Use filter_path for improving the search helpers performances (#1199)

lib/Helpers.js

@@ -4,6 +4,8 @@
 
 'use strict'
 
+/* eslint camelcase: 0 */
+
 const { promisify } = require('util')
 const { ResponseError, ConfigurationError } = require('./errors')
 
@@ -29,11 +31,14 @@ class Helpers {
   /**
    * Runs a search operation. The only difference between client.search and this utility,
    * is that we are only returning the hits to the user and not the full ES response.
+   * This helper automatically adds `filter_path=hits.hits._source` to the querystring,
+   * as it will only need the documents source.
    * @param {object} params - The Elasticsearch's search parameters.
    * @param {object} options - The client optional configuration for this request.
    * @return {array} The documents that matched the request.
    */
   async search (params, options) {
+    appendFilterPath('hits.hits._source', params, true)
     const response = await this[kClient].search(params, options)
     return this[kGetHits](response.body)
   }
@@ -63,6 +68,8 @@ class Helpers {
       options.ignore = [429]
     }
     params.scroll = params.scroll || '1m'
+    appendFilterPath('_scroll_id', params, false)
+    const { method, body, index, ...querystring } = params
 
     let response = null
     for (let i = 0; i < maxRetries; i++) {
@@ -74,33 +81,31 @@ class Helpers {
       throw new ResponseError(response)
     }
 
-    let scrollId = response.body._scroll_id
+    let scroll_id = response.body._scroll_id
     let stop = false
     const clear = async () => {
       stop = true
       await this[kClient].clearScroll(
-        { body: { scroll_id: scrollId } },
+        { body: { scroll_id } },
         { ignore: [400] }
       )
     }
 
-    while (response.body.hits.hits.length > 0) {
-      scrollId = response.body._scroll_id
+    while (response.body.hits && response.body.hits.hits.length > 0) {
+      scroll_id = response.body._scroll_id
       response.clear = clear
       response.documents = this[kGetHits](response.body)
 
       yield response
 
-      if (!scrollId || stop === true) {
+      if (!scroll_id || stop === true) {
         break
       }
 
       for (let i = 0; i < maxRetries; i++) {
         response = await this[kClient].scroll({
-          scroll: params.scroll,
-          body: {
-            scroll_id: scrollId
-          }
+          ...querystring,
+          body: { scroll_id }
         }, options)
         if (response.statusCode !== 429) break
         await sleep(wait)
@@ -120,11 +125,14 @@ class Helpers {
    * }
    * ```
    * Each document is what you will find by running a scrollSearch and iterating on the hits array.
+   * This helper automatically adds `filter_path=hits.hits._source` to the querystring,
+   * as it will only need the documents source.
    * @param {object} params - The Elasticsearch's search parameters.
    * @param {object} options - The client optional configuration for this request.
    * @return {iterator} the async iterator
    */
   async * scrollDocuments (params, options) {
+    appendFilterPath('hits.hits._source', params, true)
     for await (const { documents } of this.scrollSearch(params)) {
       for (const document of documents) {
         yield document
@@ -428,4 +436,14 @@ class Helpers {
   }
 }
 
+function appendFilterPath (filter, params, force) {
+  if (params.filter_path !== undefined) {
+    params.filter_path += ',' + filter
+  } else if (params.filterPath !== undefined) {
+    params.filterPath += ',' + filter
+  } else if (force === true) {
+    params.filter_path = filter
+  }
+}
+
 module.exports = Helpers