Writing CSV Files

Once you've read a CSV, you'll usually process it somehow, then write the result back out in CSV.

The csv.DictWriter class writes a dictionary to a CSV file. Give it the output columns, write the header, then write rows:

def save_attendance(filename: str, attendees: list[dict[str, str]]) -> None:
    fieldnames = ["name", "role", "checked_in"]

    with open(filename, "w", newline="") as file:
        writer = csv.DictWriter(file, fieldnames=fieldnames)
        writer.writeheader()
        writer.writerows(attendees)

• fieldnames decides the column order
• writeheader() writes the first row
• writerows() writes many rows
• newline="" avoids blank-line weirdness on Windows

If you're generating results incrementally, the writerow() function can be used to write one row at a time in a loop:

for attendee in attendees:
    writer.writerow(attendee)

For large datasets, it's more memory-efficient to work with files (both reading and writing) by streaming rows instead of first collecting them in a list. Do you really want to load a 20GB CSV directly into RAM? Probably not.

def stream_inventory_flags(input_file: str, output_file: str) -> None:
    with (
        open(input_file, "r", newline="") as in_f,
        open(output_file, "w", newline="") as out_f,
    ):
        reader = csv.DictReader(in_f)
        writer = csv.DictWriter(out_f, fieldnames=["sku", "needs_restock"])
        writer.writeheader()

        for row in reader:
            writer.writerow(
                {
                    "sku": row["sku"],
                    "needs_restock": int(row["quantity"]) < int(row["reorder_at"]),
                }
            )