Work in progress

The content of this page was not yet updated for Godot 4.0 and may be outdated. If you know how to improve this page or you can confirm that it's up to date, feel free to open a pull request.

HTTP client class

HTTPClient provides low-level access to HTTP communication. For a higher-level interface, you may want to take a look at HTTPRequest first, which has a tutorial available here.


When exporting to Android, make sure to enable the INTERNET permission in the Android export preset before exporting the project or using one-click deploy. Otherwise, network communication of any kind will be blocked by Android.

Here's an example of using the HTTPClient class. It's just a script, so it can be run by executing:

c:\godot> godot -s

It will connect and fetch a website.

extends SceneTree

# HTTPClient demo
# This simple class can do HTTP requests; it will not block, but it needs to be polled.

func _init():
    var err = 0
    var http = # Create the Client.

    err = http.connect_to_host("", 80) # Connect to host/port.
    assert(err == OK) # Make sure connection is OK.

    # Wait until resolved and connected.
    while http.get_status() == HTTPClient.STATUS_CONNECTING or http.get_status() == HTTPClient.STATUS_RESOLVING:
        if not OS.has_feature("web"):
            yield(Engine.get_main_loop(), "idle_frame")

    assert(http.get_status() == HTTPClient.STATUS_CONNECTED) # Check if the connection was made successfully.

    # Some headers
    var headers = [
        "User-Agent: Pirulo/1.0 (Godot)",
        "Accept: */*"

    err = http.request(HTTPClient.METHOD_GET, "/ChangeLog-5.php", headers) # Request a page from the site (this one was chunked..)
    assert(err == OK) # Make sure all is OK.

    while http.get_status() == HTTPClient.STATUS_REQUESTING:
        # Keep polling for as long as the request is being processed.
        if OS.has_feature("web"):
            # Synchronous HTTP requests are not supported on the web,
            # so wait for the next main loop iteration.
            yield(Engine.get_main_loop(), "idle_frame")

    assert(http.get_status() == HTTPClient.STATUS_BODY or http.get_status() == HTTPClient.STATUS_CONNECTED) # Make sure request finished well.

    print("response? ", http.has_response()) # Site might not have a response.

    if http.has_response():
        # If there is a response...

        headers = http.get_response_headers_as_dictionary() # Get response headers.
        print("code: ", http.get_response_code()) # Show response code.
        print("**headers:\\n", headers) # Show headers.

        # Getting the HTTP Body

        if http.is_response_chunked():
            # Does it use chunks?
            print("Response is Chunked!")
            # Or just plain Content-Length
            var bl = http.get_response_body_length()
            print("Response Length: ", bl)

        # This method works for both anyway

        var rb = PackedByteArray() # Array that will hold the data.

        while http.get_status() == HTTPClient.STATUS_BODY:
            # While there is body left to be read
            # Get a chunk.
            var chunk = http.read_response_body_chunk()
            if chunk.size() == 0:
                if not OS.has_feature("web"):
                    # Got nothing, wait for buffers to fill a bit.
                    yield(Engine.get_main_loop(), "idle_frame")
                rb = rb + chunk # Append to read buffer.
        # Done!

        print("bytes got: ", rb.size())
        var text = rb.get_string_from_ascii()
        print("Text: ", text)