Merge pull request #241 from alphagov/full-taxonomy-tree

Implement support for deeply nested taxons
alphagov · Mar 15, 2024 · 9cda934 · 9cda934
2 parents 7d449fe + 0c9ef4c
commit 9cda934
Show file tree

Hide file tree

Showing 3 changed files with 83 additions and 15 deletions.
diff --git a/app/models/concerns/publishing_api/metadata.rb b/app/models/concerns/publishing_api/metadata.rb
@@ -9,6 +9,17 @@ module Metadata
     # paths in this list will have its `manual` field set accordingly.
     IMPLICIT_MANUAL_PATHS = %w[/service-manual].freeze
 
+    # 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 = [
+      # 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
+
     # Extracts a hash of structured metadata about this document.
     def metadata
       {
@@ -20,7 +31,7 @@ def metadata
         public_timestamp:,
         document_type: document_hash[:document_type],
         content_purpose_supergroup: document_hash[:content_purpose_supergroup],
-        part_of_taxonomy_tree: document_hash.dig(:links, :taxons) || [],
+        part_of_taxonomy_tree:,
         # Vertex can only currently boost on numeric fields, not booleans
         is_historic: historic? ? 1 : 0,
         government_name:,
@@ -72,6 +83,13 @@ def public_timestamp
       # rubocop:enable Rails/TimeZone
     end
 
+    def part_of_taxonomy_tree
+      TAXON_VALUES_JSON_PATHS
+        .flat_map { _1.on(document_hash) }
+        .compact
+        .uniq
+    end
+
     def historic?
       political = document_hash.dig(:details, :political) || false
       government = document_hash.dig(:expanded_links, :government)&.first

diff --git a/spec/integration/document_synchronization_spec.rb b/spec/integration/document_synchronization_spec.rb
@@ -32,7 +32,10 @@
           is_historic: 0,
           government_name: "2015 Conservative government",
           content_purpose_supergroup: "news_and_communications",
-          part_of_taxonomy_tree: %w[37d0fa26-abed-4c74-8835-b3b51ae1c8b2],
+          part_of_taxonomy_tree: an_array_matching(%w[
+            37d0fa26-abed-4c74-8835-b3b51ae1c8b2
+            f3bbdec2-0e62-4520-a7fd-6ffd5d36e03a
+          ]),
           organisations: %w[government-digital-service],
           locale: "en",
           debug: {
@@ -62,9 +65,12 @@
           document_type: "travel_advice",
           is_historic: 0,
           content_purpose_supergroup: "guidance_and_regulation",
-          part_of_taxonomy_tree: %w[
-            8f78544f-a4ed-46b4-8163-889679d119b9 71cd9f51-f492-4c3f-91ca-5ad694c26592
-          ],
+          part_of_taxonomy_tree: an_array_matching(%w[
+            91b8ef20-74e7-4552-880c-50e6d73c2ff9
+            862fdd81-0b52-41c1-9aa0-e208ac86b763
+            8f78544f-a4ed-46b4-8163-889679d119b9
+            71cd9f51-f492-4c3f-91ca-5ad694c26592
+          ]),
           organisations: %w[foreign-commonwealth-development-office],
           locale: "en",
           parts: [
@@ -122,11 +128,13 @@
           is_historic: 1,
           government_name: "2010 to 2015 Conservative and Liberal Democrat coalition government",
           content_purpose_supergroup: "news_and_communications",
-          part_of_taxonomy_tree: %w[
-            06ad07f7-1e79-462f-a192-6b2c9d92089c
+          part_of_taxonomy_tree: an_array_matching(%w[
+            6e3f3cfb-142a-41f4-a03a-fea504cc1f79
             ce9e9802-6138-4fe9-9f33-045ef213be29
+            f3bbdec2-0e62-4520-a7fd-6ffd5d36e03a
+            9597c30a-605a-4e36-8bc1-47e5cdae41b3
             3dbeb4a3-33c0-4bda-bd21-b721b0f8736f
-          ],
+          ]),
           organisations: %w[foreign-commonwealth-office],
           locale: "en",
           debug: {
@@ -274,10 +282,12 @@
           public_timestamp: 1_372_436_926,
           document_type: "worldwide_organisation",
           is_historic: 0,
-          part_of_taxonomy_tree: %w[
+          part_of_taxonomy_tree: an_array_matching(%w[
+            862fdd81-0b52-41c1-9aa0-e208ac86b763
             f1744c25-bbae-42d5-b0fa-452ccea8f802
+            91b8ef20-74e7-4552-880c-50e6d73c2ff9
             ca97c97d-30c3-4c31-86d5-a84fb37f919a
-          ],
+          ]),
           world_locations: %w[austria],
           content_purpose_supergroup: "other",
           locale: "en",
@@ -309,7 +319,12 @@
           is_historic: 0,
           government_name: "2010 to 2015 Conservative and Liberal Democrat coalition government",
           content_purpose_supergroup: "research_and_statistics",
-          part_of_taxonomy_tree: %w[f3caf326-fe33-410f-b7f4-553f4011c81e],
+          part_of_taxonomy_tree: an_array_matching(%w[
+            f3bbdec2-0e62-4520-a7fd-6ffd5d36e03a
+            e48ab80a-de80-4e83-bf59-26316856a5f9
+            f3f4b5d3-49c4-487b-bd5b-be75f11ec8c5
+            f3caf326-fe33-410f-b7f4-553f4011c81e
+          ]),
           organisations: %w[cabinet-office efficiency-and-reform-group government-digital-service],
           locale: "en",
           debug: {
@@ -378,7 +393,11 @@
           organisations: %w[foreign-commonwealth-development-office],
           document_type: "speech",
           is_historic: 0,
-          part_of_taxonomy_tree: %w[d6dba75a-42bd-4e1e-984c-2bddb6b41951],
+          part_of_taxonomy_tree: an_array_matching(%w[
+            f3bbdec2-0e62-4520-a7fd-6ffd5d36e03a
+            37d0fa26-abed-4c74-8835-b3b51ae1c8b2
+            d6dba75a-42bd-4e1e-984c-2bddb6b41951
+          ]),
           world_locations: %w[usa],
           topical_events: %w[her-majesty-queen-elizabeth-ii],
           content_purpose_supergroup: "news_and_communications",

diff --git a/spec/models/concerns/publishing_api/metadata_spec.rb b/spec/models/concerns/publishing_api/metadata_spec.rb
@@ -139,10 +139,41 @@
     describe "part_of_taxonomy_tree" do
       subject(:extracted_part_of_taxonomy_tree) { extracted_metadata[:part_of_taxonomy_tree] }
 
-      context "with a set of taxon links" do
-        let(:document_hash) { { links: { taxons: %w[0000 ffff] } } }
+      context "with a set of taxon links and their details in expanded links" do
+        let(:document_hash) do
+          {
+            expanded_links: {
+              taxons: [
+                { content_id: "0000" },
+                {
+                  content_id: "1111",
+                  links: {},
+                },
+                {
+                  content_id: "2222",
+                  links: { root_taxon: [{ content_id: "0000" }] },
+                },
+                {
+                  content_id: "3333",
+                  links: {
+                    parent_taxons: [
+                      {
+                        content_id: "4444",
+                        links: {
+                          parent_taxons: [
+                            { content_id: "5555" },
+                          ],
+                        },
+                      },
+                    ],
+                  },
+                },
+              ],
+            },
+          }
+        end
 
-        it { is_expected.to eq(%w[0000 ffff]) }
+        it { is_expected.to match_array(%w[0000 1111 2222 3333 4444 5555]) }
       end
 
       context "without taxon links" do