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('imdb').replace('\\', '/')
    out = out_parquet('imdb').replace('\\', '/')

    con.execute(f"""
        COPY (
            SELECT
                TRY_CAST(CODICE          AS INTEGER) AS codice,
                TRY_CAST(EDIZIONE        AS INTEGER) AS edizione,
                TRIM(RIFERIMENTO_IMDB)               AS riferimento_imdb
            FROM {icr_csv(con, src)}
            WHERE CODICE IS NOT NULL
        ) TO '{out}' (FORMAT PARQUET, COMPRESSION ZSTD)
    """)
    return con.execute(f"SELECT COUNT(*) FROM '{out}'").fetchone()[0]