[On Hold] Do Not Allow Directories To Be Added

[mikaell]

Summary

• This pr solves Housekeeper being able to add directories.
• No support is given for directories in Housekeeper. Conceptually not decided how to handle. Now trying to add a dir will give an error

Testing

alt1 $ pytest test alt2 try to add a directory

$ housekeeper add file myBumdle my_directory
my_directory is a directory
Aborted!

Review

• [ ] code approved by
• [ ] tests executed by
• [ ] "Merge and deploy" approved by Thanks for filling in who performed the code review and the test!

This version is a:

• [ ] MAJOR - when you make incompatible API changes
• [ ] MINOR - when you add functionality in a backwards compatible manner
• [ ] PATCH - when you make backwards compatible bug fixes or documentation/instructions

[coveralls]

Coverage increased (+0.09%) to 88.508% when pulling f388b77e444e9c09ac8badb3067ae6529eb82b52 on fix-no_dirs-200908 into 0b8dda38cee1229f99412abe23b6a6f7b9e79d13 on master.

[Mropat]

Very nice solution! However, I think we could benefit from catching the problem at its core, in AddHandler.new_file(). This method is called when any file is being added by any means to Housekeeper. The checks on path existing and being a directory can be performed there, and an appropriate error can be raised - then caught in the CLI.

[mikaell]

However, I think we could benefit from catching the problem at its core, in AddHandler.new_file().

The implementation was done analogous with other sanity checks. For example: A. Check if the file/path exists. This is done in 'add.py'.

add.py:
    if not path.exists():
        click.echo(click.style(f"File {path} does not exist", fg="red"))
        context.abort()

B. Check if bundle exists

    bundle_obj = store.bundle(bundle_name)
    if bundle_obj is None:
        click.echo(click.style(f"unknown bundle: {bundle_name}", fg="red"))

AddHandler performs no sanity checks at the moment. I haven no problem moving them. But then every sanity check should be centralised. Personally I think user I/O errors should be caught early, in the code handling I/O, leaving internal representation clean.

[Mropat]

@mikaell AddhHandler.add_bundle is not raising any errors for bundle and is expected to behave this way right now by cg (where it is imported directly) unfortunately. This behavior for now is also what would be desirable to keep when adding bundle/version through cg when version contains files that are folders. Containing the check to CLI wont fix the biggest problem however, when the main mode of adding files that are folders to HK is by adding it as a part of a version in CG.

This means then, that raising error is not desirable, but only introduce a check for folder, print a log warning, and return nothing if folder is being added! I still think it would be best to handle once and for all in AddHandler, but perhaps @barrystokman and/or @moonso could have a different take on this

[mikaell]

Pausing the pull-request until the CLI rewrite is published.

[ChrOertlin]

closing - out of date - new issue created https://github.com/Clinical-Genomics/housekeeper/issues/153