From f76edea5f84571fe844e5cf0e357ad64787ee2a0 Mon Sep 17 00:00:00 2001
From: Chocobozzz <me@florianbigard.com>
Date: Fri, 23 Oct 2020 10:14:44 +0200
Subject: [PATCH] Add ability to customize searched fields

---
 config/default.yaml                   | 29 ++++++++++++++
 server/initializers/constants.ts      | 57 ++++++++++++++++++++++++++-
 server/lib/elastic-search-channels.ts |  2 +-
 server/lib/elastic-search-videos.ts   |  2 +-
 4 files changed, 87 insertions(+), 3 deletions(-)

diff --git a/config/default.yaml b/config/default.yaml
index 425fc8e..3892431 100644
--- a/config/default.yaml
+++ b/config/default.yaml
@@ -50,6 +50,35 @@ videos-search:
   boost-languages:
     enabled: true
 
+  # Add ability to change videos search fields boost value
+  # See https://www.elastic.co/guide/en/elasticsearch/reference/current/query-dsl-multi-match-query.html for more information
+  # If boost == 0, the field will not be part of the search
+  search-fields:
+    name:
+      boost: 5
+    description:
+      boost: 1
+    tags:
+      boost: 3
+    account-display-name:
+      boost: 2
+    channel-display-name:
+      boost: 2
+
+channels-search:
+  # Add ability to change channels search fields boost value
+  # See https://www.elastic.co/guide/en/elasticsearch/reference/current/query-dsl-multi-match-query.html for more information
+  # If boost == 0, the field will not be part of the search
+  search-fields:
+    name:
+      boost: 5
+    description:
+      boost: 1
+    display-name:
+      boost: 3
+    account-display-name:
+      boost: 2
+
 api:
   # Blacklist hosts that will not be returned by the search API
   blacklist:
diff --git a/server/initializers/constants.ts b/server/initializers/constants.ts
index 6480060..c75ed92 100644
--- a/server/initializers/constants.ts
+++ b/server/initializers/constants.ts
@@ -34,6 +34,48 @@ const CONFIG = {
   VIDEOS_SEARCH: {
     BOOST_LANGUAGES: {
       ENABLED: config.get<boolean>('videos-search.boost-languages.enabled')
+    },
+    SEARCH_FIELDS: {
+      NAME: {
+        FIELD_NAME: 'name',
+        BOOST: config.get<number>('videos-search.search-fields.name.boost')
+      },
+      DESCRIPTION: {
+        FIELD_NAME: 'description',
+        BOOST: config.get<number>('videos-search.search-fields.description.boost')
+      },
+      TAGS: {
+        FIELD_NAME: 'tags',
+        BOOST: config.get<number>('videos-search.search-fields.tags.boost')
+      },
+      ACCOUNT_DISPLAY_NAME: {
+        FIELD_NAME: 'account.displayName',
+        BOOST: config.get<number>('videos-search.search-fields.account-display-name.boost')
+      },
+      CHANNEL_DISPLAY_NAME: {
+        FIELD_NAME: 'channel.displayName',
+        BOOST: config.get<number>('videos-search.search-fields.channel-display-name.boost')
+      }
+    }
+  },
+  CHANNELS_SEARCH: {
+    SEARCH_FIELDS: {
+      NAME: {
+        FIELD_NAME: 'name',
+        BOOST: config.get<number>('channels-search.search-fields.name.boost')
+      },
+      DESCRIPTION: {
+        FIELD_NAME: 'description',
+        BOOST: config.get<number>('channels-search.search-fields.description.boost')
+      },
+      DISPLAY_NAME: {
+        FIELD_NAME: 'displayName',
+        BOOST: config.get<number>('channels-search.search-fields.display-name.boost')
+      },
+      ACCOUNT_DISPLAY_NAME: {
+        FIELD_NAME: 'ownerAccount.displayName',
+        BOOST: config.get<number>('channels-search.search-fields.account-display-name.boost')
+      }
     }
   },
   INSTANCES_INDEX: {
@@ -78,7 +120,9 @@ const REQUESTS = {
 
 const ELASTIC_SEARCH_QUERY = {
   FUZZINESS: 'AUTO:4,7',
-  BOOST_LANGUAGE_VALUE: 2
+  BOOST_LANGUAGE_VALUE: 2,
+  VIDEOS_MULTI_MATCH_FIELDS: buildMultiMatchFields(CONFIG.VIDEOS_SEARCH.SEARCH_FIELDS),
+  CHANNELS_MULTI_MATCH_FIELDS: buildMultiMatchFields(CONFIG.CHANNELS_SEARCH.SEARCH_FIELDS)
 }
 
 function getWebserverUrl () {
@@ -89,6 +133,17 @@ function getWebserverUrl () {
   return CONFIG.WEBSERVER.SCHEME + '://' + CONFIG.WEBSERVER.HOSTNAME + ':' + CONFIG.WEBSERVER.PORT
 }
 
+function buildMultiMatchFields (fields: { [name: string]: { BOOST: number, FIELD_NAME: string } }) {
+  return Object.keys(fields)
+    .map(id => {
+      const obj = fields[id]
+      if (obj.BOOST <= 0) return ''
+
+      return `${obj.FIELD_NAME}^${obj.BOOST}`
+    })
+    .filter(v => !!v)
+}
+
 if (isTestInstance()) {
   SCHEDULER_INTERVALS_MS.videosIndexer = 1000 * 60 * 5 // 5 minutes
 }
diff --git a/server/lib/elastic-search-channels.ts b/server/lib/elastic-search-channels.ts
index bbdacfe..d04b649 100644
--- a/server/lib/elastic-search-channels.ts
+++ b/server/lib/elastic-search-channels.ts
@@ -84,7 +84,7 @@ async function queryChannels (search: ChannelsSearchQuery) {
         {
           multi_match: {
             query: search.search,
-            fields: [ 'name^5', 'displayName^3', 'description' ],
+            fields: ELASTIC_SEARCH_QUERY.CHANNELS_MULTI_MATCH_FIELDS,
             fuzziness: ELASTIC_SEARCH_QUERY.FUZZINESS
           }
         }
diff --git a/server/lib/elastic-search-videos.ts b/server/lib/elastic-search-videos.ts
index f1ddf0e..37ec2e2 100644
--- a/server/lib/elastic-search-videos.ts
+++ b/server/lib/elastic-search-videos.ts
@@ -116,7 +116,7 @@ async function queryVideos (search: VideosSearchQuery) {
         {
           multi_match: {
             query: search.search,
-            fields: [ 'name^5', 'description', 'tags^3' ],
+            fields: ELASTIC_SEARCH_QUERY.VIDEOS_MULTI_MATCH_FIELDS,
             fuzziness: ELASTIC_SEARCH_QUERY.FUZZINESS
           }
         }