Merge pull request #17170 from Shadowghost/better-bitrates

Rework bitrate reporting

MediaBrowser.MediaEncoding/Probing/ProbeResultNormalizer.cs

@@ -254,16 +254,38 @@ namespace MediaBrowser.MediaEncoding.Probing
                 {
                     if (mediaStream.Type == MediaStreamType.Audio && !mediaStream.BitRate.HasValue)
                     {
-                        mediaStream.BitRate = GetEstimatedAudioBitrate(mediaStream.Codec, mediaStream.Channels);
+                        mediaStream.BitRate = GetEstimatedAudioBitrate(mediaStream.Codec, mediaStream.Profile, mediaStream.Channels);
                     }
                 }
 
-                var videoStreamsBitrate = info.MediaStreams.Where(i => i.Type == MediaStreamType.Video).Select(i => i.BitRate ?? 0).Sum();
-                // If ffprobe reported the container bitrate as being the same as the video stream bitrate, then it's wrong
-                if (videoStreamsBitrate == (info.Bitrate ?? 0))
+                // ffprobe frequently omits the per-stream video bitrate (common in MP4/MKV containers).
+                // Estimate the missing video bitrate as the container bitrate minus the combined stream bitrates.
+                var videoStreams = info.MediaStreams.Where(i => i.Type == MediaStreamType.Video).ToList();
+                if (info.Bitrate.HasValue
+                    && videoStreams.Count == 1
+                    && !videoStreams[0].BitRate.HasValue)
                 {
-                    info.InferTotalBitrate(true);
+                    var otherStreams = info.MediaStreams
+                        .Where(i => i.Type != MediaStreamType.Video && !i.IsExternal)
+                        .ToList();
+
+                    // Only attribute the leftover bitrate to the video stream if every audio stream's bitrate is known.
+                    var audioBitratesKnown = otherStreams
+                        .Where(i => i.Type == MediaStreamType.Audio)
+                        .All(i => i.BitRate.HasValue);
+
+                    if (audioBitratesKnown)
+                    {
+                        var estimatedVideoBitrate = info.Bitrate.Value - otherStreams.Sum(i => i.BitRate ?? 0);
+                        if (estimatedVideoBitrate > 0)
+                        {
+                            videoStreams[0].BitRate = estimatedVideoBitrate;
+                        }
+                    }
                 }
+
+                // If the container bitrate is still unknown, infer it from the sum of the streams.
+                info.InferTotalBitrate();
             }
 
             return info;
@@ -316,54 +338,34 @@ namespace MediaBrowser.MediaEncoding.Probing
             return string.Join(',', splitFormat.Where(s => !string.IsNullOrEmpty(s)));
         }
 
-        private static int? GetEstimatedAudioBitrate(string codec, int? channels)
+        internal static int? GetEstimatedAudioBitrate(string codec, string profile, int? channels)
         {
-            if (!channels.HasValue)
+            if (!channels.HasValue || channels.Value < 1 || string.IsNullOrEmpty(codec))
             {
                 return null;
             }
 
-            var channelsValue = channels.Value;
+            // Rough typical bitrates used only as a fallback when ffprobe doesn't report a stream bitrate.
+            var channelCount = channels.Value;
+            var isMultichannel = channelCount > 2;
 
-            if (string.Equals(codec, "aac", StringComparison.OrdinalIgnoreCase)
-                || string.Equals(codec, "mp3", StringComparison.OrdinalIgnoreCase))
+            return codec.ToLowerInvariant() switch
             {
-                switch (channelsValue)
-                {
-                    case <= 2:
-                        return 192000;
-                    case >= 5:
-                        return 320000;
-                }
-            }
-
-            if (string.Equals(codec, "ac3", StringComparison.OrdinalIgnoreCase)
-                || string.Equals(codec, "eac3", StringComparison.OrdinalIgnoreCase))
-            {
-                switch (channelsValue)
-                {
-                    case <= 2:
-                        return 192000;
-                    case >= 5:
-                        return 640000;
-                }
-            }
-
-            if (string.Equals(codec, "flac", StringComparison.OrdinalIgnoreCase)
-                || string.Equals(codec, "alac", StringComparison.OrdinalIgnoreCase))
-            {
-                switch (channelsValue)
-                {
-                    case <= 2:
-                        return 960000;
-                    case >= 5:
-                        return 2880000;
-                }
-            }
-
-            return null;
+                "aac" or "mp3" or "mp2" => isMultichannel ? 320000 : 192000,
+                "ac3" or "eac3" => isMultichannel ? 640000 : 192000,
+                "dts" or "dca" => IsDtsLossless(profile) ? channelCount * 700000 : (isMultichannel ? 1509000 : 768000),
+                "opus" => isMultichannel ? 256000 : 128000,
+                "vorbis" => isMultichannel ? 320000 : 160000,
+                "wmav1" or "wmav2" or "wmapro" => isMultichannel ? 384000 : 192000,
+                "flac" or "alac" => channelCount * 480000,
+                "truehd" or "mlp" => channelCount * 700000,
+                _ => null
+            };
         }
 
+        private static bool IsDtsLossless(string profile)
+            => profile is not null && profile.Contains("HD MA", StringComparison.OrdinalIgnoreCase);
+
         private void FetchFromItunesInfo(string xml, MediaInfo info)
         {
             // Make things simpler and strip out the dtd
@@ -972,10 +974,12 @@ namespace MediaBrowser.MediaEncoding.Probing
                 bitrate = value;
             }
 
-            // The bitrate info of FLAC musics and some videos is included in formatInfo.
+            // The bitrate info of FLAC audio is included in formatInfo.
+            // Don't do this for video streams: formatInfo.BitRate is the overall container
+            // bitrate (video + audio + subtitles + overhead), not the video bitrate.
             if (bitrate == 0
                 && formatInfo is not null
-                && (stream.Type == MediaStreamType.Video || (isAudio && stream.Type == MediaStreamType.Audio)))
+                && isAudio && stream.Type == MediaStreamType.Audio)
             {
                 // If the stream info doesn't have a bitrate get the value from the media format info
                 if (int.TryParse(formatInfo.BitRate, CultureInfo.InvariantCulture, out value))
@@ -1260,9 +1264,16 @@ namespace MediaBrowser.MediaEncoding.Probing
             }
 
             var duration = GetDictionaryValue(streamInfo.Tags, "DURATION-eng") ?? GetDictionaryValue(streamInfo.Tags, "DURATION");
-            if (TimeSpan.TryParse(duration, out var parsedDuration))
+            if (!string.IsNullOrEmpty(duration))
             {
-                return parsedDuration.TotalSeconds;
+                // Matroska DURATION tags use nanosecond precision (e.g. "00:00:05.023000000"), but
+                // TimeSpan only supports up to 7 fractional digits (ticks). Trim the surplus digits so
+                // these durations parse instead of being silently dropped.
+                duration = DurationOverPrecisionRegex().Replace(duration, "$1");
+                if (TimeSpan.TryParse(duration, CultureInfo.InvariantCulture, out var parsedDuration))
+                {
+                    return parsedDuration.TotalSeconds;
+                }
             }
 
             return null;
@@ -1764,5 +1775,8 @@ namespace MediaBrowser.MediaEncoding.Probing
 
         [GeneratedRegex("(?<name>.*) \\((?<instrument>.*)\\)")]
         private static partial Regex PerformerRegex();
+
+        [GeneratedRegex(@"(\.\d{7})\d+")]
+        private static partial Regex DurationOverPrecisionRegex();
     }
 }

tests/Jellyfin.MediaEncoding.Tests/Probing/ProbeResultNormalizerTests.cs

@@ -1,6 +1,7 @@
 using System;
 using System.Globalization;
 using System.IO;
+using System.Linq;
 using System.Text.Json;
 using Jellyfin.Data.Enums;
 using Jellyfin.Extensions.Json;
@@ -56,6 +57,43 @@ namespace Jellyfin.MediaEncoding.Tests.Probing
         public void IsNearSquarePixelSar_DetectsCorrectly(string? sar, bool expected)
             => Assert.Equal(expected, ProbeResultNormalizer.IsNearSquarePixelSar(sar));
 
+        [Theory]
+        // Lossy codecs, mono/stereo and multichannel.
+        [InlineData("aac", null, 2, 192000)]
+        [InlineData("mp3", null, 2, 192000)]
+        [InlineData("mp2", null, 2, 192000)]
+        [InlineData("aac", null, 6, 320000)]
+        [InlineData("ac3", null, 2, 192000)]
+        [InlineData("eac3", null, 6, 640000)]
+        [InlineData("opus", null, 2, 128000)]
+        [InlineData("vorbis", null, 6, 320000)]
+        [InlineData("wmav2", null, 2, 192000)]
+        // DTS: the lossy core (any non-MA profile, or none) is flat and caps at 5.1...
+        [InlineData("dts", null, 2, 768000)]
+        [InlineData("dts", "DTS", 6, 1509000)]
+        [InlineData("dts", "DTS-HD HRA", 8, 1509000)]
+        // ...while lossless DTS-HD MA scales per channel like other lossless codecs.
+        [InlineData("dts", "DTS-HD MA", 6, 4200000)]
+        [InlineData("dts", "DTS-HD MA + DTS:X", 8, 5600000)]
+        // Lossless codecs scale per channel.
+        [InlineData("flac", null, 2, 960000)]
+        [InlineData("flac", null, 6, 2880000)]
+        [InlineData("flac", null, 8, 3840000)]
+        [InlineData("alac", null, 6, 2880000)]
+        [InlineData("truehd", null, 2, 1400000)]
+        [InlineData("truehd", null, 6, 4200000)]
+        [InlineData("truehd", "Dolby TrueHD + Dolby Atmos", 8, 5600000)]
+        // 3-4 channel audio must use the multichannel estimate, not return null.
+        [InlineData("aac", null, 3, 320000)]
+        [InlineData("ac3", null, 4, 640000)]
+        // Codec matching is case-insensitive.
+        [InlineData("AAC", null, 2, 192000)]
+        // Unknown codec or unknown channel count cannot be estimated.
+        [InlineData("pcm_s16le", null, 2, null)]
+        [InlineData("aac", null, null, null)]
+        public void GetEstimatedAudioBitrate_ReturnsExpected(string codec, string? profile, int? channels, int? expected)
+            => Assert.Equal(expected, ProbeResultNormalizer.GetEstimatedAudioBitrate(codec, profile, channels));
+
         [Fact]
         public void GetMediaInfo_MetaData_Success()
         {
@@ -71,7 +109,10 @@ namespace Jellyfin.MediaEncoding.Tests.Probing
             Assert.Equal("4:3", res.VideoStream.AspectRatio);
             Assert.Equal(25f, res.VideoStream.AverageFrameRate);
             Assert.Equal(8, res.VideoStream.BitDepth);
-            Assert.Equal(69432, res.VideoStream.BitRate);
+            // ffprobe reports no per-stream video bitrate here. The container bitrate must not be
+            // misreported as the video bitrate, and the other streams' bitrates exceed the container
+            // bitrate in this sample, so no sensible video bitrate can be inferred (see #16248).
+            Assert.Null(res.VideoStream.BitRate);
             Assert.Equal("h264", res.VideoStream.Codec);
             Assert.Equal("1/50", res.VideoStream.CodecTimeBase);
             Assert.Equal(240, res.VideoStream.Height);
@@ -321,6 +362,73 @@ namespace Jellyfin.MediaEncoding.Tests.Probing
             Assert.True(res.VideoStream.IsDefault);
         }
 
+        [Fact]
+        public void GetMediaInfo_MissingVideoBitrate_EstimatedFromContainer()
+        {
+            // ffprobe did not report a per-stream video bitrate. The video bitrate must be estimated
+            // as the container bitrate minus the other (audio) stream bitrates, not reported as the
+            // whole container bitrate (see #16248).
+            var bytes = File.ReadAllBytes("Test Data/Probing/video_missing_video_bitrate.json");
+
+            var internalMediaInfoResult = JsonSerializer.Deserialize<InternalMediaInfoResult>(bytes, _jsonOptions);
+            MediaInfo res = _probeResultNormalizer.GetMediaInfo(internalMediaInfoResult, VideoType.VideoFile, false, "Test Data/Probing/video_missing_video_bitrate.mp4", MediaProtocol.File);
+
+            Assert.Equal(2, res.MediaStreams.Count);
+
+            Assert.NotNull(res.VideoStream);
+            Assert.Equal(MediaStreamType.Video, res.VideoStream.Type);
+
+            var audioStream = res.MediaStreams.First(i => i.Type == MediaStreamType.Audio);
+            Assert.Equal(128000, audioStream.BitRate);
+
+            // Container bitrate (5128000) minus the audio bitrate (128000).
+            Assert.Equal(5000000, res.VideoStream.BitRate);
+
+            // The container bitrate itself must remain the overall container bitrate.
+            Assert.Equal(5128000, res.Bitrate);
+        }
+
+        [Fact]
+        public void GetMediaInfo_NanosecondDurationTag_BitrateComputedFromBytes()
+        {
+            // The stream carries NUMBER_OF_BYTES and a nanosecond-precision DURATION tag but no
+            // bitrate. TimeSpan only supports 7 fractional digits, so the 9-digit DURATION must be
+            // trimmed for the duration to parse and the bitrate to be computed (bytes * 8 / seconds).
+            var bytes = File.ReadAllBytes("Test Data/Probing/video_nanosecond_duration_bitrate.json");
+
+            var internalMediaInfoResult = JsonSerializer.Deserialize<InternalMediaInfoResult>(bytes, _jsonOptions);
+            MediaInfo res = _probeResultNormalizer.GetMediaInfo(internalMediaInfoResult, VideoType.VideoFile, false, "Test Data/Probing/video_nanosecond_duration_bitrate.mkv", MediaProtocol.File);
+
+            Assert.NotNull(res.VideoStream);
+
+            // 10000000 bytes * 8 / 100 seconds.
+            Assert.Equal(800000, res.VideoStream.BitRate);
+        }
+
+        [Fact]
+        public void GetMediaInfo_MissingVideoBitrate_UnknownAudioBitrate_NotEstimated()
+        {
+            // ffprobe reported no per-stream video bitrate and the audio bitrate cannot be estimated
+            // (the audio stream has no channel count, so GetEstimatedAudioBitrate returns null). The
+            // video bitrate must be left unset rather than wrongly absorbing the unaccounted audio
+            // bitrate (see #16248).
+            var bytes = File.ReadAllBytes("Test Data/Probing/video_missing_video_bitrate_unknown_audio.json");
+
+            var internalMediaInfoResult = JsonSerializer.Deserialize<InternalMediaInfoResult>(bytes, _jsonOptions);
+            MediaInfo res = _probeResultNormalizer.GetMediaInfo(internalMediaInfoResult, VideoType.VideoFile, false, "Test Data/Probing/video_missing_video_bitrate_unknown_audio.mp4", MediaProtocol.File);
+
+            Assert.Equal(2, res.MediaStreams.Count);
+
+            Assert.NotNull(res.VideoStream);
+            Assert.Null(res.VideoStream.BitRate);
+
+            var audioStream = res.MediaStreams.First(i => i.Type == MediaStreamType.Audio);
+            Assert.Null(audioStream.BitRate);
+
+            // The overall container bitrate is still reported.
+            Assert.Equal(5128000, res.Bitrate);
+        }
+
         [Fact]
         public void GetMediaInfo_VideoWithSingleFrameMjpeg_Success()
         {

tests/Jellyfin.MediaEncoding.Tests/Test Data/Probing/video_missing_video_bitrate.json

@@ -0,0 +1,113 @@
+{
+    "streams": [
+        {
+            "index": 0,
+            "codec_name": "h264",
+            "codec_long_name": "H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10",
+            "profile": "High",
+            "codec_type": "video",
+            "codec_time_base": "1/48",
+            "codec_tag_string": "avc1",
+            "codec_tag": "0x31637661",
+            "width": 1920,
+            "height": 1080,
+            "coded_width": 1920,
+            "coded_height": 1080,
+            "closed_captions": 0,
+            "has_b_frames": 2,
+            "pix_fmt": "yuv420p",
+            "level": 40,
+            "chroma_location": "left",
+            "refs": 1,
+            "is_avc": "true",
+            "nal_length_size": "4",
+            "r_frame_rate": "24/1",
+            "avg_frame_rate": "24/1",
+            "time_base": "1/12288",
+            "start_pts": 0,
+            "start_time": "0.000000",
+            "duration_ts": 3686400,
+            "duration": "300.000000",
+            "bits_per_raw_sample": "8",
+            "nb_frames": "7200",
+            "disposition": {
+                "default": 1,
+                "dub": 0,
+                "original": 0,
+                "comment": 0,
+                "lyrics": 0,
+                "karaoke": 0,
+                "forced": 0,
+                "hearing_impaired": 0,
+                "visual_impaired": 0,
+                "clean_effects": 0,
+                "attached_pic": 0,
+                "timed_thumbnails": 0
+            },
+            "tags": {
+                "language": "und",
+                "handler_name": "VideoHandler"
+            }
+        },
+        {
+            "index": 1,
+            "codec_name": "aac",
+            "codec_long_name": "AAC (Advanced Audio Coding)",
+            "profile": "LC",
+            "codec_type": "audio",
+            "codec_time_base": "1/48000",
+            "codec_tag_string": "mp4a",
+            "codec_tag": "0x6134706d",
+            "sample_fmt": "fltp",
+            "sample_rate": "48000",
+            "channels": 2,
+            "channel_layout": "stereo",
+            "bits_per_sample": 0,
+            "r_frame_rate": "0/0",
+            "avg_frame_rate": "0/0",
+            "time_base": "1/48000",
+            "start_pts": 0,
+            "start_time": "0.000000",
+            "duration_ts": 14400000,
+            "duration": "300.000000",
+            "bit_rate": "128000",
+            "nb_frames": "14063",
+            "disposition": {
+                "default": 1,
+                "dub": 0,
+                "original": 0,
+                "comment": 0,
+                "lyrics": 0,
+                "karaoke": 0,
+                "forced": 0,
+                "hearing_impaired": 0,
+                "visual_impaired": 0,
+                "clean_effects": 0,
+                "attached_pic": 0,
+                "timed_thumbnails": 0
+            },
+            "tags": {
+                "language": "eng",
+                "handler_name": "SoundHandler"
+            }
+        }
+    ],
+    "format": {
+        "filename": "test.1080p.mp4",
+        "nb_streams": 2,
+        "nb_programs": 0,
+        "format_name": "mov,mp4,m4a,3gp,3g2,mj2",
+        "format_long_name": "QuickTime / MOV",
+        "start_time": "0.000000",
+        "duration": "300.000000",
+        "size": "192000000",
+        "bit_rate": "5128000",
+        "probe_score": 100,
+        "tags": {
+            "major_brand": "isom",
+            "minor_version": "512",
+            "compatible_brands": "isomiso2avc1mp41",
+            "encoder": "Lavf58.20.100"
+        }
+    }
+}

tests/Jellyfin.MediaEncoding.Tests/Test Data/Probing/video_missing_video_bitrate_unknown_audio.json

@@ -0,0 +1,110 @@
+{
+    "streams": [
+        {
+            "index": 0,
+            "codec_name": "h264",
+            "codec_long_name": "H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10",
+            "profile": "High",
+            "codec_type": "video",
+            "codec_time_base": "1/48",
+            "codec_tag_string": "avc1",
+            "codec_tag": "0x31637661",
+            "width": 1920,
+            "height": 1080,
+            "coded_width": 1920,
+            "coded_height": 1080,
+            "closed_captions": 0,
+            "has_b_frames": 2,
+            "pix_fmt": "yuv420p",
+            "level": 40,
+            "chroma_location": "left",
+            "refs": 1,
+            "is_avc": "true",
+            "nal_length_size": "4",
+            "r_frame_rate": "24/1",
+            "avg_frame_rate": "24/1",
+            "time_base": "1/12288",
+            "start_pts": 0,
+            "start_time": "0.000000",
+            "duration_ts": 3686400,
+            "duration": "300.000000",
+            "bits_per_raw_sample": "8",
+            "nb_frames": "7200",
+            "disposition": {
+                "default": 1,
+                "dub": 0,
+                "original": 0,
+                "comment": 0,
+                "lyrics": 0,
+                "karaoke": 0,
+                "forced": 0,
+                "hearing_impaired": 0,
+                "visual_impaired": 0,
+                "clean_effects": 0,
+                "attached_pic": 0,
+                "timed_thumbnails": 0
+            },
+            "tags": {
+                "language": "und",
+                "handler_name": "VideoHandler"
+            }
+        },
+        {
+            "index": 1,
+            "codec_name": "dts",
+            "codec_long_name": "DCA (DTS Coherent Acoustics)",
+            "profile": "DTS-HD MA",
+            "codec_type": "audio",
+            "codec_time_base": "1/48000",
+            "codec_tag_string": "[0][0][0][0]",
+            "codec_tag": "0x0000",
+            "sample_fmt": "s32p",
+            "sample_rate": "48000",
+            "bits_per_sample": 0,
+            "r_frame_rate": "0/0",
+            "avg_frame_rate": "0/0",
+            "time_base": "1/48000",
+            "start_pts": 0,
+            "start_time": "0.000000",
+            "duration_ts": 14400000,
+            "duration": "300.000000",
+            "nb_frames": "14063",
+            "disposition": {
+                "default": 1,
+                "dub": 0,
+                "original": 0,
+                "comment": 0,
+                "lyrics": 0,
+                "karaoke": 0,
+                "forced": 0,
+                "hearing_impaired": 0,
+                "visual_impaired": 0,
+                "clean_effects": 0,
+                "attached_pic": 0,
+                "timed_thumbnails": 0
+            },
+            "tags": {
+                "language": "eng",
+                "handler_name": "SoundHandler"
+            }
+        }
+    ],
+    "format": {
+        "filename": "test.1080p.mp4",
+        "nb_streams": 2,
+        "nb_programs": 0,
+        "format_name": "mov,mp4,m4a,3gp,3g2,mj2",
+        "format_long_name": "QuickTime / MOV",
+        "start_time": "0.000000",
+        "duration": "300.000000",
+        "size": "192000000",
+        "bit_rate": "5128000",
+        "probe_score": 100,
+        "tags": {
+            "major_brand": "isom",
+            "minor_version": "512",
+            "compatible_brands": "isomiso2avc1mp41",
+            "encoder": "Lavf58.20.100"
+        }
+    }
+}

tests/Jellyfin.MediaEncoding.Tests/Test Data/Probing/video_nanosecond_duration_bitrate.json

@@ -0,0 +1,49 @@
+{
+    "streams": [
+        {
+            "index": 0,
+            "codec_name": "h264",
+            "codec_long_name": "H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10",
+            "profile": "High",
+            "codec_type": "video",
+            "codec_tag_string": "[0][0][0][0]",
+            "codec_tag": "0x0000",
+            "width": 1920,
+            "height": 1080,
+            "coded_width": 1920,
+            "coded_height": 1080,
+            "has_b_frames": 2,
+            "pix_fmt": "yuv420p",
+            "level": 40,
+            "r_frame_rate": "24/1",
+            "avg_frame_rate": "24/1",
+            "time_base": "1/1000",
+            "start_pts": 0,
+            "start_time": "0.000000",
+            "disposition": {
+                "default": 1
+            },
+            "tags": {
+                "language": "eng",
+                "BPS-eng": "",
+                "DURATION-eng": "00:01:40.000000000",
+                "NUMBER_OF_FRAMES-eng": "2400",
+                "NUMBER_OF_BYTES-eng": "10000000"
+            }
+        }
+    ],
+    "format": {
+        "filename": "video_nanosecond_duration_bitrate.mkv",
+        "nb_streams": 1,
+        "nb_programs": 0,
+        "format_name": "matroska,webm",
+        "format_long_name": "Matroska / WebM",
+        "start_time": "0.000000",
+        "duration": "100.000000",
+        "size": "10001000",
+        "probe_score": 100,
+        "tags": {
+            "encoder": "libebml v1.4.2 + libmatroska v1.6.4"
+        }
+    }
+}