data/squirrel_mini_tasks.json

[
  {
    "task_id": "ecommerce-syntax-0",
    "type": "syntax",
    "domain": "ecommerce",
    "ddl": "CREATE TABLE IF NOT EXISTS ecommerce_db.order_events (entity_id STRING, event_timestamp TIMESTAMP, dim_0_code STRING, dim_1_code STRING, dim_2_code STRING, dim_3_code STRING, dim_4_code STRING, dim_5_code STRING, dim_6_code STRING, dim_7_code STRING, dim_8_code STRING, dim_9_code STRING, metric_0_amount DOUBLE, metric_1_amount DOUBLE, metric_2_amount DOUBLE, metric_3_amount DOUBLE, metric_4_amount DOUBLE, metric_5_amount DOUBLE, metric_6_amount DOUBLE, metric_7_amount DOUBLE, metric_8_amount DOUBLE, metric_9_amount DOUBLE, metric_10_amount DOUBLE, metric_11_amount DOUBLE, metric_12_amount DOUBLE, metric_13_amount DOUBLE, status_flag STRING, event_date STRING);\nCREATE TABLE IF NOT EXISTS ecommerce_db.customer_profile (entity_id STRING, segment_code STRING, region_code STRING, tier_level STRING);\nCREATE TABLE IF NOT EXISTS ecommerce_db.product_catalog (dim_0_code STRING, category_code STRING, subcategory_code STRING);\nCREATE TABLE IF NOT EXISTS ecommerce_db.legacy_order_events_snapshot (segment_code STRING, region_code STRING, category_code STRING, entity_count BIGINT, total_metric_0_amount DOUBLE, snapshot_date STRING);",
    "issue_sql": "-- ETL pipeline for ecommerce daily aggregation\nWITH base_events AS (\n    SELECT\n        order_events.entity_id,\n        order_events.event_timestamp,\n        order_events.dim_0_code,\n        order_events.dim_1_code,\n        order_events.dim_2_code,\n        order_events.dim_3_code,\n        order_events.dim_4_code,\n        order_events.metric_0_amount,\n        order_events.metric_1_amount,\n        order_events.metric_2_amount,\n        order_events.metric_3_amount,\n        order_events.metric_4_amount,\n        order_events.metric_5_amount,\n        order_events.metric_6_amount,\n        order_events.status_flag\n    FROM ecommerce_db.order_events order_events\n    WHERE order_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        customer_profile.segment_code,\n        customer_profile.region_code,\n        customer_profile.tier_level,\n        product_catalog.category_code,\n        product_catalog.subcategory_code\n    FROM base_events\n    JOIN ecommerce_db.customer_profile customer_profile\n        ON base_events.entity_id = customer_profile.entity_id\n    LEFT JOIN ecommerce_db.product_catalog product_catalog\n        ON base_events.dim_0_code = product_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM ecommerce_db.legacy_order_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.ecommerce_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "reference_sql": "-- ETL pipeline for ecommerce daily aggregation\nWITH base_events AS (\n    SELECT\n        order_events.entity_id,\n        order_events.event_timestamp,\n        order_events.dim_0_code,\n        order_events.dim_1_code,\n        order_events.dim_2_code,\n        order_events.dim_3_code,\n        order_events.dim_4_code,\n        order_events.metric_0_amount,\n        order_events.metric_1_amount,\n        order_events.metric_2_amount,\n        order_events.metric_3_amount,\n        order_events.metric_4_amount,\n        order_events.metric_5_amount,\n        order_events.metric_6_amount,\n        order_events.status_flag\n    FROM ecommerce_db.order_events order_events\n    WHERE order_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        customer_profile.segment_code,\n        customer_profile.region_code,\n        customer_profile.tier_level,\n        product_catalog.category_code,\n        product_catalog.subcategory_code\n    FROM base_events\n    JOIN ecommerce_db.customer_profile customer_profile\n        ON base_events.entity_id = customer_profile.entity_id\n    LEFT JOIN ecommerce_db.product_catalog product_catalog\n        ON base_events.dim_0_code = product_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM ecommerce_db.legacy_order_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.ecommerce_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "error_message": "ParseException: mismatched input 'AS' expecting 'END' near 'value_tier'",
    "level1_error_type": "Grammar & Structure",
    "level2_error_type": "CASE Expression",
    "level3_error_type": "Missing END in CASE WHEN"
  },
  {
    "task_id": "ecommerce-semantic-0",
    "type": "semantic",
    "domain": "ecommerce",
    "ddl": "CREATE TABLE IF NOT EXISTS ecommerce_db.order_events (entity_id STRING, event_timestamp TIMESTAMP, dim_0_code STRING, dim_1_code STRING, dim_2_code STRING, dim_3_code STRING, dim_4_code STRING, dim_5_code STRING, dim_6_code STRING, dim_7_code STRING, dim_8_code STRING, dim_9_code STRING, metric_0_amount DOUBLE, metric_1_amount DOUBLE, metric_2_amount DOUBLE, metric_3_amount DOUBLE, metric_4_amount DOUBLE, metric_5_amount DOUBLE, metric_6_amount DOUBLE, metric_7_amount DOUBLE, metric_8_amount DOUBLE, metric_9_amount DOUBLE, metric_10_amount DOUBLE, metric_11_amount DOUBLE, metric_12_amount DOUBLE, metric_13_amount DOUBLE, status_flag STRING, event_date STRING);\nCREATE TABLE IF NOT EXISTS ecommerce_db.customer_profile (entity_id STRING, segment_code STRING, region_code STRING, tier_level STRING);\nCREATE TABLE IF NOT EXISTS ecommerce_db.product_catalog (dim_0_code STRING, category_code STRING, subcategory_code STRING);\nCREATE TABLE IF NOT EXISTS ecommerce_db.legacy_order_events_snapshot (segment_code STRING, region_code STRING, category_code STRING, entity_count BIGINT, total_metric_0_amount DOUBLE, snapshot_date STRING);",
    "issue_sql": "-- ETL pipeline for ecommerce daily aggregation\nWITH base_events AS (\n    SELECT\n        order_events.entity_id,\n        order_events.event_timestamp,\n        order_events.dim_0_code,\n        order_events.dim_1_code,\n        order_events.dim_2_code,\n        order_events.dim_3_code,\n        order_events.dim_4_code,\n        order_events.metric_0_amount,\n        order_events.metric_1_amount,\n        order_events.metric_2_amount,\n        order_events.metric_3_amount,\n        order_events.metric_4_amount,\n        order_events.metric_5_amount,\n        order_events.metric_6_amount,\n        order_events.status_flag\n    FROM ecommerce_db.order_events order_events\n    WHERE order_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        customer_profile.segment_code,\n        customer_profile.region_code,\n        customer_profile.tier_level,\n        product_catalog.category_code,\n        product_catalog.subcategory_code\n    FROM base_events\n    JOIN ecommerce_db.customer_profile customer_profile\n        ON base_events.entity_id = customer_profile.entity_id\n    JOIN ecommerce_db.product_catalog product_catalog\n        ON base_events.dim_0_code = product_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM ecommerce_db.legacy_order_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.ecommerce_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "reference_sql": "-- ETL pipeline for ecommerce daily aggregation\nWITH base_events AS (\n    SELECT\n        order_events.entity_id,\n        order_events.event_timestamp,\n        order_events.dim_0_code,\n        order_events.dim_1_code,\n        order_events.dim_2_code,\n        order_events.dim_3_code,\n        order_events.dim_4_code,\n        order_events.metric_0_amount,\n        order_events.metric_1_amount,\n        order_events.metric_2_amount,\n        order_events.metric_3_amount,\n        order_events.metric_4_amount,\n        order_events.metric_5_amount,\n        order_events.metric_6_amount,\n        order_events.status_flag\n    FROM ecommerce_db.order_events order_events\n    WHERE order_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        customer_profile.segment_code,\n        customer_profile.region_code,\n        customer_profile.tier_level,\n        product_catalog.category_code,\n        product_catalog.subcategory_code\n    FROM base_events\n    JOIN ecommerce_db.customer_profile customer_profile\n        ON base_events.entity_id = customer_profile.entity_id\n    LEFT JOIN ecommerce_db.product_catalog product_catalog\n        ON base_events.dim_0_code = product_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM ecommerce_db.legacy_order_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.ecommerce_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "user_query": "Some entities that should appear in the daily summary (e.g. those without a matching category record) are missing from the output entirely. Expected: all entities from the base events should be retained even if category enrichment is unavailable. Please fix the bug.",
    "level1_error_type": "Joins & Grouping",
    "level2_error_type": "JOIN Type Selection",
    "level3_error_type": "Using INNER JOIN when LEFT JOIN is needed"
  },
  {
    "task_id": "finance-syntax-1",
    "type": "syntax",
    "domain": "finance",
    "ddl": "CREATE TABLE IF NOT EXISTS finance_db.transaction_ledger (entity_id STRING, event_timestamp TIMESTAMP, dim_0_code STRING, dim_1_code STRING, dim_2_code STRING, dim_3_code STRING, dim_4_code STRING, dim_5_code STRING, dim_6_code STRING, dim_7_code STRING, dim_8_code STRING, dim_9_code STRING, metric_0_amount DOUBLE, metric_1_amount DOUBLE, metric_2_amount DOUBLE, metric_3_amount DOUBLE, metric_4_amount DOUBLE, metric_5_amount DOUBLE, metric_6_amount DOUBLE, metric_7_amount DOUBLE, metric_8_amount DOUBLE, metric_9_amount DOUBLE, metric_10_amount DOUBLE, metric_11_amount DOUBLE, metric_12_amount DOUBLE, metric_13_amount DOUBLE, status_flag STRING, event_date STRING);\nCREATE TABLE IF NOT EXISTS finance_db.account_profile (entity_id STRING, segment_code STRING, region_code STRING, tier_level STRING);\nCREATE TABLE IF NOT EXISTS finance_db.instrument_catalog (dim_0_code STRING, category_code STRING, subcategory_code STRING);\nCREATE TABLE IF NOT EXISTS finance_db.legacy_transaction_ledger_snapshot (segment_code STRING, region_code STRING, category_code STRING, entity_count BIGINT, total_metric_0_amount DOUBLE, snapshot_date STRING);",
    "issue_sql": "-- ETL pipeline for finance daily aggregation\nWITH base_events AS (\n    SELECT\n        transaction_ledger.entity_id,\n        transaction_ledger.event_timestamp,\n        transaction_ledger.dim_0_code,\n        transaction_ledger.dim_1_code,\n        transaction_ledger.dim_2_code,\n        transaction_ledger.dim_3_code,\n        transaction_ledger.dim_4_code,\n        transaction_ledger.metric_0_amount,\n        transaction_ledger.metric_1_amount,\n        transaction_ledger.metric_2_amount,\n        transaction_ledger.metric_3_amount,\n        transaction_ledger.metric_4_amount,\n        transaction_ledger.metric_5_amount,\n        transaction_ledger.metric_6_amount,\n        transaction_ledger.status_flag\n    FROM finance_db.transaction_ledger transaction_ledger\n    WHERE transaction_ledger.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        account_profile.segment_code,\n        account_profile.region_code,\n        account_profile.tier_level,\n        instrument_catalog.category_code,\n        instrument_catalog.subcategory_code\n    FROM base_events\n    JOIN finance_db.account_profile account_profile\n        ON base_events.entity_id = account_profile.entity_id\n    LEFT JOIN finance_db.instrument_catalog instrument_catalog\n        ON base_events.dim_0_code = instrument_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM finance_db.legacy_transaction_ledger_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.finance_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "reference_sql": "-- ETL pipeline for finance daily aggregation\nWITH base_events AS (\n    SELECT\n        transaction_ledger.entity_id,\n        transaction_ledger.event_timestamp,\n        transaction_ledger.dim_0_code,\n        transaction_ledger.dim_1_code,\n        transaction_ledger.dim_2_code,\n        transaction_ledger.dim_3_code,\n        transaction_ledger.dim_4_code,\n        transaction_ledger.metric_0_amount,\n        transaction_ledger.metric_1_amount,\n        transaction_ledger.metric_2_amount,\n        transaction_ledger.metric_3_amount,\n        transaction_ledger.metric_4_amount,\n        transaction_ledger.metric_5_amount,\n        transaction_ledger.metric_6_amount,\n        transaction_ledger.status_flag\n    FROM finance_db.transaction_ledger transaction_ledger\n    WHERE transaction_ledger.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        account_profile.segment_code,\n        account_profile.region_code,\n        account_profile.tier_level,\n        instrument_catalog.category_code,\n        instrument_catalog.subcategory_code\n    FROM base_events\n    JOIN finance_db.account_profile account_profile\n        ON base_events.entity_id = account_profile.entity_id\n    LEFT JOIN finance_db.instrument_catalog instrument_catalog\n        ON base_events.dim_0_code = instrument_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM finance_db.legacy_transaction_ledger_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.finance_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "error_message": "ParseException: mismatched input 'AS' expecting 'END' near 'value_tier'",
    "level1_error_type": "Grammar & Structure",
    "level2_error_type": "CASE Expression",
    "level3_error_type": "Missing END in CASE WHEN"
  },
  {
    "task_id": "finance-semantic-1",
    "type": "semantic",
    "domain": "finance",
    "ddl": "CREATE TABLE IF NOT EXISTS finance_db.transaction_ledger (entity_id STRING, event_timestamp TIMESTAMP, dim_0_code STRING, dim_1_code STRING, dim_2_code STRING, dim_3_code STRING, dim_4_code STRING, dim_5_code STRING, dim_6_code STRING, dim_7_code STRING, dim_8_code STRING, dim_9_code STRING, metric_0_amount DOUBLE, metric_1_amount DOUBLE, metric_2_amount DOUBLE, metric_3_amount DOUBLE, metric_4_amount DOUBLE, metric_5_amount DOUBLE, metric_6_amount DOUBLE, metric_7_amount DOUBLE, metric_8_amount DOUBLE, metric_9_amount DOUBLE, metric_10_amount DOUBLE, metric_11_amount DOUBLE, metric_12_amount DOUBLE, metric_13_amount DOUBLE, status_flag STRING, event_date STRING);\nCREATE TABLE IF NOT EXISTS finance_db.account_profile (entity_id STRING, segment_code STRING, region_code STRING, tier_level STRING);\nCREATE TABLE IF NOT EXISTS finance_db.instrument_catalog (dim_0_code STRING, category_code STRING, subcategory_code STRING);\nCREATE TABLE IF NOT EXISTS finance_db.legacy_transaction_ledger_snapshot (segment_code STRING, region_code STRING, category_code STRING, entity_count BIGINT, total_metric_0_amount DOUBLE, snapshot_date STRING);",
    "issue_sql": "-- ETL pipeline for finance daily aggregation\nWITH base_events AS (\n    SELECT\n        transaction_ledger.entity_id,\n        transaction_ledger.event_timestamp,\n        transaction_ledger.dim_0_code,\n        transaction_ledger.dim_1_code,\n        transaction_ledger.dim_2_code,\n        transaction_ledger.dim_3_code,\n        transaction_ledger.dim_4_code,\n        transaction_ledger.metric_0_amount,\n        transaction_ledger.metric_1_amount,\n        transaction_ledger.metric_2_amount,\n        transaction_ledger.metric_3_amount,\n        transaction_ledger.metric_4_amount,\n        transaction_ledger.metric_5_amount,\n        transaction_ledger.metric_6_amount,\n        transaction_ledger.status_flag\n    FROM finance_db.transaction_ledger transaction_ledger\n    WHERE transaction_ledger.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        account_profile.segment_code,\n        account_profile.region_code,\n        account_profile.tier_level,\n        instrument_catalog.category_code,\n        instrument_catalog.subcategory_code\n    FROM base_events\n    JOIN finance_db.account_profile account_profile\n        ON base_events.entity_id = account_profile.entity_id\n    LEFT JOIN finance_db.instrument_catalog instrument_catalog\n        ON base_events.dim_0_code = instrument_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM finance_db.legacy_transaction_ledger_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.finance_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "reference_sql": "-- ETL pipeline for finance daily aggregation\nWITH base_events AS (\n    SELECT\n        transaction_ledger.entity_id,\n        transaction_ledger.event_timestamp,\n        transaction_ledger.dim_0_code,\n        transaction_ledger.dim_1_code,\n        transaction_ledger.dim_2_code,\n        transaction_ledger.dim_3_code,\n        transaction_ledger.dim_4_code,\n        transaction_ledger.metric_0_amount,\n        transaction_ledger.metric_1_amount,\n        transaction_ledger.metric_2_amount,\n        transaction_ledger.metric_3_amount,\n        transaction_ledger.metric_4_amount,\n        transaction_ledger.metric_5_amount,\n        transaction_ledger.metric_6_amount,\n        transaction_ledger.status_flag\n    FROM finance_db.transaction_ledger transaction_ledger\n    WHERE transaction_ledger.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        account_profile.segment_code,\n        account_profile.region_code,\n        account_profile.tier_level,\n        instrument_catalog.category_code,\n        instrument_catalog.subcategory_code\n    FROM base_events\n    JOIN finance_db.account_profile account_profile\n        ON base_events.entity_id = account_profile.entity_id\n    LEFT JOIN finance_db.instrument_catalog instrument_catalog\n        ON base_events.dim_0_code = instrument_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM finance_db.legacy_transaction_ledger_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.finance_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "user_query": "The daily summary table sometimes has fewer rows than expected when the current period and legacy snapshot happen to produce identical rows for a segment; those duplicate-looking rows are actually distinct facts and should both be kept. Please fix the bug.",
    "level1_error_type": "Semantics & Logic",
    "level2_error_type": "Set Operations",
    "level3_error_type": "UNION vs UNION ALL misuse"
  },
  {
    "task_id": "healthcare-syntax-2",
    "type": "syntax",
    "domain": "healthcare",
    "ddl": "CREATE TABLE IF NOT EXISTS healthcare_db.visit_events (entity_id STRING, event_timestamp TIMESTAMP, dim_0_code STRING, dim_1_code STRING, dim_2_code STRING, dim_3_code STRING, dim_4_code STRING, dim_5_code STRING, dim_6_code STRING, dim_7_code STRING, dim_8_code STRING, dim_9_code STRING, metric_0_amount DOUBLE, metric_1_amount DOUBLE, metric_2_amount DOUBLE, metric_3_amount DOUBLE, metric_4_amount DOUBLE, metric_5_amount DOUBLE, metric_6_amount DOUBLE, metric_7_amount DOUBLE, metric_8_amount DOUBLE, metric_9_amount DOUBLE, metric_10_amount DOUBLE, metric_11_amount DOUBLE, metric_12_amount DOUBLE, metric_13_amount DOUBLE, status_flag STRING, event_date STRING);\nCREATE TABLE IF NOT EXISTS healthcare_db.patient_profile (entity_id STRING, segment_code STRING, region_code STRING, tier_level STRING);\nCREATE TABLE IF NOT EXISTS healthcare_db.procedure_catalog (dim_0_code STRING, category_code STRING, subcategory_code STRING);\nCREATE TABLE IF NOT EXISTS healthcare_db.legacy_visit_events_snapshot (segment_code STRING, region_code STRING, category_code STRING, entity_count BIGINT, total_metric_0_amount DOUBLE, snapshot_date STRING);",
    "issue_sql": "-- ETL pipeline for healthcare daily aggregation\nWITH base_events AS (\n    SELECT\n        visit_events.entity_id,\n        visit_events.event_timestamp,\n        visit_events.dim_0_code,\n        visit_events.dim_1_code,\n        visit_events.dim_2_code,\n        visit_events.dim_3_code,\n        visit_events.dim_4_code,\n        visit_events.metric_0_amount,\n        visit_events.metric_1_amount,\n        visit_events.metric_2_amount,\n        visit_events.metric_3_amount,\n        visit_events.metric_4_amount,\n        visit_events.metric_5_amount,\n        visit_events.metric_6_amount,\n        visit_events.status_flag\n    FROM healthcare_db.visit_events visit_events\n    WHERE visit_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        patient_profile.segment_code,\n        patient_profile.region_code,\n        patient_profile.tier_level,\n        procedure_catalog.category_code,\n        procedure_catalog.subcategory_code\n    FROM base_events\n    JOIN healthcare_db.patient_profile patient_profile\n        ON base_events.entity_id = patient_profile.entity_id\n    LEFT JOIN healthcare_db.procedure_catalog procedure_catalog\n        ON base_events.dim_0_code = procedure_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM healthcare_db.legacy_visit_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.healthcare_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "reference_sql": "-- ETL pipeline for healthcare daily aggregation\nWITH base_events AS (\n    SELECT\n        visit_events.entity_id,\n        visit_events.event_timestamp,\n        visit_events.dim_0_code,\n        visit_events.dim_1_code,\n        visit_events.dim_2_code,\n        visit_events.dim_3_code,\n        visit_events.dim_4_code,\n        visit_events.metric_0_amount,\n        visit_events.metric_1_amount,\n        visit_events.metric_2_amount,\n        visit_events.metric_3_amount,\n        visit_events.metric_4_amount,\n        visit_events.metric_5_amount,\n        visit_events.metric_6_amount,\n        visit_events.status_flag\n    FROM healthcare_db.visit_events visit_events\n    WHERE visit_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        patient_profile.segment_code,\n        patient_profile.region_code,\n        patient_profile.tier_level,\n        procedure_catalog.category_code,\n        procedure_catalog.subcategory_code\n    FROM base_events\n    JOIN healthcare_db.patient_profile patient_profile\n        ON base_events.entity_id = patient_profile.entity_id\n    LEFT JOIN healthcare_db.procedure_catalog procedure_catalog\n        ON base_events.dim_0_code = procedure_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM healthcare_db.legacy_visit_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.healthcare_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "error_message": "ParseException: mismatched input 'AS' expecting 'END' near 'value_tier'",
    "level1_error_type": "Grammar & Structure",
    "level2_error_type": "CASE Expression",
    "level3_error_type": "Missing END in CASE WHEN"
  },
  {
    "task_id": "healthcare-semantic-2",
    "type": "semantic",
    "domain": "healthcare",
    "ddl": "CREATE TABLE IF NOT EXISTS healthcare_db.visit_events (entity_id STRING, event_timestamp TIMESTAMP, dim_0_code STRING, dim_1_code STRING, dim_2_code STRING, dim_3_code STRING, dim_4_code STRING, dim_5_code STRING, dim_6_code STRING, dim_7_code STRING, dim_8_code STRING, dim_9_code STRING, metric_0_amount DOUBLE, metric_1_amount DOUBLE, metric_2_amount DOUBLE, metric_3_amount DOUBLE, metric_4_amount DOUBLE, metric_5_amount DOUBLE, metric_6_amount DOUBLE, metric_7_amount DOUBLE, metric_8_amount DOUBLE, metric_9_amount DOUBLE, metric_10_amount DOUBLE, metric_11_amount DOUBLE, metric_12_amount DOUBLE, metric_13_amount DOUBLE, status_flag STRING, event_date STRING);\nCREATE TABLE IF NOT EXISTS healthcare_db.patient_profile (entity_id STRING, segment_code STRING, region_code STRING, tier_level STRING);\nCREATE TABLE IF NOT EXISTS healthcare_db.procedure_catalog (dim_0_code STRING, category_code STRING, subcategory_code STRING);\nCREATE TABLE IF NOT EXISTS healthcare_db.legacy_visit_events_snapshot (segment_code STRING, region_code STRING, category_code STRING, entity_count BIGINT, total_metric_0_amount DOUBLE, snapshot_date STRING);",
    "issue_sql": "-- ETL pipeline for healthcare daily aggregation\nWITH base_events AS (\n    SELECT\n        visit_events.entity_id,\n        visit_events.event_timestamp,\n        visit_events.dim_0_code,\n        visit_events.dim_1_code,\n        visit_events.dim_2_code,\n        visit_events.dim_3_code,\n        visit_events.dim_4_code,\n        visit_events.metric_0_amount,\n        visit_events.metric_1_amount,\n        visit_events.metric_2_amount,\n        visit_events.metric_3_amount,\n        visit_events.metric_4_amount,\n        visit_events.metric_5_amount,\n        visit_events.metric_6_amount,\n        visit_events.status_flag\n    FROM healthcare_db.visit_events visit_events\n    WHERE visit_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        patient_profile.segment_code,\n        patient_profile.region_code,\n        patient_profile.tier_level,\n        procedure_catalog.category_code,\n        procedure_catalog.subcategory_code\n    FROM base_events\n    JOIN healthcare_db.patient_profile patient_profile\n        ON base_events.dim_0_code = patient_profile.entity_id\n    LEFT JOIN healthcare_db.procedure_catalog procedure_catalog\n        ON base_events.dim_0_code = procedure_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM healthcare_db.legacy_visit_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.healthcare_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "reference_sql": "-- ETL pipeline for healthcare daily aggregation\nWITH base_events AS (\n    SELECT\n        visit_events.entity_id,\n        visit_events.event_timestamp,\n        visit_events.dim_0_code,\n        visit_events.dim_1_code,\n        visit_events.dim_2_code,\n        visit_events.dim_3_code,\n        visit_events.dim_4_code,\n        visit_events.metric_0_amount,\n        visit_events.metric_1_amount,\n        visit_events.metric_2_amount,\n        visit_events.metric_3_amount,\n        visit_events.metric_4_amount,\n        visit_events.metric_5_amount,\n        visit_events.metric_6_amount,\n        visit_events.status_flag\n    FROM healthcare_db.visit_events visit_events\n    WHERE visit_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        patient_profile.segment_code,\n        patient_profile.region_code,\n        patient_profile.tier_level,\n        procedure_catalog.category_code,\n        procedure_catalog.subcategory_code\n    FROM base_events\n    JOIN healthcare_db.patient_profile patient_profile\n        ON base_events.entity_id = patient_profile.entity_id\n    LEFT JOIN healthcare_db.procedure_catalog procedure_catalog\n        ON base_events.dim_0_code = procedure_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM healthcare_db.legacy_visit_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.healthcare_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "user_query": "The join between base events and the profile dimension looks wrong: segment_code and region_code values in the output don't correspond to the correct entity anymore, producing mismatched enrichment. Please fix the bug so profile attributes are joined on the correct key.",
    "level1_error_type": "Joins & Grouping",
    "level2_error_type": "JOIN Logic",
    "level3_error_type": "Wrong join key used"
  },
  {
    "task_id": "logistics-syntax-3",
    "type": "syntax",
    "domain": "logistics",
    "ddl": "CREATE TABLE IF NOT EXISTS logistics_db.shipment_events (entity_id STRING, event_timestamp TIMESTAMP, dim_0_code STRING, dim_1_code STRING, dim_2_code STRING, dim_3_code STRING, dim_4_code STRING, dim_5_code STRING, dim_6_code STRING, dim_7_code STRING, dim_8_code STRING, dim_9_code STRING, metric_0_amount DOUBLE, metric_1_amount DOUBLE, metric_2_amount DOUBLE, metric_3_amount DOUBLE, metric_4_amount DOUBLE, metric_5_amount DOUBLE, metric_6_amount DOUBLE, metric_7_amount DOUBLE, metric_8_amount DOUBLE, metric_9_amount DOUBLE, metric_10_amount DOUBLE, metric_11_amount DOUBLE, metric_12_amount DOUBLE, metric_13_amount DOUBLE, status_flag STRING, event_date STRING);\nCREATE TABLE IF NOT EXISTS logistics_db.carrier_profile (entity_id STRING, segment_code STRING, region_code STRING, tier_level STRING);\nCREATE TABLE IF NOT EXISTS logistics_db.route_catalog (dim_0_code STRING, category_code STRING, subcategory_code STRING);\nCREATE TABLE IF NOT EXISTS logistics_db.legacy_shipment_events_snapshot (segment_code STRING, region_code STRING, category_code STRING, entity_count BIGINT, total_metric_0_amount DOUBLE, snapshot_date STRING);",
    "issue_sql": "-- ETL pipeline for logistics daily aggregation\nWITH base_events AS (\n    SELECT\n        shipment_events.entity_id,\n        shipment_events.event_timestamp,\n        shipment_events.dim_0_code,\n        shipment_events.dim_1_code,\n        shipment_events.dim_2_code,\n        shipment_events.dim_3_code,\n        shipment_events.dim_4_code,\n        shipment_events.metric_0_amount,\n        shipment_events.metric_1_amount,\n        shipment_events.metric_2_amount,\n        shipment_events.metric_3_amount,\n        shipment_events.metric_4_amount,\n        shipment_events.metric_5_amount,\n        shipment_events.metric_6_amount,\n        shipment_events.status_flag\n    FROM logistics_db.shipment_events shipment_events\n    WHERE shipment_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        carrier_profile.segment_code,\n        carrier_profile.region_code,\n        carrier_profile.tier_level,\n        route_catalog.category_code,\n        route_catalog.subcategory_code\n    FROM base_events\n    JOIN logistics_db.carrier_profile carrier_profile\n        ON base_events.entity_id = carrier_profile.entity_id\n    LEFT JOIN logistics_db.route_catalog route_catalog\n        ON base_events.dim_0_code = route_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM logistics_db.legacy_shipment_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.logistics_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "reference_sql": "-- ETL pipeline for logistics daily aggregation\nWITH base_events AS (\n    SELECT\n        shipment_events.entity_id,\n        shipment_events.event_timestamp,\n        shipment_events.dim_0_code,\n        shipment_events.dim_1_code,\n        shipment_events.dim_2_code,\n        shipment_events.dim_3_code,\n        shipment_events.dim_4_code,\n        shipment_events.metric_0_amount,\n        shipment_events.metric_1_amount,\n        shipment_events.metric_2_amount,\n        shipment_events.metric_3_amount,\n        shipment_events.metric_4_amount,\n        shipment_events.metric_5_amount,\n        shipment_events.metric_6_amount,\n        shipment_events.status_flag\n    FROM logistics_db.shipment_events shipment_events\n    WHERE shipment_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        carrier_profile.segment_code,\n        carrier_profile.region_code,\n        carrier_profile.tier_level,\n        route_catalog.category_code,\n        route_catalog.subcategory_code\n    FROM base_events\n    JOIN logistics_db.carrier_profile carrier_profile\n        ON base_events.entity_id = carrier_profile.entity_id\n    LEFT JOIN logistics_db.route_catalog route_catalog\n        ON base_events.dim_0_code = route_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM logistics_db.legacy_shipment_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.logistics_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "error_message": "ParseException: mismatched input 'AS' expecting 'END' near 'value_tier'",
    "level1_error_type": "Grammar & Structure",
    "level2_error_type": "CASE Expression",
    "level3_error_type": "Missing END in CASE WHEN"
  },
  {
    "task_id": "logistics-semantic-3",
    "type": "semantic",
    "domain": "logistics",
    "ddl": "CREATE TABLE IF NOT EXISTS logistics_db.shipment_events (entity_id STRING, event_timestamp TIMESTAMP, dim_0_code STRING, dim_1_code STRING, dim_2_code STRING, dim_3_code STRING, dim_4_code STRING, dim_5_code STRING, dim_6_code STRING, dim_7_code STRING, dim_8_code STRING, dim_9_code STRING, metric_0_amount DOUBLE, metric_1_amount DOUBLE, metric_2_amount DOUBLE, metric_3_amount DOUBLE, metric_4_amount DOUBLE, metric_5_amount DOUBLE, metric_6_amount DOUBLE, metric_7_amount DOUBLE, metric_8_amount DOUBLE, metric_9_amount DOUBLE, metric_10_amount DOUBLE, metric_11_amount DOUBLE, metric_12_amount DOUBLE, metric_13_amount DOUBLE, status_flag STRING, event_date STRING);\nCREATE TABLE IF NOT EXISTS logistics_db.carrier_profile (entity_id STRING, segment_code STRING, region_code STRING, tier_level STRING);\nCREATE TABLE IF NOT EXISTS logistics_db.route_catalog (dim_0_code STRING, category_code STRING, subcategory_code STRING);\nCREATE TABLE IF NOT EXISTS logistics_db.legacy_shipment_events_snapshot (segment_code STRING, region_code STRING, category_code STRING, entity_count BIGINT, total_metric_0_amount DOUBLE, snapshot_date STRING);",
    "issue_sql": "-- ETL pipeline for logistics daily aggregation\nWITH base_events AS (\n    SELECT\n        shipment_events.entity_id,\n        shipment_events.event_timestamp,\n        shipment_events.dim_0_code,\n        shipment_events.dim_1_code,\n        shipment_events.dim_2_code,\n        shipment_events.dim_3_code,\n        shipment_events.dim_4_code,\n        shipment_events.metric_0_amount,\n        shipment_events.metric_1_amount,\n        shipment_events.metric_2_amount,\n        shipment_events.metric_3_amount,\n        shipment_events.metric_4_amount,\n        shipment_events.metric_5_amount,\n        shipment_events.metric_6_amount,\n        shipment_events.status_flag\n    FROM logistics_db.shipment_events shipment_events\n    WHERE shipment_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        carrier_profile.segment_code,\n        carrier_profile.region_code,\n        carrier_profile.tier_level,\n        route_catalog.category_code,\n        route_catalog.subcategory_code\n    FROM base_events\n    JOIN logistics_db.carrier_profile carrier_profile\n        ON base_events.entity_id = carrier_profile.entity_id\n    JOIN logistics_db.route_catalog route_catalog\n        ON base_events.dim_0_code = route_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM logistics_db.legacy_shipment_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.logistics_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "reference_sql": "-- ETL pipeline for logistics daily aggregation\nWITH base_events AS (\n    SELECT\n        shipment_events.entity_id,\n        shipment_events.event_timestamp,\n        shipment_events.dim_0_code,\n        shipment_events.dim_1_code,\n        shipment_events.dim_2_code,\n        shipment_events.dim_3_code,\n        shipment_events.dim_4_code,\n        shipment_events.metric_0_amount,\n        shipment_events.metric_1_amount,\n        shipment_events.metric_2_amount,\n        shipment_events.metric_3_amount,\n        shipment_events.metric_4_amount,\n        shipment_events.metric_5_amount,\n        shipment_events.metric_6_amount,\n        shipment_events.status_flag\n    FROM logistics_db.shipment_events shipment_events\n    WHERE shipment_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        carrier_profile.segment_code,\n        carrier_profile.region_code,\n        carrier_profile.tier_level,\n        route_catalog.category_code,\n        route_catalog.subcategory_code\n    FROM base_events\n    JOIN logistics_db.carrier_profile carrier_profile\n        ON base_events.entity_id = carrier_profile.entity_id\n    LEFT JOIN logistics_db.route_catalog route_catalog\n        ON base_events.dim_0_code = route_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM logistics_db.legacy_shipment_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.logistics_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "user_query": "Some entities that should appear in the daily summary (e.g. those without a matching category record) are missing from the output entirely. Expected: all entities from the base events should be retained even if category enrichment is unavailable. Please fix the bug.",
    "level1_error_type": "Joins & Grouping",
    "level2_error_type": "JOIN Type Selection",
    "level3_error_type": "Using INNER JOIN when LEFT JOIN is needed"
  },
  {
    "task_id": "education-syntax-4",
    "type": "syntax",
    "domain": "education",
    "ddl": "CREATE TABLE IF NOT EXISTS education_db.enrollment_events (entity_id STRING, event_timestamp TIMESTAMP, dim_0_code STRING, dim_1_code STRING, dim_2_code STRING, dim_3_code STRING, dim_4_code STRING, dim_5_code STRING, dim_6_code STRING, dim_7_code STRING, dim_8_code STRING, dim_9_code STRING, metric_0_amount DOUBLE, metric_1_amount DOUBLE, metric_2_amount DOUBLE, metric_3_amount DOUBLE, metric_4_amount DOUBLE, metric_5_amount DOUBLE, metric_6_amount DOUBLE, metric_7_amount DOUBLE, metric_8_amount DOUBLE, metric_9_amount DOUBLE, metric_10_amount DOUBLE, metric_11_amount DOUBLE, metric_12_amount DOUBLE, metric_13_amount DOUBLE, status_flag STRING, event_date STRING);\nCREATE TABLE IF NOT EXISTS education_db.student_profile (entity_id STRING, segment_code STRING, region_code STRING, tier_level STRING);\nCREATE TABLE IF NOT EXISTS education_db.course_catalog (dim_0_code STRING, category_code STRING, subcategory_code STRING);\nCREATE TABLE IF NOT EXISTS education_db.legacy_enrollment_events_snapshot (segment_code STRING, region_code STRING, category_code STRING, entity_count BIGINT, total_metric_0_amount DOUBLE, snapshot_date STRING);",
    "issue_sql": "-- ETL pipeline for education daily aggregation\nWITH base_events AS (\n    SELECT\n        enrollment_events.entity_id,\n        enrollment_events.event_timestamp,\n        enrollment_events.dim_0_code,\n        enrollment_events.dim_1_code,\n        enrollment_events.dim_2_code,\n        enrollment_events.dim_3_code,\n        enrollment_events.dim_4_code,\n        enrollment_events.metric_0_amount,\n        enrollment_events.metric_1_amount,\n        enrollment_events.metric_2_amount,\n        enrollment_events.metric_3_amount,\n        enrollment_events.metric_4_amount,\n        enrollment_events.metric_5_amount,\n        enrollment_events.metric_6_amount,\n        enrollment_events.status_flag\n    FROM education_db.enrollment_events enrollment_events\n    WHERE enrollment_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        student_profile.segment_code,\n        student_profile.region_code,\n        student_profile.tier_level,\n        course_catalog.category_code,\n        course_catalog.subcategory_code\n    FROM base_events\n    JOIN education_db.student_profile student_profile\n        ON base_events.entity_id = student_profile.entity_id\n    LEFT JOIN education_db.course_catalog course_catalog\n        ON base_events.dim_0_code = course_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM education_db.legacy_enrollment_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.education_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "reference_sql": "-- ETL pipeline for education daily aggregation\nWITH base_events AS (\n    SELECT\n        enrollment_events.entity_id,\n        enrollment_events.event_timestamp,\n        enrollment_events.dim_0_code,\n        enrollment_events.dim_1_code,\n        enrollment_events.dim_2_code,\n        enrollment_events.dim_3_code,\n        enrollment_events.dim_4_code,\n        enrollment_events.metric_0_amount,\n        enrollment_events.metric_1_amount,\n        enrollment_events.metric_2_amount,\n        enrollment_events.metric_3_amount,\n        enrollment_events.metric_4_amount,\n        enrollment_events.metric_5_amount,\n        enrollment_events.metric_6_amount,\n        enrollment_events.status_flag\n    FROM education_db.enrollment_events enrollment_events\n    WHERE enrollment_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        student_profile.segment_code,\n        student_profile.region_code,\n        student_profile.tier_level,\n        course_catalog.category_code,\n        course_catalog.subcategory_code\n    FROM base_events\n    JOIN education_db.student_profile student_profile\n        ON base_events.entity_id = student_profile.entity_id\n    LEFT JOIN education_db.course_catalog course_catalog\n        ON base_events.dim_0_code = course_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM education_db.legacy_enrollment_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.education_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "error_message": "ParseException: mismatched input 'AS' expecting 'END' near 'value_tier'",
    "level1_error_type": "Grammar & Structure",
    "level2_error_type": "CASE Expression",
    "level3_error_type": "Missing END in CASE WHEN"
  },
  {
    "task_id": "education-semantic-4",
    "type": "semantic",
    "domain": "education",
    "ddl": "CREATE TABLE IF NOT EXISTS education_db.enrollment_events (entity_id STRING, event_timestamp TIMESTAMP, dim_0_code STRING, dim_1_code STRING, dim_2_code STRING, dim_3_code STRING, dim_4_code STRING, dim_5_code STRING, dim_6_code STRING, dim_7_code STRING, dim_8_code STRING, dim_9_code STRING, metric_0_amount DOUBLE, metric_1_amount DOUBLE, metric_2_amount DOUBLE, metric_3_amount DOUBLE, metric_4_amount DOUBLE, metric_5_amount DOUBLE, metric_6_amount DOUBLE, metric_7_amount DOUBLE, metric_8_amount DOUBLE, metric_9_amount DOUBLE, metric_10_amount DOUBLE, metric_11_amount DOUBLE, metric_12_amount DOUBLE, metric_13_amount DOUBLE, status_flag STRING, event_date STRING);\nCREATE TABLE IF NOT EXISTS education_db.student_profile (entity_id STRING, segment_code STRING, region_code STRING, tier_level STRING);\nCREATE TABLE IF NOT EXISTS education_db.course_catalog (dim_0_code STRING, category_code STRING, subcategory_code STRING);\nCREATE TABLE IF NOT EXISTS education_db.legacy_enrollment_events_snapshot (segment_code STRING, region_code STRING, category_code STRING, entity_count BIGINT, total_metric_0_amount DOUBLE, snapshot_date STRING);",
    "issue_sql": "-- ETL pipeline for education daily aggregation\nWITH base_events AS (\n    SELECT\n        enrollment_events.entity_id,\n        enrollment_events.event_timestamp,\n        enrollment_events.dim_0_code,\n        enrollment_events.dim_1_code,\n        enrollment_events.dim_2_code,\n        enrollment_events.dim_3_code,\n        enrollment_events.dim_4_code,\n        enrollment_events.metric_0_amount,\n        enrollment_events.metric_1_amount,\n        enrollment_events.metric_2_amount,\n        enrollment_events.metric_3_amount,\n        enrollment_events.metric_4_amount,\n        enrollment_events.metric_5_amount,\n        enrollment_events.metric_6_amount,\n        enrollment_events.status_flag\n    FROM education_db.enrollment_events enrollment_events\n    WHERE enrollment_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        student_profile.segment_code,\n        student_profile.region_code,\n        student_profile.tier_level,\n        course_catalog.category_code,\n        course_catalog.subcategory_code\n    FROM base_events\n    JOIN education_db.student_profile student_profile\n        ON base_events.entity_id = student_profile.entity_id\n    LEFT JOIN education_db.course_catalog course_catalog\n        ON base_events.dim_0_code = course_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM education_db.legacy_enrollment_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.education_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "reference_sql": "-- ETL pipeline for education daily aggregation\nWITH base_events AS (\n    SELECT\n        enrollment_events.entity_id,\n        enrollment_events.event_timestamp,\n        enrollment_events.dim_0_code,\n        enrollment_events.dim_1_code,\n        enrollment_events.dim_2_code,\n        enrollment_events.dim_3_code,\n        enrollment_events.dim_4_code,\n        enrollment_events.metric_0_amount,\n        enrollment_events.metric_1_amount,\n        enrollment_events.metric_2_amount,\n        enrollment_events.metric_3_amount,\n        enrollment_events.metric_4_amount,\n        enrollment_events.metric_5_amount,\n        enrollment_events.metric_6_amount,\n        enrollment_events.status_flag\n    FROM education_db.enrollment_events enrollment_events\n    WHERE enrollment_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        student_profile.segment_code,\n        student_profile.region_code,\n        student_profile.tier_level,\n        course_catalog.category_code,\n        course_catalog.subcategory_code\n    FROM base_events\n    JOIN education_db.student_profile student_profile\n        ON base_events.entity_id = student_profile.entity_id\n    LEFT JOIN education_db.course_catalog course_catalog\n        ON base_events.dim_0_code = course_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM education_db.legacy_enrollment_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.education_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "user_query": "The daily summary table sometimes has fewer rows than expected when the current period and legacy snapshot happen to produce identical rows for a segment; those duplicate-looking rows are actually distinct facts and should both be kept. Please fix the bug.",
    "level1_error_type": "Semantics & Logic",
    "level2_error_type": "Set Operations",
    "level3_error_type": "UNION vs UNION ALL misuse"
  },
  {
    "task_id": "telecom-syntax-5",
    "type": "syntax",
    "domain": "telecom",
    "ddl": "CREATE TABLE IF NOT EXISTS telecom_db.usage_events (entity_id STRING, event_timestamp TIMESTAMP, dim_0_code STRING, dim_1_code STRING, dim_2_code STRING, dim_3_code STRING, dim_4_code STRING, dim_5_code STRING, dim_6_code STRING, dim_7_code STRING, dim_8_code STRING, dim_9_code STRING, metric_0_amount DOUBLE, metric_1_amount DOUBLE, metric_2_amount DOUBLE, metric_3_amount DOUBLE, metric_4_amount DOUBLE, metric_5_amount DOUBLE, metric_6_amount DOUBLE, metric_7_amount DOUBLE, metric_8_amount DOUBLE, metric_9_amount DOUBLE, metric_10_amount DOUBLE, metric_11_amount DOUBLE, metric_12_amount DOUBLE, metric_13_amount DOUBLE, status_flag STRING, event_date STRING);\nCREATE TABLE IF NOT EXISTS telecom_db.subscriber_profile (entity_id STRING, segment_code STRING, region_code STRING, tier_level STRING);\nCREATE TABLE IF NOT EXISTS telecom_db.plan_catalog (dim_0_code STRING, category_code STRING, subcategory_code STRING);\nCREATE TABLE IF NOT EXISTS telecom_db.legacy_usage_events_snapshot (segment_code STRING, region_code STRING, category_code STRING, entity_count BIGINT, total_metric_0_amount DOUBLE, snapshot_date STRING);",
    "issue_sql": "-- ETL pipeline for telecom daily aggregation\nWITH base_events AS (\n    SELECT\n        usage_events.entity_id,\n        usage_events.event_timestamp,\n        usage_events.dim_0_code,\n        usage_events.dim_1_code,\n        usage_events.dim_2_code,\n        usage_events.dim_3_code,\n        usage_events.dim_4_code,\n        usage_events.metric_0_amount,\n        usage_events.metric_1_amount,\n        usage_events.metric_2_amount,\n        usage_events.metric_3_amount,\n        usage_events.metric_4_amount,\n        usage_events.metric_5_amount,\n        usage_events.metric_6_amount,\n        usage_events.status_flag\n    FROM telecom_db.usage_events usage_events\n    WHERE usage_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        subscriber_profile.segment_code,\n        subscriber_profile.region_code,\n        subscriber_profile.tier_level,\n        plan_catalog.category_code,\n        plan_catalog.subcategory_code\n    FROM base_events\n    JOIN telecom_db.subscriber_profile subscriber_profile\n        ON base_events.entity_id = subscriber_profile.entity_id\n    LEFT JOIN telecom_db.plan_catalog plan_catalog\n        ON base_events.dim_0_code = plan_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM telecom_db.legacy_usage_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.telecom_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "reference_sql": "-- ETL pipeline for telecom daily aggregation\nWITH base_events AS (\n    SELECT\n        usage_events.entity_id,\n        usage_events.event_timestamp,\n        usage_events.dim_0_code,\n        usage_events.dim_1_code,\n        usage_events.dim_2_code,\n        usage_events.dim_3_code,\n        usage_events.dim_4_code,\n        usage_events.metric_0_amount,\n        usage_events.metric_1_amount,\n        usage_events.metric_2_amount,\n        usage_events.metric_3_amount,\n        usage_events.metric_4_amount,\n        usage_events.metric_5_amount,\n        usage_events.metric_6_amount,\n        usage_events.status_flag\n    FROM telecom_db.usage_events usage_events\n    WHERE usage_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        subscriber_profile.segment_code,\n        subscriber_profile.region_code,\n        subscriber_profile.tier_level,\n        plan_catalog.category_code,\n        plan_catalog.subcategory_code\n    FROM base_events\n    JOIN telecom_db.subscriber_profile subscriber_profile\n        ON base_events.entity_id = subscriber_profile.entity_id\n    LEFT JOIN telecom_db.plan_catalog plan_catalog\n        ON base_events.dim_0_code = plan_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM telecom_db.legacy_usage_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.telecom_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "error_message": "ParseException: mismatched input 'AS' expecting 'END' near 'value_tier'",
    "level1_error_type": "Grammar & Structure",
    "level2_error_type": "CASE Expression",
    "level3_error_type": "Missing END in CASE WHEN"
  },
  {
    "task_id": "telecom-semantic-5",
    "type": "semantic",
    "domain": "telecom",
    "ddl": "CREATE TABLE IF NOT EXISTS telecom_db.usage_events (entity_id STRING, event_timestamp TIMESTAMP, dim_0_code STRING, dim_1_code STRING, dim_2_code STRING, dim_3_code STRING, dim_4_code STRING, dim_5_code STRING, dim_6_code STRING, dim_7_code STRING, dim_8_code STRING, dim_9_code STRING, metric_0_amount DOUBLE, metric_1_amount DOUBLE, metric_2_amount DOUBLE, metric_3_amount DOUBLE, metric_4_amount DOUBLE, metric_5_amount DOUBLE, metric_6_amount DOUBLE, metric_7_amount DOUBLE, metric_8_amount DOUBLE, metric_9_amount DOUBLE, metric_10_amount DOUBLE, metric_11_amount DOUBLE, metric_12_amount DOUBLE, metric_13_amount DOUBLE, status_flag STRING, event_date STRING);\nCREATE TABLE IF NOT EXISTS telecom_db.subscriber_profile (entity_id STRING, segment_code STRING, region_code STRING, tier_level STRING);\nCREATE TABLE IF NOT EXISTS telecom_db.plan_catalog (dim_0_code STRING, category_code STRING, subcategory_code STRING);\nCREATE TABLE IF NOT EXISTS telecom_db.legacy_usage_events_snapshot (segment_code STRING, region_code STRING, category_code STRING, entity_count BIGINT, total_metric_0_amount DOUBLE, snapshot_date STRING);",
    "issue_sql": "-- ETL pipeline for telecom daily aggregation\nWITH base_events AS (\n    SELECT\n        usage_events.entity_id,\n        usage_events.event_timestamp,\n        usage_events.dim_0_code,\n        usage_events.dim_1_code,\n        usage_events.dim_2_code,\n        usage_events.dim_3_code,\n        usage_events.dim_4_code,\n        usage_events.metric_0_amount,\n        usage_events.metric_1_amount,\n        usage_events.metric_2_amount,\n        usage_events.metric_3_amount,\n        usage_events.metric_4_amount,\n        usage_events.metric_5_amount,\n        usage_events.metric_6_amount,\n        usage_events.status_flag\n    FROM telecom_db.usage_events usage_events\n    WHERE usage_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        subscriber_profile.segment_code,\n        subscriber_profile.region_code,\n        subscriber_profile.tier_level,\n        plan_catalog.category_code,\n        plan_catalog.subcategory_code\n    FROM base_events\n    JOIN telecom_db.subscriber_profile subscriber_profile\n        ON base_events.dim_0_code = subscriber_profile.entity_id\n    LEFT JOIN telecom_db.plan_catalog plan_catalog\n        ON base_events.dim_0_code = plan_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM telecom_db.legacy_usage_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.telecom_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "reference_sql": "-- ETL pipeline for telecom daily aggregation\nWITH base_events AS (\n    SELECT\n        usage_events.entity_id,\n        usage_events.event_timestamp,\n        usage_events.dim_0_code,\n        usage_events.dim_1_code,\n        usage_events.dim_2_code,\n        usage_events.dim_3_code,\n        usage_events.dim_4_code,\n        usage_events.metric_0_amount,\n        usage_events.metric_1_amount,\n        usage_events.metric_2_amount,\n        usage_events.metric_3_amount,\n        usage_events.metric_4_amount,\n        usage_events.metric_5_amount,\n        usage_events.metric_6_amount,\n        usage_events.status_flag\n    FROM telecom_db.usage_events usage_events\n    WHERE usage_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        subscriber_profile.segment_code,\n        subscriber_profile.region_code,\n        subscriber_profile.tier_level,\n        plan_catalog.category_code,\n        plan_catalog.subcategory_code\n    FROM base_events\n    JOIN telecom_db.subscriber_profile subscriber_profile\n        ON base_events.entity_id = subscriber_profile.entity_id\n    LEFT JOIN telecom_db.plan_catalog plan_catalog\n        ON base_events.dim_0_code = plan_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM telecom_db.legacy_usage_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.telecom_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "user_query": "The join between base events and the profile dimension looks wrong: segment_code and region_code values in the output don't correspond to the correct entity anymore, producing mismatched enrichment. Please fix the bug so profile attributes are joined on the correct key.",
    "level1_error_type": "Joins & Grouping",
    "level2_error_type": "JOIN Logic",
    "level3_error_type": "Wrong join key used"
  },
  {
    "task_id": "real_estate-syntax-6",
    "type": "syntax",
    "domain": "real_estate",
    "ddl": "CREATE TABLE IF NOT EXISTS real_estate_db.listing_events (entity_id STRING, event_timestamp TIMESTAMP, dim_0_code STRING, dim_1_code STRING, dim_2_code STRING, dim_3_code STRING, dim_4_code STRING, dim_5_code STRING, dim_6_code STRING, dim_7_code STRING, dim_8_code STRING, dim_9_code STRING, metric_0_amount DOUBLE, metric_1_amount DOUBLE, metric_2_amount DOUBLE, metric_3_amount DOUBLE, metric_4_amount DOUBLE, metric_5_amount DOUBLE, metric_6_amount DOUBLE, metric_7_amount DOUBLE, metric_8_amount DOUBLE, metric_9_amount DOUBLE, metric_10_amount DOUBLE, metric_11_amount DOUBLE, metric_12_amount DOUBLE, metric_13_amount DOUBLE, status_flag STRING, event_date STRING);\nCREATE TABLE IF NOT EXISTS real_estate_db.agent_profile (entity_id STRING, segment_code STRING, region_code STRING, tier_level STRING);\nCREATE TABLE IF NOT EXISTS real_estate_db.property_catalog (dim_0_code STRING, category_code STRING, subcategory_code STRING);\nCREATE TABLE IF NOT EXISTS real_estate_db.legacy_listing_events_snapshot (segment_code STRING, region_code STRING, category_code STRING, entity_count BIGINT, total_metric_0_amount DOUBLE, snapshot_date STRING);",
    "issue_sql": "-- ETL pipeline for real_estate daily aggregation\nWITH base_events AS (\n    SELECT\n        listing_events.entity_id,\n        listing_events.event_timestamp,\n        listing_events.dim_0_code,\n        listing_events.dim_1_code,\n        listing_events.dim_2_code,\n        listing_events.dim_3_code,\n        listing_events.dim_4_code,\n        listing_events.metric_0_amount,\n        listing_events.metric_1_amount,\n        listing_events.metric_2_amount,\n        listing_events.metric_3_amount,\n        listing_events.metric_4_amount,\n        listing_events.metric_5_amount,\n        listing_events.metric_6_amount,\n        listing_events.status_flag\n    FROM real_estate_db.listing_events listing_events\n    WHERE listing_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        agent_profile.segment_code,\n        agent_profile.region_code,\n        agent_profile.tier_level,\n        property_catalog.category_code,\n        property_catalog.subcategory_code\n    FROM base_events\n    JOIN real_estate_db.agent_profile agent_profile\n        ON base_events.entity_id = agent_profile.entity_id\n    LEFT JOIN real_estate_db.property_catalog property_catalog\n        ON base_events.dim_0_code = property_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM real_estate_db.legacy_listing_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.real_estate_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "reference_sql": "-- ETL pipeline for real_estate daily aggregation\nWITH base_events AS (\n    SELECT\n        listing_events.entity_id,\n        listing_events.event_timestamp,\n        listing_events.dim_0_code,\n        listing_events.dim_1_code,\n        listing_events.dim_2_code,\n        listing_events.dim_3_code,\n        listing_events.dim_4_code,\n        listing_events.metric_0_amount,\n        listing_events.metric_1_amount,\n        listing_events.metric_2_amount,\n        listing_events.metric_3_amount,\n        listing_events.metric_4_amount,\n        listing_events.metric_5_amount,\n        listing_events.metric_6_amount,\n        listing_events.status_flag\n    FROM real_estate_db.listing_events listing_events\n    WHERE listing_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        agent_profile.segment_code,\n        agent_profile.region_code,\n        agent_profile.tier_level,\n        property_catalog.category_code,\n        property_catalog.subcategory_code\n    FROM base_events\n    JOIN real_estate_db.agent_profile agent_profile\n        ON base_events.entity_id = agent_profile.entity_id\n    LEFT JOIN real_estate_db.property_catalog property_catalog\n        ON base_events.dim_0_code = property_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM real_estate_db.legacy_listing_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.real_estate_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "error_message": "ParseException: mismatched input 'AS' expecting 'END' near 'value_tier'",
    "level1_error_type": "Grammar & Structure",
    "level2_error_type": "CASE Expression",
    "level3_error_type": "Missing END in CASE WHEN"
  },
  {
    "task_id": "real_estate-semantic-6",
    "type": "semantic",
    "domain": "real_estate",
    "ddl": "CREATE TABLE IF NOT EXISTS real_estate_db.listing_events (entity_id STRING, event_timestamp TIMESTAMP, dim_0_code STRING, dim_1_code STRING, dim_2_code STRING, dim_3_code STRING, dim_4_code STRING, dim_5_code STRING, dim_6_code STRING, dim_7_code STRING, dim_8_code STRING, dim_9_code STRING, metric_0_amount DOUBLE, metric_1_amount DOUBLE, metric_2_amount DOUBLE, metric_3_amount DOUBLE, metric_4_amount DOUBLE, metric_5_amount DOUBLE, metric_6_amount DOUBLE, metric_7_amount DOUBLE, metric_8_amount DOUBLE, metric_9_amount DOUBLE, metric_10_amount DOUBLE, metric_11_amount DOUBLE, metric_12_amount DOUBLE, metric_13_amount DOUBLE, status_flag STRING, event_date STRING);\nCREATE TABLE IF NOT EXISTS real_estate_db.agent_profile (entity_id STRING, segment_code STRING, region_code STRING, tier_level STRING);\nCREATE TABLE IF NOT EXISTS real_estate_db.property_catalog (dim_0_code STRING, category_code STRING, subcategory_code STRING);\nCREATE TABLE IF NOT EXISTS real_estate_db.legacy_listing_events_snapshot (segment_code STRING, region_code STRING, category_code STRING, entity_count BIGINT, total_metric_0_amount DOUBLE, snapshot_date STRING);",
    "issue_sql": "-- ETL pipeline for real_estate daily aggregation\nWITH base_events AS (\n    SELECT\n        listing_events.entity_id,\n        listing_events.event_timestamp,\n        listing_events.dim_0_code,\n        listing_events.dim_1_code,\n        listing_events.dim_2_code,\n        listing_events.dim_3_code,\n        listing_events.dim_4_code,\n        listing_events.metric_0_amount,\n        listing_events.metric_1_amount,\n        listing_events.metric_2_amount,\n        listing_events.metric_3_amount,\n        listing_events.metric_4_amount,\n        listing_events.metric_5_amount,\n        listing_events.metric_6_amount,\n        listing_events.status_flag\n    FROM real_estate_db.listing_events listing_events\n    WHERE listing_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        agent_profile.segment_code,\n        agent_profile.region_code,\n        agent_profile.tier_level,\n        property_catalog.category_code,\n        property_catalog.subcategory_code\n    FROM base_events\n    JOIN real_estate_db.agent_profile agent_profile\n        ON base_events.entity_id = agent_profile.entity_id\n    JOIN real_estate_db.property_catalog property_catalog\n        ON base_events.dim_0_code = property_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM real_estate_db.legacy_listing_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.real_estate_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "reference_sql": "-- ETL pipeline for real_estate daily aggregation\nWITH base_events AS (\n    SELECT\n        listing_events.entity_id,\n        listing_events.event_timestamp,\n        listing_events.dim_0_code,\n        listing_events.dim_1_code,\n        listing_events.dim_2_code,\n        listing_events.dim_3_code,\n        listing_events.dim_4_code,\n        listing_events.metric_0_amount,\n        listing_events.metric_1_amount,\n        listing_events.metric_2_amount,\n        listing_events.metric_3_amount,\n        listing_events.metric_4_amount,\n        listing_events.metric_5_amount,\n        listing_events.metric_6_amount,\n        listing_events.status_flag\n    FROM real_estate_db.listing_events listing_events\n    WHERE listing_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        agent_profile.segment_code,\n        agent_profile.region_code,\n        agent_profile.tier_level,\n        property_catalog.category_code,\n        property_catalog.subcategory_code\n    FROM base_events\n    JOIN real_estate_db.agent_profile agent_profile\n        ON base_events.entity_id = agent_profile.entity_id\n    LEFT JOIN real_estate_db.property_catalog property_catalog\n        ON base_events.dim_0_code = property_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM real_estate_db.legacy_listing_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.real_estate_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "user_query": "Some entities that should appear in the daily summary (e.g. those without a matching category record) are missing from the output entirely. Expected: all entities from the base events should be retained even if category enrichment is unavailable. Please fix the bug.",
    "level1_error_type": "Joins & Grouping",
    "level2_error_type": "JOIN Type Selection",
    "level3_error_type": "Using INNER JOIN when LEFT JOIN is needed"
  },
  {
    "task_id": "entertainment-syntax-7",
    "type": "syntax",
    "domain": "entertainment",
    "ddl": "CREATE TABLE IF NOT EXISTS entertainment_db.stream_events (entity_id STRING, event_timestamp TIMESTAMP, dim_0_code STRING, dim_1_code STRING, dim_2_code STRING, dim_3_code STRING, dim_4_code STRING, dim_5_code STRING, dim_6_code STRING, dim_7_code STRING, dim_8_code STRING, dim_9_code STRING, metric_0_amount DOUBLE, metric_1_amount DOUBLE, metric_2_amount DOUBLE, metric_3_amount DOUBLE, metric_4_amount DOUBLE, metric_5_amount DOUBLE, metric_6_amount DOUBLE, metric_7_amount DOUBLE, metric_8_amount DOUBLE, metric_9_amount DOUBLE, metric_10_amount DOUBLE, metric_11_amount DOUBLE, metric_12_amount DOUBLE, metric_13_amount DOUBLE, status_flag STRING, event_date STRING);\nCREATE TABLE IF NOT EXISTS entertainment_db.viewer_profile (entity_id STRING, segment_code STRING, region_code STRING, tier_level STRING);\nCREATE TABLE IF NOT EXISTS entertainment_db.title_catalog (dim_0_code STRING, category_code STRING, subcategory_code STRING);\nCREATE TABLE IF NOT EXISTS entertainment_db.legacy_stream_events_snapshot (segment_code STRING, region_code STRING, category_code STRING, entity_count BIGINT, total_metric_0_amount DOUBLE, snapshot_date STRING);",
    "issue_sql": "-- ETL pipeline for entertainment daily aggregation\nWITH base_events AS (\n    SELECT\n        stream_events.entity_id,\n        stream_events.event_timestamp,\n        stream_events.dim_0_code,\n        stream_events.dim_1_code,\n        stream_events.dim_2_code,\n        stream_events.dim_3_code,\n        stream_events.dim_4_code,\n        stream_events.metric_0_amount,\n        stream_events.metric_1_amount,\n        stream_events.metric_2_amount,\n        stream_events.metric_3_amount,\n        stream_events.metric_4_amount,\n        stream_events.metric_5_amount,\n        stream_events.metric_6_amount,\n        stream_events.status_flag\n    FROM entertainment_db.stream_events stream_events\n    WHERE stream_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        viewer_profile.segment_code,\n        viewer_profile.region_code,\n        viewer_profile.tier_level,\n        title_catalog.category_code,\n        title_catalog.subcategory_code\n    FROM base_events\n    JOIN entertainment_db.viewer_profile viewer_profile\n        ON base_events.entity_id = viewer_profile.entity_id\n    LEFT JOIN entertainment_db.title_catalog title_catalog\n        ON base_events.dim_0_code = title_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM entertainment_db.legacy_stream_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.entertainment_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "reference_sql": "-- ETL pipeline for entertainment daily aggregation\nWITH base_events AS (\n    SELECT\n        stream_events.entity_id,\n        stream_events.event_timestamp,\n        stream_events.dim_0_code,\n        stream_events.dim_1_code,\n        stream_events.dim_2_code,\n        stream_events.dim_3_code,\n        stream_events.dim_4_code,\n        stream_events.metric_0_amount,\n        stream_events.metric_1_amount,\n        stream_events.metric_2_amount,\n        stream_events.metric_3_amount,\n        stream_events.metric_4_amount,\n        stream_events.metric_5_amount,\n        stream_events.metric_6_amount,\n        stream_events.status_flag\n    FROM entertainment_db.stream_events stream_events\n    WHERE stream_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        viewer_profile.segment_code,\n        viewer_profile.region_code,\n        viewer_profile.tier_level,\n        title_catalog.category_code,\n        title_catalog.subcategory_code\n    FROM base_events\n    JOIN entertainment_db.viewer_profile viewer_profile\n        ON base_events.entity_id = viewer_profile.entity_id\n    LEFT JOIN entertainment_db.title_catalog title_catalog\n        ON base_events.dim_0_code = title_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM entertainment_db.legacy_stream_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.entertainment_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "error_message": "ParseException: mismatched input 'AS' expecting 'END' near 'value_tier'",
    "level1_error_type": "Grammar & Structure",
    "level2_error_type": "CASE Expression",
    "level3_error_type": "Missing END in CASE WHEN"
  },
  {
    "task_id": "entertainment-semantic-7",
    "type": "semantic",
    "domain": "entertainment",
    "ddl": "CREATE TABLE IF NOT EXISTS entertainment_db.stream_events (entity_id STRING, event_timestamp TIMESTAMP, dim_0_code STRING, dim_1_code STRING, dim_2_code STRING, dim_3_code STRING, dim_4_code STRING, dim_5_code STRING, dim_6_code STRING, dim_7_code STRING, dim_8_code STRING, dim_9_code STRING, metric_0_amount DOUBLE, metric_1_amount DOUBLE, metric_2_amount DOUBLE, metric_3_amount DOUBLE, metric_4_amount DOUBLE, metric_5_amount DOUBLE, metric_6_amount DOUBLE, metric_7_amount DOUBLE, metric_8_amount DOUBLE, metric_9_amount DOUBLE, metric_10_amount DOUBLE, metric_11_amount DOUBLE, metric_12_amount DOUBLE, metric_13_amount DOUBLE, status_flag STRING, event_date STRING);\nCREATE TABLE IF NOT EXISTS entertainment_db.viewer_profile (entity_id STRING, segment_code STRING, region_code STRING, tier_level STRING);\nCREATE TABLE IF NOT EXISTS entertainment_db.title_catalog (dim_0_code STRING, category_code STRING, subcategory_code STRING);\nCREATE TABLE IF NOT EXISTS entertainment_db.legacy_stream_events_snapshot (segment_code STRING, region_code STRING, category_code STRING, entity_count BIGINT, total_metric_0_amount DOUBLE, snapshot_date STRING);",
    "issue_sql": "-- ETL pipeline for entertainment daily aggregation\nWITH base_events AS (\n    SELECT\n        stream_events.entity_id,\n        stream_events.event_timestamp,\n        stream_events.dim_0_code,\n        stream_events.dim_1_code,\n        stream_events.dim_2_code,\n        stream_events.dim_3_code,\n        stream_events.dim_4_code,\n        stream_events.metric_0_amount,\n        stream_events.metric_1_amount,\n        stream_events.metric_2_amount,\n        stream_events.metric_3_amount,\n        stream_events.metric_4_amount,\n        stream_events.metric_5_amount,\n        stream_events.metric_6_amount,\n        stream_events.status_flag\n    FROM entertainment_db.stream_events stream_events\n    WHERE stream_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        viewer_profile.segment_code,\n        viewer_profile.region_code,\n        viewer_profile.tier_level,\n        title_catalog.category_code,\n        title_catalog.subcategory_code\n    FROM base_events\n    JOIN entertainment_db.viewer_profile viewer_profile\n        ON base_events.entity_id = viewer_profile.entity_id\n    LEFT JOIN entertainment_db.title_catalog title_catalog\n        ON base_events.dim_0_code = title_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM entertainment_db.legacy_stream_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.entertainment_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "reference_sql": "-- ETL pipeline for entertainment daily aggregation\nWITH base_events AS (\n    SELECT\n        stream_events.entity_id,\n        stream_events.event_timestamp,\n        stream_events.dim_0_code,\n        stream_events.dim_1_code,\n        stream_events.dim_2_code,\n        stream_events.dim_3_code,\n        stream_events.dim_4_code,\n        stream_events.metric_0_amount,\n        stream_events.metric_1_amount,\n        stream_events.metric_2_amount,\n        stream_events.metric_3_amount,\n        stream_events.metric_4_amount,\n        stream_events.metric_5_amount,\n        stream_events.metric_6_amount,\n        stream_events.status_flag\n    FROM entertainment_db.stream_events stream_events\n    WHERE stream_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        viewer_profile.segment_code,\n        viewer_profile.region_code,\n        viewer_profile.tier_level,\n        title_catalog.category_code,\n        title_catalog.subcategory_code\n    FROM base_events\n    JOIN entertainment_db.viewer_profile viewer_profile\n        ON base_events.entity_id = viewer_profile.entity_id\n    LEFT JOIN entertainment_db.title_catalog title_catalog\n        ON base_events.dim_0_code = title_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM entertainment_db.legacy_stream_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.entertainment_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "user_query": "The daily summary table sometimes has fewer rows than expected when the current period and legacy snapshot happen to produce identical rows for a segment; those duplicate-looking rows are actually distinct facts and should both be kept. Please fix the bug.",
    "level1_error_type": "Semantics & Logic",
    "level2_error_type": "Set Operations",
    "level3_error_type": "UNION vs UNION ALL misuse"
  },
  {
    "task_id": "travel-syntax-8",
    "type": "syntax",
    "domain": "travel",
    "ddl": "CREATE TABLE IF NOT EXISTS travel_db.booking_events (entity_id STRING, event_timestamp TIMESTAMP, dim_0_code STRING, dim_1_code STRING, dim_2_code STRING, dim_3_code STRING, dim_4_code STRING, dim_5_code STRING, dim_6_code STRING, dim_7_code STRING, dim_8_code STRING, dim_9_code STRING, metric_0_amount DOUBLE, metric_1_amount DOUBLE, metric_2_amount DOUBLE, metric_3_amount DOUBLE, metric_4_amount DOUBLE, metric_5_amount DOUBLE, metric_6_amount DOUBLE, metric_7_amount DOUBLE, metric_8_amount DOUBLE, metric_9_amount DOUBLE, metric_10_amount DOUBLE, metric_11_amount DOUBLE, metric_12_amount DOUBLE, metric_13_amount DOUBLE, status_flag STRING, event_date STRING);\nCREATE TABLE IF NOT EXISTS travel_db.traveler_profile (entity_id STRING, segment_code STRING, region_code STRING, tier_level STRING);\nCREATE TABLE IF NOT EXISTS travel_db.fare_catalog (dim_0_code STRING, category_code STRING, subcategory_code STRING);\nCREATE TABLE IF NOT EXISTS travel_db.legacy_booking_events_snapshot (segment_code STRING, region_code STRING, category_code STRING, entity_count BIGINT, total_metric_0_amount DOUBLE, snapshot_date STRING);",
    "issue_sql": "-- ETL pipeline for travel daily aggregation\nWITH base_events AS (\n    SELECT\n        booking_events.entity_id,\n        booking_events.event_timestamp,\n        booking_events.dim_0_code,\n        booking_events.dim_1_code,\n        booking_events.dim_2_code,\n        booking_events.dim_3_code,\n        booking_events.dim_4_code,\n        booking_events.metric_0_amount,\n        booking_events.metric_1_amount,\n        booking_events.metric_2_amount,\n        booking_events.metric_3_amount,\n        booking_events.metric_4_amount,\n        booking_events.metric_5_amount,\n        booking_events.metric_6_amount,\n        booking_events.status_flag\n    FROM travel_db.booking_events booking_events\n    WHERE booking_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        traveler_profile.segment_code,\n        traveler_profile.region_code,\n        traveler_profile.tier_level,\n        fare_catalog.category_code,\n        fare_catalog.subcategory_code\n    FROM base_events\n    JOIN travel_db.traveler_profile traveler_profile\n        ON base_events.entity_id = traveler_profile.entity_id\n    LEFT JOIN travel_db.fare_catalog fare_catalog\n        ON base_events.dim_0_code = fare_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM travel_db.legacy_booking_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.travel_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "reference_sql": "-- ETL pipeline for travel daily aggregation\nWITH base_events AS (\n    SELECT\n        booking_events.entity_id,\n        booking_events.event_timestamp,\n        booking_events.dim_0_code,\n        booking_events.dim_1_code,\n        booking_events.dim_2_code,\n        booking_events.dim_3_code,\n        booking_events.dim_4_code,\n        booking_events.metric_0_amount,\n        booking_events.metric_1_amount,\n        booking_events.metric_2_amount,\n        booking_events.metric_3_amount,\n        booking_events.metric_4_amount,\n        booking_events.metric_5_amount,\n        booking_events.metric_6_amount,\n        booking_events.status_flag\n    FROM travel_db.booking_events booking_events\n    WHERE booking_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        traveler_profile.segment_code,\n        traveler_profile.region_code,\n        traveler_profile.tier_level,\n        fare_catalog.category_code,\n        fare_catalog.subcategory_code\n    FROM base_events\n    JOIN travel_db.traveler_profile traveler_profile\n        ON base_events.entity_id = traveler_profile.entity_id\n    LEFT JOIN travel_db.fare_catalog fare_catalog\n        ON base_events.dim_0_code = fare_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM travel_db.legacy_booking_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.travel_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "error_message": "ParseException: mismatched input 'AS' expecting 'END' near 'value_tier'",
    "level1_error_type": "Grammar & Structure",
    "level2_error_type": "CASE Expression",
    "level3_error_type": "Missing END in CASE WHEN"
  },
  {
    "task_id": "travel-semantic-8",
    "type": "semantic",
    "domain": "travel",
    "ddl": "CREATE TABLE IF NOT EXISTS travel_db.booking_events (entity_id STRING, event_timestamp TIMESTAMP, dim_0_code STRING, dim_1_code STRING, dim_2_code STRING, dim_3_code STRING, dim_4_code STRING, dim_5_code STRING, dim_6_code STRING, dim_7_code STRING, dim_8_code STRING, dim_9_code STRING, metric_0_amount DOUBLE, metric_1_amount DOUBLE, metric_2_amount DOUBLE, metric_3_amount DOUBLE, metric_4_amount DOUBLE, metric_5_amount DOUBLE, metric_6_amount DOUBLE, metric_7_amount DOUBLE, metric_8_amount DOUBLE, metric_9_amount DOUBLE, metric_10_amount DOUBLE, metric_11_amount DOUBLE, metric_12_amount DOUBLE, metric_13_amount DOUBLE, status_flag STRING, event_date STRING);\nCREATE TABLE IF NOT EXISTS travel_db.traveler_profile (entity_id STRING, segment_code STRING, region_code STRING, tier_level STRING);\nCREATE TABLE IF NOT EXISTS travel_db.fare_catalog (dim_0_code STRING, category_code STRING, subcategory_code STRING);\nCREATE TABLE IF NOT EXISTS travel_db.legacy_booking_events_snapshot (segment_code STRING, region_code STRING, category_code STRING, entity_count BIGINT, total_metric_0_amount DOUBLE, snapshot_date STRING);",
    "issue_sql": "-- ETL pipeline for travel daily aggregation\nWITH base_events AS (\n    SELECT\n        booking_events.entity_id,\n        booking_events.event_timestamp,\n        booking_events.dim_0_code,\n        booking_events.dim_1_code,\n        booking_events.dim_2_code,\n        booking_events.dim_3_code,\n        booking_events.dim_4_code,\n        booking_events.metric_0_amount,\n        booking_events.metric_1_amount,\n        booking_events.metric_2_amount,\n        booking_events.metric_3_amount,\n        booking_events.metric_4_amount,\n        booking_events.metric_5_amount,\n        booking_events.metric_6_amount,\n        booking_events.status_flag\n    FROM travel_db.booking_events booking_events\n    WHERE booking_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        traveler_profile.segment_code,\n        traveler_profile.region_code,\n        traveler_profile.tier_level,\n        fare_catalog.category_code,\n        fare_catalog.subcategory_code\n    FROM base_events\n    JOIN travel_db.traveler_profile traveler_profile\n        ON base_events.dim_0_code = traveler_profile.entity_id\n    LEFT JOIN travel_db.fare_catalog fare_catalog\n        ON base_events.dim_0_code = fare_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM travel_db.legacy_booking_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.travel_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "reference_sql": "-- ETL pipeline for travel daily aggregation\nWITH base_events AS (\n    SELECT\n        booking_events.entity_id,\n        booking_events.event_timestamp,\n        booking_events.dim_0_code,\n        booking_events.dim_1_code,\n        booking_events.dim_2_code,\n        booking_events.dim_3_code,\n        booking_events.dim_4_code,\n        booking_events.metric_0_amount,\n        booking_events.metric_1_amount,\n        booking_events.metric_2_amount,\n        booking_events.metric_3_amount,\n        booking_events.metric_4_amount,\n        booking_events.metric_5_amount,\n        booking_events.metric_6_amount,\n        booking_events.status_flag\n    FROM travel_db.booking_events booking_events\n    WHERE booking_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        traveler_profile.segment_code,\n        traveler_profile.region_code,\n        traveler_profile.tier_level,\n        fare_catalog.category_code,\n        fare_catalog.subcategory_code\n    FROM base_events\n    JOIN travel_db.traveler_profile traveler_profile\n        ON base_events.entity_id = traveler_profile.entity_id\n    LEFT JOIN travel_db.fare_catalog fare_catalog\n        ON base_events.dim_0_code = fare_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM travel_db.legacy_booking_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.travel_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "user_query": "The join between base events and the profile dimension looks wrong: segment_code and region_code values in the output don't correspond to the correct entity anymore, producing mismatched enrichment. Please fix the bug so profile attributes are joined on the correct key.",
    "level1_error_type": "Joins & Grouping",
    "level2_error_type": "JOIN Logic",
    "level3_error_type": "Wrong join key used"
  },
  {
    "task_id": "energy-syntax-9",
    "type": "syntax",
    "domain": "energy",
    "ddl": "CREATE TABLE IF NOT EXISTS energy_db.meter_events (entity_id STRING, event_timestamp TIMESTAMP, dim_0_code STRING, dim_1_code STRING, dim_2_code STRING, dim_3_code STRING, dim_4_code STRING, dim_5_code STRING, dim_6_code STRING, dim_7_code STRING, dim_8_code STRING, dim_9_code STRING, metric_0_amount DOUBLE, metric_1_amount DOUBLE, metric_2_amount DOUBLE, metric_3_amount DOUBLE, metric_4_amount DOUBLE, metric_5_amount DOUBLE, metric_6_amount DOUBLE, metric_7_amount DOUBLE, metric_8_amount DOUBLE, metric_9_amount DOUBLE, metric_10_amount DOUBLE, metric_11_amount DOUBLE, metric_12_amount DOUBLE, metric_13_amount DOUBLE, status_flag STRING, event_date STRING);\nCREATE TABLE IF NOT EXISTS energy_db.site_profile (entity_id STRING, segment_code STRING, region_code STRING, tier_level STRING);\nCREATE TABLE IF NOT EXISTS energy_db.tariff_catalog (dim_0_code STRING, category_code STRING, subcategory_code STRING);\nCREATE TABLE IF NOT EXISTS energy_db.legacy_meter_events_snapshot (segment_code STRING, region_code STRING, category_code STRING, entity_count BIGINT, total_metric_0_amount DOUBLE, snapshot_date STRING);",
    "issue_sql": "-- ETL pipeline for energy daily aggregation\nWITH base_events AS (\n    SELECT\n        meter_events.entity_id,\n        meter_events.event_timestamp,\n        meter_events.dim_0_code,\n        meter_events.dim_1_code,\n        meter_events.dim_2_code,\n        meter_events.dim_3_code,\n        meter_events.dim_4_code,\n        meter_events.metric_0_amount,\n        meter_events.metric_1_amount,\n        meter_events.metric_2_amount,\n        meter_events.metric_3_amount,\n        meter_events.metric_4_amount,\n        meter_events.metric_5_amount,\n        meter_events.metric_6_amount,\n        meter_events.status_flag\n    FROM energy_db.meter_events meter_events\n    WHERE meter_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        site_profile.segment_code,\n        site_profile.region_code,\n        site_profile.tier_level,\n        tariff_catalog.category_code,\n        tariff_catalog.subcategory_code\n    FROM base_events\n    JOIN energy_db.site_profile site_profile\n        ON base_events.entity_id = site_profile.entity_id\n    LEFT JOIN energy_db.tariff_catalog tariff_catalog\n        ON base_events.dim_0_code = tariff_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM energy_db.legacy_meter_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.energy_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "reference_sql": "-- ETL pipeline for energy daily aggregation\nWITH base_events AS (\n    SELECT\n        meter_events.entity_id,\n        meter_events.event_timestamp,\n        meter_events.dim_0_code,\n        meter_events.dim_1_code,\n        meter_events.dim_2_code,\n        meter_events.dim_3_code,\n        meter_events.dim_4_code,\n        meter_events.metric_0_amount,\n        meter_events.metric_1_amount,\n        meter_events.metric_2_amount,\n        meter_events.metric_3_amount,\n        meter_events.metric_4_amount,\n        meter_events.metric_5_amount,\n        meter_events.metric_6_amount,\n        meter_events.status_flag\n    FROM energy_db.meter_events meter_events\n    WHERE meter_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        site_profile.segment_code,\n        site_profile.region_code,\n        site_profile.tier_level,\n        tariff_catalog.category_code,\n        tariff_catalog.subcategory_code\n    FROM base_events\n    JOIN energy_db.site_profile site_profile\n        ON base_events.entity_id = site_profile.entity_id\n    LEFT JOIN energy_db.tariff_catalog tariff_catalog\n        ON base_events.dim_0_code = tariff_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM energy_db.legacy_meter_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.energy_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "error_message": "ParseException: mismatched input 'AS' expecting 'END' near 'value_tier'",
    "level1_error_type": "Grammar & Structure",
    "level2_error_type": "CASE Expression",
    "level3_error_type": "Missing END in CASE WHEN"
  },
  {
    "task_id": "energy-semantic-9",
    "type": "semantic",
    "domain": "energy",
    "ddl": "CREATE TABLE IF NOT EXISTS energy_db.meter_events (entity_id STRING, event_timestamp TIMESTAMP, dim_0_code STRING, dim_1_code STRING, dim_2_code STRING, dim_3_code STRING, dim_4_code STRING, dim_5_code STRING, dim_6_code STRING, dim_7_code STRING, dim_8_code STRING, dim_9_code STRING, metric_0_amount DOUBLE, metric_1_amount DOUBLE, metric_2_amount DOUBLE, metric_3_amount DOUBLE, metric_4_amount DOUBLE, metric_5_amount DOUBLE, metric_6_amount DOUBLE, metric_7_amount DOUBLE, metric_8_amount DOUBLE, metric_9_amount DOUBLE, metric_10_amount DOUBLE, metric_11_amount DOUBLE, metric_12_amount DOUBLE, metric_13_amount DOUBLE, status_flag STRING, event_date STRING);\nCREATE TABLE IF NOT EXISTS energy_db.site_profile (entity_id STRING, segment_code STRING, region_code STRING, tier_level STRING);\nCREATE TABLE IF NOT EXISTS energy_db.tariff_catalog (dim_0_code STRING, category_code STRING, subcategory_code STRING);\nCREATE TABLE IF NOT EXISTS energy_db.legacy_meter_events_snapshot (segment_code STRING, region_code STRING, category_code STRING, entity_count BIGINT, total_metric_0_amount DOUBLE, snapshot_date STRING);",
    "issue_sql": "-- ETL pipeline for energy daily aggregation\nWITH base_events AS (\n    SELECT\n        meter_events.entity_id,\n        meter_events.event_timestamp,\n        meter_events.dim_0_code,\n        meter_events.dim_1_code,\n        meter_events.dim_2_code,\n        meter_events.dim_3_code,\n        meter_events.dim_4_code,\n        meter_events.metric_0_amount,\n        meter_events.metric_1_amount,\n        meter_events.metric_2_amount,\n        meter_events.metric_3_amount,\n        meter_events.metric_4_amount,\n        meter_events.metric_5_amount,\n        meter_events.metric_6_amount,\n        meter_events.status_flag\n    FROM energy_db.meter_events meter_events\n    WHERE meter_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        site_profile.segment_code,\n        site_profile.region_code,\n        site_profile.tier_level,\n        tariff_catalog.category_code,\n        tariff_catalog.subcategory_code\n    FROM base_events\n    JOIN energy_db.site_profile site_profile\n        ON base_events.entity_id = site_profile.entity_id\n    JOIN energy_db.tariff_catalog tariff_catalog\n        ON base_events.dim_0_code = tariff_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM energy_db.legacy_meter_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.energy_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "reference_sql": "-- ETL pipeline for energy daily aggregation\nWITH base_events AS (\n    SELECT\n        meter_events.entity_id,\n        meter_events.event_timestamp,\n        meter_events.dim_0_code,\n        meter_events.dim_1_code,\n        meter_events.dim_2_code,\n        meter_events.dim_3_code,\n        meter_events.dim_4_code,\n        meter_events.metric_0_amount,\n        meter_events.metric_1_amount,\n        meter_events.metric_2_amount,\n        meter_events.metric_3_amount,\n        meter_events.metric_4_amount,\n        meter_events.metric_5_amount,\n        meter_events.metric_6_amount,\n        meter_events.status_flag\n    FROM energy_db.meter_events meter_events\n    WHERE meter_events.event_date BETWEEN '${start_date}' AND '${end_date}'\n      AND status_flag = 'ACTIVE'\n),\nenriched_events AS (\n    SELECT\n        base_events.entity_id,\n        base_events.event_timestamp,\n        base_events.dim_0_code,\n        base_events.dim_1_code,\n        base_events.dim_2_code,\n        base_events.dim_3_code,\n        base_events.dim_4_code,\n        COALESCE(base_events.metric_0_amount, 0) AS metric_0_amount,\n        COALESCE(base_events.metric_1_amount, 0) AS metric_1_amount,\n        COALESCE(base_events.metric_2_amount, 0) AS metric_2_amount,\n        COALESCE(base_events.metric_3_amount, 0) AS metric_3_amount,\n        COALESCE(base_events.metric_4_amount, 0) AS metric_4_amount,\n        COALESCE(base_events.metric_5_amount, 0) AS metric_5_amount,\n        COALESCE(base_events.metric_6_amount, 0) AS metric_6_amount,\n        site_profile.segment_code,\n        site_profile.region_code,\n        site_profile.tier_level,\n        tariff_catalog.category_code,\n        tariff_catalog.subcategory_code\n    FROM base_events\n    JOIN energy_db.site_profile site_profile\n        ON base_events.entity_id = site_profile.entity_id\n    LEFT JOIN energy_db.tariff_catalog tariff_catalog\n        ON base_events.dim_0_code = tariff_catalog.dim_0_code\n),\nwindowed_metrics AS (\n    SELECT\n        entity_id,\n        segment_code,\n        region_code,\n        category_code,\n        metric_0_amount,\n        metric_1_amount,\n        metric_2_amount,\n        metric_3_amount,\n        metric_4_amount,\n        metric_5_amount,\n        metric_6_amount,\n        ROW_NUMBER() OVER (PARTITION BY entity_id ORDER BY event_timestamp DESC) AS recency_rank,\n        SUM(metric_0_amount) OVER (PARTITION BY entity_id, segment_code) AS segment_total,\n        AVG(metric_1_amount) OVER (PARTITION BY region_code ORDER BY event_timestamp ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_avg_metric_1\n    FROM enriched_events\n),\naggregated AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        COUNT(DISTINCT entity_id) AS entity_count,\n        SUM(metric_0_amount) AS total_metric_0_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_0_amount ELSE NULL END) AS latest_metric_0_amount,\n        SUM(metric_1_amount) AS total_metric_1_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_1_amount ELSE NULL END) AS latest_metric_1_amount,\n        SUM(metric_2_amount) AS total_metric_2_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_2_amount ELSE NULL END) AS latest_metric_2_amount,\n        SUM(metric_3_amount) AS total_metric_3_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_3_amount ELSE NULL END) AS latest_metric_3_amount,\n        SUM(metric_4_amount) AS total_metric_4_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_4_amount ELSE NULL END) AS latest_metric_4_amount,\n        SUM(metric_5_amount) AS total_metric_5_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_5_amount ELSE NULL END) AS latest_metric_5_amount,\n        SUM(metric_6_amount) AS total_metric_6_amount,\n        MAX(CASE WHEN recency_rank = 1 THEN metric_6_amount ELSE NULL END) AS latest_metric_6_amount,\n        MAX(segment_total) AS segment_total,\n        AVG(rolling_avg_metric_1) AS avg_rolling_metric_1\n    FROM windowed_metrics\n    GROUP BY\n        segment_code,\n        region_code,\n        category_code\n),\nclassified AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count,\n        total_metric_0_amount,\n        total_metric_1_amount,\n        segment_total,\n        avg_rolling_metric_1,\n        CASE\n            WHEN total_metric_0_amount > 100000 THEN 'HIGH'\n            WHEN total_metric_0_amount > 10000 THEN 'MEDIUM'\n            ELSE 'LOW'\n        END AS value_tier,\n        CONCAT_WS('_', segment_code, region_code) AS segment_region_key,\n        ROUND(total_metric_1_amount / NULLIF(entity_count, 0), 2) AS per_entity_metric_1\n    FROM aggregated\n),\nlegacy_snapshot AS (\n    SELECT\n        segment_code,\n        region_code,\n        category_code,\n        entity_count AS entity_count,\n        total_metric_0_amount AS total_metric_0_amount,\n        'legacy' AS data_source\n    FROM energy_db.legacy_meter_events_snapshot\n    WHERE snapshot_date = '${end_date}'\n)\nINSERT OVERWRITE TABLE fake_base_test.energy_daily_summary\nPARTITION (processing_date = '${end_date}')\nSELECT\n    classified.segment_code,\n    classified.region_code,\n    classified.category_code,\n    classified.entity_count,\n    classified.total_metric_0_amount,\n    classified.total_metric_1_amount,\n    classified.segment_total,\n    classified.avg_rolling_metric_1,\n    classified.value_tier,\n    classified.segment_region_key,\n    classified.per_entity_metric_1,\n    'current' AS data_source\nFROM classified\nUNION ALL\nSELECT\n    legacy_snapshot.segment_code,\n    legacy_snapshot.region_code,\n    legacy_snapshot.category_code,\n    legacy_snapshot.entity_count,\n    legacy_snapshot.total_metric_0_amount,\n    0 AS total_metric_1_amount,\n    0 AS segment_total,\n    0 AS avg_rolling_metric_1,\n    'LEGACY' AS value_tier,\n    CONCAT_WS('_', legacy_snapshot.segment_code, legacy_snapshot.region_code) AS segment_region_key,\n    0 AS per_entity_metric_1,\n    legacy_snapshot.data_source\nFROM legacy_snapshot;",
    "user_query": "Some entities that should appear in the daily summary (e.g. those without a matching category record) are missing from the output entirely. Expected: all entities from the base events should be retained even if category enrichment is unavailable. Please fix the bug.",
    "level1_error_type": "Joins & Grouping",
    "level2_error_type": "JOIN Type Selection",
    "level3_error_type": "Using INNER JOIN when LEFT JOIN is needed"
  }
]