tasks/offline-compute/PrestoSQL/prestosql_005_en/task.md

metadata

id: offline-compute_PrestoSQL_prestosql_005
name: Hot Table Governance Metadata Wide Table Aggregation
category: offline-compute/PrestoSQL
timeout_seconds: 900
modality: pure-text
engine: prestosql

Prompt

Task Objective: From the hot table governance metadata detail table, filter valid hot table records (heat > 0, excluding temporary/test tables), aggregate by database name and table name to take the maximum of each metric, and write the results to the output table.

Input Table:

• internal_platform_db.ads_gov_cost_table_govern_detail_df_prestosql_005 (hot table governance metadata detail table)
  • imp_date BIGINT — date partition
  • db_name STRING — database name
  • table_name STRING — table name
  • heat BIGINT — heat
  • owner STRING — owner
  • storage_detail DOUBLE — storage detail
  • task_cnt BIGINT — task count
  • table_type STRING — table type
  • govern_status STRING — governance status
  • last_access_time STRING — last access time
  • table_comment STRING — table comment
  • partition_cnt BIGINT — partition count
  • file_format STRING — file format
  • table_size DOUBLE — table size
  • avg_file_size DOUBLE — average file size
  • is_partitioned INT — whether partitioned
  • lifecycle_days INT — lifecycle days
  • create_time STRING — creation time
  • modify_time STRING — modification time
  • project_name STRING — project name

Filter Conditions:

• imp_date = 20260507
• heat > 0
• table_name NOT LIKE '%temp%'
• db_name NOT LIKE '%test%'

Output Requirements:

• Target table: internal_platform_db.dwd_hot_metadata_table_cand_prestosql_005
• Output fields and order: dt STRING, db_name STRING, table_name STRING, owner STRING, storage_detail DOUBLE, heat BIGINT, task_cnt BIGINT, table_type STRING, govern_status STRING, last_access_time STRING, table_comment STRING, partition_cnt BIGINT, file_format STRING, table_size DOUBLE, avg_file_size DOUBLE, is_partitioned INT, lifecycle_days INT, create_time STRING, modify_time STRING, project_name STRING
• dt: Fixed value '20260507'
• Group by db_name, table_name, apply MAX() to all other fields
• If the target table does not exist, first create the table, then write the data
• Use Presto/Trino SQL syntax; do not use Hive/Spark SQL dialects

Environment and Execution Notes:

• Presto is running, connected via the Hive catalog
• Execute SQL: presto-cli --catalog hive --schema internal_platform_db -f /tmp_workspace/result.sql
• After writing result.sql, you must execute it yourself to verify that it runs successfully and produces correct data