perf: Switch most of the ES query to filter context

helpers/deduplicates/helpers.js

@@ -23,6 +23,8 @@ module.exports = {
   buildDuplicatesFromEsHits,
   buildSources,
   unwrapEsHits,
+  unwrapResult,
+  getHits,
 };
 
 /*
@@ -187,10 +189,18 @@ function buildDuplicateFromDocObject (docObject, sessionName, rules) {
   );
 }
 
+function unwrapResult (result) {
+  return unwrapEsHits(getHits(result));
+}
+
 function unwrapEsHits (hits) {
   return hits.map(unwrapEsHit);
 }
 
 function unwrapEsHit (hit) {
   return hit._source;
 }
+
+function getHits (result) {
+  return result.body.hits.hits;
+}

index.js

@@ -42,7 +42,7 @@ function deduplicate (docObject) {
       const ignoredFields = await searchIgnoredBySourceUid(docObject.sourceUid);
       const request = buildQuery(docObject, ignoredFields);
 
-      if (request.query.bool.should.length === 0) {
+      if (request.query.bool.filter.bool.should.length === 0) {
         business.emit('info', `Not deduplicable {docObject}, internalId: ${docObject.technical.internalId}`);
         docObject.business.isDeduplicable = false;
         return updateDuplicatesGraph(docObject, currentSessionName);

package.json

@@ -26,6 +26,7 @@
     "colors": "^1.4.0",
     "corhal-config": "github:istex/corhal-config",
     "debug": "^4.1.1",
+    "elastic-builder": "^2.16.0",
     "fs-extra": "^7.0.1",
     "lodash": "^4.17.21",
     "nanoid": "^2.0.1",

src/deduplicateQueryBuilder.js

@@ -12,14 +12,22 @@ function buildQuery (docObject, ignoredFields = []) {
       if (_.includes(scenario[docObject.business.duplicateGenre], rule.rule) &&
           validateRequiredAndForbiddenParameters(docObject, rule, ignoredFields)
       ) {
-        request.query.bool.should.push(buildQueryFromRule(docObject, rule));
+        request.bool.should.push(buildQueryFromRule(docObject, rule));
       }
     });
   }
 
-  request.query.bool.must_not = { term: { _id: docObject.technical.internalId } };
+  request.bool.must_not = { term: { _id: docObject.technical.internalId } };
 
-  return request;
+  return {
+    query: {
+      bool: {
+        filter:
+        request,
+
+      },
+    },
+  };
 }
 
 function validateRequiredAndForbiddenParameters (docObject, rule, ignoredFields = []) {