libdatachannel-py 2025.1.0.dev8__tar.gz → 2025.1.0.dev9__tar.gz

{libdatachannel_py-2025.1.0.dev8/src/libdatachannel_py.egg-info → libdatachannel_py-2025.1.0.dev9}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: libdatachannel-py
-Version: 2025.1.0.dev8
+Version: 2025.1.0.dev9
 Summary: libdatachannel-py is a Python binding for libdatachannel.
 Home-page: https://github.com/shiguredo/libdatachannel-py
 License-Expression: Apache-2.0
@@ -39,6 +39,7 @@ Please read <https://github.com/shiguredo/oss/blob/master/README.en.md> before u
 
 - [nanobind](https://github.com/wjakob/nanobind) を利用しています
 - [Opus](https://opus-codec.org/) を利用した Opus ソフトウェアエンコード/デコードに対応しています
+- [libaom](https://aomedia.googlesource.com/aom/) を利用した AV1 ソフトウェアエンコード/デコードに対応しています
 - [OpenH264](https://www.openh264.org/) を利用した H.264 ソフトウェアエンコード/デコードに対応しています
 - [Apple Video Toolbox](https://developer.apple.com/documentation/videotoolbox) を利用した H.264/H.265 ハードウェアエンコード/デコードに対応しています
 

{libdatachannel_py-2025.1.0.dev8 → libdatachannel_py-2025.1.0.dev9}/README.md

@@ -23,6 +23,7 @@ Please read <https://github.com/shiguredo/oss/blob/master/README.en.md> before u
 
 - [nanobind](https://github.com/wjakob/nanobind) を利用しています
 - [Opus](https://opus-codec.org/) を利用した Opus ソフトウェアエンコード/デコードに対応しています
+- [libaom](https://aomedia.googlesource.com/aom/) を利用した AV1 ソフトウェアエンコード/デコードに対応しています
 - [OpenH264](https://www.openh264.org/) を利用した H.264 ソフトウェアエンコード/デコードに対応しています
 - [Apple Video Toolbox](https://developer.apple.com/documentation/videotoolbox) を利用した H.264/H.265 ハードウェアエンコード/デコードに対応しています
 

libdatachannel_py-2025.1.0.dev9/VERSION

@@ -0,0 +1 @@
+2025.1.0.dev9

{libdatachannel_py-2025.1.0.dev8 → libdatachannel_py-2025.1.0.dev9}/pyproject.toml

@@ -18,19 +18,24 @@ Source = "https://github.com/shiguredo/libdatachannel-py"
 Discord = "https://discord.gg/shiguredo"
 
 [build-system]
-requires = ["setuptools>=80.8", "wheel>=0.45.1"]
+requires = ["setuptools>=80.9", "wheel>=0.45.1"]
 build-backend = "setuptools.build_meta"
 
 [dependency-groups]
 dev = [
     "nanobind>=2.7",
     "pytest>=8",
-    "setuptools>=80.8",
+    "setuptools>=80.9",
     "build>=1.2.2.post1",
     "wheel>=0.45.1",
     "numpy>=2.2",
+    "ty",
+    "hypothesis>=6.132.0",
 ]
 
+[tool.uv.workspace]
+members = ["examples"]
+
 [tool.ruff]
 target-version = "py313"
 line-length = 100

{libdatachannel_py-2025.1.0.dev8 → libdatachannel_py-2025.1.0.dev9/src/libdatachannel_py.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: libdatachannel-py
-Version: 2025.1.0.dev8
+Version: 2025.1.0.dev9
 Summary: libdatachannel-py is a Python binding for libdatachannel.
 Home-page: https://github.com/shiguredo/libdatachannel-py
 License-Expression: Apache-2.0
@@ -39,6 +39,7 @@ Please read <https://github.com/shiguredo/oss/blob/master/README.en.md> before u
 
 - [nanobind](https://github.com/wjakob/nanobind) を利用しています
 - [Opus](https://opus-codec.org/) を利用した Opus ソフトウェアエンコード/デコードに対応しています
+- [libaom](https://aomedia.googlesource.com/aom/) を利用した AV1 ソフトウェアエンコード/デコードに対応しています
 - [OpenH264](https://www.openh264.org/) を利用した H.264 ソフトウェアエンコード/デコードに対応しています
 - [Apple Video Toolbox](https://developer.apple.com/documentation/videotoolbox) を利用した H.264/H.265 ハードウェアエンコード/デコードに対応しています
 

{libdatachannel_py-2025.1.0.dev8 → libdatachannel_py-2025.1.0.dev9}/src/libdatachannel_py.egg-info/SOURCES.txt

@@ -27,6 +27,7 @@ tests/test_message.py
 tests/test_nalunit.py
 tests/test_packetizationconfig.py
 tests/test_peerconnection.py
+tests/test_prop_audio_codec.py
 tests/test_reliability.py
 tests/test_rtppacketizer.py
 tests/test_video_codec.py

{libdatachannel_py-2025.1.0.dev8 → libdatachannel_py-2025.1.0.dev9}/tests/test_audio_codec.py

@@ -4,9 +4,11 @@ import numpy as np
 
 from libdatachannel.codec import (
     AudioCodecType,
+    AudioDecoder,
     AudioEncoder,
     AudioFrame,
     EncodedAudio,
+    create_opus_audio_decoder,
     create_opus_audio_encoder,
 )
 
@@ -34,14 +36,14 @@ def test_audio_encoder_encode():
     frame = AudioFrame()
     frame.sample_rate = 48000
     frame.timestamp = timedelta(milliseconds=1000)
-    # 48000 Hz * 20 ms = 960 samples
+    # 48000 Hz * 20 ms = 960 サンプル
     frame.pcm = np.zeros((960, 2), dtype=np.float32)
 
     encoder.encode(frame)
 
     assert callback_timestamps == [timedelta(milliseconds=1000)]
 
-    # 10ms → 20ms でエンコード
+    # 10ms -> 20ms でエンコード
     frame = AudioFrame()
     frame.sample_rate = 48000
     frame.timestamp = timedelta(milliseconds=1000)
@@ -70,3 +72,55 @@ def test_audio_encoder_encode():
     ]
 
     encoder.release()
+
+
+def test_audio_encoder_decoder():
+    """エンコード -> デコードのテスト"""
+    encoder = create_opus_audio_encoder()
+    encoder_settings = AudioEncoder.Settings()
+    encoder_settings.codec_type = AudioCodecType.OPUS
+    encoder_settings.sample_rate = 48000
+    encoder_settings.channels = 2
+    encoder_settings.bitrate = 64000
+    encoder_settings.frame_duration_ms = 20
+    assert encoder.init(encoder_settings) is True
+
+    decoder = create_opus_audio_decoder()
+    decoder_settings = AudioDecoder.Settings()
+    decoder_settings.codec_type = AudioCodecType.OPUS
+    decoder_settings.sample_rate = 48000
+    decoder_settings.channels = 2
+    assert decoder.init(decoder_settings) is True
+
+    encoded_data = []
+    decoded_frames = []
+
+    def on_encoded(encoded: EncodedAudio):
+        encoded_data.append(encoded)
+
+    def on_decoded(frame: AudioFrame):
+        decoded_frames.append(frame)
+
+    encoder.set_on_encode(on_encoded)
+    decoder.set_on_decode(on_decoded)
+
+    # 無音データでテスト
+    frame = AudioFrame()
+    frame.sample_rate = 48000
+    frame.timestamp = timedelta(milliseconds=1000)
+    frame.pcm = np.zeros((960, 2), dtype=np.float32)
+
+    encoder.encode(frame)
+    assert len(encoded_data) == 1
+
+    decoder.decode(encoded_data[0])
+    assert len(decoded_frames) == 1
+
+    decoded_frame = decoded_frames[0]
+    assert decoded_frame.sample_rate == 48000
+    assert decoded_frame.channels() == 2
+    assert decoded_frame.samples() == 960
+    assert decoded_frame.timestamp == timedelta(milliseconds=1000)
+
+    encoder.release()
+    decoder.release()

libdatachannel_py-2025.1.0.dev9/tests/test_prop_audio_codec.py

@@ -0,0 +1,97 @@
+from datetime import timedelta
+
+import numpy as np
+from hypothesis import given, settings
+from hypothesis import strategies as st
+
+from libdatachannel.codec import (
+    AudioCodecType,
+    AudioDecoder,
+    AudioEncoder,
+    AudioFrame,
+    EncodedAudio,
+    create_opus_audio_decoder,
+    create_opus_audio_encoder,
+)
+
+
+@settings(max_examples=1000)
+@given(
+    samples=st.integers(min_value=0, max_value=96000),
+    # https://www.rfc-editor.org/rfc/rfc6716#section-2.1.1
+    # Opus supports all bitrates from 6 kbit/s to 510 kbit/s.
+    bitrate=st.integers(min_value=6000, max_value=510000),
+    # https://www.rfc-editor.org/rfc/rfc6716#section-2.1.4
+    # Opus can encode frames of 2.5, 5, 10, 20, 40, or 60 ms.  It can also combine multiple frames into packets of up to 120 ms.
+    # TODO: 2.5ms は現時点では対応しない
+    # frame_duration_ms=st.sampled_from([2.5, 5, 10, 20, 40, 60, 120]),
+    frame_duration_ms=st.sampled_from([5, 10, 20, 40, 60, 120]),
+)
+def test_prop_audio_encode_decode(samples, bitrate, frame_duration_ms):
+    """PBT を利用したエンコーダー・デコーダーのテスト"""
+    encoder = create_opus_audio_encoder()
+    encoder_settings = AudioEncoder.Settings()
+    encoder_settings.codec_type = AudioCodecType.OPUS
+    encoder_settings.sample_rate = 48000
+    encoder_settings.channels = 2
+    encoder_settings.bitrate = bitrate
+    encoder_settings.frame_duration_ms = frame_duration_ms
+    assert encoder.init(encoder_settings) is True
+
+    decoder = create_opus_audio_decoder()
+    decoder_settings = AudioDecoder.Settings()
+    decoder_settings.codec_type = AudioCodecType.OPUS
+    decoder_settings.sample_rate = 48000
+    decoder_settings.channels = 2
+    assert decoder.init(decoder_settings) is True
+
+    encoded_data = []
+    decoded_frames = []
+
+    def on_encoded(encoded: EncodedAudio):
+        assert isinstance(encoded.data, np.ndarray)
+        assert len(encoded.data) > 0
+        encoded_data.append(encoded)
+
+    def on_decoded(frame: AudioFrame):
+        decoded_frames.append(frame)
+
+    encoder.set_on_encode(on_encoded)
+    decoder.set_on_decode(on_decoded)
+
+    frame = AudioFrame()
+    frame.sample_rate = 48000
+    frame.timestamp = timedelta(milliseconds=1000)
+
+    # 時間軸を生成 (サンプル数 / サンプリングレート = 秒)
+    t = np.arange(samples) / 48000
+    # 440Hz のサイン波を生成 (振幅 0.1)
+    audio_signal = np.sin(2 * np.pi * 440 * t) * 0.1
+    frame.pcm = np.column_stack((audio_signal, audio_signal)).astype(np.float32)
+
+    # エンコード実行
+    encoder.encode(frame)
+
+    # frame_duration_ms に基づいて期待されるフレーム数を計算
+    samples_per_frame = int(48000 * frame_duration_ms / 1000)
+    expected_frames = samples // samples_per_frame
+    assert len(encoded_data) == expected_frames
+
+    # デコード実行
+    for encoded in encoded_data:
+        decoder.decode(encoded)
+
+    # デコードされたフレーム数の確認
+    assert len(decoded_frames) == expected_frames
+
+    # デコードされたフレームのプロパティを確認
+    for i, decoded_frame in enumerate(decoded_frames):
+        assert decoded_frame.sample_rate == 48000
+        assert decoded_frame.channels() == 2
+        assert decoded_frame.samples() == samples_per_frame
+        # タイムスタンプの確認
+        expected_timestamp = timedelta(milliseconds=1000 + i * frame_duration_ms)
+        assert decoded_frame.timestamp == expected_timestamp
+
+    encoder.release()
+    decoder.release()

libdatachannel_py-2025.1.0.dev9/tests/test_video_codec.py

@@ -0,0 +1,513 @@
+import os
+import time
+from datetime import timedelta
+
+import numpy as np
+import pytest
+
+from libdatachannel import (
+    NalUnit,
+    NalUnitFragmentA,
+)
+from libdatachannel.codec import (
+    EncodedImage,
+    ImageFormat,
+    VideoCodecType,
+    VideoDecoder,
+    VideoEncoder,
+    VideoFrame,
+    VideoFrameBufferBGR888,
+    VideoFrameBufferI420,
+    VideoFrameBufferNV12,
+    create_aom_video_decoder,
+    create_aom_video_encoder,
+    create_openh264_video_decoder,
+    create_openh264_video_encoder,
+    create_videotoolbox_video_encoder,
+)
+
+
+def test_video_frame_buffer():
+    # I420
+    i420_buffer = VideoFrameBufferI420.create(640, 480)
+    assert i420_buffer.width() == 640
+    assert i420_buffer.height() == 480
+    assert i420_buffer.stride_y() == 640
+    assert i420_buffer.stride_u() == 320
+    assert i420_buffer.stride_v() == 320
+    assert isinstance(i420_buffer.y, np.ndarray)
+    assert isinstance(i420_buffer.u, np.ndarray)
+    assert isinstance(i420_buffer.v, np.ndarray)
+    i420_buffer.y[0] = 255
+    i420_buffer.y = np.zeros((100, 100), dtype=np.uint8)
+    assert i420_buffer.width() == 100
+
+    # NV12
+    nv12_buffer = VideoFrameBufferNV12.create(640, 480)
+    assert nv12_buffer.width() == 640
+    assert nv12_buffer.height() == 480
+    assert nv12_buffer.stride_y() == 640
+    assert nv12_buffer.stride_uv() == 640
+    assert isinstance(nv12_buffer.y, np.ndarray)
+    assert isinstance(nv12_buffer.uv, np.ndarray)
+
+    # BGR888
+    bgr888_buffer = VideoFrameBufferBGR888.create(640, 480)
+    assert bgr888_buffer.width() == 640
+    assert bgr888_buffer.height() == 480
+    assert bgr888_buffer.stride() == 640 * 3
+    assert isinstance(bgr888_buffer.bgr, np.ndarray)
+
+
+def test_video_frame_properties():
+    frame = VideoFrame()
+
+    # format による width()/height() の分岐
+    frame.i420_buffer = VideoFrameBufferI420.create(640, 360)
+    frame.nv12_buffer = VideoFrameBufferNV12.create(1280, 720)
+    frame.bgr888_buffer = VideoFrameBufferBGR888.create(1920, 1080)
+
+    frame.format = ImageFormat.I420
+    assert frame.width() == 640
+    assert frame.height() == 360
+
+    frame.format = ImageFormat.NV12
+    assert frame.width() == 1280
+    assert frame.height() == 720
+
+    frame.format = ImageFormat.BGR888
+    assert frame.width() == 1920
+    assert frame.height() == 1080
+
+    # timestamp, rid
+    ts = timedelta(microseconds=1234567)
+    frame.timestamp = ts
+    assert frame.timestamp == ts
+
+    frame.rid = "abc"
+    assert frame.rid == "abc"
+
+    # base_width, base_height
+    frame.base_width = 999
+    frame.base_height = 888
+    assert frame.base_width == 999
+    assert frame.base_height == 888
+
+
+def test_encoded_image_properties():
+    img = EncodedImage()
+
+    img.data = np.array([], dtype=np.uint8)
+    assert img.data.size == 0
+    img.data = np.array([1, 2, 3], dtype=np.uint8)
+    assert img.data.size == 3
+
+    ts = timedelta(microseconds=555000)
+    img.timestamp = ts
+    assert img.timestamp == ts
+
+    img.rid = "r0"
+    assert img.rid == "r0"
+
+
+def test_openh264():
+    openh264 = os.environ.get("OPENH264_PATH")
+    assert openh264 is not None
+    encoder = create_openh264_video_encoder(openh264)
+    settings = VideoEncoder.Settings()
+    settings.codec_type = VideoCodecType.H264
+    settings.width = 640
+    settings.height = 480
+    settings.bitrate = 1000000
+    success = encoder.init(settings)
+    assert success
+    frame = VideoFrame()
+    frame.i420_buffer = VideoFrameBufferI420.create(640, 360)
+    frame.format = ImageFormat.I420
+    frame.base_width = 640
+    frame.base_height = 480
+    frame.timestamp = timedelta(microseconds=1234567)
+
+    on_encode_called = False
+
+    def on_encode(encoded_image):
+        nonlocal on_encode_called
+        assert encoded_image.data.size > 0
+        assert encoded_image.timestamp == frame.timestamp
+        assert encoded_image.rid == frame.rid
+        on_encode_called = True
+
+    encoder.set_on_encode(on_encode)
+    encoder.encode(frame)
+    encoder.release()
+    assert on_encode_called
+
+
+@pytest.mark.skipif(
+    os.environ.get("ENABLE_VIDEOTOOLBOX") is None, reason="macOS の場合だけ実行する"
+)
+def test_videotoolbox():
+    encoder = create_videotoolbox_video_encoder()
+    settings = VideoEncoder.Settings()
+    settings.codec_type = VideoCodecType.H264
+    settings.width = 640
+    settings.height = 480
+    settings.bitrate = 1000000
+    success = encoder.init(settings)
+    assert success
+    frame = VideoFrame()
+    frame.nv12_buffer = VideoFrameBufferNV12.create(640, 360)
+    frame.format = ImageFormat.NV12
+    frame.base_width = 640
+    frame.base_height = 480
+    frame.timestamp = timedelta(microseconds=1234567)
+
+    on_encode_called = False
+
+    def on_encode(encoded_image):
+        nonlocal on_encode_called
+        assert encoded_image.data.size > 0
+        assert encoded_image.timestamp == frame.timestamp
+        assert encoded_image.rid == frame.rid
+        on_encode_called = True
+
+    encoder.set_on_encode(on_encode)
+    encoder.encode(frame)
+    time.sleep(1)
+    encoder.release()
+    assert on_encode_called
+
+
+def test_aom():
+    encoder = create_aom_video_encoder()
+    settings = VideoEncoder.Settings()
+    settings.codec_type = VideoCodecType.H264
+    settings.width = 640
+    settings.height = 480
+    settings.bitrate = 1000000
+    success = encoder.init(settings)
+    assert success
+    frame = VideoFrame()
+    frame.i420_buffer = VideoFrameBufferI420.create(640, 360)
+    frame.format = ImageFormat.I420
+    frame.base_width = 640
+    frame.base_height = 480
+    frame.timestamp = timedelta(microseconds=1234567)
+
+    on_encode_called = False
+
+    def on_encode(encoded_image):
+        nonlocal on_encode_called
+        assert encoded_image.data.size > 0
+        assert encoded_image.timestamp == frame.timestamp
+        assert encoded_image.rid == frame.rid
+        on_encode_called = True
+
+    encoder.set_on_encode(on_encode)
+    encoder.encode(frame)
+    encoder.release()
+    assert on_encode_called
+
+
+def test_openh264_encode_decode():
+    """OpenH264を使用してnumpyで生成した映像をエンコード・デコードするテスト"""
+    openh264 = os.environ.get("OPENH264_PATH")
+    assert openh264 is not None
+
+    # エンコーダーの初期化
+    encoder = create_openh264_video_encoder(openh264)
+    settings = VideoEncoder.Settings()
+    settings.codec_type = VideoCodecType.H264
+    settings.width = 320
+    settings.height = 240
+    settings.bitrate = 500000
+    settings.fps = 30
+    success = encoder.init(settings)
+    assert success
+
+    # テスト用の映像フレームを生成（グラデーションパターン）
+    width, height = 320, 240
+    i420_buffer = VideoFrameBufferI420.create(width, height)
+
+    # Y平面：グラデーション（左から右へ明るくなる）
+    y_gradient = np.linspace(16, 235, width, dtype=np.uint8)
+    y_plane = y_gradient[np.newaxis, :].repeat(height, axis=0)
+    i420_buffer.y[:, :] = y_plane
+
+    # U,V平面：単色（グレー）
+    i420_buffer.u[:, :] = 128
+    i420_buffer.v[:, :] = 128
+
+    frame = VideoFrame()
+    frame.i420_buffer = i420_buffer
+    frame.format = ImageFormat.I420
+    frame.base_width = width
+    frame.base_height = height
+    frame.timestamp = timedelta(microseconds=1000)
+
+    encoded_frames = []
+
+    def on_encode(encoded_image):
+        encoded_frames.append(encoded_image)
+        assert encoded_image.data.size > 0
+
+    encoder.set_on_encode(on_encode)
+    encoder.force_intra_next_frame()  # キーフレームを生成
+    encoder.encode(frame)
+
+    # 2フレーム目も送信
+    frame.timestamp = timedelta(microseconds=1000 + 33333)  # 30fps
+    encoder.encode(frame)
+    encoder.release()
+
+    assert len(encoded_frames) == 2
+    print(f"Encoded {len(encoded_frames)} frames")
+
+    # デコーダーでデコード
+    decoder = create_openh264_video_decoder(openh264)
+    decoder_settings = VideoDecoder.Settings()
+    decoder_settings.codec_type = VideoCodecType.H264
+    success = decoder.init(decoder_settings)
+    assert success
+
+    decoded_frames = []
+
+    def on_decode(video_frame):
+        nonlocal decoded_frames
+        decoded_frames.append(video_frame)
+        print(f"Decoded frame: {video_frame.width()}x{video_frame.height()}")
+
+    decoder.set_on_decode(on_decode)
+
+    # すべてのエンコード済みフレームをデコード
+    for i, encoded in enumerate(encoded_frames):
+        encoded_image = EncodedImage()
+        encoded_image.data = encoded.data
+        encoded_image.timestamp = encoded.timestamp
+        print(f"Decoding frame {i + 1}: {encoded_image.data.size} bytes...")
+        decoder.decode(encoded_image)
+
+    decoder.release()
+
+    assert len(decoded_frames) == 2
+    assert decoded_frames[0].width() == width
+    assert decoded_frames[0].height() == height
+    assert decoded_frames[0].format == ImageFormat.I420
+    assert decoded_frames[0].timestamp == timedelta(microseconds=1000)
+    assert decoded_frames[1].width() == width
+    assert decoded_frames[1].height() == height
+    assert decoded_frames[1].format == ImageFormat.I420
+    assert decoded_frames[1].timestamp == timedelta(microseconds=1000 + 33333)
+
+    # デコードされたフレームの内容を簡単に検証
+    # Y平面の最初と最後の値がグラデーションになっているか確認
+    decoded_y = decoded_frames[0].i420_buffer.y.reshape(height, width)
+    assert decoded_y[0, 0] < decoded_y[0, -1]  # 左端より右端の方が明るい
+
+
+def test_openh264_with_nal_units():
+    """OpenH264とNALユニット解析を使用したテスト"""
+    openh264 = os.environ.get("OPENH264_PATH")
+    assert openh264 is not None
+
+    # エンコーダーの初期化
+    encoder = create_openh264_video_encoder(openh264)
+    settings = VideoEncoder.Settings()
+    settings.codec_type = VideoCodecType.H264
+    settings.width = 320
+    settings.height = 240
+    settings.bitrate = 500000
+    settings.fps = 30
+    success = encoder.init(settings)
+    assert success
+
+    # テスト用の映像フレームを生成（グラデーションパターン）
+    width, height = 320, 240
+    i420_buffer = VideoFrameBufferI420.create(width, height)
+
+    # Y平面：グラデーション（左から右へ明るくなる）
+    y_gradient = np.linspace(16, 235, width, dtype=np.uint8)
+    y_plane = y_gradient[np.newaxis, :].repeat(height, axis=0)
+    i420_buffer.y[:, :] = y_plane
+
+    # U,V平面：単色（グレー）
+    i420_buffer.u[:, :] = 128
+    i420_buffer.v[:, :] = 128
+
+    frame = VideoFrame()
+    frame.i420_buffer = i420_buffer
+    frame.format = ImageFormat.I420
+    frame.base_width = width
+    frame.base_height = height
+    frame.timestamp = timedelta(microseconds=1000)
+
+    encoded_frames = []
+
+    def on_encode(encoded_image):
+        encoded_frames.append(encoded_image)
+
+    encoder.set_on_encode(on_encode)
+    encoder.force_intra_next_frame()  # キーフレームを生成
+    encoder.encode(frame)
+
+    # 2フレーム目も送信
+    frame.timestamp = timedelta(microseconds=1000 + 33333)
+    encoder.encode(frame)
+    encoder.release()
+
+    assert len(encoded_frames) == 2
+    print(f"Encoded {len(encoded_frames)} frames")
+
+    # 最初のフレームのNALユニットを解析
+    first_encoded = encoded_frames[0].data
+    print(f"First frame size: {first_encoded.size} bytes")
+
+    # NALユニットを分割して作成
+    nal_units_byte = bytes(first_encoded).split(b"\x00\x00\x00\x01")[1:]
+    nal_units = list(map(lambda b: NalUnit(b), nal_units_byte))
+
+    # NALユニットのヘッダー情報を検証
+    assert len(nal_units) == 3
+    assert nal_units[0].unit_type() == 7  # SPS
+    assert nal_units[1].unit_type() == 8  # PPS
+    assert nal_units[2].unit_type() == 5  # IDR
+
+    # 複数のNALユニットに分割（フラグメント化）のテスト
+    fragments = NalUnitFragmentA.fragments_from(nal_units[2], max_fragment_size=100)
+    assert len(fragments) == 5
+    print(f"Generated {len(fragments)} fragments")
+
+    # デコーダーでデコード
+    decoder = create_openh264_video_decoder(openh264)
+    decoder_settings = VideoDecoder.Settings()
+    decoder_settings.codec_type = VideoCodecType.H264
+    success = decoder.init(decoder_settings)
+    assert success
+
+    decoded_frames = []
+
+    def on_decode(video_frame):
+        nonlocal decoded_frames
+        decoded_frames.append(video_frame)
+        print(f"Decoded frame: {video_frame.width()}x{video_frame.height()}")
+
+    decoder.set_on_decode(on_decode)
+
+    # すべてのエンコード済みフレームをデコード
+    for i, encoded in enumerate(encoded_frames):
+        encoded_image = EncodedImage()
+        encoded_image.data = encoded.data
+        encoded_image.timestamp = encoded.timestamp
+        print(f"Decoding frame {i + 1}: {encoded_image.data.size} bytes...")
+        decoder.decode(encoded_image)
+
+    decoder.release()
+
+    # デコード結果の検証
+    assert len(decoded_frames) == 2
+    assert decoded_frames[0].width() == width
+    assert decoded_frames[0].height() == height
+    assert decoded_frames[0].timestamp == timedelta(microseconds=1000)
+    assert decoded_frames[1].width() == width
+    assert decoded_frames[1].height() == height
+    assert decoded_frames[1].timestamp == timedelta(microseconds=1000 + 33333)
+
+    # デコードされたフレームの内容を簡単に検証
+    decoded_y = decoded_frames[0].i420_buffer.y.reshape(height, width)
+    assert decoded_y[0, 0] < decoded_y[0, -1]  # 左端より右端の方が明るい
+
+
+def test_aom_encode_decode():
+    """AOMを使用してnumpyで生成した映像をエンコード・デコードするテスト"""
+    # エンコーダーの初期化
+    encoder = create_aom_video_encoder()
+    settings = VideoEncoder.Settings()
+    settings.codec_type = VideoCodecType.AV1
+    settings.width = 320
+    settings.height = 240
+    settings.bitrate = 500000
+    settings.fps = 30
+    success = encoder.init(settings)
+    assert success
+
+    # テスト用の映像フレームを生成（グラデーションパターン）
+    width, height = 320, 240
+    i420_buffer = VideoFrameBufferI420.create(width, height)
+
+    # Y平面：グラデーション（左から右へ明るくなる）
+    y_gradient = np.linspace(16, 235, width, dtype=np.uint8)
+    y_plane = y_gradient[np.newaxis, :].repeat(height, axis=0)
+    i420_buffer.y[:, :] = y_plane
+
+    # U,V平面：単色（グレー）
+    i420_buffer.u[:, :] = 128
+    i420_buffer.v[:, :] = 128
+
+    frame = VideoFrame()
+    frame.i420_buffer = i420_buffer
+    frame.format = ImageFormat.I420
+    frame.base_width = width
+    frame.base_height = height
+    frame.timestamp = timedelta(microseconds=1000)
+
+    encoded_frames = []
+
+    def on_encode(encoded_image):
+        encoded_frames.append(encoded_image)
+        assert encoded_image.data.size > 0
+
+    encoder.set_on_encode(on_encode)
+    encoder.force_intra_next_frame()  # キーフレームを生成
+    encoder.encode(frame)
+
+    # 2フレーム目も送信
+    frame.timestamp = timedelta(microseconds=1000 + 33333)  # 30fps
+    encoder.encode(frame)
+    encoder.release()
+
+    assert len(encoded_frames) == 2
+    print(f"Encoded {len(encoded_frames)} frames")
+
+    assert encoded_frames[0].timestamp == timedelta(microseconds=1000)
+    assert encoded_frames[1].timestamp == timedelta(microseconds=1000 + 33333)
+
+    # デコーダーでデコード
+    decoder = create_aom_video_decoder()
+    decoder_settings = VideoDecoder.Settings()
+    decoder_settings.codec_type = VideoCodecType.AV1
+    success = decoder.init(decoder_settings)
+    assert success
+
+    decoded_frames = []
+
+    def on_decode(video_frame):
+        nonlocal decoded_frames
+        decoded_frames.append(video_frame)
+        print(f"Decoded frame: {video_frame.width()}x{video_frame.height()}")
+
+    decoder.set_on_decode(on_decode)
+
+    # すべてのエンコード済みフレームをデコード
+    for i, encoded in enumerate(encoded_frames):
+        encoded_image = EncodedImage()
+        encoded_image.data = encoded.data
+        encoded_image.timestamp = encoded.timestamp
+        print(f"Decoding frame {i + 1}: {encoded_image.data.size} bytes...")
+        decoder.decode(encoded_image)
+
+    decoder.release()
+
+    assert len(decoded_frames) == 2
+    assert decoded_frames[0].width() == width
+    assert decoded_frames[0].height() == height
+    assert decoded_frames[0].format == ImageFormat.I420
+    assert decoded_frames[0].timestamp == timedelta(microseconds=1000)
+    assert decoded_frames[1].width() == width
+    assert decoded_frames[1].height() == height
+    assert decoded_frames[1].format == ImageFormat.I420
+    assert decoded_frames[1].timestamp == timedelta(microseconds=1000 + 33333)
+
+    # デコードされたフレームの内容を簡単に検証
+    # Y平面の最初と最後の値がグラデーションになっているか確認
+    decoded_y = decoded_frames[0].i420_buffer.y.reshape(height, width)
+    assert decoded_y[0, 0] < decoded_y[0, -1]  # 左端より右端の方が明るい

libdatachannel_py-2025.1.0.dev8/VERSION

@@ -1 +0,0 @@
-2025.1.0.dev8

libdatachannel_py-2025.1.0.dev8/tests/test_video_codec.py

@@ -1,202 +0,0 @@
-import os
-import time
-from datetime import timedelta
-
-import numpy as np
-import pytest
-
-from libdatachannel.codec import (
-    EncodedImage,
-    ImageFormat,
-    VideoCodecType,
-    VideoEncoder,
-    VideoFrame,
-    VideoFrameBufferBGR888,
-    VideoFrameBufferI420,
-    VideoFrameBufferNV12,
-    create_aom_video_encoder,
-    create_openh264_video_encoder,
-    create_videotoolbox_video_encoder,
-)
-
-
-def test_video_frame_buffer():
-    # I420
-    i420_buffer = VideoFrameBufferI420.create(640, 480)
-    assert i420_buffer.width() == 640
-    assert i420_buffer.height() == 480
-    assert i420_buffer.stride_y() == 640
-    assert i420_buffer.stride_u() == 320
-    assert i420_buffer.stride_v() == 320
-    assert isinstance(i420_buffer.y, np.ndarray)
-    assert isinstance(i420_buffer.u, np.ndarray)
-    assert isinstance(i420_buffer.v, np.ndarray)
-    i420_buffer.y[0] = 255
-    i420_buffer.y = np.zeros((100, 100), dtype=np.uint8)
-    assert i420_buffer.width() == 100
-
-    # NV12
-    nv12_buffer = VideoFrameBufferNV12.create(640, 480)
-    assert nv12_buffer.width() == 640
-    assert nv12_buffer.height() == 480
-    assert nv12_buffer.stride_y() == 640
-    assert nv12_buffer.stride_uv() == 640
-    assert isinstance(nv12_buffer.y, np.ndarray)
-    assert isinstance(nv12_buffer.uv, np.ndarray)
-
-    # BGR888
-    bgr888_buffer = VideoFrameBufferBGR888.create(640, 480)
-    assert bgr888_buffer.width() == 640
-    assert bgr888_buffer.height() == 480
-    assert bgr888_buffer.stride() == 640 * 3
-    assert isinstance(bgr888_buffer.bgr, np.ndarray)
-
-
-def test_video_frame_properties():
-    frame = VideoFrame()
-
-    # format による width()/height() の分岐
-    frame.i420_buffer = VideoFrameBufferI420.create(640, 360)
-    frame.nv12_buffer = VideoFrameBufferNV12.create(1280, 720)
-    frame.bgr888_buffer = VideoFrameBufferBGR888.create(1920, 1080)
-
-    frame.format = ImageFormat.I420
-    assert frame.width() == 640
-    assert frame.height() == 360
-
-    frame.format = ImageFormat.NV12
-    assert frame.width() == 1280
-    assert frame.height() == 720
-
-    frame.format = ImageFormat.BGR888
-    assert frame.width() == 1920
-    assert frame.height() == 1080
-
-    # timestamp, rid
-    ts = timedelta(microseconds=1234567)
-    frame.timestamp = ts
-    assert frame.timestamp == ts
-
-    frame.rid = "abc"
-    assert frame.rid == "abc"
-
-    # base_width, base_height
-    frame.base_width = 999
-    frame.base_height = 888
-    assert frame.base_width == 999
-    assert frame.base_height == 888
-
-
-def test_encoded_image_properties():
-    img = EncodedImage()
-
-    img.data = np.array([], dtype=np.uint8)
-    assert img.data.size == 0
-    img.data = np.array([1, 2, 3], dtype=np.uint8)
-    assert img.data.size == 3
-
-    ts = timedelta(microseconds=555000)
-    img.timestamp = ts
-    assert img.timestamp == ts
-
-    img.rid = "r0"
-    assert img.rid == "r0"
-
-
-def test_openh264():
-    openh264 = os.environ.get("OPENH264_PATH")
-    assert openh264 is not None
-    encoder = create_openh264_video_encoder(openh264)
-    settings = VideoEncoder.Settings()
-    settings.codec_type = VideoCodecType.H264
-    settings.width = 640
-    settings.height = 480
-    settings.bitrate = 1000000
-    success = encoder.init(settings)
-    assert success
-    frame = VideoFrame()
-    frame.i420_buffer = VideoFrameBufferI420.create(640, 360)
-    frame.format = ImageFormat.I420
-    frame.base_width = 640
-    frame.base_height = 480
-    frame.timestamp = timedelta(microseconds=1234567)
-
-    on_encode_called = False
-
-    def on_encode(encoded_image):
-        nonlocal on_encode_called
-        assert encoded_image.data.size > 0
-        assert encoded_image.timestamp == frame.timestamp
-        assert encoded_image.rid == frame.rid
-        on_encode_called = True
-
-    encoder.set_on_encode(on_encode)
-    encoder.encode(frame)
-    encoder.release()
-    assert on_encode_called
-
-
-@pytest.mark.skipif(
-    os.environ.get("ENABLE_VIDEOTOOLBOX") is None, reason="macOS の場合だけ実行する"
-)
-def test_videotoolbox():
-    encoder = create_videotoolbox_video_encoder()
-    settings = VideoEncoder.Settings()
-    settings.codec_type = VideoCodecType.H264
-    settings.width = 640
-    settings.height = 480
-    settings.bitrate = 1000000
-    success = encoder.init(settings)
-    assert success
-    frame = VideoFrame()
-    frame.nv12_buffer = VideoFrameBufferNV12.create(640, 360)
-    frame.format = ImageFormat.NV12
-    frame.base_width = 640
-    frame.base_height = 480
-    frame.timestamp = timedelta(microseconds=1234567)
-
-    on_encode_called = False
-
-    def on_encode(encoded_image):
-        nonlocal on_encode_called
-        assert encoded_image.data.size > 0
-        assert encoded_image.timestamp == frame.timestamp
-        assert encoded_image.rid == frame.rid
-        on_encode_called = True
-
-    encoder.set_on_encode(on_encode)
-    encoder.encode(frame)
-    time.sleep(1)
-    encoder.release()
-    assert on_encode_called
-
-
-def test_aom():
-    encoder = create_aom_video_encoder()
-    settings = VideoEncoder.Settings()
-    settings.codec_type = VideoCodecType.H264
-    settings.width = 640
-    settings.height = 480
-    settings.bitrate = 1000000
-    success = encoder.init(settings)
-    assert success
-    frame = VideoFrame()
-    frame.i420_buffer = VideoFrameBufferI420.create(640, 360)
-    frame.format = ImageFormat.I420
-    frame.base_width = 640
-    frame.base_height = 480
-    frame.timestamp = timedelta(microseconds=1234567)
-
-    on_encode_called = False
-
-    def on_encode(encoded_image):
-        nonlocal on_encode_called
-        assert encoded_image.data.size > 0
-        assert encoded_image.timestamp == frame.timestamp
-        assert encoded_image.rid == frame.rid
-        on_encode_called = True
-
-    encoder.set_on_encode(on_encode)
-    encoder.encode(frame)
-    encoder.release()
-    assert on_encode_called