import warnings
from datetime import datetime

import dagster as dg
from dagster import EnvVar
from dagster_snowflake import SnowflakeResource

# Using asset key references to avoid import chain issues
from src.utils.automation_conditions import hourly_cron_with_eager_historical_backfill_condition
from src.utils.snowflake.constants import TIME_WINDOW_FRESHNESS_POLICY_WARN_1H_FAIL_2H, PartitionExpr, Warehouse
from src.utils.snowflake.query import JinjaSQLFormatter, PythonStringSQLFormatter

warnings.filterwarnings("ignore", category=dg.BetaWarning)

REC_RUN_START_DATE = datetime.strptime('2025-01-01', '%Y-%m-%d')
REC_RUN_TABLE_NAME = "REC_RUN"


class RecRunConfig(dg.Config):
    rec_run_table_name: str = REC_RUN_TABLE_NAME
    warehouse: str = Warehouse.DIM_HOOK_MEDIUM.value


@dg.asset(
    name="rec_run",
    description="Table for recommendation run data with metrics and performance information extracted from bk_nonrealtime_event.",
    group_name="recommendation",
    partitions_def=dg.HourlyPartitionsDefinition(start_date=REC_RUN_START_DATE, end_offset=-1),
    deps=[
        dg.AssetDep("bk_nonrealtime_event"),
    ],
    backfill_policy=dg.BackfillPolicy.multi_run(max_partitions_per_run=24*7),
    owners=["team:core-pod"],
    metadata={
        "database": EnvVar("SNOWFLAKE_DB").get_value(),
        "schema": EnvVar("SNOWFLAKE_SCHEMA").get_value(),
        "table_name": REC_RUN_TABLE_NAME,
        "data_start_date": REC_RUN_START_DATE.strftime("%Y-%m-%d"),
        "cluster_by": "[p_date, p_hour, user_id]",
        "partition_expr": PartitionExpr.HOURLY.value,
        "transient": True,
        "sla_minutes": 120,
    },
    automation_condition=hourly_cron_with_eager_historical_backfill_condition,
    freshness_policy=TIME_WINDOW_FRESHNESS_POLICY_WARN_1H_FAIL_2H,
)
def rec_run(context: dg.AssetExecutionContext, snowflake: SnowflakeResource, config: RecRunConfig) -> dg.MaterializeResult:
    run_id = context.run.run_id
    logger = dg.get_dagster_logger()
    jinja_formatter = JinjaSQLFormatter()
    python_formatter = PythonStringSQLFormatter()

    # Get partition time window for processing
    partition_start = context.partition_time_window.start
    partition_end = context.partition_time_window.end
    is_multi_partition_range = context.has_partition_key_range

    fetch_params = {
        "partition_start_date": partition_start.strftime("%Y-%m-%d"),
        "partition_end_date": partition_end.strftime("%Y-%m-%d"),
        "partition_start_hour": partition_start.hour,
        "partition_end_hour": partition_end.hour, # End hour for partition window
        "rec_run_table_name": config.rec_run_table_name,
    }

    logger.info(f"Processing rec_run for partition: {partition_start} to {partition_end}")
    logger.info(f"Fetch params: {fetch_params}")

    with snowflake.get_connection() as conn:
        cursor = conn.cursor()
        logger.info(f"Using warehouse {config.warehouse}")
        warehouse_query = python_formatter.load("src/utils/snowflake/queries/use_warehouse.sql", params={"warehouse": config.warehouse}, logger=logger)
        cursor.execute(warehouse_query)

        # 1. Ensure the target table exists
        logger.info(f"Creating table {config.rec_run_table_name}...")
        create_table_query = jinja_formatter.load("src/assets/snowflake/recommendation/rec_run/table.sql", params=fetch_params, logger=logger)
        cursor.execute(create_table_query)

        # 2. Delete existing data from the target table for the partition window
        logger.info(f"Deleting existing data from {config.rec_run_table_name} for partition window {partition_start} to {partition_end}.")
        delete_query = python_formatter.load("src/utils/snowflake/queries/delete_hourly_partitions.sql", params={**fetch_params, "delete_partition_table_name": config.rec_run_table_name}, logger=logger)
        cursor.execute(delete_query)

        # 3. Insert new rec_run data
        logger.info(f"Inserting new rec_run data into {config.rec_run_table_name}.")
        insert_query = jinja_formatter.load("src/assets/snowflake/recommendation/rec_run/insert_rec_run.sql", params=fetch_params, logger=logger)
        cursor.execute(insert_query)

        # 4. Commit the transaction
        conn.commit()
        rows_inserted = cursor.rowcount

    logger.info(f"Successfully processed partition. Inserted {rows_inserted} rows.")

    return dg.MaterializeResult(
        metadata={
            "run_id": dg.MetadataValue.text(run_id),
            "table_name": config.rec_run_table_name,
            "partition_time_window_start": dg.MetadataValue.text(partition_start.isoformat()),
            "partition_time_window_end": dg.MetadataValue.text(partition_end.isoformat()),
            "dagster/row_count": rows_inserted if not is_multi_partition_range else 0,
        },
    )