sql2parquet.py

import duckdb
import os
import sqlite3
import sys

def convert_sqlite_to_parquet_with_duckdb(sqlite_db_path, output_dir, tables_to_convert=None):
    """
    Converts tables from a SQLite database to Parquet files using DuckDB.

    Args:
        sqlite_db_path (str): The path to the SQLite database file.
        output_dir (str): The directory where the Parquet files will be saved.
        tables_to_convert (list, optional): A list of table names to convert. 
                                            If None, all tables will be converted.
    """
    # --- Step 1: Create the output directory if it doesn't exist ---
    if not os.path.exists(output_dir):
        os.makedirs(output_dir)
        print(f"Created output directory: {output_dir}")

    # --- Step 2: Connect to the SQLite database and get table names ---
    # We use the built-in sqlite3 library just to get the list of tables.
    # DuckDB will do the heavy lifting of reading the data.
    try:
        conn_sqlite = sqlite3.connect(sqlite_db_path)
        cursor = conn_sqlite.cursor()
        
        # Get a list of all tables. We exclude internal SQLite tables.
        cursor.execute("SELECT name FROM sqlite_master WHERE type='table' AND name NOT LIKE 'sqlite_%';")
        all_tables = [row[0] for row in cursor.fetchall()]
        
        conn_sqlite.close()
    except sqlite3.Error as e:
        print(f"Error connecting to SQLite database: {e}")
        return

    # --- Step 3: Use DuckDB to attach the SQLite database and convert tables ---
    
    # Connect to a DuckDB in-memory database.
    # The 'read_only=False' is important for writing the Parquet files.
    # You can also use a persistent DuckDB database file here for more complex workflows.
    con_duckdb = duckdb.connect(database=':memory:', read_only=False)

    try:
        # Load the SQLite extension. It's often autoloader, but this ensures it.
        con_duckdb.execute("INSTALL sqlite; LOAD sqlite;")
        
        # Attach the SQLite database. This makes all its tables available to DuckDB.
        con_duckdb.execute(f"ATTACH '{sqlite_db_path}' AS sqlite_db (TYPE sqlite);")
        
        # Determine which tables to convert
        if tables_to_convert is None:
            tables_to_process = all_tables
        else:
            tables_to_process = tables_to_convert
            # Filter to ensure the requested tables actually exist in the database
            tables_to_process = [table for table in tables_to_process if table in all_tables]
        
        if not tables_to_process:
            print("No tables found to convert.")
            return

        print(f"Found {len(tables_to_process)} tables to convert: {', '.join(tables_to_process)}")

        # Loop through the tables and use the efficient COPY command
        for table_name in tables_to_process:
            print(f"\nConverting table: '{table_name}'...")
            
            # Define the output path for the Parquet file
            parquet_file_path = os.path.join(output_dir, f'{table_name}.parquet')
            
            # Use the powerful COPY statement. This reads directly from the attached
            # SQLite table and writes directly to the Parquet file.
            # The 'FORMAT parquet' option specifies the output format.
            # You can also add options for compression (e.g., COMPRESSION 'zstd').
            # The query is `SELECT * FROM attached_db_alias.table_name`.
            copy_query = f"COPY (SELECT * FROM sqlite_db.\"{table_name}\") TO '{parquet_file_path}' (FORMAT parquet);"
            
            try:
                con_duckdb.execute(copy_query)
                print(f"  -> Successfully saved to '{parquet_file_path}'")
            except duckdb.Error as copy_err:
                print(f"  -> ERROR converting table '{table_name}': {copy_err}")
                print("     This may be due to unsupported data types or other schema issues.")

    finally:
        # --- Step 4: Clean up (detach the database and close the connection) ---
        con_duckdb.execute("DETACH sqlite_db;")
        con_duckdb.close()
        print("\nConversion process finished.")

# --- How to use the function ---
if __name__ == '__main__':
    # Replace 'your_large_database.db' with the path to your actual SQLite file
    database_file = sys.argv[1]
    
    # Replace 'path/to/your/output_folder' with your desired output directory
    output_folder = '.'
    
    # Call the function to perform the conversion
    # You can also specify a list of tables if you don't want to convert all of them.
    # For example: tables_to_convert=['table_1', 'another_large_table']
    convert_sqlite_to_parquet_with_duckdb(database_file, output_folder)