如何高效交换含百万条记录的CSV文件指定行（保持列结构正确）

Got it, dealing with million-row CSVs means we can’t just load the whole file into memory—here are a few efficient, reliable approaches to swap specific rows while keeping your column structure intact:

Using AWK (Fastest for Large Files) #

AWK is built for processing large text files efficiently without hogging memory. This script handles CSV fields with embedded commas (quoted fields) and preserves your header row.

First, create a file named swap_rows.awk with this code:

BEGIN {
    # Properly handle quoted CSV fields (ignore commas inside quotes)
    FPAT = "([^,]+)|(\"[^\"]+\")"
    header_printed = 0
}
NR == 1 {
    # Print the header first
    print $0
    header_printed = 1
    next
}
NR == target_row1 {
    # Store the first target row
    row1 = $0
    next
}
NR == target_row2 {
    # Store the second target row, then print the first stored row
    row2 = $0
    print row1
    next
}
# Print all other rows as-is
NR != target_row1 && NR != target_row2 {
    print $0
}
# If the second target row was the last line, print it here
END {
    if (length(row2) > 0) print row2
}

Run it by replacing target_row1 and target_row2 with your actual row numbers (counting the header as row 1):

awk -v target_row1=1000 -v target_row2=500000 -f swap_rows.awk input.csv > output.csv

Using Python (Flexible & Safe for Complex CSVs) #

If you need more control or want to avoid edge cases with quoted fields/newlines, use Python’s built-in csv module—it handles CSV structure natively without parsing errors.

Here’s a reusable function:

import csv

def swap_csv_rows(input_path, output_path, row_num1, row_num2):
    # Ensure row_num1 is the smaller number to simplify logic
    if row_num1 > row_num2:
        row_num1, row_num2 = row_num2, row_num1
    
    stored_row1 = None
    stored_row2 = None

    with open(input_path, 'r', encoding='utf-8') as infile, \
         open(output_path, 'w', encoding='utf-8', newline='') as outfile:
        
        reader = csv.reader(infile)
        writer = csv.writer(outfile)

        # Write the header first
        header = next(reader)
        writer.writerow(header)

        current_row = 1  # Data rows start after the header
        for row in reader:
            current_row += 1
            if current_row == row_num1:
                stored_row1 = row
                continue
            elif current_row == row_num2:
                stored_row2 = row
                # Write the first stored row instead of the current one
                writer.writerow(stored_row1)
                continue
            # Write all other rows normally
            writer.writerow(row)
        
        # If the second target row was the last line, write it now
        if stored_row2 is not None:
            writer.writerow(stored_row2)

# Example usage: swap row 1000 and 500000 (header is row 1)
swap_csv_rows("large_input.csv", "swapped_output.csv", 1000, 500000)

Key Notes to Avoid Mistakes #

• Backup first! Always make a copy of your original CSV before running any batch operations—large files are hard to recover if something goes wrong.
• Double-check row numbers: Confirm whether your row count includes the header. The examples above treat the header as row 1; adjust the logic if you’re counting data rows starting at 1.
• Performance: AWK will outperform Python for extremely large files (10M+ rows), but Python’s csv module is safer for CSVs with messy fields (embedded commas, newlines, quotes).

内容的提问来源于stack exchange，提问作者Kamran Tariq