EBML: Get MKA properties test passing

lofty/src/ebml/element_reader.rs

@@ -146,13 +146,17 @@ ebml_master_elements! {
 	TrackEntry: {
 		id: 0xAE,
 		children: [
+			TrackNumber: { 0xD7, UnsignedInt },
+			TrackUid: { 0x73C5, UnsignedInt },
 			TrackType: { 0x83, UnsignedInt },
 			FlagEnabled: { 0xB9, UnsignedInt },
 			FlagDefault: { 0x88, UnsignedInt },
 			DefaultDuration: { 0x23E3_83, UnsignedInt },
 			TrackTimecodeScale: { 0x2331_59, Float },
 			Language: { 0x22B5_9C, String },
+			LanguageBCP47: { 0x22B59D, String },
 			CodecID: { 0x86, String },
+			CodecPrivate: { 0x63A2, Binary },
 			CodecName: { 0x258688, Utf8 },
 			CodecDelay: { 0x56AA, UnsignedInt },
 			SeekPreRoll: { 0x56BB, UnsignedInt },
@@ -168,6 +172,7 @@ ebml_master_elements! {
 			OutputSamplingFrequency: { 0x78B5, Float },
 			Channels: { 0x9F, UnsignedInt },
 			BitDepth: { 0x6264, UnsignedInt },
+			Emphasis: { 0x52F1, UnsignedInt },
 		],
 	},
 

lofty/src/ebml/properties.rs

@@ -1,3 +1,4 @@
+use super::Language;
 use crate::properties::FileProperties;
 
 /// Properties from the EBML header
@@ -113,17 +114,41 @@ impl Default for SegmentInfo {
 }
 
 /// A full descriptor for an audio track
-#[derive(Debug, Clone, PartialEq, Default)]
+#[derive(Debug, Clone, PartialEq)]
 pub struct AudioTrackDescriptor {
 	pub(crate) number: u64,
 	pub(crate) uid: u64,
-	pub(crate) language: String,
+	pub(crate) enabled: bool,
+	pub(crate) default: bool,
+	pub(crate) language: Language,
 	pub(crate) default_duration: u64,
 	pub(crate) codec_id: String,
-	pub(crate) codec_private: Vec<u8>,
+	pub(crate) codec_private: Option<Vec<u8>>,
+	pub(crate) codec_name: Option<String>,
 	pub(crate) settings: AudioTrackSettings,
 }
 
+impl Default for AudioTrackDescriptor {
+	fn default() -> Self {
+		AudioTrackDescriptor {
+			// Note, these values are not spec compliant and will hopefully be overwritten when
+			// parsing. It doesn't really matter though, since we aren't an encoder.
+			number: 0,
+			uid: 0,
+			default_duration: 0,
+			codec_id: String::new(),
+
+			// Spec-compliant defaults
+			enabled: true,
+			default: true,
+			language: Language::Iso639_2(String::from("eng")),
+			codec_private: None,
+			codec_name: None,
+			settings: AudioTrackSettings::default(),
+		}
+	}
+}
+
 impl AudioTrackDescriptor {
 	/// The track number
 	pub fn number(&self) -> u64 {
@@ -135,10 +160,20 @@ impl AudioTrackDescriptor {
 		self.uid
 	}
 
+	/// Whether the track is usable
+	pub fn is_enabled(&self) -> bool {
+		self.enabled
+	}
+
+	/// Whether the track is eligible for automatic selection
+	pub fn is_default(&self) -> bool {
+		self.default
+	}
+
 	/// The language of the track, in the Matroska languages form
 	///
 	/// NOTE: See [basics](https://matroska.org/technical/basics.html#language-codes) on language codes.
-	pub fn language(&self) -> &str {
+	pub fn language(&self) -> &Language {
 		&self.language
 	}
 
@@ -157,8 +192,13 @@ impl AudioTrackDescriptor {
 	}
 
 	/// Private data only known to the codec
-	pub fn codec_private(&self) -> &[u8] {
-		&self.codec_private
+	pub fn codec_private(&self) -> Option<&[u8]> {
+		self.codec_private.as_deref()
+	}
+
+	/// A human-readable string for the [codec_id](AudioTrackDescriptor::codec_id)
+	pub fn codec_name(&self) -> Option<&str> {
+		self.codec_name.as_deref()
 	}
 
 	/// The audio settings of the track
@@ -170,8 +210,8 @@ impl AudioTrackDescriptor {
 /// Settings for an audio track
 #[derive(Debug, Clone, PartialEq, Default)]
 pub struct AudioTrackSettings {
-	pub(crate) sampling_frequency: u32,
-	pub(crate) output_sampling_frequency: u32,
+	pub(crate) sampling_frequency: f64,
+	pub(crate) output_sampling_frequency: f64,
 	pub(crate) channels: u8,
 	pub(crate) bit_depth: Option<u8>,
 	pub(crate) emphasis: Option<EbmlAudioTrackEmphasis>,
@@ -179,14 +219,14 @@ pub struct AudioTrackSettings {
 
 impl AudioTrackSettings {
 	/// The sampling frequency of the track
-	pub fn sampling_frequency(&self) -> u32 {
+	pub fn sampling_frequency(&self) -> f64 {
 		self.sampling_frequency
 	}
 
 	/// Real output sampling frequency in Hz (used for SBR techniques).
 	///
 	/// The default value for `output_sampling_frequency` of the same TrackEntry is equal to the [`Self::sampling_frequency`].
-	pub fn output_sampling_frequency(&self) -> u32 {
+	pub fn output_sampling_frequency(&self) -> f64 {
 		self.output_sampling_frequency
 	}
 
@@ -210,7 +250,6 @@ impl AudioTrackSettings {
 #[allow(missing_docs)]
 #[derive(Debug, Copy, Clone, Eq, PartialEq)]
 pub enum EbmlAudioTrackEmphasis {
-	None = 0,
 	CdAudio = 1,
 	Reserved = 2,
 	CcitJ17 = 3,
@@ -225,6 +264,27 @@ pub enum EbmlAudioTrackEmphasis {
 	PhonoNartb = 16,
 }
 
+impl EbmlAudioTrackEmphasis {
+	/// Get the audio emphasis from a `u8`
+	pub fn from_u8(value: u8) -> Option<Self> {
+		match value {
+			1 => Some(Self::CdAudio),
+			2 => Some(Self::Reserved),
+			3 => Some(Self::CcitJ17),
+			4 => Some(Self::Fm50),
+			5 => Some(Self::Fm75),
+			10 => Some(Self::PhonoRiaa),
+			11 => Some(Self::PhonoIecN78),
+			12 => Some(Self::PhonoTeldec),
+			13 => Some(Self::PhonoEmi),
+			14 => Some(Self::PhonoColumbiaLp),
+			15 => Some(Self::PhonoLondon),
+			16 => Some(Self::PhonoNartb),
+			_ => None,
+		}
+	}
+}
+
 /// EBML audio properties
 #[derive(Debug, Clone, PartialEq, Default)]
 pub struct EbmlProperties {
@@ -248,16 +308,14 @@ impl EbmlProperties {
 		&self.extensions
 	}
 
-	/// Information from the `\EBML\Segment\Info` element
+	/// Information from the `\Segment\Info` element
 	pub fn segment_info(&self) -> &SegmentInfo {
 		&self.segment_info
 	}
 
 	/// All audio tracks in the file
 	///
-	/// This includes all audio tracks in the Matroska `\EBML\Segment\Tracks` element.
-	///
-	/// NOTE: The first audio track is **always** the default audio track.
+	/// This includes all audio tracks in the Matroska `\Segment\Tracks` element.
 	pub fn audio_tracks(&self) -> &[AudioTrackDescriptor] {
 		&self.audio_tracks
 	}
@@ -265,11 +323,9 @@ impl EbmlProperties {
 	/// Information about the default audio track
 	///
 	/// The information is extracted from the first audio track with its default flag set
-	/// in the `\EBML\Segment\Tracks` element.
-	///
-	/// NOTE: This will always return `Some` unless [`ParseOptions::read_properties`](crate::config::ParseOptions::read_properties) is set to `false`.
+	/// in the `\Segment\Tracks` element.
 	pub fn default_audio_track(&self) -> Option<&AudioTrackDescriptor> {
-		self.audio_tracks.first()
+		self.audio_tracks.iter().find(|track| track.default)
 	}
 }
 
@@ -283,7 +339,7 @@ impl From<EbmlProperties> for FileProperties {
 			duration: todo!("Support duration"),
 			overall_bitrate: todo!("Support bitrate"),
 			audio_bitrate: todo!("Support bitrate"),
-			sample_rate: Some(default_audio_track.settings.sampling_frequency),
+			sample_rate: Some(default_audio_track.settings.sampling_frequency as u32),
 			bit_depth: default_audio_track.settings.bit_depth,
 			channels: Some(default_audio_track.settings.channels),
 			channel_mask: todo!("Channel mask"),

lofty/src/ebml/read/segment_tracks.rs

@@ -3,6 +3,7 @@ use crate::ebml::element_reader::{
 	ChildElementDescriptor, ElementChildIterator, ElementIdent, ElementReaderYield,
 };
 use crate::ebml::properties::EbmlProperties;
+use crate::ebml::{AudioTrackDescriptor, EbmlAudioTrackEmphasis, Language};
 use crate::error::Result;
 
 use std::io::{Read, Seek};
@@ -10,17 +11,15 @@ use std::io::{Read, Seek};
 pub(super) fn read_from<R>(
 	children_reader: &mut ElementChildIterator<'_, R>,
 	parse_options: ParseOptions,
-	_properties: &mut EbmlProperties,
+	properties: &mut EbmlProperties,
 ) -> Result<()>
 where
 	R: Read + Seek,
 {
-	let mut audio_tracks = Vec::new();
-
 	while let Some(child) = children_reader.next()? {
 		match child {
 			ElementReaderYield::Master((ElementIdent::TrackEntry, _size)) => {
-				read_track_entry(children_reader, parse_options, &mut audio_tracks)?;
+				read_track_entry(children_reader, parse_options, &mut properties.audio_tracks)?;
 			},
 			ElementReaderYield::Eof => break,
 			_ => {
@@ -32,30 +31,30 @@ where
 	Ok(())
 }
 
-#[derive(Default)]
-struct AudioTrack {
-	default: bool,
-	enabled: bool,
-	codec_id: String,
-	codec_name: String,
-}
-
 const AUDIO_TRACK_TYPE: u64 = 2;
 
 fn read_track_entry<R>(
 	children_reader: &mut ElementChildIterator<'_, R>,
-	_parse_options: ParseOptions,
-	audio_tracks: &mut Vec<AudioTrack>,
+	parse_options: ParseOptions,
+	audio_tracks: &mut Vec<AudioTrackDescriptor>,
 ) -> Result<()>
 where
 	R: Read + Seek,
 {
-	let mut track = AudioTrack::default();
+	let mut track = AudioTrackDescriptor::default();
 
 	while let Some(child) = children_reader.next()? {
 		match child {
 			ElementReaderYield::Child((ChildElementDescriptor { ident, .. }, size)) => {
 				match ident {
+					ElementIdent::TrackNumber => {
+						let track_number = children_reader.read_unsigned_int(size.value())?;
+						track.number = track_number;
+					},
+					ElementIdent::TrackUid => {
+						let track_uid = children_reader.read_unsigned_int(size.value())?;
+						track.uid = track_uid;
+					},
 					ElementIdent::TrackType => {
 						let track_type = children_reader.read_unsigned_int(size.value())?;
 						log::trace!("Encountered new track of type: {}", track_type);
@@ -80,18 +79,27 @@ where
 						let _timecode_scale = children_reader.read_float(size.value())?;
 					},
 					ElementIdent::Language => {
-						let _language = children_reader.read_string(size.value())?;
+						let language = children_reader.read_string(size.value())?;
+						track.language = Language::Iso639_2(language);
+					},
+					ElementIdent::LanguageBCP47 => {
+						let language = children_reader.read_string(size.value())?;
+						track.language = Language::Bcp47(language);
 					},
 					ElementIdent::CodecID => {
 						let codec_id = children_reader.read_string(size.value())?;
 						track.codec_id = codec_id;
 					},
+					ElementIdent::CodecPrivate => {
+						let codec_private = children_reader.read_binary(size.value())?;
+						track.codec_private = Some(codec_private);
+					},
 					ElementIdent::CodecDelay => {
 						let _codec_delay = children_reader.read_unsigned_int(size.value())?;
 					},
 					ElementIdent::CodecName => {
 						let codec_name = children_reader.read_utf8(size.value())?;
-						track.codec_name = codec_name;
+						track.codec_name = Some(codec_name);
 					},
 					ElementIdent::SeekPreRoll => {
 						let _seek_pre_roll = children_reader.read_unsigned_int(size.value())?;
@@ -101,7 +109,7 @@ where
 			},
 			ElementReaderYield::Master((id, size)) => match id {
 				ElementIdent::Audio => {
-					children_reader.skip(size.value())?;
+					read_audio_settings(&mut children_reader.children(), parse_options, &mut track)?
 				},
 				_ => {
 					unreachable!("Unhandled master element in TrackEntry: {:?}", id);
@@ -114,12 +122,59 @@ where
 		}
 	}
 
-	if !track.enabled {
-		log::debug!("Skipping disabled track");
-		return Ok(());
-	}
-
 	audio_tracks.push(track);
 
 	Ok(())
 }
+
+fn read_audio_settings<R>(
+	children_reader: &mut ElementChildIterator<'_, R>,
+	_parse_options: ParseOptions,
+	audio_track: &mut AudioTrackDescriptor,
+) -> Result<()>
+where
+	R: Read + Seek,
+{
+	while let Some(child) = children_reader.next()? {
+		match child {
+			ElementReaderYield::Child((ChildElementDescriptor { ident, .. }, size)) => {
+				match ident {
+					ElementIdent::SamplingFrequency => {
+						let sampling_frequency = children_reader.read_float(size.value())?;
+						audio_track.settings.sampling_frequency = sampling_frequency;
+					},
+					ElementIdent::OutputSamplingFrequency => {
+						let output_sampling_frequency = children_reader.read_float(size.value())?;
+						audio_track.settings.output_sampling_frequency = output_sampling_frequency;
+					},
+					ElementIdent::Channels => {
+						let channels = children_reader.read_unsigned_int(size.value())? as u8;
+						audio_track.settings.channels = channels;
+					},
+					ElementIdent::BitDepth => {
+						let bit_depth = children_reader.read_unsigned_int(size.value())? as u8;
+						audio_track.settings.bit_depth = Some(bit_depth);
+					},
+					ElementIdent::Emphasis => {
+						let emphasis = children_reader.read_unsigned_int(size.value())?;
+						if emphasis == 0 {
+							continue; // No emphasis
+						}
+
+						audio_track.settings.emphasis =
+							EbmlAudioTrackEmphasis::from_u8(emphasis as u8);
+					},
+					_ => {
+						unreachable!("Unhandled child element in Audio: {child:?}");
+					},
+				}
+			},
+			ElementReaderYield::Eof => break,
+			_ => {
+				unreachable!("Unhandled child element in Audio: {child:?}");
+			},
+		}
+	}
+
+	Ok(())
+}

lofty/src/properties/tests.rs

@@ -3,7 +3,7 @@ use crate::ape::{ApeFile, ApeProperties};
 use crate::config::ParseOptions;
 use crate::ebml::{
 	AudioTrackDescriptor, AudioTrackSettings, EbmlFile, EbmlHeaderProperties, EbmlProperties,
-	SegmentInfo,
+	Language, SegmentInfo,
 };
 use crate::file::AudioFile;
 use crate::flac::{FlacFile, FlacProperties};
@@ -90,17 +90,20 @@ fn MKA_PROPERTIES() -> EbmlProperties {
 			writing_app: String::from("Lavf60.3.100"),
 		},
 		audio_tracks: vec![AudioTrackDescriptor {
-			number: 0,
-			uid: 0,
-			language: String::new(),
+			number: 1,
+			uid: 18181673715630629642,
+			enabled: true,
+			default: false,
+			language: Language::Iso639_2(String::from("und")),
 			default_duration: 0,
-			codec_id: String::new(),
-			codec_private: vec![],
+			codec_id: String::from("A_VORBIS"),
+			codec_private: None,
+			codec_name: None,
 			settings: AudioTrackSettings {
-				sampling_frequency: 0,
-				output_sampling_frequency: 0,
-				channels: 0,
-				bit_depth: None,
+				sampling_frequency: 48000.0,
+				output_sampling_frequency: 0.0,
+				channels: 2,
+				bit_depth: Some(32),
 				emphasis: None,
 			},
 		}],