Firestore Full Text Search Package

If you want full text search capabilities in Firestore, you have limited options... until now. Firebase recommends Algolia. Yes, Google recommends Algolia instead of a Google product to have search capabilities in a Google product. I digress. Basically, you have these options:

---

UPDATE 10/23/22 - I have since retired this package. I recommend using my j-firebase package.

---

• Algolia - semi supports relational data
• Elastic Search - there are many forks as well, but ultimately you don't have all options out of the box
• MeiliSearch - host it yourself... see here
• Typesense - self-hosting as well

You can also use pretty much any search engine that you can connect your data to using firebase trigger functions. However, I like to keep my data in Firebase.

I have a hate/love relationship with Firestore like everyone else, but I wrote a package that allows you to full-text search data in Firestore, by creating indexes in Firestore. While I had three different ideas to do this, see here for my article, there is only one package I think works for most people...

Relevant Search

1.) npm i adv-firestore-functions

2.) Add the function to a onWrite trigger function like so:

import { soundex, relevantIndex } from 'adv-firestore-functions';

...
functions.firestore
    .document('posts/{docId}')
    .onWrite(async (change: any, context: any) => {

...

   await relevantIndex(change, context, {
      fields: ['content', 'summary'],
      filterFunc: soundex
   });

}

The soundex function allows you to do a fuzzy search. While not perfect, it should take care of most of your needs. I made it optional in case you don't want to use it, or you want to use your own changing algorithm.

3.) Add / Edit / Delete a new Post (or whatever collection) - Everything is changed and updated automatically on the backend.

4.) Make sure to add rules for your collection as well as _search and _events collection. They both need to be writable by Firestore, while _search needs to be readable on the client side by anyone.

5.) Use my pre-written Firestore Callable Function to read the indexed data on the front end:

import { relevantSearch } from 'adv-firestore-functions';
...

exports.searchPosts = functions.https.onCall((q, context) => {

  return await relevantSearch({
    query: q.query,
    col: 'posts',
  });
...
}

or write your own:

• Observable Version - This is my Trigram Search, but same premise
• Promise Version - Backend Version could easily be implemented on Front End

If you have other needs, check out my other searches:

• Which Full-Text Search Index Do I Use? - my Fireblog.io
• adv-firestore-functions - GitHub project

While I may or may not be done with Firestore, see Dear Firebase Team..., I wrote this package a while ago and needed to share it.

Post on GitHub if you have issues or ideas.

Thanks,

J

Comments

[Stuart Tett]

I've put this in my onWrite function. But is there a way to just run it against all items in an existing collection? I don't have have a "change" value to use for example if I were to implement it in a functions.pubsub.schedule function

[Jonathan Gamble]

No, you need change to get the values. You could write your own. Check out the frontend version.

dev.to/jdgamble555/quick-firestore...