Defines metadata for a LiveView.

This must be returned from the __live__ callback.

It accepts:

• :container - an optional tuple for the HTML tag and DOM attributes to be used for the LiveView container. For example: {:li, style: "color: blue;"}.

• :layout - configures the layout the LiveView will be rendered in. This layout can be overridden by on mount/3 or via the :layout option in Phoenix.LiveView.Router.live_session/2

• :log - configures the log level for the LiveView, either false or a log level

• :on_mount - a list of tuples with module names and argument to be invoked as on_mount hooks

Uses LiveView in the current module to mark it a LiveView.

use Phoenix.LiveView,
  container: {:tr, class: "colorized"},
  layout: {MyAppWeb.Layouts, :app},
  log: :info

Options

• :container - an optional tuple for the HTML tag and DOM attributes to be used for the LiveView container. For example: {:li, style: "color: blue;"}. See Phoenix.Component.live_render/3 for more information and examples.

• :global_prefixes - the global prefixes to use for components. See Global Attributes in Phoenix.Component for more information.

• :layout - configures the layout the LiveView will be rendered in. This layout can be overridden by on mount/3 or via the :layout option in Phoenix.LiveView.Router.live_session/2

• :log - configures the log level for the LiveView, either false or a log level

Allows an upload for the provided name.

Options

• :accept - Required. A list of unique file extensions (such as ".jpeg") or mime type (such as "image/jpeg" or "image/*"). You may also pass the atom :any instead of a list to support to allow any kind of file. For example, [".jpeg"], :any, etc.

• :max_entries - The maximum number of selected files to allow per file input. Defaults to 1.

• :max_file_size - The maximum file size in bytes to allow to be uploaded. Defaults 8MB. For example, 12_000_000.

• :chunk_size - The chunk size in bytes to send when uploading. Defaults 64_000.

• :chunk_timeout - The time in milliseconds to wait before closing the upload channel when a new chunk has not been received. Defaults to 10_000.

• :external - A 2-arity function for generating metadata for external client uploaders. This function must return either {:ok, meta, socket} or {:error, meta, socket} where meta is a map. See the Uploads section for example usage.

• :progress - An optional 3-arity function for receiving progress events.

• :auto_upload - Instructs the client to upload the file automatically on file selection instead of waiting for form submits. Defaults to false.

• :writer - A module implementing the Phoenix.LiveView.UploadWriter behaviour to use for writing the uploaded chunks. Defaults to writing to a temporary file for consumption. See the Phoenix.LiveView.UploadWriter docs for custom usage.

Raises when a previously allowed upload under the same name is still active.

Examples

allow_upload(socket, :avatar, accept: ~w(.jpg .jpeg), max_entries: 2)
allow_upload(socket, :avatar, accept: :any)

For consuming files automatically as they are uploaded, you can pair auto_upload: true with a custom progress function to consume the entries as they are completed. For example:

allow_upload(socket, :avatar, accept: :any, progress: &handle_progress/3, auto_upload: true)

defp handle_progress(:avatar, entry, socket) do
  if entry.done? do
    uploaded_file =
      consume_uploaded_entry(socket, entry, fn %{} = meta ->
        {:ok, ...}
      end)

    {:noreply, put_flash(socket, :info, "file #{uploaded_file.name} uploaded")}
  else
    {:noreply, socket}
  end
end

Assigns keys asynchronously.

Wraps your function in a task linked to the caller, errors are wrapped. Each key passed to assign_async/3 will be assigned to an %AsyncResult{} struct holding the status of the operation and the result when the function completes.

The task is only started when the socket is connected.

Options

• :supervisor - allows you to specify a Task.Supervisor to supervise the task.
• :reset - remove previous results during async operation when true. Possible values are true, false, or a list of keys to reset. Defaults to false.

Examples

def mount(%{"slug" => slug}, _, socket) do
  {:ok,
   socket
   |> assign(:foo, "bar")
   |> assign_async(:org, fn -> {:ok, %{org: fetch_org!(slug)}} end)
   |> assign_async([:profile, :rank], fn -> {:ok, %{profile: ..., rank: ...}} end)}
end

See the moduledoc for more information.

assign_async/3 and send_update/3

Since the code inside assign_async/3 runs in a separate process, send_update(Component, data) does not work inside assign_async/3, since send_update/2 assumes it is running inside the LiveView process. The solution is to explicitly send the update to the LiveView:

parent = self()
assign_async(socket, :org, fn ->
  # ...
  send_update(parent, Component, data)
end)

Attaches the given fun by name for the lifecycle stage into socket.

Note: This function is for server-side lifecycle callbacks. For client-side hooks, see the JS Interop guide.

Hooks provide a mechanism to tap into key stages of the LiveView lifecycle in order to bind/update assigns, intercept events, patches, and regular messages when necessary, and to inject common functionality. Use attach_hook/4 on any of the following lifecycle stages: :handle_params, :handle_event, :handle_info, :handle_async, and :after_render. To attach a hook to the :mount stage, use on_mount/1.

Note: only :after_render and :handle_event hooks are currently supported in LiveComponents.

Return Values

Lifecycle hooks take place immediately before a given lifecycle callback is invoked on the LiveView. With the exception of :after_render, a hook may return {:halt, socket} to halt the reduction, otherwise it must return {:cont, socket} so the operation may continue until all hooks have been invoked for the current stage.

For :after_render hooks, the socket itself must be returned. Any updates to the socket assigns will not trigger a new render or diff calculation to the client.

Halting the lifecycle

Note that halting from a hook will halt the entire lifecycle stage. This means that when a hook returns {:halt, socket} then the LiveView callback will not be invoked. This has some implications.

Implications for plugin authors

When defining a plugin that matches on specific callbacks, you must define a catch-all clause, as your hook will be invoked even for events you may not be interested in.

Implications for end-users

Allowing a hook to halt the invocation of the callback means that you can attach hooks to intercept specific events before detaching themselves, while allowing other events to continue normally.

Replying to events

Hooks attached to the :handle_event stage are able to reply to client events by returning {:halt, reply, socket}. This is useful especially for JavaScript interoperability because a client hook can push an event and receive a reply.

Examples

Attaching and detaching a hook:

def mount(_params, _session, socket) do
  socket =
    attach_hook(socket, :my_hook, :handle_event, fn
      "very-special-event", _params, socket ->
        # Handle the very special event and then detach the hook
        {:halt, detach_hook(socket, :my_hook, :handle_event)}

      _event, _params, socket ->
        {:cont, socket}
    end)

  {:ok, socket}
end

Replying to a client event:

# JavaScript:
# /**
#  * @type {Object.<string, import("phoenix_live_view").ViewHook>}
#  */
# let Hooks = {}
# Hooks.ClientHook = {
#   mounted() {
#     this.pushEvent("ClientHook:mounted", {hello: "world"}, (reply) => {
#       console.log("received reply:", reply)
#     })
#   }
# }
# let liveSocket = new LiveSocket("/live", Socket, {hooks: Hooks, ...})

def render(assigns) do
  ~H"""
  <div id="my-client-hook" phx-hook="ClientHook"></div>
  """
end

def mount(_params, _session, socket) do
  socket =
    attach_hook(socket, :reply_on_client_hook_mounted, :handle_event, fn
      "ClientHook:mounted", params, socket ->
        {:halt, params, socket}

      _, _, socket ->
        {:cont, socket}
    end)

  {:ok, socket}
end

Cancels an async operation if one exists.

Accepts either the %AsyncResult{} when using assign_async/3 or the key passed to start_async/3.

The underlying process will be killed with the provided reason, or with {:shutdown, :cancel} if no reason is passed. For assign_async/3 operations, the :failed field will be set to {:exit, reason}. For start_async/3, the handle_async/3 callback will receive {:exit, reason} as the result.

Returns the %Phoenix.LiveView.Socket{}.

Examples

cancel_async(socket, :preview)
cancel_async(socket, :preview, :my_reason)
cancel_async(socket, socket.assigns.preview)

Cancels an upload for the given entry.

Examples

<%= for entry <- @uploads.avatar.entries do %>
  ...
  <button phx-click="cancel-upload" phx-value-ref={entry.ref}>cancel</button>
<% end %>

def handle_event("cancel-upload", %{"ref" => ref}, socket) do
  {:noreply, cancel_upload(socket, :avatar, ref)}
end

Clears the flash.

Examples

iex> clear_flash(socket)

Clearing the flash can also be triggered on the client and natively handled by LiveView using the lv:clear-flash event.

For example:

<p class="alert" phx-click="lv:clear-flash">
  <%= Phoenix.Flash.get(@flash, :info) %>
</p>

Clears a key from the flash.

Examples

iex> clear_flash(socket, :info)

Clearing the flash can also be triggered on the client and natively handled by LiveView using the lv:clear-flash event.

For example:

<p class="alert" phx-click="lv:clear-flash" phx-value-key="info">
  <%= Phoenix.Flash.get(@flash, :info) %>
</p>

Returns true if the socket is connected.

Useful for checking the connectivity status when mounting the view. For example, on initial page render, the view is mounted statically, rendered, and the HTML is sent to the client. Once the client connects to the server, a LiveView is then spawned and mounted statefully within a process. Use connected?/1 to conditionally perform stateful work, such as subscribing to pubsub topics, sending messages, etc.

Examples

defmodule DemoWeb.ClockLive do
  use Phoenix.LiveView
  ...
  def mount(_params, _session, socket) do
    if connected?(socket), do: :timer.send_interval(1000, self(), :tick)

    {:ok, assign(socket, date: :calendar.local_time())}
  end

  def handle_info(:tick, socket) do
    {:noreply, assign(socket, date: :calendar.local_time())}
  end
end

Consumes the uploaded entries.

Raises when there are still entries in progress. Typically called when submitting a form to handle the uploaded entries alongside the form data. For form submissions, it is guaranteed that all entries have completed before the submit event is invoked. Once entries are consumed, they are removed from the upload.

The function passed to consume may return a tagged tuple of the form {:ok, my_result} to collect results about the consumed entries, or {:postpone, my_result} to collect results, but postpone the file consumption to be performed later.

A list of all my_result values produced by the passed function is returned, regardless of whether they were consumed or postponed.

Examples

def handle_event("save", _params, socket) do
  uploaded_files =
    consume_uploaded_entries(socket, :avatar, fn %{path: path}, _entry ->
      dest = Path.join("priv/static/uploads", Path.basename(path))
      File.cp!(path, dest)
      {:ok, ~p"/uploads/#{Path.basename(dest)}"}
    end)
  {:noreply, update(socket, :uploaded_files, &(&1 ++ uploaded_files))}
end

Consumes an individual uploaded entry.

Raises when the entry is still in progress. Typically called when submitting a form to handle the uploaded entries alongside the form data. Once entries are consumed, they are removed from the upload.

This is a lower-level feature than consume_uploaded_entries/3 and useful for scenarios where you want to consume entries as they are individually completed.

Like consume_uploaded_entries/3, the function passed to consume may return a tagged tuple of the form {:ok, my_result} to collect results about the consumed entries, or {:postpone, my_result} to collect results, but postpone the file consumption to be performed later.

Examples

def handle_event("save", _params, socket) do
  case uploaded_entries(socket, :avatar) do
    {[_|_] = entries, []} ->
      uploaded_files = for entry <- entries do
        consume_uploaded_entry(socket, entry, fn %{path: path} ->
          dest = Path.join("priv/static/uploads", Path.basename(path))
          File.cp!(path, dest)
          {:ok, ~p"/uploads/#{Path.basename(dest)}"}
        end)
      end
      {:noreply, update(socket, :uploaded_files, &(&1 ++ uploaded_files))}

    _ ->
      {:noreply, socket}
  end
end

Detaches a hook with the given name from the lifecycle stage.

Note: This function is for server-side lifecycle callbacks. For client-side hooks, see the JS Interop guide.

If no hook is found, this function is a no-op.

Examples

def handle_event(_, socket) do
  {:noreply, detach_hook(socket, :hook_that_was_attached, :handle_event)}
end

Revokes a previously allowed upload from allow_upload/3.

Examples

disallow_upload(socket, :avatar)

Accesses a given connect info key from the socket.

The following keys are supported: :peer_data, :trace_context_headers, :x_headers, :uri, and :user_agent.

The connect information is available only during mount. During disconnected render, all keys are available. On connected render, only the keys explicitly declared in your socket are available. See Phoenix.Endpoint.socket/3 for a complete description of the keys.

Examples

The first step is to declare the connect_info you want to receive. Typically, it includes at least the session, but you must include all other keys you want to access on connected mount, such as :peer_data:

socket "/live", Phoenix.LiveView.Socket,
  websocket: [connect_info: [:peer_data, session: @session_options]]

Those values can now be accessed on the connected mount as get_connect_info/2:

def mount(_params, _session, socket) do
  peer_data = get_connect_info(socket, :peer_data)
  {:ok, assign(socket, ip: peer_data.address)}
end

If the key is not available, usually because it was not specified in connect_info, it returns nil.

Accesses the connect params sent by the client for use on connected mount.

Connect params are only sent when the client connects to the server and only remain available during mount. nil is returned when called in a disconnected state and a RuntimeError is raised if called after mount.

Reserved params

The following params have special meaning in LiveView:

• "_csrf_token" - the CSRF Token which must be explicitly set by the user when connecting
• "_mounts" - the number of times the current LiveView is mounted. It is 0 on first mount, then increases on each reconnect. It resets when navigating away from the current LiveView or on errors
• "_track_static" - set automatically with a list of all href/src from tags with the phx-track-static annotation in them. If there are no such tags, nothing is sent
• "_live_referer" - sent by the client as the referer URL when a live navigation has occurred from push_navigate or client link navigate.

Examples

def mount(_params, _session, socket) do
  {:ok, assign(socket, width: get_connect_params(socket)["width"] || @width)}
end

Declares a module callback to be invoked on the LiveView's mount.

The function within the given module, which must be named on_mount, will be invoked before both disconnected and connected mounts. The hook has the option to either halt or continue the mounting process as usual. If you wish to redirect the LiveView, you must halt, otherwise an error will be raised.

Tip: if you need to define multiple on_mount callbacks, avoid defining multiple modules. Instead, pass a tuple and use pattern matching to handle different cases:

def on_mount(:admin, _params, _session, socket) do
  {:cont, socket}
end

def on_mount(:user, _params, _session, socket) do
  {:cont, socket}
end

And then invoke it as:

on_mount {MyAppWeb.SomeHook, :admin}
on_mount {MyAppWeb.SomeHook, :user}

Registering on_mount hooks can be useful to perform authentication as well as add custom behaviour to other callbacks via attach_hook/4.

The on_mount callback can return a keyword list of options as a third element in the return tuple. These options are identical to what can optionally be returned in mount/3.

Examples

The following is an example of attaching a hook via Phoenix.LiveView.Router.live_session/3:

# lib/my_app_web/live/init_assigns.ex
defmodule MyAppWeb.InitAssigns do
  @moduledoc """
  Ensures common `assigns` are applied to all LiveViews attaching this hook.
  """
  import Phoenix.LiveView
  import Phoenix.Component

  def on_mount(:default, _params, _session, socket) do
    {:cont, assign(socket, :page_title, "DemoWeb")}
  end

  def on_mount(:user, params, session, socket) do
    # code
  end

  def on_mount(:admin, _params, _session, socket) do
    {:cont, socket, layout: {DemoWeb.Layouts, :admin}}
  end
end

# lib/my_app_web/router.ex
defmodule MyAppWeb.Router do
  use MyAppWeb, :router

  # pipelines, plugs, etc.

  live_session :default, on_mount: MyAppWeb.InitAssigns do
    scope "/", MyAppWeb do
      pipe_through :browser
      live "/", PageLive, :index
    end
  end

  live_session :authenticated, on_mount: {MyAppWeb.InitAssigns, :user} do
    scope "/", MyAppWeb do
      pipe_through [:browser, :require_user]
      live "/profile", UserLive.Profile, :index
    end
  end

  live_session :admins, on_mount: {MyAppWeb.InitAssigns, :admin} do
    scope "/admin", MyAppWeb.Admin do
      pipe_through [:browser, :require_user, :require_admin]
      live "/", AdminLive.Index, :index
    end
  end
end

Pushes an event to the client.

Events can be handled in two ways:

1. They can be handled on window via addEventListener. A "phx:" prefix will be added to the event name.

2. They can be handled inside a hook via handleEvent.

Events are dispatched to all active hooks on the client who are handling the given event. If you need to scope events, then this must be done by namespacing them.

Events pushed during push_navigate are currently discarded, as the LiveView is immediately dismounted.

Hook example

If you push a "scores" event from your LiveView:

{:noreply, push_event(socket, "scores", %{points: 100, user: "josé"})}

A hook declared via phx-hook can handle it via handleEvent:

this.handleEvent("scores", data => ...)

window example

All events are also dispatched on the window. This means you can handle them by adding listeners. For example, if you want to remove an element from the page, you can do this:

{:noreply, push_event(socket, "remove-el", %{id: "foo-bar"})}

And now in your app.js you can register and handle it:

window.addEventListener(
  "phx:remove-el",
  e => document.getElementById(e.detail.id).remove()
)

Annotates the socket for navigation to another LiveView.

The current LiveView will be shutdown and a new one will be mounted in its place, without reloading the whole page. This can also be used to remount the same LiveView, in case you want to start fresh. If you want to navigate to the same LiveView without remounting it, use push_patch/2 instead.

Options

• :to - the required path to link to. It must always be a local path
• :replace - the flag to replace the current history or push a new state. Defaults false.

Examples

{:noreply, push_navigate(socket, to: "/")}
{:noreply, push_navigate(socket, to: "/", replace: true)}

Annotates the socket for navigation within the current LiveView.

When navigating to the current LiveView, handle_params/3 is immediately invoked to handle the change of params and URL state. Then the new state is pushed to the client, without reloading the whole page while also maintaining the current scroll position. For live navigation to another LiveView, use push_navigate/2.

Options

• :to - the required path to link to. It must always be a local path
• :replace - the flag to replace the current history or push a new state. Defaults false.

Examples

{:noreply, push_patch(socket, to: "/")}
{:noreply, push_patch(socket, to: "/", replace: true)}

Adds a flash message to the socket to be displayed.

Note: While you can use put_flash/3 inside a Phoenix.LiveComponent, components have their own @flash assigns. The @flash assign in a component is only copied to its parent LiveView if the component calls push_navigate/2 or push_patch/2.

Note: You must also place the Phoenix.LiveView.Router.fetch_live_flash/2 plug in your browser's pipeline in place of fetch_flash for LiveView flash messages be supported, for example:

import Phoenix.LiveView.Router

pipeline :browser do
  ...
  plug :fetch_live_flash
end

In a typical LiveView application, the message will be rendered by the CoreComponents’ flash/1 component. It is up to this function to determine what kind of messages it supports. By default, the :info and :error kinds are handled.

Examples

iex> put_flash(socket, :info, "It worked!")
iex> put_flash(socket, :error, "You can't access that page")

Puts a new private key and value in the socket.

Privates are not change tracked. This storage is meant to be used by users and libraries to hold state that doesn't require change tracking. The keys should be prefixed with the app/library name.

Examples

Key values can be placed in private:

put_private(socket, :myapp_meta, %{foo: "bar"})

And then retrieved:

socket.private[:myapp_meta]

Annotates the socket for redirect to a destination path.

Note: LiveView redirects rely on instructing client to perform a window.location update on the provided redirect location. The whole page will be reloaded and all state will be discarded.

Options

• :to - the path to redirect to. It must always be a local path
• :external - an external path to redirect to. Either a string or {scheme, url} to redirect to a custom scheme

Examples

{:noreply, redirect(socket, to: "/")}
{:noreply, redirect(socket, external: "https://example.com")}

Configures which function to use to render a LiveView/LiveComponent.

By default, LiveView invokes the render/1 function in the same module the LiveView/LiveComponent is defined, passing assigns as its sole argument. This function allows you to set a different rendering function.

One possible use case for this function is to set a different template on disconnected render. When the user first accesses a LiveView, we will perform a disconnected render to send to the browser. This is useful for several reasons, such as reducing the time to first paint and for search engine indexing.

However, when LiveView is gated behind an authentication page, it may be useful to render a placeholder on disconnected render and perform the full render once the WebSocket connects. This can be achieved with render_with/2 and is particularly useful on complex pages (such as dashboards and reports).

To do so, you must simply invoke render_with(socket, &some_function_component/1), configuring your socket with a new rendering function.

Asynchronously updates a Phoenix.LiveComponent with new assigns.

The pid argument is optional and it defaults to the current process, which means the update instruction will be sent to a component running on the same LiveView. If the current process is not a LiveView or you want to send updates to a live component running on another LiveView, you should explicitly pass the LiveView's pid instead.

The second argument can be either the value of the @myself or the module of the live component. If you pass the module, then the :id that identifies the component must be passed as part of the assigns.

When the component receives the update, update_many/1 will be invoked if it is defined, otherwise update/2 is invoked with the new assigns. If update/2 is not defined all assigns are simply merged into the socket. The assigns received as the first argument of the update/2 callback will only include the new assigns passed from this function. Pre-existing assigns may be found in socket.assigns.

While a component may always be updated from the parent by updating some parent assigns which will re-render the child, thus invoking update/2 on the child component, send_update/3 is useful for updating a component that entirely manages its own state, as well as messaging between components mounted in the same LiveView.

Examples

def handle_event("cancel-order", _, socket) do
  ...
  send_update(Cart, id: "cart", status: "cancelled")
  {:noreply, socket}
end

def handle_event("cancel-order-asynchronously", _, socket) do
  ...
  pid = self()

  Task.Supervisor.start_child(MyTaskSup, fn ->
    # Do something asynchronously
    send_update(pid, Cart, id: "cart", status: "cancelled")
  end)

  {:noreply, socket}
end

def render(assigns) do
  ~H"""
  <.some_component on_complete={&send_update(@myself, completed: &1)} />
  """
end

Similar to send_update/3 but the update will be delayed according to the given time_in_milliseconds.

It returns a reference which can be cancelled with Process.cancel_timer/1.

Examples

def handle_event("cancel-order", _, socket) do
  ...
  send_update_after(Cart, [id: "cart", status: "cancelled"], 3000)
  {:noreply, socket}
end

def handle_event("cancel-order-asynchronously", _, socket) do
  ...
  pid = self()

  Task.start(fn ->
    # Do something asynchronously
    send_update_after(pid, Cart, [id: "cart", status: "cancelled"], 3000)
  end)

  {:noreply, socket}
end

Wraps your function in an asynchronous task and invokes a callback name to handle the result.

The task is linked to the caller and errors/exits are wrapped. The result of the task is sent to the handle_async/3 callback of the caller LiveView or LiveComponent.

The task is only started when the socket is connected.

Options

• :supervisor - allows you to specify a Task.Supervisor to supervise the task.

Examples

def mount(%{"id" => id}, _, socket) do
  {:ok,
   socket
   |> assign(:org, AsyncResult.loading())
   |> start_async(:my_task, fn -> fetch_org!(id) end)}
end

def handle_async(:my_task, {:ok, fetched_org}, socket) do
  %{org: org} = socket.assigns
  {:noreply, assign(socket, :org, AsyncResult.ok(org, fetched_org))}
end

def handle_async(:my_task, {:exit, reason}, socket) do
  %{org: org} = socket.assigns
  {:noreply, assign(socket, :org, AsyncResult.failed(org, {:exit, reason}))}
end

See the moduledoc for more information.

Returns true if the socket is connected and the tracked static assets have changed.

This function is useful to detect if the client is running on an outdated version of the marked static files. It works by comparing the static paths sent by the client with the one on the server.

Note: this functionality requires Phoenix v1.5.2 or later.

To use this functionality, the first step is to annotate which static files you want to be tracked by LiveView, with the phx-track-static. For example:

<link phx-track-static rel="stylesheet" href={~p"/assets/app.css"} />
<script defer phx-track-static type="text/javascript" src={~p"/assets/app.js"}></script>

Now, whenever LiveView connects to the server, it will send a copy src or href attributes of all tracked statics and compare those values with the latest entries computed by mix phx.digest in the server.

The tracked statics on the client will match the ones on the server the huge majority of times. However, if there is a new deployment, those values may differ. You can use this function to detect those cases and show a banner to the user, asking them to reload the page. To do so, first set the assign on mount:

def mount(params, session, socket) do
  {:ok, assign(socket, static_changed?: static_changed?(socket))}
end

And then in your views:

<%= if @static_changed? do %>
  <div id="reload-static">
    The app has been updated. Click here to <a href="#" onclick="window.location.reload()">reload</a>.
  </div>
<% end %>

If you prefer, you can also send a JavaScript script that immediately reloads the page.

Note: only set phx-track-static on your own assets. For example, do not set it in external JavaScript files:

<script defer phx-track-static type="text/javascript" src="https://ajax.googleapis.com/ajax/libs/jquery/3.4.1/jquery.min.js"></script>

Because you don't actually serve the file above, LiveView will interpret the static above as missing, and this function will return true.

@spec stream(
  %Phoenix.LiveView.Socket{
    assigns: term(),
    endpoint: term(),
    fingerprints: term(),
    host_uri: term(),
    id: term(),
    parent_pid: term(),
    private: term(),
    redirected: term(),
    root_pid: term(),
    router: term(),
    transport_pid: term(),
    view: term()
  },
  name :: atom() | String.t(),
  items :: Enumerable.t(),
  opts :: Keyword.t()
) :: %Phoenix.LiveView.Socket{
  assigns: term(),
  endpoint: term(),
  fingerprints: term(),
  host_uri: term(),
  id: term(),
  parent_pid: term(),
  private: term(),
  redirected: term(),
  root_pid: term(),
  router: term(),
  transport_pid: term(),
  view: term()
}

Assigns a new stream to the socket or inserts items into an existing stream. Returns an updated socket.

Streams are a mechanism for managing large collections on the client without keeping the resources on the server.

• name - A string or atom name of the key to place under the @streams assign.
• items - An enumerable of items to insert.

The following options are supported:

• :at - The index to insert or update the items in the collection on the client. By default -1 is used, which appends the items to the parent DOM container. A value of 0 prepends the items.

  Note that this operation is equal to inserting the items one by one, each at the given index. Therefore, when inserting multiple items at an index other than -1, the UI will display the items in reverse order:

  stream(socket, :songs, [song1, song2, song3], at: 0)

  In this case the UI will prepend song1, then song2 and then song3, so it will show song3, song2, song1 and then any previously inserted items.

  To insert in the order of the list, use Enum.reverse/1:

  stream(socket, :songs, Enum.reverse([song1, song2, song3]), at: 0)

• :reset - A boolean to reset the stream on the client or not. Defaults to false.

• :limit - An optional positive or negative number of results to limit on the UI on the client. As new items are streamed, the UI will remove existing items to maintain the limit. For example, to limit the stream to the last 10 items in the UI while appending new items, pass a negative value:

  stream(socket, :songs, songs, at: -1, limit: -10)

  Likewise, to limit the stream to the first 10 items, while prepending new items, pass a positive value:

  stream(socket, :songs, songs, at: 0, limit: 10)

Once a stream is defined, a new @streams assign is available containing the name of the defined streams. For example, in the above definition, the stream may be referenced as @streams.songs in your template. Stream items are temporary and freed from socket state immediately after the render/1 function is invoked (or a template is rendered from disk).

By default, calling stream/4 on an existing stream will bulk insert the new items on the client while leaving the existing items in place. Streams may also be reset when calling stream/4, which we discuss below.

Resetting a stream

To empty a stream container on the client, you can pass :reset with an empty list:

stream(socket, :songs, [], reset: true)

Or you can replace the entire stream on the client with a new collection:

stream(socket, :songs, new_songs, reset: true)

Limiting a stream

It is often useful to limit the number of items in the UI while allowing the server to stream new items in a fire-and-forget fashion. This prevents the server from overwhelming the client with new results while also opening up powerful features like virtualized infinite scrolling. See a complete bidirectional infinite scrolling example with stream limits in the scroll events guide

When a stream exceeds the limit on the client, the existing items will be pruned based on the number of items in the stream container and the limit direction. A positive limit will prune items from the end of the container, while a negative limit will prune items from the beginning of the container.

Note that the limit is not enforced on the first mount/3 render (when no websocket connection was established yet), as it means more data than necessary has been loaded. In such cases, you should only load and pass the desired amount of items to the stream.

When inserting single items using stream_insert/4, the limit needs to be passed as an option for it to be enforced on the client:

stream_insert(socket, :songs, song, limit: -10)

Required DOM attributes

For stream items to be trackable on the client, the following requirements must be met:

1. The parent DOM container must include a phx-update="stream" attribute, along with a unique DOM id.
2. Each stream item must include its DOM id on the item's element.

Note

Failing to place phx-update="stream" on the immediate parent for each stream will result in broken behavior.

Also, do not alter the generated DOM ids, e.g., by prefixing them. Doing so will result in broken behavior.

When consuming a stream in a template, the DOM id and item is passed as a tuple, allowing convenient inclusion of the DOM id for each item. For example:

<table>
  <tbody id="songs" phx-update="stream">
    <tr
      :for={{dom_id, song} <- @streams.songs}
      id={dom_id}
    >
      <td><%= song.title %></td>
      <td><%= song.duration %></td>
    </tr>
  </tbody>
</table>

We consume the stream in a for comprehension by referencing the @streams.songs assign. We used the computed DOM id to populate the <tr> id, then we render the table row as usual.

Now stream_insert/3 and stream_delete/3 may be issued and new rows will be inserted or deleted from the client.

Handling the empty case

When rendering a list of items, it is common to show a message for the empty case. But when using streams, we cannot rely on Enum.empty?/1 or similar approaches to check if the list is empty. Instead we can use the CSS :only-child selector and show the message client side:

<table>
  <tbody id="songs" phx-update="stream">
    <tr id="songs-empty" class="only:block hidden">
      <td colspan="2">No songs found</td>
    </tr>
    <tr
      :for={{dom_id, song} <- @streams.songs}
      id={dom_id}
    >
      <td><%= song.title %></td>
      <td><%= song.duration %></td>
    </tr>
  </tbody>
</table>

Non-stream items in stream containers

In the section on handling the empty case, we showed how to render a message when the stream is empty by rendering a non-stream item inside the stream container.

Note that for non-stream items inside a phx-update="stream" container, the following needs to be considered:

1. Items can be added and updated, but not removed, even if the stream is reset.

This means that if you try to conditionally render a non-stream item inside a stream container, it won't be removed if it was rendered once.

2. Items are affected by the :at option.

For example, when you render a non-stream item at the beginning of the stream container and then prepend items (with at: 0) to the stream, the non-stream item will be pushed down.

@spec stream_configure(
  %Phoenix.LiveView.Socket{
    assigns: term(),
    endpoint: term(),
    fingerprints: term(),
    host_uri: term(),
    id: term(),
    parent_pid: term(),
    private: term(),
    redirected: term(),
    root_pid: term(),
    router: term(),
    transport_pid: term(),
    view: term()
  },
  name :: atom() | String.t(),
  opts :: Keyword.t()
) :: %Phoenix.LiveView.Socket{
  assigns: term(),
  endpoint: term(),
  fingerprints: term(),
  host_uri: term(),
  id: term(),
  parent_pid: term(),
  private: term(),
  redirected: term(),
  root_pid: term(),
  router: term(),
  transport_pid: term(),
  view: term()
}

Configures a stream.

The following options are supported:

• :dom_id - An optional function to generate each stream item's DOM id. The function accepts each stream item and converts the item to a string id. By default, the :id field of a map or struct will be used if the item has such a field, and will be prefixed by the name hyphenated with the id. For example, the following examples are equivalent:

  stream(socket, :songs, songs)
  
  socket
  |> stream_configure(:songs, dom_id: &("songs-#{&1.id}"))
  |> stream(:songs, songs)

A stream must be configured before items are inserted, and once configured, a stream may not be re-configured. To ensure a stream is only configured a single time in a LiveComponent, use the mount/1 callback. For example:

def mount(socket) do
  {:ok, stream_configure(socket, :songs, dom_id: &("songs-#{&1.id}"))}
end

def update(assigns, socket) do
  {:ok, stream(socket, :songs, ...)}
end

Returns an updated socket.

@spec stream_delete(
  %Phoenix.LiveView.Socket{
    assigns: term(),
    endpoint: term(),
    fingerprints: term(),
    host_uri: term(),
    id: term(),
    parent_pid: term(),
    private: term(),
    redirected: term(),
    root_pid: term(),
    router: term(),
    transport_pid: term(),
    view: term()
  },
  name :: atom() | String.t(),
  item :: any()
) :: %Phoenix.LiveView.Socket{
  assigns: term(),
  endpoint: term(),
  fingerprints: term(),
  host_uri: term(),
  id: term(),
  parent_pid: term(),
  private: term(),
  redirected: term(),
  root_pid: term(),
  router: term(),
  transport_pid: term(),
  view: term()
}

Deletes an item from the stream.

The item's DOM is computed from the :dom_id provided in the stream/3 definition. Delete information for this DOM id is sent to the client and the item's element is removed from the DOM, following the same behavior of element removal, such as invoking phx-remove commands and executing client hook destroyed() callbacks.

Examples

def handle_event("delete", %{"id" => id}, socket) do
  song = get_song!(id)
  {:noreply, stream_delete(socket, :songs, song)}
end

See stream_delete_by_dom_id/3 to remove an item without requiring the original data structure.

Returns an updated socket.

@spec stream_delete_by_dom_id(
  %Phoenix.LiveView.Socket{
    assigns: term(),
    endpoint: term(),
    fingerprints: term(),
    host_uri: term(),
    id: term(),
    parent_pid: term(),
    private: term(),
    redirected: term(),
    root_pid: term(),
    router: term(),
    transport_pid: term(),
    view: term()
  },
  name :: atom() | String.t(),
  id :: String.t()
) :: %Phoenix.LiveView.Socket{
  assigns: term(),
  endpoint: term(),
  fingerprints: term(),
  host_uri: term(),
  id: term(),
  parent_pid: term(),
  private: term(),
  redirected: term(),
  root_pid: term(),
  router: term(),
  transport_pid: term(),
  view: term()
}

Deletes an item from the stream given its computed DOM id.

Returns an updated socket.

Behaves just like stream_delete/3, but accept the precomputed DOM id, which allows deleting from a stream without fetching or building the original stream data structure.

Examples

def render(assigns) do
  ~H"""
  <table>
    <tbody id="songs" phx-update="stream">
      <tr
        :for={{dom_id, song} <- @streams.songs}
        id={dom_id}
      >
        <td><%= song.title %></td>
        <td><button phx-click={JS.push("delete", value: %{id: dom_id})}>delete</button></td>
      </tr>
    </tbody>
  </table>
  """
end

def handle_event("delete", %{"id" => dom_id}, socket) do
  {:noreply, stream_delete_by_dom_id(socket, :songs, dom_id)}
end

@spec stream_insert(
  %Phoenix.LiveView.Socket{
    assigns: term(),
    endpoint: term(),
    fingerprints: term(),
    host_uri: term(),
    id: term(),
    parent_pid: term(),
    private: term(),
    redirected: term(),
    root_pid: term(),
    router: term(),
    transport_pid: term(),
    view: term()
  },
  name :: atom() | String.t(),
  item :: any(),
  opts :: Keyword.t()
) :: %Phoenix.LiveView.Socket{
  assigns: term(),
  endpoint: term(),
  fingerprints: term(),
  host_uri: term(),
  id: term(),
  parent_pid: term(),
  private: term(),
  redirected: term(),
  root_pid: term(),
  router: term(),
  transport_pid: term(),
  view: term()
}

Inserts a new item or updates an existing item in the stream.

Returns an updated socket.

See stream/4 for inserting multiple items at once.

The following options are supported:

• :at - The index to insert or update the item in the collection on the client. By default, the item is appended to the parent DOM container. This is the same as passing a limit of -1. If the item already exists in the parent DOM container then it will be updated in place.

• :limit - A limit of items to maintain in the UI. A limit passed to stream/4 does not affect subsequent calls to stream_insert/4, therefore the limit must be passed here as well in order to be enforced. See stream/4 for more information on limiting streams.

Examples

Imagine you define a stream on mount with a single item:

stream(socket, :songs, [%Song{id: 1, title: "Song 1"}])

Then, in a callback such as handle_info or handle_event, you can append a new song:

stream_insert(socket, :songs, %Song{id: 2, title: "Song 2"})

Or prepend a new song with at: 0:

stream_insert(socket, :songs, %Song{id: 2, title: "Song 2"}, at: 0)

Or update an existing song (in this case the :at option has no effect):

stream_insert(socket, :songs, %Song{id: 1, title: "Song 1 updated"}, at: 0)

Or append a new song while limiting the stream to the last 10 items:

stream_insert(socket, :songs, %Song{id: 2, title: "Song 2"}, limit: -10)

Updating Items

As shown, an existing item on the client can be updated by issuing a stream_insert for the existing item. When the client updates an existing item, the item will remain in the same location as it was previously, and will not be moved to the end of the parent children. To both update an existing item and move it to another position, issue a stream_delete, followed by a stream_insert. For example:

song = get_song!(id)

socket
|> stream_delete(:songs, song)
|> stream_insert(:songs, song, at: -1)

See stream_delete/3 for more information on deleting items.

Returns the transport pid of the socket.

Raises ArgumentError if the socket is not connected.

Examples

iex> transport_pid(socket)
#PID<0.107.0>

Returns the completed and in progress entries for the upload.

Examples

case uploaded_entries(socket, :photos) do
  {[_ | _] = completed, []} ->
    # all entries are completed

  {[], [_ | _] = in_progress} ->
    # all entries are still in progress
end