Fix "hybrid" autocomplete query for OTUs #4994

Author: kleintom

CHANGELOG.md

@@ -35,6 +35,7 @@ This project <em>does not yet</em> adhere to [Semantic Versioning](https://semve
 - Monograph facilitator: Determination label is not visible in Safari browser
 - Checklist importer crashing on empty `originalNameUsageID` in some cases
 - Source and Repository autocompletes sometimes miss results, repository usage counts were sometimes wrong [#4990]
+- OTU autocomplete sometimes returns unuseful results instead of an exact match, even when an exact match exists [#4994]
 
 [#4937]: https://github.com/SpeciesFileGroup/taxonworks/issues/4937
 [#4944]: https://github.com/SpeciesFileGroup/taxonworks/issues/4944
@@ -43,6 +44,7 @@ This project <em>does not yet</em> adhere to [Semantic Versioning](https://semve
 [#4978]: https://github.com/SpeciesFileGroup/taxonworks/issues/4978
 [#4983]: https://github.com/SpeciesFileGroup/taxonworks/issues/4983
 [#4990]: https://github.com/SpeciesFileGroup/taxonworks/issues/4990
+[#4994]: https://github.com/SpeciesFileGroup/taxonworks/issues/4994
 
 ## [0.63.1] - 2026-06-22
 

lib/queries/otu/autocomplete.rb

@@ -113,15 +113,20 @@ def otu_name_similarity
       end
 
       # For names like Tapinoma CASC_2231
+      #
+      # Note this intentionally does not use `terms` (Queries::Query#terms) -
+      # that always returns a 2-element array of wildcarded copies of the
+      # *whole* query_string, not the query_string split into words.
       def autocomplete_taxon_name_hybrid
-        if terms.length == 2
-          base_query
+        parts = query_string.to_s.split(/\s+/)
+        return nil unless parts.length == 2
+
+        genus_term, otu_term = parts
+
+        base_query
           .joins(:taxon_name)
-          .where('taxon_names.cached % ? AND otus.name % ?', terms.first, terms.second)
+          .where('taxon_names.cached % ? AND otus.name % ?', genus_term, otu_term)
           .order('taxon_names.cached, otus.name, length(taxon_names.cached), length(otus.name)')
-        else
-          nil
-        end
       end
 
       # @return [Scope]

spec/lib/queries/otu/autocomplete_spec.rb

@@ -231,8 +231,44 @@
       expect(query.autocomplete.first).to eq(otu2)
     end
 
+    # Regression: `terms` (Queries::Query#terms) always returns a 2-element
+    # array of wildcarded copies of the *whole* query_string, never the
+    # query_string split on whitespace. `autocomplete_taxon_name_hybrid` was
+    # (mis)written assuming `terms.first`/`terms.second` were the genus and
+    # otu.name (e.g. "Tapinoma CASC_2231") words, so `terms.length == 2` was
+    # vacuously true for any query, and it trigram-matched the *entire*
+    # phrase against both `taxon_names.cached` and `otus.name` instead of
+    # matching each word against its own column.
+    context '#autocomplete_taxon_name_hybrid' do
+      let!(:tapinoma) { Protonym.create!(name: 'Tapinoma', rank_class: Ranks.lookup(:iczn, 'genus'), parent: root) }
+      let!(:target) { Otu.create!(taxon_name: tapinoma, name: 'CASC_2231', project_id: project_id) }
+
+      specify 'matches genus term to taxon_names.cached and otu term to otus.name, independently' do
+        # This otu.name redundantly restates the genus (as happens in real
+        # data, e.g. "Scaphoideus menoni_E26" under genus "Scaphoideus").
+        # Under the old whole-phrase bug this shared genus text was enough
+        # to pass both trigram conditions even though the otu-specific part
+        # ("sp2_Z9") has nothing to do with the query's otu term
+        # ("CASC_2231").
+        decoy = Otu.create!(taxon_name: tapinoma, name: 'Tapinoma sp2_Z9', project_id: project_id)
+
+        q = Queries::Otu::Autocomplete.new('Tapinoma CASC_2231', project_id: project_id)
+        r = q.autocomplete_taxon_name_hybrid.to_a
+
+        expect(r).to include(target)
+        expect(r).not_to include(decoy)
+      end
 
+      specify 'does not run for single-word queries' do
+        q = Queries::Otu::Autocomplete.new('Tapinoma', project_id: project_id)
+        expect(q.autocomplete_taxon_name_hybrid).to be_nil
+      end
 
+      specify 'does not run for queries with more than two words' do
+        q = Queries::Otu::Autocomplete.new('Tapinoma CASC 2231', project_id: project_id)
+        expect(q.autocomplete_taxon_name_hybrid).to be_nil
+      end
+    end
 
   end