pikepdf-compress.py

"""Compute with pike-pdf color."""
import os
from PIL import Image
from pikepdf import Pdf, PdfImage, Name
import zlib

# input_file_path = "108.pdf"
output_file_path = "_pike_pdf.pdf"

resize_factor = 0.5
convert_to_gray_scale = True

example = Pdf.open(input_file_path, allow_overwriting_input=True)
computed_images: list[str] = []
# Grab all the pages and all the images in every page.
for page in example.pages:
    if imagelists := list(page.images.keys()):
        for image in imagelists:
            rawimage = page.images[image]
            # compute images > 300kB
            if rawimage.Length > 50000 and image not in computed_images:
                computed_images.append(image)
                pdfimage = PdfImage(rawimage)
                pillowimage = pdfimage.as_pil_image()
                width, height = pillowimage.size
                width, height = int(width * resize_factor), int(height * resize_factor)
                if (height, width) != (0,0):
                    pillowimage = pillowimage.resize((width, height), Image.HAMMING)
                    if convert_to_gray_scale:
                        pillowimage = pillowimage.convert('L')
                    rawimage.write(zlib.compress(pillowimage.tobytes()), filter=Name("/FlateDecode"))
                    rawimage.ColorSpace = Name("/DeviceGray")
                    rawimage.Width, rawimage.Height = width, height

example.save(output_file_path)

initial_size = os.path.getsize(input_file_path)
final_size = os.path.getsize(output_file_path)
ratio = 1 - (final_size / initial_size)
print("Compression by {0:.0%}.".format(ratio))
print("Final file size with pike is {0:.1f}MB".format(final_size / 1000000))
print("Done.")