--- id: offline-compute_HiveSQL_hivesql_005 name: 从 internal_platform_db.dws_mq_production_featur category: offline-compute/HiveSQL timeout_seconds: 600 modality: pure-text engine: hivesql --- ## Prompt 任务目标:汇总当天有生产量的 消息队列MQ topic 维度信息及近7/30/90天生产统计,落地为 topic 治理项明细表。 输入: - 表:internal_platform_db.dws_mq_production_feature_d_increase_query_engine_013 - 分区字段:dt(STRING,格式 YYYYMMDD) - 关键字段:business_id、business_name、topic、cluster_set、tenant、namespaces、system_belong、dw_appgroup、in_charge、description、create_time、modify_time、cluster_id、cluster_type、cluster_name、bg、category_name、total_produce_pkg_d、production_days_last_7d、total_produce_pkg_last_7d、production_days_last_30d、total_produce_pkg_last_30d、production_days_last_90d、total_produce_pkg_last_90d、is_consumed、has_consumption_days_last_7d、has_consumption_days_last_30d、has_consumption_days_last_90d 处理规则: 1. 过滤条件:dt = '20260507' AND total_produce_pkg_last_90d > 0 2. 无 Join,单表处理 3. 派生字段: - mq_full_topic:若 tenant 和 namespaces 均非 NULL,则拼接为 'persistent://' + tenant + '/' + namespaces + '/' + topic;否则直接取 topic - app_group:取自 dw_appgroup - bid_incharge:取自 in_charge - bid_description:取自 description - bid_create_time:取自 create_time - bid_modify_time:取自 modify_time - hitted_gov_items:固定为 NULL - governance_benefit_estimate:固定为 NULL 输出要求: - 输出字段顺序:dt、business_id、business_name、topic、cluster_set、mq_full_topic、system_belong、app_group、bid_incharge、bid_description、bid_create_time、bid_modify_time、cluster_id、cluster_type、cluster_name、bg、category_name、total_produce_pkg_d、production_days_last_7d、total_produce_pkg_last_7d、production_days_last_30d、total_produce_pkg_last_30d、production_days_last_90d、total_produce_pkg_last_90d、is_consumed、has_consumption_days_last_7d、has_consumption_days_last_30d、has_consumption_days_last_90d、hitted_gov_items、governance_benefit_estimate 写入要求: - 目标表:internal_platform_db.ads_mq_topic_governance_item_d_cand_query_engine_013 - 分区字段:dt(STRING,格式 YYYYMMDD) - 写入方式:INSERT OVERWRITE 按分区写入 - 若目标表不存在,先按 Hive 标准建表(ORC存储、按 dt 分区),再写入 请将最终 HiveSQL 写入 result.sql 并执行。