mlt_core/encoder/
model.rs

1use std::borrow::Cow;
2#[cfg(feature = "unstable-v2")]
3use std::collections::HashSet;
4
5use derive_debug::Dbg;
6
7#[cfg(feature = "unstable-v2")]
8use crate::decoder::RleLayout;
9use crate::decoder::{DictionaryType, FastPForKind, GeometryValues, PhysicalEncoding, StreamType};
10use crate::encoder::geometry::VertexBufferType;
11use crate::encoder::{IntEncoder, StagedId, StagedProperty};
12#[cfg(feature = "unstable-v2")]
13use crate::encoder::{StagedInterior, StagedMValue, StagedNested, StagedNode};
14#[cfg(feature = "unstable-v2")]
15use crate::tile::MAX_NESTED_DEPTH;
16use crate::tile::{ColumnRole, Extent, reject_taken_name};
17use crate::{MltError, MltResult};
18
19/// Owned variant of `Unknown`.
20#[derive(Debug, Clone, Default, PartialEq)]
21pub struct EncodedUnknown {
22    pub(crate) tag: u8,
23    pub(crate) value: Vec<u8>,
24}
25
26impl EncodedUnknown {
27    pub fn new(tag: u8, value: Vec<u8>) -> MltResult<Self> {
28        if tag == 1 {
29            return Err(MltError::ParsingColumnType(tag));
30        }
31        Ok(Self { tag, value })
32    }
33
34    #[must_use]
35    pub fn tag(&self) -> u32 {
36        u32::from(self.tag)
37    }
38
39    #[must_use]
40    pub fn data(&self) -> &[u8] {
41        &self.value
42    }
43}
44
45/// Parameters derived from the vertex set of a feature collection, used to
46/// normalize coordinates before space-filling-curve key computation.
47#[derive(Debug, Clone, Copy, PartialEq)]
48pub struct CurveParams {
49    pub shift: u32,
50    pub bits: u32,
51}
52
53impl Default for CurveParams {
54    fn default() -> Self {
55        Self { shift: 0, bits: 1 }
56    }
57}
58
59impl CurveParams {
60    /// Compute params from a flat `[x0, y0, x1, y1, …]` vertex slice.
61    #[must_use]
62    pub fn from_vertices(vertices: &[i32]) -> Self {
63        if vertices.is_empty() {
64            return Self::default();
65        }
66        let (min, max) = vertices
67            .iter()
68            .fold((i32::MAX, i32::MIN), |(mn, mx), &v| (mn.min(v), mx.max(v)));
69        crate::codecs::hilbert::hilbert_curve_params_from_bounds(min, max)
70    }
71}
72
73/// Columnar layer data being prepared for encoding (stage 2 of the encoding pipeline).
74///
75/// Holds fully-owned columnar data. Constructed directly (synthetics, benches) or
76/// converted from [`TileLayer`](crate::TileLayer).
77/// Consumed by encoding via [`StagedLayer::encode_into`] or `StagedLayer::encode_explicit`
78/// (with explicit encoding mode enabled).
79#[derive(Debug, PartialEq, Clone)]
80pub struct StagedLayer {
81    pub(crate) name: String,
82    pub(crate) extent: Extent,
83    pub(crate) id: StagedId,
84    pub(crate) geometry: GeometryValues,
85    pub(crate) properties: Vec<StagedProperty>,
86    /// Vertex-scoped columns, which only a v2 layer can be written with.
87    #[cfg(feature = "unstable-v2")]
88    pub(crate) m_values: Vec<StagedMValue>,
89    /// Nested columns, which only a v2 layer can be written with.
90    #[cfg(feature = "unstable-v2")]
91    pub(crate) nested: Vec<StagedNested>,
92}
93
94#[cfg_attr(not(feature = "__private"), allow(dead_code))]
95impl StagedLayer {
96    pub fn new(
97        name: impl Into<String>,
98        extent: u32,
99        id: StagedId,
100        geometry: GeometryValues,
101        properties: Vec<StagedProperty>,
102    ) -> MltResult<Self> {
103        let (name, extent, _) = validate_staged(name, extent, &id, &geometry, &properties)?;
104        Ok(Self {
105            name,
106            extent,
107            id,
108            geometry,
109            properties,
110            #[cfg(feature = "unstable-v2")]
111            m_values: Vec::new(),
112            #[cfg(feature = "unstable-v2")]
113            nested: Vec::new(),
114        })
115    }
116
117    /// As [`Self::new`], with the layer's vertex-scoped columns.
118    #[cfg(feature = "unstable-v2")]
119    pub fn with_m_values(
120        name: impl Into<String>,
121        extent: u32,
122        id: StagedId,
123        geometry: GeometryValues,
124        properties: Vec<StagedProperty>,
125        m_values: Vec<StagedMValue>,
126    ) -> MltResult<Self> {
127        let (name, extent, columns) = validate_staged(name, extent, &id, &geometry, &properties)?;
128        validate_m_values(&geometry, &columns, &m_values)?;
129        Ok(Self {
130            name,
131            extent,
132            id,
133            geometry,
134            properties,
135            m_values,
136            nested: Vec::new(),
137        })
138    }
139
140    /// As [`Self::new`], with both the layer's vertex-scoped and its nested columns.
141    #[cfg(feature = "unstable-v2")]
142    pub fn with_nested(
143        name: impl Into<String>,
144        extent: u32,
145        id: StagedId,
146        geometry: GeometryValues,
147        properties: Vec<StagedProperty>,
148        m_values: Vec<StagedMValue>,
149        nested: Vec<StagedNested>,
150    ) -> MltResult<Self> {
151        let (name, extent, columns) = validate_staged(name, extent, &id, &geometry, &properties)?;
152        validate_m_values(&geometry, &columns, &m_values)?;
153        validate_nested(&geometry, &columns, &m_values, &nested)?;
154        Ok(Self {
155            name,
156            extent,
157            id,
158            geometry,
159            properties,
160            m_values,
161            nested,
162        })
163    }
164
165    #[must_use]
166    pub fn name(&self) -> &str {
167        &self.name
168    }
169
170    #[must_use]
171    pub fn extent(&self) -> Extent {
172        self.extent
173    }
174
175    #[must_use]
176    pub fn id(&self) -> &StagedId {
177        &self.id
178    }
179
180    #[must_use]
181    pub fn geometry(&self) -> &GeometryValues {
182        &self.geometry
183    }
184
185    #[must_use]
186    pub fn properties(&self) -> &[StagedProperty] {
187        &self.properties
188    }
189
190    #[cfg(feature = "unstable-v2")]
191    #[must_use]
192    pub fn m_values(&self) -> &[StagedMValue] {
193        &self.m_values
194    }
195
196    #[cfg(feature = "unstable-v2")]
197    #[must_use]
198    pub fn nested(&self) -> &[StagedNested] {
199        &self.nested
200    }
201}
202
203/// Check a staged layer's name, extent and columns, returning the validated fields and the column names.
204fn validate_staged<'p>(
205    name: impl Into<String>,
206    extent: u32,
207    id: &StagedId,
208    geometry: &GeometryValues,
209    properties: &'p [StagedProperty],
210) -> MltResult<(String, Extent, Vec<Cow<'p, str>>)> {
211    let name = name.into();
212    if name.is_empty() {
213        return Err(MltError::MissingLayerName);
214    }
215    let extent = Extent::new(extent)?;
216    let feature_count = geometry.feature_count();
217    if let Some(actual) = id.feature_count()
218        && actual != feature_count
219    {
220        return Err(MltError::StagedFeatureCountMismatch {
221            column: "id".into(),
222            expected: feature_count,
223            actual,
224        });
225    }
226    // A shared dictionary's `name()` is only its prefix, which may repeat.
227    // Its real columns are `{prefix}{suffix}`, so those are the names the layer's namespace holds.
228    let mut columns: Vec<Cow<'p, str>> = Vec::with_capacity(properties.len());
229    for property in properties {
230        let actual = property.feature_count();
231        if actual != feature_count {
232            return Err(MltError::StagedFeatureCountMismatch {
233                column: property.name().to_string(),
234                expected: feature_count,
235                actual,
236            });
237        }
238        if let StagedProperty::SharedDict(sd) = property {
239            for item in &sd.items {
240                let name = format!("{}{}", sd.prefix, item.suffix);
241                reject_taken_name(&name, ColumnRole::Property, property_columns(&columns))?;
242                columns.push(Cow::Owned(name));
243            }
244        } else {
245            reject_taken_name(
246                property.name(),
247                ColumnRole::Property,
248                property_columns(&columns),
249            )?;
250            columns.push(Cow::Borrowed(property.name()));
251        }
252    }
253    Ok((name, extent, columns))
254}
255
256/// Pair every property column's name with its kind, as the uniqueness check reads them.
257fn property_columns<'a>(
258    columns: &'a [Cow<'a, str>],
259) -> impl Iterator<Item = (&'a str, ColumnRole)> + 'a {
260    columns
261        .iter()
262        .map(|name| (name.as_ref(), ColumnRole::Property))
263}
264
265/// Check every m-value column against the geometry it runs over: a unique name, a mask of one
266/// bit per feature, and one value per vertex of every feature it marks present.
267#[cfg(feature = "unstable-v2")]
268fn validate_m_values(
269    geometry: &GeometryValues,
270    columns: &[Cow<'_, str>],
271    m_values: &[StagedMValue],
272) -> MltResult<()> {
273    let feature_count = geometry.feature_count();
274    let mut seen: Vec<&str> = Vec::with_capacity(m_values.len());
275    for column in m_values {
276        reject_taken_name(
277            column.name(),
278            ColumnRole::MValue,
279            property_columns(columns).chain(seen.iter().map(|n| (*n, ColumnRole::MValue))),
280        )?;
281        seen.push(column.name());
282        if let Some(actual) = column.feature_count()
283            && actual != feature_count
284        {
285            return Err(MltError::StagedFeatureCountMismatch {
286                column: column.name().to_string(),
287                expected: feature_count,
288                actual,
289            });
290        }
291        let mut expected = 0;
292        for index in 0..feature_count {
293            let present = column
294                .presence
295                .as_ref()
296                .is_none_or(|mask| mask.get(index).copied().unwrap_or(false));
297            if present {
298                expected += geometry.vertex_count(index)?;
299            }
300        }
301        if expected != column.values().count() {
302            return Err(MltError::MValueColumnLengthMismatch {
303                name: column.name().to_string(),
304                expected,
305                actual: column.values().count(),
306            });
307        }
308    }
309    Ok(())
310}
311
312/// Check every nested column against the layer: a unique name, a mask of one bit
313/// per feature, and a tree the wire can hold.
314#[cfg(feature = "unstable-v2")]
315fn validate_nested(
316    geometry: &GeometryValues,
317    columns: &[Cow<'_, str>],
318    m_values: &[StagedMValue],
319    nested: &[StagedNested],
320) -> MltResult<()> {
321    let feature_count = geometry.feature_count();
322    let mut seen: Vec<&str> = Vec::with_capacity(nested.len());
323    for column in nested {
324        reject_taken_name(
325            column.name(),
326            ColumnRole::Nested,
327            property_columns(columns)
328                .chain(m_values.iter().map(|c| (c.name(), ColumnRole::MValue)))
329                .chain(seen.iter().map(|n| (*n, ColumnRole::Nested))),
330        )?;
331        seen.push(column.name());
332        if column.root.parent_count() != feature_count {
333            return Err(MltError::StagedFeatureCountMismatch {
334                column: column.name().to_string(),
335                expected: feature_count,
336                actual: column.root.parent_count(),
337            });
338        }
339        validate_nested_interior(&column.root, 1)?;
340    }
341    Ok(())
342}
343
344/// Check one node of a nested tree against the depth the wire allows, then its own children.
345#[cfg(feature = "unstable-v2")]
346fn validate_nested_node(node: &StagedNode, depth: usize) -> MltResult<()> {
347    if depth > MAX_NESTED_DEPTH {
348        return Err(MltError::NestedTooDeep(depth));
349    }
350    match node {
351        StagedNode::Leaf(_) => Ok(()),
352        StagedNode::Interior(interior) => validate_nested_interior(interior, depth),
353    }
354}
355
356/// Check an interior node's children: unique field names, and no level the wire cannot reach.
357#[cfg(feature = "unstable-v2")]
358fn validate_nested_interior(interior: &StagedInterior, depth: usize) -> MltResult<()> {
359    match interior {
360        StagedInterior::Struct(node) => {
361            let mut seen: HashSet<&str> = HashSet::with_capacity(node.fields.len());
362            for (name, field) in &node.fields {
363                if !seen.insert(name.as_str()) {
364                    return Err(MltError::DuplicateFieldName(name.clone()));
365                }
366                validate_nested_node(field, depth + 1)?;
367            }
368            Ok(())
369        }
370        StagedInterior::List(node) => validate_nested_node(&node.element, depth + 1),
371        StagedInterior::Map(node) => validate_nested_node(&node.value, depth + 1),
372    }
373}
374
375/// Which wire format layers are encoded to.
376#[cfg(feature = "unstable-v2")]
377#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, Default)]
378#[cfg_attr(feature = "arbitrary", derive(arbitrary::Arbitrary))]
379pub enum WireVersion {
380    /// Tag `0x01` - the stable v1 format.
381    #[default]
382    V01,
383    /// Tag `0x02` - the experimental v2 format (see `docs/migrating-to-v2.md`).
384    ///
385    /// Currently limited to ID, scalar, string, shared-dictionary and geometry columns.
386    V02,
387}
388
389#[cfg(feature = "unstable-v2")]
390impl WireVersion {
391    /// The layer tag byte identifying this format on the wire.
392    #[must_use]
393    pub(crate) fn tag(self) -> u8 {
394        match self {
395            Self::V01 => 1,
396            Self::V02 => 2,
397        }
398    }
399
400    /// The `FastPFor` block size and word order used by this format.
401    #[must_use]
402    pub(crate) fn fastpfor_kind(self) -> FastPForKind {
403        match self {
404            Self::V01 => FastPForKind::Block256Be,
405            Self::V02 => FastPForKind::Block128Le,
406        }
407    }
408
409    /// The RLE stream data layout used by this format.
410    #[must_use]
411    pub(crate) fn rle_layout(self) -> RleLayout {
412        match self {
413            Self::V01 => RleLayout::Split,
414            Self::V02 => RleLayout::Interleaved,
415        }
416    }
417
418    /// The header bytes a raw stream of fixed-width words saves over any other physical encoding.
419    /// v2 leaves such a stream's `byte_length` varint out, since its count and width give it.
420    #[must_use]
421    pub(crate) fn raw_header_saving(self) -> usize {
422        match self {
423            Self::V01 => 0,
424            Self::V02 => 1,
425        }
426    }
427}
428
429/// Global encoder settings controlling which optimization strategies are attempted.
430#[derive(Debug, Clone, Copy, PartialEq, Hash)]
431#[expect(
432    clippy::struct_excessive_bools,
433    reason = "enums would not model this better, not a state machine"
434)]
435pub struct EncoderConfig {
436    /// The wire format to encode layers to.
437    #[cfg(feature = "unstable-v2")]
438    wire_version: WireVersion,
439    /// Generate tessellation data for polygons and multi-polygons.
440    tessellate: bool,
441    /// Try sorting features by the Z-order (Morton) curve index of their first vertex.
442    attempt_spatial_morton_sort: bool,
443    /// Try sorting features by the Hilbert curve index of their first vertex.
444    attempt_spatial_hilbert_sort: bool,
445    /// Try sorting features by their feature ID in ascending order.
446    attempt_id_sort: bool,
447    /// Allow `FSST` string compression
448    allow_fsst: bool,
449    /// Allow `FastPFOR` integer compression
450    allow_fastpfor: bool,
451    /// Allow string grouping into shared dictionaries
452    allow_shared_dict: bool,
453    /// Allow the v2-only float dictionary encoding
454    #[cfg(feature = "unstable-v2")]
455    allow_float_dict: bool,
456    /// Allow the v2-only Framed, Exception-Free ALP float encoding
457    #[cfg(feature = "unstable-v2")]
458    allow_float_alp: bool,
459    /// Allow the v2-only bit-packed physical encoding for dictionary code streams
460    #[cfg(feature = "unstable-v2")]
461    allow_packed_dict_codes: bool,
462    /// Code a nested struct's or map's row shapes instead of per-field presence or per-entry keys.
463    #[cfg(feature = "unstable-v2")]
464    allow_row_shapes: bool,
465    /// Let a tessellated all-polygon layer store its triangles without the outlines.
466    #[cfg(feature = "unstable-v2")]
467    allow_triangles_only: bool,
468    /// Allow the v2-only second-order delta encodings of integer and vertex streams
469    #[cfg(feature = "unstable-v2")]
470    allow_delta2: bool,
471}
472impl Default for EncoderConfig {
473    fn default() -> Self {
474        Self {
475            #[cfg(feature = "unstable-v2")]
476            wire_version: WireVersion::V01,
477            tessellate: false,
478            attempt_spatial_morton_sort: true,
479            attempt_spatial_hilbert_sort: true,
480            attempt_id_sort: true,
481            allow_fsst: true,
482            allow_fastpfor: true,
483            allow_shared_dict: true,
484            // Off by default while the encoding is still being measured.
485            #[cfg(feature = "unstable-v2")]
486            allow_float_dict: false,
487            #[cfg(feature = "unstable-v2")]
488            allow_float_alp: false,
489            #[cfg(feature = "unstable-v2")]
490            allow_row_shapes: false,
491            #[cfg(feature = "unstable-v2")]
492            allow_packed_dict_codes: false,
493            #[cfg(feature = "unstable-v2")]
494            allow_triangles_only: false,
495            #[cfg(feature = "unstable-v2")]
496            allow_delta2: false,
497        }
498    }
499}
500
501impl EncoderConfig {
502    #[cfg(feature = "unstable-v2")]
503    #[must_use]
504    pub fn wire_version(self) -> WireVersion {
505        self.wire_version
506    }
507
508    #[must_use]
509    pub fn tessellate(self) -> bool {
510        self.tessellate
511    }
512
513    #[must_use]
514    pub fn attempt_spatial_morton_sort(self) -> bool {
515        self.attempt_spatial_morton_sort
516    }
517
518    #[must_use]
519    pub fn attempt_spatial_hilbert_sort(self) -> bool {
520        self.attempt_spatial_hilbert_sort
521    }
522
523    #[must_use]
524    pub fn attempt_id_sort(self) -> bool {
525        self.attempt_id_sort
526    }
527
528    #[must_use]
529    pub fn allow_fsst(self) -> bool {
530        self.allow_fsst
531    }
532
533    #[must_use]
534    pub fn allow_fastpfor(self) -> bool {
535        self.allow_fastpfor
536    }
537
538    /// The `FastPFor` encoding to race, or `None` when `FastPFor` is switched off.
539    #[must_use]
540    pub(crate) fn fastpfor(self) -> Option<PhysicalEncoding> {
541        #[cfg(feature = "unstable-v2")]
542        let kind = self.wire_version.fastpfor_kind();
543        #[cfg(not(feature = "unstable-v2"))]
544        let kind = FastPForKind::Block256Be;
545        self.allow_fastpfor
546            .then_some(PhysicalEncoding::FastPFor(kind))
547    }
548
549    #[must_use]
550    pub fn allow_shared_dict(self) -> bool {
551        self.allow_shared_dict
552    }
553
554    /// Whether a nested struct or map may code its row shapes, which only v2 can express.
555    #[cfg(feature = "unstable-v2")]
556    #[must_use]
557    pub fn allow_row_shapes(self) -> bool {
558        self.allow_row_shapes
559    }
560
561    /// Let a nested struct or map replace per-field presence and per-entry keys with row shapes.
562    #[cfg(feature = "unstable-v2")]
563    #[must_use]
564    pub fn with_row_shapes(mut self, enabled: bool) -> Self {
565        self.allow_row_shapes = enabled;
566        self
567    }
568
569    /// Whether 8-bit integer columns stay 8-bit, which only v2 can express.
570    #[must_use]
571    pub fn allow_8bit_ints(self) -> bool {
572        #[cfg(feature = "unstable-v2")]
573        return self.wire_version != WireVersion::V01;
574        #[cfg(not(feature = "unstable-v2"))]
575        return false;
576    }
577
578    /// Whether float columns may use a dictionary, which only v2 can express.
579    #[cfg(feature = "unstable-v2")]
580    #[must_use]
581    pub fn allow_float_dict(self) -> bool {
582        self.allow_float_dict && self.wire_version != WireVersion::V01
583    }
584
585    /// Whether float columns may use Framed, Exception-Free ALP, which only v2 can express.
586    #[cfg(feature = "unstable-v2")]
587    #[must_use]
588    pub fn allow_float_alp(self) -> bool {
589        self.allow_float_alp && self.wire_version != WireVersion::V01
590    }
591
592    #[cfg(feature = "unstable-v2")]
593    #[must_use]
594    pub fn with_wire_version(mut self, version: WireVersion) -> Self {
595        self.wire_version = version;
596        self
597    }
598
599    #[must_use]
600    pub fn with_tessellation(mut self, enabled: bool) -> Self {
601        self.tessellate = enabled;
602        self
603    }
604
605    /// Whether a tessellated layer may drop its outlines, which only v2 can express.
606    #[cfg(feature = "unstable-v2")]
607    #[must_use]
608    pub fn allow_triangles_only(self) -> bool {
609        self.allow_triangles_only && self.wire_version != WireVersion::V01
610    }
611
612    /// Let a tessellated layer of only polygons, with no m-values, store its triangles without the outlines.
613    /// Each of its polygons then decodes as the triangles it was cut into.
614    #[cfg(feature = "unstable-v2")]
615    #[must_use]
616    pub fn with_triangles_only(mut self, enabled: bool) -> Self {
617        self.allow_triangles_only = enabled;
618        self
619    }
620
621    /// Whether integer and vertex streams may store the deltas of their deltas, which only v2 can express.
622    #[cfg(feature = "unstable-v2")]
623    #[must_use]
624    pub fn allow_delta2(self) -> bool {
625        self.allow_delta2 && self.wire_version != WireVersion::V01
626    }
627
628    /// Allow integer and vertex streams to store the deltas of their deltas.
629    #[cfg(feature = "unstable-v2")]
630    #[must_use]
631    pub fn with_delta2(mut self, enabled: bool) -> Self {
632        self.allow_delta2 = enabled;
633        self
634    }
635
636    #[must_use]
637    pub fn with_spatial_morton_sort(mut self, enabled: bool) -> Self {
638        self.attempt_spatial_morton_sort = enabled;
639        self
640    }
641
642    #[must_use]
643    pub fn with_spatial_hilbert_sort(mut self, enabled: bool) -> Self {
644        self.attempt_spatial_hilbert_sort = enabled;
645        self
646    }
647
648    #[must_use]
649    pub fn with_id_sort(mut self, enabled: bool) -> Self {
650        self.attempt_id_sort = enabled;
651        self
652    }
653
654    #[must_use]
655    pub fn with_fsst(mut self, enabled: bool) -> Self {
656        self.allow_fsst = enabled;
657        self
658    }
659
660    #[must_use]
661    pub fn with_fastpfor(mut self, enabled: bool) -> Self {
662        self.allow_fastpfor = enabled;
663        self
664    }
665
666    #[must_use]
667    pub fn with_shared_dict(mut self, enabled: bool) -> Self {
668        self.allow_shared_dict = enabled;
669        self
670    }
671
672    /// Allow float columns to store one code per value into a dictionary of the distinct ones.
673    /// Off by default, and only v2 can express it.
674    /// A column takes it only when it comes out strictly smaller in stored bytes.
675    #[cfg(feature = "unstable-v2")]
676    #[must_use]
677    pub fn with_float_dict(mut self, enabled: bool) -> Self {
678        self.allow_float_dict = enabled;
679        self
680    }
681
682    /// Allow float columns to store each value as a decimal-scaled integer.
683    #[cfg(feature = "unstable-v2")]
684    #[must_use]
685    pub fn with_float_alp(mut self, enabled: bool) -> Self {
686        self.allow_float_alp = enabled;
687        self
688    }
689
690    /// Whether dictionary code streams may be stored bit-packed.
691    #[cfg(feature = "unstable-v2")]
692    #[must_use]
693    pub fn allow_packed_dict_codes(self) -> bool {
694        self.allow_packed_dict_codes && self.wire_version != WireVersion::V01
695    }
696
697    /// Allow dictionary code streams to store every code in the same
698    /// `ceil(log2(dict_len))` bits instead of a varint each.
699    #[cfg(feature = "unstable-v2")]
700    #[must_use]
701    pub fn with_packed_dict_codes(mut self, enabled: bool) -> Self {
702        self.allow_packed_dict_codes = enabled;
703        self
704    }
705}
706
707/// How to encode a string column.
708#[derive(Debug, Clone, Copy, PartialEq, Eq)]
709pub enum StrEncoding {
710    Plain,
711    Dict,
712    Fsst,
713    FsstDict,
714    FrontDict,
715    FsstFrontDict,
716}
717
718/// How to encode a float column, pinned rather than costed against the alternatives.
719#[derive(Debug, Clone, Copy, PartialEq, Eq)]
720pub enum FloatEncoding {
721    None,
722    #[cfg(feature = "unstable-v2")]
723    Dict,
724    #[cfg(feature = "unstable-v2")]
725    Alp,
726}
727
728#[derive(Debug, Clone, Copy, PartialEq)]
729pub enum ColumnKind {
730    Id,
731    Geometry,
732    Property,
733}
734
735/// Where one string stream set sits: the column that holds it, and the path
736/// through a nested column's tree that leads to it.
737///
738/// A flat column's path is empty, so its streams are named exactly as before.
739#[cfg(feature = "unstable-v2")]
740#[derive(Clone, Copy, Debug)]
741pub struct StrAt<'a> {
742    pub name: &'a str,
743    pub subname: &'a str,
744}
745
746#[cfg(feature = "unstable-v2")]
747impl<'a> StrAt<'a> {
748    #[must_use]
749    pub const fn flat(name: &'a str) -> Self {
750        Self { name, subname: "" }
751    }
752
753    #[must_use]
754    pub const fn nested(name: &'a str, subname: &'a str) -> Self {
755        Self { name, subname }
756    }
757
758    pub(crate) fn ctx(self, stream_type: StreamType) -> StreamCtx<'a> {
759        StreamCtx::prop2(stream_type, self.name, self.subname)
760    }
761
762    /// The name an [`ExplicitEncoder`] pins an encoding by, the column's plus the path.
763    pub(crate) fn qualified(self) -> Cow<'a, str> {
764        StreamCtx::prop2(StreamType::Present, self.name, self.subname).qualified()
765    }
766}
767
768/// Context for per-stream encoding decisions in [`ExplicitEncoder`] callbacks.
769#[derive(Clone, Copy, Debug, PartialEq)]
770pub struct StreamCtx<'a> {
771    pub kind: ColumnKind,
772    pub stream_type: StreamType,
773    pub name: &'a str,
774    pub subname: &'a str,
775}
776impl<'a> StreamCtx<'a> {
777    /// Stream with a logical sub-part (e.g. string column `"lengths"` / `"offsets"`, shared-dict child suffix).
778    #[inline]
779    #[must_use]
780    pub const fn new(
781        kind: ColumnKind,
782        stream_type: StreamType,
783        name: &'a str,
784        subname: &'a str,
785    ) -> Self {
786        Self {
787            kind,
788            stream_type,
789            name,
790            subname,
791        }
792    }
793
794    #[inline]
795    #[must_use]
796    pub const fn id(stream_type: StreamType) -> Self {
797        Self::new(ColumnKind::Id, stream_type, "", "")
798    }
799
800    #[inline]
801    #[must_use]
802    pub const fn geom(stream_type: StreamType, name: &'a str) -> Self {
803        Self::new(ColumnKind::Geometry, stream_type, name, "")
804    }
805
806    #[inline]
807    #[must_use]
808    pub const fn prop(stream_type: StreamType, name: &'a str) -> Self {
809        Self::new(ColumnKind::Property, stream_type, name, "")
810    }
811
812    #[inline]
813    #[must_use]
814    pub const fn prop_data(name: &'a str) -> Self {
815        let stream_type = StreamType::Data(DictionaryType::None);
816        Self::new(ColumnKind::Property, stream_type, name, "")
817    }
818
819    #[inline]
820    #[must_use]
821    pub const fn prop2(stream_type: StreamType, prefix: &'a str, suffix: &'a str) -> Self {
822        Self::new(ColumnKind::Property, stream_type, prefix, suffix)
823    }
824
825    /// The name an [`ExplicitEncoder`] pins an encoding by, the column's plus any path within it.
826    #[must_use]
827    pub fn qualified(&self) -> Cow<'a, str> {
828        if self.subname.is_empty() {
829            Cow::Borrowed(self.name)
830        } else {
831            Cow::Owned(format!("{}{}", self.name, self.subname))
832        }
833    }
834}
835
836/// Explicit, deterministic encoding configuration for synthetics and tests.
837///
838/// All encoding choices are caller-specified via callbacks so one struct can cover any combination without per-stream boilerplate.
839#[derive(Dbg)]
840pub struct ExplicitEncoder {
841    /// Vertex buffer layout for geometry streams.
842    pub vertex_buffer_type: VertexBufferType,
843    /// Per-stream override for the skip-empty-stream rule used by `write_geo_u32_stream`.
844    #[dbg(skip)]
845    pub force_stream: Box<dyn for<'a> Fn(&'a StreamCtx<'a>) -> bool>,
846    /// Return the [`IntEncoder`] for a stream identified by [`StreamCtx`].
847    #[dbg(skip)]
848    pub get_int_encoder: Box<dyn for<'a> Fn(&'a StreamCtx<'a>) -> IntEncoder>,
849    /// Return the string encoding strategy for a string property column.
850    #[dbg(skip)]
851    pub get_str_encoding: Box<dyn Fn(&str) -> StrEncoding>,
852    /// Return the logical encoding for a float property column.
853    #[dbg(skip)]
854    pub get_float_encoding: Box<dyn Fn(&str) -> FloatEncoding>,
855}
856
857#[cfg(all(test, feature = "unstable-v2"))]
858mod tests {
859    use insta::assert_snapshot;
860
861    use super::*;
862    use crate::encoder::{StagedLeaf, StagedList, StagedStruct, StagedValues};
863
864    fn leaf() -> StagedNode {
865        StagedNode::Leaf(StagedLeaf::new(None, StagedValues::I32(vec![1])))
866    }
867
868    /// A tree of `depth` levels: nested lists all the way down to one leaf.
869    fn lists(depth: usize) -> StagedInterior {
870        let mut node = leaf();
871        for _ in 2..depth {
872            node = StagedNode::Interior(StagedInterior::List(StagedList::new(None, vec![1], node)));
873        }
874        StagedInterior::List(StagedList::new(None, vec![1], node))
875    }
876
877    #[test]
878    fn a_struct_repeating_a_field_name() {
879        let root =
880            StagedInterior::Struct(StagedStruct::new(None, vec![("a", leaf()), ("a", leaf())]));
881        assert_snapshot!(
882            validate_nested_interior(&root, 1).unwrap_err(),
883            @"duplicate field name in a struct node: a"
884        );
885    }
886
887    #[test]
888    fn a_tree_one_level_past_the_limit() {
889        assert_snapshot!(
890            validate_nested_interior(&lists(MAX_NESTED_DEPTH + 1), 1).unwrap_err(),
891            @"a nested column is 9 levels deep, more than the 8 the format allows"
892        );
893    }
894
895    #[test]
896    fn a_tree_at_the_limit() {
897        validate_nested_interior(&lists(MAX_NESTED_DEPTH), 1).expect("a tree at the limit");
898    }
899}
mlt_core/encoder/model.rs

mlt_core/encoder/
model.rs
