mlt_core/dump/
model.rs

1//! Data model for the annotated binary dump (see [`crate::dump`]).
2
3use std::fmt::{Display, Formatter, Result as FmtResult};
4
5use serde::ser::SerializeStruct as _;
6use serde::{Serialize, Serializer};
7
8#[cfg(feature = "unstable-v2")]
9use crate::wire::Alp;
10use crate::wire::StreamMeta;
11
12/// Whether a region is tile metadata or an opaque data payload.
13#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize)]
14#[serde(rename_all = "camelCase")]
15pub enum RegionKind {
16    /// Framing, schema, or stream-header bytes, annotated byte- and bit-for-byte.
17    Meta,
18    /// A stream payload, rendered as raw hex plus best-effort decoded values.
19    DataBlob,
20}
21
22/// How a [`RegionKind::DataBlob`] payload is decoded for display.
23///
24/// Best-effort: on any decode error the renderer falls back to raw hex.
25#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize)]
26#[serde(tag = "kind", rename_all = "camelCase")]
27pub enum DecodeHint {
28    /// Nullability bitmap (byte-RLE -> packed bits).
29    Presence,
30    /// Boolean data stream (byte-RLE -> bools).
31    Bool,
32    /// Signed 32-bit integers (`i8`/`i32` columns).
33    I32,
34    /// Unsigned 32-bit integers (`u8`/`u32` columns, offsets, lengths).
35    U32,
36    /// Signed 64-bit integers (`i64` columns).
37    I64,
38    /// ALP offsets, put back on their frame of reference.
39    #[cfg(feature = "unstable-v2")]
40    Alp(#[serde(serialize_with = "serialize_alp")] Alp),
41    /// Unsigned 64-bit integers (`u64` columns, 64-bit ids).
42    U64,
43    /// 32-bit floats.
44    F32,
45    /// 64-bit floats.
46    F64,
47    /// Opaque bytes (string / dictionary / FSST payloads); shown as hex + UTF-8 preview.
48    Bytes,
49    /// A raw LSB0 bitfield, not a stream: `ceil(num_values/8)` packed bytes.
50    #[cfg(feature = "unstable-v2")]
51    PackedBits,
52}
53
54/// ALP's parameters flattened beside the hint's tag, keeping `Alp`'s fields crate-private.
55#[cfg(feature = "unstable-v2")]
56fn serialize_alp<S: Serializer>(alp: &Alp, s: S) -> Result<S::Ok, S::Error> {
57    let mut st = s.serialize_struct("Alp", 3)?;
58    st.serialize_field("e", &alp.scale.e)?;
59    st.serialize_field("f", &alp.scale.f)?;
60    st.serialize_field("base", &alp.base)?;
61    st.end()
62}
63
64/// One sub-field of a bit-packed byte, e.g. a nibble of `stream_type`.
65#[derive(Debug, Clone, Serialize)]
66pub struct BitField {
67    /// Inclusive high bit index (7..=0, MSB first).
68    hi: u8,
69    /// Inclusive low bit index.
70    lo: u8,
71    /// The extracted field value, shifted down to bit 0.
72    #[serde(serialize_with = "serialize_as_number")]
73    raw: u64,
74    /// Human-readable meaning, e.g. `"physical = VarInt"`.
75    meaning: String,
76}
77
78impl BitField {
79    /// One field of a packed byte, located by the same mask constant the parser reads it
80    /// with, so the dump cannot drift from the wire format.
81    ///
82    /// `raw` is the masked bits shifted down to bit 0, so it renders as the field's own
83    /// value rather than its in-byte position. `mask` must be one contiguous run of bits.
84    pub fn mask(mask: u8, byte: u8, meaning: impl Into<String>) -> Self {
85        let (hi, lo) = mask_bounds(mask);
86        Self {
87            hi,
88            lo,
89            raw: u64::from((byte & mask) >> lo),
90            meaning: meaning.into(),
91        }
92    }
93
94    /// A one-bit flag, worded as a phrase instead of a `0`/`1`.
95    ///
96    /// The renderer already prefixes every bit line with `bit N = V ->`, so `when_set`
97    /// and `when_clear` should say what the bit asserts about the tile rather than
98    /// repeat its value.
99    #[must_use]
100    pub fn flag(mask: u8, byte: u8, when_set: &str, when_clear: &str) -> Self {
101        debug_assert_eq!(mask.count_ones(), 1, "a flag occupies exactly one bit");
102        let meaning = if byte & mask == 0 {
103            when_clear
104        } else {
105            when_set
106        };
107        Self::mask(mask, byte, meaning)
108    }
109
110    /// What this field means in prose, without the `bit N = V ->` prefix
111    /// [`Display`] puts in front of it.
112    #[must_use]
113    pub fn meaning(&self) -> &str {
114        &self.meaning
115    }
116}
117
118impl Display for BitField {
119    /// `bit 7 = 1 -> meaning`, or `bits 6-4 = 011 -> meaning` for a multi-bit field.
120    ///
121    /// The value is printed as wide as the field, so the leading zeros of a nibble are
122    /// not mistaken for a narrower field. The caller adds any indent and color.
123    fn fmt(&self, f: &mut Formatter<'_>) -> FmtResult {
124        if self.hi == self.lo {
125            write!(f, "bit {}", self.hi)?;
126        } else {
127            write!(f, "bits {}-{}", self.hi, self.lo)?;
128        }
129        let width = usize::from(self.hi - self.lo + 1);
130        write!(f, " = {:0width$b} -> {}", self.raw, self.meaning)
131    }
132}
133
134/// Inclusive `(hi, lo)` bit indices spanned by `mask`.
135#[expect(
136    clippy::cast_possible_truncation,
137    reason = "a u8 mask spans at most 8 bits"
138)]
139fn mask_bounds(mask: u8) -> (u8, u8) {
140    debug_assert!(mask != 0, "a bit field needs at least one bit");
141    let lo = mask.trailing_zeros() as u8;
142    let hi = (u8::BITS - 1 - mask.leading_zeros()) as u8;
143    debug_assert_eq!(
144        u32::from(hi - lo) + 1,
145        mask.count_ones(),
146        "a bit field's mask must be one contiguous run of bits"
147    );
148    (hi, lo)
149}
150
151/// Stream metadata attached to a [`RegionKind::DataBlob`] so the renderer can decode it.
152#[derive(Debug, Clone, Copy)]
153pub struct BlobInfo {
154    pub meta: StreamMeta,
155    pub hint: DecodeHint,
156}
157
158impl Serialize for BlobInfo {
159    /// Flattens `meta` into display strings, so the wire vocabulary stays out of the JSON API.
160    fn serialize<S: Serializer>(&self, s: S) -> Result<S::Ok, S::Error> {
161        let mut st = s.serialize_struct("BlobInfo", 5)?;
162        st.serialize_field("streamType", &self.meta.stream_type.to_string())?;
163        st.serialize_field("logical", &self.meta.encoding.logical.to_string())?;
164        st.serialize_field("physical", &self.meta.encoding.physical.to_string())?;
165        st.serialize_field("numValues", &self.meta.num_values)?;
166        st.serialize_field("hint", &self.hint)?;
167        st.end()
168    }
169}
170
171/// Serialize an offset or a count as a 32-bit number.
172///
173/// Only the payload values a tile actually holds are 64-bit, and a transport that maps
174/// 64-bit integers to `BigInt` would otherwise hand every offset across as one.
175fn serialize_as_number<T, S>(v: &T, s: S) -> Result<S::Ok, S::Error>
176where
177    T: Copy + TryInto<u32>,
178    S: Serializer,
179{
180    let n = (*v)
181        .try_into()
182        .map_err(|_| serde::ser::Error::custom("offset or count exceeds u32"))?;
183    s.serialize_u32(n)
184}
185
186/// A single annotated span of the tile buffer.
187///
188/// Emitted in pre-order. Containers bracket their children and may overlap them.
189/// Leaf regions partition the buffer exactly; the coverage test relies on this.
190#[derive(Debug, Clone, Serialize)]
191#[serde(rename_all = "camelCase")]
192pub struct Region {
193    /// Absolute byte offset into the tile buffer.
194    #[serde(serialize_with = "serialize_as_number")]
195    pub offset: usize,
196    #[serde(serialize_with = "serialize_as_number")]
197    pub len: usize,
198    /// Nesting depth, for indentation.
199    #[serde(serialize_with = "serialize_as_number")]
200    pub depth: usize,
201    /// Short label, e.g. `"column[2].type"` or `"num_values"`.
202    pub label: String,
203    /// Rendered scalar value (varint value, string, enum name), if any.
204    pub value: Option<String>,
205    /// Bit-level breakdown; empty unless this is a bit-packed byte.
206    pub bits: Vec<BitField>,
207    pub kind: RegionKind,
208    /// True for structural groups that span their children (excluded from coverage).
209    pub container: bool,
210    /// Present for `DataBlob` regions that carry decodable stream metadata.
211    pub blob: Option<BlobInfo>,
212}
213
214/// The full annotation of a tile: a flat, depth-tagged region list.
215#[derive(Serialize)]
216#[serde(rename_all = "camelCase")]
217pub struct DumpTree {
218    #[serde(serialize_with = "serialize_as_number")]
219    pub buf_len: usize,
220    /// Regions in pre-order (containers before their children).
221    pub regions: Vec<Region>,
222}
223
224/// Label of the synthetic leaf covering the bytes a bailed-out walk never reached.
225pub const UNANNOTATED: &str = "<unannotated>";
226
227/// The `idx`-th top-level layer's regions, or `None` when the tile has no such layer.
228///
229/// Containers span their children, so the layer is selected by byte range, not by depth.
230#[must_use]
231pub fn filter_layer(tree: &DumpTree, idx: usize) -> Option<DumpTree> {
232    let layer = tree
233        .regions
234        .iter()
235        .filter(|r| r.depth == 0 && r.container)
236        .nth(idx)?;
237    let start = layer.offset;
238    let end = layer.offset + layer.len;
239    let regions = tree
240        .regions
241        .iter()
242        .filter(|r| r.offset >= start && r.offset + r.len <= end)
243        .cloned()
244        .collect();
245    Some(DumpTree {
246        buf_len: tree.buf_len,
247        regions,
248    })
249}
mlt_core/dump/model.rs

mlt_core/dump/
model.rs
