diff options
Diffstat (limited to 'remuxer')
| -rw-r--r-- | remuxer/Cargo.toml | 4 | ||||
| -rw-r--r-- | remuxer/src/extract.rs | 17 | ||||
| -rw-r--r-- | remuxer/src/fragment.rs | 109 | ||||
| -rw-r--r-- | remuxer/src/lib.rs | 66 | ||||
| -rw-r--r-- | remuxer/src/matroska_to_mpeg4.rs | 36 | ||||
| -rw-r--r-- | remuxer/src/matroska_to_webm.rs | 84 | ||||
| -rw-r--r-- | remuxer/src/metadata.rs | 123 | ||||
| -rw-r--r-- | remuxer/src/remux.rs | 572 | ||||
| -rw-r--r-- | remuxer/src/seek_index.rs | 33 |
9 files changed, 642 insertions, 402 deletions
diff --git a/remuxer/Cargo.toml b/remuxer/Cargo.toml index 2313dcc..16713df 100644 --- a/remuxer/Cargo.toml +++ b/remuxer/Cargo.toml @@ -13,3 +13,7 @@ log = { workspace = true } serde = { version = "1.0.217", features = ["derive"] } bincode = { version = "2.0.0-rc.3", features = ["serde"] } + +ebml-struct = { git = "https://codeberg.org/metamuffin/ebml-struct", features = [ + "bincode", +] } diff --git a/remuxer/src/extract.rs b/remuxer/src/extract.rs index 12e4003..15c1e9d 100644 --- a/remuxer/src/extract.rs +++ b/remuxer/src/extract.rs @@ -5,29 +5,22 @@ */ use crate::seek_index::get_seek_index; use anyhow::{anyhow, bail}; -use jellybase::common::LocalTrack; use jellymatroska::{block::Block, read::EbmlReader, Master, MatroskaTag}; use log::debug; use std::{fs::File, io::BufReader, path::PathBuf}; pub type TrackExtract = Vec<(u64, Option<u64>, Vec<u8>)>; -pub fn extract_track(path_base: PathBuf, track_info: LocalTrack) -> anyhow::Result<TrackExtract> { - let source_path = path_base.join(track_info.path); - let file = File::open(&source_path)?; +pub fn extract_track(path: PathBuf, track: u64) -> anyhow::Result<TrackExtract> { + let file = File::open(&path)?; let mut reader = EbmlReader::new(BufReader::new(file)); - let index = get_seek_index(&source_path)?; - let index = index - .get(&(track_info.track as u64)) - .ok_or(anyhow!("track missing"))?; + let index = get_seek_index(&path)?; + let index = index.get(&track).ok_or(anyhow!("track missing"))?; let mut out = Vec::new(); for b in &index.blocks { reader.seek(b.source_off, MatroskaTag::BlockGroup(Master::Start))?; let (duration, block) = read_group(&mut reader)?; - assert_eq!( - track_info.track, block.track as usize, - "seek index is wrong" - ); + assert_eq!(track, block.track, "seek index is wrong"); out.push((b.pts, duration, block.data)) } Ok(out) diff --git a/remuxer/src/fragment.rs b/remuxer/src/fragment.rs index 9fa68f3..0da1ed5 100644 --- a/remuxer/src/fragment.rs +++ b/remuxer/src/fragment.rs @@ -5,11 +5,12 @@ */ use crate::{ - ebml_header, ebml_segment_info, ebml_track_entry, seek_index::get_seek_index, + ebml_header, ebml_segment_info, ebml_track_entry, + metadata::{matroska_metadata, MatroskaMetadata}, + seek_index::get_seek_index, segment_extractor::SegmentExtractIter, }; use anyhow::{anyhow, Context, Result}; -use jellybase::common::{LocalTrack, Node, SourceTrackKind}; use jellymatroska::{read::EbmlReader, write::EbmlWriter, Master, MatroskaTag}; use log::{debug, info}; use std::{ @@ -19,34 +20,35 @@ use std::{ path::Path, }; -const FRAGMENT_LENGTH: f64 = 2.; +const FRAGMENT_LENGTH: f64 = 5.; -pub fn fragment_index( - path_base: &Path, - item: &Node, - local_track: &LocalTrack, - track_index: usize, -) -> Result<Vec<Range<f64>>> { - let media_info = item.media.as_ref().unwrap(); - let source_path = path_base.join(&local_track.path); - let index = get_seek_index(&source_path)?; +pub fn fragment_index(path: &Path, track: u64) -> Result<Vec<Range<f64>>> { + let meta = matroska_metadata(path)?; + let duration = media_duration(&meta); + let force_kf = meta + .as_ref() + .tracks + .as_ref() + .unwrap() + .entries + .iter() + .find(|t| t.track_number == track) + .unwrap() + .track_type + == 17; + + let index = get_seek_index(&path)?; let index = index - .get(&(local_track.track as u64)) + .get(&track) .ok_or(anyhow!("seek index track missing"))?; - // everything is a keyframe (even though nothing is...) - let force_kf = matches!( - media_info.tracks[track_index].kind, - SourceTrackKind::Subtitles { .. } - ); - let n_kf = if force_kf { index.blocks.len() } else { index.keyframes.len() }; - let average_kf_interval = media_info.duration / n_kf as f64; + let average_kf_interval = duration / n_kf as f64; let kf_per_frag = (FRAGMENT_LENGTH / average_kf_interval).ceil() as usize; debug!("average keyframe interval: {average_kf_interval}"); debug!(" => keyframes per frag {kf_per_frag}"); @@ -72,7 +74,7 @@ pub fn fragment_index( index.keyframes.get((i + 1) * kf_per_frag).copied() } .map(|i| index.blocks[i].pts as f64 / 1000.) - .unwrap_or(media_info.duration); + .unwrap_or(duration); start..end }) .collect()) @@ -80,45 +82,46 @@ pub fn fragment_index( pub fn write_fragment_into( writer: impl Write, - path_base: &Path, - item: &Node, - local_track: &LocalTrack, - track: usize, + path: &Path, + track: u64, webm: bool, + title: &str, n: usize, ) -> anyhow::Result<()> { - info!("writing fragment {n} of {:?} (track {track})", item.title); - let mut output = EbmlWriter::new(BufWriter::new(writer), 0); - let media_info = item.media.as_ref().unwrap(); - let info = media_info + let meta = matroska_metadata(path)?; + let duration = media_duration(&meta); + let track_meta = meta + .as_ref() .tracks - .get(track) - .ok_or(anyhow!("track not available"))? - .to_owned(); - let source_path = path_base.join(&local_track.path); + .as_ref() + .unwrap() + .entries + .iter() + .find(|t| t.track_number == track) + .unwrap(); + let force_kf = track_meta.track_type == 17; + + info!("writing fragment {n} of {:?} (track {track})", title); + let mut output = EbmlWriter::new(BufWriter::new(writer), 0); let mapped = 1; - info!( - "\t- {track} {source_path:?} ({} => {mapped})", - local_track.track - ); - info!("\t {}", info); - let file = File::open(&source_path).context("opening source file")?; - let index = get_seek_index(&source_path)?; + info!("\t- {track} {path:?} ({} => {mapped})", track); + // info!("\t {}", info); + let file = File::open(&path).context("opening source file")?; + let index = get_seek_index(&path)?; let index = index - .get(&(local_track.track as u64)) + .get(&track) .ok_or(anyhow!("track missing 2"))? .to_owned(); debug!("\t seek index: {} blocks loaded", index.blocks.len()); let mut reader = EbmlReader::new(BufReader::new(file)); - let force_kf = matches!(info.kind, SourceTrackKind::Subtitles { .. }); let n_kf = if force_kf { index.blocks.len() } else { index.keyframes.len() }; - - let average_kf_interval = media_info.duration / n_kf as f64; + debug!("{duration} {n_kf}"); + let average_kf_interval = duration / n_kf as f64; let kf_per_frag = (FRAGMENT_LENGTH / average_kf_interval).ceil() as usize; debug!("average keyframe interval: {average_kf_interval}"); debug!(" => keyframes per frag {kf_per_frag}"); @@ -144,25 +147,20 @@ pub fn write_fragment_into( .blocks .get(end_block_index) .map(|b| b.pts) - .unwrap_or((media_info.duration * 1000.) as u64); + .unwrap_or((duration * 1000.) as u64); output.write_tag(&ebml_header(webm))?; output.write_tag(&MatroskaTag::Segment(Master::Start))?; output.write_tag(&ebml_segment_info( - format!("{}: {info}", item.title.clone().unwrap_or_default()), + title.to_string(), (last_block_pts - start_block.pts) as f64 / 1000., ))?; output.write_tag(&MatroskaTag::Tracks(Master::Collected(vec![ - ebml_track_entry( - mapped, - local_track.track as u64 * 100, // TODO something else that is unique to the track - &info, - local_track.codec_private.clone(), - ), + ebml_track_entry(mapped, track_meta), ])))?; reader.seek(start_block.source_off, MatroskaTag::Cluster(Master::Start))?; - let mut reader = SegmentExtractIter::new(&mut reader, local_track.track as u64); + let mut reader = SegmentExtractIter::new(&mut reader, track); { // TODO this one caused fragments to get dropped by MSE for no reason @@ -214,3 +212,8 @@ pub fn write_fragment_into( debug!("wrote {} bytes", output.position()); Ok(()) } + +fn media_duration(m: &MatroskaMetadata) -> f64 { + let info = m.info.as_ref().unwrap(); + (info.duration.unwrap_or_default() * info.timestamp_scale as f64) / 1_000_000_000. +} diff --git a/remuxer/src/lib.rs b/remuxer/src/lib.rs index a98ffad..931d5e6 100644 --- a/remuxer/src/lib.rs +++ b/remuxer/src/lib.rs @@ -3,18 +3,22 @@ which is licensed under the GNU Affero General Public License (version 3); see /COPYING. Copyright (C) 2025 metamuffin <metamuffin.org> */ +#![feature(random, exit_status_error)] pub mod extract; pub mod fragment; +pub mod metadata; +pub mod matroska_to_mpeg4; pub mod remux; pub mod seek_index; pub mod segment_extractor; pub mod trim_writer; +pub mod matroska_to_webm; +use ebml_struct::matroska::TrackEntry; pub use fragment::write_fragment_into; -pub use remux::remux_stream_into; - -use jellybase::common::{SourceTrack, SourceTrackKind}; use jellymatroska::{Master, MatroskaTag}; +pub use matroska_to_mpeg4::matroska_to_mpeg4; +pub use remux::remux_stream_into; pub fn ebml_header(webm: bool) -> MatroskaTag { MatroskaTag::Ebml(Master::Collected(vec![ @@ -41,66 +45,56 @@ pub fn ebml_segment_info(title: String, duration: f64) -> MatroskaTag { ])) } -pub fn ebml_track_entry( - number: u64, - uid: u64, - track: &SourceTrack, - codec_private: Option<Vec<u8>>, -) -> MatroskaTag { +pub fn ebml_track_entry(number: u64, track: &TrackEntry) -> MatroskaTag { let mut els = vec![ MatroskaTag::TrackNumber(number), - MatroskaTag::TrackUID(uid), MatroskaTag::FlagLacing(track.flag_lacing), MatroskaTag::Language(track.language.clone()), - MatroskaTag::CodecID(track.codec.clone()), + MatroskaTag::CodecID(track.codec_id.clone()), MatroskaTag::CodecDelay(track.codec_delay), MatroskaTag::SeekPreRoll(track.seek_pre_roll), ]; if let Some(d) = &track.default_duration { els.push(MatroskaTag::DefaultDuration(*d)); } - match track.kind { - SourceTrackKind::Video { - width, - height, - display_height, - display_width, - display_unit, - fps, - } => { + match track.track_type { + 1 => { + let video = track.video.as_ref().unwrap(); els.push(MatroskaTag::TrackType(1)); let mut props = vec![ - MatroskaTag::PixelWidth(width), - MatroskaTag::PixelHeight(height), + MatroskaTag::PixelWidth(video.pixel_width), + MatroskaTag::PixelHeight(video.pixel_height), ]; - props.push(MatroskaTag::DisplayWidth(display_width.unwrap_or(width))); - props.push(MatroskaTag::DisplayHeight(display_height.unwrap_or(height))); - props.push(MatroskaTag::DisplayUnit(display_unit)); - if let Some(fps) = fps { + props.push(MatroskaTag::DisplayWidth( + video.display_width.unwrap_or(video.pixel_width), + )); + props.push(MatroskaTag::DisplayHeight( + video.display_height.unwrap_or(video.pixel_height), + )); + props.push(MatroskaTag::DisplayUnit(video.display_unit)); + if let Some(fps) = video.frame_rate { props.push(MatroskaTag::FrameRate(fps)) } els.push(MatroskaTag::Video(Master::Collected(props))) } - SourceTrackKind::Audio { - channels, - sample_rate, - bit_depth, - } => { + 2 => { + let audio = track.audio.as_ref().unwrap(); els.push(MatroskaTag::TrackType(2)); let mut props = vec![ - MatroskaTag::SamplingFrequency(sample_rate), - MatroskaTag::Channels(channels.try_into().unwrap()), + MatroskaTag::SamplingFrequency(audio.sampling_frequency), + MatroskaTag::Channels(audio.channels), ]; - if let Some(bit_depth) = bit_depth { + if let Some(bit_depth) = audio.bit_depth { props.push(MatroskaTag::BitDepth(bit_depth.try_into().unwrap())); } els.push(MatroskaTag::Audio(Master::Collected(props))); } - SourceTrackKind::Subtitles => { + 17 => { els.push(MatroskaTag::TrackType(17)); } + _ => unreachable!(), } - if let Some(d) = &codec_private { + if let Some(d) = &track.codec_private { els.push(MatroskaTag::CodecPrivate(d.clone())); } MatroskaTag::TrackEntry(Master::Collected(els)) diff --git a/remuxer/src/matroska_to_mpeg4.rs b/remuxer/src/matroska_to_mpeg4.rs new file mode 100644 index 0000000..e8268e7 --- /dev/null +++ b/remuxer/src/matroska_to_mpeg4.rs @@ -0,0 +1,36 @@ +/* + This file is part of jellything (https://codeberg.org/metamuffin/jellything) + which is licensed under the GNU Affero General Public License (version 3); see /COPYING. + Copyright (C) 2025 metamuffin <metamuffin.org> +*/ +use anyhow::Result; +use std::{ + fs::{remove_file, File}, + io::{copy, Read, Write}, + process::{Command, Stdio}, + random::random, +}; + +pub fn matroska_to_mpeg4( + mut input: impl Read + Send + 'static, + mut output: impl Write, +) -> Result<()> { + let path = format!("/tmp/jellything-tc-hack-{:016x}", random::<u64>()); + let args = format!( + "-hide_banner -loglevel warning -f matroska -i pipe:0 -copyts -c copy -f mp4 -movflags frag_keyframe+empty_moov {path}" + ); + let mut child = Command::new("ffmpeg") + .args(args.split(" ")) + .stdin(Stdio::piped()) + .stderr(Stdio::inherit()) + .spawn()?; + + let mut stdin = child.stdin.take().unwrap(); + copy(&mut input, &mut stdin)?; + drop(stdin); + child.wait()?.exit_ok()?; + copy(&mut File::open(&path)?, &mut output)?; + remove_file(path)?; + + Ok(()) +} diff --git a/remuxer/src/matroska_to_webm.rs b/remuxer/src/matroska_to_webm.rs new file mode 100644 index 0000000..b9a1819 --- /dev/null +++ b/remuxer/src/matroska_to_webm.rs @@ -0,0 +1,84 @@ +use crate::ebml_track_entry; +use anyhow::Context; +use ebml_struct::{ + ids::*, + matroska::{Cluster, Ebml, Info, Tracks}, + read::{EbmlReadExt, TagRead}, + write::TagWrite, +}; +use jellymatroska::{read::EbmlReader, write::EbmlWriter, Master, MatroskaTag}; +use log::warn; +use std::io::{BufReader, BufWriter, ErrorKind, Read, Seek, Write}; + +pub fn matroska_to_webm( + input: impl Read + Seek + 'static, + output: impl Write, +) -> anyhow::Result<()> { + let mut output = EbmlWriter::new(BufWriter::new(output), 0); + let mut input = EbmlReader::new(BufReader::new(input)); + + Ebml { + ebml_version: 1, + ebml_read_version: 1, + ebml_max_id_length: 4, + ebml_max_size_length: 8, + doc_type: "webm".to_string(), + doc_type_version: 4, + doc_type_read_version: 2, + doc_type_extensions: vec![], + } + .write(&mut output)?; + output.write_tag(&MatroskaTag::Segment(Master::Start))?; + + let (x, mut ebml) = input.read_tag()?; + assert_eq!(x, EL_EBML); + let ebml = Ebml::read(&mut ebml).unwrap(); + assert!(ebml.doc_type == "matroska" || ebml.doc_type == "webm"); + let (x, mut segment) = input.read_tag()?; + assert_eq!(x, EL_SEGMENT); + + loop { + let (x, mut seg) = match segment.read_tag() { + Ok(o) => o, + Err(e) if e.kind() == ErrorKind::UnexpectedEof => break, + Err(e) => return Err(e.into()), + }; + match x { + EL_INFO => { + let info = Info::read(&mut seg).context("info")?; + output.write_tag(&{ + MatroskaTag::Info(Master::Collected(vec![ + MatroskaTag::TimestampScale(info.timestamp_scale), + MatroskaTag::Duration(info.duration.unwrap_or_default()), + MatroskaTag::Title(info.title.unwrap_or_default()), + MatroskaTag::MuxingApp("jellyremux".to_string()), + MatroskaTag::WritingApp("jellything".to_string()), + ])) + })?; + } + EL_TRACKS => { + let tracks = Tracks::read(&mut seg).context("tracks")?; + output.write_tag(&MatroskaTag::Tracks(Master::Collected( + tracks + .entries + .into_iter() + .map(|t| ebml_track_entry(t.track_number, &t)) + .collect(), + )))?; + } + EL_VOID | EL_CRC32 | EL_CUES | EL_SEEKHEAD | EL_ATTACHMENTS | EL_TAGS => { + seg.consume()?; + } + EL_CLUSTER => { + let cluster = Cluster::read(&mut seg).context("cluster")?; + // TODO mixing both ebml libraries :))) + cluster.write(&mut output)?; + } + id => { + warn!("unknown top-level element {id:x}"); + seg.consume()?; + } + } + } + Ok(()) +} diff --git a/remuxer/src/metadata.rs b/remuxer/src/metadata.rs new file mode 100644 index 0000000..c8a5f8f --- /dev/null +++ b/remuxer/src/metadata.rs @@ -0,0 +1,123 @@ +/* + This file is part of jellything (https://codeberg.org/metamuffin/jellything) + which is licensed under the GNU Affero General Public License (version 3); see /COPYING. + Copyright (C) 2025 metamuffin <metamuffin.org> +*/ +use anyhow::{Context, Result}; +use bincode::{Decode, Encode}; +use ebml_struct::{ + ids::*, + matroska::*, + read::{EbmlReadExt, TagRead}, +}; +use jellybase::{ + assetfed::AssetInner, + cache::{cache_file, cache_memory}, + common::Asset, +}; +use log::{info, warn}; +use std::{ + fs::File, + io::{BufReader, ErrorKind, Read, Write}, + path::Path, + sync::Arc, +}; + +#[derive(Debug, Encode, Decode, Clone)] +pub struct MatroskaMetadata { + pub info: Option<Info>, + pub tracks: Option<Tracks>, + pub cover: Option<Asset>, + pub chapters: Option<Chapters>, + pub tags: Option<Tags>, + pub infojson: Option<Vec<u8>>, +} +pub fn checked_matroska_metadata(path: &Path) -> Result<Arc<Option<MatroskaMetadata>>> { + cache_memory( + &["mkmeta-check-v1", path.to_string_lossy().as_ref()], + || { + let mut magic = [0; 4]; + File::open(path)?.read_exact(&mut magic).ok(); + if !matches!(magic, [0x1A, 0x45, 0xDF, 0xA3]) { + return Ok(None); + } + Ok(Some((*matroska_metadata(path)?).clone())) + }, + ) +} +pub fn matroska_metadata(path: &Path) -> Result<Arc<MatroskaMetadata>> { + cache_memory(&["mkmeta-v3", path.to_string_lossy().as_ref()], || { + info!("reading {path:?}"); + let mut file = BufReader::new(File::open(path)?); + let mut file = file.by_ref().take(u64::MAX); + + let (x, mut ebml) = file.read_tag()?; + assert_eq!(x, EL_EBML); + let ebml = Ebml::read(&mut ebml).unwrap(); + assert!(ebml.doc_type == "matroska" || ebml.doc_type == "webm"); + let (x, mut segment) = file.read_tag()?; + assert_eq!(x, EL_SEGMENT); + + let mut info = None; + let mut infojson = None; + let mut tracks = None; + let mut cover = None; + let mut chapters = None; + let mut tags = None; + loop { + let (x, mut seg) = match segment.read_tag() { + Ok(o) => o, + Err(e) if e.kind() == ErrorKind::UnexpectedEof => break, + Err(e) => return Err(e.into()), + }; + match x { + EL_INFO => info = Some(Info::read(&mut seg).context("info")?), + EL_TRACKS => tracks = Some(Tracks::read(&mut seg).context("tracks")?), + EL_CHAPTERS => chapters = Some(Chapters::read(&mut seg).context("chapters")?), + EL_TAGS => tags = Some(Tags::read(&mut seg).context("tags")?), + EL_ATTACHMENTS => { + let attachments = Attachments::read(&mut seg).context("attachments")?; + for f in attachments.files { + match f.name.as_str() { + "info.json" => { + infojson = Some(f.data); + } + "cover.webp" | "cover.png" | "cover.jpg" | "cover.jpeg" + | "cover.avif" => { + cover = Some( + AssetInner::Cache(cache_file( + &["att-cover", path.to_string_lossy().as_ref()], + move |mut file| { + file.write_all(&f.data)?; + Ok(()) + }, + )?) + .ser(), + ) + } + _ => (), + } + } + } + EL_VOID | EL_CRC32 | EL_CUES | EL_SEEKHEAD => { + seg.consume()?; + } + EL_CLUSTER => { + break; + } + id => { + warn!("unknown top-level element {id:x}"); + seg.consume()?; + } + } + } + Ok(MatroskaMetadata { + chapters, + cover, + info, + infojson, + tags, + tracks, + }) + }) +} diff --git a/remuxer/src/remux.rs b/remuxer/src/remux.rs index 0507f1e..a44c58b 100644 --- a/remuxer/src/remux.rs +++ b/remuxer/src/remux.rs @@ -3,333 +3,311 @@ which is licensed under the GNU Affero General Public License (version 3); see /COPYING. Copyright (C) 2025 metamuffin <metamuffin.org> */ -use crate::{ - ebml_header, ebml_track_entry, seek_index::get_seek_index, - segment_extractor::SegmentExtractIter, trim_writer::TrimWriter, -}; -use anyhow::{anyhow, Context}; -use jellybase::common::{ - seek_index::{BlockIndex, SeekIndex}, - LocalTrack, Node, SourceTrack, -}; -use jellymatroska::{ - read::EbmlReader, - write::{bad_vint_length, vint_length, EbmlWriter}, - Master, MatroskaTag, -}; -use log::{debug, info, trace, warn}; -use std::{ - fs::File, - io::{BufReader, BufWriter, Seek, SeekFrom, Write}, - ops::Range, - path::PathBuf, - sync::Arc, - time::Instant, -}; +use jellybase::common::Node; +use std::{io::Write, ops::Range, path::PathBuf}; -struct ClusterLayout { - position: usize, - timestamp: u64, - source_offsets: Vec<Option<u64>>, - blocks: Vec<(usize, BlockIndex)>, -} +// struct ClusterLayout { +// position: usize, +// timestamp: u64, +// source_offsets: Vec<Option<u64>>, +// blocks: Vec<(usize, BlockIndex)>, +// } pub fn remux_stream_into( - writer: impl Write, - range: Range<usize>, - path_base: PathBuf, - item: &Node, - track_sources: Vec<LocalTrack>, - selection: Vec<usize>, - webm: bool, + _writer: impl Write, + _range: Range<usize>, + _path_base: PathBuf, + _item: &Node, + _selection: Vec<usize>, + _webm: bool, ) -> anyhow::Result<()> { - info!("remuxing {:?} to have tracks {selection:?}", item.title); - let writer = TrimWriter::new(BufWriter::new(writer), range.clone()); - let mut output = EbmlWriter::new(writer, 0); + // info!("remuxing {:?} to have tracks {selection:?}", item.title); + // let writer = TrimWriter::new(BufWriter::new(writer), range.clone()); + // let mut output = EbmlWriter::new(writer, 0); - struct ReaderC { - info: SourceTrack, - reader: EbmlReader, - mapped: u64, - index: Arc<SeekIndex>, - source_track_index: usize, - codec_private: Option<Vec<u8>>, - layouting_progress_index: usize, - } + // struct ReaderC { + // info: SourceTrack, + // reader: EbmlReader, + // mapped: u64, + // index: Arc<SeekIndex>, + // source_track_index: usize, + // codec_private: Option<Vec<u8>>, + // layouting_progress_index: usize, + // } - let timing_cp = Instant::now(); + // let timing_cp = Instant::now(); - let mut inputs = selection - .iter() - .enumerate() - .map(|(index, sel)| { - let info = item - .media - .as_ref() - .unwrap() - .tracks - .get(*sel) - .ok_or(anyhow!("track not available"))? - .to_owned(); - let private = &track_sources[index]; - let source_path = path_base.join(&private.path); - let mapped = index as u64 + 1; - info!("\t- {sel} {source_path:?} ({} => {mapped})", private.track); - info!("\t {}", info); - let file = File::open(&source_path).context("opening source file")?; - let index = get_seek_index(&source_path)?; - let index = index - .get(&(private.track as u64)) - .ok_or(anyhow!("track missing 3"))? - .to_owned(); - debug!("\t seek index: {} blocks loaded", index.blocks.len()); - let reader = EbmlReader::new(BufReader::new(file)); - Ok(ReaderC { - index, - reader, - info, - mapped, - source_track_index: private.track, - codec_private: private.codec_private.clone(), - layouting_progress_index: 0, - }) - }) - .collect::<anyhow::Result<Vec<_>>>()?; + // let mut inputs = selection + // .iter() + // .enumerate() + // .map(|(index, sel)| { + // let info = item + // .media + // .as_ref() + // .unwrap() + // .tracks + // .get(*sel) + // .ok_or(anyhow!("track not available"))? + // .to_owned(); + // let source_path = path_base.join(&private.path); + // let mapped = index as u64 + 1; + // info!("\t- {sel} {source_path:?} ({} => {mapped})", private.track); + // info!("\t {}", info); + // let file = File::open(&source_path).context("opening source file")?; + // let index = get_seek_index(&source_path)?; + // let index = index + // .get(&(private.track as u64)) + // .ok_or(anyhow!("track missing 3"))? + // .to_owned(); + // debug!("\t seek index: {} blocks loaded", index.blocks.len()); + // let reader = EbmlReader::new(BufReader::new(file)); + // Ok(ReaderC { + // index, + // reader, + // info, + // mapped, + // source_track_index: private.track, + // codec_private: private.codec_private.clone(), + // layouting_progress_index: 0, + // }) + // }) + // .collect::<anyhow::Result<Vec<_>>>()?; - info!("(perf) prepare inputs: {:?}", Instant::now() - timing_cp); - let timing_cp = Instant::now(); + // info!("(perf) prepare inputs: {:?}", Instant::now() - timing_cp); + // let timing_cp = Instant::now(); - output.write_tag(&ebml_header(webm))?; + // output.write_tag(&ebml_header(webm))?; - output.write_tag(&MatroskaTag::Segment(Master::Start))?; - let segment_offset = output.position(); + // output.write_tag(&MatroskaTag::Segment(Master::Start))?; + // let segment_offset = output.position(); - output.write_tag(&MatroskaTag::Info(Master::Collected(vec![ - MatroskaTag::TimestampScale(1_000_000), - MatroskaTag::Duration(item.media.as_ref().unwrap().duration * 1000.0), - MatroskaTag::Title(item.title.clone().unwrap_or_default()), - MatroskaTag::MuxingApp("jellyremux".to_string()), - MatroskaTag::WritingApp("jellything".to_string()), - ])))?; + // output.write_tag(&MatroskaTag::Info(Master::Collected(vec![ + // MatroskaTag::TimestampScale(1_000_000), + // MatroskaTag::Duration(item.media.as_ref().unwrap().duration * 1000.0), + // MatroskaTag::Title(item.title.clone().unwrap_or_default()), + // MatroskaTag::MuxingApp("jellyremux".to_string()), + // MatroskaTag::WritingApp("jellything".to_string()), + // ])))?; - let tracks_header = inputs - .iter_mut() - .map(|rc| ebml_track_entry(rc.mapped, rc.mapped, &rc.info, rc.codec_private.take())) - .collect(); - output.write_tag(&MatroskaTag::Tracks(Master::Collected(tracks_header)))?; + // let tracks_header = inputs + // .iter_mut() + // .map(|rc| ebml_track_entry(rc.mapped, rc.mapped, &rc.info, rc.codec_private.take())) + // .collect(); + // output.write_tag(&MatroskaTag::Tracks(Master::Collected(tracks_header)))?; - let mut segment_layout: Vec<ClusterLayout> = { - let mut cluster_pts = 0; - let mut clusters = vec![]; - let mut cluster = vec![]; - let mut source_offsets = vec![None; inputs.len()]; - let mut gp = 0usize; // cluster position (in the segment) - let mut p = 0usize; // block position (in the cluster) - loop { - let (track, block) = { - let mut best_block = BlockIndex { - pts: u64::MAX, - size: 0, - source_off: 0, - }; - let mut best_track = 0; - for (i, r) in inputs.iter().enumerate() { - if let Some(v) = r.index.blocks.get(r.layouting_progress_index) { - if v.pts < best_block.pts { - best_block = v.to_owned(); - best_track = i; - } - }; - } - (best_track, best_block) - }; - inputs[track].layouting_progress_index += 1; - source_offsets[track].get_or_insert(block.source_off); - if block.pts > cluster_pts + 1_000 { - let cluster_content_size = 1 + 1 // timestamp {tag, size} - + bad_vint_length(cluster_pts) // timestamp tag value - + p; - let cluster_size = 4 // tag length - + vint_length(cluster_content_size as u64) // size varint - + cluster_content_size; - clusters.push(ClusterLayout { - position: gp, // relative to the first cluster - timestamp: cluster_pts, - source_offsets, - blocks: std::mem::take(&mut cluster), - }); + // let mut segment_layout: Vec<ClusterLayout> = { + // let mut cluster_pts = 0; + // let mut clusters = vec![]; + // let mut cluster = vec![]; + // let mut source_offsets = vec![None; inputs.len()]; + // let mut gp = 0usize; // cluster position (in the segment) + // let mut p = 0usize; // block position (in the cluster) + // loop { + // let (track, block) = { + // let mut best_block = BlockIndex { + // pts: u64::MAX, + // size: 0, + // source_off: 0, + // }; + // let mut best_track = 0; + // for (i, r) in inputs.iter().enumerate() { + // if let Some(v) = r.index.blocks.get(r.layouting_progress_index) { + // if v.pts < best_block.pts { + // best_block = v.to_owned(); + // best_track = i; + // } + // }; + // } + // (best_track, best_block) + // }; + // inputs[track].layouting_progress_index += 1; + // source_offsets[track].get_or_insert(block.source_off); + // if block.pts > cluster_pts + 1_000 { + // let cluster_content_size = 1 + 1 // timestamp {tag, size} + // + bad_vint_length(cluster_pts) // timestamp tag value + // + p; + // let cluster_size = 4 // tag length + // + vint_length(cluster_content_size as u64) // size varint + // + cluster_content_size; + // clusters.push(ClusterLayout { + // position: gp, // relative to the first cluster + // timestamp: cluster_pts, + // source_offsets, + // blocks: std::mem::take(&mut cluster), + // }); - cluster_pts = block.pts; - source_offsets = vec![None; inputs.len()]; - gp += cluster_size; - p = 0; - } - if block.pts == u64::MAX { - break; - } + // cluster_pts = block.pts; + // source_offsets = vec![None; inputs.len()]; + // gp += cluster_size; + // p = 0; + // } + // if block.pts == u64::MAX { + // break; + // } - let simpleblock_size = 1 + 2 + 1 // block {tracknum, pts_off, flags} - // TODO does not work, if more than 127 tracks are present - + block.size; // block payload - p += 1; // simpleblock tag - p += vint_length(simpleblock_size as u64); // simpleblock size vint - p += simpleblock_size; + // let simpleblock_size = 1 + 2 + 1 // block {tracknum, pts_off, flags} + // // TODO does not work, if more than 127 tracks are present + // + block.size; // block payload + // p += 1; // simpleblock tag + // p += vint_length(simpleblock_size as u64); // simpleblock size vint + // p += simpleblock_size; - cluster.push((track, block)) - } - info!("segment layout computed ({} clusters)", clusters.len()); - clusters - }; - info!( - "(perf) compute segment layout: {:?}", - Instant::now() - timing_cp - ); - let timing_cp = Instant::now(); + // cluster.push((track, block)) + // } + // info!("segment layout computed ({} clusters)", clusters.len()); + // clusters + // }; + // info!( + // "(perf) compute segment layout: {:?}", + // Instant::now() - timing_cp + // ); + // let timing_cp = Instant::now(); - let max_cue_size = 4 // cues id - + 8 // cues len - + ( // cues content - 1 // cp id - + 1 // cp len - + ( // cp content - 1 // ctime id, - + 1 // ctime len - + 8 // ctime content uint - + ( // ctps - 1 // ctp id - + 8 // ctp len - + (// ctp content - 1 // ctrack id - + 1 // ctrack size - + 1 // ctrack content int - // TODO this breaks if inputs.len() >= 127 - + 1 // ccp id - + 1 // ccp len - + 8 // ccp content offset - ) - ) - ) * inputs.len() - ) * segment_layout.len() - + 1 // void id - + 8; // void len + // let max_cue_size = 4 // cues id + // + 8 // cues len + // + ( // cues content + // 1 // cp id + // + 1 // cp len + // + ( // cp content + // 1 // ctime id, + // + 1 // ctime len + // + 8 // ctime content uint + // + ( // ctps + // 1 // ctp id + // + 8 // ctp len + // + (// ctp content + // 1 // ctrack id + // + 1 // ctrack size + // + 1 // ctrack content int + // // TODO this breaks if inputs.len() >= 127 + // + 1 // ccp id + // + 1 // ccp len + // + 8 // ccp content offset + // ) + // ) + // ) * inputs.len() + // ) * segment_layout.len() + // + 1 // void id + // + 8; // void len - let first_cluster_offset_predict = max_cue_size + output.position(); + // let first_cluster_offset_predict = max_cue_size + output.position(); - // make the cluster position relative to the segment start as they should - segment_layout - .iter_mut() - .for_each(|e| e.position += first_cluster_offset_predict - segment_offset); + // // make the cluster position relative to the segment start as they should + // segment_layout + // .iter_mut() + // .for_each(|e| e.position += first_cluster_offset_predict - segment_offset); - output.write_tag(&MatroskaTag::Cues(Master::Collected( - segment_layout - .iter() - .map(|cluster| { - MatroskaTag::CuePoint(Master::Collected( - Some(MatroskaTag::CueTime(cluster.timestamp)) - .into_iter() - // TODO: Subtitles should not have cues for every cluster - .chain(inputs.iter().map(|i| { - MatroskaTag::CueTrackPositions(Master::Collected(vec![ - MatroskaTag::CueTrack(i.mapped), - MatroskaTag::CueClusterPosition(cluster.position as u64), - ])) - })) - .collect(), - )) - }) - .collect(), - )))?; - output.write_padding(first_cluster_offset_predict)?; - let first_cluster_offset = output.position(); - assert_eq!(first_cluster_offset, first_cluster_offset_predict); + // output.write_tag(&MatroskaTag::Cues(Master::Collected( + // segment_layout + // .iter() + // .map(|cluster| { + // MatroskaTag::CuePoint(Master::Collected( + // Some(MatroskaTag::CueTime(cluster.timestamp)) + // .into_iter() + // // TODO: Subtitles should not have cues for every cluster + // .chain(inputs.iter().map(|i| { + // MatroskaTag::CueTrackPositions(Master::Collected(vec![ + // MatroskaTag::CueTrack(i.mapped), + // MatroskaTag::CueClusterPosition(cluster.position as u64), + // ])) + // })) + // .collect(), + // )) + // }) + // .collect(), + // )))?; + // output.write_padding(first_cluster_offset_predict)?; + // let first_cluster_offset = output.position(); + // assert_eq!(first_cluster_offset, first_cluster_offset_predict); - let mut skip = 0; - // TODO binary search - for (i, cluster) in segment_layout.iter().enumerate() { - if (cluster.position + segment_offset) >= range.start { - break; - } - skip = i; - } + // let mut skip = 0; + // // TODO binary search + // for (i, cluster) in segment_layout.iter().enumerate() { + // if (cluster.position + segment_offset) >= range.start { + // break; + // } + // skip = i; + // } - if skip != 0 { - info!("skipping {skip} clusters"); - output.seek(SeekFrom::Start( - (segment_layout[skip].position + segment_offset) as u64, - ))?; - } + // if skip != 0 { + // info!("skipping {skip} clusters"); + // output.seek(SeekFrom::Start( + // (segment_layout[skip].position + segment_offset) as u64, + // ))?; + // } - struct ReaderD<'a> { - stream: SegmentExtractIter<'a>, - mapped: u64, - } + // struct ReaderD<'a> { + // stream: SegmentExtractIter<'a>, + // mapped: u64, + // } - let mut track_readers = inputs - .iter_mut() - .enumerate() - .map(|(i, inp)| { - inp.reader - .seek( - // the seek target might be a hole; we continue until the next cluster of that track. - // this should be fine since tracks are only read according to segment_layout - find_first_cluster_with_off(&segment_layout, skip, i) - .ok_or(anyhow!("cluster hole at eof"))?, - MatroskaTag::Cluster(Master::Start), // TODO shouldn't this be a child of cluster? - ) - .context("seeking in input")?; - let stream = SegmentExtractIter::new(&mut inp.reader, inp.source_track_index as u64); + // let mut track_readers = inputs + // .iter_mut() + // .enumerate() + // .map(|(i, inp)| { + // inp.reader + // .seek( + // // the seek target might be a hole; we continue until the next cluster of that track. + // // this should be fine since tracks are only read according to segment_layout + // find_first_cluster_with_off(&segment_layout, skip, i) + // .ok_or(anyhow!("cluster hole at eof"))?, + // MatroskaTag::Cluster(Master::Start), // TODO shouldn't this be a child of cluster? + // ) + // .context("seeking in input")?; + // let stream = SegmentExtractIter::new(&mut inp.reader, inp.source_track_index as u64); - Ok(ReaderD { - mapped: inp.mapped, - stream, - }) - }) - .collect::<anyhow::Result<Vec<_>>>()?; + // Ok(ReaderD { + // mapped: inp.mapped, + // stream, + // }) + // }) + // .collect::<anyhow::Result<Vec<_>>>()?; - info!("(perf) seek inputs: {:?}", Instant::now() - timing_cp); + // info!("(perf) seek inputs: {:?}", Instant::now() - timing_cp); - for (cluster_index, cluster) in segment_layout.into_iter().enumerate().skip(skip) { - debug!( - "writing cluster {cluster_index} (pts_base={}) with {} blocks", - cluster.timestamp, - cluster.blocks.len() - ); - { - let cue_error = cluster.position as i64 - (output.position() - segment_offset) as i64; - if cue_error != 0 { - warn!("calculation was {} bytes off", cue_error); - } - } + // for (cluster_index, cluster) in segment_layout.into_iter().enumerate().skip(skip) { + // debug!( + // "writing cluster {cluster_index} (pts_base={}) with {} blocks", + // cluster.timestamp, + // cluster.blocks.len() + // ); + // { + // let cue_error = cluster.position as i64 - (output.position() - segment_offset) as i64; + // if cue_error != 0 { + // warn!("calculation was {} bytes off", cue_error); + // } + // } - let mut cluster_blocks = vec![MatroskaTag::Timestamp(cluster.timestamp)]; - for (block_track, index_block) in cluster.blocks { - let track_reader = &mut track_readers[block_track]; - // TODO handle duration - let mut block = track_reader.stream.next_block()?.0; + // let mut cluster_blocks = vec![MatroskaTag::Timestamp(cluster.timestamp)]; + // for (block_track, index_block) in cluster.blocks { + // let track_reader = &mut track_readers[block_track]; + // // TODO handle duration + // let mut block = track_reader.stream.next_block()?.0; - assert_eq!(index_block.size, block.data.len(), "seek index is wrong"); + // assert_eq!(index_block.size, block.data.len(), "seek index is wrong"); - block.track = track_reader.mapped; - block.timestamp_off = (index_block.pts - cluster.timestamp).try_into().unwrap(); - trace!("n={} tso={}", block.track, block.timestamp_off); + // block.track = track_reader.mapped; + // block.timestamp_off = (index_block.pts - cluster.timestamp).try_into().unwrap(); + // trace!("n={} tso={}", block.track, block.timestamp_off); - cluster_blocks.push(MatroskaTag::SimpleBlock(block)) - } - output.write_tag(&MatroskaTag::Cluster(Master::Collected(cluster_blocks)))?; - } - // output.write_tag(&MatroskaTag::Segment(Master::End))?; - Ok(()) + // cluster_blocks.push(MatroskaTag::SimpleBlock(block)) + // } + // output.write_tag(&MatroskaTag::Cluster(Master::Collected(cluster_blocks)))?; + // } + // // output.write_tag(&MatroskaTag::Segment(Master::End))?; + // Ok(()) + todo!() } -fn find_first_cluster_with_off( - segment_layout: &[ClusterLayout], - skip: usize, - track: usize, -) -> Option<u64> { - for cluster in segment_layout.iter().skip(skip) { - if let Some(off) = cluster.source_offsets[track] { - return Some(off); - } - } - None -} +// fn find_first_cluster_with_off( +// segment_layout: &[ClusterLayout], +// skip: usize, +// track: usize, +// ) -> Option<u64> { +// for cluster in segment_layout.iter().skip(skip) { +// if let Some(off) = cluster.source_offsets[track] { +// return Some(off); +// } +// } +// None +// } diff --git a/remuxer/src/seek_index.rs b/remuxer/src/seek_index.rs index bd351d9..7296d93 100644 --- a/remuxer/src/seek_index.rs +++ b/remuxer/src/seek_index.rs @@ -4,10 +4,8 @@ Copyright (C) 2025 metamuffin <metamuffin.org> */ use anyhow::{Context, Result}; -use jellybase::{ - cache::cache_memory, - common::seek_index::{BlockIndex, SeekIndex}, -}; +use bincode::{Decode, Encode}; +use jellybase::cache::cache_memory; use jellymatroska::{ block::Block, read::EbmlReader, @@ -17,6 +15,33 @@ use jellymatroska::{ use log::{debug, info, trace, warn}; use std::{collections::BTreeMap, fs::File, io::BufReader, path::Path, sync::Arc}; +pub const SEEK_INDEX_VERSION: u32 = 0x5eef1de4; + +#[derive(Debug, Clone, Decode, Encode)] +pub struct SeekIndex { + pub version: u32, + pub blocks: Vec<BlockIndex>, + pub keyframes: Vec<usize>, +} + +#[derive(Debug, Clone, Decode, Encode)] +pub struct BlockIndex { + pub pts: u64, + // pub duration: Option<u64>, + pub source_off: u64, // points to start of SimpleBlock or BlockGroup (not the Block inside it) + pub size: usize, +} + +impl Default for SeekIndex { + fn default() -> Self { + Self { + version: SEEK_INDEX_VERSION, + blocks: Vec::new(), + keyframes: Vec::new(), + } + } +} + pub fn get_seek_index(path: &Path) -> anyhow::Result<Arc<BTreeMap<u64, Arc<SeekIndex>>>> { cache_memory(&["seekindex", path.to_str().unwrap()], move || { info!("generating seek index for {path:?}"); |