def update_user_info(user_info, recommend_result):
""" update recommend_result to user_info
@return None
"""
#@note:暂时不获取item_feature结构（不管是news还是item类，为了计算用户的feature和feature偏好的pv部分）；
#@note:item_id列表的push由外围世荣处保证
for reason,item_list in recommend_result.reason2itemidlist.iteritems():
add_reason(user_info, reason)
for v in item_list:
resource_type = ResourceType.get_resource_type(v.item_id)
assert(resource_type)
user_info.recent_resource_visit_info.setdefault(resource_type, VisitInfo())
user_info.recent_resource_visit_info[resource_type].pv_count += 1
#set online user_feature_reason and user resource visit info
feature_list = ItemFeatureList()
for reason,weight in user_info.recent_push_reason_info.iteritems():
feature = feature_list.feature.add()
feature.feature_name = reason
feature.weight = weight
uid = user_info.uid
errCode = model.set_online_user_feature_reason(uid, feature_list.SerializeToString())
if not errCode:
logger.warn("fail to set_online_user_feature_reason, uid:%s" % uid)
for resource_type,visit_info in user_info.recent_resource_visit_info.iteritems():
errCode = model.set_online_user_resource_visitinfo(uid, resource_type, visit_info.SerializeToString())
if not errCode:
logger.warn("fail to set_online_user_resource_visitinfo, uid:%s, resource_type:%s" % (uid, resource_type))
#feedback for show action
update_user_feature_list_for_show_action(user_info, recommend_result)
def filter_and_compose_recommend_item_list(user_info, reason, item_id_list_proto, result):
""" try set result[reason] = item_id_list_proto and compose result if no dedup, update user_info
@return None
"""
logger.debug("try to add recommend, reason:%s, item_id_list_proto:%s" % (reason, item_id_list_proto))
if not check_reason(user_info, reason):
logger.debug("existing feature_type, ignore:%s" % reason)
return
result_item_id_list = []
resource_type = ResourceType.get_resource_type(reason)
for v in item_id_list_proto.item_id:
if v in user_info.recent_push_item_id_list:
continue
result_item_id_list.append(v)
user_info.recent_push_item_id_list.add(v)
max_recommend_item_count_per_reason = get_max_recommend_item_count_per_resource(resource_type)
if len(result_item_id_list) >= max_recommend_item_count_per_reason:
logger.debug("real_list size:%s, select part:%s" % (len(item_id_list_proto.item_id),
max_recommend_item_count_per_reason))
break
if not result_item_id_list:
logger.info("all item id list has been pushed")
logger.debug("all item id list has been pushed:%s" % item_id_list_proto)
return
if not result_item_id_list:
return
result_recommend_item = [RecommendItem(item_id=v) for v in result_item_id_list]
result.add(reason, result_recommend_item)
add_reason(user_info, reason)
return
def raw_get_news_online_item_by_feature_list(user_info, resource_type, feature_name_list_proto,
count, max_recommend_item_count_per_reason, fetch_more_times, backup_feature_count):
""" talk with news online item interface by feature list, @return RecommendItemContainer
@note: should filter visited feature in user_info for user_feature_list, and update user_info
"""
result = RecommendItemContainer()
input = FeatureNameLimitList()
logger.debug("feature list proto:%s" % feature_name_list_proto)
for i in feature_name_list_proto.feature:
if not check_reason(user_info, i.feature_name):
continue
#@note: temp logic to ignore some feature type
continue_out = False
for prefix in NEWS_IGNORE_FEATURE_NAME_PREFIX:
if i.feature_name.startswith(prefix):
logger.info("ignore feature:%s" % i.feature_name)
continue_out = True
break
if continue_out:
continue
cur = input.feature_name_limit.add()
cur.feature_name = i.feature_name
cur.limit = int(count * max_recommend_item_count_per_reason * fetch_more_times)
if len(input.feature_name_limit) >= count + backup_feature_count:
logger.info("break for enough feature count:%s,feature list:%s" %
(len(input.feature_name_limit), input))
break
#logger.debug("feature_name:%s,limit:%s" % (cur.feature_name, cur.limit))
if not input.feature_name_limit:
logger.info("no new feature to get for news:%s, feature_list size:%s" % (resource_type, len(feature_name_list_proto.feature)))
logger.debug("current feature_list :%s" % (feature_name_list_proto))
return result
try:
#socket_out = urllib2.urlopen(NEWS_ONLINE_URL_SEARCH_BY_QUERY, urllib.urlencode(input.SerializeToString()))
socket_out = urllib2.urlopen(NEWS_ONLINE_URL_SEARCH_BY_QUERY, input.SerializeToString(),
timeout=online_clicklog_feedback.NEWS_TIMEOUT_SECONDS)
return_data = socket_out.read()
except:
logger.error("fail to talk with online news server,featurelists:%s,exceptions:%s" %
(input, traceback.format_exc()))
return result
try:
output = NewsFeatureItemList()
output.ParseFromString(return_data)
logger.debug("get news online item list result,user_info:%s,result:%s" % (user_info, output))
except:
logger.error("bad parse news return value:%s" % traceback.format_exc())
return result
for reason_item_id_list in output.feature_item:
if not len(reason_item_id_list.item_feature_info):
continue
reason = reason_item_id_list.feature_name
resource_type = ResourceType.get_resource_type(reason)
item_list = []
for i in reason_item_id_list.item_feature_info:
if i.item_id in user_info.recent_push_item_id_list:
logger.debug("ignore existing item id:%s" % i.item_id)
continue
item = RecommendItem()
item.item_id = i.item_id
item.item_info_json = i.item_info
item.feature_name_list = i.item_feature_list
item_list.append(item)
user_info.recent_push_item_id_list.add(i.item_id)
logger.info("item_list:%s" %item_list)
if len(item_list) >= max_recommend_item_count_per_reason:
logger.debug("real_list size:%s, select part:%s" % (len(reason_item_id_list.item_feature_info),
max_recommend_item_count_per_reason))
break
if not item_list or len(item_list) < max_recommend_item_count_per_reason:
logger.info("not enough data, discard:%s" % (item_list))
ignore_reason(user_info, reason)
#fetch not enough data, discard this reason
continue
result.add(reason, item_list)
add_reason(user_info, reason)
if result.size() >= count:
break
if not result:
logger.info("get empty get_news_online_item_by_feature_list")
return result
