Fix 500 error in transcripts.

[auraz]

When ID was removed from unit html (in https://github.com/edx/edx-platform/pull/1680) , transcripts were left broken, as they use id (location) and get_item(item_location) to obtain transcripts files.

This is CAT-1 blades bug: https://edx-wiki.atlassian.net/browse/BLD-530 (Can't import or add transcripts for video)

In longer term - @cahrens what should be changed in transcripts files to support (https://github.com/edx/edx-platform/pull/1680) change?

@polesye can we add tests for this case?

@Lyla FYI

[cahrens]

Sorry about that-- I didn't realize that was how the ID was being obtained for transcripts. We need to change the transcripts python code to expect locators instead of locations. We actually have a story for that (STUD-870), which we should be working on soon.

[cahrens]

Is there a more explicit test that you can add that verifies the transcripts are working properly?

[auraz]

@cahrens Thank you. Great, that you have story for that.

[polesye]

@cahrens
If ajax request in some reason fails (for example, 500 error) error message will be shown on frontend.
This additional assertion checks if some error message is shown after each ajax request.
Is this test okay for you?

[cahrens]

I guess that is OK if there is no other way to tell if uploading and modifying the transcripts actually worked. I was hoping for an integration test that uploaded or modified transcripts, then verified in the video instance that the transcripts are correct after the change. But perhaps that is not easy to test?

[polesye]

The last test in the transcripts.feature should cover situation described by you.

[cahrens]

But the existing test did not catch the bug I introduced. Didn't I completely break adding and updating transcripts?

[polesye]

Good point. I have updated this test.
It should catch the bug now.

[cahrens]

👍 after rebase.