Fixes #3354 - Handles FileNotFoundError for assets

[karlcow]

This PR fixes issue #3354

Proposed PR background

Basically in case the actual files where missing, we got a FileNotFound spiraling up and returning a 500.

This now will help a bit more by failing gracefully. The resource still does not exist but we will be able to see it in the network panel of devtools. The uris will carry a ?missing_file?

[miketaylr]

Thanks, this is a nice simple improvement.

How hard would it be to write some unit tests for this? Give it a fixture file, assert expected checksum, give it a non-existant file, assert missing_file returned?

[karlcow]

@miketaylr do you remember if there was a reason to put the get_checksum inside? it's slightly weird.

[karlcow]

This was done a long time ago
133cffe

:)

[miketaylr]

do you remember if there was a reason to put the get_checksum inside? it's slightly weird.

you might have left out a word. inside _____?

[karlcow]

@miketaylr do you remember if there was a rationale for doing this nesting instead of two flat functions?

def bust_cache(file_path):
    def get_checksum(file_path):
         …
         return
    return

[karlcow]

Let's see the test coverage.

coverage report -m webcompat/templates/__init__.py 
Name                              Stmts   Miss  Cover   Missing
---------------------------------------------------------------
webcompat/templates/__init__.py      75     19    75%   32-37, 42, 47-61, 96

We can do better.

Let me complete this PR. Do not merge for now.

[karlcow]

I have switched to draft to not break anything.

[karlcow]

(env) ~/code/webcompat.com-karlcow % coverage report -m webcompat/templates/__init__.py 
Name                              Stmts   Miss  Cover   Missing
---------------------------------------------------------------
webcompat/templates/__init__.py      79      1    99%   99

ok better the 1% missing is because we have one case which is not tested because the function doesn't make sense. I will open another issue for it.

[karlcow]

ok better the 1% missing is because we have one case which is not tested because the function doesn't make sense. I will open another issue for it.

opened in #3356

[miketaylr]

@miketaylr do you remember if there was a rationale for doing this nesting instead of two flat functions?

Ah, no -- no memory. Nest functions seems nicer and more pythonic.

[miketaylr]

Thanks @karlcow