vole_document/encode/
mod.rs

1//! Encoding: candidate portfolio, complete-cost court, and decode-before-commit.
2
3pub mod candidates;
4pub mod court;
5
6/// Encoder-only search governance, compiled only under the non-default,
7/// dependency-free `dsfb-search` feature. Encoder-only: the decode path never
8/// imports it (see `docs/adr/0022-encoder-only-search-governance.md`).
9#[cfg(feature = "dsfb-search")]
10pub mod governor;
11
12use crate::accounting::CostBreakdown;
13use crate::encode::candidates::{Candidate, CandidateKind};
14use crate::error::{Error, Result};
15use crate::integrity::{sha256, to_hex};
16use crate::limits::Limits;
17
18/// A report describing encode-time decisions and physical attribution.
19#[derive(Debug, Clone)]
20pub struct EncodeReport {
21    /// Winning candidate family.
22    pub kind: CandidateKind,
23    /// Source length in bytes.
24    pub source_len: u64,
25    /// Serialized `.voldoc` length in bytes.
26    pub encoded_len: u64,
27    /// Physical byte attribution of the serialized descriptor.
28    pub cost: CostBreakdown,
29    /// Lower-case hex SHA-256 of the source.
30    pub sha256_hex: String,
31    /// Number of candidates evaluated.
32    pub candidates_evaluated: u32,
33    /// Reconstruction work (DRA instruction count) of the winner.
34    pub graph_ops: usize,
35}
36
37impl EncodeReport {
38    /// Source bytes divided by encoded bytes (0.0 when the source is empty).
39    pub fn compression_ratio(&self) -> f64 {
40        if self.encoded_len == 0 {
41            return 0.0;
42        }
43        self.source_len as f64 / self.encoded_len as f64
44    }
45}
46
47/// Encode `input` exactly, returning the serialized descriptor and a report.
48///
49/// Every returned byte sequence has already been round-tripped through the
50/// normative decoder and byte-compared against `input`.
51pub fn encode(input: &[u8], limits: Limits) -> Result<(Vec<u8>, EncodeReport)> {
52    encode_with(input, limits, None)
53}
54
55/// Encode `input` exactly, optionally forcing a single candidate family.
56///
57/// With `force == None` this is the ordinary complete-cost court over every
58/// proposed candidate. With `force == Some(kind)` the candidate set is first
59/// filtered to proposals of exactly that kind, then the *same* court runs over
60/// the filtered set: the forced lane is still serialized, decoded, and
61/// byte-compared before it may be returned, so forcing never weakens exactness
62/// and determinism.
63///
64/// Returns [`Error::usage`] when `force` names a kind that this input does not
65/// propose (for example PDF_CHANNELS on a non-PDF, or BYTE_RANS on empty input),
66/// rather than silently substituting another lane. That failure is the honest
67/// signal that the mechanism does not apply.
68pub fn encode_with(
69    input: &[u8],
70    limits: Limits,
71    force: Option<CandidateKind>,
72) -> Result<(Vec<u8>, EncodeReport)> {
73    encode_impl(input, limits, force, false)
74}
75
76/// Encode one forced candidate for the direct-build path, attaching the advisory
77/// observation index to the descriptor **before** the court serializes it
78/// (Phase 18.3).
79///
80/// The returned bytes are byte-identical to
81/// `with_observation_index(&encode_with(input, limits, Some(kind))?.0)` whenever
82/// the plain encode carries no index: the index is a pure function of the
83/// (unchanged) program, and serialization is deterministic, so serializing the
84/// enriched descriptor once emits exactly the bytes the enrich-after-serialize
85/// path would. The source-sized payload is therefore serialized once instead of
86/// twice, and the extra parse + reserialize of the whole descriptor is removed
87/// from the direct-build wall. Exactness is unweakened: the same court
88/// serializes, parses, materializes, and byte-compares the enriched authority.
89///
90/// Enrichment is advisory: if the enriched lane is refused for any reason, this
91/// falls back to exactly the plain forced lane, so attaching the index can never
92/// turn a build that would otherwise succeed into a failure.
93#[cfg(feature = "field")]
94pub(crate) fn encode_with_observation_index(
95    input: &[u8],
96    limits: Limits,
97    kind: CandidateKind,
98) -> Result<(Vec<u8>, EncodeReport)> {
99    encode_impl(input, limits, Some(kind), true)
100}
101
102fn encode_impl(
103    input: &[u8],
104    limits: Limits,
105    force: Option<CandidateKind>,
106    enrich_index: bool,
107) -> Result<(Vec<u8>, EncodeReport)> {
108    let mut court = court::Court::new();
109    if let Some(kind) = force {
110        // Forced lane: build only that family (no portfolio), so e.g. `--force raw`
111        // for a runtime ingest costs one candidate, not the whole search.
112        let candidate = candidates::propose_forced(input, limits, kind)?.ok_or_else(|| {
113            Error::usage(format!("candidate {kind:?} is not proposed for this input"))
114        })?;
115        if enrich_index {
116            let enriched = Candidate {
117                kind,
118                descriptor: candidate.descriptor.with_observation_index(limits),
119            };
120            if court.offer(input, enriched, limits).is_err() {
121                // The index is ignorable; never let it break an otherwise exact
122                // build. Re-propose the plain lane and price that instead.
123                let plain = candidates::propose_forced(input, limits, kind)?.ok_or_else(|| {
124                    Error::usage(format!("candidate {kind:?} is not proposed for this input"))
125                })?;
126                court = court::Court::new();
127                court.offer(input, plain, limits)?;
128            }
129        } else {
130            court.offer(input, candidate, limits)?;
131        }
132    } else {
133        // Stream the portfolio through the court one candidate at a time: only the
134        // current best (plus the candidate being priced) stays resident, bounding
135        // encoder memory even when every candidate holds a source-sized payload.
136        candidates::propose_each(input, limits, |c| court.offer(input, c, limits))?;
137    }
138    let result = court.finish()?;
139    let report = EncodeReport {
140        kind: result.kind,
141        source_len: input.len() as u64,
142        encoded_len: result.bytes.len() as u64,
143        cost: result.cost,
144        sha256_hex: to_hex(&sha256(input)),
145        candidates_evaluated: result.candidates_evaluated,
146        graph_ops: result.graph_ops,
147    };
148    Ok((result.bytes, report))
149}
150
151#[cfg(test)]
152mod tests {
153    use super::*;
154
155    #[test]
156    fn encode_decode_identity() {
157        let input: Vec<u8> = (0..=255u8).cycle().take(10_000).collect();
158        let (bytes, report) = encode(&input, Limits::DEFAULT).unwrap();
159        assert_eq!(report.source_len, input.len() as u64);
160        assert_eq!(report.encoded_len, bytes.len() as u64);
161        let (out, _) = crate::materialize::decode_to_bytes(&bytes, Limits::DEFAULT).unwrap();
162        assert_eq!(out, input);
163    }
164
165    #[test]
166    fn deterministic_output() {
167        let input = b"determinism check".to_vec();
168        let (a, _) = encode(&input, Limits::DEFAULT).unwrap();
169        let (b, _) = encode(&input, Limits::DEFAULT).unwrap();
170        assert_eq!(a, b);
171    }
172
173    #[test]
174    fn encode_with_none_matches_encode() {
175        let input = b"the court must not care how it was invoked".repeat(40);
176        let (auto, auto_report) = encode(&input, Limits::DEFAULT).unwrap();
177        let (explicit, explicit_report) = encode_with(&input, Limits::DEFAULT, None).unwrap();
178        assert_eq!(auto, explicit, "None must equal the unforced court");
179        assert_eq!(auto_report.kind, explicit_report.kind);
180        assert_eq!(auto_report.encoded_len, explicit_report.encoded_len);
181        assert_eq!(
182            auto_report.candidates_evaluated,
183            explicit_report.candidates_evaluated
184        );
185    }
186
187    #[cfg(feature = "field")]
188    #[test]
189    fn forced_encode_omits_index_and_enrichment_is_byte_identical() {
190        let input = b"the enriched authority must be byte-identical".repeat(64);
191
192        // The public forced lane is unchanged: it emits no observation index, so
193        // `encode` bytes stay byte-identical where the index is absent.
194        let (plain, _) = encode_with(&input, Limits::DEFAULT, Some(CandidateKind::Raw)).unwrap();
195        let parsed = crate::container::Descriptor::parse(&plain, Limits::DEFAULT).unwrap();
196        assert!(parsed.descriptor.observation_index.is_none());
197
198        // Enrich-before-serialize must equal enrich-after-serialize byte for byte.
199        let after = crate::field::ingest::with_observation_index(&plain, Limits::DEFAULT).unwrap();
200        let (before, report) =
201            encode_with_observation_index(&input, Limits::DEFAULT, CandidateKind::Raw).unwrap();
202        assert_eq!(
203            before, after,
204            "the one-pass enriched authority must equal the two-pass bytes"
205        );
206        assert_eq!(report.candidates_evaluated, 1);
207        assert_eq!(report.kind, CandidateKind::Raw);
208
209        // The enriched authority decodes exactly and still verifies.
210        let enriched = crate::container::Descriptor::parse(&before, Limits::DEFAULT).unwrap();
211        assert!(enriched.descriptor.observation_index.is_some());
212        let (out, _) = crate::materialize::decode_to_bytes(&before, Limits::DEFAULT).unwrap();
213        assert_eq!(out, input);
214    }
215
216    #[test]
217    fn encode_with_force_raw_is_exact() {
218        // RAW normally loses to RLE on this input, so forcing RAW proves the
219        // filter really overrides the court's own choice.
220        let input = vec![0u8; 8192];
221        let (_, auto_report) = encode(&input, Limits::DEFAULT).unwrap();
222        assert_ne!(auto_report.kind, CandidateKind::Raw);
223
224        let (bytes, report) =
225            encode_with(&input, Limits::DEFAULT, Some(CandidateKind::Raw)).unwrap();
226        assert_eq!(report.kind, CandidateKind::Raw);
227        assert_eq!(report.candidates_evaluated, 1);
228        let (out, _) = crate::materialize::decode_to_bytes(&bytes, Limits::DEFAULT).unwrap();
229        assert_eq!(out, input);
230    }
231
232    #[cfg(feature = "rans")]
233    #[test]
234    fn encode_with_force_byte_rans_is_exact() {
235        let input = b"The quick brown fox jumps over the lazy dog. ".repeat(800);
236        let (bytes, report) =
237            encode_with(&input, Limits::DEFAULT, Some(CandidateKind::ByteRans)).unwrap();
238        assert_eq!(report.kind, CandidateKind::ByteRans);
239        assert_eq!(report.candidates_evaluated, 1);
240        let (out, _) = crate::materialize::decode_to_bytes(&bytes, Limits::DEFAULT).unwrap();
241        assert_eq!(out, input);
242    }
243
244    #[test]
245    fn encode_with_unproposed_kind_errors() {
246        // A non-PDF cannot propose PDF_CHANNELS, so forcing it must fail with a
247        // typed Usage error rather than silently falling back to another lane.
248        let input = b"definitely not a pdf, just text".to_vec();
249        let err = encode_with(&input, Limits::DEFAULT, Some(CandidateKind::PdfChannels))
250            .expect_err("PDF_CHANNELS is not proposed for a non-PDF");
251        assert_eq!(err.class(), crate::error::ErrorClass::Usage);
252        assert!(
253            err.message().contains("is not proposed"),
254            "unexpected message: {}",
255            err.message()
256        );
257    }
258}
vole_document/encode/mod.rs

vole_document/encode/
mod.rs
