python合并pdf

 

import os, time
from PyPDF2 import PdfFileReader, PdfFileWriter


def get_file_list(path):
    file_list = [os.path.join(root, filepath)
                 for root, dirs, files in os.walk(path)
                 for filepath in files
                 if str(filepath).endswith('.pdf')]
    return file_list if file_list else []


def merge_pdf(filepath, outfile):
    output = PdfFileWriter()
    output_pages = 0
    file_list = get_file_list(filepath)
    if file_list:
        for file in file_list:
            print('full path: %s' % file)
            # 读取源pdf文件
            input = PdfFileReader(open(file, 'rb'))

            # 获取源pdf文件页数
            page_count = input.getNumPages()
            output_pages += page_count
            print('page number: %d' % page_count)

            # 将page添加至output
            for i in range(page_count):
                output.addPage(input.getPage(i))

        print('total pages: %d' % output_pages)

        # 写入到目标pdf文件中
        output_stream = open(os.path.join(filepath, outfile), 'wb')
        output.write(output_stream)
        output_stream.close()
        print('merge pdf done')
    else:
        print('no pdf files')


def main():
    commence = time.time()
    file_path = r'e:\python'  # 存放合并pdf的文件夹
    output_file = 'merged_pdf.pdf'  # 合并后的pdf文件名
    merge_pdf(file_path, output_file)
    closure = time.time()
    print('elapsed %s seconds' % (closure - commence))


main()