Datasets:

dicemy
/

DataClawEval

-- mysql_003 database initialization: create tables + load seed data
-- Executed via: mysql -u root < init_db.sql
--
-- NOTE: This includes ALL data from build_inputs, AC_rerun, and landing phases
-- to produce the expected 58-row output matching expected.csv.

-- Fix: MySQL root default auth_socket blocks pymysql TCP connections,
-- switch to mysql_native_password so pymysql (Python) can connect.
ALTER USER 'root'@'localhost' IDENTIFIED WITH mysql_native_password BY 'root123';
FLUSH PRIVILEGES;

CREATE DATABASE IF NOT EXISTS internal_platform_db
  DEFAULT CHARACTER SET utf8mb4
  DEFAULT COLLATE utf8mb4_unicode_ci;

USE internal_platform_db;

-- Input Table 1: aide_offline_inference_info_fixed
CREATE TABLE IF NOT EXISTS aide_offline_inference_info_fixed_mysql_003 (
    id INT COMMENT 'id',
    wsid INT COMMENT 'wsid',
    name VARCHAR(256) COMMENT 'name',
    `desc` VARCHAR(256) COMMENT 'desc',
    inference_id INT,
    service_id INT,
    model_ids VARCHAR(256),
    input_config VARCHAR(256),
    output_config VARCHAR(256),
    visible INT,
    create_time VARCHAR(256),
    update_time VARCHAR(256),
    status VARCHAR(256),
    pipeline VARCHAR(256),
    modifier_id INT,
    source INT,
    users VARCHAR(256),
    celery_id VARCHAR(256),
    base VARCHAR(256),
    model_change INT,
    inference_config VARCHAR(256),
    location VARCHAR(256),
    activity_id INT,
    api_token VARCHAR(256),
    activity_body VARCHAR(256),
    file_id INT,
    max_wait_time_in_day INT,
    is_timeout INT,
    model_type VARCHAR(256),
    job_type VARCHAR(256),
    submit_type VARCHAR(256),
    item_retry_inference_count INT,
    pass_n_config VARCHAR(256),
    enable_debug INT,
    debug_ttl INT,
    retry_count INT,
    debug_expire_time VARCHAR(256),
    enable_proxy INT,
    enable_post_process INT,
    enable_only_model_deploy INT,
    enable_global_cache INT,
    service_name VARCHAR(256),
    custom_eval_task_config VARCHAR(256),
    is_custom_eval_task INT,
    custom_eval_service_id INT,
    spark_config VARCHAR(256),
    dt BIGINT
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4;

-- build_inputs: original seed data
INSERT INTO aide_offline_inference_info_fixed_mysql_003 VALUES
(1001, 100, 'task_alpha', 'desc1', 1, 1, '{"mould_id": 501, "mould_name": "model_A"}', '{}', '{}', 1, '2026-05-07 06:30:00', '2026-05-07 06:55:00', 'RUNNING', '{}', 1, 1, 'user1', 'cel1', 'JOB', 0, '{"config":"val1"}', 'sh', 0, '', '', 201, 3, 0, 'STATIC_MODEL', 'NORMAL', 'API', 1, '', 0, 0, 3, '', 0, 0, 0, 0, '', '', 0, 0, '', 2026050700),
(1002, 100, 'task_beta', 'desc2', 2, 2, '{"mould_id": 502, "mould_name": "model_B"}', '{}', '{}', 1, '2026-05-07 05:00:00', '2026-05-07 06:50:00', 'FINISH', '{}', 1, 1, 'user2', 'cel2', 'JOB', 0, '{"config":"val2"}', 'sh', 0, '', '', 202, 3, 0, 'STATIC_MODEL', 'NORMAL', 'WEB', 1, '', 0, 0, 3, '', 0, 0, 0, 0, '', '', 0, 0, '', 2026050700),
(1003, 100, 'task_gamma', 'desc3', 3, 3, '{"mould_id": 503, "mould_name": "model_C"}', '{}', '{}', 1, '2026-05-07 06:10:00', '2026-05-07 06:40:00', 'FAILED', '{}', 1, 1, 'user3', 'cel3', 'JOB', 0, '{"config":"val3"}', 'bj', 0, '', '', 203, 3, 0, 'STATIC_MODEL', 'NORMAL', 'API', 1, '', 0, 0, 3, '', 0, 0, 0, 0, '', '', 0, 0, '', 2026050700),
(1004, 100, 'task_delta', 'desc4', 4, 4, '{"mould_id": 504, "mould_name": "model_D"}', '{}', '{}', 1, '2026-04-01 10:00:00', '2026-04-01 11:00:00', 'FINISH', '{}', 1, 1, 'user4', 'cel4', 'JOB', 0, '{}', 'sh', 0, '', '', 204, 3, 0, 'STATIC_MODEL', 'NORMAL', 'API', 1, '', 0, 0, 3, '', 0, 0, 0, 0, '', '', 0, 0, '', 2026050700),
(1005, 100, 'task_epsilon', 'desc5', 5, 5, '{"mould_id": 505, "mould_name": "model_E"}', '{}', '{}', 1, '2026-05-07 06:20:00', '2026-05-07 06:45:00', 'KILL', '{}', 1, 1, 'user5', 'cel5', 'OTHER', 0, '{}', 'sh', 0, '', '', 205, 3, 0, 'STATIC_MODEL', 'NORMAL', 'API', 1, '', 0, 0, 3, '', 0, 0, 0, 0, '', '', 0, 0, '', 2026050700);

-- AC_rerun insert: id=1001 (WEB/submit_type, model_x, create_time=00:10)
INSERT INTO aide_offline_inference_info_fixed_mysql_003 VALUES
(1001, 100, 'task_alpha', 'desc1', 1, 1, '{"mould_id": 501, "mould_name": "model_x"}', '', '', 1, '2026-05-07 00:10:00', '2026-05-07 00:30:00', 'RUNNING', '', 0, 0, '', '', 'JOB', 0, '{"batch":10}', 'shanghai', 0, '', '', 2001, 7, 0, 'STATIC_MODEL', 'BATCH', 'WEB', 3, '', 0, 0, 2, '', 0, 0, 0, 0, 'svc_alpha', '', 0, 0, '', 2026050700),
(1002, 100, 'task_beta', 'desc2', 2, 2, '{"mould_id": 502, "mould_name": "model_y"}', '', '', 1, '2026-05-06 23:50:00', '2026-05-06 23:55:00', 'FINISH', '', 0, 0, '', '', 'JOB', 0, '{"batch":20}', 'beijing', 0, '', '', 2002, 7, 0, 'STATIC_MODEL', 'BATCH', 'API', 3, '', 0, 0, 2, '', 0, 0, 0, 0, 'svc_beta', '', 0, 0, '', 2026050700);

-- landing insert: id=1001 (API/submit_type, model_a), id=1002 (WEB, model_b), id=1003 (SDK, model_c)
INSERT INTO aide_offline_inference_info_fixed_mysql_003 VALUES
(1001, 100, 'task_alpha', 'desc1', 1, 1, '{"mould_id": 501, "mould_name": "model_a"}', '{}', '{}', 1, '2026-05-07 00:10:00', '2026-05-07 00:30:00', 'RUNNING', '{}', 1, 1, 'user1', 'cel1', 'JOB', 0, '{"key":"val"}', 'bj', 0, 'token1', '{}', 201, 3, 0, 'STATIC_MODEL', 'BATCH', 'API', 1, '{}', 0, 3600, 1, NULL, 0, 0, 0, 0, 'svc1', '{}', 0, 0, '{}', 2026050700),
(1002, 100, 'task_beta', 'desc2', 2, 2, '{"mould_id": 502, "mould_name": "model_b"}', '{}', '{}', 1, '2026-05-07 00:15:00', '2026-05-07 00:45:00', 'FINISH', '{}', 1, 1, 'user2', 'cel2', 'JOB', 0, '{"key2":"val2"}', 'sh', 0, 'token2', '{}', 202, 3, 0, 'STATIC_MODEL', 'BATCH', 'WEB', 1, '{}', 0, 3600, 1, NULL, 0, 0, 0, 0, 'svc2', '{}', 0, 0, '{}', 2026050700),
(1003, 100, 'task_gamma', 'desc3', 3, 3, '{"mould_id": 503, "mould_name": "model_c"}', '{}', '{}', 1, '2026-05-07 00:20:00', '2026-05-07 00:50:00', 'FAILED', '{}', 1, 1, 'user3', 'cel3', 'JOB', 0, '{"key3":"val3"}', 'gz', 0, 'token3', '{}', 203, 3, 0, 'STATIC_MODEL', 'BATCH', 'SDK', 1, '{}', 0, 3600, 1, NULL, 0, 0, 0, 0, 'svc3', '{}', 0, 0, '{}', 2026050700);

-- 诱饵 (main-table filter decoys): each passes ALL other WHERE conditions but
-- fails exactly ONE filter, so the correct GT removes it (baseline unchanged at
-- 58 rows). If that single filter is dropped, the decoy leaks into the output.
-- id=1006: base='OTHER' (only the base='JOB' filter removes it)
-- id=1007: model_type='DYNAMIC_MODEL' (only the model_type filter removes it)
INSERT INTO aide_offline_inference_info_fixed_mysql_003 VALUES
(1006, 100, 'task_decoy_base', 'descD', 1, 1, '{"mould_id": 501, "mould_name": "model_a"}', '{}', '{}', 1, '2026-05-07 00:30:00', '2026-05-07 00:31:00', 'RUNNING', '{}', 1, 1, 'user1', 'cel1', 'OTHER', 0, '{"key":"val"}', 'bj', 0, 'token1', '{}', 201, 3, 0, 'STATIC_MODEL', 'NORMAL', 'API', 1, '{}', 0, 3600, 1, NULL, 0, 0, 0, 0, 'svc1', '{}', 0, 0, '{}', 2026050700),
(1007, 100, 'task_decoy_mt', 'descD', 1, 1, '{"mould_id": 501, "mould_name": "model_a"}', '{}', '{}', 1, '2026-05-07 00:30:00', '2026-05-07 00:31:00', 'RUNNING', '{}', 1, 1, 'user1', 'cel1', 'JOB', 0, '{"key":"val"}', 'bj', 0, 'token1', '{}', 201, 3, 0, 'DYNAMIC_MODEL', 'NORMAL', 'API', 1, '{}', 0, 3600, 1, NULL, 0, 0, 0, 0, 'svc1', '{}', 0, 0, '{}', 2026050700);


-- Input Table 2: aide_offline_inference_dataset_info_h
CREATE TABLE IF NOT EXISTS aide_offline_inference_dataset_info_h_mysql_003 (
    id INT,
    file_id INT,
    offline_inference_id INT,
    stage_id INT,
    result_file_name VARCHAR(256),
    result_file_path VARCHAR(256),
    inference_result_data_url VARCHAR(256),
    dataset_act_num BIGINT,
    dataset_fin_num BIGINT,
    create_time VARCHAR(256),
    update_time VARCHAR(256),
    dataset_err_num BIGINT,
    inference_failed_result_data_url VARCHAR(256),
    dataset_exp_num BIGINT,
    remaining_time VARCHAR(256),
    output_file_id INT,
    error_file_id INT,
    inference_unprocessed_data_url VARCHAR(256),
    hifs_result_file_path VARCHAR(256),
    task_waiting_info VARCHAR(256),
    etl_stamp VARCHAR(256)
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4;

-- build_inputs
INSERT INTO aide_offline_inference_dataset_info_h_mysql_003 VALUES
(1, 201, 1001, 1, 'res1.csv', '/path/res1', 'http://url1', 1000, 950, '2026-05-07 06:30:00', '2026-05-07 06:55:00', 10, '', 1000, '5m', 1, 1, '', '', '', ''),
(2, 202, 1002, 2, 'res2.csv', '/path/res2', 'http://url2', 500, 500, '2026-05-07 05:00:00', '2026-05-07 06:50:00', 0, '', 500, '0m', 2, 2, '', '', '', ''),
(3, 203, 1003, 3, 'res3.csv', '/path/res3', 'http://url3', 200, 100, '2026-05-07 06:10:00', '2026-05-07 06:40:00', 50, '', 200, '10m', 3, 3, '', '', '', '');

-- AC_rerun insert: dataset for 3001/3002 stage
INSERT INTO aide_offline_inference_dataset_info_h_mysql_003 VALUES
(4, 2001, 1001, 3001, 'result1.csv', '/path/r1', 'http://url1', 1000, 900, '2026-05-07 00:11:00', '2026-05-07 00:20:00', 50, 'http://err1', 1000, '10min', 10, 11, 'http://unp1', '/hifs/r1', '', ''),
(5, 2002, 1002, 3002, 'result2.csv', '/path/r2', 'http://url2', 500, 480, '2026-05-06 23:51:00', '2026-05-06 23:55:00', 20, 'http://err2', 500, '5min', 20, 21, 'http://unp2', '/hifs/r2', '', '');

-- landing insert: dataset for 301/302/303 stages with small numbers
INSERT INTO aide_offline_inference_dataset_info_h_mysql_003 VALUES
(6, 201, 1001, 1, 'result1.csv', '/path/r1', 'http://url1', 100, 80, '2026-05-07 00:12:00', '2026-05-07 00:30:00', 5, 'http://err1', 100, '10min', 301, 401, 'http://unp1', '/hifs/r1', '{}', 'stamp1'),
(7, 202, 1002, 2, 'result2.csv', '/path/r2', 'http://url2', 200, 200, '2026-05-07 00:16:00', '2026-05-07 00:45:00', 0, 'http://err2', 200, '0min', 302, 402, 'http://unp2', '/hifs/r2', '{}', 'stamp2'),
(8, 203, 1003, 3, 'result3.csv', '/path/r3', 'http://url3', 150, 50, '2026-05-07 00:22:00', '2026-05-07 00:50:00', 30, 'http://err3', 150, '20min', 303, 403, 'http://unp3', '/hifs/r3', '{}', 'stamp3');


-- Input Table 3: aide_offline_inference_stage_info_h
CREATE TABLE IF NOT EXISTS aide_offline_inference_stage_info_h_mysql_003 (
    id INT,
    offline_inference_id INT,
    mould_id INT,
    deployment_id INT,
    scaleup_id INT,
    mq_info_id INT,
    dataset_exp_num INT,
    dataset_act_num INT,
    dataset_fin_num INT,
    status VARCHAR(256),
    create_time VARCHAR(256),
    update_time VARCHAR(256),
    modifier_id INT,
    error_msg VARCHAR(256),
    etl_stamp VARCHAR(256)
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4;

-- build_inputs
INSERT INTO aide_offline_inference_stage_info_h_mysql_003 VALUES
(301, 1001, 501, 1, 1, 1, 1000, 1000, 950, 'RUNNING', '2026-05-07 06:30:00', '2026-05-07 06:55:00', 1, '', ''),
(302, 1002, 502, 2, 2, 2, 500, 500, 500, 'FINISH', '2026-05-07 05:00:00', '2026-05-07 06:50:00', 1, '', ''),
(303, 1003, 503, 3, 3, 3, 200, 200, 100, 'FAILED', '2026-05-07 06:10:00', '2026-05-07 06:40:00', 1, 'error', '');

-- AC_rerun insert: stage 3001/3002
INSERT INTO aide_offline_inference_stage_info_h_mysql_003 VALUES
(3001, 1001, 501, 1, 1, 1, 1000, 1000, 900, 'RUNNING', '2026-05-07 00:11:00', '2026-05-07 00:20:00', 0, '', ''),
(3002, 1002, 502, 2, 2, 2, 500, 500, 480, 'FINISH', '2026-05-06 23:51:00', '2026-05-06 23:55:00', 0, '', '');

-- landing insert: stage 5001/5002/5003
INSERT INTO aide_offline_inference_stage_info_h_mysql_003 VALUES
(5001, 1001, 501, 1, 1, 1, 100, 100, 80, 'RUNNING', '2026-05-07 00:11:00', '2026-05-07 00:30:00', 1, NULL, 'stamp1'),
(5002, 1002, 502, 2, 2, 2, 200, 200, 200, 'FINISH', '2026-05-07 00:16:00', '2026-05-07 00:45:00', 1, NULL, 'stamp2'),
(5003, 1003, 503, 3, 3, 3, 150, 150, 50, 'FAILED', '2026-05-07 00:21:00', '2026-05-07 00:50:00', 1, 'timeout', 'stamp3');


-- Input Table 4: aide_offline_inference_pipeline_time
CREATE TABLE IF NOT EXISTS aide_offline_inference_pipeline_time_mysql_003 (
    id BIGINT,
    task_id INT,
    stage_id INT,
    step INT,
    step_desc VARCHAR(256),
    start_time VARCHAR(256),
    end_time VARCHAR(256),
    status VARCHAR(256),
    time_cost_in_second BIGINT,
    etl_stamp VARCHAR(256)
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4;

-- build_inputs
INSERT INTO aide_offline_inference_pipeline_time_mysql_003 VALUES
(1, 1001, 301, 1, '任务下发', '2026-05-07 06:30:00', '2026-05-07 06:32:00', 'DONE', 120, ''),
(2, 1001, 301, 2, '离线推理', '2026-05-07 06:32:00', '2026-05-07 06:50:00', 'DONE', 1080, ''),
(3, 1002, 302, 1, '任务下发', '2026-05-07 05:00:00', '2026-05-07 05:01:00', 'DONE', 60, ''),
(4, 1002, 302, 2, '离线推理', '2026-05-07 05:01:00', '2026-05-07 06:30:00', 'DONE', 5340, ''),
(5, 1003, 303, 1, '任务下发', '2026-05-07 06:10:00', '2026-05-07 06:11:00', 'DONE', 60, ''),
(6, 1003, 303, 2, '离线推理', '2026-05-07 06:11:00', NULL, 'RUNNING', 0, '');

-- 诱饵 (pipeline end_time IS NOT NULL filter): an unfinished step (end_time NULL)
-- with a NON-ZERO time_cost for an existing task_id. Correct GT excludes it
-- (SUM for task 1001 stays 1200); if the end_time filter is dropped it leaks into
-- the SUM (1200 -> 1700), so the rule becomes detectable. Baseline unchanged.
INSERT INTO aide_offline_inference_pipeline_time_mysql_003 VALUES
(7, 1001, 301, 3, '离线推理', '2026-05-07 06:50:00', NULL, 'RUNNING', 500, '');

-- NOTE: AC_rerun pipeline_time INSERTs are skipped because they did not execute
-- successfully in the original Hive (wrong table names / ID conflicts).
-- The expected.csv shows only build_inputs pipeline data in the aggregation results.


-- Input Table 5: aide_offline_inference_file_info_h
CREATE TABLE IF NOT EXISTS aide_offline_inference_file_info_h_mysql_003 (
    create_time VARCHAR(256),
    file_download_url VARCHAR(256),
    file_name VARCHAR(256),
    file_path VARCHAR(256),
    file_type VARCHAR(256),
    id INT,
    operator VARCHAR(256),
    scan_result VARCHAR(256),
    scan_status VARCHAR(256),
    storage_conf VARCHAR(256),
    storage_type VARCHAR(256),
    update_time VARCHAR(256),
    dt VARCHAR(256),
    etl_stamp VARCHAR(256)
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4;

-- build_inputs
INSERT INTO aide_offline_inference_file_info_h_mysql_003 VALUES
('2026-05-07 06:25:00', 'http://dl1', 'file1.csv', '/path/f1', 'CSV', 201, 'submitter_A', '', '', '', 'COS', '2026-05-07 06:30:00', '2026050700', ''),
('2026-05-07 04:55:00', 'http://dl2', 'file2.csv', '/path/f2', 'CSV', 202, 'submitter_B', '', '', '', 'COS', '2026-05-07 05:00:00', '2026050700', ''),
('2026-05-07 06:05:00', 'http://dl3', 'file3.csv', '/path/f3', 'CSV', 203, 'submitter_C', '', '', '', 'COS', '2026-05-07 06:10:00', '2026050700', '');

-- AC_rerun insert: file 2001/2002 (user_zhang/user_li)
INSERT INTO aide_offline_inference_file_info_h_mysql_003 VALUES
('2026-05-06 23:00:00', 'http://dl1', 'input1.csv', '/path/input1', 'csv', 2001, 'user_zhang', '', 'PASS', '', 'COS', '2026-05-07 00:05:00', '2026050700', ''),
('2026-05-06 22:00:00', 'http://dl2', 'input2.csv', '/path/input2', 'csv', 2002, 'user_li', '', 'PASS', '', 'COS', '2026-05-06 22:30:00', '2026050700', '');

-- landing insert: file 201/202/203 (submitter_a/b/c)
INSERT INTO aide_offline_inference_file_info_h_mysql_003 VALUES
('2026-05-07 00:09:00', 'http://dl1', 'input1.jsonl', '/path/f1', 'jsonl', 201, 'submitter_a', 'OK', 'PASS', '{}', 'cos', '2026-05-07 00:09:00', '2026050700', 'stamp1'),
('2026-05-07 00:14:00', 'http://dl2', 'input2.jsonl', '/path/f2', 'jsonl', 202, 'submitter_b', 'OK', 'PASS', '{}', 'cos', '2026-05-07 00:14:00', '2026050700', 'stamp2'),
('2026-05-07 00:19:00', 'http://dl3', 'input3.jsonl', '/path/f3', 'jsonl', 203, 'submitter_c', 'OK', 'PASS', '{}', 'cos', '2026-05-07 00:19:00', '2026050700', 'stamp3');

-- 诱饵 (file current-hour dt filter): a row in a DIFFERENT partition (dt='2026050623')
-- for an existing file id=2001 with a LATER update_time than the current-partition
-- row. Correct GT filters by dt='2026050700' and never sees it (baseline unchanged).
-- If the dt filter is dropped, ROW_NUMBER (ORDER BY update_time DESC) would pick this
-- decoy, changing submit_operator/submit_time — so the rule becomes detectable.
INSERT INTO aide_offline_inference_file_info_h_mysql_003 VALUES
('2026-05-06 23:00:00', 'http://dlX', 'inputX.csv', '/path/inputX', 'csv', 2001, 'decoy_operator', '', 'PASS', '', 'COS', '2026-05-07 09:00:00', '2026050623', '');


-- Input Table 6: aide_mould_h
-- NOTE: This table has ~90 columns. Many VARCHAR columns would exceed MySQL's
-- 65535-byte row size limit under utf8mb4. Non-key, non-filter columns are changed to TEXT.
CREATE TABLE IF NOT EXISTS aide_mould_h_mysql_003 (
    id INT,
    create_time VARCHAR(64),
    update_time VARCHAR(64),
    name VARCHAR(128),
    `desc` TEXT,
    path TEXT,
    checkpoint TEXT,
    `enum` VARCHAR(64),
    source VARCHAR(64),
    job_id INT,
    modifier_id INT,
    dep TEXT,
    scale TEXT,
    total_parameters TEXT,
    activate_parameters TEXT,
    stage VARCHAR(64),
    style VARCHAR(64),
    wsid INT,
    pub INT,
    category VARCHAR(64),
    queue_name VARCHAR(128),
    art INT,
    spaces TEXT,
    spaces_json TEXT,
    template TEXT,
    ch_location TEXT,
    location_infos TEXT,
    flows TEXT,
    amount TEXT,
    context_len TEXT,
    file TEXT,
    remark1 TEXT,
    remark2 TEXT,
    ver_style VARCHAR(64),
    disable INT,
    old_path TEXT,
    tokenizer TEXT,
    pattern VARCHAR(64),
    rlhf_pattern VARCHAR(64),
    online INT,
    source_path TEXT,
    scene_text_type VARCHAR(64),
    scene_type VARCHAR(64),
    wwfs_path TEXT,
    compression_strategy VARCHAR(64),
    compression_path TEXT,
    vit_input_resolution VARCHAR(64),
    relate_base_mould TEXT,
    relate_base_id INT,
    relate_base_enum VARCHAR(64),
    father_mould TEXT,
    visual_structure TEXT,
    only_save_lora INT,
    cos_bucket TEXT,
    bucket_region VARCHAR(64),
    old_compression_path TEXT,
    disable_execute INT,
    task_instance_name TEXT,
    train_task_name TEXT,
    train_mode VARCHAR(64),
    sft_tokenizer TEXT,
    model_structure VARCHAR(64),
    benchmark_id VARCHAR(64),
    scene_train VARCHAR(64),
    image_model_id VARCHAR(64),
    trigger_prompt TEXT,
    image_tag VARCHAR(64),
    model_type VARCHAR(64),
    from_inference_id VARCHAR(64),
    from_inference_data TEXT,
    hf_model_path TEXT,
    hf_version INT,
    config_content TEXT,
    audio_encoder_path TEXT,
    audio_decode_before_path TEXT,
    audio_decode_after_path TEXT,
    audio_llm_path TEXT,
    extra_scene_text_type VARCHAR(64),
    copy_state VARCHAR(64),
    copy_tips TEXT,
    use_audio_embeddings INT,
    open_source_type VARCHAR(64),
    icon TEXT,
    manufacturer VARCHAR(64),
    disable_deploy INT,
    show_platform VARCHAR(64),
    extra_info TEXT,
    is_delete INT,
    manufacturer_series VARCHAR(64),
    storage_type VARCHAR(64),
    nas_server TEXT,
    thought_type VARCHAR(64),
    model_source VARCHAR(64),
    deleted INT,
    copy_from_location_id INT,
    etl_stamp VARCHAR(64)
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4;

-- build_inputs: model_A/B/C with img_tag_A/B/C
INSERT INTO aide_mould_h_mysql_003 VALUES
(501, '2026-01-01 00:00:00', '2026-01-01 00:00:00', 'model_A', '', '', '', '', '', 0, 0, '', '', '', '', '', '', 100, 0, '', '', 0, '', '', '', '', '', '', '', '', '', '', '', '', 0, '', '', '', '', 0, '', '', '', '', '', '', '', '', 0, '', '', '', 0, '', '', '', 0, '', '', '', '', '', '', '', '', '', 'img_tag_A', '', '', '', '', 0, '', '', '', '', '', '', '', '', 0, '', '', '', 0, '', '', 0, '', '', '', '', '', 0, 0, ''),
(502, '2026-01-01 00:00:00', '2026-01-01 00:00:00', 'model_B', '', '', '', '', '', 0, 0, '', '', '', '', '', '', 100, 0, '', '', 0, '', '', '', '', '', '', '', '', '', '', '', '', 0, '', '', '', '', 0, '', '', '', '', '', '', '', '', 0, '', '', '', 0, '', '', '', 0, '', '', '', '', '', '', '', '', '', 'img_tag_B', '', '', '', '', 0, '', '', '', '', '', '', '', '', 0, '', '', '', 0, '', '', 0, '', '', '', '', '', 0, 0, ''),
(503, '2026-01-01 00:00:00', '2026-01-01 00:00:00', 'model_C', '', '', '', '', '', 0, 0, '', '', '', '', '', '', 100, 0, '', '', 0, '', '', '', '', '', '', '', '', '', '', '', '', 0, '', '', '', '', 0, '', '', '', '', '', '', '', '', 0, '', '', '', 0, '', '', '', 0, '', '', '', '', '', '', '', '', '', 'img_tag_C', '', '', '', '', 0, '', '', '', '', '', '', '', '', 0, '', '', '', 0, '', '', 0, '', '', '', '', '', 0, 0, '');

-- AC_rerun insert: model_x with image_tag=v1.0-alpha
INSERT INTO aide_mould_h_mysql_003 VALUES
(501, '2026-01-01 00:00:00', '2026-05-01 00:00:00', 'model_x', '', '', '', '', '', 0, 0, '', '', '', '', '', '', 100, 0, '', '', 0, '', '', '', '', '', '', '', '', '', '', '', '', 0, '', '', '', '', 0, '', '', '', '', '', '', '', '', 0, '', '', '', 0, '', '', '', 0, '', '', '', '', '', '', '', '', '', 'v1.0-alpha', '', '', '', '', 0, '', '', '', '', '', '', '', '', 0, '', '', '', 0, '', '', 0, '', '', '', '', '', 0, 0, '');

-- NOTE: landing inserts for aide_mould_h were skipped because the original Hive SQL
-- had column count mismatches that caused those INSERTs to fail silently.

-- Input Table 7: task_instance_wait_time_stats
CREATE TABLE IF NOT EXISTS task_instance_wait_time_stats_mysql_003 (
    dt VARCHAR(256),
    instance_uuid VARCHAR(256),
    task_instance_id VARCHAR(256),
    state_from VARCHAR(256),
    state_to VARCHAR(256),
    time_start DOUBLE,
    time_end DOUBLE,
    first_wait_time DOUBLE,
    wait_time_status VARCHAR(256),
    create_time DOUBLE,
    state VARCHAR(256),
    host_num DOUBLE,
    host_gpu_num DOUBLE,
    task_id VARCHAR(256),
    is_elasticity SMALLINT,
    is_mixing_task SMALLINT,
    is_exist_cluster SMALLINT,
    quota_type VARCHAR(256),
    location VARCHAR(256),
    business_flag VARCHAR(256),
    gpuname VARCHAR(256),
    account VARCHAR(256),
    business_tag VARCHAR(256),
    gpu_num DOUBLE,
    service_id BIGINT,
    service_name VARCHAR(256)
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4;

-- build_inputs: task_alpha_301_OFFLINE (60,120), task_beta_302_OFFLINE (80)
INSERT INTO task_instance_wait_time_stats_mysql_003 VALUES
('20260507', 'uuid1', 'ti1', 'READY', 'RUNNING', 1746582600.0, 1746582660.0, 60.0, 'OK', 1746582600.0, 'RUNNING', 1.0, 8.0, 't1', 0, 0, 1, 'dedicated', 'sh', 'nextgen_platform', 'A100', 'user1', 'tag1', 8.0, 1, 'task_alpha_301_OFFLINE'),
('20260507', 'uuid2', 'ti2', 'READY', 'RUNNING', 1746582600.0, 1746582720.0, 120.0, 'OK', 1746582600.0, 'RUNNING', 1.0, 8.0, 't2', 0, 0, 1, 'dedicated', 'sh', 'nextgen_platform', 'A100', 'user1', 'tag1', 8.0, 1, 'task_alpha_301_OFFLINE'),
('20260507', 'uuid3', 'ti3', 'READY', 'RUNNING', 1746582600.0, 1746582680.0, 80.0, 'OK', 1746582600.0, 'RUNNING', 1.0, 8.0, 't3', 0, 0, 1, 'dedicated', 'bj', 'nextgen_platform', 'A100', 'user2', 'tag2', 8.0, 2, 'task_beta_302_OFFLINE');

-- AC_rerun insert: task_alpha_3001_OFFLINE (60), task_beta_3002_OFFLINE (120)
INSERT INTO task_instance_wait_time_stats_mysql_003 VALUES
('20260507', 'uuid-001', 'ti-001', 'READY', 'RUNNING', 1746576600.0, 1746576660.0, 60.0, 'OK', 1746576500.0, 'RUNNING', 1.0, 8.0, 'task-001', 0, 0, 1, 'dedicated', 'shanghai', 'teg', 'A100', 'user1', 'ai', 8.0, 100, 'task_alpha_3001_OFFLINE'),
('20260507', 'uuid-002', 'ti-002', 'READY', 'RUNNING', 1746576700.0, 1746576820.0, 120.0, 'OK', 1746576600.0, 'RUNNING', 1.0, 4.0, 'task-002', 0, 0, 1, 'dedicated', 'beijing', 'teg', 'V100', 'user2', 'ai', 4.0, 200, 'task_beta_3002_OFFLINE');

-- landing insert: task_alpha_5001_OFFLINE (300), task_beta_5002_OFFLINE (150), task_gamma_5003_OFFLINE (200)
INSERT INTO task_instance_wait_time_stats_mysql_003 VALUES
('20260507', 'uuid1', 'inst1', 'READY', 'RUNNING', 1746576600.0, 1746576900.0, 300.0, 'OK', 1746576600.0, 'RUNNING', 1.0, 8.0, 'task1', 0, 0, 1, 'dedicated', 'bj', 'inference', 'A100', 'user1', 'tag1', 8.0, 1, 'task_alpha_5001_OFFLINE'),
('20260507', 'uuid2', 'inst2', 'READY', 'RUNNING', 1746576700.0, 1746576850.0, 150.0, 'OK', 1746576700.0, 'RUNNING', 1.0, 4.0, 'task2', 0, 0, 1, 'shared', 'sh', 'inference', 'V100', 'user2', 'tag2', 4.0, 2, 'task_beta_5002_OFFLINE'),
('20260507', 'uuid3', 'inst3', 'READY', 'RUNNING', 1746576800.0, 1746577000.0, 200.0, 'OK', 1746576800.0, 'RUNNING', 1.0, 8.0, 'task3', 0, 0, 1, 'dedicated', 'gz', 'inference', 'A100', 'user3', 'tag3', 8.0, 3, 'task_gamma_5003_OFFLINE');

-- 诱饵 (latest-partition MAX(dt) pick): a STALE partition (dt='20260506') row for
-- an existing service with a wildly different first_wait_time. Correct GT uses
-- MAX(dt)='20260507' and ignores it (baseline unchanged). If the partition pick is
-- broken (e.g. MIN(dt)), this stale value leaks in and the wait stats diverge.
INSERT INTO task_instance_wait_time_stats_mysql_003 VALUES
('20260506', 'uuidS', 'instS', 'READY', 'RUNNING', 1746490000.0, 1746499999.0, 9999.0, 'OK', 1746490000.0, 'RUNNING', 1.0, 8.0, 'taskS', 0, 0, 1, 'dedicated', 'bj', 'inference', 'A100', 'user1', 'tag1', 8.0, 1, 'task_alpha_301_OFFLINE');


-- Output Table: dwd_aide_offline_inference_feature_cand
CREATE TABLE IF NOT EXISTS dwd_aide_offline_inference_feature_cand_mysql_003 (
    dt VARCHAR(256) COMMENT '分区字段',
    create_time VARCHAR(256) COMMENT '创建时间',
    ct INT COMMENT '计数标识',
    id BIGINT COMMENT '离线推理任务ID',
    name VARCHAR(256) COMMENT '任务名称',
    servingName VARCHAR(256) COMMENT '服务名称',
    base VARCHAR(256) COMMENT '基础类型',
    status VARCHAR(256) COMMENT '状态',
    submit_type VARCHAR(256) COMMENT '提交类型',
    mould_id VARCHAR(256) COMMENT '模型ID',
    mould_name VARCHAR(256) COMMENT '模型名称',
    time_cost_in_second BIGINT COMMENT '总耗时(秒)',
    dataset_act_num BIGINT COMMENT '数据集激活数量',
    dataset_fin_num BIGINT COMMENT '数据集完成数量',
    dataset_err_num BIGINT COMMENT '数据集错误数量',
    max_instance_first_wait_time DOUBLE COMMENT '最大首次等待时长(秒)',
    avg_instance_first_wait_time DOUBLE COMMENT '平均首次等待时长(秒)',
    task_dispatch_time BIGINT COMMENT '任务下发耗时(秒)',
    offline_inference_time BIGINT COMMENT '离线推理耗时(秒)',
    inference_config VARCHAR(256) COMMENT '推理配置',
    submit_operator VARCHAR(256) COMMENT '提交人',
    submit_time VARCHAR(256) COMMENT '提交时间',
    image_tag VARCHAR(256) COMMENT '镜像标签'
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4;