A Python script to convert .xlsx files to .csv with UTF-8 encoding, supporting optional output file specification and overwrite protection.

xlsx_to_csv.py

#!/usr/bin/env python3

"""
Convert an Excel (.xlsx) file to a CSV (UTF-8) file.

Usage:
    python3 xlsx_to_csv.py [--method {pandas,soffice}] input.xlsx [output.csv]

Options:
    -h, --help                  Show this help message and exit.
    --method {pandas,soffice}   Conversion method to use (default: pandas).

Features:
    - Accepts an input .xlsx file as an argument.
    - Optionally allows specifying an output .csv filename.
    - Defaults to using the same basename for the output file if not specified.
    - Outputs to stdout if no output filename is provided (only for pandas method).
    - Prompts before overwriting an existing file.
    - Uses the 'soffice' command (LibreOffice) when selected and available.
"""

import argparse
import os
import sys
import subprocess
import tempfile
import shutil
import pandas as pd

def convert_with_pandas(input_file: str, output_file: str | None = None) -> None:
    """Converts an Excel file to a CSV file using pandas."""
    try:
        df = pd.read_excel(input_file, engine='openpyxl')
        if output_file:
            if os.path.exists(output_file):
                confirm = input(f"File '{output_file}' exists. Overwrite? (y/n): ").strip().lower()
                if confirm != 'y':
                    print("Operation cancelled.")
                    return
            df.to_csv(output_file, index=False, encoding='utf-8')
            print(f"Converted '{input_file}' to '{output_file}'")
        else:
            df.to_csv(sys.stdout, index=False, encoding='utf-8')
    except Exception as e:
        print(f"Error during pandas conversion: {e}", file=sys.stderr)

def convert_with_soffice(input_file: str, output_file: str | None = None) -> None:
    """Converts an Excel file to a CSV file using the soffice command."""
    soffice_path = shutil.which('soffice')
    if not soffice_path:
        print("Error: 'soffice' command not found. Please install LibreOffice or select the pandas method.", file=sys.stderr)
        sys.exit(1)
    
    # Determine the output directory and target filename.
    input_basename = os.path.splitext(os.path.basename(input_file))[0] + ".csv"

    if output_file:
        outdir = os.path.abspath(os.path.dirname(output_file)) or os.getcwd()
        target_csv = os.path.join(outdir, input_basename)
        
        # Check if output_file exists and prompt for overwrite.
        if os.path.exists(output_file) and os.path.abspath(output_file) != os.path.abspath(target_csv):
            confirm = input(f"File '{output_file}' exists. Overwrite? (y/n): ").strip().lower()
            if confirm != 'y':
                print("Operation cancelled.")
                return

        try:
            # Run soffice conversion with specified output directory.
            subprocess.run([soffice_path, "--headless", "--convert-to", "csv", "--outdir", outdir, input_file],
                           check=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
        except subprocess.CalledProcessError as e:
            print(f"Error during soffice conversion: {e}", file=sys.stderr)
            return

        # If the generated file name does not match the desired output_file, rename it.
        abs_target_csv = os.path.abspath(target_csv)
        abs_output_file = os.path.abspath(output_file)
        if abs_target_csv != abs_output_file:
            try:
                os.replace(abs_target_csv, abs_output_file)
                target_csv = abs_output_file
            except Exception as e:
                print(f"Error renaming output file: {e}", file=sys.stderr)
                return
        print(f"Converted '{input_file}' to '{target_csv}'")
    else:
        # When no output file is provided, convert to a temporary directory and print to stdout.
        with tempfile.TemporaryDirectory() as tmpdir:
            try:
                subprocess.run([soffice_path, "--headless", "--convert-to", "csv", "--outdir", tmpdir, input_file],
                               check=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
            except subprocess.CalledProcessError as e:
                print(f"Error during soffice conversion: {e}", file=sys.stderr)
                return
            temp_csv = os.path.join(tmpdir, input_basename)
            try:
                with open(temp_csv, encoding='utf-8') as f:
                    sys.stdout.write(f.read())
            except Exception as e:
                print(f"Error reading temporary CSV file: {e}", file=sys.stderr)

def main() -> None:
    """Main function to handle argument parsing and execution."""
    parser = argparse.ArgumentParser(
        description="Convert an Excel (.xlsx) file to a CSV (UTF-8) file."
    )
    parser.add_argument("input_file", help="Path to the input .xlsx file.")
    parser.add_argument("output_file", nargs="?", help="Optional output .csv filename.")
    parser.add_argument("--method", choices=["pandas", "soffice"], default="soffice",
                        help="Conversion method to use (default: soffice).")
    
    args = parser.parse_args()
    
    input_file = args.input_file
    # For pandas method, output to stdout if output_file is not provided.
    output_file = args.output_file or (f"{os.path.splitext(input_file)[0]}.csv" if args.method == "soffice" else None)
    
    if args.method == "soffice":
        convert_with_soffice(input_file, output_file)
    else:
        convert_with_pandas(input_file, output_file)

if __name__ == "__main__":
    main()