structured_zstd/encoding/parameters.rs
1//! Fine-grained compression parameters — the drop-in equivalent of C
2//! zstd's advanced `ZSTD_CCtx_setParameter` surface.
3//!
4//! [`CompressionLevel`] selects a
5//! whole tuning preset in one knob. This module exposes the individual
6//! knobs underneath it — window/hash/chain/search logs, the match
7//! strategy, and the long-distance-matching (LDM) block — so callers
8//! can override a level's defaults for domain-specific tuning.
9//!
10//! # Builder
11//!
12//! [`CompressionParameters`] is built through
13//! [`CompressionParameters::builder`], which takes an explicit base
14//! [`CompressionLevel`] (there is no
15//! implicit default). Every knob left unset inherits that base level's
16//! resolved value, so a builder that overrides nothing reproduces plain
17//! level-based compression byte-for-byte.
18//!
19//! ```rust
20//! use structured_zstd::encoding::{CompressionLevel, CompressionParameters, Strategy};
21//!
22//! let params = CompressionParameters::builder(CompressionLevel::Level(19))
23//! .window_log(22)
24//! .strategy(Strategy::Btultra2)
25//! .enable_long_distance_matching(true)
26//! .build()
27//! .expect("parameters within bounds");
28//! ```
29//!
30//! # Bounds
31//!
32//! Every knob has an inclusive `[lower, upper]` range, queryable via
33//! [`CParameter::bounds`] (the analogue of `ZSTD_cParam_getBounds`).
34//! [`CompressionParametersBuilder::build`] validates each set knob and
35//! returns [`ParameterError::OutOfBounds`] for the first violation.
36//!
37//! # Long-distance matching (LDM)
38//!
39//! LDM is **off at every [`CompressionLevel`] preset**, matching upstream `libzstd.so.1` where `ZSTD_compress(..., level)`
40//! never enables LDM — even at level 22. It is activated either by
41//! [`CompressionParametersBuilder::enable_long_distance_matching`] or by any of
42//! the `ldm_*` setters, which each imply `enable_long_distance_matching(true)`.
43//! It works under every [`Strategy`]: the optimal parsers weigh its matches as
44//! candidates, the others take them outright and search only the bytes between
45//! them, as upstream does. A dictionary loaded under it is indexed for it too.
46
47use crate::encoding::CompressionLevel;
48
49/// Match-finder strategy — the drop-in equivalent of C zstd's
50/// `ZSTD_strategy` enum (`ZSTD_fast` … `ZSTD_btultra2`). The numeric
51/// ordinals match upstream (`fast = 1` … `btultra2 = 9`), so
52/// [`Strategy::ordinal`] / [`Strategy::from_ordinal`] round-trip with
53/// the C `ZSTD_c_strategy` parameter value.
54#[derive(Copy, Clone, Debug, PartialEq, Eq, PartialOrd, Ord)]
55pub enum Strategy {
56 /// `ZSTD_fast` (1) — single-table fast finder.
57 Fast,
58 /// `ZSTD_dfast` (2) — two parallel hash tables.
59 Dfast,
60 /// `ZSTD_greedy` (3) — commit the first acceptable match, no lookahead.
61 Greedy,
62 /// `ZSTD_lazy` (4) — one-position lazy lookahead.
63 Lazy,
64 /// `ZSTD_lazy2` (5) — two-position lazy lookahead.
65 Lazy2,
66 /// `ZSTD_btlazy2` (6) — binary-tree-assisted lazy2.
67 Btlazy2,
68 /// `ZSTD_btopt` (7) — optimal parser, no ultra refinements.
69 Btopt,
70 /// `ZSTD_btultra` (8) — optimal parser with refined price tables.
71 Btultra,
72 /// `ZSTD_btultra2` (9) — optimal parser with two-pass dynamic stats.
73 Btultra2,
74}
75
76impl Strategy {
77 /// Upstream `ZSTD_strategy` ordinal (`fast = 1` … `btultra2 = 9`).
78 pub const fn ordinal(self) -> u32 {
79 match self {
80 Self::Fast => 1,
81 Self::Dfast => 2,
82 Self::Greedy => 3,
83 Self::Lazy => 4,
84 Self::Lazy2 => 5,
85 Self::Btlazy2 => 6,
86 Self::Btopt => 7,
87 Self::Btultra => 8,
88 Self::Btultra2 => 9,
89 }
90 }
91
92 /// Construct from an upstream `ZSTD_strategy` ordinal. Returns
93 /// `None` outside `1..=9`.
94 pub const fn from_ordinal(ordinal: u32) -> Option<Self> {
95 Some(match ordinal {
96 1 => Self::Fast,
97 2 => Self::Dfast,
98 3 => Self::Greedy,
99 4 => Self::Lazy,
100 5 => Self::Lazy2,
101 6 => Self::Btlazy2,
102 7 => Self::Btopt,
103 8 => Self::Btultra,
104 9 => Self::Btultra2,
105 _ => return None,
106 })
107 }
108
109 /// Internal runtime strategy tag.
110 pub(crate) const fn tag(self) -> crate::encoding::strategy::StrategyTag {
111 use crate::encoding::strategy::StrategyTag;
112 match self {
113 Self::Fast => StrategyTag::Fast,
114 Self::Dfast => StrategyTag::Dfast,
115 Self::Greedy => StrategyTag::Greedy,
116 // Lazy / Lazy2 ride the runtime `Lazy` tag (the lazy lookahead
117 // depth carries the variance, see `lazy_depth`). `Btlazy2`
118 // keeps its own tag: `Lazy` resolves to the Row finder, while
119 // btlazy2 is a binary-tree search and must stay on the
120 // HashChain/BT storage.
121 Self::Lazy | Self::Lazy2 => StrategyTag::Lazy,
122 Self::Btlazy2 => StrategyTag::Btlazy2,
123 Self::Btopt => StrategyTag::BtOpt,
124 Self::Btultra => StrategyTag::BtUltra,
125 Self::Btultra2 => StrategyTag::BtUltra2,
126 }
127 }
128
129 /// Lazy lookahead depth for the greedy/lazy band (0/1/2). `Optimal`
130 /// strategies report 2 (the depth their hash-chain seed walk runs at).
131 pub(crate) const fn lazy_depth(self) -> u8 {
132 match self {
133 Self::Fast | Self::Dfast | Self::Greedy => 0,
134 Self::Lazy => 1,
135 _ => 2,
136 }
137 }
138}
139
140/// Whether literals are entropy-coded, the drop-in equivalent of C zstd's
141/// `ZSTD_c_literalCompressionMode` (`ZSTD_ps_auto` / `ZSTD_ps_enable` /
142/// `ZSTD_ps_disable`).
143#[derive(Copy, Clone, Debug, Default, PartialEq, Eq)]
144pub enum LiteralCompressionMode {
145 /// The level decides: literals are stored raw on the fast strategy's
146 /// acceleration (negative) levels, where the Huffman pass costs more
147 /// speed than it saves, and compressed everywhere else.
148 #[default]
149 Auto,
150 /// Compress literals on every level, the negative ones included. A
151 /// block whose literals do not shrink still stores them raw.
152 Enable,
153 /// Never compress literals: every literal section is stored raw.
154 Disable,
155}
156
157/// One tunable compression parameter — the analogue of a C zstd
158/// `ZSTD_cParameter`. Used to query bounds via [`CParameter::bounds`].
159#[derive(Copy, Clone, Debug, PartialEq, Eq)]
160#[non_exhaustive]
161pub enum CParameter {
162 /// Maximum back-reference distance, `log2`. C `ZSTD_c_windowLog`.
163 WindowLog,
164 /// Match-finder hash table size, `log2`. C `ZSTD_c_hashLog`.
165 HashLog,
166 /// Match-finder chain table size, `log2`. C `ZSTD_c_chainLog`.
167 ChainLog,
168 /// Number of search attempts, `log2`. C `ZSTD_c_searchLog`.
169 SearchLog,
170 /// Minimum match length in bytes. C `ZSTD_c_minMatch`.
171 MinMatch,
172 /// "Good enough" match length that ends the search. C `ZSTD_c_targetLength`.
173 TargetLength,
174 /// Match-finder [`Strategy`] (1..=9). C `ZSTD_c_strategy`.
175 Strategy,
176 /// LDM enable flag (0/1). C `ZSTD_c_enableLongDistanceMatching`.
177 EnableLongDistanceMatching,
178 /// LDM hash table size, `log2`. C `ZSTD_c_ldmHashLog`.
179 LdmHashLog,
180 /// LDM minimum match length in bytes. C `ZSTD_c_ldmMinMatch`.
181 LdmMinMatch,
182 /// LDM bucket size, `log2`. C `ZSTD_c_ldmBucketSizeLog`.
183 LdmBucketSizeLog,
184 /// LDM hash-insertion rate, `log2`. C `ZSTD_c_ldmHashRateLog`.
185 LdmHashRateLog,
186}
187
188/// Inclusive `[lower_bound, upper_bound]` range for a [`CParameter`],
189/// the drop-in equivalent of C zstd's `ZSTD_bounds`.
190#[derive(Copy, Clone, Debug, PartialEq, Eq)]
191pub struct Bounds {
192 /// Smallest accepted value (inclusive).
193 pub lower_bound: i64,
194 /// Largest accepted value (inclusive).
195 pub upper_bound: i64,
196}
197
198impl Bounds {
199 /// Whether `value` falls within `[lower_bound, upper_bound]`.
200 pub const fn contains(&self, value: i64) -> bool {
201 value >= self.lower_bound && value <= self.upper_bound
202 }
203
204 /// `value` moved to the nearest end of the range when it falls outside.
205 ///
206 /// ```
207 /// use structured_zstd::encoding::CParameter;
208 /// assert_eq!(CParameter::MinMatch.bounds().clamp(9), 7);
209 /// assert_eq!(CParameter::MinMatch.bounds().clamp(5), 5);
210 /// ```
211 pub const fn clamp(&self, value: i64) -> i64 {
212 if value < self.lower_bound {
213 self.lower_bound
214 } else if value > self.upper_bound {
215 self.upper_bound
216 } else {
217 value
218 }
219 }
220}
221
222impl CParameter {
223 /// Inclusive value bounds for this parameter, mirroring
224 /// `ZSTD_cParam_getBounds`. Window/hash/chain logs cap at 30 (the
225 /// encoder's match-finder ceiling) rather than the 31 C allows on
226 /// 64-bit, because the back-reference history is indexed with `u32`
227 /// positions over a `2 * window` eviction band.
228 pub const fn bounds(self) -> Bounds {
229 let (lower_bound, upper_bound) = match self {
230 // ZSTD_WINDOWLOG_MIN .. encoder ceiling.
231 Self::WindowLog => (10, 30),
232 // ZSTD_HASHLOG_MIN .. ZSTD_HASHLOG_MAX.
233 Self::HashLog => (6, 30),
234 // ZSTD_CHAINLOG_MIN .. ZSTD_CHAINLOG_MAX (64-bit).
235 Self::ChainLog => (6, 30),
236 // ZSTD_SEARCHLOG_MIN .. ZSTD_SEARCHLOG_MAX (64-bit).
237 Self::SearchLog => (1, 30),
238 // ZSTD_MINMATCH_MIN .. ZSTD_MINMATCH_MAX.
239 Self::MinMatch => (3, 7),
240 // ZSTD_TARGETLENGTH_MIN .. ZSTD_TARGETLENGTH_MAX.
241 Self::TargetLength => (0, 131_072),
242 // ZSTD_fast .. ZSTD_btultra2.
243 Self::Strategy => (1, 9),
244 // Boolean flag.
245 Self::EnableLongDistanceMatching => (0, 1),
246 // ZSTD_LDM_HASHLOG_MIN .. ZSTD_LDM_HASHLOG_MAX.
247 Self::LdmHashLog => (6, 30),
248 // ZSTD_LDM_MINMATCH_MIN .. ZSTD_LDM_MINMATCH_MAX.
249 Self::LdmMinMatch => (4, 4096),
250 // ZSTD_LDM_BUCKETSIZELOG_MIN .. ZSTD_LDM_BUCKETSIZELOG_MAX.
251 Self::LdmBucketSizeLog => (1, 8),
252 // ZSTD_LDM_HASHRATELOG_MIN .. ZSTD_WINDOWLOG_MAX - ZSTD_HASHLOG_MIN.
253 Self::LdmHashRateLog => (0, 24),
254 };
255 Bounds {
256 lower_bound,
257 upper_bound,
258 }
259 }
260}
261
262/// The match-finder parameters a numeric level selects for a source of a given
263/// size, the drop-in equivalent of C zstd's `ZSTD_getCParams`.
264///
265/// The size matters: the reference's level table has a row per source-size
266/// tier, and the tier changes the strategy as well as the table widths, so one
267/// level can run a different match-finder on a small input than on a large
268/// one. This is the selection the encoder itself makes for a frame without a
269/// dictionary; knobs set through [`CompressionParameters`] override it.
270///
271/// # Examples
272///
273/// ```
274/// use structured_zstd::encoding::{LevelParameters, Strategy};
275///
276/// // Level 11 is a lazy2 level on a large or unknown-size source...
277/// let large = LevelParameters::for_level(11, None, 0);
278/// assert_eq!(large.strategy, Strategy::Lazy2);
279///
280/// // ...and the optimal parser on one of 16 KiB or less, whose window is
281/// // also cut down to the source.
282/// let small = LevelParameters::for_level(11, Some(4096), 0);
283/// assert_eq!(small.strategy, Strategy::Btopt);
284/// assert_eq!(small.window_log, 12);
285/// ```
286#[derive(Copy, Clone, Debug, PartialEq, Eq)]
287pub struct LevelParameters {
288 /// Back-reference window, `log2`. C `windowLog`.
289 pub window_log: u32,
290 /// Chain / binary-tree table size, `log2`. C `chainLog`.
291 pub chain_log: u32,
292 /// Hash table size, `log2`. C `hashLog`.
293 pub hash_log: u32,
294 /// Search attempts per position, `log2`. C `searchLog`.
295 pub search_log: u32,
296 /// Match-finder hash width, in bytes. C `minMatch`.
297 pub min_match: u32,
298 /// Length that ends a search early; on a negative level, the step size
299 /// minus one. C `targetLength`.
300 pub target_length: u32,
301 /// The match-finder strategy. C `strategy`.
302 pub strategy: Strategy,
303}
304
305impl LevelParameters {
306 /// Parameters for `level` compressing a source of `source_size` bytes with
307 /// a `dictionary_size`-byte dictionary (`0` for none).
308 ///
309 /// `level` is on the reference's scale: `0` is the default level, levels
310 /// above [`CompressionLevel::MAX_LEVEL`] clamp to it, and a negative level
311 /// is an acceleration factor. `None` is a source of unknown size, which is
312 /// sized as a large one; `Some(0)` is a source that is really empty.
313 /// C `ZSTD_getCParams` spells unknown as `0`, so a caller porting from it
314 /// maps `0` to `None`.
315 pub fn for_level(level: i32, source_size: Option<u64>, dictionary_size: usize) -> Self {
316 let size = source_size.unwrap_or(crate::encoding::cparams::CONTENTSIZE_UNKNOWN);
317 let cp = crate::encoding::cparams::get_cparams(level, size, dictionary_size);
318 Self {
319 window_log: cp.window_log,
320 chain_log: cp.chain_log,
321 hash_log: cp.hash_log,
322 search_log: cp.search_log,
323 min_match: cp.min_match,
324 target_length: cp.target_length,
325 // Every row of the level table names one of the nine strategies.
326 strategy: Strategy::from_ordinal(cp.strategy)
327 .expect("the level table only holds strategies 1..=9"),
328 }
329 }
330
331 /// These parameters fitted to a source of `source_size` bytes and a
332 /// `dictionary_size`-byte dictionary, the drop-in equivalent of C zstd's
333 /// `ZSTD_adjustCParams`: a window wider than the data comes down to it, and
334 /// the tables shrink with the window. `None` is a source of unknown size,
335 /// which leaves the tables alone.
336 ///
337 /// Every field must already lie within its [`CParameter::bounds`].
338 ///
339 /// # Examples
340 ///
341 /// ```
342 /// use structured_zstd::encoding::LevelParameters;
343 ///
344 /// let level_19 = LevelParameters::for_level(19, None, 0);
345 /// let small = level_19.adjusted(Some(1000), 0);
346 /// assert_eq!(small.window_log, 10);
347 /// assert!(small.hash_log <= 11);
348 /// assert_eq!(level_19.adjusted(None, 0), level_19);
349 /// ```
350 pub fn adjusted(self, source_size: Option<u64>, dictionary_size: usize) -> Self {
351 let cp = crate::encoding::cparams::CParams {
352 window_log: self.window_log,
353 chain_log: self.chain_log,
354 hash_log: self.hash_log,
355 search_log: self.search_log,
356 min_match: self.min_match,
357 target_length: self.target_length,
358 strategy: self.strategy.ordinal(),
359 };
360 let size = source_size.unwrap_or(crate::encoding::cparams::CONTENTSIZE_UNKNOWN);
361 let cp = crate::encoding::cparams::adjust_cparams(cp, size, dictionary_size, false);
362 Self {
363 window_log: cp.window_log,
364 chain_log: cp.chain_log,
365 hash_log: cp.hash_log,
366 search_log: cp.search_log,
367 min_match: cp.min_match,
368 target_length: cp.target_length,
369 strategy: self.strategy,
370 }
371 }
372}
373
374/// Error returned by [`CompressionParametersBuilder::build`] when a knob
375/// is set outside its [`CParameter::bounds`].
376#[derive(Copy, Clone, Debug, PartialEq, Eq)]
377#[non_exhaustive]
378pub enum ParameterError {
379 /// A parameter was set to a value outside its inclusive bounds.
380 OutOfBounds {
381 /// Which parameter violated its range.
382 parameter: CParameter,
383 /// The rejected value.
384 value: i64,
385 /// The inclusive `[lower, upper]` range it had to fall within.
386 bounds: Bounds,
387 },
388}
389
390impl core::fmt::Display for ParameterError {
391 fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
392 match self {
393 Self::OutOfBounds {
394 parameter,
395 value,
396 bounds,
397 } => write!(
398 f,
399 "compression parameter {parameter:?} = {value} out of bounds \
400 [{}, {}]",
401 bounds.lower_bound, bounds.upper_bound
402 ),
403 }
404 }
405}
406
407#[cfg(feature = "std")]
408impl std::error::Error for ParameterError {}
409
410/// LDM tuning overrides — every knob is `Option`, falling back to the
411/// strategy-derived upstream zstd default (`LdmParams::adjust_for`) when unset.
412#[derive(Copy, Clone, Debug, Default, PartialEq, Eq)]
413pub(crate) struct LdmOverride {
414 pub(crate) hash_log: Option<u32>,
415 pub(crate) min_match: Option<u32>,
416 pub(crate) bucket_size_log: Option<u32>,
417 pub(crate) hash_rate_log: Option<u32>,
418}
419
420/// Internal per-knob override set consumed by the match-generator's
421/// `reset` path. Every field left `None` inherits the base level's
422/// resolved value, so the default path is byte-identical to level-based
423/// compression.
424#[derive(Copy, Clone, Debug, Default, PartialEq, Eq)]
425pub(crate) struct ParamOverrides {
426 pub(crate) window_log: Option<u8>,
427 pub(crate) hash_log: Option<u32>,
428 pub(crate) chain_log: Option<u32>,
429 pub(crate) search_log: Option<u32>,
430 pub(crate) min_match: Option<u32>,
431 pub(crate) target_length: Option<u32>,
432 pub(crate) strategy: Option<Strategy>,
433 /// `Some` when `enable_long_distance_matching(true)` was set; carries
434 /// the (possibly empty) LDM knob overrides.
435 pub(crate) ldm: Option<LdmOverride>,
436 /// Whether literals are entropy-coded; `Auto` leaves it to the level.
437 pub(crate) literal_compression: LiteralCompressionMode,
438}
439
440impl ParamOverrides {
441 /// Whether any knob overrides the base level. An all-`None`
442 /// override is a no-op the `reset` path can skip entirely, keeping
443 /// the default level-based geometry byte-identical.
444 pub(crate) fn is_empty(&self) -> bool {
445 self.window_log.is_none()
446 && self.hash_log.is_none()
447 && self.chain_log.is_none()
448 && self.search_log.is_none()
449 && self.min_match.is_none()
450 && self.target_length.is_none()
451 && self.strategy.is_none()
452 && self.ldm.is_none()
453 && self.literal_compression == LiteralCompressionMode::Auto
454 }
455}
456
457/// Fully-resolved fine-grained compression parameters. Build through
458/// [`CompressionParameters::builder`]; pass to
459/// [`FrameCompressor::set_parameters`](crate::encoding::FrameCompressor::set_parameters)
460/// or [`compress_with_parameters`](crate::encoding::compress_with_parameters).
461///
462/// Wraps a base [`CompressionLevel`] plus the set of knobs that override it. A parameter set that
463/// overrides nothing is equivalent to compressing at its base level.
464#[derive(Copy, Clone, Debug, PartialEq, Eq)]
465pub struct CompressionParameters {
466 level: CompressionLevel,
467 overrides: ParamOverrides,
468}
469
470impl CompressionParameters {
471 /// Start a builder from a base compression level. Knobs left unset
472 /// inherit that level's resolved defaults.
473 pub fn builder(level: CompressionLevel) -> CompressionParametersBuilder {
474 CompressionParametersBuilder {
475 level,
476 window_log: None,
477 hash_log: None,
478 chain_log: None,
479 search_log: None,
480 min_match: None,
481 target_length: None,
482 strategy: None,
483 enable_ldm: false,
484 ldm: LdmOverride::default(),
485 literal_compression: LiteralCompressionMode::Auto,
486 }
487 }
488
489 /// The base compression level these parameters override.
490 pub fn level(&self) -> CompressionLevel {
491 self.level
492 }
493
494 /// Whether long-distance matching is enabled.
495 pub fn long_distance_matching_enabled(&self) -> bool {
496 self.overrides.ldm.is_some()
497 }
498
499 /// How literals are entropy-coded (see [`LiteralCompressionMode`]).
500 pub fn literal_compression_mode(&self) -> LiteralCompressionMode {
501 self.overrides.literal_compression
502 }
503
504 /// The window, `log2`, the encoder gives a frame of `source_size` bytes
505 /// (`None` = unknown) with `dictionary_size` bytes of dictionary content
506 /// under these parameters: resolved exactly as the encoder resolves it.
507 ///
508 /// ```
509 /// use structured_zstd::encoding::{CompressionLevel, CompressionParameters};
510 ///
511 /// let params = CompressionParameters::builder(CompressionLevel::Level(19))
512 /// .window_log(23)
513 /// .build()
514 /// .unwrap();
515 /// assert_eq!(params.window_log_for(None, 0), 23);
516 /// assert!(params.window_log_for(Some(1000), 0) < 23);
517 /// ```
518 pub fn window_log_for(&self, source_size: Option<u64>, dictionary_size: usize) -> u32 {
519 let sizes = crate::encoding::DictionarySizes {
520 content: dictionary_size,
521 serialized: dictionary_size,
522 };
523 let params =
524 crate::encoding::levels::config::frame_level_params(self, source_size, Some(sizes));
525 u32::from(params.window_log)
526 }
527
528 /// These parameters as a dictionary's own, driving a frame of
529 /// `source_size` bytes (`None` = unknown): an explicit window is raised to
530 /// cover the source, as far as 512 KiB, as upstream raises a prepared
531 /// dictionary's window for a frame of known size
532 /// (`ZSTD_compressBegin_usingCDict_internal`, zstd_compress.c:5844-5852).
533 /// A window left to the level is resolved for the source and covers it.
534 ///
535 /// ```
536 /// use structured_zstd::encoding::{CompressionLevel, CompressionParameters};
537 ///
538 /// let params = CompressionParameters::builder(CompressionLevel::Level(3))
539 /// .window_log(10)
540 /// .build()
541 /// .unwrap();
542 /// let source = Some(1 << 20);
543 /// assert_eq!(params.for_dictionary_frame(source).window_log_for(source, 0), 19);
544 /// assert_eq!(params.for_dictionary_frame(None).window_log_for(None, 0), 10);
545 /// ```
546 pub fn for_dictionary_frame(mut self, source_size: Option<u64>) -> Self {
547 const RAISE_LIMIT: u64 = 1 << 19;
548 let known =
549 source_size.filter(|size| *size != crate::encoding::cparams::CONTENTSIZE_UNKNOWN);
550 if let (Some(window), Some(source)) = (self.overrides.window_log, known) {
551 // Bounded by `RAISE_LIMIT`, so it fits and its log is at most 19.
552 let limited = source.min(RAISE_LIMIT) as u32;
553 let source_log = if limited > 1 {
554 (u32::BITS - (limited - 1).leading_zeros()) as u8
555 } else {
556 1
557 };
558 self.overrides.window_log = Some(window.max(source_log));
559 }
560 self
561 }
562
563 pub(crate) fn overrides(&self) -> ParamOverrides {
564 self.overrides
565 }
566}
567
568/// Builder for [`CompressionParameters`]. Each setter records one knob;
569/// [`Self::build`] validates them against [`CParameter::bounds`].
570#[derive(Copy, Clone, Debug)]
571pub struct CompressionParametersBuilder {
572 level: CompressionLevel,
573 window_log: Option<u32>,
574 hash_log: Option<u32>,
575 chain_log: Option<u32>,
576 search_log: Option<u32>,
577 min_match: Option<u32>,
578 target_length: Option<u32>,
579 strategy: Option<Strategy>,
580 enable_ldm: bool,
581 ldm: LdmOverride,
582 literal_compression: LiteralCompressionMode,
583}
584
585impl CompressionParametersBuilder {
586 /// Decide whether literals are entropy-coded, overriding the level's own
587 /// choice. C `ZSTD_c_literalCompressionMode`.
588 ///
589 /// ```rust
590 /// use structured_zstd::encoding::{
591 /// compress_with_parameters, CompressionLevel, CompressionParameters,
592 /// LiteralCompressionMode,
593 /// };
594 ///
595 /// // Literal-heavy input: 32 distinct symbols and nothing for the match
596 /// // finder to repeat, so only entropy-coding the literals shrinks it.
597 /// let text: Vec<u8> = (0..8192u32)
598 /// .map(|i| b'a' + (i.wrapping_mul(2_654_435_761) >> 27) as u8)
599 /// .collect();
600 /// // Level -3 stores literals raw by default; asking for compression
601 /// // makes the frame smaller on input like this.
602 /// let compressed_literals = CompressionParameters::builder(CompressionLevel::Level(-3))
603 /// .literal_compression(LiteralCompressionMode::Enable)
604 /// .build()
605 /// .unwrap();
606 /// let plain = compress_with_parameters(
607 /// &text[..],
608 /// &CompressionParameters::builder(CompressionLevel::Level(-3)).build().unwrap(),
609 /// );
610 /// let coded = compress_with_parameters(&text[..], &compressed_literals);
611 /// assert!(coded.len() < plain.len());
612 /// ```
613 pub fn literal_compression(mut self, mode: LiteralCompressionMode) -> Self {
614 self.literal_compression = mode;
615 self
616 }
617
618 /// Override the maximum back-reference distance (`log2`). C
619 /// `ZSTD_c_windowLog`.
620 pub fn window_log(mut self, value: u32) -> Self {
621 self.window_log = Some(value);
622 self
623 }
624
625 /// Override the match-finder hash table size (`log2`). C `ZSTD_c_hashLog`.
626 pub fn hash_log(mut self, value: u32) -> Self {
627 self.hash_log = Some(value);
628 self
629 }
630
631 /// Override the match-finder chain table size (`log2`). C `ZSTD_c_chainLog`.
632 pub fn chain_log(mut self, value: u32) -> Self {
633 self.chain_log = Some(value);
634 self
635 }
636
637 /// Override the search-attempts count (`log2`). C `ZSTD_c_searchLog`.
638 pub fn search_log(mut self, value: u32) -> Self {
639 self.search_log = Some(value);
640 self
641 }
642
643 /// Override the minimum match length in bytes. C `ZSTD_c_minMatch`.
644 pub fn min_match(mut self, value: u32) -> Self {
645 self.min_match = Some(value);
646 self
647 }
648
649 /// Override the "good enough" target match length. C `ZSTD_c_targetLength`.
650 pub fn target_length(mut self, value: u32) -> Self {
651 self.target_length = Some(value);
652 self
653 }
654
655 /// Override the match-finder [`Strategy`]. C `ZSTD_c_strategy`.
656 pub fn strategy(mut self, value: Strategy) -> Self {
657 self.strategy = Some(value);
658 self
659 }
660
661 /// Enable or disable long-distance matching. C
662 /// `ZSTD_c_enableLongDistanceMatching`. Off at every level preset.
663 /// This is the explicit activation toggle; the `ldm_*` knob setters
664 /// also enable LDM implicitly. The flag is plain last-write-wins, so
665 /// a trailing `enable_long_distance_matching(false)` disables LDM even
666 /// if an earlier `ldm_*` call set a knob (the knob is then ignored at
667 /// [`build`](Self::build)).
668 pub fn enable_long_distance_matching(mut self, enable: bool) -> Self {
669 self.enable_ldm = enable;
670 self
671 }
672
673 /// Override the LDM hash table size (`log2`). C `ZSTD_c_ldmHashLog`.
674 /// Implies [`Self::enable_long_distance_matching(true)`](Self::enable_long_distance_matching).
675 pub fn ldm_hash_log(mut self, value: u32) -> Self {
676 self.enable_ldm = true;
677 self.ldm.hash_log = Some(value);
678 self
679 }
680
681 /// Override the LDM minimum match length. C `ZSTD_c_ldmMinMatch`.
682 /// Implies [`Self::enable_long_distance_matching(true)`](Self::enable_long_distance_matching).
683 pub fn ldm_min_match(mut self, value: u32) -> Self {
684 self.enable_ldm = true;
685 self.ldm.min_match = Some(value);
686 self
687 }
688
689 /// Override the LDM bucket size (`log2`). C `ZSTD_c_ldmBucketSizeLog`.
690 /// Implies [`Self::enable_long_distance_matching(true)`](Self::enable_long_distance_matching).
691 pub fn ldm_bucket_size_log(mut self, value: u32) -> Self {
692 self.enable_ldm = true;
693 self.ldm.bucket_size_log = Some(value);
694 self
695 }
696
697 /// Override the LDM hash-insertion rate (`log2`). C `ZSTD_c_ldmHashRateLog`.
698 /// Implies [`Self::enable_long_distance_matching(true)`](Self::enable_long_distance_matching).
699 pub fn ldm_hash_rate_log(mut self, value: u32) -> Self {
700 self.enable_ldm = true;
701 self.ldm.hash_rate_log = Some(value);
702 self
703 }
704
705 /// Validate every set knob against [`CParameter::bounds`] and
706 /// produce the resolved [`CompressionParameters`].
707 ///
708 /// # Errors
709 ///
710 /// Returns [`ParameterError::OutOfBounds`] for the first knob whose
711 /// value falls outside its inclusive range.
712 pub fn build(self) -> Result<CompressionParameters, ParameterError> {
713 check(CParameter::WindowLog, self.window_log)?;
714 check(CParameter::HashLog, self.hash_log)?;
715 check(CParameter::ChainLog, self.chain_log)?;
716 check(CParameter::SearchLog, self.search_log)?;
717 check(CParameter::MinMatch, self.min_match)?;
718 check(CParameter::TargetLength, self.target_length)?;
719 if let Some(s) = self.strategy {
720 check(CParameter::Strategy, Some(s.ordinal()))?;
721 }
722 let ldm = if self.enable_ldm {
723 check(CParameter::LdmHashLog, self.ldm.hash_log)?;
724 check(CParameter::LdmMinMatch, self.ldm.min_match)?;
725 check(CParameter::LdmBucketSizeLog, self.ldm.bucket_size_log)?;
726 check(CParameter::LdmHashRateLog, self.ldm.hash_rate_log)?;
727 Some(self.ldm)
728 } else {
729 None
730 };
731 Ok(CompressionParameters {
732 level: self.level,
733 overrides: ParamOverrides {
734 // `window_log` is bounds-checked at <= 30, so the cast is lossless.
735 window_log: self.window_log.map(|v| v as u8),
736 hash_log: self.hash_log,
737 chain_log: self.chain_log,
738 search_log: self.search_log,
739 min_match: self.min_match,
740 target_length: self.target_length,
741 strategy: self.strategy,
742 ldm,
743 literal_compression: self.literal_compression,
744 },
745 })
746 }
747}
748
749/// Validate one optional knob against its bounds.
750fn check(parameter: CParameter, value: Option<u32>) -> Result<(), ParameterError> {
751 if let Some(value) = value {
752 let bounds = parameter.bounds();
753 let value = i64::from(value);
754 if !bounds.contains(value) {
755 return Err(ParameterError::OutOfBounds {
756 parameter,
757 value,
758 bounds,
759 });
760 }
761 }
762 Ok(())
763}
764
765#[cfg(test)]
766mod tests;