Hi there I would like to propose a repository & NPM transfer of this repo to the @metalsmith org.
This plugin would be refactored with a common setup for core plugins & renamed to @metalsmith/search-index.

Your name would remain in the LICENSE file, and a new line would be added in front (c) webketje.
I would add you as a member to the metalsmith org (temporarily) so you can initiate a Transfer of the repo.
You would need to deprecate the NPM package instructing users to use @metalsmith/search-index instead.

If you do not further maintain this package, or respond to this request in a month's time, I will fork the project and move ahead as I did with @metalsmith/postcss

Nice plugin! Is it possible to add multiples refs in the store items (or something along those lines)? For example, if you index by 'filePath', and then use lunr.js to search through searchIndex.json, how would you grab the title? Is there a way to make 'title' and 'filePath' easily accessible at the same time, using the same json file? I'm trying to use this to make a search-suggest sort of thing. Any tips appreciated. Thanks in advance.

I'm not sure if this is a bug for you, or for Lunr, but seeing as it has to do with the index, and this is what I'm using to build the index, I thought I would start here.

So I'm getting the error Uncaught TypeError: Cannot read property 'tf' of undefined. It seems to have to do with parsing vector similarities. The code seems to be looking for tf on a document in the tokenStore.
https://cl.ly/1B3f1m0B391b

My lunr configs look like:

      metalsmithLunr({
        ref: 'id',
        indexPath: 'api/search-index.json',
        fields: {
          headline: 5,
          subheadline: 3,
          category: 5,
          tags: 3,
          contents: 1,
        },
        preprocess: function(content) {
          const tr = (str) => {
            const map = {"а":"a","б":"b",/*  truncated for issue  */"І":"I","і":"i"};
            let new_str = "", char, substitute, n = str.length;
            for(let i = 0; i < n; i++) {
                char = str[i];
                substitute = map[char];
                new_str += substitute ? substitute : char;
            }
            return new_str;
          };
          return tr(
            content.replace(/<[^>]+>/g, ' ') // Strip HTML
          ) // Transliterate foreign characters
            .replace(/[^\w]/g, ' ') // Strip Punctuation
          ;
        },
      }),

And the client side code looks like:

    let ip = new Promise((resolve, reject) => {
      $.getJSON('/api/search-index.json', data => {
        console.log('SEARCH DATA', data);
        resolve(lunr.Index.load(data));
      });
    });

and then later:

      if (keywords) {
        SearchService.results = SearchService.index.search(keywords);
      }

I can provide the index data or the serialized index object if you would like.

It would be nice if one could edit the lunr pipeline through the options object. E.g., to support stemming of other languages than english.

I tried using this plugin with v2.0.0 of Lunr and I am getting this error :

Once I switched back to an older version : 0.5.12. It seemed to work properly.

This is my lunr snippet from the build file:

.use(lunr({
        preprocess: function(content) {
        // Replace all occurrences of __title__ with the current file's title metadata.
        return content.replace(/__title__/g, this.title);
        }
 }))

How do I strip HTML tags ??

Ref gives filename by default, can i add more fields like "ref"