Script for merging all pdf inside a folder as a single pdf

merge_pdf.py

"""
Script for merging all pdf inside a folder as a single pdf

Requirement: PyPDF2
usage: python merge_pdf.py -d <input_directory> -o output.pdf

Author: Gtindo Dev (https://gtindo.dev)
License: ISC
"""


from PyPDF2 import PdfFileReader, PdfFileWriter

import argparse
import os

parser = argparse.ArgumentParser(description='Merge multiple pdf files')
parser.add_argument('--dir', '-d', help='Directory that contain pdf to join')
parser.add_argument('--output', '-o', help="Output pdf name")
args = parser.parse_args()


def is_pdf(filename):
    if filename.split(".")[1] == "pdf":
        return True

    return False


def join_by_directory(dir_path, output_file):
    pdf_writer = PdfFileWriter()

    try:
        directory = os.listdir(dir_path)

        print("[+] Start Merging files of {}".format(dir_path))

        for file in directory:
            if not is_pdf(file):
                continue

            pdf_path = os.path.join(dir_path, file)
            pdf_reader = PdfFileReader(pdf_path)

            print("[-] Merging file {}".format(file))

            for page in range(pdf_reader.getNumPages()):
                pdf_writer.addPage(pdf_reader.getPage(page))

        with open(output_file, 'wb') as out:
            pdf_writer.write(out)

        print("[x] Pdf joined successfully !")

    except FileNotFoundError:
        print("[Error] - Directory {} does not exists.".format(dir_path))
        exit(0)


if __name__ == '__main__':
    if args.output is None:
        print("Output file name is missing")
        exit(0)

    if args.dir is None:
        print("Directory path is missing")
        exit(0)

    join_by_directory(args.dir, args.output)