validate if a URL is an image

[jabbalaci]

I have a feature request: validate a URL if it's an image or not.

Rationale: I'm writing a scraper that extracts all links. From this list I want to filter out just the image links. I think it could be done simply by analyzing the URL string: if it contains .jpg or .png or .gif then it's an image.

Edit: here is my current solution

IMAGE_EXTENSIONS = ['jpg', 'jpeg', 'png', 'gif']    # can be extended

def is_image(path):
    """
    Path can be a URL or a local file. Decide if it's an image or not.
    """
    path = path.lower()
    for x in IMAGE_EXTENSIONS:
        if x in path:
            return True
    #
    return False

[kvesteri]

I'm fine with adding this as long as its named a bit differently, maybe omething like image_url. Also please not that your code has a bug it would return True for image url such as 'myjpg.doc'

[jabbalaci]

Let's add dot to the extensions then:

IMAGE_EXTENSIONS = ['.jpg', '.jpeg', '.png', '.gif']

I've already seen URLs like this: http://host/pic.jpg?w=20&h=20, thus instead of endswith I do substring match.

Do you want a pull request?

[kvesteri]

Yeah you can create a PR but you need to do much smarter url parsing. Consider also filename urls such as: http://host/my.jpg.xlsx

Or what if the hostname contains one the image extensions? This needs to be bulletproof.

[jabbalaci]

If you want a bulletproof solution, then:

1. if it's a local file, check the header of the file (see the file command)
2. if it's a URL, download the header (HEAD) and check the mimetype

Both of them are slow.

[kvesteri]

That's why this should be named image_url not image. If we are going to add image validator we should make it work as you suggested. image_url validator should only validate the url string using smart string parsing.

[jabbalaci]

Here is a gist of it: https://gist.github.com/jabbalaci/a1bf5cafc223ac887db0 . Maybe it would be better if you integrated it. You know better where you want it to be placed.

[Sinistral2099]

Thought I had it. Ran into problems. Still working on something for this.

[yozachar]

Today, URLs do not need to end in any specific extension to locate media. For example: https://images.unsplash.com/photo-1520446266423-6daca23fe8c7.

A plausible method to validate if the URL points to an image is to actually HTTP GET the content at the URL and analyse it (or perform and HTTP HEAD request and check mime type, as suggested in #13 (comment)). I think that is beyond the scope of the library, at least for now.