Offline / local search for Docusaurus v2+ that works behind your firewall.
Feature Highlights:
- Supports multiple documentation versions
- Supports documentation written in languages other than English
- Highlights search results
- Customized parsers for docs, blogs, and general pages
- Lazy-loads the index
Note: We use the open source algolia/autocomplete library for the searchbox. This library is just used as the frontend, and, unlike the default Docusaurus search experience (algolia/docsearch), does not connect to any Algolia or third-party servers.
npm install @cmfcmf/docusaurus-search-localor
yarn add @cmfcmf/docusaurus-search-localAdd this plugin to the plugins array in docusaurus.config.js.
module.exports = {
// ...
plugins: [require.resolve("@cmfcmf/docusaurus-search-local")],
// or, if you want to specify options:
// ...
plugins: [
[
require.resolve("@cmfcmf/docusaurus-search-local"),
{
// Options here
},
],
],
};The following options are available (defaults are shown below):
{
// whether to index docs pages
indexDocs: true,
// Whether to also index the titles of the parent categories in the sidebar of a doc page.
// 0 disables this feature.
// 1 indexes the direct parent category in the sidebar of a doc page
// 2 indexes up to two nested parent categories of a doc page
// 3...
//
// Do _not_ use Infinity, the value must be a JSON-serializable integer.
indexDocSidebarParentCategories: 0,
// whether to index blog pages
indexBlog: true,
// whether to index static pages
// /404.html is never indexed
indexPages: false,
// language of your documentation, see next section
language: "en",
// setting this to "none" will prevent the default CSS to be included. The default CSS
// comes from autocomplete-theme-classic, which you can read more about here:
// https://www.algolia.com/doc/ui-libraries/autocomplete/api-reference/autocomplete-theme-classic/
// When you want to overwrite CSS variables defined by the default theme, make sure to suffix your
// overwrites with `!important`, because they might otherwise not be applied as expected. See the
// following comment for more information: https://github.com/cmfcmf/docusaurus-search-local/issues/107#issuecomment-1119831938.
style: undefined,
// The maximum number of search results shown to the user. This does _not_ affect performance of
// searches, but simply does not display additional search results that have been found.
maxSearchResults: 8,
// lunr.js-specific settings
lunr: {
// When indexing your documents, their content is split into "tokens".
// Text entered into the search box is also tokenized.
// This setting configures the separator used to determine where to split the text into tokens.
// By default, it splits the text at whitespace and dashes.
//
// Note: Does not work for "ja" and "th" languages, since these use a different tokenizer.
tokenizerSeparator: /[\s\-]+/,
// https://lunrjs.com/guides/customising.html#similarity-tuning
//
// This parameter controls the importance given to the length of a document and its fields. This
// value must be between 0 and 1, and by default it has a value of 0.75. Reducing this value
// reduces the effect of different length documents on a term’s importance to that document.
b: 0.75,
// This controls how quickly the boost given by a common word reaches saturation. Increasing it
// will slow down the rate of saturation and lower values result in quicker saturation. The
// default value is 1.2. If the collection of documents being indexed have high occurrences
// of words that are not covered by a stop word filter, these words can quickly dominate any
// similarity calculation. In these cases, this value can be reduced to get more balanced results.
k1: 1.2,
// By default, we rank pages where the search term appears in the title higher than pages where
// the search term appears in just the text. This is done by "boosting" title matches with a
// higher value than content matches. The concrete boosting behavior can be controlled by changing
// the following settings.
titleBoost: 5,
contentBoost: 1,
tagsBoost: 3,
parentCategoriesBoost: 2, // Only used when indexDocSidebarParentCategories > 0
}
}You can now use the search bar to search your documentation.
Important: Search only works for the statically built documentation (i.e., after you ran npm run docusaurus build in your documentation folder).
Search does not work in development (i.e., when running npm run docusaurus start).
If you want to test search locally, first build the documentation with npm run docusaurus build, and then serve it via npm run docusaurus serve.
Use the language option if your documentation is not written in English. You can either specify a single language or an array of multiple languages.
The following languages are available:
ar, da, de, en, es, fi, fr, hi, hu, it, ja, nl, no, pt, ro, ru, sv, th, tr, vi, zh
Important: For Chinese language support (zh), you also have to install the nodejieba npm package at ^2.5.0.
Documentation versions created with the official Docusaurus docs plugin are supported.
The search bar defaults to the latest version (not next, but the latest version defined in versions.json) when not on a documentation page (e.g., when looking at a blog post or a static page).
If the user visits a documentation page, the version is extracted from the page and search will only search the documentation of that version.
The searchbar placeholder text always reflects the currently detected documentation version.
This plugin supports documentation using Docusaurus i18n out of the box.. Please contribute additional translations by creating a new translation file in the codeTranslations subfolder and submitting a PR.
You can also adjust translations by modifiying the translations in <yourfolder>/i18n/<locale>/code.json that start with cmfcmf/d-s-l..
Read more at: https://docusaurus.io/docs/i18n/tutorial#translate-json-files
If building your documentation produces an error, you can build it in debug mode to figure out
which page is causing it. To do so, simply set the DEBUG environment variable when building
your documentation: DEBUG=1 npm run docusaurus build.
Please see the CONTRIBUTING.md file for further information.
MIT