dvb_si/descriptors/

any.rs

//! Unified descriptor dispatch: [`AnyDescriptor`] + [`parse_loop`].
//!
//! [`AnyDescriptor`] is generated from a single declarative list
//! (`declare_descriptors!`) — one line per crate-implemented descriptor tag.
//! The list is the single source of truth: it produces the enum, the
//! `From<T>` conversions, and the tag → type dispatcher, and a drift test
//! pins each tag literal to the type's [`crate::traits::DescriptorDef::TAG`].
//!
//! [`parse_loop`] lazily walks a raw descriptor loop (the variable-length
//! `descriptor()` sequence inside tables), yielding one [`AnyDescriptor`] per
//! entry. It never panics: a malformed entry whose length is known yields an
//! `Err` and iteration continues; a truncated final header/body yields one
//! final `Err` and then fuses.
//!
//! ```
//! use dvb_si::descriptors::{parse_loop, AnyDescriptor};
//!
//! // A two-descriptor loop: short_event (tag 0x4D, "eng" / "News") then an
//! // unrecognised private tag 0xA7 — the walker yields a typed value for the
//! // first and `Unknown` for the second, never panicking.
//! let loop_bytes = [
//!     0x4D, 0x07, b'e', b'n', b'g', 0x02, b'H', b'i', 0x00, // short_event
//!     0xA7, 0x02, 0xCA, 0xFE,                               // unknown 0xA7
//! ];
//! let items: Vec<_> = parse_loop(&loop_bytes).collect();
//! assert_eq!(items.len(), 2);
//! match items[0].as_ref().unwrap() {
//!     AnyDescriptor::ShortEvent(se) => {
//!         assert_eq!(se.language_code.as_str(), "eng");
//!         assert_eq!(se.event_name.decode(), "Hi");
//!     }
//!     other => panic!("expected ShortEvent, got {other:?}"),
//! }
//! assert!(matches!(items[1].as_ref().unwrap(), AnyDescriptor::Unknown { tag: 0xA7, .. }));
//! ```
//!
//! # Adding a descriptor
//!
//! 1. Create the module with the wire layout, a `pub const TAG: u8`, and the
//!    symmetric [`broadcast_common::Parse`]/[`broadcast_common::Serialize`] impls +
//!    round-trip tests (copy an existing module).
//! 2. `impl DescriptorDef` for the type (`TAG` from the module const, `NAME`
//!    in SCREAMING_SNAKE without the `_descriptor` suffix).
//! 3. Add one line to the `declare_descriptors!` invocation below — the enum
//!    variant, dispatcher arm, and drift test are generated from it.
//! 4. The integration completeness test walks the generated
//!    [`AnyDescriptor::DISPATCHED_TAGS`] automatically — no test edits needed.

use alloc::boxed::Box;
#[cfg(feature = "serde")]
use alloc::string::ToString;

/// Declares [`AnyDescriptor`] + its dispatcher from one tag list.
///
/// Each line is `Variant = 0xTAG => module::Type[<'a>]`. The optional trailing
/// `@no_dispatch …` section adds variants that are NOT reachable from the
/// generated dispatcher (private / context-dependent tags such as 0x83
/// logical_channel) — the variant exists for callers that opt in via the
/// registry, but `dispatch` never produces it.
macro_rules! declare_descriptors {
    (
        $lt:lifetime;
        $( $variant:ident = $tag:literal => $($path:ident)::+ $(<$plt:lifetime>)? ),+ $(,)?
        $( ; @no_dispatch $( $nd_variant:ident => $($nd_path:ident)::+ $(<$nd_plt:lifetime>)? ),+ $(,)? )?
    ) => {
        /// Every crate-implemented descriptor, plus an `Unknown` fallthrough.
        ///
        /// serde uses external tagging with camelCase variant keys —
        /// a parsed short_event_descriptor serializes as `{"shortEvent": {…}}`.
        /// Variant names map 1:1 to the descriptor modules; see each module
        /// for the wire layout.
        #[derive(Debug)]
        #[cfg_attr(feature = "serde", derive(serde::Serialize))]
        #[cfg_attr(feature = "serde", serde(rename_all = "camelCase"))]
        // Every variant is covariant in `$lt`: typed variants hold only
        // lifetime-parametrised views, `Unknown` holds `&$lt [u8]`, and the
        // `Other` value is a `'static` `Box<dyn DescriptorObject>`. The derive
        // accepts the `'static` field unchanged, so the impl is sound.
        #[cfg_attr(feature = "yoke", derive(yoke::Yokeable))]
        #[non_exhaustive]
        pub enum AnyDescriptor<$lt> {
            $(
                #[allow(missing_docs)]
                $variant($($path)::+ $(<$plt>)?),
            )+
            $($(
                #[allow(missing_docs)]
                $nd_variant($($nd_path)::+ $(<$nd_plt>)?),
            )+)?
            /// Runtime-registered custom descriptor (see [`DescriptorRegistry`]).
            ///
            /// [`DescriptorRegistry`]: crate::descriptors::registry::DescriptorRegistry
            Other {
                /// The raw descriptor_tag byte.
                tag: u8,
                /// The parsed, type-erased descriptor value. Call `downcast_ref`
                /// on it (see [`DescriptorObject`](crate::descriptors::registry::DescriptorObject))
                /// to recover the concrete type.
                #[cfg_attr(
                    feature = "serde",
                    serde(serialize_with = "crate::descriptors::registry::serialize_erased")
                )]
                value: Box<dyn crate::descriptors::registry::DescriptorObject>,
            },
            /// Tag with no typed implementation; `body` is the payload sans
            /// the 2-byte (tag, length) header.
            Unknown {
                /// The raw descriptor_tag byte.
                tag: u8,
                /// The raw payload bytes (descriptor_length bytes).
                body: &$lt [u8],
            },
        }

        $(
            impl<$lt> From<$($path)::+ $(<$plt>)?> for AnyDescriptor<$lt> {
                fn from(d: $($path)::+ $(<$plt>)?) -> Self {
                    Self::$variant(d)
                }
            }
        )+
        $($(
            impl<$lt> From<$($nd_path)::+ $(<$nd_plt>)?> for AnyDescriptor<$lt> {
                fn from(d: $($nd_path)::+ $(<$nd_plt>)?) -> Self {
                    Self::$nd_variant(d)
                }
            }
        )+)?

        impl<$lt> AnyDescriptor<$lt> {
            /// Every tag the generated dispatcher routes (excludes `@no_dispatch`
            /// variants and [`AnyDescriptor::Unknown`]).
            pub const DISPATCHED_TAGS: &'static [u8] = &[$($tag),+];

            /// Diagnostic name of the contained descriptor — the type's
            /// [`DescriptorDef::NAME`](crate::traits::DescriptorDef::NAME)
            /// (`"SHORT_EVENT"`, `"NETWORK_NAME"`, …); `"CUSTOM"` for
            /// [`AnyDescriptor::Other`] (runtime-registered) and `"UNKNOWN"`
            /// for [`AnyDescriptor::Unknown`].
            #[must_use]
            pub fn name(&self) -> &'static str {
                match self {
                    $(
                        Self::$variant(_) =>
                            <$($path)::+ as crate::traits::DescriptorDef>::NAME,
                    )+
                    $($(
                        Self::$nd_variant(_) =>
                            <$($nd_path)::+ as crate::traits::DescriptorDef>::NAME,
                    )+)?
                    Self::Other { .. } => "CUSTOM",
                    Self::Unknown { .. } => "UNKNOWN",
                }
            }

            /// Re-serialize the typed descriptor to its full wire form
            /// (2-byte header included). `None` for [`AnyDescriptor::Other`]
            /// (type-erased); `Unknown` re-emits its raw bytes. Used by the
            /// real-capture byte-identity survey.
            #[doc(hidden)]
            pub fn reserialize(&self) -> Option<core::result::Result<alloc::vec::Vec<u8>, alloc::string::String>> {
                use broadcast_common::Serialize;
                fn go<S: Serialize>(d: &S) -> core::result::Result<alloc::vec::Vec<u8>, alloc::string::String>
                where
                    S::Error: core::fmt::Debug,
                {
                    let mut out = alloc::vec![0u8; d.serialized_len()];
                    d.serialize_into(&mut out).map_err(|e| alloc::format!("{e:?}"))?;
                    Ok(out)
                }
                match self {
                    $( Self::$variant(d) => Some(go(d)), )+
                    $($( Self::$nd_variant(d) => Some(go(d)), )+)?
                    Self::Other { .. } => None,
                    Self::Unknown { tag, body } => {
                        Some(
                            broadcast_common::len::fit_u8(body.len(), "descriptor_length")
                                .map(|n| {
                                    let mut v = alloc::vec![*tag, n];
                                    v.extend_from_slice(body);
                                    v
                                })
                                .map_err(|e| alloc::format!("{e:?}")),
                        )
                    }
                }
            }

            /// Parse one full descriptor (2-byte header included) by its tag.
            ///
            /// `None` means no typed implementation exists for `tag` (the
            /// caller turns that into [`AnyDescriptor::Unknown`]). `Some(Err)`
            /// is a typed parse failure for a recognised tag.
            pub(crate) fn dispatch(tag: u8, full: &$lt [u8]) -> Option<crate::Result<Self>> {
                use broadcast_common::Parse;
                match tag {
                    $(
                        $tag => Some(<$($path)::+>::parse(full).map(Self::$variant)),
                    )+
                    _ => None,
                }
            }
        }

        #[cfg(test)]
        mod macro_drift {
            #[test]
            fn tag_literals_match_descriptor_def() {
                use crate::traits::DescriptorDef;
                $(
                    assert_eq!(
                        $tag,
                        <$($path)::+ as DescriptorDef>::TAG,
                        concat!("tag literal drift for ", stringify!($variant)),
                    );
                    assert!(
                        !<$($path)::+ as DescriptorDef>::NAME.is_empty(),
                        concat!("empty NAME for ", stringify!($variant)),
                    );
                )+
                $($(
                    assert!(
                        !<$($nd_path)::+ as DescriptorDef>::NAME.is_empty(),
                        concat!("empty NAME for ", stringify!($nd_variant)),
                    );
                )+)?
            }
        }
    };
}

declare_descriptors! {'a;
    // MPEG-2 systems descriptors (ISO/IEC 13818-1) used outside table context.
    VideoStream = 0x02 => crate::descriptors::video_stream::VideoStreamDescriptor,
    AudioStream = 0x03 => crate::descriptors::audio_stream::AudioStreamDescriptor,
    Hierarchy = 0x04 => crate::descriptors::hierarchy::HierarchyDescriptor,
    Registration = 0x05 => crate::descriptors::registration::RegistrationDescriptor<'a>,
    DataStreamAlignment = 0x06 => crate::descriptors::data_stream_alignment::DataStreamAlignmentDescriptor,
    TargetBackgroundGrid = 0x07 => crate::descriptors::target_background_grid::TargetBackgroundGridDescriptor,
    VideoWindow = 0x08 => crate::descriptors::video_window::VideoWindowDescriptor,
    Ca = 0x09 => crate::descriptors::ca::CaDescriptor<'a>,
    Iso639Language = 0x0A => crate::descriptors::iso_639_language::Iso639LanguageDescriptor,
    SystemClock = 0x0B => crate::descriptors::system_clock::SystemClockDescriptor,
    MultiplexBufferUtilization = 0x0C => crate::descriptors::multiplex_buffer_utilization::MultiplexBufferUtilizationDescriptor,
    Copyright = 0x0D => crate::descriptors::copyright::CopyrightDescriptor<'a>,
    MaximumBitrate = 0x0E => crate::descriptors::maximum_bitrate::MaximumBitrateDescriptor,
    PrivateDataIndicator = 0x0F => crate::descriptors::private_data_indicator::PrivateDataIndicatorDescriptor,
    SmoothingBuffer = 0x10 => crate::descriptors::smoothing_buffer::SmoothingBufferDescriptor,
    Std = 0x11 => crate::descriptors::std::StdDescriptor,
    Ibp = 0x12 => crate::descriptors::ibp::IbpDescriptor,
    CarouselIdentifier = 0x13 => crate::descriptors::carousel_identifier::CarouselIdentifierDescriptor<'a>,
    AssociationTag = 0x14 => crate::descriptors::association_tag::AssociationTagDescriptor<'a>,
    DeferredAssociationTags = 0x15 => crate::descriptors::deferred_association_tags::DeferredAssociationTagsDescriptor<'a>,
    Mpeg4Video = 0x1B => crate::descriptors::mpeg4_video::Mpeg4VideoDescriptor,
    Mpeg4Audio = 0x1C => crate::descriptors::mpeg4_audio::Mpeg4AudioDescriptor,
    Iod = 0x1D => crate::descriptors::iod::IodDescriptor<'a>,
    Sl = 0x1E => crate::descriptors::sl::SlDescriptor,
    Fmc = 0x1F => crate::descriptors::fmc::FmcDescriptor,
    ExternalEsId = 0x20 => crate::descriptors::external_es_id::ExternalEsIdDescriptor,
    Muxcode = 0x21 => crate::descriptors::muxcode::MuxcodeDescriptor<'a>,
    FmxBufferSize = 0x22 => crate::descriptors::fmx_buffer_size::FmxBufferSizeDescriptor<'a>,
    MultiplexBuffer = 0x23 => crate::descriptors::multiplex_buffer::MultiplexBufferDescriptor,
    ContentLabeling = 0x24 => crate::descriptors::content_labeling::ContentLabelingDescriptor<'a>,
    MetadataPointer = 0x25 => crate::descriptors::metadata_pointer::MetadataPointerDescriptor<'a>,
    Metadata = 0x26 => crate::descriptors::metadata::MetadataDescriptor<'a>,
    MetadataStd = 0x27 => crate::descriptors::metadata_std::MetadataStdDescriptor,
    AvcVideo = 0x28 => crate::descriptors::avc_video::AvcVideoDescriptor,
    AvcTimingAndHrd = 0x2A => crate::descriptors::avc_timing_and_hrd::AvcTimingAndHrdDescriptor,
    Mpeg2AacAudio = 0x2B => crate::descriptors::mpeg2_aac_audio::Mpeg2AacAudioDescriptor,
    FlexMuxTiming = 0x2C => crate::descriptors::flex_mux_timing::FlexMuxTimingDescriptor,
    Mpeg4Text = 0x2D => crate::descriptors::mpeg4_text::Mpeg4TextDescriptor<'a>,
    Mpeg4AudioExtension = 0x2E => crate::descriptors::mpeg4_audio_extension::Mpeg4AudioExtensionDescriptor,
    AuxiliaryVideoStream = 0x2F => crate::descriptors::auxiliary_video_stream::AuxiliaryVideoStreamDescriptor<'a>,
    SvcExtension = 0x30 => crate::descriptors::svc_extension::SvcExtensionDescriptor,
    MvcExtension = 0x31 => crate::descriptors::mvc_extension::MvcExtensionDescriptor,
    J2kVideo = 0x32 => crate::descriptors::j2k_video::J2kVideoDescriptor<'a>,
    MvcOperationPoint = 0x33 => crate::descriptors::mvc_operation_point::MvcOperationPointDescriptor,
    Mpeg2StereoscopicVideoFormat = 0x34 => crate::descriptors::stereoscopic::Mpeg2StereoscopicVideoFormatDescriptor,
    StereoscopicProgramInfo = 0x35 => crate::descriptors::stereoscopic::StereoscopicProgramInfoDescriptor,
    StereoscopicVideoInfo = 0x36 => crate::descriptors::stereoscopic::StereoscopicVideoInfoDescriptor,
    TransportProfile = 0x37 => crate::descriptors::transport_profile::TransportProfileDescriptor,
    HevcVideo = 0x38 => crate::descriptors::hevc_video::HevcVideoDescriptor,
    VvcVideo = 0x39 => crate::descriptors::vvc_video::VvcVideoDescriptor,
    EvcVideo = 0x3A => crate::descriptors::evc_video::EvcVideoDescriptor,
    Mpeg2Extension = 0x3F => crate::descriptors::mpeg2_extension::Mpeg2ExtensionDescriptor,
    // DVB descriptors (ETSI EN 300 468) — contiguous 0x40..=0x7F.
    NetworkName = 0x40 => crate::descriptors::network_name::NetworkNameDescriptor<'a>,
    ServiceList = 0x41 => crate::descriptors::service_list::ServiceListDescriptor,
    Stuffing = 0x42 => crate::descriptors::stuffing::StuffingDescriptor<'a>,
    SatelliteDeliverySystem = 0x43 => crate::descriptors::satellite_delivery_system::SatelliteDeliverySystemDescriptor,
    CableDeliverySystem = 0x44 => crate::descriptors::cable_delivery_system::CableDeliverySystemDescriptor,
    VbiData = 0x45 => crate::descriptors::vbi_data::VbiDataDescriptor<'a>,
    VbiTeletext = 0x46 => crate::descriptors::vbi_teletext::VbiTeletextDescriptor,
    BouquetName = 0x47 => crate::descriptors::bouquet_name::BouquetNameDescriptor<'a>,
    Service = 0x48 => crate::descriptors::service::ServiceDescriptor<'a>,
    CountryAvailability = 0x49 => crate::descriptors::country_availability::CountryAvailabilityDescriptor,
    Linkage = 0x4A => crate::descriptors::linkage::LinkageDescriptor<'a>,
    NvodReference = 0x4B => crate::descriptors::nvod_reference::NvodReferenceDescriptor,
    TimeShiftedService = 0x4C => crate::descriptors::time_shifted_service::TimeShiftedServiceDescriptor,
    ShortEvent = 0x4D => crate::descriptors::short_event::ShortEventDescriptor<'a>,
    ExtendedEvent = 0x4E => crate::descriptors::extended_event::ExtendedEventDescriptor<'a>,
    TimeShiftedEvent = 0x4F => crate::descriptors::time_shifted_event::TimeShiftedEventDescriptor,
    Component = 0x50 => crate::descriptors::component::ComponentDescriptor<'a>,
    Mosaic = 0x51 => crate::descriptors::mosaic::MosaicDescriptor,
    StreamIdentifier = 0x52 => crate::descriptors::stream_identifier::StreamIdentifierDescriptor,
    CaIdentifier = 0x53 => crate::descriptors::ca_identifier::CaIdentifierDescriptor,
    Content = 0x54 => crate::descriptors::content::ContentDescriptor,
    ParentalRating = 0x55 => crate::descriptors::parental_rating::ParentalRatingDescriptor,
    Teletext = 0x56 => crate::descriptors::teletext::TeletextDescriptor,
    Telephone = 0x57 => crate::descriptors::telephone::TelephoneDescriptor<'a>,
    LocalTimeOffset = 0x58 => crate::descriptors::local_time_offset::LocalTimeOffsetDescriptor,
    Subtitling = 0x59 => crate::descriptors::subtitling::SubtitlingDescriptor,
    TerrestrialDeliverySystem = 0x5A => crate::descriptors::terrestrial_delivery_system::TerrestrialDeliverySystemDescriptor,
    MultilingualNetworkName = 0x5B => crate::descriptors::multilingual_network_name::MultilingualNetworkNameDescriptor<'a>,
    MultilingualBouquetName = 0x5C => crate::descriptors::multilingual_bouquet_name::MultilingualBouquetNameDescriptor<'a>,
    MultilingualServiceName = 0x5D => crate::descriptors::multilingual_service_name::MultilingualServiceNameDescriptor<'a>,
    MultilingualComponent = 0x5E => crate::descriptors::multilingual_component::MultilingualComponentDescriptor<'a>,
    PrivateDataSpecifier = 0x5F => crate::descriptors::private_data_specifier::PrivateDataSpecifierDescriptor,
    ServiceMove = 0x60 => crate::descriptors::service_move::ServiceMoveDescriptor,
    ShortSmoothingBuffer = 0x61 => crate::descriptors::short_smoothing_buffer::ShortSmoothingBufferDescriptor<'a>,
    FrequencyList = 0x62 => crate::descriptors::frequency_list::FrequencyListDescriptor,
    PartialTransportStream = 0x63 => crate::descriptors::partial_transport_stream::PartialTransportStreamDescriptor,
    DataBroadcast = 0x64 => crate::descriptors::data_broadcast::DataBroadcastDescriptor<'a>,
    Scrambling = 0x65 => crate::descriptors::scrambling::ScramblingDescriptor,
    DataBroadcastId = 0x66 => crate::descriptors::data_broadcast_id::DataBroadcastIdDescriptor<'a>,
    TransportStream = 0x67 => crate::descriptors::transport_stream::TransportStreamDescriptor<'a>,
    Dsng = 0x68 => crate::descriptors::dsng::DsngDescriptor<'a>,
    Pdc = 0x69 => crate::descriptors::pdc::PdcDescriptor,
    Ac3 = 0x6A => crate::descriptors::ac3::Ac3Descriptor<'a>,
    AncillaryData = 0x6B => crate::descriptors::ancillary_data::AncillaryDataDescriptor,
    CellList = 0x6C => crate::descriptors::cell_list::CellListDescriptor,
    CellFrequencyLink = 0x6D => crate::descriptors::cell_frequency_link::CellFrequencyLinkDescriptor,
    AnnouncementSupport = 0x6E => crate::descriptors::announcement_support::AnnouncementSupportDescriptor,
    ApplicationSignalling = 0x6F => crate::descriptors::application_signalling::ApplicationSignallingDescriptor,
    AdaptationFieldData = 0x70 => crate::descriptors::adaptation_field_data::AdaptationFieldDataDescriptor,
    ServiceIdentifier = 0x71 => crate::descriptors::service_identifier::ServiceIdentifierDescriptor<'a>,
    ServiceAvailability = 0x72 => crate::descriptors::service_availability::ServiceAvailabilityDescriptor,
    DefaultAuthority = 0x73 => crate::descriptors::default_authority::DefaultAuthorityDescriptor<'a>,
    RelatedContent = 0x74 => crate::descriptors::related_content::RelatedContentDescriptor,
    TvaId = 0x75 => crate::descriptors::tva_id::TvaIdDescriptor,
    ContentIdentifier = 0x76 => crate::descriptors::content_identifier::ContentIdentifierDescriptor<'a>,
    TimeSliceFecIdentifier = 0x77 => crate::descriptors::time_slice_fec_identifier::TimeSliceFecIdentifierDescriptor<'a>,
    EcmRepetitionRate = 0x78 => crate::descriptors::ecm_repetition_rate::EcmRepetitionRateDescriptor<'a>,
    S2SatelliteDeliverySystem = 0x79 => crate::descriptors::s2_satellite_delivery_system::S2SatelliteDeliverySystemDescriptor,
    EnhancedAc3 = 0x7A => crate::descriptors::enhanced_ac3::EnhancedAc3Descriptor<'a>,
    Dts = 0x7B => crate::descriptors::dts::DtsDescriptor<'a>,
    Aac = 0x7C => crate::descriptors::aac::AacDescriptor<'a>,
    XaitLocation = 0x7D => crate::descriptors::xait_location::XaitLocationDescriptor,
    FtaContentManagement = 0x7E => crate::descriptors::fta_content_management::FtaContentManagementDescriptor,
    Extension = 0x7F => crate::descriptors::extension::ExtensionDescriptor<'a>;
    // Private / context-dependent: variant exists but is NOT auto-dispatched.
    // 0x83 logical_channel requires private_data_specifier context; enabled
    // via the descriptor registry (Task 4).
    @no_dispatch
    // ORDER: this no-dispatch tail is ordered by descriptor tag (0x3E, 0x40, 0x83, 0x86, 0x87, 0xD1, 0xD2, 0xD3), not alphabetically.
    TuwaAvs3Video => crate::descriptors::tuwa_avs3_video::TuwaAvs3VideoDescriptor,
    Avs2Video => crate::descriptors::avs2_video::Avs2VideoDescriptor,
    LogicalChannel => crate::descriptors::logical_channel::LogicalChannelDescriptor,
    NordigLogicalChannelV1 => crate::descriptors::nordig::NordigLogicalChannelV1,
    ChinaCaptionService => crate::descriptors::china_caption_service::ChinaCaptionServiceDescriptor,
    AtscCaptionService => crate::descriptors::atsc_caption_service::AtscCaptionServiceDescriptor,
    NordigLogicalChannelV2 => crate::descriptors::nordig::NordigLogicalChannelV2,
    Avs3Video => crate::descriptors::avs3_video::Avs3VideoDescriptor,
    Avs3Audio => crate::descriptors::avs3_audio::Avs3AudioDescriptor,
    Avs2Audio => crate::descriptors::avs2_audio::Avs2AudioDescriptor,
}

/// Lazily walk a raw descriptor loop. Never panics.
///
/// Per-descriptor parse errors yield `Err` and iteration continues (the
/// descriptor_length field bounds each entry, so the walker can always
/// advance past a malformed body). A truncated final header or body yields
/// one `Err` and then the iterator fuses (returns `None` forever after).
#[must_use]
pub fn parse_loop(bytes: &[u8]) -> DescriptorIter<'_> {
    DescriptorIter {
        bytes,
        pos: 0,
        fused: false,
    }
}

/// Extract the next descriptor-loop entry (tag + full-entry slice) from the
/// raw byte stream. Shared by [`DescriptorIter::next`] and
/// [`RegistryIter::next`](crate::descriptors::registry::RegistryIter::next).
///
/// On success, advances `pos` past the entry and returns `(tag, full)`.
/// On truncation, sets `fused = true` and returns `Some(Err(..))`.
/// When the loop is exhausted, returns `None`.
pub(crate) fn next_loop_entry<'a>(
    bytes: &'a [u8],
    pos: &mut usize,
    fused: &mut bool,
) -> Option<crate::Result<(u8, &'a [u8])>> {
    if *fused || *pos >= bytes.len() {
        return None;
    }
    let rem = &bytes[*pos..];
    if rem.len() < 2 {
        *fused = true;
        return Some(Err(crate::Error::BufferTooShort(
            broadcast_common::error::BufferTooShort {
                need: 2,
                have: rem.len(),
                what: "descriptor header in loop",
            },
        )));
    }
    let tag = rem[0];
    let len = rem[1] as usize;
    let total = 2 + len;
    if rem.len() < total {
        *fused = true;
        return Some(Err(crate::Error::BufferTooShort(
            broadcast_common::error::BufferTooShort {
                need: total,
                have: rem.len(),
                what: "descriptor body in loop",
            },
        )));
    }
    let full = &rem[..total];
    *pos += total;
    Some(Ok((tag, full)))
}

/// Iterator over a raw descriptor loop; see [`parse_loop`].
#[derive(Debug, Clone)]
pub struct DescriptorIter<'a> {
    bytes: &'a [u8],
    pos: usize,
    fused: bool,
}

impl<'a> Iterator for DescriptorIter<'a> {
    type Item = crate::Result<AnyDescriptor<'a>>;

    fn next(&mut self) -> Option<Self::Item> {
        let (tag, full) = match next_loop_entry(self.bytes, &mut self.pos, &mut self.fused)? {
            Ok(v) => v,
            Err(e) => return Some(Err(e)),
        };
        Some(match AnyDescriptor::dispatch(tag, full) {
            Some(res) => res,
            None => Ok(AnyDescriptor::Unknown {
                tag,
                body: &full[2..],
            }),
        })
    }
}

impl core::iter::FusedIterator for DescriptorIter<'_> {}

/// A raw descriptor loop, borrowed from the section. Zero-copy: walk it
/// typed via [`DescriptorLoop::iter`]; serde serializes the typed walk.
///
/// This is the table-loop analogue of [`crate::text::DvbText`]: it wraps the
/// raw `descriptor()` sequence (the variable-length region inside a table) and
/// decodes — i.e. dispatches each entry to a typed [`AnyDescriptor`] — only on
/// demand. Parsing stays zero-copy; the typed walk happens when you call
/// [`DescriptorLoop::iter`] or serialize.
///
/// ```
/// use dvb_si::descriptors::{AnyDescriptor, DescriptorLoop};
///
/// // short_event (tag 0x4D, "eng" / "Hi") then an unknown private tag 0xA7.
/// let raw = [
///     0x4D, 0x07, b'e', b'n', b'g', 0x02, b'H', b'i', 0x00,
///     0xA7, 0x02, 0xCA, 0xFE,
/// ];
/// let loop_ = DescriptorLoop::new(&raw);
/// let items: Vec<_> = loop_.iter().collect();
/// assert_eq!(items.len(), 2);
/// assert!(matches!(items[0].as_ref().unwrap(), AnyDescriptor::ShortEvent(_)));
/// assert_eq!(loop_.raw(), &raw[..]); // bytes preserved verbatim
/// ```
#[derive(Clone, Copy, PartialEq, Eq, Hash, Default)]
#[cfg_attr(feature = "yoke", derive(yoke::Yokeable))]
pub struct DescriptorLoop<'a>(&'a [u8]);

/// Bytes of a descriptor header: `descriptor_tag` (1) + `descriptor_length` (1).
const DESC_HEADER_LEN: usize = 2;

impl<'a> DescriptorLoop<'a> {
    /// Wrap a raw descriptor-loop slice (the `descriptor()` bytes only — no
    /// enclosing length field).
    #[must_use]
    pub const fn new(raw: &'a [u8]) -> Self {
        Self(raw)
    }

    /// The raw wire bytes of the loop, verbatim. These are what a serializer
    /// writes back; use them for the byte length of the loop.
    #[must_use]
    pub const fn raw(&self) -> &'a [u8] {
        self.0
    }

    /// Lazily walk the loop, yielding one typed [`AnyDescriptor`] per entry
    /// (or [`AnyDescriptor::Unknown`] for tags with no implementation).
    /// Delegates to [`parse_loop`]; never panics.
    #[must_use]
    pub fn iter(&self) -> DescriptorIter<'a> {
        parse_loop(self.0)
    }

    /// Walk the loop's `(tag, body)` pairs by the TLV structure only — never
    /// typed-parses a body. Each entry is `descriptor_tag` (1 byte) +
    /// `descriptor_length` (1 byte) + that many body bytes; a truncated final
    /// entry (declared length running past the buffer) simply ends iteration.
    ///
    /// Use this for body-agnostic structural scans (e.g. "is tag X present?")
    /// where a malformed or empty body should not hide the tag — unlike
    /// [`iter`](Self::iter), which yields `Err`/`Unknown` for bodies that fail
    /// to typed-parse.
    pub fn raw_tags(&self) -> impl Iterator<Item = (u8, &'a [u8])> {
        let b = self.0;
        let mut pos = 0usize;
        core::iter::from_fn(move || {
            if pos + DESC_HEADER_LEN > b.len() {
                return None;
            }
            let tag = b[pos];
            let len = b[pos + 1] as usize;
            let end = pos + DESC_HEADER_LEN + len;
            if end > b.len() {
                return None; // truncated final entry
            }
            let body = &b[pos + DESC_HEADER_LEN..end];
            pos = end;
            Some((tag, body))
        })
    }

    /// True if the loop contains a descriptor with `tag`, regardless of whether
    /// its body parses (or even fits). Pure structural walk: checks the
    /// `descriptor_tag` byte at each entry header, so an empty (`[tag, 0x00]`)
    /// or truncated (`[tag, 0x01]`) descriptor still counts as present.
    #[must_use]
    pub fn contains_tag(&self, tag: u8) -> bool {
        let b = self.0;
        let mut pos = 0usize;
        while pos < b.len() {
            if b[pos] == tag {
                return true;
            }
            // Need the length byte to advance to the next header; without it
            // the tag byte just checked was the last thing in the loop.
            if pos + 1 >= b.len() {
                break;
            }
            pos += DESC_HEADER_LEN + b[pos + 1] as usize;
        }
        false
    }

    /// Walk this loop through a [`DescriptorRegistry`](crate::descriptors::registry::DescriptorRegistry)
    /// so registered private/custom descriptors (and PDS-scoped tags) are
    /// produced as [`AnyDescriptor::Other`]; mirrors [`iter`](Self::iter)
    /// otherwise.
    ///
    /// See [`DescriptorRegistry::parse_loop`](crate::descriptors::registry::DescriptorRegistry::parse_loop)
    /// for precedence rules and PDS-scoped dispatch.
    #[must_use]
    pub fn iter_with<'r>(
        &self,
        registry: &'r crate::descriptors::registry::DescriptorRegistry,
    ) -> crate::descriptors::registry::RegistryIter<'r, 'a> {
        registry.parse_loop(self.0)
    }

    /// Walk this loop through both a [`DescriptorRegistry`](crate::descriptors::registry::DescriptorRegistry) and an
    /// [`ExtensionRegistry`](crate::descriptors::extension::registry::ExtensionRegistry),
    /// so that custom-registered extension bodies (tag `0x7F` with a known
    /// `descriptor_tag_extension`) are surfaced as
    /// [`ExtIterItem::CustomExtension`](crate::descriptors::registry::ExtIterItem::CustomExtension) with the type-erased value available
    /// for downcast. All other descriptors follow the normal
    /// [`iter_with`](Self::iter_with) precedence.
    ///
    /// This is the only way to reach private extension bodies during a
    /// descriptor-loop walk; [`iter_with`](Self::iter_with) alone produces
    /// the built-in [`ExtensionBody::Raw`](crate::descriptors::extension::ExtensionBody::Raw)
    /// for unrecognised `descriptor_tag_extension` values.
    #[must_use]
    pub fn iter_with_extensions<'r>(
        &self,
        desc_reg: &'r crate::descriptors::registry::DescriptorRegistry,
        ext_reg: &'r crate::descriptors::extension::registry::ExtensionRegistry,
    ) -> crate::descriptors::registry::ExtRegistryIter<'r, 'a> {
        crate::descriptors::registry::ExtRegistryIter::new(desc_reg, ext_reg, self.0, None)
    }

    /// [`iter_with_extensions`](Self::iter_with_extensions) starting inside a
    /// scope an enclosing loop opened (`inherited`, from
    /// [`ExtRegistryIter::avs_scope`](crate::descriptors::registry::ExtRegistryIter::avs_scope)):
    /// a `program_info` 'AVSV'/'AVSA' registration scopes the AVS descriptors of
    /// the ES_info loops (see `DescriptorRegistry::parse_loop_in_scope`).
    #[must_use]
    pub fn iter_with_extensions_in_scope<'r>(
        &self,
        desc_reg: &'r crate::descriptors::registry::DescriptorRegistry,
        ext_reg: &'r crate::descriptors::extension::registry::ExtensionRegistry,
        inherited: Option<u32>,
    ) -> crate::descriptors::registry::ExtRegistryIter<'r, 'a> {
        crate::descriptors::registry::ExtRegistryIter::new(desc_reg, ext_reg, self.0, inherited)
    }
}

impl core::ops::Deref for DescriptorLoop<'_> {
    /// Derefs to the raw wire bytes — `len()`/indexing are **byte counts for
    /// serialization, not entry counts**. To count entries, use
    /// [`DescriptorLoop::iter`].
    type Target = [u8];
    fn deref(&self) -> &[u8] {
        self.0
    }
}

impl<'a> From<&'a [u8]> for DescriptorLoop<'a> {
    fn from(raw: &'a [u8]) -> Self {
        Self(raw)
    }
}

impl core::fmt::Debug for DescriptorLoop<'_> {
    /// Cheap: prints the byte length, not the decoded entries.
    fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
        write!(f, "DescriptorLoop(<{} bytes>)", self.0.len())
    }
}

impl<'a> IntoIterator for &DescriptorLoop<'a> {
    type Item = crate::Result<AnyDescriptor<'a>>;
    type IntoIter = DescriptorIter<'a>;
    fn into_iter(self) -> Self::IntoIter {
        self.iter()
    }
}

#[cfg(feature = "serde")]
impl serde::Serialize for DescriptorLoop<'_> {
    /// Serializes as a sequence of the typed walk: each `Ok(d)` becomes the
    /// [`AnyDescriptor`] (camelCase external tagging), and each `Err(e)`
    /// becomes a `{"parseError": "<Display>"}` map — parse errors are surfaced,
    /// never silently dropped.
    fn serialize<S: serde::Serializer>(&self, s: S) -> Result<S::Ok, S::Error> {
        struct Entry<'a>(crate::Result<AnyDescriptor<'a>>);
        impl serde::Serialize for Entry<'_> {
            fn serialize<S: serde::Serializer>(&self, s: S) -> Result<S::Ok, S::Error> {
                match &self.0 {
                    Ok(d) => d.serialize(s),
                    Err(e) => {
                        use serde::ser::SerializeMap;
                        let mut m = s.serialize_map(Some(1))?;
                        m.serialize_entry("parseError", &e.to_string())?;
                        m.end()
                    }
                }
            }
        }
        s.collect_seq(self.iter().map(Entry))
    }
}
// Serialize-only: the typed walk decodes DVB text and dispatches per-tag —
// there is no lossless way to reconstruct the raw loop bytes from the
// serialized form. Structs holding a DescriptorLoop derive Serialize only.
// To reconstruct, keep the wire bytes and re-`parse` the table.

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn contains_tag_detects_present_tag_regardless_of_body() {
        // Well-formed AC-3 (0x6A) with a flags byte.
        assert!(DescriptorLoop::new(&[0x6A, 0x01, 0x80]).contains_tag(0x6A));
        // Empty body — malformed for AC-3, but the tag is present.
        assert!(DescriptorLoop::new(&[0x6A, 0x00]).contains_tag(0x6A));
        // Truncated — declares 1 body byte that isn't there; tag still present.
        assert!(DescriptorLoop::new(&[0x6A, 0x01]).contains_tag(0x6A));
        // Present after an earlier well-formed entry.
        assert!(DescriptorLoop::new(&[0x09, 0x02, 0x00, 0x00, 0x7A, 0x00]).contains_tag(0x7A));
        // Absent.
        assert!(!DescriptorLoop::new(&[0x09, 0x02, 0x00, 0x00]).contains_tag(0x6A));
        assert!(!DescriptorLoop::new(&[]).contains_tag(0x6A));
    }

    #[test]
    fn raw_tags_walks_tlv_without_typed_parsing() {
        // Two complete entries: tag 0x6A (empty body) + tag 0x40 (2-byte body).
        let loop_ = DescriptorLoop::new(&[0x6A, 0x00, 0x40, 0x02, 0xAA, 0xBB]);
        let pairs: Vec<_> = loop_.raw_tags().collect();
        assert_eq!(pairs, vec![(0x6A, &[][..]), (0x40, &[0xAA, 0xBB][..])]);
        // First entry [0x40,len=1,0xAA] complete, then a truncated [0x6A, len=5]
        // entry whose declared body runs past the buffer.
        let trunc = DescriptorLoop::new(&[0x40, 0x01, 0xAA, 0x6A, 0x05]);
        let tags: Vec<u8> = trunc.raw_tags().map(|(t, _)| t).collect();
        assert_eq!(tags, vec![0x40]); // the truncated 0x6A entry is dropped by raw_tags
        // ...but contains_tag still sees the truncated tag.
        assert!(trunc.contains_tag(0x6A));
    }

    #[test]
    fn unknown_tag_yields_unknown_with_body_sans_header() {
        // tag 0xA7 (no typed impl), length 2, body [0xDE, 0xAD].
        let bytes = [0xA7, 0x02, 0xDE, 0xAD];
        let items: Vec<_> = parse_loop(&bytes).collect();
        assert_eq!(items.len(), 1);
        match items[0].as_ref().unwrap() {
            AnyDescriptor::Unknown { tag, body } => {
                assert_eq!(*tag, 0xA7);
                assert_eq!(*body, &[0xDE, 0xAD]);
            }
            other => panic!("expected Unknown, got {other:?}"),
        }
    }

    #[test]
    fn empty_loop_yields_nothing() {
        assert_eq!(parse_loop(&[]).count(), 0);
    }

    #[test]
    fn logical_channel_0x83_is_not_dispatched() {
        // 0x83 has a variant but no dispatcher entry → Unknown, never panics.
        let bytes = [0x83, 0x04, 0x00, 0x01, 0xFC, 0x01];
        let items: Vec<_> = parse_loop(&bytes).collect();
        assert_eq!(items.len(), 1);
        assert!(matches!(
            items[0].as_ref().unwrap(),
            AnyDescriptor::Unknown { tag: 0x83, .. }
        ));
    }

    #[test]
    fn descriptor_loop_iter_matches_parse_loop() {
        let raw = [
            0x4D, 0x07, b'e', b'n', b'g', 0x02, b'H', b'i', 0x00, // short_event
            0xA7, 0x02, 0xCA, 0xFE, // unknown 0xA7
        ];
        let via_loop: Vec<_> = DescriptorLoop::new(&raw)
            .iter()
            .map(|r| format!("{r:?}"))
            .collect();
        let via_fn: Vec<_> = parse_loop(&raw).map(|r| format!("{r:?}")).collect();
        assert_eq!(via_loop, via_fn);
        // raw()/Deref expose the wire bytes (byte length, not entry count).
        assert_eq!(DescriptorLoop::new(&raw).raw(), &raw[..]);
        assert_eq!(DescriptorLoop::new(&raw).len(), raw.len());
        // IntoIterator for &DescriptorLoop.
        let count = (&DescriptorLoop::new(&raw)).into_iter().count();
        assert_eq!(count, 2);
    }

    #[test]
    fn descriptor_loop_debug_is_cheap() {
        let raw = [0x4D, 0x02, 0x01, 0x02];
        assert_eq!(
            format!("{:?}", DescriptorLoop::new(&raw)),
            "DescriptorLoop(<4 bytes>)"
        );
    }

    #[test]
    fn iter_with_custom_tag_yields_other() {
        use crate::descriptors::registry::DescriptorRegistry;
        use crate::traits::DescriptorDef;

        #[derive(Debug, PartialEq, Eq)]
        #[cfg_attr(feature = "serde", derive(serde::Serialize))]
        struct MyTag0xA7 {
            x: u8,
        }

        impl<'a> broadcast_common::Parse<'a> for MyTag0xA7 {
            type Error = crate::error::Error;
            fn parse(bytes: &'a [u8]) -> crate::Result<Self> {
                if bytes.len() < 3 {
                    return Err(crate::error::Error::BufferTooShort(
                        broadcast_common::error::BufferTooShort {
                            need: 3,
                            have: bytes.len(),
                            what: "MyTag0xA7",
                        },
                    ));
                }
                Ok(Self { x: bytes[2] })
            }
        }

        impl DescriptorDef<'_> for MyTag0xA7 {
            const TAG: u8 = 0xA7;
            const NAME: &'static str = "MY_TAG_0xA7";
        }

        let mut reg = DescriptorRegistry::new();
        reg.register::<MyTag0xA7>();

        let raw = [
            0x4D, 0x07, b'e', b'n', b'g', 0x02, b'H', b'i', 0x00, 0xA7, 0x02, 0xCA, 0xFE,
        ];
        let loop_ = DescriptorLoop::new(&raw);
        let items: Vec<_> = loop_.iter_with(&reg).collect::<Result<_, _>>().unwrap();
        assert_eq!(items.len(), 2);
        assert!(matches!(items[0], AnyDescriptor::ShortEvent(_)));
        match &items[1] {
            AnyDescriptor::Other { tag, value } => {
                assert_eq!(*tag, 0xA7);
                assert_eq!(value.downcast_ref::<MyTag0xA7>().unwrap().x, 0xCA);
            }
            other => panic!("expected Other, got {other:?}"),
        }
    }

    #[test]
    fn iter_with_empty_registry_matches_iter_for_builtin() {
        let raw = [
            0x4D, 0x07, b'e', b'n', b'g', 0x02, b'H', b'i', 0x00, 0xA7, 0x02, 0xCA, 0xFE,
        ];
        let loop_ = DescriptorLoop::new(&raw);
        let reg = crate::descriptors::registry::DescriptorRegistry::new();
        let via_iter: Vec<_> = loop_.iter().collect();
        let via_iter_with: Vec<_> = loop_.iter_with(&reg).collect();

        assert_eq!(via_iter.len(), via_iter_with.len());
        for (a, b) in via_iter.iter().zip(via_iter_with.iter()) {
            match (a, b) {
                (Ok(AnyDescriptor::ShortEvent(_)), Ok(AnyDescriptor::ShortEvent(_))) => {}
                (
                    Ok(AnyDescriptor::Unknown { tag: t1, body: b1 }),
                    Ok(AnyDescriptor::Unknown { tag: t2, body: b2 }),
                ) => {
                    assert_eq!(t1, t2);
                    assert_eq!(b1, b2);
                }
                (Err(_), Err(_)) => {}
                (l, r) => panic!("mismatch: {l:?} vs {r:?}"),
            }
        }
    }

    #[cfg(feature = "serde")]
    #[test]
    fn descriptor_loop_serializes_typed_unknown_and_parse_error() {
        // [valid short_event, unknown tag 0xA7, truncated final entry].
        // A truncated final entry (declared len 5, only 1 body byte present)
        // makes the walker yield a final Err → {"parseError": …}.
        let raw = [
            0x4D, 0x07, b'e', b'n', b'g', 0x02, b'H', b'i', 0x00, // short_event
            0xA7, 0x02, 0xCA, 0xFE, // unknown 0xA7
            0x55, 0x05, 0x00, // parental_rating header claims 5 bytes; only 1 present
        ];
        let v = serde_json::to_value(DescriptorLoop::new(&raw)).unwrap();
        let arr = v.as_array().expect("sequence");
        assert_eq!(arr.len(), 3);
        // 1. typed short_event under the camelCase variant key.
        assert!(arr[0].get("shortEvent").is_some(), "got {}", arr[0]);
        assert_eq!(arr[0]["shortEvent"]["event_name"], "Hi");
        // 2. unknown tag carries its raw body bytes.
        let unknown = arr[1].get("unknown").expect("unknown variant");
        assert_eq!(unknown["tag"], 0xA7);
        assert_eq!(unknown["body"], serde_json::json!([0xCA, 0xFE]));
        // 3. truncated entry → parseError, never silently dropped.
        assert!(
            arr[2].get("parseError").is_some(),
            "expected parseError, got {}",
            arr[2]
        );
    }
}