Validation regex pattern for branch_label display default values in Auspice config JSON is too strict

[huddlej]

Current Behavior

For seasonal flu builds, we need to annotate alternate clade labels including one called proposed_subclade. We want to show these proposed subclades by default in some analyses, so we need to set the branch_label key in the Auspice config JSON's display_defaults to the proposed_subclade field.

When I do this, augur export v2 fails during validation of the Auspice config JSON with the following error:

.display_defaults.branch_label "proposed_subclade" failed pattern validation for "^(none|[a-zA-Z0-9]+)$"

Expected behavior

Validation should allow hyphens, underscores, and spaces in the branch label. For example, proposed_subclade, proposed-subclade, and Proposed subclade are all valid branch labels for the final Auspice JSON, but they are only disallowed when selected as display defaults.

Possible solution

Update the regex in the Auspice config JSON schema to be: ^(none|[-_ a-zA-Z0-9]+)$.

[jameshadfield]

The display_defaults regex should match the regex of allowed keys for labels in the main schema:

"^(none|[a-zA-Z0-9]+)$" # current display_default.branch_label
"^[a-zA-Z0-9]+$": {     # currently allowed keys for labels as set on node.branch_attrs

which was the intention described in the commit message which updated the display default. (The addition of none in the former regex is nice as a hint, but we also should ensure you can't have a label key of "none"!)

However these regexes are overly restrictive - Auspice can (I think) display any string as label names/values:

Although the URL query may need some attention -- the above is ?branchLabel=[key] _ → ⁰ anything goes%3F (auspice correctly loads this from URL state, but I suspect it's invalid in some sense).

So I think the issue here requires a few things:

• decide what values you want to allow in auspice, potentially dictated by any URL query constraints
• Update the two regexes in the (separate) JSON schemas in augur
• Disallow none as a valid label key

[huddlej]

decide what values you want to allow in auspice, potentially dictated by any URL query constraints

I'd vote for at least hyphen, underscore, and spaces to be added to what's supported now. Is there any reason not to allow anything besides none, though? The URL encoding shouldn't be an issue...

[tsibley]

☝️