import duckdb
from config.settings import csv_path, out_parquet
from utils.csv_reader import icr_csv


def run(con: duckdb.DuckDBPyConnection) -> int:
    src = csv_path('veg').replace('\\', '/')
    out = out_parquet('veg').replace('\\', '/')

    con.execute(f"""
        COPY (
            SELECT
                TRY_CAST(CCPROD   AS INTEGER) AS ccprod,
                TRIM(CCPRODEDIZ)              AS ccprodediz,
                TRIM(DDATINIVALID)            AS ddatinivalid,
                TRIM(DDATFINVALID)            AS ddatfinvalid,
                TRIM(DESCR_VEG)               AS descr_veg
            FROM {icr_csv(con, src)}
            WHERE CCPROD IS NOT NULL
        ) TO '{out}' (FORMAT PARQUET, COMPRESSION ZSTD)
    """)
    return con.execute(f"SELECT COUNT(*) FROM '{out}'").fetchone()[0]