Merge pull request #252 from alphagov/truncate-taxons

Fix sync failures due to excessive taxons

app/models/concerns/publishing_api/metadata.rb

@@ -12,13 +12,17 @@ module Metadata
     # Taxons can be deeply nested, so we need to make sure we extract all of their content IDs all
     # the way down.
     TAXON_VALUES_JSON_PATHS = [
+      # Root and parent taxons
+      #
+      # Note: these come *before* the direct taxons on purpose, as the top two levels of taxons are
+      # facetable, and we set a limit on the maximum number of taxons to index. This way, they are
+      # less at risk of being truncated.
+      "$.expanded_links.taxons..links.root_taxon[*].content_id",
+      "$.expanded_links.taxons..links.parent_taxons[*].content_id",
       # Direct taxons
       "$.expanded_links.taxons[*].content_id",
-      # Parent taxons
-      "$.expanded_links.taxons..links.parent_taxons[*].content_id",
-      # Root taxon (note: that's still an array!)
-      "$.expanded_links.taxons..links.root_taxon[*].content_id",
     ].map { JsonPath.new(_1, use_symbols: true) }.freeze
+    MAX_TAXON_COUNT = 250
 
     # Extracts a hash of structured metadata about this document.
     def metadata
@@ -88,6 +92,7 @@ def part_of_taxonomy_tree
         .flat_map { _1.on(document_hash) }
         .compact
         .uniq
+        .first(MAX_TAXON_COUNT)
     end
 
     def historic?

spec/models/concerns/publishing_api/metadata_spec.rb

@@ -173,7 +173,21 @@
           }
         end
 
-        it { is_expected.to match_array(%w[0000 1111 2222 3333 4444 5555]) }
+        it { is_expected.to eq(%w[0000 4444 5555 1111 2222 3333]) }
+      end
+
+      context "with an excessive number of taxon links" do
+        let(:document_hash) do
+          {
+            expanded_links: {
+              taxons: Array.new(260) { { content_id: sprintf("%04d", _1) } },
+            },
+          }
+        end
+
+        it "is truncated to 250 taxons" do
+          expect(extracted_part_of_taxonomy_tree.count).to eq(250)
+        end
       end
 
       context "without taxon links" do