Elegua: the best Svelte micro client router you'll ever see.
It's hopefully very straighforward: there are no components, just some stores reflecting current path/parts of url, and a resolve()
function for more complex (regexp/named routes) routings. The rest is just your plain Svelte logical blocks.
{#if $path === '/'}
<h1>Home page</h1>
{:else if $path === '/about'}
<h1>About page</h1>
{:else if resolve($path, '/blog/:slug')}
<Post slug={$params('slug')}/>
{:else}
<h1>404</h1>
<p>Not found: {$path}</p>
{/if}
Online: https://elegua.netlify.app/
... or run yourself:
git clone http://github.com/howesteve/elegua
cd elegua
pnpm install
pnpm run dev
<Route>
, <Link>
or any other components. Uses regular {#if}/{:else}
blocks from Svelte to control routing/rendering. It's all stores/functions./
or /blog
/xxx/:group/:id
) (yes, they can be nested)/id/[0-9]+\.json
)preventUnload()
and preventChange()
.<Route>
or <Link>
, or complicated filesystem logic. I wanted to get rid of those and just use plain Svelte logic blocks for routing.src/routes/blog/[slug]/+page.js
src/routes/blog/page/[page]/+page.js
src/routes/blog/page/[page]/+page.server.js
src/routes/blog/category/[category]/page/[page]/+page.server.js
... then all the boilerplate to make it work, just make me sick. Pages and pages and pages of docs just to learn how to re-learn routing! Shoot me. I have no patience.http://github.com/howesteve/elegua
Most of this repository is a demo applicaion for Elegua. But Elegua itself is a single file.
You mean, besides reading the source code? :) Here it goes.
pnpm i -D elegua
npm i -D elegua
yarn i -D elegua
Elegua's routing is designed expecting routes are mainly guided by path
, of course. However, you can route by hash, searchParams, url or anything else you want.
Every time the current browser's url changes, Elegua will update its stores and your routing logic will do the rest.
:warning: WARNING: It's best to define routes in your main application's page and not in subpages that are lazily loaded, otherwise routes might not be defined when you think they are, and that could lead to unexpected results.
A more comprehensive example of how routing looks like in Elegua:
<script lang="ts">
import { path, resolve, params, match } from 'elegua';
</script>
<main>
<!-- Simple main page route -->
{#if $path === '/'}
<Home />
<!-- /about - another static route. -->
{:else if $path === '/about'}
<About />
<!-- Nested paths routing - you can use any path -->
{:else if $path === '/blog/main-post'}
<MainBlogPost />
<!-- Static page resolving using resolve() function -->
{:else if resolve($path, '/users')}
<Users />
<!-- Named param: $match will have the matching param "slug" -->
{:else if resolve($path, '/blog/:slug')}
<PostPage slug={$match['slug']} />
<!-- Multiple named paths: $match will have all the matching params -->
{:else if resolve($path, '/blog/:author/:slug')}
<PostPage author={$params['author']} slug={$params['slug']} />
<!-- RegExp route: $match will have the matching params, same as with named paths -->
{:else if resolve($path, /\/authors\/([0-9]+)/)}
<Author id={$match[1]} />
{:else}
<!-- Fallback/error route "*" - this will get matched/rendered if nothing else did. You can inspect $path, $url, etc as usual to see what user tried to reach -->
<Error />
{/if}
</main>
resolve(path: string, route: string|RegExp)
The resolve(path, route)
function is Elegua's core route resolver. It accepts plain string, named or regExp route param, and while in a template block, it's designed to be used with the $path
argument to be called reactively whenever the url changes
route
is a plain string, $match
and $params
will be empty. route
is a named params route, $match
will be empty but $params
will have each named param set.route
is a regexp route, $match
and $params
will both be set.See below for examples:
<!-- fixed route -->
{#if resolve($path, '/')}
<h1>Home page</h2>
<!-- named params route -->
{:else if resolve($path, '/blog/:post_id')}
<Post id={$params["post_id"]}/>
<!-- regexp route -->
{:else if resolve($path, /users\/([a-zA-Z])+/)}
<p>This is the page for user {$match[1]}
{:else}
<!-- None of the above matched? Render the error route -->
<h1>Error 404</h1>
<p>Page {$path} not found</p>
{/if}
Note
Implementation detail. One might wonder why I left a
$path
param in theresolve(path, route)
api - it might seem cumbersome, why not using justresolve('/')
? It's because otherwise Svelte wouldn't know it has to re-render the template containing the resolve block. With$path
explicitly appearing in template block, Svelte will re-render it every time$path
changes, andresolve()
gets called. Otherwise, path changes would not be perceived.
goto(href: string|URL)
The goto(href)
method navigates to some url/path programmatically. Internally, it uses history.pushState()
. Calls to goto(href)
will trigger updates in all the expected reactive stores: $path, $url, $hash, $oldUrl, etc. - and also will update the current browser's url.
preventChange : (()=> boolean|undefined) | undefined
This function allows setting a callback for preventing exiting a route/page from either clicking <a>
links or using goto()
. If the callback function returns true
, the link change will be blocked. If it returns anything else, it will be allowed. For instance, for preventing going away from a form has been changed:
preventChange(() => {
if (formIsDirty) {
alert('Please save or cancel changes before navigating away.');
return true;
}
});
To unset the callback, just call preventChange()
without arguments:
preventChange()
For a more useful example of calling preventChange()
inside the onMount()
/onDestroy()
handlers, check our demo app and it's source.
Note
This method will not prevent the user from closing the window/reloading the page. That is accomplished by handiling the beforeload event. But check the
preventUnload
action for a nice wrapper.
I don't know of any other routers that have this feature.
preventUnload(callback: () => boolean | string| undefined)
This is a svelte action that will prevent the user from closing the current window if a condition is met. It can be used in set with preventChange
to prevent the user from closing or navigating away from the changed forms.
<svelte:window use:preventUnload={() => formIsdirty()} />
...
</svelte:window>
Although this is not a part of the routing system per se, I think it's a nice addition to Elegua. Same as with preventChange
, I don't know about any other routers that have this feature.
dynamic(path: string, routes: DynamicRoute[], defaultRoute?: ComponentType): ComponentType|undefined
type DynamicRoute = [string | RegExp, ComponentType, any?]
The dynamic()
method is a very special one; it allows dynamic routing using Elegua:
<script lang="ts">
import {dynamic} from 'elegua';
import Home from './Home.svelte';
import About from './About.svelte';
import Blog from './Blog.svelte';
import Error from './Error404.svelte';
</script>
<svelte:component this={dynamic($path, [
['/', Home],
['/about', About],
['/blog', Blog],
['/blog/:post', Post],
], Error)} />
Sometimes you might have a lot of static routes that point to components and don't want to define them one by one in {$if}
blocks, or you just don't know the routes beforehand. For such cases, dynamic()
is very handy.
The arguments are:
path
: the current $path
routes
: the defined routes, i.e. an Array\<DynamicRoute>:
route
: a string or regexp; the same you'd use in a resolve()
calldefaultRoute
: an optional default route that will be rendered if no other route matches. This is typically used for displaying 404/error pages.If you need to inspect any Elegua variables inside your dynamically rendered component, please do it from inside the component. For instance, in the example above, the Post
component could be implemented as:
<p>This is post id {$params["post"]}</p>
or:
<script>
onMount(()=> {
if ($params['post']) {
<!-- Load post from server -->
}
})
</script>
A writable store that reflects the current url's path. If you load http://localhost/blog
, $path
will be set as "/blog"
.
If you set $path
using path.set("/")
or $path = "/"
, it will update the value store and the browser will load home route (similar to using goto("/")
).
This is used very often for simple, fixed routings.
<script lang="ts">
import { path } from 'elegua';
</script>
<h1>Your are now in the {$path} page.</h1>
<p>Set path:</p>
<button on:click|preventDefault={() => ($path = '/blog')}>Set $path='/blog'</button>
Use Svelte's {#if}
blocks for routing using $path
:
<script lang="ts">
import { path } from 'elegua';
import Home from './Home.svelte';
import Blog from './Blog.svelte';
</script>
{#if $path === '/'}
<Home />
{:else if $path === '/blog'}
<Blog />
{/if}
Note
Routing using
$path
as stated above (e.g.$path === '/blog'
) works fine, but$match
and$params
will be unchanged and might be reflecting the values from the latest previousresolve()
call. If that's not what you want, route static paths usingresolve()
:
<script lang="ts">
import { path } from 'elegua';
</script>
{#if resolve($path, '/')}
<h1>Home page. {$match} and {$params} are empty.</h1>
{/if}
This writable store is a URL object for the current loaded url. Anytime the browser's url changes, $url
will be updated. And if you update $url
, the current browser's url will change to reflect it as well.
You can inspect/use url.pathname
, url.hash
, url.searchParams
, etc.
<script lang="ts">
import { url } from 'elegua';
</script>
Current page: {$url.pathname}
<br />Current path: {$url.pathname}
<br />Current hash: {$url.hash}
<br />Current searchParams: {$url.searchParams}
Using $url
, you can handle any kind of routing. For instance, loading a post by hash using $url
:
<script lang="ts">
import { url } from 'elegua';
let post: Post;
url.subscribe(async (u) => {
// Loading post by hash
post = await getPost(u.hash);
});
</script>
<Post data={post} />
A writable store that reflects the current url's hash. If you load http://localhost/#xxx
, $hash
will be set as "xxx"
. If you call hash.set('xxx')
, $hash
will be set to 'xxx'
, and the browser url will be updated.
Important
Elegua strips the
#
symbol is from thehash
string, i.e. on urlhttp://localhost/blog#myhash
,$hash
will bemyhash
instead of#myhash
.
<script lang="ts">
import Router, { hash } from 'elegua';
</script>
<h1>The current hash is: {$hash}</h1>
Internally, it works by monitoring both popstate and hashchange events.
This readable store is a URLSearchParams object for the current url. For instance, if you load:http://localhost/blog?x=1
and call $searchParams.get('x')
, you'll get "1"
(yes, a string). For changing a searchParams
value, call $searchParams.set("x", "1")
. Check the URLSearchParams reference for other methods.
Elegua has reactive searchParams.set()
and searchParams.delete()
and searchParams.append()
methods for convenience; if you use them, the current browser's url and history will automatically be updated, and if you change the url values, searchParams
will reflect them.
Reading from searchParams
:
<script lang="ts">
import { searparams } from 'elegua';
</script>
{#if resolve($path, '/blog')}
<!-- when you load /blog/searchparams?x=1-->
Param <code>x</code> is {$searchParams.get('x')}.
<br />(should print "1")
{/if}
Routing using searchParams
:
<script lang="ts">
import { searparams } from 'elegua';
</script>
<!-- when you load, ex /blog?x=1-->
{@const x = $searchParams.get('x')}
{#if x === 1}
Param <code>x</code> is 1.
{:else if x === 2}
Param <code>x</code> is 2.
{:else}
Param <code>x</code> is {x}.
{/if}
<br />(should print "1")
Setting a searchParam
(reactive - browser url will change):
<button
on:click|preventDefault={() => {
$searchParams.set('x', '1');
}}>Set x</button
>
Removing a searchParam
(reactive - browser url will change):
<button
on:click|preventDefault={() => {
$searchParams.delete('x');
}}>Remove x</button
>
Appending a searchParam
(reactive - browser url will change):
<button
on:click|preventDefault={() => {
$searchParams.append('y', 1);
}}>Append('y', 1)</button
>
searchParam
changesThese helper $searchParam
methods above perform both the searchParam
update and a history push at the same time. I implemented this way to make the most common case, i.e. single searchParam
updates, easier. They call $url.searchParams.xxx()
internally.
But somethimes, you might want to change multiple searchParam
and do a single history push for them all after those changes, avoiding ending up with a lot of history items for every change.
For that, use the original $url.searchParams.xxx()
methods for each change, and "commit" them all at once at the end by calling goto
:
// do all changes you want using original URL.searchParams methods
$url.searchParams.set('a', 1);
$url.searchParams.set('b', 2);
$url.searchParams.append('c', 3);
//(...)
// then push all above changes into history just once
goto($url)
Note
searchParam.append()
will perform exactly asURLSearchParam.append()
, i.e. if you call it multiple times, it will append params multiple times.
This readable store will be set after a resolve(path, route)
call. If the route is a regexp route, match[x]
will have all matching results.
$match
is just the return from the regexp.exec().groups()
called internally to resolve the route - so, it's number indexed.
For instance, when loading http://localhost/blog/my-post
or http://localhost/users/my-user
<script lang="ts">
import { matches } from 'elegua';
</script>
{#if resolve($path, '/blog/:slug')}
Blog post {match[1]} (="my-post")
{:else if resolve($path, /users\/([a-zA-Z\-])+/)}
User: {match[1]} (="my-user")
{/if}
Important
$match is only updated after a
resolve()
call. Specifically,$path
-based routing will not update this store.
This store contains the named (variable) parts of a match, after a resolve
call; is kinda similar to $match, but whereas $match is indexed by number, $params
expects strings.
<script lang="ts">
import { params } from 'elegua';
</script>
{#if resolve($path, '/blog/:slug')}
You are rendering the {$params['slug']} blog post.
{/if}
$params
will also have any regexp named groups:
<script lang="ts">
import { params } from 'elegua';
</script>
{#if resolve($path, /\/blog\/(?<slug>[a-zA-Z0-9]+)/)}
You are rendering the {$params['slug']} blog post.
{/if}
Important
Just as with $match,
$params
is only updated after aresolve()
call. Specifically,$path
-based routing will not update these stores.
A store for the old (previous) url before the last change.
<script lang="ts">
import { path, oldUrl } from 'elegua';
</script>
<h1>Your are now in the {$path} page, coming from {$oldUrl.pathname}.</h1>
$ pnpm create vite@latest elegua-test --template svelte
$ cd elegua-test
$ pnpm install -D elegua
Then, edit src/App.svelte
(i.e. "code ."
if you use vscode) and replace it with something like:
<script>
import { path } from "elegua";
</script>
<header>
<nav>
<ul id="menu">
<li><a href="https://github.com/howesteve/elegua/blob/master/" class:selected={$path === "/"}>Home</a></li>
<li><a href="https://github.com/howesteve/elegua/blob/master/about" class:selected={$path === "/about"}>About</a></li>
</ul>
</nav>
</header>
<main>
{#if $path === "/"}
<h1>Home page</h1>
{:else if $path === "/about"}
<h1>About page</h1>
{:else}
<h1>Error 404</h1>
<p>Not found</p>
{/if}
</main>
<style>
a {
text-decoration: none;
}
a:hover {
text-decoration: underline;
}
a:visited {
color: inherit;
}
.selected {
font-weight: bolder;
}
ul#menu li {
display: inline;
margin-right: 1em;
}
header {
font-size: large;
}
</style>
Just use regular Svelte if
blocks. When nothing else matches, show your error page.
<script lang="ts">
import { path } from 'elegua';
</script>
{#if $path === '/'}
<h1>Home page</h1>
{:else}
<p>No such path: {$path}</p>
{/if}
Named routes are routes that can have named variables, prefixed with ":"
, such as /blog/:slug
:
<script lang="ts">
import { resolve } from 'elegua';
</script>
{#if resolve('/blog/:slug')}
<Post id={$params['slug']} />
{/if}
The $params store will reflect the params by name. Internally, this is implemented using a RegExp route.
If the last resolve()
call did not use named routes/or regexp matching (i.e. a hash match), $params will be empty.
A named route will resolve with any string. If you need more control of what's matched by a named route, you should either be using regexp routes or a more specialized subrouting:
<script lang="ts">
import { resolve } from 'elegua';
</script>
{#if resolve('/blog/:slug')}
{#if $params['slug'] == "1"}
<Post1/>
{:else if $params['slug'] == "2"}
<Post2/>
{:else if $params['slug'] == "3"}
<Post3/>
{:else}
<h1>Post not found</h1>
{/if}
{/if}
If you might want a route to match only on certain specific path patterns; ex: /users/123
. For that, use regexp routes by passing a regexp as route in the resolve(path, route)
method:
{#if resolve($path, /\/users/([0-9]+)/)}
<p>Rendering user {$match[1]} page</p>
{/if}
\/users
will not match this route.\/users\/howe
will not match this route.\/users\/123
will match this route, and $match[1]
will be 123
You could use other patterns in the same way. Ex:
resolve($path, '\/users\/(howe|steve)')
=> $match[1] will match "/users/howe"
or "/users/steve"
resolve($path, '\/users\/([a-zA-Z\_\\])*')
=> inspect $match[1], $match[2]Named groups work as expected, and captured groups will be reflected in $params
. After (and only after ) resolve()
is called, $match
and $params
will be redefined.
{#if resolve($path, /users\/(?<user_id>[0-9]+)/)}
<p>User by $match: {$match && $match[1]}</p>
<p>User by $param: {$param['user_id']}</p>
{/if}
Often you want to highlight a nav menu item when user is on that page, so that he can see at a glance where they are. For instance, if you are in "/about"
and your nav menu has the following links:
BLOG | ORDERS | ABOUT
... and you want ABOUT
highlighted when user is on that page (/about
).
In this case, just set a dynamic class inspecting $path
:
<script lang="ts">
import { path } from 'elegua';
</script>
<nav>
<a href={'/blog'} class:selected={$path === '/blog' || $path.startsWith('/blog/')}>BLOG </a>
<a href={'/orders'} class:selected={$path === '/orders'}>ORDERS </a>
<a href={'/about'} class:selected={$path === '/about'}>ABOUT </a>
</nav>
<style>
.selected {
font-weight: bold;
}
</style>
Now when you are on /about
, nav menu will show something as:
BLOG | ORDERS | ABOUT
Subscribe to $url. It's the DOM URL object for the current browser's url. Then you can do anything you want with it.
No need to bloat Elegua with that. Just use <meta refresh="url...">
:
<script lang="ts">
import { path } from 'elegua';
</script>
{#if $path ==="/old_blog"}
<svelte:head>
<meta http-equiv="refresh" content="3; url = /blog"/>
</sve
Blog has changed path; redirecting...
{/if}
goto()
is another way to do it:
<script lang="ts">
import { goto } from 'elegua';
import { onMount } from 'svelte';
onMount(() => {
goto('/new_path');
});
</script>
I would have implemented this better, but Vite only allows string literals in import.meta.glob()
calls, so I end up impolthis had to be manual.
await Promise.all(
// '/src/posts' is fixed below because vite only accepts literals.
// If posts dir gets moved, it has to be updated accordingly
Object.entries(import.meta.glob('/src/posts/**/*.md')).map(async ([path, resolver]) => {
// ... handle path/resolver, vite style
})
);
There is an example of this on the demo page.
Just use history.replaceState()
:
history.replaceState({}, '', '/blog');
Of course. That's the point about this lib.
Same as usual. Ex: loading post by hash (async):
http://localhost/blog/#mypost
<script lang="ts">
import { path, hash } from 'elegua';
</script>
{#if $path === '/blog'}
{#await loadPost($hash)}
<p>Loading post {$hash}...</p>
{:then post}
<Post data={post} />
{/await}
{/if}
Sure:
{#if $path.startsWith("/blog")}
<!-- Only pages starting with "/blog" from now on: /blog, /blog/post1, /blog/post2... -->
{/if}
{#if $path.test(/myregexp/)}
<!-- Match $path against regexp -->
{/if}
If you want, you can let the browser route the links natively, skipping Elegua. Just add the data-native-router
attribute to attribute to the link:
<a href="https://github.com/howesteve/elegua/blob/master/blog" data-native-router>Blog</a>
When you can click such links, you can see the browser's "refresh" animation active, indicating the link was loaded from server, skipping Elegua.
Your server must redirect all requests to /index.html
so that Elegua gets loaded and handle the routing by itself; otherwise, the server will try to route and you'll probably not get what you were hoping for.
For instance, if you load /blog
without setting up the server to load /index.html
on all requests, it will reply with a 404
error.
In netlify, where the demo is hosted, this is done by adding a _redirect
file with the following contents:
/* /index.html 200
Yes - see here.
Version 1.x used a <Route>
component; this version routes using resolve()
and $path
, which is much cleaner and more flexible.
No, I'm not benchmarking a client router. However if you care to see the source code, you'll see it's very fast.
That is the Yoruba deity of the paths, directions, and crossroads. Elegua controls the routes.
MIT