M9.9: Matroska/WebM demuxer (WIP: EBML reader, metadata, cluster scan, packets, seek)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
Brandon Thomas
parent
3f3265bb35
commit
77a096884b
|
|
@ -1484,6 +1484,13 @@ dependencies = [
|
|||
"thiserror 2.0.21",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "filmcraft-matroska"
|
||||
version = "0.1.0"
|
||||
dependencies = [
|
||||
"serde_json",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "filmcraft-media"
|
||||
version = "0.1.0"
|
||||
|
|
|
|||
|
|
@ -17,6 +17,7 @@ filmcraft-geom = { path = "crates/geom" }
|
|||
filmcraft-color = { path = "crates/color" }
|
||||
filmcraft-bitstream = { path = "crates/bitstream" }
|
||||
filmcraft-isobmff = { path = "crates/isobmff" }
|
||||
filmcraft-matroska = { path = "crates/matroska" }
|
||||
filmcraft-h264 = { path = "crates/h264" }
|
||||
filmcraft-prores = { path = "crates/prores" }
|
||||
filmcraft-aac = { path = "crates/aac" }
|
||||
|
|
|
|||
|
|
@ -0,0 +1,15 @@
|
|||
[package]
|
||||
name = "filmcraft-matroska"
|
||||
description = "Clean-room Matroska (MKV) and WebM demuxer"
|
||||
version.workspace = true
|
||||
edition.workspace = true
|
||||
license.workspace = true
|
||||
rust-version.workspace = true
|
||||
|
||||
[lints]
|
||||
workspace = true
|
||||
|
||||
[dependencies]
|
||||
|
||||
[dev-dependencies]
|
||||
serde_json = { workspace = true }
|
||||
|
|
@ -0,0 +1,225 @@
|
|||
//! Matroska CodecID → codec mapping (Matroska codec specifications, RFC 9559 §5.1.4.1.21 and the
|
||||
//! codec mappings registry).
|
||||
|
||||
/// Codec of a track, derived from `CodecID` and `CodecPrivate`.
|
||||
///
|
||||
/// Configuration records are passed through as the raw bytes found in `CodecPrivate`, which for
|
||||
/// the ISO codecs are exactly the ISOBMFF configuration boxes' payloads (avcC, hvcC, av1C, ASC…).
|
||||
#[derive(Clone, Debug, PartialEq)]
|
||||
pub enum Codec {
|
||||
/// `V_MPEG4/ISO/AVC`; payload is an `AVCDecoderConfigurationRecord` (avcC).
|
||||
Avc { avcc: Vec<u8> },
|
||||
/// `V_MPEGH/ISO/HEVC`; payload is an `HEVCDecoderConfigurationRecord` (hvcC).
|
||||
Hevc { hvcc: Vec<u8> },
|
||||
/// `V_VP8`.
|
||||
Vp8,
|
||||
/// `V_VP9`; `CodecPrivate` (optional) holds VP9 codec feature metadata.
|
||||
Vp9 { private: Vec<u8> },
|
||||
/// `V_AV1`; payload is an `AV1CodecConfigurationRecord` (av1C).
|
||||
Av1 { av1c: Vec<u8> },
|
||||
/// `V_PRORES`; `fourcc` from `CodecPrivate` (e.g. `apcn`), if present.
|
||||
ProRes { fourcc: Option<[u8; 4]> },
|
||||
/// `V_MJPEG`, or `V_MS/VFW/FOURCC` with an `MJPG` compression FourCC.
|
||||
Mjpeg,
|
||||
/// `V_MS/VFW/FOURCC` with any other FourCC; `bitmap_info_header` is the whole `CodecPrivate`.
|
||||
VfwFourcc { fourcc: [u8; 4], bitmap_info_header: Vec<u8> },
|
||||
/// `A_AAC` (and the legacy `A_AAC/MPEG{2,4}/<profile>` IDs); `asc` is the
|
||||
/// `AudioSpecificConfig` (synthesised for legacy IDs without `CodecPrivate`).
|
||||
Aac { asc: Vec<u8> },
|
||||
/// `A_OPUS`; `head` is the `OpusHead` identification header.
|
||||
Opus { head: Vec<u8> },
|
||||
/// `A_VORBIS`; the three Vorbis headers (identification, comment, setup), split from the
|
||||
/// Xiph-laced `CodecPrivate`.
|
||||
Vorbis { headers: Vec<Vec<u8>> },
|
||||
/// `A_FLAC`; `private` is `fLaC` + metadata blocks (STREAMINFO first).
|
||||
Flac { private: Vec<u8> },
|
||||
/// `A_PCM/INT/LIT`, `A_PCM/INT/BIG`, `A_PCM/FLOAT/IEEE`. Bit depth comes from the Audio element.
|
||||
Pcm { float: bool, big_endian: bool, bits: u16 },
|
||||
/// `A_AC3`.
|
||||
Ac3,
|
||||
/// `A_EAC3`.
|
||||
Eac3,
|
||||
/// `A_MPEG/L3`.
|
||||
Mp3,
|
||||
/// `A_MPEG/L2`.
|
||||
Mp2,
|
||||
/// `S_TEXT/UTF8` (SubRip-style plain text).
|
||||
SubRip,
|
||||
/// `S_TEXT/WEBVTT`.
|
||||
WebVtt,
|
||||
/// `S_TEXT/ASS` / `S_TEXT/SSA`; `header` is the script header from `CodecPrivate`.
|
||||
Ass { header: Vec<u8>, ssa: bool },
|
||||
/// Any other CodecID (kept verbatim along with `Track::codec_private`).
|
||||
Other(String),
|
||||
}
|
||||
|
||||
impl Codec {
|
||||
/// Short codec name, matching FFmpeg's `codec_name` where one exists (useful for logs/UI).
|
||||
pub fn name(&self) -> String {
|
||||
match self {
|
||||
Codec::Avc { .. } => "h264".into(),
|
||||
Codec::Hevc { .. } => "hevc".into(),
|
||||
Codec::Vp8 => "vp8".into(),
|
||||
Codec::Vp9 { .. } => "vp9".into(),
|
||||
Codec::Av1 { .. } => "av1".into(),
|
||||
Codec::ProRes { .. } => "prores".into(),
|
||||
Codec::Mjpeg => "mjpeg".into(),
|
||||
Codec::VfwFourcc { fourcc, .. } => String::from_utf8_lossy(fourcc).trim().to_ascii_lowercase(),
|
||||
Codec::Aac { .. } => "aac".into(),
|
||||
Codec::Opus { .. } => "opus".into(),
|
||||
Codec::Vorbis { .. } => "vorbis".into(),
|
||||
Codec::Flac { .. } => "flac".into(),
|
||||
Codec::Pcm { float, big_endian, bits } => {
|
||||
let k = if *float {
|
||||
'f'
|
||||
} else if *bits == 8 {
|
||||
'u'
|
||||
} else {
|
||||
's'
|
||||
};
|
||||
if *bits == 8 { format!("pcm_{k}8") } else { format!("pcm_{k}{bits}{}", if *big_endian { "be" } else { "le" }) }
|
||||
}
|
||||
Codec::Ac3 => "ac3".into(),
|
||||
Codec::Eac3 => "eac3".into(),
|
||||
Codec::Mp3 => "mp3".into(),
|
||||
Codec::Mp2 => "mp2".into(),
|
||||
Codec::SubRip => "subrip".into(),
|
||||
Codec::WebVtt => "webvtt".into(),
|
||||
Codec::Ass { ssa, .. } => if *ssa { "ssa" } else { "ass" }.into(),
|
||||
Codec::Other(id) => id.clone(),
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
const AAC_RATES: [u32; 13] = [96000, 88200, 64000, 48000, 44100, 32000, 24000, 22050, 16000, 12000, 11025, 8000, 7350];
|
||||
|
||||
/// Build a 2-byte (or 5-byte with SBR extension) AudioSpecificConfig for legacy `A_AAC/...` IDs.
|
||||
fn synth_asc(object_type: u8, rate: f64, out_rate: Option<f64>, channels: u64, sbr: bool) -> Vec<u8> {
|
||||
let rate_index = |r: f64| AAC_RATES.iter().position(|&x| x as f64 == r).unwrap_or(4) as u8;
|
||||
let ri = rate_index(rate);
|
||||
let ch = channels.min(7) as u8;
|
||||
let mut v = vec![(object_type << 3) | (ri >> 1), ((ri & 1) << 7) | (ch << 3)];
|
||||
if sbr {
|
||||
// backward-compatible explicit signalling: sync 0x2b7, SBR object type 5, flag, ext rate index
|
||||
let ext = rate_index(out_rate.unwrap_or(rate * 2.0));
|
||||
v.push(0x56);
|
||||
v.push(0xE5);
|
||||
v.push(0x80 | (ext << 3));
|
||||
}
|
||||
v
|
||||
}
|
||||
|
||||
/// Split a Xiph-laced `CodecPrivate` (Vorbis/Theora headers).
|
||||
fn split_xiph(p: &[u8]) -> Vec<Vec<u8>> {
|
||||
let Some(&n) = p.first() else { return Vec::new() };
|
||||
let mut pos = 1;
|
||||
let mut sizes = Vec::new();
|
||||
for _ in 0..n {
|
||||
let mut s = 0usize;
|
||||
loop {
|
||||
let Some(&b) = p.get(pos) else { return Vec::new() };
|
||||
pos += 1;
|
||||
s += b as usize;
|
||||
if b != 255 {
|
||||
break;
|
||||
}
|
||||
}
|
||||
sizes.push(s);
|
||||
}
|
||||
let mut out = Vec::new();
|
||||
for s in sizes {
|
||||
let Some(h) = p.get(pos..pos + s) else { return Vec::new() };
|
||||
out.push(h.to_vec());
|
||||
pos += s;
|
||||
}
|
||||
out.push(p[pos.min(p.len())..].to_vec());
|
||||
out
|
||||
}
|
||||
|
||||
/// Map a CodecID + CodecPrivate (+ audio params for legacy AAC / PCM) to a [`Codec`].
|
||||
pub(crate) fn map_codec(id: &str, private: &[u8], audio: Option<&crate::AudioInfo>) -> Codec {
|
||||
let p = private.to_vec();
|
||||
match id {
|
||||
"V_MPEG4/ISO/AVC" => Codec::Avc { avcc: p },
|
||||
"V_MPEGH/ISO/HEVC" => Codec::Hevc { hvcc: p },
|
||||
"V_VP8" => Codec::Vp8,
|
||||
"V_VP9" => Codec::Vp9 { private: p },
|
||||
"V_AV1" => Codec::Av1 { av1c: p },
|
||||
"V_PRORES" => Codec::ProRes { fourcc: private.get(..4).and_then(|f| f.try_into().ok()) },
|
||||
"V_MJPEG" => Codec::Mjpeg,
|
||||
"V_MS/VFW/FOURCC" => {
|
||||
let fourcc: [u8; 4] = private.get(16..20).and_then(|f| f.try_into().ok()).unwrap_or(*b" ");
|
||||
match &fourcc {
|
||||
b"MJPG" | b"mjpg" | b"AVRn" | b"AVDJ" => Codec::Mjpeg,
|
||||
b"H264" | b"h264" | b"avc1" | b"X264" | b"x264" => Codec::Avc { avcc: Vec::new() },
|
||||
b"apch" | b"apcn" | b"apcs" | b"apco" | b"ap4h" | b"ap4x" => Codec::ProRes { fourcc: Some(fourcc) },
|
||||
_ => Codec::VfwFourcc { fourcc, bitmap_info_header: p },
|
||||
}
|
||||
}
|
||||
"A_AAC" => Codec::Aac { asc: p },
|
||||
"A_OPUS" => Codec::Opus { head: p },
|
||||
"A_VORBIS" => Codec::Vorbis { headers: split_xiph(private) },
|
||||
"A_FLAC" => Codec::Flac { private: p },
|
||||
"A_PCM/INT/LIT" | "A_PCM/INT/BIG" | "A_PCM/FLOAT/IEEE" => {
|
||||
let bits = audio.and_then(|a| a.bit_depth).unwrap_or(if id == "A_PCM/FLOAT/IEEE" { 32 } else { 16 }) as u16;
|
||||
Codec::Pcm { float: id == "A_PCM/FLOAT/IEEE", big_endian: id == "A_PCM/INT/BIG", bits }
|
||||
}
|
||||
"A_AC3" | "A_AC3/BSID9" | "A_AC3/BSID10" => Codec::Ac3,
|
||||
"A_EAC3" => Codec::Eac3,
|
||||
"A_MPEG/L3" => Codec::Mp3,
|
||||
"A_MPEG/L2" => Codec::Mp2,
|
||||
"S_TEXT/UTF8" => Codec::SubRip,
|
||||
"S_TEXT/WEBVTT" => Codec::WebVtt,
|
||||
"S_TEXT/ASS" | "S_ASS" => Codec::Ass { header: p, ssa: false },
|
||||
"S_TEXT/SSA" | "S_SSA" => Codec::Ass { header: p, ssa: true },
|
||||
_ if id.starts_with("A_AAC/") => {
|
||||
if !private.is_empty() {
|
||||
return Codec::Aac { asc: p };
|
||||
}
|
||||
let (rate, out_rate, ch) = audio.map_or((48000.0, None, 2), |a| (a.sampling_frequency, a.output_sampling_frequency, a.channels));
|
||||
let object_type = if id.ends_with("/MAIN") {
|
||||
1
|
||||
} else if id.ends_with("/SSR") {
|
||||
3
|
||||
} else if id.ends_with("/LTP") {
|
||||
4
|
||||
} else {
|
||||
2
|
||||
};
|
||||
let sbr = id.ends_with("/SBR");
|
||||
Codec::Aac { asc: synth_asc(object_type, rate, out_rate, ch, sbr) }
|
||||
}
|
||||
_ => Codec::Other(id.to_string()),
|
||||
}
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
|
||||
#[test]
|
||||
fn mapping() {
|
||||
assert_eq!(map_codec("V_MPEG4/ISO/AVC", &[1, 2], None), Codec::Avc { avcc: vec![1, 2] });
|
||||
assert_eq!(map_codec("V_PRORES", b"apcn", None), Codec::ProRes { fourcc: Some(*b"apcn") });
|
||||
let mut bih = vec![0u8; 40];
|
||||
bih[16..20].copy_from_slice(b"MJPG");
|
||||
assert_eq!(map_codec("V_MS/VFW/FOURCC", &bih, None), Codec::Mjpeg);
|
||||
bih[16..20].copy_from_slice(b"FFV1");
|
||||
assert!(matches!(map_codec("V_MS/VFW/FOURCC", &bih, None), Codec::VfwFourcc { fourcc, .. } if &fourcc == b"FFV1"));
|
||||
let a = crate::AudioInfo { sampling_frequency: 44100.0, channels: 2, bit_depth: Some(24), output_sampling_frequency: None };
|
||||
assert_eq!(map_codec("A_PCM/INT/LIT", &[], Some(&a)).name(), "pcm_s24le");
|
||||
assert_eq!(map_codec("A_PCM/FLOAT/IEEE", &[], Some(&a)), Codec::Pcm { float: true, big_endian: false, bits: 24 });
|
||||
// legacy AAC LC 44.1k stereo → 0x12 0x10
|
||||
assert_eq!(map_codec("A_AAC/MPEG4/LC", &[], Some(&a)), Codec::Aac { asc: vec![0x12, 0x10] });
|
||||
assert_eq!(map_codec("S_TEXT/UTF8", &[], None).name(), "subrip");
|
||||
assert_eq!(map_codec("X_WHATEVER", &[], None), Codec::Other("X_WHATEVER".into()));
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn xiph_split() {
|
||||
let mut p = vec![2u8, 3, 2];
|
||||
p.extend_from_slice(&[1, 1, 1, 2, 2, 3, 3, 3, 3]);
|
||||
let h = split_xiph(&p);
|
||||
assert_eq!(h, vec![vec![1, 1, 1], vec![2, 2], vec![3, 3, 3, 3]]);
|
||||
}
|
||||
}
|
||||
File diff suppressed because it is too large
Load Diff
|
|
@ -0,0 +1,474 @@
|
|||
//! EBML primitives (RFC 8794): variable-length integers, element headers, typed values, CRC-32,
|
||||
//! and Matroska block lacing (RFC 9559 §10.3).
|
||||
|
||||
use crate::error::{Result, invalid};
|
||||
|
||||
/// Length in bytes of a VINT whose first byte is `b` (1..=8), or `None` for a zero first byte.
|
||||
#[inline]
|
||||
pub fn vint_len(b: u8) -> Option<usize> {
|
||||
if b == 0 { None } else { Some(b.leading_zeros() as usize + 1) }
|
||||
}
|
||||
|
||||
/// Element data size: known, or "unknown" (all value bits set).
|
||||
#[derive(Clone, Copy, Debug, PartialEq, Eq)]
|
||||
pub enum Size {
|
||||
Known(u64),
|
||||
Unknown,
|
||||
}
|
||||
|
||||
impl Size {
|
||||
pub fn known(self) -> Option<u64> {
|
||||
match self {
|
||||
Size::Known(n) => Some(n),
|
||||
Size::Unknown => None,
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/// Decode an unsigned VINT (marker bit stripped). Returns `(value, length, all_ones)`.
|
||||
pub fn read_vint(data: &[u8]) -> Option<(u64, usize, bool)> {
|
||||
let first = *data.first()?;
|
||||
let len = vint_len(first)?;
|
||||
if data.len() < len {
|
||||
return None;
|
||||
}
|
||||
let mask = if len == 8 { 0 } else { 0xFFu8 >> len };
|
||||
let mut v = (first & mask) as u64;
|
||||
for &b in &data[1..len] {
|
||||
v = (v << 8) | b as u64;
|
||||
}
|
||||
let all_ones = v == (1u64 << (7 * len)) - 1;
|
||||
Some((v, len, all_ones))
|
||||
}
|
||||
|
||||
/// Decode an element data size. All-ones is the reserved "unknown size".
|
||||
pub fn read_size(data: &[u8]) -> Option<(Size, usize)> {
|
||||
let (v, len, ones) = read_vint(data)?;
|
||||
Some((if ones { Size::Unknown } else { Size::Known(v) }, len))
|
||||
}
|
||||
|
||||
/// Decode an element ID (marker bits kept, 1..=4 bytes). Rejects all-zero / all-ones value bits.
|
||||
pub fn read_id(data: &[u8]) -> Option<(u32, usize)> {
|
||||
let first = *data.first()?;
|
||||
let len = vint_len(first)?;
|
||||
if len > 4 || data.len() < len {
|
||||
return None;
|
||||
}
|
||||
let (v, _, ones) = read_vint(&data[..len])?;
|
||||
if ones || v == 0 {
|
||||
return None;
|
||||
}
|
||||
let mut id = 0u32;
|
||||
for &b in &data[..len] {
|
||||
id = (id << 8) | b as u32;
|
||||
}
|
||||
Some((id, len))
|
||||
}
|
||||
|
||||
/// Decode a signed VINT as used by EBML lacing (value − (2^(7·len−1) − 1)).
|
||||
pub fn read_svint(data: &[u8]) -> Option<(i64, usize)> {
|
||||
let (v, len, _) = read_vint(data)?;
|
||||
let bias = (1i64 << (7 * len - 1)) - 1;
|
||||
Some((v as i64 - bias, len))
|
||||
}
|
||||
|
||||
/// Encode an element ID (already carrying its marker bits).
|
||||
pub fn write_id(out: &mut Vec<u8>, id: u32) {
|
||||
let n = if id >= 0x100_0000 {
|
||||
4
|
||||
} else if id >= 0x1_0000 {
|
||||
3
|
||||
} else if id >= 0x100 {
|
||||
2
|
||||
} else {
|
||||
1
|
||||
};
|
||||
out.extend_from_slice(&id.to_be_bytes()[4 - n..]);
|
||||
}
|
||||
|
||||
/// Encode a data size as a VINT using the shortest length (or `min_len` if larger).
|
||||
pub fn write_size(out: &mut Vec<u8>, size: u64, min_len: usize) {
|
||||
let mut len = 1;
|
||||
while len < 8 && size >= (1u64 << (7 * len)) - 1 {
|
||||
len += 1;
|
||||
}
|
||||
let len = len.max(min_len).min(8);
|
||||
let v = size | (1u64 << (7 * len));
|
||||
out.extend_from_slice(&v.to_be_bytes()[8 - len..]);
|
||||
}
|
||||
|
||||
/// Encode the "unknown size" marker of the given length.
|
||||
pub fn write_unknown_size(out: &mut Vec<u8>, len: usize) {
|
||||
out.push(0xFFu8 >> (len - 1));
|
||||
out.extend(std::iter::repeat_n(0xFF, len - 1));
|
||||
}
|
||||
|
||||
/// Big-endian unsigned integer of 0..=8 bytes.
|
||||
pub fn uint(data: &[u8]) -> u64 {
|
||||
data.iter().take(8).fold(0u64, |a, &b| (a << 8) | b as u64)
|
||||
}
|
||||
|
||||
/// Big-endian two's-complement signed integer of 0..=8 bytes.
|
||||
pub fn int(data: &[u8]) -> i64 {
|
||||
let n = data.len().min(8);
|
||||
if n == 0 {
|
||||
return 0;
|
||||
}
|
||||
let v = uint(&data[..n]);
|
||||
let shift = 64 - 8 * n as u32;
|
||||
((v << shift) as i64) >> shift
|
||||
}
|
||||
|
||||
/// IEEE float of 0, 4 or 8 bytes.
|
||||
pub fn float(data: &[u8]) -> f64 {
|
||||
match data.len() {
|
||||
4 => f32::from_be_bytes(data.try_into().unwrap_or([0; 4])) as f64,
|
||||
8 => f64::from_be_bytes(data.try_into().unwrap_or([0; 8])),
|
||||
_ => 0.0,
|
||||
}
|
||||
}
|
||||
|
||||
/// String / UTF-8 value: bytes up to the first NUL, lossily decoded.
|
||||
pub fn string(data: &[u8]) -> String {
|
||||
let end = data.iter().position(|&b| b == 0).unwrap_or(data.len());
|
||||
String::from_utf8_lossy(&data[..end]).into_owned()
|
||||
}
|
||||
|
||||
/// One element inside an in-memory master element.
|
||||
#[derive(Clone, Copy, Debug)]
|
||||
pub struct Elem<'a> {
|
||||
pub id: u32,
|
||||
/// Offset of the element header, relative to the start of the parent's data.
|
||||
pub offset: usize,
|
||||
pub header_len: usize,
|
||||
pub data: &'a [u8],
|
||||
/// The size field was "unknown" (data runs to the end of the parent).
|
||||
pub unknown_size: bool,
|
||||
/// The declared size overran the parent and was truncated.
|
||||
pub truncated: bool,
|
||||
}
|
||||
|
||||
/// Iterator over the children of an in-memory master element.
|
||||
///
|
||||
/// Damaged data ends the iteration (see [`Children::damaged`]); oversize elements are truncated to
|
||||
/// the parent's end.
|
||||
pub struct Children<'a> {
|
||||
data: &'a [u8],
|
||||
pos: usize,
|
||||
damaged: bool,
|
||||
}
|
||||
|
||||
impl<'a> Children<'a> {
|
||||
pub fn new(data: &'a [u8]) -> Self {
|
||||
Self { data, pos: 0, damaged: false }
|
||||
}
|
||||
/// True if iteration stopped on an undecodable header.
|
||||
pub fn damaged(&self) -> bool {
|
||||
self.damaged
|
||||
}
|
||||
}
|
||||
|
||||
impl<'a> Iterator for Children<'a> {
|
||||
type Item = Elem<'a>;
|
||||
fn next(&mut self) -> Option<Elem<'a>> {
|
||||
let rest = self.data.get(self.pos..)?;
|
||||
if rest.is_empty() {
|
||||
return None;
|
||||
}
|
||||
let Some((id, il)) = read_id(rest) else {
|
||||
self.damaged = true;
|
||||
self.pos = self.data.len();
|
||||
return None;
|
||||
};
|
||||
let Some((size, sl)) = read_size(&rest[il..]) else {
|
||||
self.damaged = true;
|
||||
self.pos = self.data.len();
|
||||
return None;
|
||||
};
|
||||
let hl = il + sl;
|
||||
let avail = (rest.len() - hl) as u64;
|
||||
let (len, unknown, truncated) = match size {
|
||||
Size::Known(n) if n <= avail => (n as usize, false, false),
|
||||
Size::Known(_) => (avail as usize, false, true),
|
||||
Size::Unknown => (avail as usize, true, false),
|
||||
};
|
||||
let e = Elem { id, offset: self.pos, header_len: hl, data: &rest[hl..hl + len], unknown_size: unknown, truncated };
|
||||
self.pos += hl + len;
|
||||
Some(e)
|
||||
}
|
||||
}
|
||||
|
||||
/// CRC-32 (ISO 3309 / IEEE 802.3, reflected polynomial 0xEDB88320), as used by EBML CRC-32 elements.
|
||||
pub fn crc32(data: &[u8]) -> u32 {
|
||||
crc32_update(0xFFFF_FFFF, data) ^ 0xFFFF_FFFF
|
||||
}
|
||||
|
||||
/// Incremental form: start with `0xFFFF_FFFF`, finish with `^ 0xFFFF_FFFF`.
|
||||
pub fn crc32_update(mut crc: u32, data: &[u8]) -> u32 {
|
||||
static TABLE: std::sync::OnceLock<[u32; 256]> = std::sync::OnceLock::new();
|
||||
let t = TABLE.get_or_init(|| {
|
||||
let mut t = [0u32; 256];
|
||||
for (i, e) in t.iter_mut().enumerate() {
|
||||
let mut c = i as u32;
|
||||
for _ in 0..8 {
|
||||
c = if c & 1 != 0 { 0xEDB8_8320 ^ (c >> 1) } else { c >> 1 };
|
||||
}
|
||||
*e = c;
|
||||
}
|
||||
t
|
||||
});
|
||||
for &b in data {
|
||||
crc = t[((crc ^ b as u32) & 0xFF) as usize] ^ (crc >> 8);
|
||||
}
|
||||
crc
|
||||
}
|
||||
|
||||
/// Check a master element's leading CRC-32 child, if present. `Some(true/false)` if there is one.
|
||||
pub fn verify_crc(master_data: &[u8]) -> Option<bool> {
|
||||
let mut it = Children::new(master_data);
|
||||
let first = it.next()?;
|
||||
if first.id != crate::ids::CRC32 || first.data.len() != 4 {
|
||||
return None;
|
||||
}
|
||||
let stored = u32::from_le_bytes(first.data.try_into().ok()?);
|
||||
let rest = &master_data[first.offset + first.header_len + 4..];
|
||||
Some(crc32(rest) == stored)
|
||||
}
|
||||
|
||||
/// Block lacing mode (flags bits 1–2).
|
||||
#[derive(Clone, Copy, Debug, PartialEq, Eq)]
|
||||
pub enum Lacing {
|
||||
None,
|
||||
Xiph,
|
||||
FixedSize,
|
||||
Ebml,
|
||||
}
|
||||
|
||||
impl Lacing {
|
||||
pub fn from_flags(flags: u8) -> Lacing {
|
||||
match (flags >> 1) & 3 {
|
||||
0 => Lacing::None,
|
||||
1 => Lacing::Xiph,
|
||||
2 => Lacing::FixedSize,
|
||||
_ => Lacing::Ebml,
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/// Split laced block payload. `data` starts right after the block's flags byte and runs to the end
|
||||
/// of the block. Returns `(offset, len)` of each frame relative to `data`.
|
||||
pub fn parse_lacing(lacing: Lacing, data: &[u8]) -> Result<Vec<(usize, usize)>> {
|
||||
if lacing == Lacing::None {
|
||||
return Ok(vec![(0, data.len())]);
|
||||
}
|
||||
let Some(&n1) = data.first() else { return invalid("laced block without frame count") };
|
||||
let count = n1 as usize + 1;
|
||||
let mut pos = 1usize;
|
||||
let mut sizes = Vec::with_capacity(count);
|
||||
match lacing {
|
||||
Lacing::None => unreachable!(),
|
||||
Lacing::Xiph => {
|
||||
for _ in 0..count - 1 {
|
||||
let mut s = 0usize;
|
||||
loop {
|
||||
let Some(&b) = data.get(pos) else { return invalid("truncated Xiph lace header") };
|
||||
pos += 1;
|
||||
s += b as usize;
|
||||
if b != 255 {
|
||||
break;
|
||||
}
|
||||
}
|
||||
sizes.push(s);
|
||||
}
|
||||
}
|
||||
Lacing::Ebml => {
|
||||
if count > 1 {
|
||||
let Some((first, l, _)) = read_vint(&data[pos..]) else { return invalid("bad EBML lace size") };
|
||||
pos += l;
|
||||
let mut prev = first as i64;
|
||||
sizes.push(first as usize);
|
||||
for _ in 1..count - 1 {
|
||||
let Some((d, l)) = read_svint(&data[pos..]) else { return invalid("bad EBML lace delta") };
|
||||
pos += l;
|
||||
prev += d;
|
||||
if prev < 0 {
|
||||
return invalid("negative EBML lace size");
|
||||
}
|
||||
sizes.push(prev as usize);
|
||||
}
|
||||
}
|
||||
}
|
||||
Lacing::FixedSize => {
|
||||
let rest = data.len() - pos;
|
||||
if !rest.is_multiple_of(count) {
|
||||
return invalid("fixed-size lacing does not divide block");
|
||||
}
|
||||
sizes.extend(std::iter::repeat_n(rest / count, count - 1));
|
||||
}
|
||||
}
|
||||
let used: usize = sizes.iter().try_fold(0usize, |a, &s| a.checked_add(s)).unwrap_or(usize::MAX);
|
||||
if pos > data.len() || used > data.len() - pos {
|
||||
return invalid("lace sizes exceed block");
|
||||
}
|
||||
sizes.push(data.len() - pos - used);
|
||||
let mut out = Vec::with_capacity(count);
|
||||
for s in sizes {
|
||||
out.push((pos, s));
|
||||
pos += s;
|
||||
}
|
||||
Ok(out)
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
|
||||
#[test]
|
||||
fn vint_roundtrip() {
|
||||
for &v in &[0u64, 1, 126, 127, 128, 16382, 16383, 16384, 1 << 20, (1 << 56) - 2] {
|
||||
let mut b = Vec::new();
|
||||
write_size(&mut b, v, 1);
|
||||
let (s, l) = read_size(&b).unwrap();
|
||||
assert_eq!(s, Size::Known(v), "value {v}");
|
||||
assert_eq!(l, b.len());
|
||||
}
|
||||
// 127 needs 2 bytes because 0xFF is "unknown"
|
||||
let mut b = Vec::new();
|
||||
write_size(&mut b, 127, 1);
|
||||
assert_eq!(b, [0x40, 0x7F]);
|
||||
// padded length
|
||||
let mut b = Vec::new();
|
||||
write_size(&mut b, 5, 8);
|
||||
assert_eq!(b, [1, 0, 0, 0, 0, 0, 0, 5]);
|
||||
assert_eq!(read_size(&b).unwrap(), (Size::Known(5), 8));
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn vint_unknown_and_invalid() {
|
||||
assert_eq!(read_size(&[0xFF]).unwrap(), (Size::Unknown, 1));
|
||||
assert_eq!(read_size(&[0x01, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF]).unwrap(), (Size::Unknown, 8));
|
||||
for l in 1..=8 {
|
||||
let mut b = Vec::new();
|
||||
write_unknown_size(&mut b, l);
|
||||
assert_eq!(read_size(&b).unwrap(), (Size::Unknown, l));
|
||||
}
|
||||
assert!(read_size(&[0x00, 1, 2]).is_none());
|
||||
assert!(read_size(&[0x40]).is_none()); // truncated
|
||||
assert_eq!(read_vint(&[0x81]).unwrap(), (1, 1, false));
|
||||
assert_eq!(read_vint(&[0x10, 0x00, 0x00, 0x01]).unwrap(), (1, 4, false));
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn ids() {
|
||||
assert_eq!(read_id(&[0x1A, 0x45, 0xDF, 0xA3, 0x99]).unwrap(), (crate::ids::EBML, 4));
|
||||
assert_eq!(read_id(&[0xA3]).unwrap(), (0xA3, 1));
|
||||
assert_eq!(read_id(&[0x42, 0x86]).unwrap(), (0x4286, 2));
|
||||
assert!(read_id(&[0xFF]).is_none()); // reserved
|
||||
assert!(read_id(&[0x80]).is_none()); // zero value
|
||||
assert!(read_id(&[0x08, 0, 0, 0, 1]).is_none()); // 5-byte ID
|
||||
for id in [0xA3u32, 0x4286, 0x2AD7B1, 0x1F43B675] {
|
||||
let mut b = Vec::new();
|
||||
write_id(&mut b, id);
|
||||
assert_eq!(read_id(&b).unwrap(), (id, b.len()));
|
||||
}
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn signed_vint() {
|
||||
assert_eq!(read_svint(&[0xBF]).unwrap(), (0, 1)); // 63 - 63
|
||||
assert_eq!(read_svint(&[0x80]).unwrap(), (-63, 1));
|
||||
assert_eq!(read_svint(&[0xC0]).unwrap(), (1, 1));
|
||||
assert_eq!(read_svint(&[0x5F, 0xFF]).unwrap(), (0, 2));
|
||||
assert_eq!(read_svint(&[0x60, 0x00]).unwrap(), (1, 2));
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn values() {
|
||||
assert_eq!(uint(&[]), 0);
|
||||
assert_eq!(uint(&[1, 0]), 256);
|
||||
assert_eq!(int(&[0xFF]), -1);
|
||||
assert_eq!(int(&[0xFF, 0xFE]), -2);
|
||||
assert_eq!(int(&[0x7F]), 127);
|
||||
assert_eq!(float(&1.5f32.to_be_bytes()), 1.5);
|
||||
assert_eq!(float(&2.25f64.to_be_bytes()), 2.25);
|
||||
assert_eq!(string(b"abc\0\0"), "abc");
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn crc() {
|
||||
assert_eq!(crc32(b"123456789"), 0xCBF4_3926);
|
||||
let payload = [0x42u8, 0x86, 0x81, 0x01];
|
||||
let mut m = vec![0xBF, 0x84];
|
||||
m.extend_from_slice(&crc32(&payload).to_le_bytes());
|
||||
m.extend_from_slice(&payload);
|
||||
assert_eq!(verify_crc(&m), Some(true));
|
||||
m[7] ^= 1;
|
||||
assert_eq!(verify_crc(&m), Some(false));
|
||||
assert_eq!(verify_crc(&payload), None);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn children_iter() {
|
||||
// two elements, then one whose size overruns
|
||||
let d = [0x42, 0x86, 0x81, 0x01, 0xEC, 0x82, 0, 0, 0xA3, 0x85, 1, 2];
|
||||
let v: Vec<_> = Children::new(&d).collect();
|
||||
assert_eq!(v.len(), 3);
|
||||
assert_eq!(v[0].id, 0x4286);
|
||||
assert_eq!(v[0].data, &[1]);
|
||||
assert_eq!(v[1].id, 0xEC);
|
||||
assert_eq!(v[2].data, &[1, 2]);
|
||||
assert!(v[2].truncated);
|
||||
let mut it = Children::new(&[0x00, 0x01]);
|
||||
assert!(it.next().is_none());
|
||||
assert!(it.damaged());
|
||||
}
|
||||
|
||||
fn frames<'a>(d: &'a [u8], f: &[(usize, usize)]) -> Vec<&'a [u8]> {
|
||||
f.iter().map(|&(o, l)| &d[o..o + l]).collect()
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn lacing_xiph() {
|
||||
// 3 frames: 300, 2, rest(3)
|
||||
let mut d = vec![2u8, 255, 45, 2];
|
||||
d.extend(std::iter::repeat_n(7u8, 300));
|
||||
d.extend_from_slice(&[1, 2]);
|
||||
d.extend_from_slice(&[3, 4, 5]);
|
||||
let f = parse_lacing(Lacing::Xiph, &d).unwrap();
|
||||
assert_eq!(f.iter().map(|x| x.1).collect::<Vec<_>>(), [300, 2, 3]);
|
||||
assert_eq!(frames(&d, &f)[2], &[3, 4, 5]);
|
||||
// exact multiple of 255 needs a terminating 0
|
||||
let mut d = vec![1u8, 255, 0];
|
||||
d.extend(std::iter::repeat_n(1u8, 255));
|
||||
d.push(9);
|
||||
let f = parse_lacing(Lacing::Xiph, &d).unwrap();
|
||||
assert_eq!(f.iter().map(|x| x.1).collect::<Vec<_>>(), [255, 1]);
|
||||
assert!(parse_lacing(Lacing::Xiph, &[1, 200, 1]).is_err());
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn lacing_ebml() {
|
||||
// 4 frames: 5, 3 (delta -2), 4 (delta +1), rest 2
|
||||
let mut d = vec![3u8, 0x85, 0xBF - 2, 0xBF + 1];
|
||||
d.extend_from_slice(&[1; 5]);
|
||||
d.extend_from_slice(&[2; 3]);
|
||||
d.extend_from_slice(&[3; 4]);
|
||||
d.extend_from_slice(&[4; 2]);
|
||||
let f = parse_lacing(Lacing::Ebml, &d).unwrap();
|
||||
assert_eq!(f.iter().map(|x| x.1).collect::<Vec<_>>(), [5, 3, 4, 2]);
|
||||
assert_eq!(frames(&d, &f), [&[1u8; 5][..], &[2; 3], &[3; 4], &[4; 2]]);
|
||||
assert!(parse_lacing(Lacing::Ebml, &[1, 0x90]).is_err());
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn lacing_fixed_and_none() {
|
||||
let d = [2u8, 1, 1, 2, 2, 3, 3];
|
||||
let f = parse_lacing(Lacing::FixedSize, &d).unwrap();
|
||||
assert_eq!(frames(&d, &f), [&[1u8, 1][..], &[2, 2], &[3, 3]]);
|
||||
assert!(parse_lacing(Lacing::FixedSize, &[2, 1, 1]).is_err());
|
||||
assert_eq!(parse_lacing(Lacing::None, &[9, 9]).unwrap(), [(0, 2)]);
|
||||
assert_eq!(Lacing::from_flags(0x80 | 0x06), Lacing::Ebml);
|
||||
assert_eq!(Lacing::from_flags(0x02), Lacing::Xiph);
|
||||
assert_eq!(Lacing::from_flags(0x04), Lacing::FixedSize);
|
||||
}
|
||||
}
|
||||
|
|
@ -0,0 +1,53 @@
|
|||
use std::fmt;
|
||||
use std::io;
|
||||
|
||||
/// Errors produced by the demuxer.
|
||||
#[derive(Debug)]
|
||||
pub enum Error {
|
||||
/// Underlying I/O failure.
|
||||
Io(io::Error),
|
||||
/// Structurally invalid data that could not be recovered from.
|
||||
Invalid(String),
|
||||
/// Valid but unsupported feature (e.g. zlib-compressed or encrypted tracks when reading data).
|
||||
Unsupported(String),
|
||||
/// Not an EBML file, or an EBML file whose DocType is not `matroska`/`webm`.
|
||||
NotMatroska(String),
|
||||
/// A track index was out of range.
|
||||
NoSuchTrack(usize),
|
||||
/// A sample index was out of range.
|
||||
NoSuchSample(usize),
|
||||
}
|
||||
|
||||
impl fmt::Display for Error {
|
||||
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
|
||||
match self {
|
||||
Error::Io(e) => write!(f, "I/O error: {e}"),
|
||||
Error::Invalid(s) => write!(f, "invalid data: {s}"),
|
||||
Error::Unsupported(s) => write!(f, "unsupported: {s}"),
|
||||
Error::NotMatroska(s) => write!(f, "not a Matroska/WebM file: {s}"),
|
||||
Error::NoSuchTrack(i) => write!(f, "track index {i} out of range"),
|
||||
Error::NoSuchSample(i) => write!(f, "sample index {i} out of range"),
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
impl std::error::Error for Error {
|
||||
fn source(&self) -> Option<&(dyn std::error::Error + 'static)> {
|
||||
match self {
|
||||
Error::Io(e) => Some(e),
|
||||
_ => None,
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
impl From<io::Error> for Error {
|
||||
fn from(e: io::Error) -> Self {
|
||||
Error::Io(e)
|
||||
}
|
||||
}
|
||||
|
||||
pub type Result<T> = std::result::Result<T, Error>;
|
||||
|
||||
pub(crate) fn invalid<T>(msg: impl Into<String>) -> Result<T> {
|
||||
Err(Error::Invalid(msg.into()))
|
||||
}
|
||||
|
|
@ -0,0 +1,206 @@
|
|||
//! EBML / Matroska element IDs (RFC 8794, RFC 9559). IDs keep their VINT marker bits.
|
||||
#![allow(dead_code)]
|
||||
|
||||
// EBML header + global elements
|
||||
pub const EBML: u32 = 0x1A45_DFA3;
|
||||
pub const EBML_VERSION: u32 = 0x4286;
|
||||
pub const EBML_READ_VERSION: u32 = 0x42F7;
|
||||
pub const EBML_MAX_ID_LENGTH: u32 = 0x42F2;
|
||||
pub const EBML_MAX_SIZE_LENGTH: u32 = 0x42F3;
|
||||
pub const DOC_TYPE: u32 = 0x4282;
|
||||
pub const DOC_TYPE_VERSION: u32 = 0x4287;
|
||||
pub const DOC_TYPE_READ_VERSION: u32 = 0x4285;
|
||||
pub const CRC32: u32 = 0xBF;
|
||||
pub const VOID: u32 = 0xEC;
|
||||
|
||||
// Segment and level-1 elements
|
||||
pub const SEGMENT: u32 = 0x1853_8067;
|
||||
pub const SEEK_HEAD: u32 = 0x114D_9B74;
|
||||
pub const INFO: u32 = 0x1549_A966;
|
||||
pub const TRACKS: u32 = 0x1654_AE6B;
|
||||
pub const CLUSTER: u32 = 0x1F43_B675;
|
||||
pub const CUES: u32 = 0x1C53_BB6B;
|
||||
pub const ATTACHMENTS: u32 = 0x1941_A469;
|
||||
pub const CHAPTERS: u32 = 0x1043_A770;
|
||||
pub const TAGS: u32 = 0x1254_C367;
|
||||
|
||||
/// True for IDs that can only appear directly inside a Segment (or are the EBML/Segment headers):
|
||||
/// used to find the end of unknown-size elements.
|
||||
pub fn is_top_level(id: u32) -> bool {
|
||||
matches!(id, EBML | SEGMENT | SEEK_HEAD | INFO | TRACKS | CLUSTER | CUES | ATTACHMENTS | CHAPTERS | TAGS)
|
||||
}
|
||||
|
||||
// SeekHead
|
||||
pub const SEEK: u32 = 0x4DBB;
|
||||
pub const SEEK_ID: u32 = 0x53AB;
|
||||
pub const SEEK_POSITION: u32 = 0x53AC;
|
||||
|
||||
// Info
|
||||
pub const SEGMENT_UUID: u32 = 0x73A4;
|
||||
pub const SEGMENT_FILENAME: u32 = 0x7384;
|
||||
pub const TIMESTAMP_SCALE: u32 = 0x2A_D7B1;
|
||||
pub const DURATION: u32 = 0x4489;
|
||||
pub const DATE_UTC: u32 = 0x4461;
|
||||
pub const TITLE: u32 = 0x7BA9;
|
||||
pub const MUXING_APP: u32 = 0x4D80;
|
||||
pub const WRITING_APP: u32 = 0x5741;
|
||||
|
||||
// Cluster
|
||||
pub const TIMESTAMP: u32 = 0xE7;
|
||||
pub const POSITION: u32 = 0xA7;
|
||||
pub const PREV_SIZE: u32 = 0xAB;
|
||||
pub const SIMPLE_BLOCK: u32 = 0xA3;
|
||||
pub const BLOCK_GROUP: u32 = 0xA0;
|
||||
pub const ENCRYPTED_BLOCK: u32 = 0xAF;
|
||||
pub const SILENT_TRACKS: u32 = 0x5854;
|
||||
pub const BLOCK: u32 = 0xA1;
|
||||
pub const BLOCK_ADDITIONS: u32 = 0x75A1;
|
||||
pub const BLOCK_DURATION: u32 = 0x9B;
|
||||
pub const REFERENCE_PRIORITY: u32 = 0xFA;
|
||||
pub const REFERENCE_BLOCK: u32 = 0xFB;
|
||||
pub const CODEC_STATE: u32 = 0xA4;
|
||||
pub const DISCARD_PADDING: u32 = 0x75A2;
|
||||
|
||||
pub fn is_cluster_child(id: u32) -> bool {
|
||||
matches!(id, TIMESTAMP | POSITION | PREV_SIZE | SIMPLE_BLOCK | BLOCK_GROUP | ENCRYPTED_BLOCK | SILENT_TRACKS | CRC32 | VOID)
|
||||
}
|
||||
|
||||
// Tracks
|
||||
pub const TRACK_ENTRY: u32 = 0xAE;
|
||||
pub const TRACK_NUMBER: u32 = 0xD7;
|
||||
pub const TRACK_UID: u32 = 0x73C5;
|
||||
pub const TRACK_TYPE: u32 = 0x83;
|
||||
pub const FLAG_ENABLED: u32 = 0xB9;
|
||||
pub const FLAG_DEFAULT: u32 = 0x88;
|
||||
pub const FLAG_FORCED: u32 = 0x55AA;
|
||||
pub const FLAG_HEARING_IMPAIRED: u32 = 0x55AB;
|
||||
pub const FLAG_VISUAL_IMPAIRED: u32 = 0x55AC;
|
||||
pub const FLAG_ORIGINAL: u32 = 0x55AE;
|
||||
pub const FLAG_COMMENTARY: u32 = 0x55AF;
|
||||
pub const FLAG_LACING: u32 = 0x9C;
|
||||
pub const DEFAULT_DURATION: u32 = 0x23_E383;
|
||||
pub const TRACK_TIMESTAMP_SCALE: u32 = 0x23_314F;
|
||||
pub const NAME: u32 = 0x536E;
|
||||
pub const LANGUAGE: u32 = 0x22_B59C;
|
||||
pub const LANGUAGE_BCP47: u32 = 0x22_B59D;
|
||||
pub const CODEC_ID: u32 = 0x86;
|
||||
pub const CODEC_PRIVATE: u32 = 0x63A2;
|
||||
pub const CODEC_NAME: u32 = 0x25_8688;
|
||||
pub const CODEC_DELAY: u32 = 0x56AA;
|
||||
pub const SEEK_PRE_ROLL: u32 = 0x56BB;
|
||||
|
||||
// Video
|
||||
pub const VIDEO: u32 = 0xE0;
|
||||
pub const FLAG_INTERLACED: u32 = 0x9A;
|
||||
pub const FIELD_ORDER: u32 = 0x9D;
|
||||
pub const STEREO_MODE: u32 = 0x53B8;
|
||||
pub const ALPHA_MODE: u32 = 0x53C0;
|
||||
pub const PIXEL_WIDTH: u32 = 0xB0;
|
||||
pub const PIXEL_HEIGHT: u32 = 0xBA;
|
||||
pub const PIXEL_CROP_BOTTOM: u32 = 0x54AA;
|
||||
pub const PIXEL_CROP_TOP: u32 = 0x54BB;
|
||||
pub const PIXEL_CROP_LEFT: u32 = 0x54CC;
|
||||
pub const PIXEL_CROP_RIGHT: u32 = 0x54DD;
|
||||
pub const DISPLAY_WIDTH: u32 = 0x54B0;
|
||||
pub const DISPLAY_HEIGHT: u32 = 0x54BA;
|
||||
pub const DISPLAY_UNIT: u32 = 0x54B2;
|
||||
pub const COLOUR_SPACE: u32 = 0x2E_B524;
|
||||
pub const COLOUR: u32 = 0x55B0;
|
||||
|
||||
// Colour
|
||||
pub const MATRIX_COEFFICIENTS: u32 = 0x55B1;
|
||||
pub const BITS_PER_CHANNEL: u32 = 0x55B2;
|
||||
pub const CHROMA_SUBSAMPLING_HORZ: u32 = 0x55B3;
|
||||
pub const CHROMA_SUBSAMPLING_VERT: u32 = 0x55B4;
|
||||
pub const CB_SUBSAMPLING_HORZ: u32 = 0x55B5;
|
||||
pub const CB_SUBSAMPLING_VERT: u32 = 0x55B6;
|
||||
pub const CHROMA_SITING_HORZ: u32 = 0x55B7;
|
||||
pub const CHROMA_SITING_VERT: u32 = 0x55B8;
|
||||
pub const RANGE: u32 = 0x55B9;
|
||||
pub const TRANSFER_CHARACTERISTICS: u32 = 0x55BA;
|
||||
pub const PRIMARIES: u32 = 0x55BB;
|
||||
pub const MAX_CLL: u32 = 0x55BC;
|
||||
pub const MAX_FALL: u32 = 0x55BD;
|
||||
pub const MASTERING_METADATA: u32 = 0x55D0;
|
||||
pub const PRIMARY_R_X: u32 = 0x55D1;
|
||||
pub const PRIMARY_R_Y: u32 = 0x55D2;
|
||||
pub const PRIMARY_G_X: u32 = 0x55D3;
|
||||
pub const PRIMARY_G_Y: u32 = 0x55D4;
|
||||
pub const PRIMARY_B_X: u32 = 0x55D5;
|
||||
pub const PRIMARY_B_Y: u32 = 0x55D6;
|
||||
pub const WHITE_POINT_X: u32 = 0x55D7;
|
||||
pub const WHITE_POINT_Y: u32 = 0x55D8;
|
||||
pub const LUMINANCE_MAX: u32 = 0x55D9;
|
||||
pub const LUMINANCE_MIN: u32 = 0x55DA;
|
||||
|
||||
// Audio
|
||||
pub const AUDIO: u32 = 0xE1;
|
||||
pub const SAMPLING_FREQUENCY: u32 = 0xB5;
|
||||
pub const OUTPUT_SAMPLING_FREQUENCY: u32 = 0x78B5;
|
||||
pub const CHANNELS: u32 = 0x9F;
|
||||
pub const BIT_DEPTH: u32 = 0x6264;
|
||||
|
||||
// ContentEncodings
|
||||
pub const CONTENT_ENCODINGS: u32 = 0x6D80;
|
||||
pub const CONTENT_ENCODING: u32 = 0x6240;
|
||||
pub const CONTENT_ENCODING_ORDER: u32 = 0x5031;
|
||||
pub const CONTENT_ENCODING_SCOPE: u32 = 0x5032;
|
||||
pub const CONTENT_ENCODING_TYPE: u32 = 0x5033;
|
||||
pub const CONTENT_COMPRESSION: u32 = 0x5034;
|
||||
pub const CONTENT_COMP_ALGO: u32 = 0x4254;
|
||||
pub const CONTENT_COMP_SETTINGS: u32 = 0x4255;
|
||||
pub const CONTENT_ENCRYPTION: u32 = 0x5035;
|
||||
|
||||
// Cues
|
||||
pub const CUE_POINT: u32 = 0xBB;
|
||||
pub const CUE_TIME: u32 = 0xB3;
|
||||
pub const CUE_TRACK_POSITIONS: u32 = 0xB7;
|
||||
pub const CUE_TRACK: u32 = 0xF7;
|
||||
pub const CUE_CLUSTER_POSITION: u32 = 0xF1;
|
||||
pub const CUE_RELATIVE_POSITION: u32 = 0xF0;
|
||||
pub const CUE_DURATION: u32 = 0xB2;
|
||||
pub const CUE_BLOCK_NUMBER: u32 = 0x5378;
|
||||
|
||||
// Attachments
|
||||
pub const ATTACHED_FILE: u32 = 0x61A7;
|
||||
pub const FILE_DESCRIPTION: u32 = 0x467E;
|
||||
pub const FILE_NAME: u32 = 0x466E;
|
||||
pub const FILE_MEDIA_TYPE: u32 = 0x4660;
|
||||
pub const FILE_DATA: u32 = 0x465C;
|
||||
pub const FILE_UID: u32 = 0x46AE;
|
||||
|
||||
// Chapters
|
||||
pub const EDITION_ENTRY: u32 = 0x45B9;
|
||||
pub const EDITION_UID: u32 = 0x45BC;
|
||||
pub const EDITION_FLAG_HIDDEN: u32 = 0x45BD;
|
||||
pub const EDITION_FLAG_DEFAULT: u32 = 0x45DB;
|
||||
pub const EDITION_FLAG_ORDERED: u32 = 0x45DD;
|
||||
pub const CHAPTER_ATOM: u32 = 0xB6;
|
||||
pub const CHAPTER_UID: u32 = 0x73C4;
|
||||
pub const CHAPTER_STRING_UID: u32 = 0x5654;
|
||||
pub const CHAPTER_TIME_START: u32 = 0x91;
|
||||
pub const CHAPTER_TIME_END: u32 = 0x92;
|
||||
pub const CHAPTER_FLAG_HIDDEN: u32 = 0x98;
|
||||
pub const CHAPTER_FLAG_ENABLED: u32 = 0x4598;
|
||||
pub const CHAPTER_DISPLAY: u32 = 0x80;
|
||||
pub const CHAP_STRING: u32 = 0x85;
|
||||
pub const CHAP_LANGUAGE: u32 = 0x437C;
|
||||
pub const CHAP_LANGUAGE_BCP47: u32 = 0x437D;
|
||||
pub const CHAP_COUNTRY: u32 = 0x437E;
|
||||
|
||||
// Tags
|
||||
pub const TAG: u32 = 0x7373;
|
||||
pub const TARGETS: u32 = 0x63C0;
|
||||
pub const TARGET_TYPE_VALUE: u32 = 0x68CA;
|
||||
pub const TARGET_TYPE: u32 = 0x63CA;
|
||||
pub const TAG_TRACK_UID: u32 = 0x63C5;
|
||||
pub const TAG_EDITION_UID: u32 = 0x63C9;
|
||||
pub const TAG_CHAPTER_UID: u32 = 0x63C4;
|
||||
pub const TAG_ATTACHMENT_UID: u32 = 0x63C6;
|
||||
pub const SIMPLE_TAG: u32 = 0x67C8;
|
||||
pub const TAG_NAME: u32 = 0x45A3;
|
||||
pub const TAG_LANGUAGE: u32 = 0x447A;
|
||||
pub const TAG_LANGUAGE_BCP47: u32 = 0x447B;
|
||||
pub const TAG_DEFAULT: u32 = 0x4484;
|
||||
pub const TAG_STRING: u32 = 0x4487;
|
||||
pub const TAG_BINARY: u32 = 0x4485;
|
||||
|
|
@ -0,0 +1,32 @@
|
|||
//! Clean-room Matroska (MKV) and WebM demuxer, written from RFC 8794 (EBML), RFC 9559 (Matroska)
|
||||
//! and the WebM container guidelines.
|
||||
//!
|
||||
//! - [`open`] / [`open_with`] parse the EBML header and the first Segment (Info, Tracks, SeekHead,
|
||||
//! Cues, Chapters, Tags, Attachments) and, by default, scan every Cluster's block headers to build
|
||||
//! per-track [`Sample`] tables — the same shape as `filmcraft-isobmff`, so callers can treat both
|
||||
//! containers alike. Frame data is read on demand with [`MkvFile::read_sample`].
|
||||
//! - [`Demuxer`] iterates [`Packet`]s in file order (from any [`ByteSource`], `Read + Seek`, or a
|
||||
//! byte slice) and seeks to the keyframe preceding a time, via the index, Cues, or a cluster scan.
|
||||
//!
|
||||
//! Layer L0: no dependencies beyond `std`, no `unsafe`, builds for `wasm32-unknown-unknown`.
|
||||
|
||||
mod codec;
|
||||
mod demux;
|
||||
pub mod ebml;
|
||||
mod error;
|
||||
mod ids;
|
||||
mod meta;
|
||||
mod source;
|
||||
mod track;
|
||||
|
||||
pub use codec::Codec;
|
||||
pub use demux::{Demuxer, Keyframe, MkvFile, OpenOptions, Packet, SeekPoint, open, open_with};
|
||||
pub use error::{Error, Result};
|
||||
pub use meta::{Attachment, Chapter, ChapterDisplay, ClusterInfo, CuePoint, CuePosition, Edition, SeekEntry, SegmentInfo, SimpleTag, Tag, TagTargets};
|
||||
pub use source::{ByteSource, ReadSeekSource};
|
||||
pub use track::{AudioInfo, Colour, ContentEncoding, MasteringMetadata, Sample, Track, TrackKind, VideoInfo};
|
||||
|
||||
/// Element IDs used by the demuxer (marker bits included), for callers inspecting raw elements.
|
||||
pub mod element_ids {
|
||||
pub use crate::ids::*;
|
||||
}
|
||||
|
|
@ -0,0 +1,526 @@
|
|||
//! Segment-level metadata: Info, SeekHead, Cues, Chapters, Tags, Attachments — types and parsers
|
||||
//! for their in-memory element data.
|
||||
|
||||
use crate::codec::map_codec;
|
||||
use crate::ebml::{Children, float, int, string, uint};
|
||||
use crate::ids::*;
|
||||
use crate::track::{AudioInfo, Colour, ContentEncoding, MasteringMetadata, Track, TrackKind, VideoInfo, gcd};
|
||||
|
||||
/// `Info` element.
|
||||
#[derive(Clone, Debug, PartialEq)]
|
||||
pub struct SegmentInfo {
|
||||
/// Nanoseconds per timestamp tick (default 1 000 000 = 1 ms).
|
||||
pub timestamp_scale: u64,
|
||||
/// `Duration` in timestamp ticks (float), if present.
|
||||
pub duration: Option<f64>,
|
||||
pub title: Option<String>,
|
||||
pub muxing_app: String,
|
||||
pub writing_app: String,
|
||||
/// `DateUTC`: nanoseconds since 2001-01-01T00:00:00 UTC.
|
||||
pub date_utc: Option<i64>,
|
||||
pub segment_uuid: Option<[u8; 16]>,
|
||||
}
|
||||
|
||||
impl Default for SegmentInfo {
|
||||
fn default() -> Self {
|
||||
Self {
|
||||
timestamp_scale: 1_000_000,
|
||||
duration: None,
|
||||
title: None,
|
||||
muxing_app: String::new(),
|
||||
writing_app: String::new(),
|
||||
date_utc: None,
|
||||
segment_uuid: None,
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/// One `SeekHead` entry: level-1 element ID and its position relative to the Segment data start.
|
||||
#[derive(Clone, Copy, Debug, PartialEq, Eq)]
|
||||
pub struct SeekEntry {
|
||||
pub id: u32,
|
||||
pub position: u64,
|
||||
}
|
||||
|
||||
/// One `CueTrackPositions` of a `CuePoint`.
|
||||
#[derive(Clone, Copy, Debug, PartialEq, Eq)]
|
||||
pub struct CuePosition {
|
||||
pub track: u64,
|
||||
/// Cluster position relative to the Segment data start.
|
||||
pub cluster_position: u64,
|
||||
/// Block position relative to the Cluster data start.
|
||||
pub relative_position: Option<u64>,
|
||||
pub block_number: Option<u64>,
|
||||
pub duration: Option<u64>,
|
||||
}
|
||||
|
||||
/// One `CuePoint`.
|
||||
#[derive(Clone, Debug, PartialEq, Eq)]
|
||||
pub struct CuePoint {
|
||||
/// In timestamp ticks.
|
||||
pub time: u64,
|
||||
pub positions: Vec<CuePosition>,
|
||||
}
|
||||
|
||||
/// A cluster found while scanning.
|
||||
#[derive(Clone, Copy, Debug, PartialEq, Eq)]
|
||||
pub struct ClusterInfo {
|
||||
/// Absolute offset of the Cluster element header.
|
||||
pub offset: u64,
|
||||
/// Absolute offset of the first child.
|
||||
pub data_start: u64,
|
||||
/// Absolute end offset (for unknown-size clusters, where the next top-level element starts).
|
||||
pub end: u64,
|
||||
/// Cluster `Timestamp` in ticks.
|
||||
pub timestamp: u64,
|
||||
}
|
||||
|
||||
/// One `ChapterDisplay`.
|
||||
#[derive(Clone, Debug, Default, PartialEq)]
|
||||
pub struct ChapterDisplay {
|
||||
pub string: String,
|
||||
pub languages: Vec<String>,
|
||||
pub languages_bcp47: Vec<String>,
|
||||
pub countries: Vec<String>,
|
||||
}
|
||||
|
||||
/// `ChapterAtom` (possibly nested).
|
||||
#[derive(Clone, Debug, PartialEq)]
|
||||
pub struct Chapter {
|
||||
pub uid: u64,
|
||||
pub string_uid: Option<String>,
|
||||
pub start_ns: u64,
|
||||
pub end_ns: Option<u64>,
|
||||
pub hidden: bool,
|
||||
pub enabled: bool,
|
||||
pub displays: Vec<ChapterDisplay>,
|
||||
pub children: Vec<Chapter>,
|
||||
}
|
||||
|
||||
impl Chapter {
|
||||
/// First display string, if any.
|
||||
pub fn title(&self) -> Option<&str> {
|
||||
self.displays.first().map(|d| d.string.as_str())
|
||||
}
|
||||
}
|
||||
|
||||
/// `EditionEntry`.
|
||||
#[derive(Clone, Debug, Default, PartialEq)]
|
||||
pub struct Edition {
|
||||
pub uid: Option<u64>,
|
||||
pub hidden: bool,
|
||||
pub default: bool,
|
||||
pub ordered: bool,
|
||||
pub chapters: Vec<Chapter>,
|
||||
}
|
||||
|
||||
/// `Targets` of a `Tag`. Empty UID lists mean "the whole segment".
|
||||
#[derive(Clone, Debug, PartialEq)]
|
||||
pub struct TagTargets {
|
||||
pub type_value: u64,
|
||||
pub type_name: Option<String>,
|
||||
pub track_uids: Vec<u64>,
|
||||
pub edition_uids: Vec<u64>,
|
||||
pub chapter_uids: Vec<u64>,
|
||||
pub attachment_uids: Vec<u64>,
|
||||
}
|
||||
|
||||
impl Default for TagTargets {
|
||||
fn default() -> Self {
|
||||
Self { type_value: 50, type_name: None, track_uids: Vec::new(), edition_uids: Vec::new(), chapter_uids: Vec::new(), attachment_uids: Vec::new() }
|
||||
}
|
||||
}
|
||||
|
||||
/// `SimpleTag` (possibly nested).
|
||||
#[derive(Clone, Debug, PartialEq)]
|
||||
pub struct SimpleTag {
|
||||
pub name: String,
|
||||
pub language: String,
|
||||
pub default: bool,
|
||||
pub string: Option<String>,
|
||||
pub binary: Option<Vec<u8>>,
|
||||
pub children: Vec<SimpleTag>,
|
||||
}
|
||||
|
||||
/// `Tag`.
|
||||
#[derive(Clone, Debug, Default, PartialEq)]
|
||||
pub struct Tag {
|
||||
pub targets: TagTargets,
|
||||
pub simple: Vec<SimpleTag>,
|
||||
}
|
||||
|
||||
/// `AttachedFile` (listed only; the payload is located by `data_offset`/`data_size`).
|
||||
#[derive(Clone, Debug, Default, PartialEq)]
|
||||
pub struct Attachment {
|
||||
pub uid: u64,
|
||||
pub name: String,
|
||||
pub media_type: String,
|
||||
pub description: Option<String>,
|
||||
pub data_offset: u64,
|
||||
pub data_size: u64,
|
||||
}
|
||||
|
||||
pub(crate) fn parse_info(d: &[u8]) -> SegmentInfo {
|
||||
let mut info = SegmentInfo::default();
|
||||
for e in Children::new(d) {
|
||||
match e.id {
|
||||
TIMESTAMP_SCALE => {
|
||||
let v = uint(e.data);
|
||||
if v > 0 {
|
||||
info.timestamp_scale = v;
|
||||
}
|
||||
}
|
||||
DURATION => info.duration = Some(float(e.data)).filter(|v| v.is_finite() && *v >= 0.0),
|
||||
TITLE => info.title = Some(string(e.data)),
|
||||
MUXING_APP => info.muxing_app = string(e.data),
|
||||
WRITING_APP => info.writing_app = string(e.data),
|
||||
DATE_UTC => info.date_utc = Some(int(e.data)),
|
||||
SEGMENT_UUID => info.segment_uuid = e.data.try_into().ok(),
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
info
|
||||
}
|
||||
|
||||
pub(crate) fn parse_seekhead(d: &[u8]) -> Vec<SeekEntry> {
|
||||
let mut out = Vec::new();
|
||||
for s in Children::new(d).filter(|e| e.id == SEEK) {
|
||||
let (mut id, mut pos) = (None, None);
|
||||
for e in Children::new(s.data) {
|
||||
match e.id {
|
||||
SEEK_ID => id = Some(uint(e.data) as u32),
|
||||
SEEK_POSITION => pos = Some(uint(e.data)),
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
if let (Some(id), Some(position)) = (id, pos) {
|
||||
out.push(SeekEntry { id, position });
|
||||
}
|
||||
}
|
||||
out
|
||||
}
|
||||
|
||||
pub(crate) fn parse_cues(d: &[u8], out: &mut Vec<CuePoint>) {
|
||||
for cp in Children::new(d).filter(|e| e.id == CUE_POINT) {
|
||||
let mut time = None;
|
||||
let mut positions = Vec::new();
|
||||
for e in Children::new(cp.data) {
|
||||
match e.id {
|
||||
CUE_TIME => time = Some(uint(e.data)),
|
||||
CUE_TRACK_POSITIONS => {
|
||||
let mut p = CuePosition { track: 0, cluster_position: u64::MAX, relative_position: None, block_number: None, duration: None };
|
||||
for c in Children::new(e.data) {
|
||||
match c.id {
|
||||
CUE_TRACK => p.track = uint(c.data),
|
||||
CUE_CLUSTER_POSITION => p.cluster_position = uint(c.data),
|
||||
CUE_RELATIVE_POSITION => p.relative_position = Some(uint(c.data)),
|
||||
CUE_BLOCK_NUMBER => p.block_number = Some(uint(c.data)),
|
||||
CUE_DURATION => p.duration = Some(uint(c.data)),
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
if p.track != 0 && p.cluster_position != u64::MAX {
|
||||
positions.push(p);
|
||||
}
|
||||
}
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
if let Some(time) = time
|
||||
&& !positions.is_empty()
|
||||
{
|
||||
out.push(CuePoint { time, positions });
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
fn parse_chapter(d: &[u8]) -> Chapter {
|
||||
let mut c = Chapter { uid: 0, string_uid: None, start_ns: 0, end_ns: None, hidden: false, enabled: true, displays: Vec::new(), children: Vec::new() };
|
||||
for e in Children::new(d) {
|
||||
match e.id {
|
||||
CHAPTER_UID => c.uid = uint(e.data),
|
||||
CHAPTER_STRING_UID => c.string_uid = Some(string(e.data)),
|
||||
CHAPTER_TIME_START => c.start_ns = uint(e.data),
|
||||
CHAPTER_TIME_END => c.end_ns = Some(uint(e.data)),
|
||||
CHAPTER_FLAG_HIDDEN => c.hidden = uint(e.data) != 0,
|
||||
CHAPTER_FLAG_ENABLED => c.enabled = uint(e.data) != 0,
|
||||
CHAPTER_DISPLAY => {
|
||||
let mut disp = ChapterDisplay::default();
|
||||
for x in Children::new(e.data) {
|
||||
match x.id {
|
||||
CHAP_STRING => disp.string = string(x.data),
|
||||
CHAP_LANGUAGE => disp.languages.push(string(x.data)),
|
||||
CHAP_LANGUAGE_BCP47 => disp.languages_bcp47.push(string(x.data)),
|
||||
CHAP_COUNTRY => disp.countries.push(string(x.data)),
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
if disp.languages.is_empty() {
|
||||
disp.languages.push("eng".into());
|
||||
}
|
||||
c.displays.push(disp);
|
||||
}
|
||||
CHAPTER_ATOM => c.children.push(parse_chapter(e.data)),
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
c
|
||||
}
|
||||
|
||||
pub(crate) fn parse_chapters(d: &[u8], out: &mut Vec<Edition>) {
|
||||
for ed in Children::new(d).filter(|e| e.id == EDITION_ENTRY) {
|
||||
let mut edition = Edition::default();
|
||||
for e in Children::new(ed.data) {
|
||||
match e.id {
|
||||
EDITION_UID => edition.uid = Some(uint(e.data)),
|
||||
EDITION_FLAG_HIDDEN => edition.hidden = uint(e.data) != 0,
|
||||
EDITION_FLAG_DEFAULT => edition.default = uint(e.data) != 0,
|
||||
EDITION_FLAG_ORDERED => edition.ordered = uint(e.data) != 0,
|
||||
CHAPTER_ATOM => edition.chapters.push(parse_chapter(e.data)),
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
out.push(edition);
|
||||
}
|
||||
}
|
||||
|
||||
fn parse_simple_tag(d: &[u8]) -> SimpleTag {
|
||||
let mut t = SimpleTag { name: String::new(), language: "und".into(), default: true, string: None, binary: None, children: Vec::new() };
|
||||
for e in Children::new(d) {
|
||||
match e.id {
|
||||
TAG_NAME => t.name = string(e.data),
|
||||
TAG_LANGUAGE => t.language = string(e.data),
|
||||
TAG_DEFAULT => t.default = uint(e.data) != 0,
|
||||
TAG_STRING => t.string = Some(string(e.data)),
|
||||
TAG_BINARY => t.binary = Some(e.data.to_vec()),
|
||||
SIMPLE_TAG => t.children.push(parse_simple_tag(e.data)),
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
t
|
||||
}
|
||||
|
||||
pub(crate) fn parse_tags(d: &[u8], out: &mut Vec<Tag>) {
|
||||
for tg in Children::new(d).filter(|e| e.id == TAG) {
|
||||
let mut tag = Tag::default();
|
||||
for e in Children::new(tg.data) {
|
||||
match e.id {
|
||||
TARGETS => {
|
||||
for x in Children::new(e.data) {
|
||||
match x.id {
|
||||
TARGET_TYPE_VALUE => tag.targets.type_value = uint(x.data),
|
||||
TARGET_TYPE => tag.targets.type_name = Some(string(x.data)),
|
||||
TAG_TRACK_UID => tag.targets.track_uids.push(uint(x.data)),
|
||||
TAG_EDITION_UID => tag.targets.edition_uids.push(uint(x.data)),
|
||||
TAG_CHAPTER_UID => tag.targets.chapter_uids.push(uint(x.data)),
|
||||
TAG_ATTACHMENT_UID => tag.targets.attachment_uids.push(uint(x.data)),
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
}
|
||||
SIMPLE_TAG => tag.simple.push(parse_simple_tag(e.data)),
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
out.push(tag);
|
||||
}
|
||||
}
|
||||
|
||||
fn parse_colour(d: &[u8]) -> Colour {
|
||||
let mut c = Colour::default();
|
||||
for e in Children::new(d) {
|
||||
let v = || Some(uint(e.data));
|
||||
match e.id {
|
||||
MATRIX_COEFFICIENTS => c.matrix_coefficients = v(),
|
||||
BITS_PER_CHANNEL => c.bits_per_channel = v(),
|
||||
CHROMA_SUBSAMPLING_HORZ => c.chroma_subsampling_horz = v(),
|
||||
CHROMA_SUBSAMPLING_VERT => c.chroma_subsampling_vert = v(),
|
||||
CB_SUBSAMPLING_HORZ => c.cb_subsampling_horz = v(),
|
||||
CB_SUBSAMPLING_VERT => c.cb_subsampling_vert = v(),
|
||||
CHROMA_SITING_HORZ => c.chroma_siting_horz = v(),
|
||||
CHROMA_SITING_VERT => c.chroma_siting_vert = v(),
|
||||
RANGE => c.range = v(),
|
||||
TRANSFER_CHARACTERISTICS => c.transfer_characteristics = v(),
|
||||
PRIMARIES => c.primaries = v(),
|
||||
MAX_CLL => c.max_cll = v(),
|
||||
MAX_FALL => c.max_fall = v(),
|
||||
MASTERING_METADATA => {
|
||||
let mut m = MasteringMetadata::default();
|
||||
for x in Children::new(e.data) {
|
||||
let f = float(x.data);
|
||||
match x.id {
|
||||
PRIMARY_R_X => m.primaries[0].0 = f,
|
||||
PRIMARY_R_Y => m.primaries[0].1 = f,
|
||||
PRIMARY_G_X => m.primaries[1].0 = f,
|
||||
PRIMARY_G_Y => m.primaries[1].1 = f,
|
||||
PRIMARY_B_X => m.primaries[2].0 = f,
|
||||
PRIMARY_B_Y => m.primaries[2].1 = f,
|
||||
WHITE_POINT_X => m.white_point.0 = f,
|
||||
WHITE_POINT_Y => m.white_point.1 = f,
|
||||
LUMINANCE_MAX => m.luminance_max = f,
|
||||
LUMINANCE_MIN => m.luminance_min = f,
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
c.mastering = Some(m);
|
||||
}
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
c
|
||||
}
|
||||
|
||||
fn parse_video(d: &[u8]) -> VideoInfo {
|
||||
let mut v = VideoInfo::default();
|
||||
for e in Children::new(d) {
|
||||
let u = uint(e.data);
|
||||
match e.id {
|
||||
PIXEL_WIDTH => v.pixel_width = u as u32,
|
||||
PIXEL_HEIGHT => v.pixel_height = u as u32,
|
||||
DISPLAY_WIDTH => v.display_width = Some(u as u32),
|
||||
DISPLAY_HEIGHT => v.display_height = Some(u as u32),
|
||||
DISPLAY_UNIT => v.display_unit = u,
|
||||
PIXEL_CROP_TOP => v.crop.0 = u as u32,
|
||||
PIXEL_CROP_BOTTOM => v.crop.1 = u as u32,
|
||||
PIXEL_CROP_LEFT => v.crop.2 = u as u32,
|
||||
PIXEL_CROP_RIGHT => v.crop.3 = u as u32,
|
||||
FLAG_INTERLACED => v.flag_interlaced = u,
|
||||
FIELD_ORDER => v.field_order = Some(u),
|
||||
STEREO_MODE => v.stereo_mode = Some(u),
|
||||
ALPHA_MODE => v.alpha_mode = u,
|
||||
COLOUR_SPACE => v.colour_space = e.data.try_into().ok(),
|
||||
COLOUR => v.colour = Some(parse_colour(e.data)),
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
v
|
||||
}
|
||||
|
||||
fn parse_audio(d: &[u8]) -> AudioInfo {
|
||||
let mut a = AudioInfo::default();
|
||||
for e in Children::new(d) {
|
||||
match e.id {
|
||||
SAMPLING_FREQUENCY => a.sampling_frequency = float(e.data),
|
||||
OUTPUT_SAMPLING_FREQUENCY => a.output_sampling_frequency = Some(float(e.data)),
|
||||
CHANNELS => a.channels = uint(e.data),
|
||||
BIT_DEPTH => a.bit_depth = Some(uint(e.data)),
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
a
|
||||
}
|
||||
|
||||
fn parse_encodings(d: &[u8]) -> Vec<ContentEncoding> {
|
||||
let mut list: Vec<(u64, ContentEncoding)> = Vec::new();
|
||||
for ce in Children::new(d).filter(|e| e.id == CONTENT_ENCODING) {
|
||||
let (mut order, mut scope, mut ty) = (0u64, 1u64, 0u64);
|
||||
let mut comp: Option<(u64, Vec<u8>)> = None;
|
||||
for e in Children::new(ce.data) {
|
||||
match e.id {
|
||||
CONTENT_ENCODING_ORDER => order = uint(e.data),
|
||||
CONTENT_ENCODING_SCOPE => scope = uint(e.data),
|
||||
CONTENT_ENCODING_TYPE => ty = uint(e.data),
|
||||
CONTENT_COMPRESSION => {
|
||||
let (mut algo, mut settings) = (0u64, Vec::new());
|
||||
for x in Children::new(e.data) {
|
||||
match x.id {
|
||||
CONTENT_COMP_ALGO => algo = uint(x.data),
|
||||
CONTENT_COMP_SETTINGS => settings = x.data.to_vec(),
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
comp = Some((algo, settings));
|
||||
}
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
let enc = if ty == 1 {
|
||||
ContentEncoding::Encryption { scope }
|
||||
} else {
|
||||
let (algo, settings) = comp.unwrap_or((0, Vec::new()));
|
||||
if algo == 3 { ContentEncoding::HeaderStripping { bytes: settings, scope } } else { ContentEncoding::Compression { algo, settings, scope } }
|
||||
};
|
||||
list.push((order, enc));
|
||||
}
|
||||
// highest order is applied last on muxing → first to undo; keep decode order
|
||||
list.sort_by_key(|e| std::cmp::Reverse(e.0));
|
||||
list.into_iter().map(|e| e.1).collect()
|
||||
}
|
||||
|
||||
pub(crate) fn parse_tracks(d: &[u8], timestamp_scale: u64) -> Vec<Track> {
|
||||
let g = gcd(timestamp_scale, 1_000_000_000);
|
||||
let timebase = (timestamp_scale / g, 1_000_000_000 / g);
|
||||
let mut out = Vec::new();
|
||||
for te in Children::new(d).filter(|e| e.id == TRACK_ENTRY) {
|
||||
let mut t = Track {
|
||||
number: 0,
|
||||
uid: 0,
|
||||
kind: TrackKind::Other(0),
|
||||
enabled: true,
|
||||
default: true,
|
||||
forced: false,
|
||||
hearing_impaired: false,
|
||||
lacing: true,
|
||||
default_duration_ns: None,
|
||||
name: String::new(),
|
||||
language: "eng".into(),
|
||||
language_bcp47: None,
|
||||
codec_id: String::new(),
|
||||
codec_private: Vec::new(),
|
||||
codec_name: String::new(),
|
||||
codec: crate::Codec::Other(String::new()),
|
||||
codec_delay_ns: 0,
|
||||
seek_pre_roll_ns: 0,
|
||||
video: None,
|
||||
audio: None,
|
||||
encodings: Vec::new(),
|
||||
timebase,
|
||||
samples: Vec::new(),
|
||||
sync: Vec::new(),
|
||||
};
|
||||
for e in Children::new(te.data) {
|
||||
match e.id {
|
||||
TRACK_NUMBER => t.number = uint(e.data),
|
||||
TRACK_UID => t.uid = uint(e.data),
|
||||
TRACK_TYPE => t.kind = TrackKind::from_type(uint(e.data)),
|
||||
FLAG_ENABLED => t.enabled = uint(e.data) != 0,
|
||||
FLAG_DEFAULT => t.default = uint(e.data) != 0,
|
||||
FLAG_FORCED => t.forced = uint(e.data) != 0,
|
||||
FLAG_HEARING_IMPAIRED => t.hearing_impaired = uint(e.data) != 0,
|
||||
FLAG_LACING => t.lacing = uint(e.data) != 0,
|
||||
DEFAULT_DURATION => t.default_duration_ns = Some(uint(e.data)).filter(|&v| v > 0),
|
||||
NAME => t.name = string(e.data),
|
||||
LANGUAGE => t.language = string(e.data),
|
||||
LANGUAGE_BCP47 => t.language_bcp47 = Some(string(e.data)),
|
||||
CODEC_ID => t.codec_id = string(e.data),
|
||||
CODEC_PRIVATE => t.codec_private = e.data.to_vec(),
|
||||
CODEC_NAME => t.codec_name = string(e.data),
|
||||
CODEC_DELAY => t.codec_delay_ns = uint(e.data),
|
||||
SEEK_PRE_ROLL => t.seek_pre_roll_ns = uint(e.data),
|
||||
VIDEO => t.video = Some(parse_video(e.data)),
|
||||
AUDIO => t.audio = Some(parse_audio(e.data)),
|
||||
CONTENT_ENCODINGS => t.encodings = parse_encodings(e.data),
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
if t.number == 0 {
|
||||
continue;
|
||||
}
|
||||
if t.kind == TrackKind::Audio && t.audio.is_none() {
|
||||
t.audio = Some(AudioInfo::default());
|
||||
}
|
||||
// CodecPrivate may itself be header-stripped (scope bit 2 = private data)
|
||||
for enc in &t.encodings {
|
||||
if let ContentEncoding::HeaderStripping { bytes, scope } = enc
|
||||
&& scope & 2 != 0
|
||||
{
|
||||
let mut p = bytes.clone();
|
||||
p.extend_from_slice(&t.codec_private);
|
||||
t.codec_private = p;
|
||||
}
|
||||
}
|
||||
t.codec = map_codec(&t.codec_id, &t.codec_private, t.audio.as_ref());
|
||||
out.push(t);
|
||||
}
|
||||
out
|
||||
}
|
||||
|
|
@ -0,0 +1,173 @@
|
|||
//! Random-access byte sources for the demuxer.
|
||||
|
||||
use std::cell::RefCell;
|
||||
use std::io::{self, Read, Seek, SeekFrom};
|
||||
use std::sync::Arc;
|
||||
|
||||
/// Random-access, read-only byte source (file, in-memory buffer, web Blob…).
|
||||
///
|
||||
/// Same shape as `filmcraft_isobmff::ByteSource`, so one adapter can serve both demuxers.
|
||||
pub trait ByteSource {
|
||||
/// Total length in bytes.
|
||||
fn len(&self) -> u64;
|
||||
/// Fill `buf` entirely with bytes starting at `offset`; fail with `UnexpectedEof` if past the end.
|
||||
fn read_at(&self, offset: u64, buf: &mut [u8]) -> io::Result<()>;
|
||||
/// True if the source holds no bytes.
|
||||
fn is_empty(&self) -> bool {
|
||||
self.len() == 0
|
||||
}
|
||||
}
|
||||
|
||||
fn eof() -> io::Error {
|
||||
io::Error::new(io::ErrorKind::UnexpectedEof, "read past end of byte source")
|
||||
}
|
||||
|
||||
impl ByteSource for [u8] {
|
||||
fn len(&self) -> u64 {
|
||||
<[u8]>::len(self) as u64
|
||||
}
|
||||
fn read_at(&self, offset: u64, buf: &mut [u8]) -> io::Result<()> {
|
||||
let start = usize::try_from(offset).map_err(|_| eof())?;
|
||||
let end = start.checked_add(buf.len()).ok_or_else(eof)?;
|
||||
let src = self.get(start..end).ok_or_else(eof)?;
|
||||
buf.copy_from_slice(src);
|
||||
Ok(())
|
||||
}
|
||||
}
|
||||
|
||||
impl ByteSource for Vec<u8> {
|
||||
fn len(&self) -> u64 {
|
||||
self.as_slice().len() as u64
|
||||
}
|
||||
fn read_at(&self, offset: u64, buf: &mut [u8]) -> io::Result<()> {
|
||||
self.as_slice().read_at(offset, buf)
|
||||
}
|
||||
}
|
||||
|
||||
impl<T: ByteSource + ?Sized> ByteSource for &T {
|
||||
fn len(&self) -> u64 {
|
||||
(**self).len()
|
||||
}
|
||||
fn read_at(&self, offset: u64, buf: &mut [u8]) -> io::Result<()> {
|
||||
(**self).read_at(offset, buf)
|
||||
}
|
||||
}
|
||||
|
||||
impl<T: ByteSource + ?Sized> ByteSource for Arc<T> {
|
||||
fn len(&self) -> u64 {
|
||||
(**self).len()
|
||||
}
|
||||
fn read_at(&self, offset: u64, buf: &mut [u8]) -> io::Result<()> {
|
||||
(**self).read_at(offset, buf)
|
||||
}
|
||||
}
|
||||
|
||||
impl<T: ByteSource + ?Sized> ByteSource for Box<T> {
|
||||
fn len(&self) -> u64 {
|
||||
(**self).len()
|
||||
}
|
||||
fn read_at(&self, offset: u64, buf: &mut [u8]) -> io::Result<()> {
|
||||
(**self).read_at(offset, buf)
|
||||
}
|
||||
}
|
||||
|
||||
#[cfg(not(target_arch = "wasm32"))]
|
||||
impl ByteSource for std::fs::File {
|
||||
fn len(&self) -> u64 {
|
||||
self.metadata().map(|m| m.len()).unwrap_or(0)
|
||||
}
|
||||
#[cfg(unix)]
|
||||
fn read_at(&self, offset: u64, buf: &mut [u8]) -> io::Result<()> {
|
||||
std::os::unix::fs::FileExt::read_exact_at(self, buf, offset)
|
||||
}
|
||||
#[cfg(windows)]
|
||||
fn read_at(&self, mut offset: u64, mut buf: &mut [u8]) -> io::Result<()> {
|
||||
use std::os::windows::fs::FileExt;
|
||||
while !buf.is_empty() {
|
||||
let n = self.seek_read(buf, offset)?;
|
||||
if n == 0 {
|
||||
return Err(eof());
|
||||
}
|
||||
offset += n as u64;
|
||||
buf = &mut buf[n..];
|
||||
}
|
||||
Ok(())
|
||||
}
|
||||
#[cfg(not(any(unix, windows)))]
|
||||
fn read_at(&self, offset: u64, buf: &mut [u8]) -> io::Result<()> {
|
||||
let mut f = self;
|
||||
f.seek(SeekFrom::Start(offset))?;
|
||||
f.read_exact(buf)
|
||||
}
|
||||
}
|
||||
|
||||
/// Adapts any `Read + Seek` (e.g. `BufReader<File>`, `Cursor<Vec<u8>>`) into a [`ByteSource`].
|
||||
///
|
||||
/// The length is taken once at construction; the stream position is tracked so sequential reads
|
||||
/// do not issue redundant seeks.
|
||||
pub struct ReadSeekSource<R> {
|
||||
inner: RefCell<(R, u64)>,
|
||||
len: u64,
|
||||
}
|
||||
|
||||
impl<R: Read + Seek> ReadSeekSource<R> {
|
||||
/// Wrap a reader; seeks to the end once to learn the length.
|
||||
pub fn new(mut r: R) -> io::Result<Self> {
|
||||
let len = r.seek(SeekFrom::End(0))?;
|
||||
Ok(Self { inner: RefCell::new((r, len)), len })
|
||||
}
|
||||
/// Unwrap the reader.
|
||||
pub fn into_inner(self) -> R {
|
||||
self.inner.into_inner().0
|
||||
}
|
||||
}
|
||||
|
||||
impl<R: Read + Seek> ByteSource for ReadSeekSource<R> {
|
||||
fn len(&self) -> u64 {
|
||||
self.len
|
||||
}
|
||||
fn read_at(&self, offset: u64, buf: &mut [u8]) -> io::Result<()> {
|
||||
if offset.checked_add(buf.len() as u64).is_none_or(|e| e > self.len) {
|
||||
return Err(eof());
|
||||
}
|
||||
let mut g = self.inner.borrow_mut();
|
||||
let (r, pos) = &mut *g;
|
||||
if *pos != offset {
|
||||
r.seek(SeekFrom::Start(offset))?;
|
||||
*pos = offset;
|
||||
}
|
||||
let res = r.read_exact(buf);
|
||||
match res {
|
||||
Ok(()) => {
|
||||
*pos = offset + buf.len() as u64;
|
||||
Ok(())
|
||||
}
|
||||
Err(e) => {
|
||||
*pos = u64::MAX; // unknown; force a seek next time
|
||||
Err(e)
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
|
||||
#[test]
|
||||
fn slice_and_reader_sources() {
|
||||
let v = vec![1u8, 2, 3, 4];
|
||||
let mut b = [0u8; 2];
|
||||
v.read_at(1, &mut b).unwrap();
|
||||
assert_eq!(b, [2, 3]);
|
||||
assert!(v.read_at(3, &mut b).is_err());
|
||||
assert!(v.read_at(u64::MAX, &mut b).is_err());
|
||||
let r = ReadSeekSource::new(io::Cursor::new(v.clone())).unwrap();
|
||||
assert_eq!(r.len(), 4);
|
||||
r.read_at(2, &mut b).unwrap();
|
||||
assert_eq!(b, [3, 4]);
|
||||
r.read_at(0, &mut b).unwrap();
|
||||
assert_eq!(b, [1, 2]);
|
||||
assert!(r.read_at(3, &mut b).is_err());
|
||||
}
|
||||
}
|
||||
|
|
@ -0,0 +1,280 @@
|
|||
//! Track descriptions (RFC 9559 §5.1.4).
|
||||
|
||||
use crate::codec::Codec;
|
||||
|
||||
/// Track type (`TrackType`).
|
||||
#[derive(Clone, Copy, Debug, PartialEq, Eq)]
|
||||
pub enum TrackKind {
|
||||
Video,
|
||||
Audio,
|
||||
Complex,
|
||||
Logo,
|
||||
Subtitle,
|
||||
Buttons,
|
||||
Control,
|
||||
Metadata,
|
||||
Other(u8),
|
||||
}
|
||||
|
||||
impl TrackKind {
|
||||
pub(crate) fn from_type(t: u64) -> TrackKind {
|
||||
match t {
|
||||
1 => TrackKind::Video,
|
||||
2 => TrackKind::Audio,
|
||||
3 => TrackKind::Complex,
|
||||
0x10 => TrackKind::Logo,
|
||||
0x11 => TrackKind::Subtitle,
|
||||
0x12 => TrackKind::Buttons,
|
||||
0x20 => TrackKind::Control,
|
||||
0x21 => TrackKind::Metadata,
|
||||
n => TrackKind::Other(n as u8),
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/// SMPTE ST 2086 mastering display metadata (`MasteringMetadata`). Chromaticities are CIE 1931 xy,
|
||||
/// luminance in cd/m².
|
||||
#[derive(Clone, Debug, Default, PartialEq)]
|
||||
pub struct MasteringMetadata {
|
||||
/// Red, green, blue primaries as (x, y).
|
||||
pub primaries: [(f64, f64); 3],
|
||||
pub white_point: (f64, f64),
|
||||
pub luminance_max: f64,
|
||||
pub luminance_min: f64,
|
||||
}
|
||||
|
||||
/// `Colour` element. Code points use ITU-T H.273 values (as Matroska does).
|
||||
#[derive(Clone, Debug, Default, PartialEq)]
|
||||
pub struct Colour {
|
||||
/// `MatrixCoefficients` (H.273; 2 = unspecified).
|
||||
pub matrix_coefficients: Option<u64>,
|
||||
pub bits_per_channel: Option<u64>,
|
||||
pub chroma_subsampling_horz: Option<u64>,
|
||||
pub chroma_subsampling_vert: Option<u64>,
|
||||
pub cb_subsampling_horz: Option<u64>,
|
||||
pub cb_subsampling_vert: Option<u64>,
|
||||
pub chroma_siting_horz: Option<u64>,
|
||||
pub chroma_siting_vert: Option<u64>,
|
||||
/// `Range`: 0 unspecified, 1 broadcast (limited), 2 full, 3 defined by matrix/transfer.
|
||||
pub range: Option<u64>,
|
||||
/// `TransferCharacteristics` (H.273).
|
||||
pub transfer_characteristics: Option<u64>,
|
||||
/// `Primaries` (H.273).
|
||||
pub primaries: Option<u64>,
|
||||
/// Maximum content light level (cd/m²).
|
||||
pub max_cll: Option<u64>,
|
||||
/// Maximum frame-average light level (cd/m²).
|
||||
pub max_fall: Option<u64>,
|
||||
pub mastering: Option<MasteringMetadata>,
|
||||
}
|
||||
|
||||
impl Colour {
|
||||
/// True if `Range` says full range.
|
||||
pub fn full_range(&self) -> bool {
|
||||
self.range == Some(2)
|
||||
}
|
||||
}
|
||||
|
||||
/// `Video` element.
|
||||
#[derive(Clone, Debug, Default, PartialEq)]
|
||||
pub struct VideoInfo {
|
||||
pub pixel_width: u32,
|
||||
pub pixel_height: u32,
|
||||
/// `DisplayWidth`/`DisplayHeight`, in `display_unit` (0 = pixels, 3 = display aspect ratio).
|
||||
pub display_width: Option<u32>,
|
||||
pub display_height: Option<u32>,
|
||||
pub display_unit: u64,
|
||||
/// Crop (top, bottom, left, right) in pixels.
|
||||
pub crop: (u32, u32, u32, u32),
|
||||
/// `FlagInterlaced`: 0 undetermined, 1 interlaced, 2 progressive.
|
||||
pub flag_interlaced: u64,
|
||||
/// `FieldOrder` (0 progressive, 1 tff, 2 undetermined, 6 bff, 9/14 interleaved).
|
||||
pub field_order: Option<u64>,
|
||||
pub stereo_mode: Option<u64>,
|
||||
/// `AlphaMode`: 1 if BlockAdditions carry alpha.
|
||||
pub alpha_mode: u64,
|
||||
/// `ColourSpace` FourCC (uncompressed video).
|
||||
pub colour_space: Option<[u8; 4]>,
|
||||
pub colour: Option<Colour>,
|
||||
}
|
||||
|
||||
impl VideoInfo {
|
||||
/// Pixel aspect ratio derived from the display size (pixels or aspect-ratio units), reduced.
|
||||
pub fn pixel_aspect(&self) -> (u32, u32) {
|
||||
let (pw, ph) = (self.pixel_width.saturating_sub(self.crop.2 + self.crop.3) as u64, self.pixel_height.saturating_sub(self.crop.0 + self.crop.1) as u64);
|
||||
let (Some(dw), Some(dh)) = (self.display_width, self.display_height) else { return (1, 1) };
|
||||
if pw == 0 || ph == 0 || dw == 0 || dh == 0 || !matches!(self.display_unit, 0 | 3) {
|
||||
return (1, 1);
|
||||
}
|
||||
// PAR = (dw/dh) / (pw/ph)
|
||||
let (n, d) = (dw as u64 * ph, dh as u64 * pw);
|
||||
let g = gcd(n, d);
|
||||
((n / g) as u32, (d / g) as u32)
|
||||
}
|
||||
}
|
||||
|
||||
pub(crate) fn gcd(mut a: u64, mut b: u64) -> u64 {
|
||||
while b != 0 {
|
||||
(a, b) = (b, a % b);
|
||||
}
|
||||
a.max(1)
|
||||
}
|
||||
|
||||
/// `Audio` element.
|
||||
#[derive(Clone, Debug, PartialEq)]
|
||||
pub struct AudioInfo {
|
||||
/// `SamplingFrequency` (Hz).
|
||||
pub sampling_frequency: f64,
|
||||
/// `OutputSamplingFrequency` (e.g. SBR output rate).
|
||||
pub output_sampling_frequency: Option<f64>,
|
||||
pub channels: u64,
|
||||
pub bit_depth: Option<u64>,
|
||||
}
|
||||
|
||||
impl Default for AudioInfo {
|
||||
fn default() -> Self {
|
||||
Self { sampling_frequency: 8000.0, output_sampling_frequency: None, channels: 1, bit_depth: None }
|
||||
}
|
||||
}
|
||||
|
||||
/// One `ContentEncoding`.
|
||||
#[derive(Clone, Debug, PartialEq)]
|
||||
pub enum ContentEncoding {
|
||||
/// `ContentCompAlgo` 3: the given bytes were stripped from the start of every frame.
|
||||
HeaderStripping { bytes: Vec<u8>, scope: u64 },
|
||||
/// Other compression (0 zlib, 1 bzlib, 2 lzo1x): not supported for reading frame data.
|
||||
Compression { algo: u64, settings: Vec<u8>, scope: u64 },
|
||||
/// Encryption (not supported for reading frame data).
|
||||
Encryption { scope: u64 },
|
||||
}
|
||||
|
||||
/// One entry of a track's sample (frame) table.
|
||||
#[derive(Clone, Copy, Debug, PartialEq, Eq)]
|
||||
pub struct Sample {
|
||||
/// Byte offset of the (stored) frame data in the file.
|
||||
pub offset: u64,
|
||||
/// Stored size (excluding header-stripped bytes; see [`Track::sample_size`]).
|
||||
pub size: u32,
|
||||
/// Presentation timestamp in the track timebase (`TimestampScale` ticks).
|
||||
pub pts: i64,
|
||||
/// Duration in timebase ticks (from `BlockDuration` or `DefaultDuration`), 0 if unknown.
|
||||
pub duration: u64,
|
||||
pub keyframe: bool,
|
||||
/// Index of the owning cluster in [`crate::MkvFile::clusters`].
|
||||
pub cluster: u32,
|
||||
/// Offset of the block element (SimpleBlock / BlockGroup) header.
|
||||
pub block_offset: u64,
|
||||
/// Index of this frame within a laced block.
|
||||
pub lace: u16,
|
||||
}
|
||||
|
||||
/// A Matroska track (`TrackEntry`) plus, once indexed, its sample table.
|
||||
#[derive(Clone, Debug)]
|
||||
pub struct Track {
|
||||
pub number: u64,
|
||||
pub uid: u64,
|
||||
pub kind: TrackKind,
|
||||
pub enabled: bool,
|
||||
pub default: bool,
|
||||
pub forced: bool,
|
||||
pub hearing_impaired: bool,
|
||||
pub lacing: bool,
|
||||
/// `DefaultDuration` in nanoseconds.
|
||||
pub default_duration_ns: Option<u64>,
|
||||
pub name: String,
|
||||
/// `Language` (ISO 639-2, default `eng`).
|
||||
pub language: String,
|
||||
pub language_bcp47: Option<String>,
|
||||
pub codec_id: String,
|
||||
pub codec_private: Vec<u8>,
|
||||
pub codec_name: String,
|
||||
pub codec: Codec,
|
||||
/// `CodecDelay` in nanoseconds.
|
||||
pub codec_delay_ns: u64,
|
||||
/// `SeekPreRoll` in nanoseconds.
|
||||
pub seek_pre_roll_ns: u64,
|
||||
pub video: Option<VideoInfo>,
|
||||
pub audio: Option<AudioInfo>,
|
||||
pub encodings: Vec<ContentEncoding>,
|
||||
/// Timebase numerator/denominator in seconds: `TimestampScale / 1e9`, reduced.
|
||||
pub timebase: (u64, u64),
|
||||
/// Frame table (empty unless the file was opened with indexing).
|
||||
pub samples: Vec<Sample>,
|
||||
pub(crate) sync: Vec<u32>,
|
||||
}
|
||||
|
||||
impl Track {
|
||||
/// Bytes prepended to every frame by header-stripping compression, if any.
|
||||
pub fn stripped_header(&self) -> Option<&[u8]> {
|
||||
self.encodings.iter().find_map(|e| match e {
|
||||
ContentEncoding::HeaderStripping { bytes, scope } if scope & 1 != 0 => Some(bytes.as_slice()),
|
||||
_ => None,
|
||||
})
|
||||
}
|
||||
|
||||
/// True if frames can be returned verbatim or with header stripping undone.
|
||||
pub fn frames_readable(&self) -> bool {
|
||||
self.encodings.iter().all(|e| match e {
|
||||
ContentEncoding::HeaderStripping { .. } => true,
|
||||
ContentEncoding::Compression { scope, .. } | ContentEncoding::Encryption { scope } => scope & 1 == 0,
|
||||
})
|
||||
}
|
||||
|
||||
/// Decoded frame size of sample `i` (stored size plus stripped header).
|
||||
pub fn sample_size(&self, i: usize) -> Option<u64> {
|
||||
let s = self.samples.get(i)?;
|
||||
Some(s.size as u64 + self.stripped_header().map_or(0, |h| h.len() as u64))
|
||||
}
|
||||
|
||||
/// Indices of keyframe samples, ascending by file order.
|
||||
pub fn keyframes(&self) -> &[u32] {
|
||||
&self.sync
|
||||
}
|
||||
|
||||
/// Timebase ticks → nanoseconds.
|
||||
pub fn ticks_to_ns(&self, t: i64) -> i64 {
|
||||
(t as i128 * self.timebase.0 as i128 * 1_000_000_000 / self.timebase.1 as i128) as i64
|
||||
}
|
||||
|
||||
/// Nanoseconds → timebase ticks (floor).
|
||||
pub fn ns_to_ticks(&self, ns: i64) -> i64 {
|
||||
let n = ns as i128 * self.timebase.1 as i128;
|
||||
let d = self.timebase.0 as i128 * 1_000_000_000;
|
||||
n.div_euclid(d) as i64
|
||||
}
|
||||
|
||||
/// Index of the last sample with `pts <= t` (in file order), or the first sample.
|
||||
pub fn sample_at_pts(&self, t: i64) -> Option<usize> {
|
||||
if self.samples.is_empty() {
|
||||
return None;
|
||||
}
|
||||
let mut best: Option<usize> = None;
|
||||
for (i, s) in self.samples.iter().enumerate() {
|
||||
if s.pts <= t && best.is_none_or(|b| s.pts >= self.samples[b].pts) {
|
||||
best = Some(i);
|
||||
}
|
||||
}
|
||||
Some(best.unwrap_or(0))
|
||||
}
|
||||
|
||||
/// Index of the keyframe with the greatest `pts <= t` (the first keyframe if none precedes `t`).
|
||||
pub fn keyframe_before_pts(&self, t: i64) -> Option<usize> {
|
||||
let mut best: Option<usize> = None;
|
||||
for &k in &self.sync {
|
||||
let s = &self.samples[k as usize];
|
||||
if s.pts <= t && best.is_none_or(|b| s.pts >= self.samples[b].pts) {
|
||||
best = Some(k as usize);
|
||||
}
|
||||
}
|
||||
best.or_else(|| self.sync.first().map(|&k| k as usize))
|
||||
}
|
||||
|
||||
/// Index of the nearest keyframe at or before sample `index` in file order.
|
||||
pub fn sync_sample_before(&self, index: usize) -> usize {
|
||||
match self.sync.binary_search(&(index as u32)) {
|
||||
Ok(i) => self.sync[i] as usize,
|
||||
Err(0) => 0,
|
||||
Err(i) => self.sync[i - 1] as usize,
|
||||
}
|
||||
}
|
||||
}
|
||||
Loading…
Reference in New Issue