Merge pull request #68 from alphagov/attachments-extra

Extract attachment fields into additional searchable text
alphagov · Oct 26, 2023 · f2de0a6 · f2de0a6
2 parents 2d29d08 + 87dddff
commit f2de0a6
Show file tree

Hide file tree

Showing 2 changed files with 18 additions and 5 deletions.
diff --git a/lib/document_sync_worker/document/publish.rb b/lib/document_sync_worker/document/publish.rb
@@ -24,7 +24,7 @@ class Publish < Base
         $.details.metadata.hidden_indexable_content
         $.details.metadata.project_code
         $.details.acronym
-        $.details.attachments[*].title
+        $.details.attachments[*]['title','isbn','unique_reference','command_paper_number','hoc_paper_number']
       ].map { JsonPath.new(_1) }.freeze
       ADDITIONAL_SEARCHABLE_TEXT_VALUES_SEPARATOR = "\n".freeze
 
@@ -88,7 +88,10 @@ def url
 
       def additional_searchable_text
         values = ADDITIONAL_SEARCHABLE_TEXT_VALUES_JSON_PATHS.map { _1.on(document_hash) }
-        values.flatten.join(ADDITIONAL_SEARCHABLE_TEXT_VALUES_SEPARATOR)
+        values
+          .flatten
+          .compact_blank
+          .join(ADDITIONAL_SEARCHABLE_TEXT_VALUES_SEPARATOR)
       end
 
       def public_timestamp

diff --git a/spec/lib/document_sync_worker/document/publish_spec.rb b/spec/lib/document_sync_worker/document/publish_spec.rb
@@ -209,14 +209,24 @@
           {
             "details" => {
               "attachments" => [
-                { "title" => "A report" },
-                { "title" => "Another report" },
+                {
+                  "title" => "A report",
+                  "isbn" => "1234567890123",
+                  "unique_reference" => "ABCDEF",
+                  "command_paper_number" => "",
+                },
+                {
+                  "title" => "Another report",
+                  "isbn" => "",
+                  "command_paper_number" => "CPN1234",
+                  "hoc_paper_number" => "ADHOC",
+                },
               ],
             },
           }
         end
 
-        it { is_expected.to eq("A report\nAnother report") }
+        it { is_expected.to eq("A report\n1234567890123\nABCDEF\nAnother report\nCPN1234\nADHOC") }
       end
     end