Search

Built-in full-text search. The search index is built at Docusaurus build time, serialized to JSON, and queried at runtime in a background worker so the UI stays responsive.

Summary

When search is enabled, the theme reads the generated HTML after the Docusaurus build, extracts titles, headings, body text, and descriptions, and produces a search index.

At runtime, a Web Worker loads the index and processes queries without blocking the main thread. Matching terms are highlighted on the page.

How It Works

Build-Time Indexing

Route filtering — each route is classified as docs, blog, or page. Routes matching ignorePatterns are skipped. Routes whose type is disabled (e.g., indexBlog: false) are skipped.
HTML parsing — the built HTML for each route is parsed to extract searchable content. The title, description meta tag, h2/h3 headings with anchors, and article body text are extracted.
Index build — documents are indexed with field boosts: title (10×), headings (5×), body (1×). Language-specific stemmers are loaded for non-English languages.
Serialization — the index and document manifest are written to JSON. If hashed is enabled, the filename includes a content hash so browsers pick up updates immediately.

Runtime Search

Web Worker — a dedicated worker loads the serialized index on first query.
Query execution — the search engine processes the query with fuzzy matching (configurable distance 0–2).
Result rendering — results are returned to the main thread. Matching terms are highlighted on the target page when highlightSearchTermsOnTargetPage is enabled.

Configuration

Chinese Segmentation

Chinese text has no spaces between words, so the default stemmer cannot tokenize it. When zh is included in language, the preset expects @node-rs/jieba to be installed as a peer dependency.

Install it in any site that enables Chinese search:

npm install @node-rs/jieba

If zh is configured but @node-rs/jieba is missing, the build throws an error with the install command.

Options

Option	Type	Default	Description
`language`	`string[]`	`['en']`	Languages for stemming. Accepts ISO 639-1 language codes such as `en`, `de`, `ja`.
`indexDocs`	`boolean`	`true`	Index documentation pages.
`indexBlog`	`boolean`	`true`	Index blog posts.
`indexPages`	`boolean`	`false`	Index custom pages.
`hashed`	`boolean`	`true`	Append a content hash to the index filename so browsers pick up updates immediately.
`searchResultLimits`	`number`	`8`	Maximum number of search results to display.
`highlightSearchTermsOnTargetPage`	`boolean`	`true`	Highlight matching terms on the target page after navigation.
`searchBarShortcutKeymap`	`string`	`'mod+k'`	Keyboard shortcut to focus the search bar.
`fuzzyMatchingDistance`	`number`	`1`	Levenshtein distance for fuzzy matching (0 = exact, 1 = 1 typo, 2 = 2 typos).
`ignorePatterns`	`string[]`	`[]`	Route patterns to exclude from indexing (e.g., `'/docs/tags/**'`).
`docsRouteBasePath`	`string`	`'docs'`	Base path for documentation routes.

Output Files

After build, search produces these files in the output directory:

File	Purpose
`search-index-{hash}.json`	Serialized search index plus document manifest.
`search-manifest.json`	Points to the current index file.