This will be a quick update as I'm on vacation and should be busy playing XBox, but I've got an update to my post on using Algolia with Eleventy. Please read it first as this post won't make much sense without it. In that post, I described a way to use Algolia with Eleventy. The process was basically:

• Generate a JSON index of your content.
• Push the index to Algolia
• Use their JavaScript library to add a search

The second bullet point was the crucial one. I made use of algolia-indexing to handle updating my index. This utility created a copy of your index, did atomic updates focusing only on what changed, and then copied back your index. This utility was super cool, but it wouldn't work for me. Algolia has a max of 10k records per index on their free tier, which is a very generous amount, but my blog has a bit over six thousand posts.

I was prepared to "give up" when Algolia, just yesterday morning, announced friendlier pricing. Follow that link for details, but the part that was most important to me was that indexing operations (adding, updating, and removing records) was now free. That means I could adapt a "delete everything and re-upload data" approach for my site.

Now to be clear, this means that for a few seconds, my search index is blank. I'm ok with that. Heck, as far as I can tell I'm the only one who actually uses my search page. But you do want to keep that in mind before you consider using the approach I describe below. If you're below 5K records, I'd definitely suggest using algolia-indexing instead.

Ok, so with that in mind, here's what I'm using on the blog.

/// HANDLE ALOGLIA
// first, get my index
let dataResp = await fetch('https://www.raymondcamden.com/algolia.json');

let data = await dataResp.json();
console.log('Successfully got the data, size of articles '+data.length, data[0].title);

let host = `https://${algCredentials.appId}.algolia.net`;

//first clear
let resp = await fetch(host + `/1/indexes/${algCredentials.indexName}/clear`, {
	method:'POST',
	headers: {
	'X-Algolia-Application-Id':algCredentials.appId,
	'X-Algolia-API-Key':algCredentials.apiKey
	}
});
let result = await resp.json();
console.log('clear result is '+JSON.stringify(result));

let batch = {
	"requests":[]
};

data.forEach(d => {
	batch.requests.push({
	'action':'addObject',
	'body':d
	})
});
console.log('batch data done');

//then batch
resp = await fetch(host + `/1/indexes/${algCredentials.indexName}/batch`, {
	method:'POST',
	body: JSON.stringify(batch),
	headers: {
	'X-Algolia-Application-Id':algCredentials.appId,
	'X-Algolia-API-Key':algCredentials.apiKey
	}
});

result = await resp.json();
if(result.objectIDs) console.log(`i had ${result.objectIDs.length} objects added`);

As I described in my last post, I generate a JSON file representing all my content. In my Netlify `deploy-succeeded' function, I fetch that file so I can use it for my updates.

The first thing I do is use the clear endpoint on my index, which does what you would assume, empty the index. Then I use the batch endpoint to add all my data. Notice I have to 'reshape' the JSON into an array of commands (addObject). I kinda wish Algolia had a "addAll" type endpoint that just took an array of objects. That would make the POST a bit smaller too. But it wasn't a big deal to create that data for the batch call.

Speaking of time - Netlify has a limit of 10 seconds for serverless function executions. While trying to get Algolia working on my site last month I asked for an extension and got the max, 20 seconds. In my testing locally, I saw the duration typically take around 8 seconds, but sometimes it took 11. Most of the time spent was in network calls, so I don't believe it's going to be an issue when running on Netlify's CDNs, but it's something to keep in mind while testing on your local machine. The netlify dev command will enforce a 10 second limit so you may get errors there that you would not get in production.

I think you could also argue that it doesn't make sense to re-index your data in every build, especially since you may update your site in a way that doesn't impact content (like changing a header graphic for example), but I'm ok with this for now.

Of course, I also had to update my search page as well. In my last post I explained how I used the JavaScript wrapper to build a simple search interface with Vue.js. For my site, I made it a bit nicer. Here's the method I use for my search:

async doSearch() {
	this.results = null;
	if(this.search === '') return;
	this.searching = true;
	this.noResults = false;
	console.log('search for '+this.search);
	let resultsRaw = await this.index.search(this.search,{
		attributesToRetrieve:['title', 'url','date'],
		attributesToSnippet:['content'],
		hitsPerPage: 50
	});

	let options = { year: 'numeric', month: 'long', day: 'numeric' };
	let formatter = new Intl.DateTimeFormat('en-US',options);
	resultsRaw.hits.forEach(h => {
		h.date = formatter.format(new Date(h.date));
	});
	this.results = resultsRaw.hits;
	this.numResults = resultsRaw.nbHits;
	this.searching = false;
	this.noResults = this.results.length === 0;
}

A few things. First, notice I'm asking for date in the results. Showing the date for my results is important so I can see which results are more recent. Next, I ask for a snippet of the content property, this lets me display part of the content in the result. Finally, I make use of the Intl API to format the dates returned in my results. I'd normally use a Vue filter for that, but as they are deprecated in Vue 3, I'm trying to wean myself off of them. Here's a screen shot of how my search page renders:

Let me be clear - this could look nicer. Blame me, not Algolia. But I'm incredibly happy to be able to rip out Google (and their ads) and replace it with a simpler, nicer system! You can find all the code for my implementation here: https://github.com/cfjedimaster/raymondcamden2020 As always, questions and comments are welcome!