"Apply Rule to Collection" -- py3 related?

[jennaj]

Tool: Apply Rule to Collection (Galaxy Version 1.0.0)
Tutorial: Problem at step 7 here: https://galaxyproject.github.io/training-material/topics/galaxy-data-manipulation/tutorials/upload-rules/tutorial.html#apply-rules-to-existing-collections
Reported at Galaxy Help: https://help.galaxyproject.org/t/problem-applying-regular-expression-wrong-number-of-groups-found-in-tutorial-collections-rule-based-uploader/2915
Reproducible: https://usegalaxy.org/u/jen/h/test-coll

Screenshots

Set it up:

Review:

Submit: ...job submission fails with a red pop-up

ping @mvdbeek @jmchilton @natefoo

[jennaj]

Found more about the problem.

When 2 columns are entered to break up the name into identifiers, that value does not persist.

1. Clicking back into the rule to edit it shows the value has flipped back to 1 (the default).
2. Try to apply the 1 == get an error (expected)
3. Change to 2 again, apply, no error in rule builder but from does not have the 3rd column "C" defined.

Swapping the column order around triggers the tool GUI rule summary to actually state "undefined" for column "C" WHEN it is the last identifier defined. If internal to the list, nothing prints out on tool form for the "C" identifier column (doesn't even print a space -- instead, prints nothing).

Submitting fails for all cases, no workaround, but also didn't exhaustingly test parsing out the A/B/C identifiers every way possible.

Source of the problem??

Start from end of the last query in the original exact above, where the tool form is odd (missing "C" column as defined) and the job will not submit >> click back into "Edit" on the form, then click into the grouping by regex rule. Isn't 2 groups anymore, is a 1.

Removing the first default rule for identifier0, get an error in the rule builder. Expected -- all columns undefined and error for reg expression rule to capture groups.

Column "C" as first column set to an identifier.

Note how the column doesn't have a "name". Same as when "C" is any other internal column (is the second column in tuto method at this point).

Column "C" as last (third) set to an identifier.

Note how the column now has a "name" printed out to the tool form at least (even though undefined and unusable).

I hope that helps to track down the issue. I am not sure yet if usegalaxy.org (at py3) only or not. Testing on usegalaxy.eu next. If on both, will assume is a /galaxy issue and will put a ticket there & link it back in here.

[jennaj]

Same problem at usegalaxy.org.

The number of "groups" falls back to the default 1 value

It doesn't matter how many groups are parsed out. Only the first two A and B column get defined (due to "groups" being 1). Any in the middle = nothing printed. Any at the end == undefined exposed.

Few screenshots

extended regex + other assignments

how those rules end up on the tool form

Looks like a Galaxy problem. Reporting to /galaxy repo.

[jennaj]

@jennaj Retest based on galaxyproject/galaxy#10390