fix(mp4): refuse a video track with no resolved dimensions

Resolution::pixels() returned (0, 0) for Unknown, and the MP4 sink wrote
it verbatim into tkhd (ISO/IEC 14496-12 8.3.2) and VisualSampleEntry
(12.1.3). Both fields are MANDATORY there, so unlike Matroska — which
omits the optional PixelWidth/PixelHeight elements — MP4 has nothing to
leave out. The result was a structurally complete file that passes every
container check, declares a 0x0 video track, cannot be rendered, and is
written with no error anywhere.

WHY IT WAS POSSIBLE, which is the part worth keeping:

pixels() previously fabricated 1920x1080 for Unknown. That was wrong but
playable, so this sink never needed a guard and the absence of one was
invisible. Changing the sentinel to (0, 0) moved the defect instead of
removing it — a zero PAIR still reads as a usable value, so the sink
stored it and serialised it.

The accessor's doc comment then ENUMERATED the callers it believed were
safe: "the Matroska sink omits the optional elements, the VobSub writer
omits its size: line, and no caller divides by either dimension." Two of
those three are true. MP4 was not on the list because MP4 has no guard
at all, and a prose list cannot enforce itself. mkv.rs's own comment
even states the principle — "the check belongs in the one accessor
rather than in each caller that remembered to write it" — and
labels/mod.rs still carried its own duplicate Unknown test long after
the accessor took that job over.

So: pixels() now returns Option. Not because Option is tidier, but
because every caller genuinely needs a DIFFERENT answer and the compiler
is the only thing that reliably makes them choose one. Matroska and the
metadata sinks take unwrap_or((0, 0)) with the reason stated at each
site; the VobSub path degrades to a palette-only .idx; MP4 fails with
E_MP4_UNKNOWN_RESOLUTION (9055).

Six call sites, not the five my first grep showed — I piped it through
`head` and acted on a truncated list. The compiler caught the sixth.
That is the same mistake as trusting a lens that reported silence.
This commit is contained in:
Matthew Jackson
2026-07-30 19:34:54 -07:00
parent 30bea12392
commit 9f25a4c454
8 changed files with 119 additions and 32 deletions
+4 -1
View File
@@ -179,7 +179,10 @@ impl Disc {
// the coded video frame the subpicture was authored against
// (720x480 NTSC / 720x576 PAL) so players place and scale the
// bitmap correctly.
let (vid_w, vid_h) = ts.video.resolution.pixels();
// format_palette guards on (0, 0) and omits its `size:` line,
// so an unresolved resolution degrades to a palette-only .idx
// rather than one claiming a 0x0 frame.
let (vid_w, vid_h) = ts.video.resolution.pixels().unwrap_or((0, 0));
let codec_data = dvd_title
.palette
.as_ref()
+42 -19
View File
@@ -945,15 +945,34 @@ impl Resolution {
/// handles it: the Matroska sink omits the optional PixelWidth/PixelHeight,
/// the VobSub `.idx` writer omits its `size:` line, and no caller divides by
/// either dimension.
pub fn pixels(&self) -> (u32, u32) {
/// `None` when the resolution never resolved.
///
/// Returns an `Option` rather than a sentinel because every caller has to
/// make a DIFFERENT decision and the compiler is the only thing that
/// reliably makes them: Matroska omits the optional PixelWidth/PixelHeight
/// elements, the metadata sinks report the dimensions as absent, and MP4
/// cannot do either — ISO/IEC 14496-12 makes width and height MANDATORY in
/// both `tkhd` (8.3.2) and VisualSampleEntry (12.1.3), so it must refuse.
///
/// This returned `(0, 0)` before, and that is the shape the sentinel
/// creates: `(0, 0)` looks like a usable pair, so the MP4 sink stored it
/// and serialised a 0x0 video track — a structurally complete file that no
/// player can render, written with no error. It returned a fabricated
/// 1920x1080 before that, which was wrong but at least playable, so the
/// sink had never needed a guard and the absence of one was invisible.
///
/// A doc comment listing which callers are safe cannot hold this: one did
/// not check, and a third kept its own duplicate `Unknown` test long after
/// the accessor took the job over.
pub fn pixels(&self) -> Option<(u32, u32)> {
match self {
Resolution::R480i | Resolution::R480p => (720, 480),
Resolution::R576i | Resolution::R576p => (720, 576),
Resolution::R720p => (1280, 720),
Resolution::R1080i | Resolution::R1080p => (1920, 1080),
Resolution::R2160p => (3840, 2160),
Resolution::R4320p => (7680, 4320),
Resolution::Unknown => (0, 0),
Resolution::R480i | Resolution::R480p => Some((720, 480)),
Resolution::R576i | Resolution::R576p => Some((720, 576)),
Resolution::R720p => Some((1280, 720)),
Resolution::R1080i | Resolution::R1080p => Some((1920, 1080)),
Resolution::R2160p => Some((3840, 2160)),
Resolution::R4320p => Some((7680, 4320)),
Resolution::Unknown => None,
}
}
@@ -3273,18 +3292,18 @@ mod tests {
fn unknown_resolution_reports_no_pixel_dimensions() {
assert_eq!(
Resolution::Unknown.pixels(),
(0, 0),
None,
"an unknown resolution must report no dimensions, not a fabricated default"
);
assert_eq!(Resolution::R480i.pixels(), (720, 480));
assert_eq!(Resolution::R480p.pixels(), (720, 480));
assert_eq!(Resolution::R576i.pixels(), (720, 576));
assert_eq!(Resolution::R576p.pixels(), (720, 576));
assert_eq!(Resolution::R720p.pixels(), (1280, 720));
assert_eq!(Resolution::R1080i.pixels(), (1920, 1080));
assert_eq!(Resolution::R1080p.pixels(), (1920, 1080));
assert_eq!(Resolution::R2160p.pixels(), (3840, 2160));
assert_eq!(Resolution::R4320p.pixels(), (7680, 4320));
assert_eq!(Resolution::R480i.pixels(), Some((720, 480)));
assert_eq!(Resolution::R480p.pixels(), Some((720, 480)));
assert_eq!(Resolution::R576i.pixels(), Some((720, 576)));
assert_eq!(Resolution::R576p.pixels(), Some((720, 576)));
assert_eq!(Resolution::R720p.pixels(), Some((1280, 720)));
assert_eq!(Resolution::R1080i.pixels(), Some((1920, 1080)));
assert_eq!(Resolution::R1080p.pixels(), Some((1920, 1080)));
assert_eq!(Resolution::R2160p.pixels(), Some((3840, 2160)));
assert_eq!(Resolution::R4320p.pixels(), Some((7680, 4320)));
}
/// Every `Unknown` variant that exposes a numeric accessor must report
@@ -3293,7 +3312,11 @@ mod tests {
/// than 0 fps, because callers divide by the numerator.
#[test]
fn no_unknown_variant_fabricates_a_numeric_value() {
assert_eq!(Resolution::Unknown.pixels(), (0, 0));
assert_eq!(
Resolution::Unknown.pixels(),
None,
"the strongest form of this rule: not even a zero pair, which reads as a usable value and was serialised into an MP4 as a 0x0 track"
);
assert_eq!(FrameRate::Unknown.as_fraction(), (0, 1));
assert_eq!(AudioChannels::Unknown.count(), 0);
assert_eq!(SampleRate::Unknown.hz(), 0.0);
+15 -3
View File
@@ -207,6 +207,12 @@ pub const E_MP4_INVALID: u16 = 9049;
/// `mp4://` video track is missing its codec-configuration record
/// (`hvcC`/`avcC`), without which the sample entry can't be written.
pub const E_MP4_MISSING_CODEC_PRIVATE: u16 = 9050;
/// `mp4://` video track has no resolved frame dimensions. ISO/IEC 14496-12
/// makes width and height mandatory in both `tkhd` (8.3.2) and
/// VisualSampleEntry (12.1.3), so unlike Matroska there is no element to omit:
/// the sink would have to write 0x0, producing a structurally complete file no
/// player can render. Refuse instead.
pub const E_MP4_UNKNOWN_RESOLUTION: u16 = 9055;
/// READ CAPACITY returned a short or overflowing transfer.
pub const E_DISC_CAPACITY_MALFORMED: u16 = 9047;
@@ -548,6 +554,9 @@ pub enum Error {
Mp4Invalid,
/// `mp4://` video track is missing its `hvcC`/`avcC` configuration record.
Mp4MissingCodecPrivate,
/// `mp4://` video track has no resolved frame dimensions. See
/// [`E_MP4_UNKNOWN_RESOLUTION`].
Mp4UnknownResolution,
PesFrameTooLarge {
size: usize,
},
@@ -737,6 +746,7 @@ impl Error {
Error::Mp4NoVideoTrack => E_MP4_NO_VIDEO_TRACK,
Error::Mp4Invalid => E_MP4_INVALID,
Error::Mp4MissingCodecPrivate => E_MP4_MISSING_CODEC_PRIVATE,
Error::Mp4UnknownResolution => E_MP4_UNKNOWN_RESOLUTION,
Error::PesFrameTooLarge { .. } => E_PES_FRAME_TOO_LARGE,
Error::PesInvalidMagic => E_PES_INVALID_MAGIC,
Error::PesTrackTooLarge { .. } => E_PES_TRACK_TOO_LARGE,
@@ -992,9 +1002,10 @@ impl From<Error> for std::io::Error {
// mp4:// demux errors: a malformed/truncated source file
// (E_MP4_INVALID), or a source whose tracks the mux can't use — no
// video track / missing codec-private config. All are invalid data.
E_MP4_NO_VIDEO_TRACK | E_MP4_INVALID | E_MP4_MISSING_CODEC_PRIVATE => {
std::io::ErrorKind::InvalidData
}
E_MP4_NO_VIDEO_TRACK
| E_MP4_INVALID
| E_MP4_MISSING_CODEC_PRIVATE
| E_MP4_UNKNOWN_RESOLUTION => std::io::ErrorKind::InvalidData,
// 9030 ExtentNotUnitAligned: a malformed/non-AACS-aligned
// extent was handed to the prefetch producer.
9030 => std::io::ErrorKind::InvalidInput,
@@ -1696,6 +1707,7 @@ mod tests {
(Error::Mp4NoVideoTrack, E_MP4_NO_VIDEO_TRACK),
(Error::Mp4Invalid, E_MP4_INVALID),
(Error::Mp4MissingCodecPrivate, E_MP4_MISSING_CODEC_PRIVATE),
(Error::Mp4UnknownResolution, E_MP4_UNKNOWN_RESOLUTION),
(Error::M2tsPacketMalformed, E_M2TS_PACKET_MALFORMED),
(Error::ExtentNotUnitAligned, E_EXTENT_NOT_UNIT_ALIGNED),
(Error::DiscCapacityMalformed, E_DISC_CAPACITY_MALFORMED),
+1 -5
View File
@@ -301,11 +301,7 @@ pub fn fill_defaults(titles: &mut [crate::disc::DiscTitle]) {
// Unknown resolution: pass (0, 0) so the label omits the
// resolution token rather than tagging it a fabricated
// 1080p.
let px = if matches!(v.resolution, crate::disc::Resolution::Unknown) {
(0, 0)
} else {
v.resolution.pixels()
};
let px = v.resolution.pixels().unwrap_or((0, 0));
v.label = generate_video_label(
&v.codec,
px,
+3 -1
View File
@@ -137,7 +137,9 @@ fn stream_json(s: &DiscStream) -> serde_json::Value {
use serde_json::json;
match s {
DiscStream::Video(v) => {
let (w, h) = v.resolution.pixels();
// Absent dimensions serialise as 0 here; the JSON consumer reads
// this as informational metadata, not as a mux input.
let (w, h) = v.resolution.pixels().unwrap_or((0, 0));
let (fps_num, fps_den) = v.frame_rate.as_fraction();
let mut o = json!({
"kind": "video",
+3 -1
View File
@@ -385,7 +385,9 @@ impl MkvTrack {
// that used to sit here existed only because `pixels()` fabricated a
// 1920x1080 default; it does not any more, so the check belongs in the
// one accessor rather than in each caller that remembered to write it.
let (w, h) = v.resolution.pixels();
// None -> 0, and the writer already omits the optional
// PixelWidth/PixelHeight elements on 0 (RFC 9559 5.1.4.1.28-29).
let (w, h) = v.resolution.pixels().unwrap_or((0, 0));
let (num, den) = v.frame_rate.as_fraction();
let default_duration_ns = if num > 0 {
(1_000_000_000u64 * den as u64) / num as u64
+49 -1
View File
@@ -316,7 +316,15 @@ impl<W: Write + Seek> Mp4Sink<W> {
.get(i)
.and_then(|c| c.clone())
.ok_or(crate::error::Error::Mp4MissingCodecPrivate)?;
let (w, h) = v.resolution.pixels();
// ISO/IEC 14496-12 makes width/height mandatory in tkhd
// (8.3.2) and VisualSampleEntry (12.1.3) — unlike Matroska
// there is no element to omit. Writing 0x0 yields a
// structurally complete file no player can render, with no
// error anywhere; refuse instead.
let (w, h) = v
.resolution
.pixels()
.ok_or(crate::error::Error::Mp4UnknownResolution)?;
tracks.push(Track {
media: Media::Video,
track_id,
@@ -1155,6 +1163,46 @@ mod tests {
assert_eq!(r.included, vec![1], "only the AC-3 audio is carried");
}
/// A video track whose resolution never resolved must FAIL the mux, not be
/// written as a 0x0 track.
///
/// ISO/IEC 14496-12 makes width and height mandatory in both `tkhd` (8.3.2)
/// and VisualSampleEntry (12.1.3), so unlike Matroska — which simply omits
/// the optional PixelWidth/PixelHeight elements — MP4 has nothing to leave
/// out. Writing zeros produces a structurally complete file that passes
/// every container check and that no player can render, with no error
/// anywhere: a wrong answer that looks like a successful rip.
///
/// `Resolution::pixels()` returns `Option` for this reason. It used to
/// return a fabricated 1920x1080, then `(0, 0)`; the zero pair reads as a
/// usable value, so this sink stored it and serialised it, and the guard
/// that two of the three sinks have was never needed here and so was never
/// written.
#[test]
fn a_video_track_with_no_resolved_resolution_is_an_error_not_a_zero_sized_track() {
let DiscStream::Video(mut v) = hevc_video() else {
unreachable!("hevc_video builds a video stream")
};
v.resolution = Resolution::Unknown;
let t = title(
vec![DiscStream::Video(v), audio(Codec::Ac3, "eng")],
vec![Some(vec![0x01, 0x02, 0x03]), None],
);
let err = match Mp4Sink::create(std::io::Cursor::new(Vec::new()), &t) {
Ok(_) => panic!("an unrenderable 0x0 track must not be written silently"),
Err(e) => e,
};
// `From<Error> for io::Error` stringifies as "E<code>[: ...]", so the
// code round-trips in the message.
assert!(
err.to_string()
.starts_with(&format!("E{}", crate::error::E_MP4_UNKNOWN_RESOLUTION)),
"the failure must name the missing dimensions, not a generic mux \
error; got {err}"
);
}
#[test]
fn no_video_track_is_an_error() {
let t = title(vec![audio(Codec::Ac3, "eng")], vec![None]);
+2 -1
View File
@@ -224,7 +224,8 @@ impl MapHeader {
let stream = match video {
Some(v) => {
let (width, height) = v.resolution.pixels();
// Informational map; absent dimensions report as 0.
let (width, height) = v.resolution.pixels().unwrap_or((0, 0));
StreamInfo {
codec: fvi_codec_id(v.codec),
width,