Generate a Search Index you can query via a static JSON blob or a Netlify Function!
You may not need this - There are other ways to add search to your site, like using Algolia or Vanilla JS with a custom search Index.
However, you may wish to have a way to generate this index based ONLY on crawling your generated static site, or you may wish to do index searches in a serverless function instead of making your user download the entire index and run clientside.
- Demo site: https://netlify-plugin-search-index.netlify.app
- Demo search function: https://netlify-plugin-search-index.netlify.app/.netlify/functions/search?s=web
- Demo JSON blob: https://netlify-plugin-search-index.netlify.app/searchIndex.json
To install, add the plugin in your netlify.toml
. No config is required but we show the default options here.
Generating both serverless function and clientside JSON
[[build]]
functions = functions # must specify a functions folder for this to work
[[plugins]]
package = netlify-plugin-search-index
# all inputs is optional, we just show you the defaults below
# [plugins.inputs]
# ignore = ["/ignore-this-file.html"] # don't index this file
# generatedFunctionName = search # change the name of generated folder in case of conflicts, use `null` to turn off
# publishDirJSONFileName = searchIndex # also use null to turn off
Without config, this would generate:
- a function at
https://yoursite.netlify.com/.netlify/functions/search
and - a clientside JSON blob at
https://yoursite.netlify.com/searchIndex.json
Generating serverless function only
To use this plugin only for the generated serveless function, supply null
to the publishDirJSONFileName
:
[[plugins]]
package = netlify-plugin-search-index
[plugins.inputs]
generatedFunctionName = mySearchFunction
publishDirJSONFileName = null
This would generate a Netlify function at https://yoursite.netlify.com/.netlify/functions/mySearchFunction
which you can query with https://yoursite.netlify.com/.netlify/functions/mySearchFunction?search=foo
.
Generating clientside JSON only
To use this plugin only for the clientside JSON file, supply null
to the generatedFunctionName
:
[[plugins]]
package = netlify-plugin-search-index
[plugins.inputs] =
generatedFunctionName = null
publishDirJSONFileName = mySearchIndex # you can use / to nest in a directory
This would generate a clientside JSON at https://yoursite.netlify.com/mySearchIndex.json
.
Supplying null
to both generatedFunctionName
and publishDirJSONFileName
would be meaningless (because there would be nothing to generate) and cause an error.
Your project probably contains some content files that you don't want your users to search. Pass an array of paths (or regex) to the files you don’t want to be indexed to dismiss them:
[[plugins]]
package = netlify-plugin-search-index
[plugins.inputs] =
exclude = ['/ignore-this-file.html', /^\/devnull\/.*/]
At the moment of writing, it is not possible to pass custom options to FuseJS. If it's something that you would like to see implemented, let us know! ✌️
After your project is built:
- this plugin goes through your HTML files
- extracts their metadata (title, description, keywords) with
unified
- converts them to searchable content, weighted by field type
- stores them as a JSON blob in
/searchIndex/searchIndex.json
- generates a Netlify Function that fuzzy searches against a query string with fuse.js
You can use this plugin in two ways:
- Client-side: You can simple require the JSON blob in your clientside JavaScript if it isn't too big:
// app.js import searchIndex from './searchIndex.json'
- Serverless-side: You can use the generated function that reads the JSON and returns fuzzy search results to be lighter on your frontend. The generated function is available at
.netlify/functions/searchIndex
and you can use it with a search term like.netlify/functions/searchIndex?s=foo
or.netlify/functions/searchIndex?search=foo
:// app.js document.getElementById('myForm').addEventListener('submit', async event => { event.preventDefault() const result = await fetch(`/.netlify/functions/searchIndex?search=${event.target.searchText.value}&limit=25`).then(x => x.json()) document.getElementById('result').innerText = JSON.stringify(result, null, 2) })
You can use an optional limit
parameter to limit the length of returned results.
Under the hood, the search function uses fuse.js and in future we may expose more configurations for this. Fuse.js search is fuzzy, therefore for short strings you may get a lot of results that don't seem relevant. In future we may make Fuzzy matching opt-in, and just use String.toLowerCase
and String.includes
- please see more details here.
We had to use patch-package to fix this bug: paulmillr/readdirp#157 - readdirp
is a dependency of ``copy-template-dir` which we use.
Hopefully it will be resolved or we can just fork copy-template-dir
entirely locally.
WE ARE SEEKING MAINTAINERS.
Micro Todo:
- expose fuse.js and html parse search options for more configurability
- filtering results
- support non html files?