Add audiences and format to transformer for query field. Add to mappi…

…ng and relevancy factors
wellcomecollection · Jan 14, 2025 · e038202 · e038202
1 parent e53fce4
commit e038202
Show file tree

Hide file tree

Showing 4 changed files with 54 additions and 3 deletions.
diff --git a/api/src/queries/events.ts b/api/src/queries/events.ts
@@ -5,7 +5,16 @@ import { TermsFilter } from './common';
 export const eventsQuery = (queryString: string): QueryDslQueryContainer => ({
   multi_match: {
     query: queryString,
-    fields: ['id', 'query.title.*^100', 'query.caption.*^10'],
+    fields: [
+      'id',
+      'query.title.*^100',
+      'query.caption.*^10',
+      'query.series.*^80',
+      'query.series.contributors*^8',
+      'query.series.contributors.keyword^80',
+      'query.format.*^80',
+      'query.audiences.*^80',
+    ],
     operator: 'or',
     type: 'cross_fields',
     minimum_should_match: '-25%',

diff --git a/pipeline/src/indices/events.ts b/pipeline/src/indices/events.ts
@@ -90,6 +90,40 @@ export const mappings = {
             },
           },
         },
+        audiences: {
+          type: 'text',
+          fields: {
+            shingles: {
+              type: 'text',
+              analyzer: 'english_shingle_analyzer',
+            },
+            cased: {
+              type: 'text',
+              analyzer: 'english_cased_analyzer',
+            },
+            keyword: {
+              type: 'keyword',
+              normalizer: 'keyword_lowercase',
+            },
+          },
+        },
+        format: {
+          type: 'text',
+          fields: {
+            shingles: {
+              type: 'text',
+              analyzer: 'english_shingle_analyzer',
+            },
+            cased: {
+              type: 'text',
+              analyzer: 'english_cased_analyzer',
+            },
+            keyword: {
+              type: 'keyword',
+              normalizer: 'keyword_lowercase',
+            },
+          },
+        },
         'times.startDateTime': {
           type: 'date',
           format: 'date_optional_time',

diff --git a/pipeline/src/transformers/eventDocument.ts b/pipeline/src/transformers/eventDocument.ts
@@ -153,6 +153,8 @@ export const transformEventDocument = (
       ? { type: 'PrismicImage' as const, ...primaryImage.image }
       : undefined;
 
+  const series = transformSeries(document);
+
   const format = transformFormat(document);
 
   const locations = transformLocations(document);
@@ -181,14 +183,18 @@ export const transformEventDocument = (
         locations,
         interpretations,
         audiences,
-        series: transformSeries(document),
+        series,
         isAvailableOnline,
       },
       query: {
         linkedIdentifiers: linkedDocumentIdentifiers(document),
         title: asTitle(title),
         caption: primaryImage?.caption && asText(primaryImage.caption),
-        series: transformSeries(document),
+        series,
+        format: format.label,
+        audiences: audiences
+          .map(audience => audience.label)
+          .filter(isNotUndefined),
         times: {
           startDateTime: documentTimes
             .map(time => time.startDateTime)

diff --git a/pipeline/src/types/transformed/index.ts b/pipeline/src/types/transformed/index.ts
@@ -80,6 +80,8 @@ export type ElasticsearchEventDocument = {
     title: string;
     caption?: string;
     series: Series;
+    format?: string;
+    audiences?: string[];
     times: { startDateTime: Date[] };
   };
   filter: {