手上有一個PDF想要餵給chatPDF,免費版限制不能超過10MB,用pypdf2 寫一個Script但是分割出來的檔案大小還比原本的大
Python3.8
謝謝你花時間 幫忙看程式互相學習
PDF檔案連結:https://www.dropbox.com/s/utn1v5dtp1dd7yt/infoblox_nios_8.6.pdf?dl=0
source code:

import os
from PyPDF2 import PdfFileReader, PdfFileWriter

def split_pdf(input_path, max_size):
    with open(input_path, 'rb') as input_file:
        input_pdf = PdfFileReader(input_file)
        num_pages = input_pdf.getNumPages()
        output = PdfFileWriter()
        current_size = 0
        current_page = 0
        while current_page < num_pages:
            output.addPage(input_pdf.getPage(current_page))
            current_size += input_pdf.getPage(current_page).getContents().get('/Length', 0)
            if current_size > max_size:
                output_path = os.path.splitext(input_path)[0] + '_part{}.pdf'.format(current_page // 100 + 1)
                with open(output_path, 'wb') as output_file:
                    output.write(output_file)
                output = PdfFileWriter()
                current_size = 0
            current_page += 1
        if not output.getNumPages() == 0:
            output_path = os.path.splitext(input_path)[0] + '_part{}.pdf'.format(current_page // 100 + 1)
            with open(output_path, 'wb') as output_file:
                output.write(output_file)

split_pdf('infoblox_nios_8.6.pdf', 9000000) # split into files of maximum 9MB