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

vole_document/encode/
mod.rs
