Skip to main content

dvb_si/descriptors/
any.rs

1//! Unified descriptor dispatch: [`AnyDescriptor`] + [`parse_loop`].
2//!
3//! [`AnyDescriptor`] is generated from a single declarative list
4//! (`declare_descriptors!`) — one line per crate-implemented descriptor tag.
5//! The list is the single source of truth: it produces the enum, the
6//! `From<T>` conversions, and the tag → type dispatcher, and a drift test
7//! pins each tag literal to the type's [`crate::traits::DescriptorDef::TAG`].
8//!
9//! [`parse_loop`] lazily walks a raw descriptor loop (the variable-length
10//! `descriptor()` sequence inside tables), yielding one [`AnyDescriptor`] per
11//! entry. It never panics: a malformed entry whose length is known yields an
12//! `Err` and iteration continues; a truncated final header/body yields one
13//! final `Err` and then fuses.
14//!
15//! ```
16//! use dvb_si::descriptors::{parse_loop, AnyDescriptor};
17//!
18//! // A two-descriptor loop: short_event (tag 0x4D, "eng" / "News") then an
19//! // unrecognised private tag 0xA7 — the walker yields a typed value for the
20//! // first and `Unknown` for the second, never panicking.
21//! let loop_bytes = [
22//!     0x4D, 0x07, b'e', b'n', b'g', 0x02, b'H', b'i', 0x00, // short_event
23//!     0xA7, 0x02, 0xCA, 0xFE,                               // unknown 0xA7
24//! ];
25//! let items: Vec<_> = parse_loop(&loop_bytes).collect();
26//! assert_eq!(items.len(), 2);
27//! match items[0].as_ref().unwrap() {
28//!     AnyDescriptor::ShortEvent(se) => {
29//!         assert_eq!(se.language_code.as_str(), "eng");
30//!         assert_eq!(se.event_name.decode(), "Hi");
31//!     }
32//!     other => panic!("expected ShortEvent, got {other:?}"),
33//! }
34//! assert!(matches!(items[1].as_ref().unwrap(), AnyDescriptor::Unknown { tag: 0xA7, .. }));
35//! ```
36//!
37//! # Adding a descriptor
38//!
39//! 1. Create the module with the wire layout, a `pub const TAG: u8`, and the
40//!    symmetric [`dvb_common::Parse`]/[`dvb_common::Serialize`] impls +
41//!    round-trip tests (copy an existing module).
42//! 2. `impl DescriptorDef` for the type (`TAG` from the module const, `NAME`
43//!    in SCREAMING_SNAKE without the `_descriptor` suffix).
44//! 3. Add one line to the `declare_descriptors!` invocation below — the enum
45//!    variant, dispatcher arm, and drift test are generated from it.
46//! 4. The integration completeness test walks the generated
47//!    [`AnyDescriptor::DISPATCHED_TAGS`] automatically — no test edits needed.
48
49use alloc::boxed::Box;
50#[cfg(feature = "serde")]
51use alloc::string::ToString;
52
53/// Declares [`AnyDescriptor`] + its dispatcher from one tag list.
54///
55/// Each line is `Variant = 0xTAG => module::Type[<'a>]`. The optional trailing
56/// `@no_dispatch …` section adds variants that are NOT reachable from the
57/// generated dispatcher (private / context-dependent tags such as 0x83
58/// logical_channel) — the variant exists for callers that opt in via the
59/// registry, but `dispatch` never produces it.
60macro_rules! declare_descriptors {
61    (
62        $lt:lifetime;
63        $( $variant:ident = $tag:literal => $($path:ident)::+ $(<$plt:lifetime>)? ),+ $(,)?
64        $( ; @no_dispatch $( $nd_variant:ident => $($nd_path:ident)::+ $(<$nd_plt:lifetime>)? ),+ $(,)? )?
65    ) => {
66        /// Every crate-implemented descriptor, plus an `Unknown` fallthrough.
67        ///
68        /// serde uses external tagging with camelCase variant keys —
69        /// a parsed short_event_descriptor serializes as `{"shortEvent": {…}}`.
70        /// Variant names map 1:1 to the descriptor modules; see each module
71        /// for the wire layout.
72        #[derive(Debug)]
73        #[cfg_attr(feature = "serde", derive(serde::Serialize))]
74        #[cfg_attr(feature = "serde", serde(rename_all = "camelCase"))]
75        // Every variant is covariant in `$lt`: typed variants hold only
76        // lifetime-parametrised views, `Unknown` holds `&$lt [u8]`, and the
77        // `Other` value is a `'static` `Box<dyn DescriptorObject>`. The derive
78        // accepts the `'static` field unchanged, so the impl is sound.
79        #[cfg_attr(feature = "yoke", derive(yoke::Yokeable))]
80        #[non_exhaustive]
81        pub enum AnyDescriptor<$lt> {
82            $(
83                #[allow(missing_docs)]
84                $variant($($path)::+ $(<$plt>)?),
85            )+
86            $($(
87                #[allow(missing_docs)]
88                $nd_variant($($nd_path)::+ $(<$nd_plt>)?),
89            )+)?
90            /// Runtime-registered custom descriptor (see [`DescriptorRegistry`]).
91            ///
92            /// [`DescriptorRegistry`]: crate::descriptors::registry::DescriptorRegistry
93            Other {
94                /// The raw descriptor_tag byte.
95                tag: u8,
96                /// The parsed, type-erased descriptor value. Call `downcast_ref`
97                /// on it (see [`DescriptorObject`](crate::descriptors::registry::DescriptorObject))
98                /// to recover the concrete type.
99                #[cfg_attr(
100                    feature = "serde",
101                    serde(serialize_with = "crate::descriptors::registry::serialize_erased")
102                )]
103                value: Box<dyn crate::descriptors::registry::DescriptorObject>,
104            },
105            /// Tag with no typed implementation; `body` is the payload sans
106            /// the 2-byte (tag, length) header.
107            Unknown {
108                /// The raw descriptor_tag byte.
109                tag: u8,
110                /// The raw payload bytes (descriptor_length bytes).
111                body: &$lt [u8],
112            },
113        }
114
115        $(
116            impl<$lt> From<$($path)::+ $(<$plt>)?> for AnyDescriptor<$lt> {
117                fn from(d: $($path)::+ $(<$plt>)?) -> Self {
118                    Self::$variant(d)
119                }
120            }
121        )+
122        $($(
123            impl<$lt> From<$($nd_path)::+ $(<$nd_plt>)?> for AnyDescriptor<$lt> {
124                fn from(d: $($nd_path)::+ $(<$nd_plt>)?) -> Self {
125                    Self::$nd_variant(d)
126                }
127            }
128        )+)?
129
130        impl<$lt> AnyDescriptor<$lt> {
131            /// Every tag the generated dispatcher routes (excludes `@no_dispatch`
132            /// variants and [`AnyDescriptor::Unknown`]).
133            pub const DISPATCHED_TAGS: &'static [u8] = &[$($tag),+];
134
135            /// Diagnostic name of the contained descriptor — the type's
136            /// [`DescriptorDef::NAME`](crate::traits::DescriptorDef::NAME)
137            /// (`"SHORT_EVENT"`, `"NETWORK_NAME"`, …); `"CUSTOM"` for
138            /// [`AnyDescriptor::Other`] (runtime-registered) and `"UNKNOWN"`
139            /// for [`AnyDescriptor::Unknown`].
140            #[must_use]
141            pub fn name(&self) -> &'static str {
142                match self {
143                    $(
144                        Self::$variant(_) =>
145                            <$($path)::+ as crate::traits::DescriptorDef>::NAME,
146                    )+
147                    $($(
148                        Self::$nd_variant(_) =>
149                            <$($nd_path)::+ as crate::traits::DescriptorDef>::NAME,
150                    )+)?
151                    Self::Other { .. } => "CUSTOM",
152                    Self::Unknown { .. } => "UNKNOWN",
153                }
154            }
155
156            /// Parse one full descriptor (2-byte header included) by its tag.
157            ///
158            /// `None` means no typed implementation exists for `tag` (the
159            /// caller turns that into [`AnyDescriptor::Unknown`]). `Some(Err)`
160            /// is a typed parse failure for a recognised tag.
161            pub(crate) fn dispatch(tag: u8, full: &$lt [u8]) -> Option<crate::Result<Self>> {
162                use dvb_common::Parse;
163                match tag {
164                    $(
165                        $tag => Some(<$($path)::+>::parse(full).map(Self::$variant)),
166                    )+
167                    _ => None,
168                }
169            }
170        }
171
172        #[cfg(test)]
173        mod macro_drift {
174            #[test]
175            fn tag_literals_match_descriptor_def() {
176                use crate::traits::DescriptorDef;
177                $(
178                    assert_eq!(
179                        $tag,
180                        <$($path)::+ as DescriptorDef>::TAG,
181                        concat!("tag literal drift for ", stringify!($variant)),
182                    );
183                    assert!(
184                        !<$($path)::+ as DescriptorDef>::NAME.is_empty(),
185                        concat!("empty NAME for ", stringify!($variant)),
186                    );
187                )+
188                $($(
189                    assert!(
190                        !<$($nd_path)::+ as DescriptorDef>::NAME.is_empty(),
191                        concat!("empty NAME for ", stringify!($nd_variant)),
192                    );
193                )+)?
194            }
195        }
196    };
197}
198
199declare_descriptors! {'a;
200    // MPEG-2 systems descriptors (ISO/IEC 13818-1) used outside table context.
201    VideoStream = 0x02 => crate::descriptors::video_stream::VideoStreamDescriptor,
202    AudioStream = 0x03 => crate::descriptors::audio_stream::AudioStreamDescriptor,
203    Hierarchy = 0x04 => crate::descriptors::hierarchy::HierarchyDescriptor,
204    Registration = 0x05 => crate::descriptors::registration::RegistrationDescriptor<'a>,
205    DataStreamAlignment = 0x06 => crate::descriptors::data_stream_alignment::DataStreamAlignmentDescriptor,
206    TargetBackgroundGrid = 0x07 => crate::descriptors::target_background_grid::TargetBackgroundGridDescriptor,
207    VideoWindow = 0x08 => crate::descriptors::video_window::VideoWindowDescriptor,
208    Ca = 0x09 => crate::descriptors::ca::CaDescriptor<'a>,
209    Iso639Language = 0x0A => crate::descriptors::iso_639_language::Iso639LanguageDescriptor,
210    SystemClock = 0x0B => crate::descriptors::system_clock::SystemClockDescriptor,
211    MultiplexBufferUtilization = 0x0C => crate::descriptors::multiplex_buffer_utilization::MultiplexBufferUtilizationDescriptor,
212    Copyright = 0x0D => crate::descriptors::copyright::CopyrightDescriptor<'a>,
213    MaximumBitrate = 0x0E => crate::descriptors::maximum_bitrate::MaximumBitrateDescriptor,
214    PrivateDataIndicator = 0x0F => crate::descriptors::private_data_indicator::PrivateDataIndicatorDescriptor,
215    SmoothingBuffer = 0x10 => crate::descriptors::smoothing_buffer::SmoothingBufferDescriptor,
216    Std = 0x11 => crate::descriptors::std::StdDescriptor,
217    Ibp = 0x12 => crate::descriptors::ibp::IbpDescriptor,
218    CarouselIdentifier = 0x13 => crate::descriptors::carousel_identifier::CarouselIdentifierDescriptor<'a>,
219    Mpeg4Video = 0x1B => crate::descriptors::mpeg4_video::Mpeg4VideoDescriptor,
220    Mpeg4Audio = 0x1C => crate::descriptors::mpeg4_audio::Mpeg4AudioDescriptor,
221    Iod = 0x1D => crate::descriptors::iod::IodDescriptor<'a>,
222    Sl = 0x1E => crate::descriptors::sl::SlDescriptor,
223    Fmc = 0x1F => crate::descriptors::fmc::FmcDescriptor,
224    ExternalEsId = 0x20 => crate::descriptors::external_es_id::ExternalEsIdDescriptor,
225    Muxcode = 0x21 => crate::descriptors::muxcode::MuxcodeDescriptor<'a>,
226    FmxBufferSize = 0x22 => crate::descriptors::fmx_buffer_size::FmxBufferSizeDescriptor<'a>,
227    MultiplexBuffer = 0x23 => crate::descriptors::multiplex_buffer::MultiplexBufferDescriptor,
228    ContentLabeling = 0x24 => crate::descriptors::content_labeling::ContentLabelingDescriptor<'a>,
229    MetadataPointer = 0x25 => crate::descriptors::metadata_pointer::MetadataPointerDescriptor<'a>,
230    Metadata = 0x26 => crate::descriptors::metadata::MetadataDescriptor<'a>,
231    MetadataStd = 0x27 => crate::descriptors::metadata_std::MetadataStdDescriptor,
232    AvcVideo = 0x28 => crate::descriptors::avc_video::AvcVideoDescriptor,
233    AvcTimingAndHrd = 0x2A => crate::descriptors::avc_timing_and_hrd::AvcTimingAndHrdDescriptor,
234    Mpeg2AacAudio = 0x2B => crate::descriptors::mpeg2_aac_audio::Mpeg2AacAudioDescriptor,
235    FlexMuxTiming = 0x2C => crate::descriptors::flex_mux_timing::FlexMuxTimingDescriptor,
236    AuxiliaryVideoStream = 0x2F => crate::descriptors::auxiliary_video_stream::AuxiliaryVideoStreamDescriptor<'a>,
237    SvcExtension = 0x30 => crate::descriptors::svc_extension::SvcExtensionDescriptor,
238    MvcExtension = 0x31 => crate::descriptors::mvc_extension::MvcExtensionDescriptor,
239    J2kVideo = 0x32 => crate::descriptors::j2k_video::J2kVideoDescriptor<'a>,
240    HevcVideo = 0x38 => crate::descriptors::hevc_video::HevcVideoDescriptor,
241    // DVB descriptors (ETSI EN 300 468) — contiguous 0x40..=0x7F.
242    NetworkName = 0x40 => crate::descriptors::network_name::NetworkNameDescriptor<'a>,
243    ServiceList = 0x41 => crate::descriptors::service_list::ServiceListDescriptor,
244    Stuffing = 0x42 => crate::descriptors::stuffing::StuffingDescriptor<'a>,
245    SatelliteDeliverySystem = 0x43 => crate::descriptors::satellite_delivery_system::SatelliteDeliverySystemDescriptor,
246    CableDeliverySystem = 0x44 => crate::descriptors::cable_delivery_system::CableDeliverySystemDescriptor,
247    VbiData = 0x45 => crate::descriptors::vbi_data::VbiDataDescriptor<'a>,
248    VbiTeletext = 0x46 => crate::descriptors::vbi_teletext::VbiTeletextDescriptor,
249    BouquetName = 0x47 => crate::descriptors::bouquet_name::BouquetNameDescriptor<'a>,
250    Service = 0x48 => crate::descriptors::service::ServiceDescriptor<'a>,
251    CountryAvailability = 0x49 => crate::descriptors::country_availability::CountryAvailabilityDescriptor,
252    Linkage = 0x4A => crate::descriptors::linkage::LinkageDescriptor<'a>,
253    NvodReference = 0x4B => crate::descriptors::nvod_reference::NvodReferenceDescriptor,
254    TimeShiftedService = 0x4C => crate::descriptors::time_shifted_service::TimeShiftedServiceDescriptor,
255    ShortEvent = 0x4D => crate::descriptors::short_event::ShortEventDescriptor<'a>,
256    ExtendedEvent = 0x4E => crate::descriptors::extended_event::ExtendedEventDescriptor<'a>,
257    TimeShiftedEvent = 0x4F => crate::descriptors::time_shifted_event::TimeShiftedEventDescriptor,
258    Component = 0x50 => crate::descriptors::component::ComponentDescriptor<'a>,
259    Mosaic = 0x51 => crate::descriptors::mosaic::MosaicDescriptor,
260    StreamIdentifier = 0x52 => crate::descriptors::stream_identifier::StreamIdentifierDescriptor,
261    CaIdentifier = 0x53 => crate::descriptors::ca_identifier::CaIdentifierDescriptor,
262    Content = 0x54 => crate::descriptors::content::ContentDescriptor,
263    ParentalRating = 0x55 => crate::descriptors::parental_rating::ParentalRatingDescriptor,
264    Teletext = 0x56 => crate::descriptors::teletext::TeletextDescriptor,
265    Telephone = 0x57 => crate::descriptors::telephone::TelephoneDescriptor<'a>,
266    LocalTimeOffset = 0x58 => crate::descriptors::local_time_offset::LocalTimeOffsetDescriptor,
267    Subtitling = 0x59 => crate::descriptors::subtitling::SubtitlingDescriptor,
268    TerrestrialDeliverySystem = 0x5A => crate::descriptors::terrestrial_delivery_system::TerrestrialDeliverySystemDescriptor,
269    MultilingualNetworkName = 0x5B => crate::descriptors::multilingual_network_name::MultilingualNetworkNameDescriptor<'a>,
270    MultilingualBouquetName = 0x5C => crate::descriptors::multilingual_bouquet_name::MultilingualBouquetNameDescriptor<'a>,
271    MultilingualServiceName = 0x5D => crate::descriptors::multilingual_service_name::MultilingualServiceNameDescriptor<'a>,
272    MultilingualComponent = 0x5E => crate::descriptors::multilingual_component::MultilingualComponentDescriptor<'a>,
273    PrivateDataSpecifier = 0x5F => crate::descriptors::private_data_specifier::PrivateDataSpecifierDescriptor,
274    ServiceMove = 0x60 => crate::descriptors::service_move::ServiceMoveDescriptor,
275    ShortSmoothingBuffer = 0x61 => crate::descriptors::short_smoothing_buffer::ShortSmoothingBufferDescriptor<'a>,
276    FrequencyList = 0x62 => crate::descriptors::frequency_list::FrequencyListDescriptor,
277    PartialTransportStream = 0x63 => crate::descriptors::partial_transport_stream::PartialTransportStreamDescriptor,
278    DataBroadcast = 0x64 => crate::descriptors::data_broadcast::DataBroadcastDescriptor<'a>,
279    Scrambling = 0x65 => crate::descriptors::scrambling::ScramblingDescriptor,
280    DataBroadcastId = 0x66 => crate::descriptors::data_broadcast_id::DataBroadcastIdDescriptor<'a>,
281    TransportStream = 0x67 => crate::descriptors::transport_stream::TransportStreamDescriptor<'a>,
282    Dsng = 0x68 => crate::descriptors::dsng::DsngDescriptor<'a>,
283    Pdc = 0x69 => crate::descriptors::pdc::PdcDescriptor,
284    Ac3 = 0x6A => crate::descriptors::ac3::Ac3Descriptor<'a>,
285    AncillaryData = 0x6B => crate::descriptors::ancillary_data::AncillaryDataDescriptor,
286    CellList = 0x6C => crate::descriptors::cell_list::CellListDescriptor,
287    CellFrequencyLink = 0x6D => crate::descriptors::cell_frequency_link::CellFrequencyLinkDescriptor,
288    AnnouncementSupport = 0x6E => crate::descriptors::announcement_support::AnnouncementSupportDescriptor,
289    ApplicationSignalling = 0x6F => crate::descriptors::application_signalling::ApplicationSignallingDescriptor,
290    AdaptationFieldData = 0x70 => crate::descriptors::adaptation_field_data::AdaptationFieldDataDescriptor,
291    ServiceIdentifier = 0x71 => crate::descriptors::service_identifier::ServiceIdentifierDescriptor<'a>,
292    ServiceAvailability = 0x72 => crate::descriptors::service_availability::ServiceAvailabilityDescriptor,
293    DefaultAuthority = 0x73 => crate::descriptors::default_authority::DefaultAuthorityDescriptor<'a>,
294    RelatedContent = 0x74 => crate::descriptors::related_content::RelatedContentDescriptor,
295    TvaId = 0x75 => crate::descriptors::tva_id::TvaIdDescriptor,
296    ContentIdentifier = 0x76 => crate::descriptors::content_identifier::ContentIdentifierDescriptor<'a>,
297    TimeSliceFecIdentifier = 0x77 => crate::descriptors::time_slice_fec_identifier::TimeSliceFecIdentifierDescriptor<'a>,
298    EcmRepetitionRate = 0x78 => crate::descriptors::ecm_repetition_rate::EcmRepetitionRateDescriptor<'a>,
299    S2SatelliteDeliverySystem = 0x79 => crate::descriptors::s2_satellite_delivery_system::S2SatelliteDeliverySystemDescriptor,
300    EnhancedAc3 = 0x7A => crate::descriptors::enhanced_ac3::EnhancedAc3Descriptor<'a>,
301    Dts = 0x7B => crate::descriptors::dts::DtsDescriptor<'a>,
302    Aac = 0x7C => crate::descriptors::aac::AacDescriptor<'a>,
303    XaitLocation = 0x7D => crate::descriptors::xait_location::XaitLocationDescriptor,
304    FtaContentManagement = 0x7E => crate::descriptors::fta_content_management::FtaContentManagementDescriptor,
305    Extension = 0x7F => crate::descriptors::extension::ExtensionDescriptor<'a>;
306    // Private / context-dependent: variant exists but is NOT auto-dispatched.
307    // 0x83 logical_channel requires private_data_specifier context; enabled
308    // via the descriptor registry (Task 4).
309    @no_dispatch
310    LogicalChannel => crate::descriptors::logical_channel::LogicalChannelDescriptor,
311    NordigLogicalChannelV1 => crate::descriptors::nordig::NordigLogicalChannelV1,
312    NordigLogicalChannelV2 => crate::descriptors::nordig::NordigLogicalChannelV2,
313}
314
315/// Lazily walk a raw descriptor loop. Never panics.
316///
317/// Per-descriptor parse errors yield `Err` and iteration continues (the
318/// descriptor_length field bounds each entry, so the walker can always
319/// advance past a malformed body). A truncated final header or body yields
320/// one `Err` and then the iterator fuses (returns `None` forever after).
321#[must_use]
322pub fn parse_loop(bytes: &[u8]) -> DescriptorIter<'_> {
323    DescriptorIter {
324        bytes,
325        pos: 0,
326        fused: false,
327    }
328}
329
330/// Extract the next descriptor-loop entry (tag + full-entry slice) from the
331/// raw byte stream. Shared by [`DescriptorIter::next`] and
332/// [`RegistryIter::next`](crate::descriptors::registry::RegistryIter::next).
333///
334/// On success, advances `pos` past the entry and returns `(tag, full)`.
335/// On truncation, sets `fused = true` and returns `Some(Err(..))`.
336/// When the loop is exhausted, returns `None`.
337pub(crate) fn next_loop_entry<'a>(
338    bytes: &'a [u8],
339    pos: &mut usize,
340    fused: &mut bool,
341) -> Option<crate::Result<(u8, &'a [u8])>> {
342    if *fused || *pos >= bytes.len() {
343        return None;
344    }
345    let rem = &bytes[*pos..];
346    if rem.len() < 2 {
347        *fused = true;
348        return Some(Err(crate::Error::BufferTooShort {
349            need: 2,
350            have: rem.len(),
351            what: "descriptor header in loop",
352        }));
353    }
354    let tag = rem[0];
355    let len = rem[1] as usize;
356    let total = 2 + len;
357    if rem.len() < total {
358        *fused = true;
359        return Some(Err(crate::Error::BufferTooShort {
360            need: total,
361            have: rem.len(),
362            what: "descriptor body in loop",
363        }));
364    }
365    let full = &rem[..total];
366    *pos += total;
367    Some(Ok((tag, full)))
368}
369
370/// Iterator over a raw descriptor loop; see [`parse_loop`].
371#[derive(Debug, Clone)]
372pub struct DescriptorIter<'a> {
373    bytes: &'a [u8],
374    pos: usize,
375    fused: bool,
376}
377
378impl<'a> Iterator for DescriptorIter<'a> {
379    type Item = crate::Result<AnyDescriptor<'a>>;
380
381    fn next(&mut self) -> Option<Self::Item> {
382        let (tag, full) = match next_loop_entry(self.bytes, &mut self.pos, &mut self.fused)? {
383            Ok(v) => v,
384            Err(e) => return Some(Err(e)),
385        };
386        Some(match AnyDescriptor::dispatch(tag, full) {
387            Some(res) => res,
388            None => Ok(AnyDescriptor::Unknown {
389                tag,
390                body: &full[2..],
391            }),
392        })
393    }
394}
395
396impl core::iter::FusedIterator for DescriptorIter<'_> {}
397
398/// A raw descriptor loop, borrowed from the section. Zero-copy: walk it
399/// typed via [`DescriptorLoop::iter`]; serde serializes the typed walk.
400///
401/// This is the table-loop analogue of [`crate::text::DvbText`]: it wraps the
402/// raw `descriptor()` sequence (the variable-length region inside a table) and
403/// decodes — i.e. dispatches each entry to a typed [`AnyDescriptor`] — only on
404/// demand. Parsing stays zero-copy; the typed walk happens when you call
405/// [`DescriptorLoop::iter`] or serialize.
406///
407/// ```
408/// use dvb_si::descriptors::{AnyDescriptor, DescriptorLoop};
409///
410/// // short_event (tag 0x4D, "eng" / "Hi") then an unknown private tag 0xA7.
411/// let raw = [
412///     0x4D, 0x07, b'e', b'n', b'g', 0x02, b'H', b'i', 0x00,
413///     0xA7, 0x02, 0xCA, 0xFE,
414/// ];
415/// let loop_ = DescriptorLoop::new(&raw);
416/// let items: Vec<_> = loop_.iter().collect();
417/// assert_eq!(items.len(), 2);
418/// assert!(matches!(items[0].as_ref().unwrap(), AnyDescriptor::ShortEvent(_)));
419/// assert_eq!(loop_.raw(), &raw[..]); // bytes preserved verbatim
420/// ```
421#[derive(Clone, Copy, PartialEq, Eq, Hash, Default)]
422#[cfg_attr(feature = "yoke", derive(yoke::Yokeable))]
423pub struct DescriptorLoop<'a>(&'a [u8]);
424
425/// Bytes of a descriptor header: `descriptor_tag` (1) + `descriptor_length` (1).
426const DESC_HEADER_LEN: usize = 2;
427
428impl<'a> DescriptorLoop<'a> {
429    /// Wrap a raw descriptor-loop slice (the `descriptor()` bytes only — no
430    /// enclosing length field).
431    #[must_use]
432    pub const fn new(raw: &'a [u8]) -> Self {
433        Self(raw)
434    }
435
436    /// The raw wire bytes of the loop, verbatim. These are what a serializer
437    /// writes back; use them for the byte length of the loop.
438    #[must_use]
439    pub const fn raw(&self) -> &'a [u8] {
440        self.0
441    }
442
443    /// Lazily walk the loop, yielding one typed [`AnyDescriptor`] per entry
444    /// (or [`AnyDescriptor::Unknown`] for tags with no implementation).
445    /// Delegates to [`parse_loop`]; never panics.
446    #[must_use]
447    pub fn iter(&self) -> DescriptorIter<'a> {
448        parse_loop(self.0)
449    }
450
451    /// Walk the loop's `(tag, body)` pairs by the TLV structure only — never
452    /// typed-parses a body. Each entry is `descriptor_tag` (1 byte) +
453    /// `descriptor_length` (1 byte) + that many body bytes; a truncated final
454    /// entry (declared length running past the buffer) simply ends iteration.
455    ///
456    /// Use this for body-agnostic structural scans (e.g. "is tag X present?")
457    /// where a malformed or empty body should not hide the tag — unlike
458    /// [`iter`](Self::iter), which yields `Err`/`Unknown` for bodies that fail
459    /// to typed-parse.
460    pub fn raw_tags(&self) -> impl Iterator<Item = (u8, &'a [u8])> {
461        let b = self.0;
462        let mut pos = 0usize;
463        core::iter::from_fn(move || {
464            if pos + DESC_HEADER_LEN > b.len() {
465                return None;
466            }
467            let tag = b[pos];
468            let len = b[pos + 1] as usize;
469            let end = pos + DESC_HEADER_LEN + len;
470            if end > b.len() {
471                return None; // truncated final entry
472            }
473            let body = &b[pos + DESC_HEADER_LEN..end];
474            pos = end;
475            Some((tag, body))
476        })
477    }
478
479    /// True if the loop contains a descriptor with `tag`, regardless of whether
480    /// its body parses (or even fits). Pure structural walk: checks the
481    /// `descriptor_tag` byte at each entry header, so an empty (`[tag, 0x00]`)
482    /// or truncated (`[tag, 0x01]`) descriptor still counts as present.
483    #[must_use]
484    pub fn contains_tag(&self, tag: u8) -> bool {
485        let b = self.0;
486        let mut pos = 0usize;
487        while pos < b.len() {
488            if b[pos] == tag {
489                return true;
490            }
491            // Need the length byte to advance to the next header; without it
492            // the tag byte just checked was the last thing in the loop.
493            if pos + 1 >= b.len() {
494                break;
495            }
496            pos += DESC_HEADER_LEN + b[pos + 1] as usize;
497        }
498        false
499    }
500
501    /// Walk this loop through a [`DescriptorRegistry`](crate::descriptors::registry::DescriptorRegistry)
502    /// so registered private/custom descriptors (and PDS-scoped tags) are
503    /// produced as [`AnyDescriptor::Other`]; mirrors [`iter`](Self::iter)
504    /// otherwise.
505    ///
506    /// See [`DescriptorRegistry::parse_loop`](crate::descriptors::registry::DescriptorRegistry::parse_loop)
507    /// for precedence rules and PDS-scoped dispatch.
508    #[must_use]
509    pub fn iter_with<'r>(
510        &self,
511        registry: &'r crate::descriptors::registry::DescriptorRegistry,
512    ) -> crate::descriptors::registry::RegistryIter<'r, 'a> {
513        registry.parse_loop(self.0)
514    }
515
516    /// Walk this loop through both a [`DescriptorRegistry`](crate::descriptors::registry::DescriptorRegistry) and an
517    /// [`ExtensionRegistry`](crate::descriptors::extension::registry::ExtensionRegistry),
518    /// so that custom-registered extension bodies (tag `0x7F` with a known
519    /// `descriptor_tag_extension`) are surfaced as
520    /// [`ExtIterItem::CustomExtension`](crate::descriptors::registry::ExtIterItem::CustomExtension) with the type-erased value available
521    /// for downcast. All other descriptors follow the normal
522    /// [`iter_with`](Self::iter_with) precedence.
523    ///
524    /// This is the only way to reach private extension bodies during a
525    /// descriptor-loop walk; [`iter_with`](Self::iter_with) alone produces
526    /// the built-in [`ExtensionBody::Raw`](crate::descriptors::extension::ExtensionBody::Raw)
527    /// for unrecognised `descriptor_tag_extension` values.
528    #[must_use]
529    pub fn iter_with_extensions<'r>(
530        &self,
531        desc_reg: &'r crate::descriptors::registry::DescriptorRegistry,
532        ext_reg: &'r crate::descriptors::extension::registry::ExtensionRegistry,
533    ) -> crate::descriptors::registry::ExtRegistryIter<'r, 'a> {
534        crate::descriptors::registry::ExtRegistryIter::new(desc_reg, ext_reg, self.0)
535    }
536}
537
538impl<'a> core::ops::Deref for DescriptorLoop<'a> {
539    /// Derefs to the raw wire bytes — `len()`/indexing are **byte counts for
540    /// serialization, not entry counts**. To count entries, use
541    /// [`DescriptorLoop::iter`].
542    type Target = [u8];
543    fn deref(&self) -> &[u8] {
544        self.0
545    }
546}
547
548impl<'a> From<&'a [u8]> for DescriptorLoop<'a> {
549    fn from(raw: &'a [u8]) -> Self {
550        Self(raw)
551    }
552}
553
554impl core::fmt::Debug for DescriptorLoop<'_> {
555    /// Cheap: prints the byte length, not the decoded entries.
556    fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
557        write!(f, "DescriptorLoop(<{} bytes>)", self.0.len())
558    }
559}
560
561impl<'a> IntoIterator for &DescriptorLoop<'a> {
562    type Item = crate::Result<AnyDescriptor<'a>>;
563    type IntoIter = DescriptorIter<'a>;
564    fn into_iter(self) -> Self::IntoIter {
565        self.iter()
566    }
567}
568
569#[cfg(feature = "serde")]
570impl serde::Serialize for DescriptorLoop<'_> {
571    /// Serializes as a sequence of the typed walk: each `Ok(d)` becomes the
572    /// [`AnyDescriptor`] (camelCase external tagging), and each `Err(e)`
573    /// becomes a `{"parseError": "<Display>"}` map — parse errors are surfaced,
574    /// never silently dropped.
575    fn serialize<S: serde::Serializer>(&self, s: S) -> Result<S::Ok, S::Error> {
576        struct Entry<'a>(crate::Result<AnyDescriptor<'a>>);
577        impl serde::Serialize for Entry<'_> {
578            fn serialize<S: serde::Serializer>(&self, s: S) -> Result<S::Ok, S::Error> {
579                match &self.0 {
580                    Ok(d) => d.serialize(s),
581                    Err(e) => {
582                        use serde::ser::SerializeMap;
583                        let mut m = s.serialize_map(Some(1))?;
584                        m.serialize_entry("parseError", &e.to_string())?;
585                        m.end()
586                    }
587                }
588            }
589        }
590        s.collect_seq(self.iter().map(Entry))
591    }
592}
593// Serialize-only: the typed walk decodes DVB text and dispatches per-tag —
594// there is no lossless way to reconstruct the raw loop bytes from the
595// serialized form. Structs holding a DescriptorLoop derive Serialize only.
596// To reconstruct, keep the wire bytes and re-`parse` the table.
597
598#[cfg(test)]
599mod tests {
600    use super::*;
601
602    #[test]
603    fn contains_tag_detects_present_tag_regardless_of_body() {
604        // Well-formed AC-3 (0x6A) with a flags byte.
605        assert!(DescriptorLoop::new(&[0x6A, 0x01, 0x80]).contains_tag(0x6A));
606        // Empty body — malformed for AC-3, but the tag is present.
607        assert!(DescriptorLoop::new(&[0x6A, 0x00]).contains_tag(0x6A));
608        // Truncated — declares 1 body byte that isn't there; tag still present.
609        assert!(DescriptorLoop::new(&[0x6A, 0x01]).contains_tag(0x6A));
610        // Present after an earlier well-formed entry.
611        assert!(DescriptorLoop::new(&[0x09, 0x02, 0x00, 0x00, 0x7A, 0x00]).contains_tag(0x7A));
612        // Absent.
613        assert!(!DescriptorLoop::new(&[0x09, 0x02, 0x00, 0x00]).contains_tag(0x6A));
614        assert!(!DescriptorLoop::new(&[]).contains_tag(0x6A));
615    }
616
617    #[test]
618    fn raw_tags_walks_tlv_without_typed_parsing() {
619        // Two complete entries: tag 0x6A (empty body) + tag 0x40 (2-byte body).
620        let loop_ = DescriptorLoop::new(&[0x6A, 0x00, 0x40, 0x02, 0xAA, 0xBB]);
621        let pairs: Vec<_> = loop_.raw_tags().collect();
622        assert_eq!(pairs, vec![(0x6A, &[][..]), (0x40, &[0xAA, 0xBB][..])]);
623        // First entry [0x40,len=1,0xAA] complete, then a truncated [0x6A, len=5]
624        // entry whose declared body runs past the buffer.
625        let trunc = DescriptorLoop::new(&[0x40, 0x01, 0xAA, 0x6A, 0x05]);
626        let tags: Vec<u8> = trunc.raw_tags().map(|(t, _)| t).collect();
627        assert_eq!(tags, vec![0x40]); // the truncated 0x6A entry is dropped by raw_tags
628                                      // ...but contains_tag still sees the truncated tag.
629        assert!(trunc.contains_tag(0x6A));
630    }
631
632    #[test]
633    fn unknown_tag_yields_unknown_with_body_sans_header() {
634        // tag 0xA7 (no typed impl), length 2, body [0xDE, 0xAD].
635        let bytes = [0xA7, 0x02, 0xDE, 0xAD];
636        let items: Vec<_> = parse_loop(&bytes).collect();
637        assert_eq!(items.len(), 1);
638        match items[0].as_ref().unwrap() {
639            AnyDescriptor::Unknown { tag, body } => {
640                assert_eq!(*tag, 0xA7);
641                assert_eq!(*body, &[0xDE, 0xAD]);
642            }
643            other => panic!("expected Unknown, got {other:?}"),
644        }
645    }
646
647    #[test]
648    fn empty_loop_yields_nothing() {
649        assert_eq!(parse_loop(&[]).count(), 0);
650    }
651
652    #[test]
653    fn logical_channel_0x83_is_not_dispatched() {
654        // 0x83 has a variant but no dispatcher entry → Unknown, never panics.
655        let bytes = [0x83, 0x04, 0x00, 0x01, 0xFC, 0x01];
656        let items: Vec<_> = parse_loop(&bytes).collect();
657        assert_eq!(items.len(), 1);
658        assert!(matches!(
659            items[0].as_ref().unwrap(),
660            AnyDescriptor::Unknown { tag: 0x83, .. }
661        ));
662    }
663
664    #[test]
665    fn descriptor_loop_iter_matches_parse_loop() {
666        let raw = [
667            0x4D, 0x07, b'e', b'n', b'g', 0x02, b'H', b'i', 0x00, // short_event
668            0xA7, 0x02, 0xCA, 0xFE, // unknown 0xA7
669        ];
670        let via_loop: Vec<_> = DescriptorLoop::new(&raw)
671            .iter()
672            .map(|r| format!("{r:?}"))
673            .collect();
674        let via_fn: Vec<_> = parse_loop(&raw).map(|r| format!("{r:?}")).collect();
675        assert_eq!(via_loop, via_fn);
676        // raw()/Deref expose the wire bytes (byte length, not entry count).
677        assert_eq!(DescriptorLoop::new(&raw).raw(), &raw[..]);
678        assert_eq!(DescriptorLoop::new(&raw).len(), raw.len());
679        // IntoIterator for &DescriptorLoop.
680        let count = (&DescriptorLoop::new(&raw)).into_iter().count();
681        assert_eq!(count, 2);
682    }
683
684    #[test]
685    fn descriptor_loop_debug_is_cheap() {
686        let raw = [0x4D, 0x02, 0x01, 0x02];
687        assert_eq!(
688            format!("{:?}", DescriptorLoop::new(&raw)),
689            "DescriptorLoop(<4 bytes>)"
690        );
691    }
692
693    #[test]
694    fn iter_with_custom_tag_yields_other() {
695        use crate::descriptors::registry::DescriptorRegistry;
696        use crate::traits::DescriptorDef;
697
698        #[derive(Debug, PartialEq, Eq)]
699        #[cfg_attr(feature = "serde", derive(serde::Serialize))]
700        struct MyTag0xA7 {
701            x: u8,
702        }
703
704        impl<'a> dvb_common::Parse<'a> for MyTag0xA7 {
705            type Error = crate::error::Error;
706            fn parse(bytes: &'a [u8]) -> crate::Result<Self> {
707                if bytes.len() < 3 {
708                    return Err(crate::error::Error::BufferTooShort {
709                        need: 3,
710                        have: bytes.len(),
711                        what: "MyTag0xA7",
712                    });
713                }
714                Ok(Self { x: bytes[2] })
715            }
716        }
717
718        impl<'a> DescriptorDef<'a> for MyTag0xA7 {
719            const TAG: u8 = 0xA7;
720            const NAME: &'static str = "MY_TAG_0xA7";
721        }
722
723        let mut reg = DescriptorRegistry::new();
724        reg.register::<MyTag0xA7>();
725
726        let raw = [
727            0x4D, 0x07, b'e', b'n', b'g', 0x02, b'H', b'i', 0x00, 0xA7, 0x02, 0xCA, 0xFE,
728        ];
729        let loop_ = DescriptorLoop::new(&raw);
730        let items: Vec<_> = loop_.iter_with(&reg).collect::<Result<_, _>>().unwrap();
731        assert_eq!(items.len(), 2);
732        assert!(matches!(items[0], AnyDescriptor::ShortEvent(_)));
733        match &items[1] {
734            AnyDescriptor::Other { tag, value } => {
735                assert_eq!(*tag, 0xA7);
736                assert_eq!(value.downcast_ref::<MyTag0xA7>().unwrap().x, 0xCA);
737            }
738            other => panic!("expected Other, got {other:?}"),
739        }
740    }
741
742    #[test]
743    fn iter_with_empty_registry_matches_iter_for_builtin() {
744        let raw = [
745            0x4D, 0x07, b'e', b'n', b'g', 0x02, b'H', b'i', 0x00, 0xA7, 0x02, 0xCA, 0xFE,
746        ];
747        let loop_ = DescriptorLoop::new(&raw);
748        let reg = crate::descriptors::registry::DescriptorRegistry::new();
749        let via_iter: Vec<_> = loop_.iter().collect();
750        let via_iter_with: Vec<_> = loop_.iter_with(&reg).collect();
751
752        assert_eq!(via_iter.len(), via_iter_with.len());
753        for (a, b) in via_iter.iter().zip(via_iter_with.iter()) {
754            match (a, b) {
755                (Ok(AnyDescriptor::ShortEvent(_)), Ok(AnyDescriptor::ShortEvent(_))) => {}
756                (
757                    Ok(AnyDescriptor::Unknown { tag: t1, body: b1 }),
758                    Ok(AnyDescriptor::Unknown { tag: t2, body: b2 }),
759                ) => {
760                    assert_eq!(t1, t2);
761                    assert_eq!(b1, b2);
762                }
763                (Err(_), Err(_)) => {}
764                (l, r) => panic!("mismatch: {l:?} vs {r:?}"),
765            }
766        }
767    }
768
769    #[cfg(feature = "serde")]
770    #[test]
771    fn descriptor_loop_serializes_typed_unknown_and_parse_error() {
772        // [valid short_event, unknown tag 0xA7, truncated final entry].
773        // A truncated final entry (declared len 5, only 1 body byte present)
774        // makes the walker yield a final Err → {"parseError": …}.
775        let raw = [
776            0x4D, 0x07, b'e', b'n', b'g', 0x02, b'H', b'i', 0x00, // short_event
777            0xA7, 0x02, 0xCA, 0xFE, // unknown 0xA7
778            0x55, 0x05, 0x00, // parental_rating header claims 5 bytes; only 1 present
779        ];
780        let v = serde_json::to_value(DescriptorLoop::new(&raw)).unwrap();
781        let arr = v.as_array().expect("sequence");
782        assert_eq!(arr.len(), 3);
783        // 1. typed short_event under the camelCase variant key.
784        assert!(arr[0].get("shortEvent").is_some(), "got {}", arr[0]);
785        assert_eq!(arr[0]["shortEvent"]["event_name"], "Hi");
786        // 2. unknown tag carries its raw body bytes.
787        let unknown = arr[1].get("unknown").expect("unknown variant");
788        assert_eq!(unknown["tag"], 0xA7);
789        assert_eq!(unknown["body"], serde_json::json!([0xCA, 0xFE]));
790        // 3. truncated entry → parseError, never silently dropped.
791        assert!(
792            arr[2].get("parseError").is_some(),
793            "expected parseError, got {}",
794            arr[2]
795        );
796    }
797}