"""
gemma.py — produce gemma.parquet (Layer 1)

Layer 1: fedele alla sorgente CSV, solo normalizzazione colonne in uppercase.
La logica di business (flag '!' sui campi V_*) viene applicata in Layer 2
da Parquet_layer1ToParquet_layer2.
"""
import duckdb
from config.settings import csv_path, out_parquet
from utils.csv_reader import gemma_csv


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

    con.execute(f"""
        CREATE OR REPLACE TABLE _gemma AS
        SELECT * FROM {gemma_csv(src)}
    """)

    cols = [row[0] for row in con.execute("DESCRIBE _gemma").fetchall()]
    select_sql = ', '.join(f'"{c}" AS "{c.upper()}"' for c in cols)

    con.execute(f"""
        COPY (
            SELECT {select_sql}
            FROM _gemma
        ) TO '{out}' (FORMAT PARQUET, COMPRESSION ZSTD)
    """)
    con.execute("DROP TABLE IF EXISTS _gemma")
    return con.execute(f"SELECT COUNT(*) FROM '{out}'").fetchone()[0]