Move RTC init helpers to bindings layer and clean up structs (#1086)

* converted from type() factory to proper ctypes.Structure subclasses

* change class to method

* bring back rtcSetLocalDescription

* rename rtc_bindings.py to datachannel.py

* rename rtc_library to datachannel_library and rearrange methods

* cleanup
This commit is contained in:
Harisreedhar
2026-05-08 13:11:46 +05:30
committed by henryruhs
parent 99fa2875a9
commit 949d9cd276
5 changed files with 187 additions and 186 deletions
+141
View File
@@ -0,0 +1,141 @@
import ctypes
from functools import lru_cache
from typing import Dict, Optional
from facefusion.common_helper import is_linux, is_macos, is_windows
from facefusion.filesystem import resolve_relative_path
from facefusion.types import DownloadSet
LOG_CB_TYPE = ctypes.CFUNCTYPE(None, ctypes.c_int, ctypes.c_char_p)
@lru_cache
def create_static_download_set() -> Dict[str, DownloadSet]: # TODO: replace once conda package is in place
binary_name = resolve_binary_file()
return\
{
'hashes':
{
'datachannel':
{
'url': 'https://huggingface.co/bluefoxcreation/libdatachannel/resolve/main/linux-x64-openssl-h264-vp8-av1-opus-libdatachannel-0.24.1.so.hash', # TODO: use url with dynamic binary_name
'path': resolve_relative_path('../.assets/binaries/' + binary_name + '.hash')
}
},
'sources':
{
'datachannel':
{
'url': 'https://huggingface.co/bluefoxcreation/libdatachannel/resolve/main/linux-x64-openssl-h264-vp8-av1-opus-libdatachannel-0.24.1.so', # TODO: use url with dynamic binary_name
'path': resolve_relative_path('../.assets/binaries/' + binary_name)
}
}
}
def resolve_binary_file() -> Optional[str]:
if is_linux():
return 'linux-x64-openssl-h264-vp8-av1-opus-libdatachannel-0.24.1.so'
if is_macos():
return 'macos-universal-openssl-h264-vp8-av1-opus-libdatachannel-0.24.1.dylib'
if is_windows():
return 'windows-x64-openssl-h264-vp8-av1-opus-datachannel-0.24.1.dll'
return None
def create_rtc_configuration() -> ctypes.Structure:
return type('RTC_CONFIGURATION', (ctypes.Structure,),
{
'_fields_':
[
('iceServers', ctypes.POINTER(ctypes.c_char_p)),
('iceServersCount', ctypes.c_int),
('proxyServer', ctypes.c_char_p),
('bindAddress', ctypes.c_char_p),
('certificateType', ctypes.c_int),
('iceTransportPolicy', ctypes.c_int),
('enableIceTcp', ctypes.c_bool),
('enableIceUdpMux', ctypes.c_bool),
('disableAutoNegotiation', ctypes.c_bool),
('forceMediaTransport', ctypes.c_bool),
('portRangeBegin', ctypes.c_ushort),
('portRangeEnd', ctypes.c_ushort),
('mtu', ctypes.c_int),
('maxMessageSize', ctypes.c_int)
]
})()
def create_rtc_packetizer_init() -> ctypes.Structure:
return type('RTC_PACKETIZER_INIT', (ctypes.Structure,),
{
'_fields_':
[
('ssrc', ctypes.c_uint32),
('cname', ctypes.c_char_p),
('payloadType', ctypes.c_uint8),
('clockRate', ctypes.c_uint32),
('sequenceNumber', ctypes.c_uint16),
('timestamp', ctypes.c_uint32),
('maxFragmentSize', ctypes.c_uint16)
]
})()
@lru_cache
def create_static_datachannel_library() -> Optional[ctypes.CDLL]:
binary_path = create_static_download_set().get('sources').get('datachannel').get('path')
if binary_path:
datachannel_library = ctypes.CDLL(binary_path)
return init_ctypes(datachannel_library)
return None
def init_ctypes(datachannel_library : ctypes.CDLL) -> ctypes.CDLL:
datachannel_library.rtcInitLogger.argtypes = [ ctypes.c_int, LOG_CB_TYPE ]
datachannel_library.rtcInitLogger.restype = None
datachannel_library.rtcInitLogger(4, LOG_CB_TYPE(0))
datachannel_library.rtcCreatePeerConnection.restype = ctypes.c_int
datachannel_library.rtcDeletePeerConnection.argtypes = [ ctypes.c_int ]
datachannel_library.rtcDeletePeerConnection.restype = ctypes.c_int
datachannel_library.rtcSetLocalDescription.argtypes = [ ctypes.c_int, ctypes.c_char_p ]
datachannel_library.rtcSetLocalDescription.restype = ctypes.c_int
datachannel_library.rtcSetRemoteDescription.argtypes = [ ctypes.c_int, ctypes.c_char_p, ctypes.c_char_p ]
datachannel_library.rtcSetRemoteDescription.restype = ctypes.c_int
datachannel_library.rtcAddTrack.argtypes = [ ctypes.c_int, ctypes.c_char_p ]
datachannel_library.rtcAddTrack.restype = ctypes.c_int
datachannel_library.rtcSendMessage.argtypes = [ ctypes.c_int, ctypes.c_void_p, ctypes.c_int ]
datachannel_library.rtcSendMessage.restype = ctypes.c_int
datachannel_library.rtcSetVP8Packetizer.restype = ctypes.c_int
datachannel_library.rtcChainRtcpSrReporter.argtypes = [ ctypes.c_int ]
datachannel_library.rtcChainRtcpSrReporter.restype = ctypes.c_int
datachannel_library.rtcSetTrackRtpTimestamp.argtypes = [ ctypes.c_int, ctypes.c_uint32 ]
datachannel_library.rtcSetTrackRtpTimestamp.restype = ctypes.c_int
datachannel_library.rtcIsOpen.argtypes = [ ctypes.c_int ]
datachannel_library.rtcIsOpen.restype = ctypes.c_bool
datachannel_library.rtcChainRtcpNackResponder.argtypes = [ ctypes.c_int, ctypes.c_uint ]
datachannel_library.rtcChainRtcpNackResponder.restype = ctypes.c_int
datachannel_library.rtcGetLocalDescription.argtypes = [ ctypes.c_int, ctypes.c_char_p, ctypes.c_int ]
datachannel_library.rtcGetLocalDescription.restype = ctypes.c_int
datachannel_library.rtcSetLocalDescription.argtypes = [ ctypes.c_int, ctypes.c_char_p ]
datachannel_library.rtcSetLocalDescription.restype = ctypes.c_int
datachannel_library.rtcSetOpusPacketizer.restype = ctypes.c_int
return datachannel_library
+31 -80
View File
@@ -1,48 +1,10 @@
import ctypes
import time
from functools import lru_cache
from typing import Dict, List, Optional
from typing import List, Optional
from facefusion.common_helper import is_linux, is_macos, is_windows
from facefusion.datachannel import create_rtc_configuration, create_rtc_packetizer_init, create_static_datachannel_library, create_static_download_set
from facefusion.download import conditional_download_hashes, conditional_download_sources
from facefusion.filesystem import resolve_relative_path
from facefusion.rtc_bindings import RTC_CONFIGURATION, RTC_PACKETIZER_INIT, init_ctypes
from facefusion.types import DownloadSet, MediaDirection, PeerConnection, RtcAudioTrack, RtcPeer, RtcVideoTrack, SdpAnswer, SdpOffer
def resolve_binary_file() -> Optional[str]:
if is_linux():
return 'linux-x64-openssl-h264-vp8-av1-opus-libdatachannel-0.24.1.so'
if is_macos():
return 'macos-universal-openssl-h264-vp8-av1-opus-libdatachannel-0.24.1.dylib'
if is_windows():
return 'windows-x64-openssl-h264-vp8-av1-opus-datachannel-0.24.1.dll'
return None
@lru_cache
def create_static_download_set() -> Dict[str, DownloadSet]: # TODO: replace once conda package is in place
binary_name = resolve_binary_file()
return\
{
'hashes':
{
'datachannel':
{
'url': 'https://huggingface.co/bluefoxcreation/libdatachannel/resolve/main/linux-x64-openssl-h264-vp8-av1-opus-libdatachannel-0.24.1.so.hash', # TODO: use url with dynamic binary_name
'path': resolve_relative_path('../.assets/binaries/' + binary_name + '.hash')
}
},
'sources':
{
'datachannel':
{
'url': 'https://huggingface.co/bluefoxcreation/libdatachannel/resolve/main/linux-x64-openssl-h264-vp8-av1-opus-libdatachannel-0.24.1.so', # TODO: use url with dynamic binary_name
'path': resolve_relative_path('../.assets/binaries/' + binary_name)
}
}
}
from facefusion.types import MediaDirection, PeerConnection, RtcAudioTrack, RtcPeer, RtcVideoTrack, SdpAnswer, SdpOffer
def pre_check() -> bool:
@@ -53,17 +15,6 @@ def pre_check() -> bool:
return conditional_download_sources(download_set.get('sources'))
@lru_cache
def create_static_rtc_library() -> Optional[ctypes.CDLL]:
binary_path = create_static_download_set().get('sources').get('datachannel').get('path')
if binary_path:
rtc_library = ctypes.CDLL(binary_path)
return init_ctypes(rtc_library)
return None
def create_peer_connection(
ice_servers : Optional[ctypes.Array[ctypes.c_char_p]] = None,
ice_servers_count : int = 0, proxy_server : Optional[bytes] = None,
@@ -78,8 +29,8 @@ def create_peer_connection(
max_packet_size : int = 0,
max_message_size : int = 0) -> PeerConnection:
rtc_library = create_static_rtc_library()
rtc_configuration = RTC_CONFIGURATION()
datachannel_library = create_static_datachannel_library()
rtc_configuration = create_rtc_configuration()
rtc_configuration.iceServers = ice_servers
rtc_configuration.iceServersCount = ice_servers_count
@@ -96,7 +47,7 @@ def create_peer_connection(
rtc_configuration.mtu = max_packet_size
rtc_configuration.maxMessageSize = max_message_size
return rtc_library.rtcCreatePeerConnection(ctypes.byref(rtc_configuration))
return datachannel_library.rtcCreatePeerConnection(ctypes.byref(rtc_configuration))
def build_media_description(media_type : str, payload_type : int, rtp_codec : str, media_direction : MediaDirection, media_id : int) -> bytes:
@@ -112,64 +63,64 @@ def build_media_description(media_type : str, payload_type : int, rtp_codec : st
def add_audio_track(peer_connection : PeerConnection, media_direction : MediaDirection) -> RtcAudioTrack:
rtc_library = create_static_rtc_library()
datachannel_library = create_static_datachannel_library()
media_description = build_media_description('audio', 111, 'opus/48000/2', media_direction, 1)
audio_track = rtc_library.rtcAddTrack(peer_connection, media_description)
audio_track = datachannel_library.rtcAddTrack(peer_connection, media_description)
audio_packetizer = RTC_PACKETIZER_INIT()
audio_packetizer = create_rtc_packetizer_init()
audio_packetizer.ssrc = 43
audio_packetizer.cname = b'audio'
audio_packetizer.payloadType = 111
audio_packetizer.clockRate = 48000
rtc_library.rtcSetOpusPacketizer(audio_track, ctypes.byref(audio_packetizer))
rtc_library.rtcChainRtcpSrReporter(audio_track)
datachannel_library.rtcSetOpusPacketizer(audio_track, ctypes.byref(audio_packetizer))
datachannel_library.rtcChainRtcpSrReporter(audio_track)
return audio_track
def add_video_track(peer_connection : PeerConnection, media_direction : MediaDirection) -> RtcVideoTrack:
rtc_library = create_static_rtc_library()
datachannel_library = create_static_datachannel_library()
media_description = build_media_description('video', 96, 'VP8/90000', media_direction, 0)
video_track = rtc_library.rtcAddTrack(peer_connection, media_description)
video_track = datachannel_library.rtcAddTrack(peer_connection, media_description)
video_packetizer = RTC_PACKETIZER_INIT()
video_packetizer = create_rtc_packetizer_init()
video_packetizer.ssrc = 42
video_packetizer.cname = b'video'
video_packetizer.payloadType = 96
video_packetizer.clockRate = 90000
video_packetizer.maxFragmentSize = 1200
rtc_library.rtcSetVP8Packetizer(video_track, ctypes.byref(video_packetizer))
rtc_library.rtcChainRtcpSrReporter(video_track)
rtc_library.rtcChainRtcpNackResponder(video_track, 512)
datachannel_library.rtcSetVP8Packetizer(video_track, ctypes.byref(video_packetizer))
datachannel_library.rtcChainRtcpSrReporter(video_track)
datachannel_library.rtcChainRtcpNackResponder(video_track, 512)
return video_track
def create_sdp(peer_connection : PeerConnection) -> Optional[SdpOffer]:
rtc_library = create_static_rtc_library()
rtc_library.rtcSetLocalDescription(peer_connection, b'offer')
datachannel_library = create_static_datachannel_library()
datachannel_library.rtcSetLocalDescription(peer_connection, b'offer')
buffer_size = 16384
buffer_string = ctypes.create_string_buffer(buffer_size)
if rtc_library.rtcGetLocalDescription(peer_connection, buffer_string, buffer_size) > 0:
if datachannel_library.rtcGetLocalDescription(peer_connection, buffer_string, buffer_size) > 0:
return buffer_string.value.decode()
return None
def negotiate_sdp(peer_connection : PeerConnection, sdp_offer : SdpOffer) -> Optional[SdpAnswer]:
rtc_library = create_static_rtc_library()
rtc_library.rtcSetRemoteDescription(peer_connection, sdp_offer.encode(), b'offer')
datachannel_library = create_static_datachannel_library()
datachannel_library.rtcSetRemoteDescription(peer_connection, sdp_offer.encode(), b'offer')
buffer_size = 16384
buffer_string = ctypes.create_string_buffer(buffer_size)
wait_limit = time.monotonic() + 5
while time.monotonic() < wait_limit:
if rtc_library.rtcGetLocalDescription(peer_connection, buffer_string, buffer_size) > 0:
if datachannel_library.rtcGetLocalDescription(peer_connection, buffer_string, buffer_size) > 0:
return buffer_string.value.decode()
time.sleep(0.05)
@@ -195,7 +146,7 @@ def handle_whep_offer(peers : List[RtcPeer], sdp_offer : SdpOffer) -> Optional[S
def send_to_peers(peers : List[RtcPeer], data : bytes) -> None:
rtc_library = create_static_rtc_library()
datachannel_library = create_static_datachannel_library()
if peers:
timestamp = int(time.monotonic() * 90000) & 0xFFFFFFFF
@@ -205,32 +156,32 @@ def send_to_peers(peers : List[RtcPeer], data : bytes) -> None:
for rtc_peer in peers:
video_track_id = rtc_peer.get('video_track')
if video_track_id and rtc_library.rtcIsOpen(video_track_id):
rtc_library.rtcSetTrackRtpTimestamp(video_track_id, timestamp)
rtc_library.rtcSendMessage(video_track_id, data_buffer, data_total)
if video_track_id and datachannel_library.rtcIsOpen(video_track_id):
datachannel_library.rtcSetTrackRtpTimestamp(video_track_id, timestamp)
datachannel_library.rtcSendMessage(video_track_id, data_buffer, data_total)
return None
def delete_peers(peers : List[RtcPeer]) -> None:
rtc_library = create_static_rtc_library()
datachannel_library = create_static_datachannel_library()
for rtc_peer in peers:
peer_connection_id = rtc_peer.get('peer_connection')
if peer_connection_id:
rtc_library.rtcDeletePeerConnection(peer_connection_id)
datachannel_library.rtcDeletePeerConnection(peer_connection_id)
peers.clear()
def is_peer_connected(peers : List[RtcPeer]) -> bool:
rtc_library = create_static_rtc_library()
datachannel_library = create_static_datachannel_library()
for rtc_peer in peers:
video_track_id = rtc_peer.get('video_track')
if video_track_id and rtc_library.rtcIsOpen(video_track_id):
if video_track_id and datachannel_library.rtcIsOpen(video_track_id):
return True
return False
-91
View File
@@ -1,91 +0,0 @@
import ctypes
RTC_CONFIGURATION = type('RtcConfiguration', (ctypes.Structure,),
{
'_fields_':
[
('iceServers', ctypes.POINTER(ctypes.c_char_p)),
('iceServersCount', ctypes.c_int),
('proxyServer', ctypes.c_char_p),
('bindAddress', ctypes.c_char_p),
('certificateType', ctypes.c_int),
('iceTransportPolicy', ctypes.c_int),
('enableIceTcp', ctypes.c_bool),
('enableIceUdpMux', ctypes.c_bool),
('disableAutoNegotiation', ctypes.c_bool),
('forceMediaTransport', ctypes.c_bool),
('portRangeBegin', ctypes.c_ushort),
('portRangeEnd', ctypes.c_ushort),
('mtu', ctypes.c_int),
('maxMessageSize', ctypes.c_int)
]
})
RTC_PACKETIZER_INIT = type('RtcPacketizerInit', (ctypes.Structure,),
{
'_fields_':
[
('ssrc', ctypes.c_uint32),
('cname', ctypes.c_char_p),
('payloadType', ctypes.c_uint8),
('clockRate', ctypes.c_uint32),
('sequenceNumber', ctypes.c_uint16),
('timestamp', ctypes.c_uint32),
('maxFragmentSize', ctypes.c_uint16),
('nalSeparator', ctypes.c_int),
('obuPacketization', ctypes.c_int),
('playoutDelayId', ctypes.c_uint8),
('playoutDelayMin', ctypes.c_uint16),
('playoutDelayMax', ctypes.c_uint16)
]
})
LOG_CB_TYPE = ctypes.CFUNCTYPE(None, ctypes.c_int, ctypes.c_char_p)
def init_ctypes(rtc_library : ctypes.CDLL) -> ctypes.CDLL:
rtc_library.rtcInitLogger.argtypes = [ ctypes.c_int, LOG_CB_TYPE ]
rtc_library.rtcInitLogger.restype = None
rtc_library.rtcInitLogger(4, LOG_CB_TYPE(0))
rtc_library.rtcCreatePeerConnection.argtypes = [ ctypes.POINTER(RTC_CONFIGURATION) ]
rtc_library.rtcCreatePeerConnection.restype = ctypes.c_int
rtc_library.rtcDeletePeerConnection.argtypes = [ ctypes.c_int ]
rtc_library.rtcDeletePeerConnection.restype = ctypes.c_int
rtc_library.rtcSetLocalDescription.argtypes = [ ctypes.c_int, ctypes.c_char_p ]
rtc_library.rtcSetLocalDescription.restype = ctypes.c_int
rtc_library.rtcSetRemoteDescription.argtypes = [ ctypes.c_int, ctypes.c_char_p, ctypes.c_char_p ]
rtc_library.rtcSetRemoteDescription.restype = ctypes.c_int
rtc_library.rtcAddTrack.argtypes = [ ctypes.c_int, ctypes.c_char_p ]
rtc_library.rtcAddTrack.restype = ctypes.c_int
rtc_library.rtcSendMessage.argtypes = [ ctypes.c_int, ctypes.c_void_p, ctypes.c_int ]
rtc_library.rtcSendMessage.restype = ctypes.c_int
rtc_library.rtcSetVP8Packetizer.argtypes = [ ctypes.c_int, ctypes.POINTER(RTC_PACKETIZER_INIT) ]
rtc_library.rtcSetVP8Packetizer.restype = ctypes.c_int
rtc_library.rtcChainRtcpSrReporter.argtypes = [ ctypes.c_int ]
rtc_library.rtcChainRtcpSrReporter.restype = ctypes.c_int
rtc_library.rtcSetTrackRtpTimestamp.argtypes = [ ctypes.c_int, ctypes.c_uint32 ]
rtc_library.rtcSetTrackRtpTimestamp.restype = ctypes.c_int
rtc_library.rtcIsOpen.argtypes = [ ctypes.c_int ]
rtc_library.rtcIsOpen.restype = ctypes.c_bool
rtc_library.rtcChainRtcpNackResponder.argtypes = [ ctypes.c_int, ctypes.c_uint ]
rtc_library.rtcChainRtcpNackResponder.restype = ctypes.c_int
rtc_library.rtcGetLocalDescription.argtypes = [ ctypes.c_int, ctypes.c_char_p, ctypes.c_int ]
rtc_library.rtcGetLocalDescription.restype = ctypes.c_int
rtc_library.rtcSetLocalDescription.argtypes = [ ctypes.c_int, ctypes.c_char_p ]
rtc_library.rtcSetLocalDescription.restype = ctypes.c_int
rtc_library.rtcSetOpusPacketizer.argtypes = [ ctypes.c_int, ctypes.POINTER(RTC_PACKETIZER_INIT) ]
rtc_library.rtcSetOpusPacketizer.restype = ctypes.c_int
return rtc_library
+7 -7
View File
@@ -10,26 +10,26 @@ from facefusion.types import SdpOffer
def create_sdp_offer() -> Optional[SdpOffer]:
rtc_library = rtc.create_static_rtc_library()
datachannel_library = rtc.create_static_datachannel_library()
peer_connection = rtc.create_peer_connection(disable_auto_negotiation = True)
rtc_library.rtcAddTrack(peer_connection, rtc.build_media_description('video', 96, 'VP8/90000', 'recvonly', 0))
rtc_library.rtcAddTrack(peer_connection, rtc.build_media_description('audio', 111, 'opus/48000/2', 'recvonly', 1))
rtc_library.rtcSetLocalDescription(peer_connection, b'offer')
datachannel_library.rtcAddTrack(peer_connection, rtc.build_media_description('video', 96, 'VP8/90000', 'recvonly', 0))
datachannel_library.rtcAddTrack(peer_connection, rtc.build_media_description('audio', 111, 'opus/48000/2', 'recvonly', 1))
datachannel_library.rtcSetLocalDescription(peer_connection, b'offer')
buffer_size = 16384
buffer_string = ctypes.create_string_buffer(buffer_size)
wait_limit = time.monotonic() + 5
while time.monotonic() < wait_limit:
if rtc_library.rtcGetLocalDescription(peer_connection, buffer_string, buffer_size) > 0:
if datachannel_library.rtcGetLocalDescription(peer_connection, buffer_string, buffer_size) > 0:
sdp = buffer_string.value.decode()
rtc_library.rtcDeletePeerConnection(peer_connection)
datachannel_library.rtcDeletePeerConnection(peer_connection)
return sdp
time.sleep(0.05)
rtc_library.rtcDeletePeerConnection(peer_connection)
datachannel_library.rtcDeletePeerConnection(peer_connection)
return None
+8 -8
View File
@@ -15,14 +15,14 @@ def test_build_media_description() -> None:
def test_create_peer_connection() -> None:
peer_connection = rtc.create_peer_connection()
rtc_library = rtc.create_static_rtc_library()
datachannel_library = rtc.create_static_datachannel_library()
assert peer_connection > 0
assert rtc_library.rtcDeletePeerConnection(peer_connection) == 0
assert datachannel_library.rtcDeletePeerConnection(peer_connection) == 0
def test_add_audio_track() -> None:
rtc_library = rtc.create_static_rtc_library()
datachannel_library = rtc.create_static_datachannel_library()
sender_connection = rtc.create_peer_connection()
sender_audio_track = rtc.add_audio_track(sender_connection, 'sendonly')
@@ -40,12 +40,12 @@ def test_add_audio_track() -> None:
assert 'opus/48000/2' in sdp_offer
assert 'opus/48000/2' in sdp_answer
assert rtc_library.rtcDeletePeerConnection(sender_connection) == 0
assert rtc_library.rtcDeletePeerConnection(receiver_connection) == 0
assert datachannel_library.rtcDeletePeerConnection(sender_connection) == 0
assert datachannel_library.rtcDeletePeerConnection(receiver_connection) == 0
def test_add_video_track() -> None:
rtc_library = rtc.create_static_rtc_library()
datachannel_library = rtc.create_static_datachannel_library()
sender_connection = rtc.create_peer_connection()
sender_video_track = rtc.add_video_track(sender_connection, 'sendonly')
@@ -63,5 +63,5 @@ def test_add_video_track() -> None:
assert 'VP8/90000' in sdp_offer
assert 'VP8/90000' in sdp_answer
assert rtc_library.rtcDeletePeerConnection(sender_connection) == 0
assert rtc_library.rtcDeletePeerConnection(receiver_connection) == 0
assert datachannel_library.rtcDeletePeerConnection(sender_connection) == 0
assert datachannel_library.rtcDeletePeerConnection(receiver_connection) == 0