Check whether a PDF file is valid with Python

Question

I get a file via a HTTP upload and need to make sure its a PDF file. The programing language is Python, but this should not matter.

I thought of the following solutions:

1. Check if the first bytes of the string are %PDF. This is not a good check but prevents the user from uploading other files accidentally.

2. Use libmagic (the file command in bash uses it). This does exactly the same check as in (1)

3. Use a library to try to read the page count out of the file. If the lib is able to read a page count it should be a valid PDF file. Problem: I don't know a Python library that can do this

Are there solutions using a library or another trick?

Answer 1

In case anyone else is having a similar setup as me using PyMuPDF (imported as fitz package) and FastAPI, this works out for me

import fitz
import os
from FastAPI import HTTPException
import tempfile

@app.post("/upload")
def upload(file: UploadFile = File(...)):
    """Upload file endpoint."""

    # 1. Check if the file is a PDF
    if file.content_type != "application/pdf":
        raise HTTPException(
            status_code=400,
            detail="File is not a PDF!",
        )
    
    # 2. Check that the file ending is correct
    if not file.filename.endswith(".pdf"):
        raise HTTPException(
            status_code=400,
            detail="Filename does not end as PDF!",
        )

    # 3. Check reading it with fitz throws no error
    try:
        # Save the uploaded file temporarily
        with tempfile.NamedTemporaryFile(delete=False) as tmp:
            tmp.write(file.file.read())
            # Set the file pointer to the beginning of the file to be able to read it again later on
            file.file.seek(0)
            tmp_path = tmp.name

        # Check if the file size is greater than 0 (i.e., not empty)
        if os.path.getsize(tmp_path) == 0:
            raise HTTPException(status_code=400, detail="Uploaded file is empty")

        # Open the file to check if it is a valid PDF via the fitz library
        doc = fitz.open(tmp_path)
        doc.close()

        # Clean up the temporary file
        os.unlink(tmp_path)
    except HTTPException as e:
        raise HTTPException(
            status_code=400, detail="Uploaded file is not a valid PDF"
        ) from e

Answer 2

By valid do you mean that it can be displayed by a PDF viewer, or that the text can be extracted? They are two very different things.

If you just want to check that it really is a PDF file that has been uploaded then the pypdf solution, or something similar, will work.

If, however, you want to check that the text can be extracted then you have found a whole world of pain! Using pdftotext would be a simple solution that would work in a majority of cases but it is by no means 100% successful. We have found many examples of PDFs that pdftotext cannot extract from but Java libraries such as iText and PDFBox can.

Answer 3

I run into the same problem but was not forced to use a programming language to manage this task. I used pypdf but was not efficient for me as it hangs infinitely on some corrupted files.

However, I found this software useful till now.

Good luck with it.

https://sourceforge.net/projects/corruptedpdfinder/

Answer 4

If you're on a Linux or OS X box, you could use Pdftotext (part of Xpdf, found here). If you pass a non-PDF to pdftotext, it will certainly bark at you, and you can use commands.getstatusoutput to get the output and parse it for these warnings.

If you're looking for a platform-independent solution, you might be able to make use of pypdf.

Edit: It's not elegant, but it looks like pypdf's PdfReader will throw an IOError(22) if you attempt to load a non-PDF.

Answer 5

The two most commonly used PDF libraries for Python are:

• pypdf
• ReportLab

Both are pure python so should be easy to install as well be cross-platform.

With pypdf it would probably be as simple as doing:

from pypdf import PdfReader
reader = PdfReader("upload.pdf")

This should be enough, but reader will now have the metadata and pages attributes if you want to do further checking.

As Carl answered, pdftotext is also a good solution, and would probably be faster on very large documents (especially ones with many cross-references). However it might be a little slower on small PDF's due to system overhead of forking a new process, etc.

Answer 6

The current solution (as of 2023) is to use pypdf and catch exceptions (and possibly analyze reader.metadata)

from pypdf import PdfReader
from pypdf.errors import PdfReadError

with open("testfile.txt", "w") as f:
    f.write("hello world!")

try:
    PdfReader("testfile.txt")
except PdfReadError:
    print("invalid PDF file")
else:
    pass

Answer 7

Here is a solution using pdfminersix, which can be installed with pip install pdfminer.six:

from pdfminer.high_level import extract_text

def is_pdf(path_to_file):
    try:
        extract_text(path_to_file)
        return True
    except:
        return False

You can also use filetype (pip install filetype):

import filetype

def is_pdf(path_to_file):
    return filetype.guess(path_to_file).mime == 'application/pdf'

Neither of these solutions is ideal.

1. The problem with the filetype solution is that it doesn't tell you if the PDF itself is readable or not. It will tell you if the file is a PDF, but it could be a corrupt PDF.
2. The pdfminer solution should only return True if the PDF is actually readable. But it is a big library and seems like overkill for such a simple function.

I've started another thread here asking how to check if a file is a valid PDF without using a library (or using a smaller one).

Answer 8

In a project if mine I need to check for the mime type of some uploaded file. I simply use the file command like this:

from subprocess import Popen, PIPE
filetype = Popen("/usr/bin/file -b --mime -", shell=True, stdout=PIPE, stdin=PIPE).communicate(file.read(1024))[0].strip()

You of course might want to move the actual command into some configuration file as also command line options vary among operating systems (e.g. mac).

If you just need to know whether it's a PDF or not and do not need to process it anyway I think the file command is a faster solution than a lib. Doing it by hand is of course also possible but the file command gives you maybe more flexibility if you want to check for different types.