Upload a valid image to an s3 bucket with its url using the requests package. Catches corrupt images using the pillow package.

upload_image_with_url.py

from copy import deepcopy
import boto3
import requests
import environ
from PIL import Image
import io

env = environ.Env()

# Download an image with the url and store it in s3.
def upload_image_with_url(url, filepath):
    try:
        headers = {
            "User-Agent": "Mozilla",
        }
        request = requests.get(url, stream=True, headers=headers, timeout=5)

        # Checks for resource not found errors.
        request.raise_for_status()

        image = Image.open(request.raw)

        # Grab the image buffer to make a deepcopy.
        # - Because image.verify invalidates the buffer and reference.
        # - Cannot deepcopy request.raw because it cannot be serialized.
        image_buffer = io.BytesIO()
        image.save(image_buffer, format=image.format)
        image_buffer.seek(0)
        image_buffer_copy = deepcopy(image_buffer)

        # Image needs to be opened again after save.
        image = Image.open(image_buffer)
        extension = image.format.lower()

        # Check if image is corrupted.
        image.verify()

        session = boto3.Session()
        s3 = session.resource("s3")

        # Construct the filepath with the extension.
        filepath = filepath + "." + extension
        bucket = s3.Bucket(env("AWS_STORAGE_BUCKET_NAME"))
        bucket.upload_fileobj(image_buffer_copy, filepath)

        image.close()
    except Exception as e:
        raise e