chore: elasticsearch

databases/elasticsearch_db.ts

@@ -19,9 +19,9 @@ import assert, {equal} from 'assert';
 
 import {Buffer} from 'buffer';
 import {createHash} from 'crypto';
-import {Client} from 'elasticsearch8';
+import {Client} from '@elastic/elasticsearch';
+import type {estypes} from '@elastic/elasticsearch';
 import {BulkObject} from './cassandra_db';
-import {MappingTypeMapping} from "elasticsearch8/lib/api/types";
 
 const schema = '2';
 
@@ -30,7 +30,7 @@ const keyToId = (key:string) => {
   return keyBuf.length > 512 ? createHash('sha512').update(keyBuf).digest('hex') : key;
 };
 
-const mappings: MappingTypeMapping = {
+const mappings: estypes.MappingTypeMapping = {
   // _id is expected to equal key, unless the UTF-8 encoded key is > 512 bytes, in which case it is
   // the hex-encoded sha512 hash of the UTF-8 encoded key.
   properties: {
@@ -39,6 +39,28 @@ const mappings: MappingTypeMapping = {
   },
 };
 
+const legacyDocToSchema2Key = (index: string, id: string, type: unknown, v1BaseIndex?: string) => {
+  const legacyType = typeof type === 'string' && type !== '' && type !== '_doc' ? type : null;
+  if (v1BaseIndex && index !== v1BaseIndex) {
+    const parts = index.slice(v1BaseIndex.length + 1).split('-');
+    if (parts.length !== 2) {
+      throw new Error(`unable to migrate records from index ${index} due to data ambiguity`);
+    }
+    if (legacyType != null) return `${parts[0]}:${decodeURIComponent(legacyType)}:${parts[1]}:${id}`;
+    const idParts = id.split(':');
+    if (idParts.length !== 2) {
+      throw new Error(`unable to migrate records from index ${index} due to data ambiguity`);
+    }
+    return `${parts[0]}:${idParts[0]}:${parts[1]}:${idParts[1]}`;
+  }
+  if (legacyType != null) return `${legacyType}:${id}`;
+  const idParts = id.split(':');
+  if (idParts.length !== 2) {
+    throw new Error(`unable to migrate records from index ${index} due to missing legacy type metadata`);
+  }
+  return `${idParts[0]}:${idParts[1]}`;
+};
+
 const migrateToSchema2 = async (client: Client, v1BaseIndex: string | undefined, v2Index: string, logger: any) => {
   let recordsMigratedLastLogged = 0;
   let recordsMigrated = 0;
@@ -60,14 +82,7 @@ const migrateToSchema2 = async (client: Client, v1BaseIndex: string | undefined,
       totals.set(index, total);
       const body = [];
       for (const {_id, _type, _source: {val}} of hits) {
-        let key = `${_type}:${_id}`;
-        if (v1BaseIndex && index !== v1BaseIndex) {
-          const parts = index.slice(v1BaseIndex.length + 1).split('-');
-          if (parts.length !== 2) {
-            throw new Error(`unable to migrate records from index ${index} due to data ambiguity`);
-          }
-          key = `${parts[0]}:${decodeURIComponent(_type)}:${parts[1]}:${_id}`;
-        }
+        const key = legacyDocToSchema2Key(index, _id, _type, v1BaseIndex);
         body.push({index: {_id: keyToId(key)}}, {key, value: JSON.parse(val)});
       }
       await client.bulk({index: v2Index, body});

docker-compose.yml

@@ -9,7 +9,7 @@ services:
           COUCHDB_USER: ueberdb
           COUCHDB_PASSWORD: ueberdb
     elasticsearch:
-        image: elasticsearch:7.17.3
+        image: elasticsearch:9.3.3
         ports:
         - 9200:9200
         environment:

package.json

@@ -37,7 +37,7 @@
     "cassandra-driver": "^4.8.0",
     "cli-table3": "^0.6.5",
     "dirty-ts": "^1.1.8",
-    "elasticsearch8": "npm:@elastic/elasticsearch@^8.19.1",
+    "@elastic/elasticsearch": "^9.3.4",
     "eslint": "^9.35.0",
     "eslint-config-etherpad": "^4.0.4",
     "eslint-plugin-import": "^2.32.0",
@@ -95,6 +95,6 @@
     }
   ],
   "engines": {
-    "node": ">=16.20.1"
+    "node": ">=22.0.0"
   }
 }