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::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    // Stream the portfolio through the court one candidate at a time: only the
74    // current best (plus the candidate being priced) stays resident, so encoder
75    // memory is bounded even when every candidate holds a source-sized payload
76    // (Phase 14: large PDFs). Same candidate set and order as `propose_all`.
77    let mut court = court::Court::new();
78    let mut matched: u32 = 0;
79    candidates::propose_each(input, limits, |c| {
80        if force.is_none_or(|k| c.kind == k) {
81            matched += 1;
82            court.offer(input, c, limits)?;
83        }
84        Ok(())
85    })?;
86    if let Some(kind) = force
87        && matched == 0
88    {
89        return Err(Error::usage(format!(
90            "candidate {kind:?} is not proposed for this input"
91        )));
92    }
93    let result = court.finish()?;
94    let report = EncodeReport {
95        kind: result.kind,
96        source_len: input.len() as u64,
97        encoded_len: result.bytes.len() as u64,
98        cost: result.cost,
99        sha256_hex: to_hex(&sha256(input)),
100        candidates_evaluated: result.candidates_evaluated,
101        graph_ops: result.graph_ops,
102    };
103    Ok((result.bytes, report))
104}
105
106#[cfg(test)]
107mod tests {
108    use super::*;
109
110    #[test]
111    fn encode_decode_identity() {
112        let input: Vec<u8> = (0..=255u8).cycle().take(10_000).collect();
113        let (bytes, report) = encode(&input, Limits::DEFAULT).unwrap();
114        assert_eq!(report.source_len, input.len() as u64);
115        assert_eq!(report.encoded_len, bytes.len() as u64);
116        let (out, _) = crate::materialize::decode_to_bytes(&bytes, Limits::DEFAULT).unwrap();
117        assert_eq!(out, input);
118    }
119
120    #[test]
121    fn deterministic_output() {
122        let input = b"determinism check".to_vec();
123        let (a, _) = encode(&input, Limits::DEFAULT).unwrap();
124        let (b, _) = encode(&input, Limits::DEFAULT).unwrap();
125        assert_eq!(a, b);
126    }
127
128    #[test]
129    fn encode_with_none_matches_encode() {
130        let input = b"the court must not care how it was invoked".repeat(40);
131        let (auto, auto_report) = encode(&input, Limits::DEFAULT).unwrap();
132        let (explicit, explicit_report) = encode_with(&input, Limits::DEFAULT, None).unwrap();
133        assert_eq!(auto, explicit, "None must equal the unforced court");
134        assert_eq!(auto_report.kind, explicit_report.kind);
135        assert_eq!(auto_report.encoded_len, explicit_report.encoded_len);
136        assert_eq!(
137            auto_report.candidates_evaluated,
138            explicit_report.candidates_evaluated
139        );
140    }
141
142    #[test]
143    fn encode_with_force_raw_is_exact() {
144        // RAW normally loses to RLE on this input, so forcing RAW proves the
145        // filter really overrides the court's own choice.
146        let input = vec![0u8; 8192];
147        let (_, auto_report) = encode(&input, Limits::DEFAULT).unwrap();
148        assert_ne!(auto_report.kind, CandidateKind::Raw);
149
150        let (bytes, report) =
151            encode_with(&input, Limits::DEFAULT, Some(CandidateKind::Raw)).unwrap();
152        assert_eq!(report.kind, CandidateKind::Raw);
153        assert_eq!(report.candidates_evaluated, 1);
154        let (out, _) = crate::materialize::decode_to_bytes(&bytes, Limits::DEFAULT).unwrap();
155        assert_eq!(out, input);
156    }
157
158    #[cfg(feature = "rans")]
159    #[test]
160    fn encode_with_force_byte_rans_is_exact() {
161        let input = b"The quick brown fox jumps over the lazy dog. ".repeat(800);
162        let (bytes, report) =
163            encode_with(&input, Limits::DEFAULT, Some(CandidateKind::ByteRans)).unwrap();
164        assert_eq!(report.kind, CandidateKind::ByteRans);
165        assert_eq!(report.candidates_evaluated, 1);
166        let (out, _) = crate::materialize::decode_to_bytes(&bytes, Limits::DEFAULT).unwrap();
167        assert_eq!(out, input);
168    }
169
170    #[test]
171    fn encode_with_unproposed_kind_errors() {
172        // A non-PDF cannot propose PDF_CHANNELS, so forcing it must fail with a
173        // typed Usage error rather than silently falling back to another lane.
174        let input = b"definitely not a pdf, just text".to_vec();
175        let err = encode_with(&input, Limits::DEFAULT, Some(CandidateKind::PdfChannels))
176            .expect_err("PDF_CHANNELS is not proposed for a non-PDF");
177        assert_eq!(err.class(), crate::error::ErrorClass::Usage);
178        assert!(
179            err.message().contains("is not proposed"),
180            "unexpected message: {}",
181            err.message()
182        );
183    }
184}
vole_document/encode/mod.rs

vole_document/encode/
mod.rs
