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 get_item_list_by_item(user_info, resource_type, count):
""" get item_list by item recommendation
@return RecommendItemContainer
"""
logger.debug("begin get_item_list_by_item")
uid = user_info.uid
result = RecommendItemContainer()
#get favor item_id列表
errCode, kv_dict = model.get_online_user_favor_item_list(uid, [resource_type])
if not is_valid_data(errCode, kv_dict, "online_user_favor_item_list", {"uid":uid, "resource_type":resource_type}):
return result
if not kv_dict:
return result
assert(resource_type in kv_dict)
try:
item_id_list_proto = ItemIdList()
item_id_list_proto.ParseFromString(kv_dict[resource_type])
reason_list = [resource_type+ITEM_RECOMMEND_FEATURE_TYPE+v for v in item_id_list_proto.item_id]
item_id_list = []
for v in item_id_list_proto.item_id:
if not check_reason(user_info, resource_type+ITEM_RECOMMEND_FEATURE_TYPE+v):
continue
item_id_list.append(v)
if not item_id_list:
logger.info("all item id list have no item2item result:%s" % item_id_list)
return result
#@note: use latest item first
item_id_list.reverse()
#recommend by item id
errCode, kv_dict = model.get_offline_item_recommend_item_list(item_id_list)
if not is_valid_data(errCode, kv_dict, "offline_item_recommend_item_list", {"item_id_list":"..."}):
return result
if not kv_dict:
for item_id in item_id_list:
reason = resource_type + ITEM_RECOMMEND_FEATURE_TYPE + item_id
ignore_reason(user_info, reason)
return result
for item_id in item_id_list:
if result.size() >= count:
break
if item_id not in kv_dict:
logger.info("no item2itemid list for id:%s" % item_id)
ignore_reason(user_info, reason)
continue
k = item_id
v = kv_dict[k]
reason = resource_type + ITEM_RECOMMEND_FEATURE_TYPE + k
item_id_list_proto = ItemIdList()
item_id_list_proto.ParseFromString(v)
filter_and_compose_recommend_item_list(user_info, reason, item_id_list_proto, result)
except:
logger.error("bad get_item_list_by_item:%s" % traceback.format_exc())
return result
def get_item_list_by_feature_list(user_info, resource_type, feature_name_list_proto, count):
""" get item list by check with all feature list
@return RecommendItemContainer
"""
logger.debug("begin get_item_list_by_feature_list")
feature_name_list = []
result = RecommendItemContainer()
filtered_feature_name_list_proto = UserFeatureList()
for v in feature_name_list_proto.feature:
feature_name = v.feature_name
if not check_reason(user_info, feature_name):
continue
feature_name_list.append(feature_name)
cur = filtered_feature_name_list_proto.feature.add()
cur.feature_name = feature_name
if not feature_name_list:
logger.info("no new reason from feature list,current feature_name size:%s" % len(feature_name_list_proto.feature))
logger.debug("current feature_name_list_proto:%s" % feature_name_list_proto)
return result
#news like
if ResourceType.RESOURCE_TYPE_NEWS == resource_type:
return get_news_online_item_by_feature_list(user_info, resource_type, filtered_feature_name_list_proto, count)
#items like
errCode, kv_dict = model.get_offline_feature_hot_item_list(feature_name_list)
if not is_valid_data(errCode, kv_dict, "offline_feature_hot_item_list", {"feature_name_list":"..."}):
return result
if not kv_dict:
for k in feature_name_list:
ignore_reason(user_info, k)
return result
try:
#logger.debug("%s" % feature_name_list)
for k in feature_name_list: #@note:ordered
logger.debug("check result with feature_name:%s" % k)
if result.size() >= count:
logger.debug("fetch enough recommend result, count:%s" % count)
break
if k not in kv_dict:
logger.info("no result for feature_name:%s" % k)
ignore_reason(user_info, k)
continue
v = kv_dict[k]
item_id_list = ItemIdList()
item_id_list.ParseFromString(v)
filter_and_compose_recommend_item_list(user_info, k, item_id_list, result)
logger.debug("get_item_list_by_feature_list,result_size:%s, uid:%s, resource_type:%s" % (result.size(), user_info.uid, resource_type))
except:
logger.error("bad ItemIdList for get_item_list_by_feature_list:%s" % traceback.format_exc())
return result
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