Type safe URL construction and deconstruction
Until now you've probably been relying on keeping various bits of code in sync, or casting types as
whatever you believe they should be to prevent errors when constructing, matching and or deconstructing URLs.
TSURL prevents the need for you to worry about keeping your path templates in sync with your URL parameters, worries about accidental //
in paths, missing URL parameters, unintentional or missing trailing slashes, URL encoding and decoding, query parameter construction/deconstruction, and type casting.
Built on top of existing libraries that you're probably already using, such as path-to-regexp
, encodeurl
query-string
, url-parse
, etc, TSURL combines their functionality to provide a type safe interface for working with URLs and paths.
This library only supports modern browsers (latest Chrome, Firefox, Safari, etc), due to a reliance on query-string@6
which only has support for modern browsers.
If you need to support older browsers (e.g. Internet Explorer) you should configure your bundler/compiler to compile both @jakesidsmith/tsurl
and query-string
.
npm i @jakesidsmith/tsurl -S
-S
is shorthand for --save
and will automatically add this to your package.json
and package-lock.json
where necessary.
If you've ever used react-router
this should be a great example of how TSURL can improve your code.
Let's define a TSURL for our user details page. This would be something like /users/id/
where id
is the user's ID. react-router
uses path-to-regexp
behind the scenes and so would require a path with the syntax /users/:id/
, but TSURL can handle this for you.
import { createTSURL, requiredString } from '@jakesidsmith/tsurl';
export const userDetailURL = createTSURL(['/users', requiredString('id')], {
trailingSlash: true,
});
We can now use this TSURL instance to supply a path to our routes.
import UserDetailPage from './users';
import { userDetailURL } from './urls';
// ...
<Route path={userDetailURL.getPathTemplate()} component={UserDetailPage} />;
This outputs a path with the same syntax as required by react-router
.
Now, if you wanted to navigate to this URL you'd probably previously have used string concatenation, a template string, or path-to-regexp
to construct the path, but none of these offer type safety. Instead now you can use your TSURL instance to construct a path. It will enforce that you provide a string for the user's ID.
const whereWeWantToGo = userDetailURL.constructPath({ id: 'abc' }, {});
This will output /users/abc/
exactly as we'd expect, but would fail type checks if the id
key was missing from the URL parameters object. The second argument is the query params object, but since we don't need any, we've just provided an empty object. Note: this method will throw an error if you somehow fail to supply the user id e.g. not using type checking, or cast your params to any
.
What if we now needed to extract the current user ID from the URL? Well, we can handle that also.
try {
const { id } = userDetailURL.deconstruct(window.location.href).urlParams;
// Do something with the id
} catch (error) {
// Bail here
}
It's sensible to wrap your deconstruct calls in a try/catch as if the URL doesn't match the schema previously provided, then this will throw an error. Unfortunately there's no way to check that the URL provided will match the schema at compile time.
Say we have a paginated users list, we may be storing the current page number in the query parameters. No worries, TSURL has got your back.
const userListURL = createTSURL(['/users'], {
trailingSlash: true,
queryParams: [optionalNumber('page')],
});
Here we've constructed a TSURL instance that will not only enforce that we provide a number (or nothing as we may not want to define the page number for the first page) when constructing a URL for this route, but also will give us sensible types for the parameters when deconstructing.
userListURL.constructPath({}, {}); // Is fine because page is optional
userListURL.constructPath({}, { page: 2 }); // Is fine because page should be a number
userListURL.constructPath({}, { page: '2' }); // Disallowed by types (would error)
userListURL.constructPath({}, { page: null }); // Disallowed by types (would error)
// The below deconstruct will handle casting the page query param to a number if found
userListURL.deconstruct(window.location.href);
// And output a type matching
interface Result {
urlParams: {};
queryParams: {
page: number | undefined;
};
}
If all of your requests are prefixed with a specific domain and or path you can provide a baseURL
and or basePath
as an option. The baseURL
may include protocol, host, port, and base path, but I'd recommend using basePath
for paths in most circumstances.
Note: the baseURL
is not affected by the normalize
option, except where a base URL with a trailing slash and a path with a leading slash would cause an unwanted double slash e.g. baseURL: https://domain.com/api/
and path /users
would output https://domain.com/api/users
instead of https://domain.com/api//users
.
Example base URLs:
'domain.com'
'https://domain.com'
'http://localhost:1234'
'/api/'
'domain.com/api/'
'https://domain.com/api'
'http://localhost:1234/api'
Example output:
const url = createTSURL(['/users', requiredString('userId')], {
baseURL: 'https://domain.com/',
basePath: '/api',
});
url.getURLTemplate();
// https://domain.com/api/users/:userId
url.getPathTemplate();
// /api/users/:userId
url.constructURL({ userId: 'abc' }, {});
// https://domain.com/api/users/abc
url.constructPath({ userId: 'abc' }, {});
// /api/users/abc
If we have our API running on a different domain we can use groups to pre-fill baseURL
, basePath
(and or any other options) for client and API URLs.
const api = createTSURLGroup({
baseUrl: 'https://server.com',
basePath: '/api',
trailingSlash: true,
});
const API_URLS = {
users: api.createTSURL(['users']),
user: api.createTSURL(['users', requiredString('userId')]),
userImages: api.createTSURL(['users', requiredString('userId'), 'images']),
};
const client = createTSURLGroup({
baseUrl: 'https://client.com',
trailingSlash: false,
});
const CLIENT_URLS = {
users: client.createTSURL(['users']),
user: client.createTSURL(['users', requiredString('userId')]),
};
If we want to match any trailing part of a URL/path we can do so using a splat
URL part.
const url = createTSURL(['user', requiredString('userId'), splat('splat')], {
baseUrl: 'https://server.com',
basePath: '/api',
trailingSlash: false,
});
url.constructPath({ userId: 'abc' }, {});
// returns '/api/user/abc'
url.constructPath({ userId: 'abc', splat: ['posts', '123'] }, {});
// returns '/api/user/abc/posts/123'
url.deconstruct('https://server.com/api/user/abc');
// returns { urlParams: { userId: 'abc', splat: undefined }, queryParams: {} }
url.deconstruct('https://server.com/api/user/123/posts/123');
// returns { urlParams: { userId: '123', splat: ['posts', '123'] }, queryParams: {} }
In some cases you want your URL templates to match a specific template, but when deconstructing you don't mind if the URL/path contains additional sub-paths. In these cases you can pass the ignoreSubPaths
option to the deconstruct method.
const url = createTSURL(['user', requiredString('userId')], {
baseUrl: 'https://server.com',
basePath: '/api',
trailingSlash: false,
});
url.deconstruct('https://server.com/api/user/123/posts/123');
// throws an error
url.deconstruct('https://server.com/api/user/123/posts/123', {
ignoreSubPaths: true,
});
// returns { urlParams: { userId: '123' }, queryParams: {} }
The createTSURL
function, which is also the default export, will construct a TSURL instance.
This takes 1 or 2 arguments:
- URL schema - an array of
strings
and or parameters - An options object (optional) -
Options
- see Options for more info
Returns TSURL
with inferred keys for URL and query params.
The createTSURLGroup
function, returns an object with a createTSURL
function.
This takes 1 argument:
- An options object -
Options
- see Options for more info
As with the named/default exported createTSURL
, but inherits options from the group.
Returns a path-to-regexp
compatible URL (including the baseURL
) string
from your defined schema.
This method takes no arguments.
Returns a path-to-regexp
compatible path (excluding the baseURL
) string
from your defined schema.
This method takes no arguments.
Returns a string
URL (including the baseURL
).
This takes 2 arguments:
- The URL params for this URL - an object with keys that match the required/optional URL params
- The Query params for this URL - an object with keys that match the required/optional query params
Note: this method will throw an error if you have not supplied required url/query params somehow (e.g. if you are not using type checking because your app is written in JavaScript, or you have cast your params to any
in TypeScript).
Returns a string
path (excluding the baseURL
).
This takes 2 arguments:
- The URL params for this URL - an object with keys that match the required/optional URL params
- The Query params for this URL - an object with keys that match the required/optional query params
Note: this method will throw an error if you have not supplied required url/query params somehow (e.g. if you are not using type checking because your app is written in JavaScript, or you have cast your params to any
in TypeScript).
Returns a string
of query params (prefixed with ?
) if any are provided.
This takes 1 argument:
- The Query params for this URL - an object with keys that match the required/optional query params
Note: this method will throw an error if you have not supplied required query params somehow (e.g. if you are not using type checking because your app is written in JavaScript, or you have cast your params to any
in TypeScript).
Returns the URL and query params extracted from a string URL/path.
This takes 1 or 2 arguments:
- URL/path -
string
- the URL you wish to extract parameters from - An options object (optional) -
Options
- see Deconstruct Options for more info
Returns an object with urlParams
and queryParams
keys. Both of these keys will be objects containing the parameters you defined in your schema e.g.
interface Result {
urlParams: {
organization: string;
};
queryParams: {
users: readonly string[];
search: string;
page: number | undefined;
};
}
Note: this method will throw an error if the URL/path does not match the previously defined schema. You should always wrap calls to deconstruct in a try/catch as the string that you provide contains no type information, and we cannot check at compile time.
The options object is the second argument to the createTSURL
function. All available options are optional.
Options include:
baseURL
-string
- base URL to prefix constructed URLs with (can include protocol, host, port, and base path e.g.https://domain.com/api
).basePath
-string | readonly string[]
- path to prefix the schema (path parts) for all URLs created with the group.trailingSlash
-boolean
- enforce or remove trailing slashes. Does nothing by default.encode
-boolean
- whether to encode the URL when constructing. Defaults totrue
.decode
-boolean
- whether to decode the URL when deconstructing. Default totrue
.normalize
-boolean
- whether to strip all double slashes from the path (//
, excluding thebaseURL
, except where this causes multiple trailing slashes e.g.createTSURL
withbaseURL: 'https://domain.com/api/'
and path parts['/users']
will constructhttps://domain.com/api/users
). Defaults totrue
.queryArrayFormat
- how to handle constructing/deconstructing query params that can have multiple values. This option is defined by thequery-string
package.queryArrayFormatSeparator
-string
- the separator to use whenqueryArrayFormat
is set toseparator
. Defaults to,
.queryParams
- an array of parameters.
This is the second argument to the deconstruct
function.
Options include:
ignoreSubPaths
-boolean
- whether the deconstruction will allow sub-paths (stuff that appears after your defined template) in the provided URL/pathignoreInvalidEnums
-boolean
- whether the deconstruction will error or omit invalid values if the URL contains values for enum restricted fields that do not adhere to the enum (does not apply torequiredEnum
fields)
There are a lot of functions that you can use to define parameters.
The URL schema supports the following:
requiredString
requiredNumber
requiredBoolean
requiredEnum
optionalString
optionalNumber
optionalBoolean
optionalEnum
splat
The query params schema supports the following:
requiredString
requiredNumber
requiredBoolean
requiredEnum
optionalString
optionalNumber
optionalBoolean
optionalEnum
requiredStringArray
requiredNumberArray
requiredBooleanArray
requiredEnumArray
optionalStringArray
optionalNumberArray
optionalBooleanArray
optionalEnumArray
Ensure you are using a compatible version of NodeJS (16) and NPM (8).
If you're using NVM you can simply:
nvm use
And ensure you have NPM 8 installed globally:
npm i npm@8 -g
Then run a clean install to get fresh dependencies:
npm ci
Run type-checking, linting and tests with:
npm test
You can fix formatting issues by running:
npm run prettier