INS-1515

Services/dataset.service.js

@@ -6,7 +6,10 @@ const mysql = require('../Components/mysql');
 const queryGenerator = require('./queryGenerator');
 const cacheKeyGenerator = require('./cacheKeyGenerator');
 const utils = require('../Utils');
-const { DATASET_RETURN_FIELDS } = require('../Utils/datasetFields.js');
+const {
+  DATASET_RETURN_FIELDS,
+  DATASET_SEARCH_RETURN_MAPPING
+} = require('../Utils/datasetFields.js');
 const FACET_FILTERS = [
   'dataset_source_repo',
   'primary_disease',
@@ -39,7 +42,7 @@ const search = async (searchText, filters, options) => {
     searchableText = utils.getSearchableText(sanitizedSearchText);
   }
 
-  query = queryGenerator.getSearchQueryV2(searchableText, filters, options, DATASET_RETURN_FIELDS);
+  query = queryGenerator.getSearchQueryV2(searchableText, filters, options, Object.keys(DATASET_SEARCH_RETURN_MAPPING));
 
   if (query == null) {
     return result;
@@ -71,32 +74,46 @@ const search = async (searchText, filters, options) => {
   }
 
   let datasets = searchResults.hits.hits.map((ds) => {
-    if (ds.inner_hits) {
-      const terms = Object.keys(ds.inner_hits);
-      const additionalHitsDict = {};
-      if (terms.length > 0) {
-        terms.forEach((t) => {
-          ds.inner_hits[t].hits.hits.forEach((hit) => {
-            if (!additionalHitsDict[hit._nested.offset]) {
-              additionalHitsDict[hit._nested.offset] = {};
-              additionalHitsDict[hit._nested.offset].source = hit._source;
-              additionalHitsDict[hit._nested.offset].highlight = [];
-            }
-            additionalHitsDict[hit._nested.offset].highlight = additionalHitsDict[hit._nested.offset].highlight.concat(hit.highlight['additional.attr_set.k']);
-          });
+    // const content = ds._source;
+    // const highlight = ds.highlight;
+
+    // Rename return fields and highlights according to mappings
+    const content = Object.keys(DATASET_SEARCH_RETURN_MAPPING).reduce((acc, key) => {
+      acc[DATASET_SEARCH_RETURN_MAPPING[key]] = ds._source[key];
+      return acc;
+    }, {});
+    const highlight = Object.keys(DATASET_SEARCH_RETURN_MAPPING).reduce((acc, key) => {
+      acc[DATASET_SEARCH_RETURN_MAPPING[key]] = ds.highlight[`${key}.search`];
+      return acc;
+    }, {});
+
+    if (!ds.inner_hits) {
+      return {content: content, highlight: highlight};
+    }
+
+    const terms = Object.keys(ds.inner_hits);
+    const additionalHitsDict = {};
+    if (terms.length > 0) {
+      terms.forEach((t) => {
+        ds.inner_hits[t].hits.hits.forEach((hit) => {
+          if (!additionalHitsDict[hit._nested.offset]) { // We currently don't use this code
+            additionalHitsDict[hit._nested.offset] = {};
+            additionalHitsDict[hit._nested.offset].source = hit._source;
+            additionalHitsDict[hit._nested.offset].highlight = [];
+          }
+          additionalHitsDict[hit._nested.offset].highlight = additionalHitsDict[hit._nested.offset].highlight.concat(hit.highlight['additional.attr_set.k']);
         });
-      }
-      const additionalHits = [];
-      for (let key in additionalHitsDict) {
-        const tmp = {};
-        tmp.content = additionalHitsDict[key].source;
-        tmp.highlight = {};
-        tmp.highlight['additional.attr_set.k'] = utils.consolidateHighlight(additionalHitsDict[key].highlight);
-        additionalHits.push(tmp);
-      }
-      return {content: ds._source, highlight: ds.highlight, additionalHits: additionalHits};
+      });
+    }
+    const additionalHits = [];
+    for (let key in additionalHitsDict) { // We currently don't use this code
+      const tmp = {};
+      tmp.content = additionalHitsDict[key].source;
+      tmp.highlight = {};
+      tmp.highlight['additional.attr_set.k'] = utils.consolidateHighlight(additionalHitsDict[key].highlight);
+      additionalHits.push(tmp);
     }
-    return {content: ds._source, highlight: ds.highlight};
+    return {content: content, highlight: highlight, additionalHits: additionalHits};
   });
   result.total = searchResults.hits.total.value;
   result.data = datasets;

Utils/datasetFields.js

@@ -1,10 +1,13 @@
+// Default sort field for dataset search
 const DATASET_DEFAULT_SORT_FIELD = 'dataset_title_sort';
-// Maps Dataset natural field names to property names
+
+// Dataset fields eligible for text search
 const DATASET_SEARCH_FIELDS = [
   // 'dataset_uuid',
   'dataset_source_repo.search',
   'dataset_title.search',
-  'description.search',
+  // 'description.search',
+  'description_anchorless.search',
   'experimental_approaches.search',
   'dataset_source_id.search',
   'dataset_source_url.search',
@@ -32,7 +35,11 @@ const DATASET_SEARCH_FIELDS = [
   'related_diseases.search',
   'related_terms.search',
 ];
+
+// Fields to highlight in dataset search results
 const DATASET_HIGHLIGHT_FIELDS = DATASET_SEARCH_FIELDS;
+
+// Fields to return in dataset search results
 const DATASET_RETURN_FIELDS = [
   // 'dataset_uuid',
   'dataset_source_repo',
@@ -65,6 +72,23 @@ const DATASET_RETURN_FIELDS = [
   'related_diseases',
   'related_terms',
 ];
+
+// Opensearch properties that need to be mapped to different return fields
+const DATASET_SEARCH_RETURN_MAPPING_EXCEPTIONS = {
+  'description_anchorless': 'description',
+};
+
+// Opensearch properties mapped to dataset search return fields
+const DATASET_SEARCH_RETURN_MAPPING = {
+  ...DATASET_RETURN_FIELDS.filter(field => !Object.values(DATASET_SEARCH_RETURN_MAPPING_EXCEPTIONS).includes(field)) // Exclude some fields
+  .reduce((acc, str) => ({ // By default, Opensearch property has same name as return field
+    ...acc,
+    [str]: str,
+  }), {}),
+  ...DATASET_SEARCH_RETURN_MAPPING_EXCEPTIONS, // Special fields are mapped here
+};
+
+// Map column names to properties for dataset CSV export
 const datasetFields = {
   'Dataset UUID': 'dataset_uuid',
   'Dataset Title': 'dataset_title',
@@ -98,5 +122,6 @@ module.exports = {
   DATASET_SEARCH_FIELDS,
   DATASET_HIGHLIGHT_FIELDS,
   DATASET_RETURN_FIELDS,
+  DATASET_SEARCH_RETURN_MAPPING,
   datasetFields,
 };