Skip to content

RussBaz/VaporHX

Repository files navigation

VaporHX - Swift Vapor + Htmx + Extensions

VaporHX is a collection of Htmx and other extensions that I made when started working with htmx in a personal project (and thus they can be quite opinionated). In any case, please feel free to discuss any changes, as I am open to new and convincing ideas.

NOTE: All the HTMX parts are fully functional and they are actively used by me, even if the documentation is incomplete. I am sorry about this but it writing docs is incredibly tedious.

Core Idea

The core idea is that you can combine your existing API endpoints with HTMX endpoints with minimal effort. The response will depend on the value of the request Accept header and the request method.

All you need to do is to call the hx(template: String) method on your Content struct and return its value. It will automatically pick the appropriate response, whether it is JSON encoded data, a full HTML page or an HTMX fragment. When HTML (HTMX) is returned, your content is injected into the specified template as a context. It uses the Leaf templating engine by default.

However, you do not have to use the Leaf engine if you do not want to. This package defines HXTemplateable protocol with a single render method that returns an html page as a string. For as long as your own templating engine implements it, you can pass its type into the hx(template:) method instead.

import VHX

// Do NOT forget to call 'configureHtmx' in your 'configure' method before trying this snippet in your project

// Also, do NOT create a folder called '--page' in your template root without changing the default VHX settings
// as it is used as a prefix for dynamically generated wrapper page templates
// and the custom template provider is the last one to be checked after the default ones are run

// Furthermore, this snippet assumes the default leaf naming conventions as they can be manually overriden

struct MyApi: Content {
    let name: String
}

func routes(_ app: Application) throws {
    // Combined API and HTMX endpoint
    // 'api.leaf' template must exist
    app.get("api") { req in
        MyApi(name: "name").hx(template: "api")
        // The return type of this function call is 'HX<MyApi>'
    }

    // HTMX only endpoint
    // 'index.leaf' template must exist
    // It will automatically select whether to return a full page or only a fragment if the generic page template was configured.
    // Otherwise, it will simply render the 'index.leaf' template
    app.get { req in
        try await req.htmx.render("index")
    }

    // Or an even quicker definition of simple 'static' routes
    app.get("static", use: staticRoute(template: "my-static-template"))
}

Basic configuration (configure.swift):

import Vapor
import VHX

// Basic config configures Leaf engine but you are not required to use it
// Basic config assumes 'index-base.leaf' template exists and that it contains '#import("body")' tag
// It will generate a dynamic template that wraps the content of the specified template for NON-htmx calls to 'htmx.render'
// It will simply plug the provided template into the 'body' slot of the base template

public func configure(_ app: Application) async throws {
    let config = HtmxConfiguration.basic()
    try configureHtmx(app, configuration: config)
}

Table of Contents

What is HTMX?

Here is my hot take: Make your backend code the single source of truth for your project, and drop most of your front end bloat in favour of updating your HTML in-place and seamlessly. Without reloading the page and with only your server side HTML templates. Learn more at htmx.org.

And here is the official intro:

  • Why should only <a> and <form> be able to make HTTP requests?
  • Why should only click & submit events trigger them?
  • Why should only GET & POST methods be available?
  • Why should you only be able to replace the entire screen?

By removing these arbitrary constraints, htmx completes HTML as a hypertext.

Lastly, here is a quick introduction to HTMX by Fireship: htmx in 100 seconds.

HTMX

Installation

SPM installation:

  • Add the package to your package dependencies
.package(url: "https://github.com/RussBaz/VaporHX.git", from: "0.0.26"),
  • Then add it to your target dependencies
.product(name: "VHX", package: "VaporHX"),

Configuration

Assuming the standard use of `configure.swift' in all the following examples.

The simplest config usign the Leaf engine (without localisation helpers):

import Vapor
import VHX

// Basic config configures Leaf engine but you are not required to use it

// Basic config assumes 'index-base.leaf' template exists and that it contains '#import("body")' tag
// It will generate a dynamic template that wraps the content of the specified template for NON-htmx calls to 'htmx.render'
// It will simply plug the provided template into the 'body' slot of the base template

public func configure(_ app: Application) async throws {
  // other configuration
  let config = HtmxConfiguration.basic()
  try configureHtmx(app, configuration: config)
  // more configuration
}

Please note that the default wrapper allows dynamically changing the base template name and the slot name. Please refer to the render function.

Otherwise, if you want to specify your own htmx page wrapper (plugs the provided template name into a dynamically generated page on NON-HTMX requests):

// The most straightforward configuration
import Vapor
import VHX

// Defining the page dynamic template generator separately
// Check the 'HXBasicLeafSource' later in this section for further details
func pageTemplate(_ template: String) -> String {
  """
  #extend("index-base"): #export("body"): #extend("\(template)") #endexport #endextend
  """
}

public func configure(_ app: Application) async throws {
  // Other configuration
  // HTMX configuration also enables leaf templating language
  try configureHtmx(app, pageTemplate: pageTemplate)
  // Later configuration and routes registration
}

Here are all the signatures:

func configureHtmx(_ app: Application, pageTemplate template: ((_ name: String) -> String)? = nil) throws
// or
func configureHtmx(_ app: Application, configuration: HtmxConfiguration) throws

// -------------------------------------------- //

// This struct stores globally available (through the Application) htmx configuration
struct HtmxConfiguration {
  var pageSource: HXLeafSource
  // A header name that will be copied back from the request when HXError is thrown
  // The header type must be UInt, otherwise 0 is returned
  // Should be used by the client when retrying
  var errorAttemptCountHeaderName: String?

  // Possible ways to init the configuration structure
  init()
  init(pagePrefix prefix: String)
  init(pagePrefix prefix: String = "--page", pageTemplate template: @escaping (_ name: String) -> String)
  init(pageSource: HXLeafSource, errorAttemptCountHeaderName: String? = nil)

  // Default basic configuration
  static func basic(pagePrefix prefix: String = "--page", baseTemplate: String = "index-base", slotName: String = "body") -> Self
}

HXLeafSource is used to generate a dynamic template that is used for wrapping HTMX fragments with the rest of the page content when it is accessed through a normal browser request.

It satisfies the following protocol:

protocol HXLeafSource: LeafSource {
  var pagePrefix: String { get }
}

Where LeafSource is a special Leaf protocol designed for customising how leaf templates are discovered.

Then VaporHX implements its implementation of this specialised protocol.

struct HXBasicLeafSource: HXLeafSource {
  let pagePrefix: String
  // This is our custom template generator
  let pageTemplate: (_ name: String) -> String
}

In order to manually initialise this struct, please use the following functions:

func hxPageLeafSource(prefix: String = "--page", template: ((_ name: String) -> String)?) -> HXLeafSource
func hxBasicPageLeafSource(prefix: String = "--page", baseTemplate: String = "index-base", slotName: String = "body") -> HXLeafSource

In our case the default pagePrefix value is --page. Therefore, everytime you ask leaf for a template prefixed with --page/ (please do not miss / after the prefix, it is always required), the default HXBasicLeafSource will return a template generated by the pageTemplate closure. Everything after the prefix with / will be passed into the page template generator and the result of this function should be a valid leaf template as a string.

The value passed to the pageTemplate method must not be empty. If it is, then HXBasicLeafSource will return a 'not found' error.

Lastly, this LeafSource implementation is registered as a last leaf source, and this means that the default search path is fully preserved.

HX Request Extensions

How to check if the incoming request is an HTMX request?

// Check this extensions property on the 'Request' object
req.htmx.prefered // Bool
// And if you want more more accuracy ...
req.htmx.prefers // Preference

// -------------------------------------------- //

// HTMX case implies an HTMX fragment
// HTML case implies standard browser request
// API case implies json api request
enum Preference {
  case htmx, html, api
}

How to automatically decide if you need to render an HTMX fragment or a full page?

// Try this method on the 'req.htmx' extension
// This method tries to mimic the 'req.view.render' api
// but it also can accept optional HXResponseHeaders
// Setting the 'page' parameter to true will force the server to always return a full page or a page fragment only otherwise
func render(_ name: String, _ context: some Encodable, page: Bool? = nil, headers: HXResponseHeaders? = nil) async throws -> Response

func render(_ name: String, page: Bool? = nil, headers: HXResponseHeaders? = nil) async throws -> Response

// If you are using a custom templating engine, then you should use this method
func render<T: HXTemplateable>(_ template: T.Type, _ context: T.Context, page: Bool? = nil, headers: HXResponseHeaderAddable? = nil) async throws -> Response

Furthermore, if you are using the default template generator, you can manually override the base template name and the slot name. Here is an example how it can be done:

// Use square brackets at the beginning of the name to override default base template and slot names

routes.get("template") { req in
  // Just square brackets without colons to override base template name only
  try await req.htmx.render("[index-custom]name")
}

routes.get("slot") { req in
  // Use the following format to override a slot name: [template:slot]
  // Using multiple colons will result in an error
  try await req.htmx.render("[index-custom:extra]name")
}

To add an HTMX specific header to a response, you can update response.headers extension:

req.htmx.response.headers // HXResponseHeaders

Otherwise, you can provide a header to any render method. It will override any previously specified headers.

To learn more about the HXResponseHeaders, please refere to the Response Headers section.

How to redirect quickly with proper HTMX headers?

// The simplest type of redirect
func redirect(to location: String, htmx: HXRedirect.Kind = .redirect, html: Redirect = .normal, refresh: Bool = false) async throws -> Response

// A helper that looks for a query parameter by the 'key' value and redirects to it
func autoRedirect(key: String = "next", htmx: HXRedirect.Kind = .redirect, html: Redirect = .normal, refresh: Bool = false) async throws -> Response

// A helper that looks for a query parameter by the 'key' value
// And then redirects to a 'through location' while preserving the query parameter from the first step during the redirect
// e.g. it can redirect from '/redirect?next=/dashboard/' to '/login?next=/dashboard/'
// by making 'through' equal to '/login'
func autoRedirect(through location: String, key: String = "next", htmx: HXRedirect.Kind = .redirect, html: Redirect = .normal, refresh: Bool = false) async throws -> Response

// A helper that redirects to 'from location' while adding the current url as query parameter with a name specified by the 'key'
// It preserves query parameteres from the original url
// e.g from /dashboard/ to /login?next=/dashboard/
// by making 'from' equal to '/login'
func autoRedirectBack(from location: String, key: String = "next", htmx: HXRedirect.Kind = .redirect, html: Redirect = .normal, refresh: Bool = false) async throws -> Response

// -------------------------------------------- //

// HXRedirect.Kind
enum Kind {
  case redirect
  case redirectAndPush
  case redirectAndReplace
}

// What is a 'Redirect' type?
// It is simply the default 'Vapor' type which you use with the 'req.redirect'

HX Extension Method and HX<MyType>

The inbuilt Content type was extended with an hx() method. This is the secret ingredient that adds automatic HTMX support to the standard responses. In addition, some other types have been extended, such as HTTPStatus and Abort.

Here is how it works:

// This is a slightly simplified version of this extension method declaration
extension Content where Self: AsyncResponseEncodable & Encodable {
  func hx(template name: String? = nil, page: Bool? = nil, headers: HXResponseHeaders? = nil) -> HX<Self>

  // For custom templating engines
  func hx<T: HXTemplateable>(template: T.Type, page: Bool? = nil, headers: HXResponseHeaders? = nil) -> HX<Self> where T.Context == Self
}

// And this is how you would use it
app.get("api") { req in
// Where 'MyApi' is some Content
  MyApi(name: "name").hx(template: "api")
  // The return type of this function call is 'HX<MyApi>'
}

One should not normally deal with the HX struct directly but in case it is ever needed, here is its definition:

typealias TemplateRenderer = (_ req: Request, _ context: T, _ page: Bool?, _ headers: HXResponseHeaders?) async throws -> Response

struct HX<T: AsyncResponseEncodable & Encodable> {
  let context: T
  let template: TemplateRenderer?
  let page: Bool?
  let htmxHeaders: HXResponseHeaders?
}

HX Templateable and Custom Templating Engines

If you would like to use your own templating engine, then each renderer should implement the following protocol:

protocol HXTemplateable {
    associatedtype Context: AsyncResponseEncodable & Encodable

    static func render(req: Request, isPage: Bool, context: Context) -> String
}

If this protocol is not enough for your use case, please open an issue and we can discuss it there.

Request Headers

// 'HTMX' request header getter on every request
req.htmx.headers

// Request header structure
// For the meaning of value of each header, please refer to the 'HTMX' docs
struct HXRequestHeaders {
  let boosted: Bool
  let currentUrl: String?
  let historyRestoreRequest: Bool
  let prompt: Bool
  let request: Bool
  let target: String?
  let triggerName: String?
  let trigger: String?
}

Response Headers

Overview

HTMX headers are defined as structs with a simple inbuilt validation. They can be added as individual headers to a Response object or as a whole collection by using HXResponseHeaders struct. The latter struct can be passed to htmx specific functions as an optional parameter, such as .htmx.render or .hx

// Cleaned up definition
struct HXResponseHeaders {
  var location: HXLocationHeader?
  var pushUrl: HXPushUrlHeader?
  var redirect: HXRedirectHeader?
  var refresh: HXRefreshHeader?
  var replaceUrl: HXReplaceUrlHeader?
  var reselect: HXReselectHeader?
  var reswap: HXReswapHeader?
  var retarget: HXRetargetHeader?
  var trigger: HXTriggerHeader?
  var triggerAfterSettle: HXTriggerAfterSettleHeader?
  var triggerAfterSwap: HXTriggerAfterSwapHeader?
}

// Example usages
// With 'hx' extension
app.get("api") { req in
  let headers = HXResponseHeaders(retarget: HXRetargetHeader("#content"))
  return MyApi(name: "name").hx(template: "api", headers: headers)
}

// With 'htmx.render' function
app.get("example") { req in
  let headers = HXResponseHeaders(retarget: HXRetargetHeader("#content"))
  return req.htmx.render("example", headers: headers)
}

// With 'add' extension method on 'Response'
// Can be used with the whole container ('HXResponseHeaders') or with individual headers (such as 'HXRetargetHeader')
// Later header values will replace earlier headers
app.get("redirect") { req in
  req.htmx.autoRedirect(through: "/login", html: .temporary).add(headers: HXResponseHeaders())
}

Location

HXLocationHeader is a type-safe constructor for a HX-Location response header. It is the most complicated response header in my opinion.

struct HXLocationHeader {
  let location: HXLocationType
}

enum HXLocationType {
  case simple(String) // Header value is just a string
  case custom(HXCustomLocation) // Header value is json
}

struct HXCustomLocation: Encodable {
  let path: String
  let target: String?
  let source: String?
  let event: String?
  let handler: String?
  let swap: HXReswapHeader?
  let values: [String]?
  let headers: [String: String]?
}

Because of the sheer volume of possible options, there is a 'fluent' api to generate a new header struct with updated property values, such as:

func set(target newTarget: String) -> Self
func set(source newSource: String?) -> Self
func set(event newEvent: String?) -> Self
func set(handler newHandler: String?) -> Self
func set(swap newSwap: HXReswapHeader?) -> Self
func set(values newValues: [String]?) -> Self
func set(headers newHeaders: [String: String]?) -> Self
// Removes all other properties and sets the location property to simple
func makeSimple(_ path: String? = nil) -> Self

A few examples of how to create a simple and more complicated headers:

let header1 = HXLocationHeader("/test") // reuslts in "HX-Location: /test"
let header2 = HXLocationHeader("/test2", target: "#testdiv") // results in "HX-Location: {"path":"/test2", "target":"#testdiv"}"
// and so on

Push Url

HXPushUrlHeader is a type-safe constructor for a HX-Push-Url response header.

struct HXPushUrlHeader {
    let url: HXPushType
}

enum HXPushType {
    case disable
    case enable(String) // where the string should contain the new value to be pushed onto the browser location history stack
}

A few examples:

let header1 = HXPushUrlHeader(url: .disable) // results in "HX-Push-Url: false"
let header2 = HXPushUrlHeader("/test") // results in "HX-Push-Url: /test"

Redirect

HXRedirectHeader is a type-safe constructor for a HX-Redirect response header. Please do not confuse this response header with the HXRedirect structure.

It is extremely simple:

struct HXRedirectHeader {
    let location: String
}

// sample usage
let header = HXRedirectHeader("/test") // results in "HX-Redirect: /test"

Refresh

To be continued...

HTMX+Leaf Demo

For those new to HTMX, this package comes bundled with an HTMX Demo that you can run locally. It consists of a few examples from the HTMX.org website.

This is a showcase of some HTMX features built with the Leaf templating engine and VaporHX. It is not a showcase of all the capabilities of this library, but a good introduction to HTMX. It is also not a production ready example.

To run the demo using Xcode

  • Open this project in Xcode
  • Switch the scheme from VHX to Demo (left hand side of the top url bar)
  • Confirm that the correct Run Target is selected (likely 'My Mac')
  • Set the custom working directory in the scheme editor to the root folder of this package as shown in the Vapor docs. Otherwise, it will display a warning in the console and it will be unable to find leaf templates.
  • Press the play / run button
  • Then head to http://localhost:8080

To run the demo using the command line

  • execute the following from the VaporHX projects root dir
swift run Demo