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query
if query is None: if query is None:
have_read_topic_id_list = list(json.loads(redis_field_val_list[0])) have_read_topic_id_list = list(json.loads(redis_field_val_list[0]))
else: else:
if offset > 0: # 首次搜索时不需要过滤已读 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])) 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: if have_read_topic_id_list == None:
have_read_topic_id_list = list() have_read_topic_id_list = list()
have_read_topic_id_list.extend(promote_topic_list) have_read_topic_id_list.extend(promote_topic_list)
...@@ -92,7 +104,7 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query ...@@ -92,7 +104,7 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query
# 有用标签召回 # 有用标签召回
if usefulrecall != -1: if usefulrecall != -1:
useful_topic_id_list = TopicUtils.userful_tag_topic_list(user_id, have_read_topic_id_list, 4, useful_topic_id_list = TopicUtils.userful_tag_topic_list(user_id, have_read_topic_id_list, 4,
"topic-high-star", "6") "topic-high-star", "6",useful_tag_list=useful_tag_list)
# user_similar_score_redis_key = "physical:user_similar_score:user_id:" + str(user_id) # 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) # redis_user_similar_score_redis_val = redis_client.get(user_similar_score_redis_key)
# user_similar_score_redis_list = json.loads( # user_similar_score_redis_list = json.loads(
...@@ -112,8 +124,7 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query ...@@ -112,8 +124,7 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query
linucb_recommend_topic_id_list = list() 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 # 推荐帖子是强插的,要保证推荐帖子不在已读里
logging.warning( logging.warning("type1:%s,type2:%s"%(type(linucb_recommend_topic_id_list),type(have_read_topic_id_list)))
"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 = 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) recommend_topic_id_list.sort(key=linucb_recommend_topic_id_list.index)
# cursor = int(str(recommend_topic_dict[b"cursor"], encoding="utf-8")) # cursor = int(str(recommend_topic_dict[b"cursor"], encoding="utf-8"))
...@@ -124,7 +135,7 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query ...@@ -124,7 +135,7 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query
if b"datadict" in recommend_topic_dict: if b"datadict" in recommend_topic_dict:
linucb_recommend_topic_id_dict = json.loads(recommend_topic_dict[b"datadict"]) 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: if linucb_recommend_topic_id_dict is not None and len(recommend_topic_list) >0:
for i in recommend_topic_list: for i in recommend_topic_list:
recommend_topic_user_list.append(linucb_recommend_topic_id_dict[str(i)]) recommend_topic_user_list.append(linucb_recommend_topic_id_dict[str(i)])
# if have_read_topic_id_list == None: # if have_read_topic_id_list == None:
...@@ -132,8 +143,8 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query ...@@ -132,8 +143,8 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query
# 用户关注标签 # 用户关注标签
redis_tag_data = redis_client.hget("physical:linucb:register_user_tag_info", user_id) 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 [] attention_tag_list = json.loads(redis_tag_data) if redis_tag_data else []
if len(recommend_topic_list) > 0: if len(recommend_topic_list)>0:
size = size - len(recommend_topic_list) size = size-len(recommend_topic_list)
have_read_topic_id_list.extend(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 = list()
...@@ -144,15 +155,11 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query ...@@ -144,15 +155,11 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query
# for topic_id in promote_recommend_topic_id_list: # for topic_id in promote_recommend_topic_id_list:
# have_read_topic_id_list_add_promote.append(topic_id) # have_read_topic_id_list_add_promote.append(topic_id)
rank_topic_id_list = list() rank_topic_id_list = list()
if size > 0: if size>0:
rank_topic_id_list = TopicUtils.get_recommend_topic_ids(user_id=user_id, tag_id=tag_id, offset=0, size=size, 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, single_size=size,query=query, query_type=query_type,
query_type=query_type,
filter_topic_id_list=have_read_topic_id_list, filter_topic_id_list=have_read_topic_id_list,
index_type=index_type, routing=topic_star_routing, 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)
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 len(recommend_topic_list) == 6 and query is None:
# if (size < 11): # if (size < 11):
...@@ -169,7 +176,7 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query ...@@ -169,7 +176,7 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query
# topic_id_list.extend(rank_topic_id_list) # topic_id_list.extend(rank_topic_id_list)
have_read_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: if len(have_read_topic_id_list) > 30000:
cut_len = 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:] have_read_topic_id_list = have_read_topic_id_list[cut_len:]
redis_dict = { redis_dict = {
"have_read_topic_list": json.dumps(have_read_topic_id_list), "have_read_topic_list": json.dumps(have_read_topic_id_list),
...@@ -178,13 +185,13 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query ...@@ -178,13 +185,13 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query
# 每个session key保存60天 # 每个session key保存60天
redis_client.expire(redis_key, 60 * 60 * 24 * 60) redis_client.expire(redis_key, 60 * 60 * 24 * 60)
if usefulrecall != -1: if usefulrecall != -1:
return recommend_topic_list, rank_topic_id_list, useful_topic_id_list return recommend_topic_list,rank_topic_id_list,useful_topic_id_list
else: else:
return recommend_topic_list, rank_topic_id_list return recommend_topic_list, rank_topic_id_list
except: except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc()) logging.error("catch exception,err_msg:%s" % traceback.format_exc())
if usefulrecall != -1: if usefulrecall != -1:
return [], [], [] return [],[],[]
else: else:
return [], [] return [], []
...@@ -199,8 +206,7 @@ def query_tag_id_by_topic(offset=0, size=10, topic_id_list=[], user_id=-1): ...@@ -199,8 +206,7 @@ def query_tag_id_by_topic(offset=0, size=10, topic_id_list=[], user_id=-1):
@bind("physical/search/home_recommend") @bind("physical/search/home_recommend")
def home_recommend(device_id="", user_id=-1, offset=0, size=10, query_type=TopicPageType.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=[]):
promote_topic_list=[], usefulrecall=-1):
""" """
:remark:首页推荐,目前只推荐日记 :remark:首页推荐,目前只推荐日记
:param session_id: :param session_id:
...@@ -233,31 +239,24 @@ def home_recommend(device_id="", user_id=-1, offset=0, size=10, query_type=Topic ...@@ -233,31 +239,24 @@ def home_recommend(device_id="", user_id=-1, offset=0, size=10, query_type=Topic
have_read_topic_id_list = list(json.loads(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: if len(have_read_topic_id_list) > offset:
recommend_topic_ids = have_read_topic_id_list[offset:offset + size] recommend_topic_ids = have_read_topic_id_list[offset:offset+size]
else: else:
recommend_topic_ids = have_read_topic_id_list[0:size] recommend_topic_ids = have_read_topic_id_list[0:size]
else: else:
if usefulrecall != -1: 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, 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,
tag_id=0, offset=0, query_type=query_type,promote_topic_list=promote_topic_list,usefulrecall=usefulrecall,useful_tag_list=useful_tag_list)
size=size, return {"linucb_topic_ids": recommend_topic_ids,"rank_topic_ids":rank_topic_ids,"useful_topic_ids":useful_topic_ids}
query_type=query_type,
promote_topic_list=promote_topic_list,
usefulrecall=usefulrecall)
return {"linucb_topic_ids": recommend_topic_ids, "rank_topic_ids": rank_topic_ids,
"useful_topic_ids": useful_topic_ids}
else: else:
recommend_topic_ids, rank_topic_ids = get_home_recommend_topic_ids(user_id, device_id, tag_id=0, recommend_topic_ids,rank_topic_ids = get_home_recommend_topic_ids(user_id, device_id, tag_id=0, offset=0, size=size,
offset=0, size=size, query_type=query_type,promote_topic_list=promote_topic_list)
query_type=query_type,
promote_topic_list=promote_topic_list)
return {"linucb_topic_ids": recommend_topic_ids, "rank_topic_ids": rank_topic_ids} return {"linucb_topic_ids": recommend_topic_ids,"rank_topic_ids":rank_topic_ids}
except: except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc()) logging.error("catch exception,err_msg:%s" % traceback.format_exc())
if usefulrecall != -1: if usefulrecall != -1:
return {"linucb_topic_ids": [], "rank_topic_ids": [], "useful_topic_ids": []} return {"linucb_topic_ids": [],"rank_topic_ids":[],"useful_topic_ids":[]}
else: else:
return {"linucb_topic_ids": [], "rank_topic_ids": []} return {"linucb_topic_ids": [], "rank_topic_ids": []}
...@@ -303,12 +302,11 @@ def home_query(device_id="", tag_id=-1, user_id=-1, query="", offset=0, size=10) ...@@ -303,12 +302,11 @@ def home_query(device_id="", tag_id=-1, user_id=-1, query="", offset=0, size=10)
if not isinstance(device_id, str): if not isinstance(device_id, str):
device_id = "" device_id = ""
recommend_topic_list, rank_topic_id_list = get_home_recommend_topic_ids(user_id, device_id, tag_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)
offset=offset, size=size, query=query) if len(rank_topic_id_list)>0 and len(rank_topic_id_list)<size:
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, recommend_topic_list, rank_topic_id_list = get_home_recommend_topic_ids(user_id, device_id, tag_id,
offset=offset, size=size, offset=offset, size=size,
query=query, disable_collpase=True) query=query,disable_collpase=True)
return {"recommend_topic_ids": rank_topic_id_list} return {"recommend_topic_ids": rank_topic_id_list}
except: except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc()) logging.error("catch exception,err_msg:%s" % traceback.format_exc())
...@@ -330,9 +328,9 @@ def topic_detail_page_recommend(device_id="", user_id=-1, topic_id=-1, topic_pic ...@@ -330,9 +328,9 @@ def topic_detail_page_recommend(device_id="", user_id=-1, topic_id=-1, topic_pic
if not isinstance(user_id, int): if not isinstance(user_id, int):
user_id = -1 user_id = -1
if user_id > 0: if user_id > 0:
redis_key = "physical:topic_detail_recommend" + ":user_id:" + str(user_id) + "topic_id:" + str(topic_id) redis_key = "physical:topic_detail_recommend" + ":user_id:" + str(user_id) + "topic_id:"+str(topic_id)
else: else:
redis_key = "physical:topic_detail_recommend" + ":device_id:" + device_id + "topic_id:" + str(topic_id) redis_key = "physical:topic_detail_recommend" + ":device_id:" + device_id + "topic_id:"+str(topic_id)
if int(offset) == 0: if int(offset) == 0:
have_read_topic_list = list() have_read_topic_list = list()
# redis_dict = { # redis_dict = {
...@@ -345,7 +343,7 @@ def topic_detail_page_recommend(device_id="", user_id=-1, topic_id=-1, topic_pic ...@@ -345,7 +343,7 @@ def topic_detail_page_recommend(device_id="", user_id=-1, topic_id=-1, topic_pic
have_read_topic_list = list() have_read_topic_list = list()
redis_field_list = [b'have_read_topic_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_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 [] 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) have_read_topic_list.append(topic_id)
...@@ -353,23 +351,18 @@ def topic_detail_page_recommend(device_id="", user_id=-1, topic_id=-1, topic_pic ...@@ -353,23 +351,18 @@ def topic_detail_page_recommend(device_id="", user_id=-1, topic_id=-1, topic_pic
topic_tag_result = list() topic_tag_result = list()
result = list() result = list()
if len(topic_tag_list) != 0: if len(topic_tag_list) != 0:
topic_tag_result = TopicUtils.top_get_topic_detail_recommend_list(user_id, topic_id, have_read_topic_list, topic_tag_result = TopicUtils.top_get_topic_detail_recommend_list(user_id,topic_id,have_read_topic_list,size,es_cli_obj,
size, es_cli_obj, index_type="topic",routing="3,4,5,6",topic_tag_list = topic_tag_list)
index_type="topic", routing="3,4,5,6",
topic_tag_list=topic_tag_list)
topic_tag_size = len(topic_tag_result) topic_tag_size = len(topic_tag_result)
have_read_topic_list.extend(topic_tag_result) have_read_topic_list.extend(topic_tag_result)
else: else:
topic_tag_size = 0 topic_tag_size = 0
if topic_tag_size < size: if topic_tag_size <size:
size = size - topic_tag_size size = size - topic_tag_size
if topic_user_id != -1: if topic_user_id != -1:
topic_user_result = TopicUtils.top_get_topic_detail_recommend_list(user_id, topic_id, topic_user_result = TopicUtils.top_get_topic_detail_recommend_list(user_id, topic_id, have_read_topic_list,
have_read_topic_list,
size, es_cli_obj, size, es_cli_obj,
index_type="topic", index_type="topic", routing="3,4,5,6",topic_user_id = topic_user_id
routing="3,4,5,6",
topic_user_id=topic_user_id
) )
topic_user_size = len(topic_user_result) topic_user_size = len(topic_user_result)
have_read_topic_list.extend(topic_user_result) have_read_topic_list.extend(topic_user_result)
...@@ -455,7 +448,7 @@ def topic_search(filters, nfilters=None, sorts_by=None, offset=0, size=10): ...@@ -455,7 +448,7 @@ def topic_search(filters, nfilters=None, sorts_by=None, offset=0, size=10):
"""帖子搜索。""" """帖子搜索。"""
try: try:
(topic_id_list, total_count) = TopicUtils.list_topic_ids(filters=filters, nfilters=nfilters, (topic_id_list,total_count) = TopicUtils.list_topic_ids(filters=filters, nfilters=nfilters,
sorts_by=sorts_by, offset=offset, size=size) sorts_by=sorts_by, offset=offset, size=size)
return { return {
...@@ -479,26 +472,10 @@ def query_topic_by_user_similarity(topic_similarity_score_dict, offset=0, size=1 ...@@ -479,26 +472,10 @@ def query_topic_by_user_similarity(topic_similarity_score_dict, offset=0, size=1
try: try:
must_topic_id_list = list(topic_similarity_score_dict.keys()) 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, topic_id_list = TopicUtils.get_recommend_topic_ids(tag_id=0, user_id=-1, offset=offset, size=size,single_size=size,
single_size=size, must_topic_id_list=must_topic_id_list,index_type="topic",routing="4,5,6")
must_topic_id_list=must_topic_id_list, index_type="topic",
routing="4,5,6")
return {"recommend_topic_ids": topic_id_list} return {"recommend_topic_ids": topic_id_list}
except: except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc()) logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return {"recommend_topic_id": []} return {"recommend_topic_id": []}
@bind("physical/search/query_topic_search")
def query_topic_search(query=""):
"""
召回搜索帖子
当搜索的内容 完全匹配 用户昵称,且该用户为推荐用户时,帖子tab中也会展示用户栏
:param query:
:return:
"""
#召回完全匹配的用户
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