PyPDF2 unable to compress pdf

I want to show an embed pdf on streamlit app which has a limitation of <2MB pdf size to be displayed.

So I am trying to compress the pdf file which a user uploads via st.file_uploader on streamit app using PyPDF2 package. Here's the code I used:

from PyPDF2 import PdfReader, PdfWriter
from io import BytesIO

def compress_pdf(pdf_file, target_size):
  # Load PDF using PyPDF2
    pdf_reader = PdfReader(pdf_file)

    # Compress PDF using PyPDF2
    output_pdf = BytesIO()
    pdf_writer = PdfWriter()
    for page in pdf_reader.pages:
        page.compress_content_streams()  # This is CPU intensive!
        pdf_writer.add_page(page)

    # Get the compressed PDF bytes
    pdf_writer.write(output_pdf)
    compressed_pdf_bytes = output_pdf.getvalue()
    print(len(compressed_pdf_bytes)) # Check output of compressed pdf

    return compressed_pdf_bytes

The above function takes in the file uploaded by the user on streamlit as below:

uploaded_file = st.sidebar.file_uploader("Upload a file", key= "uploaded_file")
print(st.session_state.uploaded_file)

if uploaded_file is not None:
    compressed_pdf_bytes= compress_pdf(uploaded_file, 2000000)

Even after doing all that, I see the file isn't compressed AT ALL.

This is the output in terminal as u see. The file uploaded (output of and the length of compressed_pdf_bytes is almost same.

#Output of actual file uploaded.
UploadedFile(id=6, name='abc.pdf', type='application/pdf', size=4588407)

#output of compressed file in bytes
4472714

If you want to try PyMuPDF: python -m pip install pymupdf.
Then do this

import fitz  # pymupdf

def compress_pdf(pdf_file, target_size):
    # Load PDF using pymupdf
    doc = fitz.open(pdf_file)
    compressed_pdf_bytes = doc.tobytes(
        deflate=True,
        garbage=4,
    )
    print(len(compressed_pdf_bytes)) # Check output of compressed pdf
    return compressed_pdf_bytes

The above solution is lossless!
You can also try to build subset fonts - if your PDFs contain fonts that are not subsetted, the gain may be significant.

If your 2 MB limit is a hard one and you are ready to sacrifice things: PyMuPDF also offers to remove images.
With some more programming effort, you can also replace images by their gray-scale equivalents.

Here is a snippet to remove all images from a PDF:

doc = fitz.open(pdf_file)
for page in doc:
    for xref in [item[0] for item in page.get_images()]:
        page.delete_image(xref)
doc.save("no-images.pdf", garbage=4, deflate=True)

You can try cpdf -squeeze in.pdf -o out.pdf which is lossless compression.

Generally speaking, if your file has been created competently, there should be little space to squeeze. Especially if it is mostly images and fonts.

It is not always possible to compress the file. At some point, it has its minimal form.

The pypdf docs about file size give some important hints. They especially point out that removing images can reduce the file size.