Struct PacketView 

#[non_exhaustive]
pub struct PacketView<'a> { /* private fields */ }

Borrowed view of one delivered packet (strict tier).

The view — including data — is valid only during the on_packet callback; the underlying packet is recycled as soon as the callback returns. Copy out what you keep.

Implementations

impl<'a> PacketView<'a>

pub fn stream_index(&self) -> usize

Output stream index (matches the on_stream_info entries).

pub fn pts(&self) -> i64

Presentation timestamp in time_base units, on the shared zero-based timeline (see applied_offset).

pub fn dts(&self) -> i64

Decode timestamp in time_base units, strictly increasing per stream. May be negative on non-anchor streams (a stream whose timeline starts earlier than the anchor stream keeps its true relative offset) and, with B-frames, ahead of pts reordering.

pub fn duration(&self) -> i64

Packet duration in time_base units. Always positive in the strict tier: the encoder’s duration is passed through; when absent it is derived (video: one CFR frame interval; audio: the codec frame size). A packet whose duration cannot be derived fails the job before delivery — this field is never a guess of zero.

pub fn time_base(&self) -> AVRational

Time base of this stream (identical to the stream’s PacketStreamInfo::time_base).

pub fn pts_us(&self) -> i64

pts in microseconds (exact rescale of the ticks).

pub fn dts_us(&self) -> i64

dts in microseconds.

pub fn duration_us(&self) -> i64

duration in microseconds.

pub fn applied_offset_us(&self) -> i64

applied_offset in microseconds.

pub fn is_key(&self) -> bool

Whether this packet is a fresh-decoder-safe random access point.

For H.264 this is true iff the access unit contains an IDR NAL unit — deliberately not the encoder’s raw AV_PKT_FLAG_KEY: with open-GOP encoding, encoders flag non-IDR recovery points as key frames, and feeding such a packet to a fresh decoder (the WebCodecs "key" contract) is not safe. Audio packets are always key.

pub fn applied_offset(&self) -> i64

The per-stream offset that was subtracted from pts/dts to move this stream onto the shared zero-based timeline, in this stream’s time_base units.

All streams share a single origin: the (dts, time_base) of the first delivered packet of the job. The anchor stream therefore starts at dts 0; other streams keep their true audio/video offset (which may be negative). original_ts = delivered_ts + applied_offset recovers the encoder timeline exactly (cross-time-base rounding is at most one tick per stream).

pub fn data(&self) -> &'a [u8]

The packet payload. H.264: one complete access unit, 4-byte length-prefixed (AVCC), parameter sets carried out-of-band in the stream configuration. AAC: one raw AAC frame.

Trait Implementations

impl<'a> Debug for PacketView<'a>

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.