Added mimetype to file data sent to model

[ajbozarth]

This is a follow up to IBM/MAX-Image-Caption-Generator#9 which added file extension verification using mimetype

Closes #38

[ptitzler]

Have you tested what happens when no mime-type can be guessed?

[ajbozarth]

I tested with the three valid extensions, any other extension would be rejected before getting to this line. Given this library guesses mime type by the file extension (IIUC), that should be enough. I originally used magic to do this (more robust, not a guess), but it requires installs outside pip, which I didn't want to add.

[ptitzler]

If I'm not mistaken the file handle returned by open is never closed ...

[ajbozarth]

Nice catch, odd that no one ever caught that before, not that it matter much given this file is opened in a thread. I'll close it.

[MLnick]

I see the Python 2.7 build on Travis failed to produce output. Do we know the issue?

[MLnick]

Changes LGTM

[stevemar]

@MLnick the travis job failed in a weird way, i'm restarting it:

No output has been received in the last 10m0s, this potentially indicates a stalled build or something wrong with the build itself.
Check the details on how to adjust your build configuration on: https://docs.travis-ci.com/user/common-build-problems/#Build-times-out-because-no-output-was-received
The build has been terminated

[stevemar]

If you use with it'll be closed at the end of the block. See the last paragraph of https://docs.python.org/2/tutorial/inputoutput.html#methods-of-file-objects

    with open(img_path, 'rb') as img_file:
        file_form = {'image': (img_path, img_file, mime_type)}

     r = requests.post(url=ml_endpoint, files=file_form)
     cap_json = r.json()

[ajbozarth]

Updated to use with per @stevemart comment

[stevemar]

reading the docs, it's fine to assume [0] here. cool.

https://docs.python.org/2/library/mimetypes.html#mimetypes.guess_type

[ajbozarth]

yeah I looked into that to make sure and the return is always a "pair"

[stevemar]

this LGTM

[stevemar]

i'm not sure if requests will like mime_type set to None if it can't guess the mimetype correctly.

[ajbozarth]

the current default is None which is what causes the error. And IIUC mimetypes.guess_type() uses the file extension in the filename to guess, and jpg, jpeg, and png all work and thats all we need.

[stevemar]

you could handle the case where someone submits a file with no extension :) but we're bikeshedding now :)

[ajbozarth]

valid extensions are already checked elsewhere in both the python code that calls this function and on the html form client-side, so I think we're good