Source code for spotify.sink

from __future__ import unicode_literals

import sys

import spotify

__all__ = ["AlsaSink", "PortAudioSink"]


class Sink(object):
    def on(self):
        """Turn on the audio sink.

        This is done automatically when the sink is instantiated, so you'll
        only need to call this method if you ever call :meth:`off` and want to
        turn the sink back on.
        """
        assert self._session.num_listeners(spotify.SessionEvent.MUSIC_DELIVERY) == 0
        self._session.on(spotify.SessionEvent.MUSIC_DELIVERY, self._on_music_delivery)

    def off(self):
        """Turn off the audio sink.

        This disconnects the sink from the relevant session events.
        """
        self._session.off(spotify.SessionEvent.MUSIC_DELIVERY, self._on_music_delivery)
        assert self._session.num_listeners(spotify.SessionEvent.MUSIC_DELIVERY) == 0
        self._close()

    def _on_music_delivery(self, session, audio_format, frames, num_frames):
        # This method is called from an internal libspotify thread and must
        # not block in any way.
        raise NotImplementedError

    def _close(self):
        pass


[docs]class AlsaSink(Sink): """Audio sink for systems using ALSA, e.g. most Linux systems. This audio sink requires `pyalsaaudio <https://pypi.python.org/pypi/pyalsaaudio>`_. pyalsaaudio is probably packaged in your Linux distribution. For example, on Debian/Ubuntu you can install it from APT:: sudo apt-get install python-alsaaudio Or, if you want to install pyalsaaudio inside a virtualenv, install the ALSA development headers from APT, then pyalsaaudio:: sudo apt-get install libasound2-dev pip install pyalsaaudio The ``device`` keyword argument is passed on to :class:`alsaaudio.PCM`. Please refer to the pyalsaaudio documentation for details. Example:: >>> import spotify >>> session = spotify.Session() >>> audio = spotify.AlsaSink(session) >>> loop = spotify.EventLoop(session) >>> loop.start() # Login, etc... >>> track = session.get_track('spotify:track:3N2UhXZI4Gf64Ku3cCjz2g') >>> track.load() >>> session.player.load(track) >>> session.player.play() # Listen to music... """ def __init__(self, session, device="default"): self._session = session self._device_name = device import alsaaudio # Crash early if not available self._alsaaudio = alsaaudio self._device = None self.on() def _on_music_delivery(self, session, audio_format, frames, num_frames): assert audio_format.sample_type == spotify.SampleType.INT16_NATIVE_ENDIAN if self._device is None: if hasattr(self._alsaaudio, "pcms"): # pyalsaaudio >= 0.8 self._device = self._alsaaudio.PCM( mode=self._alsaaudio.PCM_NONBLOCK, device=self._device_name ) else: # pyalsaaudio == 0.7 self._device = self._alsaaudio.PCM( mode=self._alsaaudio.PCM_NONBLOCK, card=self._device_name ) if sys.byteorder == "little": self._device.setformat(self._alsaaudio.PCM_FORMAT_S16_LE) else: self._device.setformat(self._alsaaudio.PCM_FORMAT_S16_BE) self._device.setrate(audio_format.sample_rate) self._device.setchannels(audio_format.channels) self._device.setperiodsize(num_frames * audio_format.frame_size()) return self._device.write(frames) def _close(self): if self._device is not None: self._device.close() self._device = None
[docs]class PortAudioSink(Sink): """Audio sink for `PortAudio <http://www.portaudio.com/>`_. PortAudio is available for many platforms, including Linux, macOS, and Windows. This audio sink requires `PyAudio <https://pypi.python.org/pypi/pyaudio>`_. PyAudio is probably packaged in your Linux distribution. On Debian/Ubuntu you can install PyAudio from APT:: sudo apt-get install python-pyaudio Or, if you want to install PyAudio inside a virtualenv, install the PortAudio development headers from APT, then PyAudio:: sudo apt-get install portaudio19-dev pip install --allow-unverified=pyaudio pyaudio On macOS you can install PortAudio using Homebrew:: brew install portaudio pip install --allow-unverified=pyaudio pyaudio For an example of how to use this class, see the :class:`AlsaSink` example. Just replace ``AlsaSink`` with ``PortAudioSink``. """ def __init__(self, session): self._session = session import pyaudio # Crash early if not available self._pyaudio = pyaudio self._device = self._pyaudio.PyAudio() self._stream = None self.on() def _on_music_delivery(self, session, audio_format, frames, num_frames): assert audio_format.sample_type == spotify.SampleType.INT16_NATIVE_ENDIAN if self._stream is None: self._stream = self._device.open( format=self._pyaudio.paInt16, channels=audio_format.channels, rate=audio_format.sample_rate, output=True, ) # XXX write() is a blocking call. There are two non-blocking # alternatives: # 1) Only feed write() with the number of frames returned by # self._stream.get_write_available() on each call. This causes buffer # underruns every third or fourth write(). # 2) Let pyaudio call a callback function when it needs data, but then # we need to introduce a thread safe buffer here which is filled when # libspotify got data and drained when pyaudio needs data. self._stream.write(frames, num_frames=num_frames) return num_frames def _close(self): if self._stream is not None: self._stream.close() self._stream = None