feat(server): improve indexer (#14698)

fix #13862 


#### PR Dependency Tree


* **PR #14698** 👈

This tree was auto-generated by
[Charcoal](https://github.com/danerwilliams/charcoal)

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->

## Summary by CodeRabbit

* **New Features**
* Enhanced search support for Chinese, Japanese, and Korean languages
with improved text segmentation and character matching.
* Added index management capabilities with table recreation
functionality.

* **Bug Fixes**
* Improved search accuracy for non-Latin scripts through updated
morphology and n-gram configuration.

* **Chores**
  * Added database migration for search index optimization.

<!-- end of auto-generated comment: release notes by coderabbit.ai -->
This commit is contained in:
DarkSky
2026-03-22 02:50:59 +08:00
committed by GitHub
parent bcf2a51d41
commit f47ee2bc8a
10 changed files with 402 additions and 50 deletions
@@ -14,6 +14,7 @@ import {
AggregateQueryDSL,
BaseQueryDSL,
HighlightDSL,
ManticoresearchProvider,
OperationOptions,
SearchNode,
SearchProvider,
@@ -130,6 +131,63 @@ export class IndexerService {
}
}
async rebuildManticoreIndexes() {
let searchProvider: SearchProvider | undefined;
try {
searchProvider = this.factory.get();
} catch (err) {
if (err instanceof SearchProviderNotFound) {
this.logger.debug('No search provider found, skip rebuilding tables');
return;
}
throw err;
}
if (!(searchProvider instanceof ManticoresearchProvider)) {
this.logger.debug(
`Search provider ${searchProvider.type} does not need manticore rebuild`
);
return;
}
const mappings = SearchTableMappingStrings[searchProvider.type];
for (const table of Object.keys(mappings) as SearchTable[]) {
await searchProvider.recreateTable(table, mappings[table]);
}
let lastWorkspaceSid = 0;
while (true) {
const workspaces = await this.models.workspace.list(
{ sid: { gt: lastWorkspaceSid } },
{ id: true, sid: true },
100
);
if (!workspaces.length) {
break;
}
for (const workspace of workspaces) {
await this.models.workspace.update(
workspace.id,
{ indexed: false },
false
);
await this.queue.add(
'indexer.indexWorkspace',
{
workspaceId: workspace.id,
},
{
jobId: `indexWorkspace/${workspace.id}`,
priority: 100,
}
);
}
lastWorkspaceSid = workspaces[workspaces.length - 1].sid;
}
}
async write<T extends SearchTable>(
table: T,
documents: UpsertTypeByTable<T>[],