mux/dts: monotonic per-frame PTS + real frame duration (fixes DVD DTS)

DVD packs several DTS core frames into one PES; the parser stamped every
access unit with that single PES PTS and duration_ns=None, so consecutive
frames collided on one timestamp — ffmpeg rejected the output as 'non
monotonically increasing dts to muxer: X >= X' (deep-decode = corrupt,
e.g. The Punisher). The UHD DTS-HD MA path (one AU per PES, distinct PTS)
was unaffected, which is why this only surfaced on DVD.

Parse the DTS core header for samples ((NBLKS+1)*32) and sample rate
(SFREQ, 48kHz fallback) to derive each AU's duration, and stamp a running
monotonic PTS: max(next_clock, own-core-PES PTS), then advance by the
frame duration. A later PES whose PTS is ahead of the clock still wins
(preserves the UHD per-PES attribution from da85f56/c49a180); frames
sharing one PES advance frame-by-frame instead of colliding.

Tests: the 3 that encoded 'same PES -> same PTS' now assert monotonic
advance; new dvd_many_cores_one_pes_are_strictly_monotonic reproduces the
Punisher bug; duration/SFREQ-fallback unit tests added.
This commit is contained in:
Matthew Jackson
2026-07-01 17:17:35 -07:00
parent b2c3540989
commit f122f08628
+183 -30
View File
@@ -38,6 +38,16 @@ pub struct DtsParser {
/// current `buf` start and rebased whenever bytes are drained from the
/// front.
pts_marks: Vec<(usize, i64)>,
/// Running monotonic timestamp: the PTS the NEXT emitted access unit must
/// not fall below. Each AU is stamped `max(next_pts_ns, its own core PES
/// PTS)` and then `next_pts_ns` advances by the AU's decoded duration. This
/// is what makes the DVD case — several DTS core frames packed in ONE PES,
/// all reporting that single PES PTS — advance frame-by-frame instead of
/// colliding on one timestamp (the "non monotonically increasing dts" mux
/// error). The UHD/DTS-HD MA case (one AU per PES with its own later PTS) is
/// unaffected: that PTS is `>= next_pts_ns`, so the AU keeps its own PES's
/// timestamp exactly as before. `PTS_UNSET` = no base captured yet.
next_pts_ns: i64,
}
impl Default for DtsParser {
@@ -52,9 +62,28 @@ impl DtsParser {
buf: Vec::with_capacity(32768),
pending_pts: 0,
pts_marks: Vec::new(),
next_pts_ns: PTS_UNSET,
}
}
/// Stamp an access unit with a monotonic PTS and advance the running clock.
/// `front` is the AU's own core-PES PTS (from [`front_pts`]); `dur_ns` its
/// decoded duration. Honors a forward jump to a new PES PTS (UHD multi-PES
/// AUs) but never emits below the running clock (DVD multi-AU-per-PES).
fn stamp_pts(&mut self, front: i64, dur_ns: i64) -> i64 {
let base = if front != PTS_UNSET && front > self.next_pts_ns {
front
} else if self.next_pts_ns != PTS_UNSET {
self.next_pts_ns
} else if front != PTS_UNSET {
front
} else {
0
};
self.next_pts_ns = base + dur_ns;
base
}
/// Drop `n` bytes from the front of `buf` and rebase the PTS markers so
/// their offsets stay relative to the new buffer start. A marker that now
/// sits at or before offset 0 is clamped to 0 (it still covers the front).
@@ -137,6 +166,9 @@ impl CodecParser for DtsParser {
self.buf.clear();
self.pts_marks.clear();
self.pending_pts = PTS_UNSET;
// A concealed gap is a timeline discontinuity: let the post-gap AU
// re-base to its own (later) PES PTS rather than the pre-gap clock.
self.next_pts_ns = PTS_UNSET;
}
if pes.data.is_empty() {
return Vec::new();
@@ -267,10 +299,13 @@ impl CodecParser for DtsParser {
};
let au: Vec<u8> = self.buf[..au_end].to_vec();
// The AU takes the PTS of the PES covering its first byte — its own
// core's PES, even if that PES preceded the one(s) carrying its
// extensions or the next core.
let au_pts = self.front_pts();
// The AU's own core PES PTS (the PES covering its first byte, even if
// that PES preceded the one(s) carrying its extensions or the next
// core), stamped monotonically: honored when it advances past the
// running clock (UHD one-AU-per-PES), but never allowed to collide
// with the previous AU when several cores share ONE PES (DVD).
let dur_ns = dts_core_duration_ns(&au) as i64;
let au_pts = self.stamp_pts(self.front_pts(), dur_ns);
frames.push(Frame {
discontinuity: false,
coding: None,
@@ -278,7 +313,7 @@ impl CodecParser for DtsParser {
pts_ns: au_pts,
keyframe: true,
data: au,
duration_ns: None,
duration_ns: Some(dur_ns as u64),
});
self.drain_front(au_end);
// After draining, the marker covering the new front (if any) carries
@@ -324,9 +359,9 @@ impl CodecParser for DtsParser {
}
// The final AU's PTS is the PES covering the buffer front (its core's
// PES). Fall back to pending_pts, clamping the sentinel to 0.
let front = self.front_pts();
let pts_ns = if front == PTS_UNSET { 0 } else { front };
let au = std::mem::take(&mut self.buf);
let dur_ns = dts_core_duration_ns(&au) as i64;
let pts_ns = self.stamp_pts(self.front_pts(), dur_ns);
self.pts_marks.clear();
vec![Frame {
discontinuity: false,
@@ -335,7 +370,7 @@ impl CodecParser for DtsParser {
pts_ns,
keyframe: true,
data: au,
duration_ns: None,
duration_ns: Some(dur_ns as u64),
}]
}
@@ -501,6 +536,49 @@ fn dts_core_frame_size(data: &[u8]) -> usize {
fsize + 1
}
/// DTS core `SFREQ` → sample rate (Hz). 4-bit index; reserved/invalid entries
/// fall back to 48 kHz (the DVD/UHD norm) so a bogus value never yields a zero
/// rate (division) or a wildly wrong frame duration.
const DTS_CORE_SAMPLE_RATES: [u32; 16] = [
48_000, // 0: invalid → fallback
8_000, 16_000, 32_000, 48_000, // 4: invalid → fallback
48_000, // 5: invalid → fallback
11_025, 22_050, 44_100, 48_000, // 9: invalid → fallback
48_000, // 10: invalid → fallback
12_000, 24_000, 48_000, 96_000, 192_000,
];
/// Samples in one DTS core frame: `(NBLKS + 1) * 32`. `NBLKS` (7 bits) is the
/// core-header PCM-sample-block count — the same field ffmpeg's `dca` decoder
/// uses to timestamp frames. Bit layout after the 32-bit sync: FTYPE(1) SHORT(5)
/// CPF(1) **NBLKS(7)** FSIZE(14) …, so NBLKS = byte4 bit0 + byte5 bits7-2.
fn dts_core_samples(data: &[u8]) -> u32 {
if data.len() < CORE_HEADER_MIN_BYTES {
return 512; // typical; only reached on a truncated header
}
let nblks = ((data[4] as u32 & 0x01) << 6) | (data[5] as u32 >> 2);
(nblks + 1) * 32
}
/// DTS core sample rate (Hz) from `SFREQ` (4 bits: byte8 bits5-2), with a 48 kHz
/// fallback for reserved indices.
fn dts_core_sample_rate(data: &[u8]) -> u32 {
if data.len() < CORE_HEADER_MIN_BYTES {
return 48_000;
}
let sfreq = (data[8] as usize >> 2) & 0x0F;
DTS_CORE_SAMPLE_RATES[sfreq]
}
/// Duration of one DTS core access unit in nanoseconds: `samples / rate`,
/// rounded to nearest. This is what lets consecutive core frames packed in a
/// single DVD PES advance monotonically instead of colliding on one PES PTS.
fn dts_core_duration_ns(data: &[u8]) -> u64 {
let samples = dts_core_samples(data) as u64;
let rate = dts_core_sample_rate(data) as u64;
(samples * 1_000_000_000 + rate / 2) / rate
}
#[cfg(test)]
mod tests {
use super::*;
@@ -521,12 +599,22 @@ mod tests {
let fsize = size - 1;
let mut data = vec![0u8; size];
data[0..4].copy_from_slice(&DTS_CORE_SYNC);
data[5] = (data[5] & 0xFC) | ((fsize >> 12) & 0x03) as u8;
// NBLKS = 15 → (15+1)*32 = 512 samples/frame (the DVD/UHD DTS-core norm).
// NBLKS is byte4 bit0 + byte5 bits7-2; here byte4 bit0 = 0, byte5 = 15<<2.
data[5] = (15u8 << 2) | ((fsize >> 12) & 0x03) as u8;
data[6] = ((fsize >> 4) & 0xFF) as u8;
data[7] = (data[7] & 0x0F) | (((fsize & 0x0F) << 4) as u8);
// SFREQ = 13 → 48 kHz (byte8 bits5-2). Only when the header byte exists.
if size > 8 {
data[8] = 13u8 << 2;
}
data
}
// 512 samples @ 48 kHz, rounded to nearest ns — the duration make_dts_core
// frames advance by. (512 * 1e9 + 24000) / 48000 = 10_666_667 ns.
const DTS_CORE_DUR_NS: i64 = (512 * 1_000_000_000 + 48_000 / 2) / 48_000;
/// A real DTS-HD EXSS substream of `total` bytes (short header form), with an
/// optional false DTS core syncword embedded in its payload (decoding to a
/// plausible core size) — to prove precise sizing, not a payload scan, bounds
@@ -657,10 +745,11 @@ mod tests {
}
#[test]
fn two_cores_back_to_back_emit_first_on_boundary() {
// The first complete unit is emitted as soon as the next core sync is
// seen; the second is held until flush. Both cores arrived in ONE PES,
// so both legitimately carry that PES's PTS.
fn two_cores_back_to_back_advance_within_one_pes() {
// Both cores arrive in ONE PES — the DVD layout. AU1 keeps the PES PTS;
// AU2 must ADVANCE by one frame duration to stay monotonic. Reusing the
// single PES PTS for both was the "non monotonically increasing dts to
// muxer" bug (fixed for 1.2.1).
let mut parser = DtsParser::new();
let mut stream = make_dts_core(512);
stream.extend_from_slice(&make_dts_core(640));
@@ -668,13 +757,18 @@ mod tests {
assert_eq!(f.len(), 1);
assert_eq!(f[0].data.len(), 512);
assert_eq!(f[0].pts_ns, pts_to_ns(90000), "AU1 keeps its PES PTS");
assert_eq!(
f[0].duration_ns,
Some(DTS_CORE_DUR_NS as u64),
"AU1 carries a real frame duration (was None)"
);
let tail = parser.flush();
assert_eq!(tail.len(), 1);
assert_eq!(tail[0].data.len(), 640);
assert_eq!(
tail[0].pts_ns,
pts_to_ns(90000),
"AU2 came in the same PES → same PTS"
pts_to_ns(90000) + DTS_CORE_DUR_NS,
"AU2 in the same PES advances one frame duration (monotonic)"
);
}
@@ -689,31 +783,33 @@ mod tests {
let mut parser = DtsParser::new();
// PES A: just core1 (held — no following core yet).
let f0 = parser.parse(&make_pes(make_dts_core(512), Some(100)));
let f0 = parser.parse(&make_pes(make_dts_core(512), Some(90000)));
assert!(f0.is_empty(), "core1 held awaiting next core");
// PES B: core2 + core3. Closes AU1 (core1) and AU2 (core2).
// PES B (realistically LATER — far past one frame): core2 + core3.
// Closes AU1 (core1) and AU2 (core2).
let mut pes_b = make_dts_core(600);
pes_b.extend_from_slice(&make_dts_core(640));
let f = parser.parse(&make_pes(pes_b, Some(200)));
let f = parser.parse(&make_pes(pes_b, Some(190000)));
assert_eq!(f.len(), 2, "AU1 and AU2 both close in this call");
assert_eq!(f[0].data.len(), 512, "AU1 = core1");
assert_eq!(
f[0].pts_ns,
pts_to_ns(100),
pts_to_ns(90000),
"AU1 keeps PES A's PTS, not the later PES B PTS"
);
assert_eq!(f[1].data.len(), 600, "AU2 = core2");
assert_eq!(
f[1].pts_ns,
pts_to_ns(200),
"AU2's core arrived in PES B → PES B PTS"
pts_to_ns(190000),
"AU2's core is in PES B → attributes to PES B PTS (ahead of the clock, so it wins)"
);
// AU3 (core3) drains on flush, also PES B PTS.
// AU3 (core3) drains on flush — 2nd core in PES B, so it advances one
// frame duration from AU2 to stay monotonic.
let tail = parser.flush();
assert_eq!(tail.len(), 1);
assert_eq!(tail[0].pts_ns, pts_to_ns(200));
assert_eq!(tail[0].pts_ns, pts_to_ns(190000) + DTS_CORE_DUR_NS);
}
#[test]
@@ -728,23 +824,80 @@ mod tests {
// AU2 (core2) held awaiting a third core.
let mut pes_a = make_dts_core(512);
pes_a.extend_from_slice(&make_dts_core(600));
let f = parser.parse(&make_pes(pes_a, Some(100)));
let f = parser.parse(&make_pes(pes_a, Some(90000)));
assert_eq!(f.len(), 1);
assert_eq!(f[0].pts_ns, pts_to_ns(100), "AU1 PES A PTS");
assert_eq!(f[0].pts_ns, pts_to_ns(90000), "AU1 PES A PTS");
// PES B: core3 — closes AU2 (core2). AU2's core was in PES A.
let f2 = parser.parse(&make_pes(make_dts_core(640), Some(200)));
// PES B (realistically later): core3 — closes AU2 (core2). AU2's core
// was in PES A, so it is PES A's 2nd frame: it advances one frame
// duration from AU1 (still on PES A's timeline, and monotonic) — it does
// NOT inherit the closing PES B PTS.
let f2 = parser.parse(&make_pes(make_dts_core(640), Some(190000)));
assert_eq!(f2.len(), 1);
assert_eq!(f2[0].data.len(), 600, "AU2 = core2");
assert_eq!(
f2[0].pts_ns,
pts_to_ns(100),
"AU2's core arrived in PES A → must keep PES A's PTS, not PES B's"
pts_to_ns(90000) + DTS_CORE_DUR_NS,
"AU2 = 2nd frame of PES A → PES A base + one frame, not the closing PES B PTS"
);
// AU3 = core3, whose own core is in PES B → jumps to PES B's PTS.
let tail = parser.flush();
assert_eq!(tail.len(), 1);
assert_eq!(tail[0].pts_ns, pts_to_ns(200), "AU3 = core3, PES B PTS");
assert_eq!(
tail[0].pts_ns,
pts_to_ns(190000),
"AU3 = core3 in PES B → PES B PTS"
);
}
#[test]
fn dvd_many_cores_one_pes_are_strictly_monotonic() {
// Punisher-DVD reproduction: a single PES carrying SEVERAL DTS core
// frames (the DVD packing) must emit STRICTLY-increasing PTSs. The old
// code stamped every AU with the one PES PTS, which ffmpeg rejected as
// "non monotonically increasing dts to muxer: X >= X".
let mut parser = DtsParser::new();
let mut stream = Vec::new();
for _ in 0..6 {
stream.extend_from_slice(&make_dts_core(512));
}
let mut frames = parser.parse(&make_pes(stream, Some(90000)));
frames.extend(parser.flush());
assert_eq!(frames.len(), 6, "all six cores emitted");
for w in frames.windows(2) {
assert!(
w[1].pts_ns > w[0].pts_ns,
"consecutive DTS AUs must STRICTLY increase: {} !> {}",
w[1].pts_ns,
w[0].pts_ns
);
}
// Each advances by exactly one frame duration, and carries that duration.
assert_eq!(frames[0].pts_ns, pts_to_ns(90000));
assert_eq!(frames[1].pts_ns, pts_to_ns(90000) + DTS_CORE_DUR_NS);
assert_eq!(frames[5].pts_ns, pts_to_ns(90000) + 5 * DTS_CORE_DUR_NS);
for f in &frames {
assert_eq!(f.duration_ns, Some(DTS_CORE_DUR_NS as u64));
}
}
#[test]
fn dts_core_duration_512_samples_48khz() {
// NBLKS=15 → (15+1)*32 = 512 samples; SFREQ=13 → 48 kHz.
let core = make_dts_core(512);
assert_eq!(dts_core_samples(&core), 512);
assert_eq!(dts_core_sample_rate(&core), 48_000);
assert_eq!(dts_core_duration_ns(&core), DTS_CORE_DUR_NS as u64);
}
#[test]
fn dts_core_sfreq_reserved_falls_back_to_48k() {
// A bogus SFREQ index must never yield a zero rate (division) — fall
// back to 48 kHz.
let mut core = make_dts_core(512);
core[8] = 0; // SFREQ = 0 (reserved)
assert_eq!(dts_core_sample_rate(&core), 48_000);
}
/// Build a minimal DTS-HD extension substream of `size` bytes (just the