Working on Android#

This page gives details on accessing Android APIs and managing other interactions on Android.

Storage paths#

If you want to store and retrieve data, you shouldn’t just save to the current directory, and not hardcode /sdcard/ or some other path either - it might differ per device.

Instead, the android module which you can add to your –requirements allows you to query the most commonly required paths:

from import app_storage_path
settings_path = app_storage_path()

from import primary_external_storage_path
primary_ext_storage = primary_external_storage_path()

from import secondary_external_storage_path
secondary_ext_storage = secondary_external_storage_path()

app_storage_path() gives you Android’s so-called “internal storage” which is specific to your app and cannot seen by others or the user. It compares best to the AppData directory on Windows.

primary_external_storage_path() returns Android’s so-called “primary external storage”, often found at /sdcard/ and potentially accessible to any other app. It compares best to the Documents directory on Windows. Requires Permission.WRITE_EXTERNAL_STORAGE to read and write to.

secondary_external_storage_path() returns Android’s so-called “secondary external storage”, often found at /storage/External_SD/. It compares best to an external disk plugged to a Desktop PC, and can after a device restart become inaccessible if removed. Requires Permission.WRITE_EXTERNAL_STORAGE to read and write to.


Even if secondary_external_storage_path returns a path the external sd card may still not be present. Only non-empty contents or a successful write indicate that it is.

Read more on all the different storage types and what to use them for in the Android documentation:

A note on permissions#

Only the internal storage is always accessible with no additional permissions. For both primary and secondary external storage, you need to obtain Permission.WRITE_EXTERNAL_STORAGE and the user may deny it. Also, if you get it, both forms of external storage may only allow your app to write to the common pre-existing folders like “Music”, “Documents”, and so on. (see the Android Docs linked above for details)

Runtime permissions#

With API level >= 21, you will need to request runtime permissions to access the SD card, the camera, and other things.

This can be done through the android module which is available per default unless you blacklist it. Use it in your app like this:

from android.permissions import request_permissions, Permission

The available permissions are listed here:

Other common tasks#

Dismissing the splash screen#

With the SDL2 bootstrap, the app’s splash screen may be visible longer than necessary (with your app already being loaded) due to a limitation with the way we check if the app has properly started. In this case, the splash screen overlaps the app gui for a short time.

To dismiss the loading screen explicitly in your code, use the android module:

from android import loadingscreen

You can call it e.g. using kivy.clock.Clock.schedule_once to run it in the first active frame of your app, or use the app build method.

Handling the back button#

Android phones always have a back button, which users expect to perform an appropriate in-app function. If you do not handle it, Kivy apps will actually shut down and appear to have crashed.

In SDL2 bootstraps, the back button appears as the escape key (keycode 27, codepoint 270). You can handle this key to perform actions when it is pressed.

For instance, in your App class in Kivy:

from kivy.core.window import Window

class YourApp(App):

   def build(self):
      return Widget() # your root widget here as normal

   def key_input(self, window, key, scancode, codepoint, modifier):
      if key == 27:
         return True  # override the default behaviour
      else:           # the key now does nothing
         return False

Pausing the App#

When the user leaves an App, it is automatically paused by Android, although it gets a few seconds to store data etc. if necessary. Once paused, there is no guarantee that your app will run again.

With Kivy, add an on_pause method to your App class, which returns True:

def on_pause(self):
    return True

With the webview bootstrap, pausing should work automatically.

Under SDL2, you can handle the appropriate events (see SDL_APP_WILLENTERBACKGROUND etc.).

Observing Activity result#

The default PythonActivity has a observer pattern for onActivityResult and onNewIntent.

android.activity.bind(eventname=callback, ...)#

This allows you to bind a callback to an Android event: - on_new_intent is the event associated to the onNewIntent java call - on_activity_result is the event associated to the onActivityResult java call


This method is not thread-safe. Call it in the mainthread of your app. (tips: use kivy.clock.mainthread decorator)

android.activity.unbind(eventname=callback, ...)#

Unregister a previously registered callback with bind().


# This example is a snippet from an NFC p2p app implemented with Kivy.

from android import activity

def on_new_intent(self, intent):
    if intent.getAction() != NfcAdapter.ACTION_NDEF_DISCOVERED:
    rawmsgs = intent.getParcelableArrayExtra(NfcAdapter.EXTRA_NDEF_MESSAGES)
    if not rawmsgs:
    for message in rawmsgs:
        message = cast(NdefMessage, message)
        payload = message.getRecords()[0].getPayload()
        print('payload: {}'.format(''.join(map(chr, payload))))

def nfc_enable(self):
    # ...

def nfc_disable(self):
    # ...

Activity lifecycle handling#

The Android Application class provides the ActivityLifecycleCallbacks interface where callbacks can be registered corresponding to activity lifecycle changes. These callbacks can be used to implement logic in the Python app when the activity changes lifecycle states.

Note that some of the callbacks are not useful in the Python app. For example, an onActivityCreated callback will never be run since the the activity’s onCreate callback will complete before the Python app is running. Similarly, saving instance state in an onActivitySaveInstanceState callback will not be helpful since the Python app doesn’t have access to the restored instance state.

android.activity.register_activity_lifecycle_callbacks(callbackname=callback, ...)#

This allows you to bind a callbacks to Activity lifecycle state changes. The callback names correspond to ActivityLifecycleCallbacks method names such as onActivityStarted. See the ActivityLifecycleCallbacks documentation for names and function signatures for the callbacks.


Unregister a ActivityLifecycleCallbacks instance previously registered with register_activity_lifecycle_callbacks().


from android.activity import register_activity_lifecycle_callbacks

def on_activity_stopped(activity):
    print('Activity is stopping')


Receiving Broadcast message#

Implementation of the android BroadcastReceiver. You can specify the callback that will receive the broadcast event, and actions or categories filters.

class android.broadcast.BroadcastReceiver#


The callback will be called in another thread than the main thread. In that thread, be careful not to access OpenGL or something like that.

__init__(callback, actions=None, categories=None)#
  • callback – function or method that will receive the event. Will receive the context and intent as argument.

  • actions – list of strings that represent an action.

  • categories – list of strings that represent a category.

For actions and categories, the string must be in lower case, without the prefix:

# In python: 'headset_plug'

Register the receiver with all the actions and categories, and start handling events.


Unregister the receiver with all the actions and categories, and stop handling events.


class TestApp(App):

    def build(self): = BroadcastReceiver(
            self.on_broadcast, actions=['headset_plug'])
        # ...

    def on_broadcast(self, context, intent):
        extras = intent.getExtras()
        headset_state = bool(extras.get('state'))
        if headset_state:
            print('The headset is plugged')
            print('The headset is unplugged')

    # Don't forget to stop and restart the receiver when the app is going
    # to pause / resume mode

    def on_pause(self):
        return True

    def on_resume(self):


Runnable is a wrapper around the Java Runnable class. This class can be used to schedule a call of a Python function into the PythonActivity thread.


from android.runnable import Runnable

def helloworld(arg):
    print 'Called from PythonActivity with arg:', arg


Or use our decorator:

from android.runnable import run_on_ui_thread

def helloworld(arg):
    print 'Called from PythonActivity with arg:', arg


This can be used to prevent errors like:

  • W/System.err( 9514): java.lang.RuntimeException: Can’t create handler inside thread that has not called Looper.prepare()

  • NullPointerException in ActivityThread.currentActivityThread()


Because the python function is called from the PythonActivity thread, you need to be careful about your own calls.

Advanced Android API use#

android for Android API access#

As mentioned above, the android Python module provides a simple wrapper around many native Android APIS, and it is included by default unless you blacklist it.

The available functionality of this module is not separately documented. You can read the source on Github.

Also please note you can replicate most functionality without it using pyjnius. (see below)

Plyer - a more comprehensive API wrapper#

Plyer provides a more thorough wrapper than android for a much larger area of platform-specific APIs, supporting not only Android but also iOS and desktop operating systems. (Though plyer is a work in progress and not all platforms support all Plyer calls yet)

Plyer does not support all APIs yet, but you can always use Pyjnius to call anything that is currently missing.

You can include Plyer in your APKs by adding the Plyer recipe to your build requirements, e.g. --requirements=plyer.

You should check the Plyer documentation for details of all supported facades (platform APIs), but as an example the following is how you would achieve vibration as described in the Pyjnius section above:

from plyer.vibrator import vibrate
vibrate(10)  # in Plyer, the argument is in seconds

This is obviously much less verbose than with Pyjnius!

Pyjnius - raw lowlevel API access#

Pyjnius lets you call the Android API directly from Python Pyjnius is works by dynamically wrapping Java classes, so you don’t have to wait for any particular feature to be pre-supported.

This is particularly useful when android and plyer don’t already provide a convenient access to the API, or you need more control.

You can include Pyjnius in your APKs by adding pyjnius to your build requirements, e.g. --requirements=flask,pyjnius. It is automatically included in any APK containing Kivy, in which case you don’t need to specify it manually.

The basic mechanism of Pyjnius is the autoclass command, which wraps a Java class. For instance, here is the code to vibrate your device:

from jnius import autoclass

# We need a reference to the Java activity running the current
# application, this reference is stored automatically by
# Kivy's PythonActivity bootstrap

# This one works with SDL2
PythonActivity = autoclass('')

activity = PythonActivity.mActivity

Context = autoclass('android.content.Context')
vibrator = activity.getSystemService(Context.VIBRATOR_SERVICE)

vibrator.vibrate(10000)  # the argument is in milliseconds

Things to note here are:

  • The class that must be wrapped depends on the bootstrap. This is because Pyjnius is using the bootstrap’s java source code to get a reference to the current activity, which the bootstraps store in the mActivity static variable. This difference isn’t always important, but it’s important to know about.

  • The code closely follows the Java API - this is exactly the same set of function calls that you’d use to achieve the same thing from Java code.

  • This is quite verbose - it’s a lot of lines to achieve a simple vibration!

These emphasise both the advantages and disadvantage of Pyjnius; you can achieve just about any API call with it (though the syntax is sometimes a little more involved, particularly if making Java classes from Python code), but it’s not Pythonic and it’s not short. These are problems that Plyer, explained below, attempts to address.

You can check the Pyjnius documentation for further details.