Indistinct search_names

[JoeCohen]

1. Prevent creating a Name (or editing a Name to create) a search_name which is the identical to, or only trivially different from, an existing search_name. I.e., different at most only in: diacritical, punctuation, case.
2. Possibly move some existing validators from app/models/name.rb to app/models/name/validation.rb.
3. Fix (merge) existing examples in the live db.

Tasks

• Validator preventing saving of indistinct Names
• search_names for all fixtures, fixing any Failures
• validate presence of search_name
• fix any errors caused by above presence validators
• remove temporary hack which ignored blank search_name
• Fix any test issues caused by above removal.
• review Rails Active Record Validations Guide
• Can I move existing existing validators from app/models/name.rb to app/models/name/validation.rb.?
• Make sure error message/flash are correct
• Improve author ending test. See Comment below
• add author content test: only alphabetical (including diacriticals), comma, period, open/close parens, ampersand
• More author ending tests
• Fix author ending message
• Fix this: https://github.com/MushroomObserver/mushroom-observer/actions/runs/13315844899/job/37189513589
• See if I can simplify form of existing Name validators
• Improve Name create form instructions
  • omit author of protocol
  • omit publication
  • omit "in"
• Make sure new validators prevent API from saving bad Names
• Can I DRY API?
• Deep Rubocop
• Add suggested manual tests to PR
• PR ready for review
• Add questions/notes:
  • full author can include [];
  • advantage of doing validations in model;
  • translated message.
• Request review
• Optional: Fix (merge) existing examples in the live db.

See 2025-02-09 Slack message:

Kuhner vs. Kühner is real. (Not just a movie with Meryl Strœp.)
As of yesterday’s db backup, there were 6 homonyms whose search_name had an exact or near miss.
Lepiota rhacodioides - exact duplicates
Cortinarius amigochrous - Kuhner vs. Kühner
Entoloma prismatospermum - trailing comma
Lactarius subvernalis - trailing comma
Macrolepiota bonaerensis - trailing comma
Melanoleuca phaeopodia - trailing comma
I suggest:
Add validator(s) which prevent creating a Name (or editing a Name to create) exact matches or near misses for search_name.
After deploy, merge the listed homonyms
I think the validators should be before_save validators, rather than in #create or #edit. Else the API could create a bad Name. (This would also let us dry the API a bit.)
There also might be some validators in name.rb which could be moved to app/models/name/validation.rb.

[nimmolo]

I don't feel qualified to comment on anything nomenclature-related, but this feels right to me.

[mo-nathan]

Thanks for doing all this work! Sounds reasonable to me. The trailing comma case is interesting. In my view, authors should only end in a letter or a period. There are a bunch of other cases of trailing commas as well as a few random ")"s, years, and alt spellings in square brackets. These just seem to create more chances for duplicates.

[JoeCohen]

@mo-nathan: Thanks for the encouragement and the comment.

• Re "doing all this work". I haven't done (or at least pushed) anything yet. The author ending stuff is old.
• Looks like this approach leads into a rabbit hole. I don't know if it's high-enough priority to dig into.
  • I wrote a validator (and a test), but rails t gets 82 failures, 3 errors.
  • Off the top of my head, my guess is that the cause is incomplete fixtures (without search_name) and tests which save Names which are incomplete or otherwise problematic.
  • I will investigate further.
• I agree with your comments about author ending. I'll have to clean up the db. (This stuff is the result of careless pasting from IF/MB.)