Khan Academy support

[mikedawson]

User story

As a system administrator or edtech developer I want to extract/convert individual Khan Academy exercises from ZIM files to a xAPI zip so that I can use Khan Academy lessons in Ustad Mobile or other edtech environment

(If work duration will exceed five days, must be broken down into multiple)

Background info on user story

• Perseus standalone
• Example lesson: https://library.kiwix.org/viewer#khanacademy_en_all_2023-03/3464a3cb0f29575a8aa5f31d89f0473e

Design tasks

• [ ] Command line interface

When design is complete: attach PDF of prototype that includes all screens added/modified here.

Implementation tasks

Each task max one day - otherwise should be split.

• [ ] Kotlin code that will output an experience API file based on a Kolibri channel ID and topic ID. Can be run on command line.
• [ ] Kotlin code that will list available channels - may require server url (e.g. default likely to be Kolibri studio)
• [ ] Kotlin code that will list available topics in given channel (with max recursion depth parameter)

When implementation is complete: developer to attach video showing completion of acceptance criteria

Acceptance criteria

• Can list available categories and units from a Khan Academy ZIM file and/or Kolibri channel
• Can extract a specified interactive exercise to static HTML/Javascript/CSS that will run standalone on any standard HTTP server (e.g. Apache) probably as a TinCan zip (Rustici launch method) as per xAPI docs
• Can extract a specified video (and any related subtitles). Video is simply saved as the video file, in the same format it was found in (e.g. no conversion/compression). Save any subtitles to the same (or other specified) folder.

[SamihMustafa]

Identified issue with zimdump: file and folder name conflicts. zimdump creates an exception folder with all the missed files. This is a known issue, tracked here: https://github.com/openzim/zim-tools/issues/414

[SamihMustafa]

Identified issue with zimdump: file and folder name conflicts. zimdump creates an exception folder with all the missed files. This is a known issue, tracked here: openzim/zim-tools#414

The proposed fix is to ensure each file has the correct extension, then move any missed files to the correct folder.

[mikedawson]

Behavior note: should have an optional command line parameter for the paths. If not manually specified, use the system path.

The logic for this is in SysPathUtil in the main Ustad repo. Normally, copy/paste is bad. I think in this case; there isn't really a practical alternative. ZIM2Xapi should not depend on the main Ustad repo.

[mikedawson]

Code review notes (as we don't have a pull request)

1. Package name should be set - e.g. com.ustadmobile.zim2xapi
2. Needs a README file with instructions on required dependencies to install. A few example commands in the README would be enough - users can get full command line detail/help using the help provided via CLI argument parser.
3. UseCases - constructor parameters should dependencies only (e.g. other use cases, serialization things, etc). Per invocation arguments (e.g. the file to process etc) should be arguments to invoke as per coding style. The UseCase is provided via DI (normally as a singleton), so per invocation arguments should not be part of the instance itself.
4. If something fails - let it throw an exception as per code style note just added. This should be caught eventually by the command line runner. E.g. if ExtractZimUseCase fails it needs to throw an exception ; otherwise when it is used programmatically in a chain, anything that is using it will continue as if everything was OK. That is not desirable. If a command fails (eg exit code is non zero), then we should also throw an exception.
5. DownloadKolibri2ZimUseCase - don't hardcode the use of docker. That could be a sensible default. Also ensure that per-invocation arguments are arguments to invoke, not constructor parameters. The UseCase could accept an argument (constructor parameter) for the command. Should be a string that can then be split up. E.g. default could be "docker run ...docker params"

[mikedawson]

Behavior note: path handling logic

If user has specified a path - use the expected name on it's own e.g. kolibri2zim

Split the path up by whitespace (e.g. if user specifies 'docker ... kolibri2zim')

For the first argument (e.g. the actual executable), if it is not a absolute path, use syspathutil to look for the absolute path

[mikedawson]

For input: allow specifying a Zim file or kolibri channel/topic (where we use kolibri2zim to get the zim).