rxjs-docs icon indicating copy to clipboard operation
rxjs-docs copied to clipboard

Add search feature for documentation

Open ashwin-sureshkumar opened this issue 7 years ago • 14 comments

A feature to search the documentation site to quickly search and navigate to the desired section

ashwin-sureshkumar avatar Oct 16 '17 18:10 ashwin-sureshkumar

cc @ladyleet

ashwin-sureshkumar avatar Oct 16 '17 18:10 ashwin-sureshkumar

Algolia could help for this. https://community.algolia.com/docsearch/

staltz avatar Oct 16 '17 18:10 staltz

@staltz - That looks great ! @ladyleet - If we end up deciding on algolia, at some point we would need to register for this.

ashwin-sureshkumar avatar Oct 16 '17 18:10 ashwin-sureshkumar

It's very easy, basically they do the hard work for you.

staltz avatar Oct 17 '17 08:10 staltz

I feel like @benlesh and @jayphelps had some words related to search. Does this do fuzzy search? Guys did you have opinions?

ladyleet avatar Oct 17 '17 11:10 ladyleet

It seems to allow 1-character typos, so yes, some amount of fuzzy search.

screen shot 2017-10-17 at 15 14 03

staltz avatar Oct 17 '17 12:10 staltz

@staltz does it supports 2byte languages for translations? (like chinese, kr, jp..)

kwonoj avatar Oct 17 '17 15:10 kwonoj

I can ping @pixelastic he could probably help with this.

manekinekko avatar Oct 18 '17 20:10 manekinekko

Thx @manekinekko !

ladyleet avatar Oct 18 '17 21:10 ladyleet

Hey everyone, thanks for pinging me :)

To add search to your documentation, I would definitely suggest using DocSearch. It's using Algolia's API under the hood, but you don't have to open an Algolia account.

All you have to do is fill the form on the website with the url of your website and an email address where we can get back to you. We will take care of defining a custom crawler config for your website (basically a bunch of CSS selectors to identify what is a title, what is a text, etc). Then, we will have a crawler running on our infra every 24 hours that will re-index your search results.

As you've noted, we handle typo-tolerance (even with more than 1 typo) out of the box. Now for multi-language search it's definitely possible to segregate results by language (people on the fr doc will only get fr results, etc).

Note that by default DocSearch works as a crawler running on our own infra, but we've open-sourced the code and you can actually run it from its docker container directly if you want. If you go that way you'll have to open an Algolia account and push your data to it, though.

I'm also pinging @phrawzty here that might be able to help

pixelastic avatar Oct 20 '17 23:10 pixelastic

this is really cool! unless @jayphelps @benlesh @kwonoj comment on anything, then let's run with this. we'll give em a few days.

ladyleet avatar Oct 21 '17 07:10 ladyleet

Hi all! I admit that I'm biased, but :+1: for DocSearch. :smile: If you need any help or advice, or just have questions about how it works or what it takes to get up and running (spoiler: it's easy), please don't hesitate to ask.

phrawzty avatar Oct 23 '17 08:10 phrawzty

@pixelastic seems really awesome, though how is it (or will be) subsidized/monetized?

jayphelps avatar Oct 26 '17 23:10 jayphelps

It's free and will always be. Docsearch is not our main product, it's a gift back to the developer community and we will never charge for it.

We just have a small logo on the dropdown that links to us, that's all.

On Fri, Oct 27, 2017, 01:27 Jay Phelps [email protected] wrote:

@pixelastic https://github.com/pixelastic seems really awesome, but how is it (or will be) subsidized/monetized?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/ReactiveX/rxjs-docs/issues/38#issuecomment-339829210, or mute the thread https://github.com/notifications/unsubscribe-auth/AARTG0mvLtHbjqteqnlsKhUDpHsqIUUnks5swRVOgaJpZM4P7C3n .

pixelastic avatar Oct 27 '17 05:10 pixelastic