Show More
@@ -118,7 +118,7 b' fn test_combine_changeset_copies() {' | |||
|
118 | 118 | // keys to copy source values. Note: the arrows for map literal syntax |
|
119 | 119 | // point **backwards** compared to the logical direction of copy! |
|
120 | 120 | |
|
121 | use crate::NULL_REVISION as NULL; | |
|
121 | use crate::revlog::NULL_REVISION as NULL; | |
|
122 | 122 | use Action::*; |
|
123 | 123 | use MergeCase::*; |
|
124 | 124 |
@@ -6,8 +6,8 b'' | |||
|
6 | 6 | // GNU General Public License version 2 or any later version. |
|
7 | 7 | |
|
8 | 8 | use crate::repo::Repo; |
|
9 | use crate::revlog::revlog::RevlogError; | |
|
10 | 9 | use crate::revlog::Node; |
|
10 | use crate::revlog::RevlogError; | |
|
11 | 11 | |
|
12 | 12 | use crate::utils::hg_path::HgPath; |
|
13 | 13 |
@@ -7,7 +7,7 b'' | |||
|
7 | 7 | |
|
8 | 8 | use crate::repo::Repo; |
|
9 | 9 | use crate::requirements; |
|
10 |
use crate |
|
|
10 | use crate::revlog::{Revlog, RevlogError}; | |
|
11 | 11 | |
|
12 | 12 | /// Kind of data to debug |
|
13 | 13 | #[derive(Debug, Copy, Clone)] |
@@ -10,7 +10,7 b' use crate::dirstate_tree::on_disk::{for_' | |||
|
10 | 10 | use crate::errors::HgError; |
|
11 | 11 | use crate::repo::Repo; |
|
12 | 12 | use crate::revlog::manifest::Manifest; |
|
13 |
use crate::revlog:: |
|
|
13 | use crate::revlog::RevlogError; | |
|
14 | 14 | use crate::utils::hg_path::HgPath; |
|
15 | 15 | use crate::DirstateError; |
|
16 | 16 | use rayon::prelude::*; |
@@ -8,7 +8,7 b' use crate::errors::{HgError, IoResultExt' | |||
|
8 | 8 | use crate::lock::{try_with_lock_no_wait, LockError}; |
|
9 | 9 | use crate::manifest::{Manifest, Manifestlog}; |
|
10 | 10 | use crate::revlog::filelog::Filelog; |
|
11 |
use crate::revlog:: |
|
|
11 | use crate::revlog::RevlogError; | |
|
12 | 12 | use crate::utils::files::get_path_from_bytes; |
|
13 | 13 | use crate::utils::hg_path::HgPath; |
|
14 | 14 | use crate::utils::SliceExt; |
@@ -1,7 +1,7 b'' | |||
|
1 | 1 | use crate::errors::HgError; |
|
2 | use crate::revlog::revlog::{Revlog, RevlogEntry, RevlogError}; | |
|
3 | 2 | use crate::revlog::Revision; |
|
4 | 3 | use crate::revlog::{Node, NodePrefix}; |
|
4 | use crate::revlog::{Revlog, RevlogEntry, RevlogError}; | |
|
5 | 5 | use crate::utils::hg_path::HgPath; |
|
6 | 6 | use crate::vfs::Vfs; |
|
7 | 7 | use itertools::Itertools; |
@@ -1,10 +1,10 b'' | |||
|
1 | 1 | use crate::errors::HgError; |
|
2 | 2 | use crate::repo::Repo; |
|
3 | 3 | use crate::revlog::path_encode::path_encode; |
|
4 | use crate::revlog::revlog::RevlogEntry; | |
|
5 | use crate::revlog::revlog::{Revlog, RevlogError}; | |
|
6 | 4 | use crate::revlog::NodePrefix; |
|
7 | 5 | use crate::revlog::Revision; |
|
6 | use crate::revlog::RevlogEntry; | |
|
7 | use crate::revlog::{Revlog, RevlogError}; | |
|
8 | 8 | use crate::utils::files::get_path_from_bytes; |
|
9 | 9 | use crate::utils::hg_path::HgPath; |
|
10 | 10 | use crate::utils::SliceExt; |
@@ -1,7 +1,7 b'' | |||
|
1 | 1 | use crate::errors::HgError; |
|
2 | use crate::revlog::revlog::{Revlog, RevlogError}; | |
|
3 | 2 | use crate::revlog::Revision; |
|
4 | 3 | use crate::revlog::{Node, NodePrefix}; |
|
4 | use crate::revlog::{Revlog, RevlogError}; | |
|
5 | 5 | use crate::utils::hg_path::HgPath; |
|
6 | 6 | use crate::utils::SliceExt; |
|
7 | 7 | use crate::vfs::Vfs; |
This diff has been collapsed as it changes many lines, (642 lines changed) Show them Hide them | |||
@@ -1,4 +1,4 b'' | |||
|
1 |
// Copyright 2018-202 |
|
|
1 | // Copyright 2018-2023 Georges Racinet <georges.racinet@octobus.net> | |
|
2 | 2 | // and Mercurial contributors |
|
3 | 3 | // |
|
4 | 4 | // This software may be used and distributed according to the terms of the |
@@ -15,7 +15,22 b' pub mod filelog;' | |||
|
15 | 15 | pub mod index; |
|
16 | 16 | pub mod manifest; |
|
17 | 17 | pub mod patch; |
|
18 | pub mod revlog; | |
|
18 | ||
|
19 | use std::borrow::Cow; | |
|
20 | use std::io::Read; | |
|
21 | use std::ops::Deref; | |
|
22 | use std::path::Path; | |
|
23 | ||
|
24 | use flate2::read::ZlibDecoder; | |
|
25 | use sha1::{Digest, Sha1}; | |
|
26 | use zstd; | |
|
27 | ||
|
28 | use self::node::{NODE_BYTES_LENGTH, NULL_NODE}; | |
|
29 | use self::nodemap_docket::NodeMapDocket; | |
|
30 | use super::index::Index; | |
|
31 | use super::nodemap::{NodeMap, NodeMapError}; | |
|
32 | use crate::errors::HgError; | |
|
33 | use crate::vfs::Vfs; | |
|
19 | 34 | |
|
20 | 35 | /// Mercurial revision numbers |
|
21 | 36 | /// |
@@ -70,3 +85,626 b' pub trait RevlogIndex {' | |||
|
70 | 85 | /// `NULL_REVISION` is not considered to be out of bounds. |
|
71 | 86 | fn node(&self, rev: Revision) -> Option<&Node>; |
|
72 | 87 | } |
|
88 | ||
|
89 | const REVISION_FLAG_CENSORED: u16 = 1 << 15; | |
|
90 | const REVISION_FLAG_ELLIPSIS: u16 = 1 << 14; | |
|
91 | const REVISION_FLAG_EXTSTORED: u16 = 1 << 13; | |
|
92 | const REVISION_FLAG_HASCOPIESINFO: u16 = 1 << 12; | |
|
93 | ||
|
94 | // Keep this in sync with REVIDX_KNOWN_FLAGS in | |
|
95 | // mercurial/revlogutils/flagutil.py | |
|
96 | const REVIDX_KNOWN_FLAGS: u16 = REVISION_FLAG_CENSORED | |
|
97 | | REVISION_FLAG_ELLIPSIS | |
|
98 | | REVISION_FLAG_EXTSTORED | |
|
99 | | REVISION_FLAG_HASCOPIESINFO; | |
|
100 | ||
|
101 | const NULL_REVLOG_ENTRY_FLAGS: u16 = 0; | |
|
102 | ||
|
103 | #[derive(Debug, derive_more::From)] | |
|
104 | pub enum RevlogError { | |
|
105 | InvalidRevision, | |
|
106 | /// Working directory is not supported | |
|
107 | WDirUnsupported, | |
|
108 | /// Found more than one entry whose ID match the requested prefix | |
|
109 | AmbiguousPrefix, | |
|
110 | #[from] | |
|
111 | Other(HgError), | |
|
112 | } | |
|
113 | ||
|
114 | impl From<NodeMapError> for RevlogError { | |
|
115 | fn from(error: NodeMapError) -> Self { | |
|
116 | match error { | |
|
117 | NodeMapError::MultipleResults => RevlogError::AmbiguousPrefix, | |
|
118 | NodeMapError::RevisionNotInIndex(rev) => RevlogError::corrupted( | |
|
119 | format!("nodemap point to revision {} not in index", rev), | |
|
120 | ), | |
|
121 | } | |
|
122 | } | |
|
123 | } | |
|
124 | ||
|
125 | fn corrupted<S: AsRef<str>>(context: S) -> HgError { | |
|
126 | HgError::corrupted(format!("corrupted revlog, {}", context.as_ref())) | |
|
127 | } | |
|
128 | ||
|
129 | impl RevlogError { | |
|
130 | fn corrupted<S: AsRef<str>>(context: S) -> Self { | |
|
131 | RevlogError::Other(corrupted(context)) | |
|
132 | } | |
|
133 | } | |
|
134 | ||
|
135 | /// Read only implementation of revlog. | |
|
136 | pub struct Revlog { | |
|
137 | /// When index and data are not interleaved: bytes of the revlog index. | |
|
138 | /// When index and data are interleaved: bytes of the revlog index and | |
|
139 | /// data. | |
|
140 | index: Index, | |
|
141 | /// When index and data are not interleaved: bytes of the revlog data | |
|
142 | data_bytes: Option<Box<dyn Deref<Target = [u8]> + Send>>, | |
|
143 | /// When present on disk: the persistent nodemap for this revlog | |
|
144 | nodemap: Option<nodemap::NodeTree>, | |
|
145 | } | |
|
146 | ||
|
147 | impl Revlog { | |
|
148 | /// Open a revlog index file. | |
|
149 | /// | |
|
150 | /// It will also open the associated data file if index and data are not | |
|
151 | /// interleaved. | |
|
152 | pub fn open( | |
|
153 | store_vfs: &Vfs, | |
|
154 | index_path: impl AsRef<Path>, | |
|
155 | data_path: Option<&Path>, | |
|
156 | use_nodemap: bool, | |
|
157 | ) -> Result<Self, HgError> { | |
|
158 | let index_path = index_path.as_ref(); | |
|
159 | let index = { | |
|
160 | match store_vfs.mmap_open_opt(&index_path)? { | |
|
161 | None => Index::new(Box::new(vec![])), | |
|
162 | Some(index_mmap) => { | |
|
163 | let index = Index::new(Box::new(index_mmap))?; | |
|
164 | Ok(index) | |
|
165 | } | |
|
166 | } | |
|
167 | }?; | |
|
168 | ||
|
169 | let default_data_path = index_path.with_extension("d"); | |
|
170 | ||
|
171 | // type annotation required | |
|
172 | // won't recognize Mmap as Deref<Target = [u8]> | |
|
173 | let data_bytes: Option<Box<dyn Deref<Target = [u8]> + Send>> = | |
|
174 | if index.is_inline() { | |
|
175 | None | |
|
176 | } else { | |
|
177 | let data_path = data_path.unwrap_or(&default_data_path); | |
|
178 | let data_mmap = store_vfs.mmap_open(data_path)?; | |
|
179 | Some(Box::new(data_mmap)) | |
|
180 | }; | |
|
181 | ||
|
182 | let nodemap = if index.is_inline() || !use_nodemap { | |
|
183 | None | |
|
184 | } else { | |
|
185 | NodeMapDocket::read_from_file(store_vfs, index_path)?.map( | |
|
186 | |(docket, data)| { | |
|
187 | nodemap::NodeTree::load_bytes( | |
|
188 | Box::new(data), | |
|
189 | docket.data_length, | |
|
190 | ) | |
|
191 | }, | |
|
192 | ) | |
|
193 | }; | |
|
194 | ||
|
195 | Ok(Revlog { | |
|
196 | index, | |
|
197 | data_bytes, | |
|
198 | nodemap, | |
|
199 | }) | |
|
200 | } | |
|
201 | ||
|
202 | /// Return number of entries of the `Revlog`. | |
|
203 | pub fn len(&self) -> usize { | |
|
204 | self.index.len() | |
|
205 | } | |
|
206 | ||
|
207 | /// Returns `true` if the `Revlog` has zero `entries`. | |
|
208 | pub fn is_empty(&self) -> bool { | |
|
209 | self.index.is_empty() | |
|
210 | } | |
|
211 | ||
|
212 | /// Returns the node ID for the given revision number, if it exists in this | |
|
213 | /// revlog | |
|
214 | pub fn node_from_rev(&self, rev: Revision) -> Option<&Node> { | |
|
215 | if rev == NULL_REVISION { | |
|
216 | return Some(&NULL_NODE); | |
|
217 | } | |
|
218 | Some(self.index.get_entry(rev)?.hash()) | |
|
219 | } | |
|
220 | ||
|
221 | /// Return the revision number for the given node ID, if it exists in this | |
|
222 | /// revlog | |
|
223 | pub fn rev_from_node( | |
|
224 | &self, | |
|
225 | node: NodePrefix, | |
|
226 | ) -> Result<Revision, RevlogError> { | |
|
227 | if node.is_prefix_of(&NULL_NODE) { | |
|
228 | return Ok(NULL_REVISION); | |
|
229 | } | |
|
230 | ||
|
231 | if let Some(nodemap) = &self.nodemap { | |
|
232 | return nodemap | |
|
233 | .find_bin(&self.index, node)? | |
|
234 | .ok_or(RevlogError::InvalidRevision); | |
|
235 | } | |
|
236 | ||
|
237 | // Fallback to linear scan when a persistent nodemap is not present. | |
|
238 | // This happens when the persistent-nodemap experimental feature is not | |
|
239 | // enabled, or for small revlogs. | |
|
240 | // | |
|
241 | // TODO: consider building a non-persistent nodemap in memory to | |
|
242 | // optimize these cases. | |
|
243 | let mut found_by_prefix = None; | |
|
244 | for rev in (0..self.len() as Revision).rev() { | |
|
245 | let index_entry = self.index.get_entry(rev).ok_or_else(|| { | |
|
246 | HgError::corrupted( | |
|
247 | "revlog references a revision not in the index", | |
|
248 | ) | |
|
249 | })?; | |
|
250 | if node == *index_entry.hash() { | |
|
251 | return Ok(rev); | |
|
252 | } | |
|
253 | if node.is_prefix_of(index_entry.hash()) { | |
|
254 | if found_by_prefix.is_some() { | |
|
255 | return Err(RevlogError::AmbiguousPrefix); | |
|
256 | } | |
|
257 | found_by_prefix = Some(rev) | |
|
258 | } | |
|
259 | } | |
|
260 | found_by_prefix.ok_or(RevlogError::InvalidRevision) | |
|
261 | } | |
|
262 | ||
|
263 | /// Returns whether the given revision exists in this revlog. | |
|
264 | pub fn has_rev(&self, rev: Revision) -> bool { | |
|
265 | self.index.get_entry(rev).is_some() | |
|
266 | } | |
|
267 | ||
|
268 | /// Return the full data associated to a revision. | |
|
269 | /// | |
|
270 | /// All entries required to build the final data out of deltas will be | |
|
271 | /// retrieved as needed, and the deltas will be applied to the inital | |
|
272 | /// snapshot to rebuild the final data. | |
|
273 | pub fn get_rev_data( | |
|
274 | &self, | |
|
275 | rev: Revision, | |
|
276 | ) -> Result<Cow<[u8]>, RevlogError> { | |
|
277 | if rev == NULL_REVISION { | |
|
278 | return Ok(Cow::Borrowed(&[])); | |
|
279 | }; | |
|
280 | Ok(self.get_entry(rev)?.data()?) | |
|
281 | } | |
|
282 | ||
|
283 | /// Check the hash of some given data against the recorded hash. | |
|
284 | pub fn check_hash( | |
|
285 | &self, | |
|
286 | p1: Revision, | |
|
287 | p2: Revision, | |
|
288 | expected: &[u8], | |
|
289 | data: &[u8], | |
|
290 | ) -> bool { | |
|
291 | let e1 = self.index.get_entry(p1); | |
|
292 | let h1 = match e1 { | |
|
293 | Some(ref entry) => entry.hash(), | |
|
294 | None => &NULL_NODE, | |
|
295 | }; | |
|
296 | let e2 = self.index.get_entry(p2); | |
|
297 | let h2 = match e2 { | |
|
298 | Some(ref entry) => entry.hash(), | |
|
299 | None => &NULL_NODE, | |
|
300 | }; | |
|
301 | ||
|
302 | hash(data, h1.as_bytes(), h2.as_bytes()) == expected | |
|
303 | } | |
|
304 | ||
|
305 | /// Build the full data of a revision out its snapshot | |
|
306 | /// and its deltas. | |
|
307 | fn build_data_from_deltas( | |
|
308 | snapshot: RevlogEntry, | |
|
309 | deltas: &[RevlogEntry], | |
|
310 | ) -> Result<Vec<u8>, HgError> { | |
|
311 | let snapshot = snapshot.data_chunk()?; | |
|
312 | let deltas = deltas | |
|
313 | .iter() | |
|
314 | .rev() | |
|
315 | .map(RevlogEntry::data_chunk) | |
|
316 | .collect::<Result<Vec<_>, _>>()?; | |
|
317 | let patches: Vec<_> = | |
|
318 | deltas.iter().map(|d| patch::PatchList::new(d)).collect(); | |
|
319 | let patch = patch::fold_patch_lists(&patches); | |
|
320 | Ok(patch.apply(&snapshot)) | |
|
321 | } | |
|
322 | ||
|
323 | /// Return the revlog data. | |
|
324 | fn data(&self) -> &[u8] { | |
|
325 | match &self.data_bytes { | |
|
326 | Some(data_bytes) => data_bytes, | |
|
327 | None => panic!( | |
|
328 | "forgot to load the data or trying to access inline data" | |
|
329 | ), | |
|
330 | } | |
|
331 | } | |
|
332 | ||
|
333 | pub fn make_null_entry(&self) -> RevlogEntry { | |
|
334 | RevlogEntry { | |
|
335 | revlog: self, | |
|
336 | rev: NULL_REVISION, | |
|
337 | bytes: b"", | |
|
338 | compressed_len: 0, | |
|
339 | uncompressed_len: 0, | |
|
340 | base_rev_or_base_of_delta_chain: None, | |
|
341 | p1: NULL_REVISION, | |
|
342 | p2: NULL_REVISION, | |
|
343 | flags: NULL_REVLOG_ENTRY_FLAGS, | |
|
344 | hash: NULL_NODE, | |
|
345 | } | |
|
346 | } | |
|
347 | ||
|
348 | /// Get an entry of the revlog. | |
|
349 | pub fn get_entry( | |
|
350 | &self, | |
|
351 | rev: Revision, | |
|
352 | ) -> Result<RevlogEntry, RevlogError> { | |
|
353 | if rev == NULL_REVISION { | |
|
354 | return Ok(self.make_null_entry()); | |
|
355 | } | |
|
356 | let index_entry = self | |
|
357 | .index | |
|
358 | .get_entry(rev) | |
|
359 | .ok_or(RevlogError::InvalidRevision)?; | |
|
360 | let start = index_entry.offset(); | |
|
361 | let end = start + index_entry.compressed_len() as usize; | |
|
362 | let data = if self.index.is_inline() { | |
|
363 | self.index.data(start, end) | |
|
364 | } else { | |
|
365 | &self.data()[start..end] | |
|
366 | }; | |
|
367 | let entry = RevlogEntry { | |
|
368 | revlog: self, | |
|
369 | rev, | |
|
370 | bytes: data, | |
|
371 | compressed_len: index_entry.compressed_len(), | |
|
372 | uncompressed_len: index_entry.uncompressed_len(), | |
|
373 | base_rev_or_base_of_delta_chain: if index_entry | |
|
374 | .base_revision_or_base_of_delta_chain() | |
|
375 | == rev | |
|
376 | { | |
|
377 | None | |
|
378 | } else { | |
|
379 | Some(index_entry.base_revision_or_base_of_delta_chain()) | |
|
380 | }, | |
|
381 | p1: index_entry.p1(), | |
|
382 | p2: index_entry.p2(), | |
|
383 | flags: index_entry.flags(), | |
|
384 | hash: *index_entry.hash(), | |
|
385 | }; | |
|
386 | Ok(entry) | |
|
387 | } | |
|
388 | ||
|
389 | /// when resolving internal references within revlog, any errors | |
|
390 | /// should be reported as corruption, instead of e.g. "invalid revision" | |
|
391 | fn get_entry_internal( | |
|
392 | &self, | |
|
393 | rev: Revision, | |
|
394 | ) -> Result<RevlogEntry, HgError> { | |
|
395 | self.get_entry(rev) | |
|
396 | .map_err(|_| corrupted(format!("revision {} out of range", rev))) | |
|
397 | } | |
|
398 | } | |
|
399 | ||
|
400 | /// The revlog entry's bytes and the necessary informations to extract | |
|
401 | /// the entry's data. | |
|
402 | #[derive(Clone)] | |
|
403 | pub struct RevlogEntry<'a> { | |
|
404 | revlog: &'a Revlog, | |
|
405 | rev: Revision, | |
|
406 | bytes: &'a [u8], | |
|
407 | compressed_len: u32, | |
|
408 | uncompressed_len: i32, | |
|
409 | base_rev_or_base_of_delta_chain: Option<Revision>, | |
|
410 | p1: Revision, | |
|
411 | p2: Revision, | |
|
412 | flags: u16, | |
|
413 | hash: Node, | |
|
414 | } | |
|
415 | ||
|
416 | impl<'a> RevlogEntry<'a> { | |
|
417 | pub fn revision(&self) -> Revision { | |
|
418 | self.rev | |
|
419 | } | |
|
420 | ||
|
421 | pub fn node(&self) -> &Node { | |
|
422 | &self.hash | |
|
423 | } | |
|
424 | ||
|
425 | pub fn uncompressed_len(&self) -> Option<u32> { | |
|
426 | u32::try_from(self.uncompressed_len).ok() | |
|
427 | } | |
|
428 | ||
|
429 | pub fn has_p1(&self) -> bool { | |
|
430 | self.p1 != NULL_REVISION | |
|
431 | } | |
|
432 | ||
|
433 | pub fn p1_entry(&self) -> Result<Option<RevlogEntry>, RevlogError> { | |
|
434 | if self.p1 == NULL_REVISION { | |
|
435 | Ok(None) | |
|
436 | } else { | |
|
437 | Ok(Some(self.revlog.get_entry(self.p1)?)) | |
|
438 | } | |
|
439 | } | |
|
440 | ||
|
441 | pub fn p2_entry(&self) -> Result<Option<RevlogEntry>, RevlogError> { | |
|
442 | if self.p2 == NULL_REVISION { | |
|
443 | Ok(None) | |
|
444 | } else { | |
|
445 | Ok(Some(self.revlog.get_entry(self.p2)?)) | |
|
446 | } | |
|
447 | } | |
|
448 | ||
|
449 | pub fn p1(&self) -> Option<Revision> { | |
|
450 | if self.p1 == NULL_REVISION { | |
|
451 | None | |
|
452 | } else { | |
|
453 | Some(self.p1) | |
|
454 | } | |
|
455 | } | |
|
456 | ||
|
457 | pub fn p2(&self) -> Option<Revision> { | |
|
458 | if self.p2 == NULL_REVISION { | |
|
459 | None | |
|
460 | } else { | |
|
461 | Some(self.p2) | |
|
462 | } | |
|
463 | } | |
|
464 | ||
|
465 | pub fn is_censored(&self) -> bool { | |
|
466 | (self.flags & REVISION_FLAG_CENSORED) != 0 | |
|
467 | } | |
|
468 | ||
|
469 | pub fn has_length_affecting_flag_processor(&self) -> bool { | |
|
470 | // Relevant Python code: revlog.size() | |
|
471 | // note: ELLIPSIS is known to not change the content | |
|
472 | (self.flags & (REVIDX_KNOWN_FLAGS ^ REVISION_FLAG_ELLIPSIS)) != 0 | |
|
473 | } | |
|
474 | ||
|
475 | /// The data for this entry, after resolving deltas if any. | |
|
476 | pub fn rawdata(&self) -> Result<Cow<'a, [u8]>, HgError> { | |
|
477 | let mut entry = self.clone(); | |
|
478 | let mut delta_chain = vec![]; | |
|
479 | ||
|
480 | // The meaning of `base_rev_or_base_of_delta_chain` depends on | |
|
481 | // generaldelta. See the doc on `ENTRY_DELTA_BASE` in | |
|
482 | // `mercurial/revlogutils/constants.py` and the code in | |
|
483 | // [_chaininfo] and in [index_deltachain]. | |
|
484 | let uses_generaldelta = self.revlog.index.uses_generaldelta(); | |
|
485 | while let Some(base_rev) = entry.base_rev_or_base_of_delta_chain { | |
|
486 | let base_rev = if uses_generaldelta { | |
|
487 | base_rev | |
|
488 | } else { | |
|
489 | entry.rev - 1 | |
|
490 | }; | |
|
491 | delta_chain.push(entry); | |
|
492 | entry = self.revlog.get_entry_internal(base_rev)?; | |
|
493 | } | |
|
494 | ||
|
495 | let data = if delta_chain.is_empty() { | |
|
496 | entry.data_chunk()? | |
|
497 | } else { | |
|
498 | Revlog::build_data_from_deltas(entry, &delta_chain)?.into() | |
|
499 | }; | |
|
500 | ||
|
501 | Ok(data) | |
|
502 | } | |
|
503 | ||
|
504 | fn check_data( | |
|
505 | &self, | |
|
506 | data: Cow<'a, [u8]>, | |
|
507 | ) -> Result<Cow<'a, [u8]>, HgError> { | |
|
508 | if self.revlog.check_hash( | |
|
509 | self.p1, | |
|
510 | self.p2, | |
|
511 | self.hash.as_bytes(), | |
|
512 | &data, | |
|
513 | ) { | |
|
514 | Ok(data) | |
|
515 | } else { | |
|
516 | if (self.flags & REVISION_FLAG_ELLIPSIS) != 0 { | |
|
517 | return Err(HgError::unsupported( | |
|
518 | "ellipsis revisions are not supported by rhg", | |
|
519 | )); | |
|
520 | } | |
|
521 | Err(corrupted(format!( | |
|
522 | "hash check failed for revision {}", | |
|
523 | self.rev | |
|
524 | ))) | |
|
525 | } | |
|
526 | } | |
|
527 | ||
|
528 | pub fn data(&self) -> Result<Cow<'a, [u8]>, HgError> { | |
|
529 | let data = self.rawdata()?; | |
|
530 | if self.is_censored() { | |
|
531 | return Err(HgError::CensoredNodeError); | |
|
532 | } | |
|
533 | self.check_data(data) | |
|
534 | } | |
|
535 | ||
|
536 | /// Extract the data contained in the entry. | |
|
537 | /// This may be a delta. (See `is_delta`.) | |
|
538 | fn data_chunk(&self) -> Result<Cow<'a, [u8]>, HgError> { | |
|
539 | if self.bytes.is_empty() { | |
|
540 | return Ok(Cow::Borrowed(&[])); | |
|
541 | } | |
|
542 | match self.bytes[0] { | |
|
543 | // Revision data is the entirety of the entry, including this | |
|
544 | // header. | |
|
545 | b'\0' => Ok(Cow::Borrowed(self.bytes)), | |
|
546 | // Raw revision data follows. | |
|
547 | b'u' => Ok(Cow::Borrowed(&self.bytes[1..])), | |
|
548 | // zlib (RFC 1950) data. | |
|
549 | b'x' => Ok(Cow::Owned(self.uncompressed_zlib_data()?)), | |
|
550 | // zstd data. | |
|
551 | b'\x28' => Ok(Cow::Owned(self.uncompressed_zstd_data()?)), | |
|
552 | // A proper new format should have had a repo/store requirement. | |
|
553 | format_type => Err(corrupted(format!( | |
|
554 | "unknown compression header '{}'", | |
|
555 | format_type | |
|
556 | ))), | |
|
557 | } | |
|
558 | } | |
|
559 | ||
|
560 | fn uncompressed_zlib_data(&self) -> Result<Vec<u8>, HgError> { | |
|
561 | let mut decoder = ZlibDecoder::new(self.bytes); | |
|
562 | if self.is_delta() { | |
|
563 | let mut buf = Vec::with_capacity(self.compressed_len as usize); | |
|
564 | decoder | |
|
565 | .read_to_end(&mut buf) | |
|
566 | .map_err(|e| corrupted(e.to_string()))?; | |
|
567 | Ok(buf) | |
|
568 | } else { | |
|
569 | let cap = self.uncompressed_len.max(0) as usize; | |
|
570 | let mut buf = vec![0; cap]; | |
|
571 | decoder | |
|
572 | .read_exact(&mut buf) | |
|
573 | .map_err(|e| corrupted(e.to_string()))?; | |
|
574 | Ok(buf) | |
|
575 | } | |
|
576 | } | |
|
577 | ||
|
578 | fn uncompressed_zstd_data(&self) -> Result<Vec<u8>, HgError> { | |
|
579 | if self.is_delta() { | |
|
580 | let mut buf = Vec::with_capacity(self.compressed_len as usize); | |
|
581 | zstd::stream::copy_decode(self.bytes, &mut buf) | |
|
582 | .map_err(|e| corrupted(e.to_string()))?; | |
|
583 | Ok(buf) | |
|
584 | } else { | |
|
585 | let cap = self.uncompressed_len.max(0) as usize; | |
|
586 | let mut buf = vec![0; cap]; | |
|
587 | let len = zstd::bulk::decompress_to_buffer(self.bytes, &mut buf) | |
|
588 | .map_err(|e| corrupted(e.to_string()))?; | |
|
589 | if len != self.uncompressed_len as usize { | |
|
590 | Err(corrupted("uncompressed length does not match")) | |
|
591 | } else { | |
|
592 | Ok(buf) | |
|
593 | } | |
|
594 | } | |
|
595 | } | |
|
596 | ||
|
597 | /// Tell if the entry is a snapshot or a delta | |
|
598 | /// (influences on decompression). | |
|
599 | fn is_delta(&self) -> bool { | |
|
600 | self.base_rev_or_base_of_delta_chain.is_some() | |
|
601 | } | |
|
602 | } | |
|
603 | ||
|
604 | /// Calculate the hash of a revision given its data and its parents. | |
|
605 | fn hash( | |
|
606 | data: &[u8], | |
|
607 | p1_hash: &[u8], | |
|
608 | p2_hash: &[u8], | |
|
609 | ) -> [u8; NODE_BYTES_LENGTH] { | |
|
610 | let mut hasher = Sha1::new(); | |
|
611 | let (a, b) = (p1_hash, p2_hash); | |
|
612 | if a > b { | |
|
613 | hasher.update(b); | |
|
614 | hasher.update(a); | |
|
615 | } else { | |
|
616 | hasher.update(a); | |
|
617 | hasher.update(b); | |
|
618 | } | |
|
619 | hasher.update(data); | |
|
620 | *hasher.finalize().as_ref() | |
|
621 | } | |
|
622 | ||
|
623 | #[cfg(test)] | |
|
624 | mod tests { | |
|
625 | use super::*; | |
|
626 | use crate::index::{IndexEntryBuilder, INDEX_ENTRY_SIZE}; | |
|
627 | use itertools::Itertools; | |
|
628 | ||
|
629 | #[test] | |
|
630 | fn test_empty() { | |
|
631 | let temp = tempfile::tempdir().unwrap(); | |
|
632 | let vfs = Vfs { base: temp.path() }; | |
|
633 | std::fs::write(temp.path().join("foo.i"), b"").unwrap(); | |
|
634 | let revlog = Revlog::open(&vfs, "foo.i", None, false).unwrap(); | |
|
635 | assert!(revlog.is_empty()); | |
|
636 | assert_eq!(revlog.len(), 0); | |
|
637 | assert!(revlog.get_entry(0).is_err()); | |
|
638 | assert!(!revlog.has_rev(0)); | |
|
639 | } | |
|
640 | ||
|
641 | #[test] | |
|
642 | fn test_inline() { | |
|
643 | let temp = tempfile::tempdir().unwrap(); | |
|
644 | let vfs = Vfs { base: temp.path() }; | |
|
645 | let node0 = Node::from_hex("2ed2a3912a0b24502043eae84ee4b279c18b90dd") | |
|
646 | .unwrap(); | |
|
647 | let node1 = Node::from_hex("b004912a8510032a0350a74daa2803dadfb00e12") | |
|
648 | .unwrap(); | |
|
649 | let node2 = Node::from_hex("dd6ad206e907be60927b5a3117b97dffb2590582") | |
|
650 | .unwrap(); | |
|
651 | let entry0_bytes = IndexEntryBuilder::new() | |
|
652 | .is_first(true) | |
|
653 | .with_version(1) | |
|
654 | .with_inline(true) | |
|
655 | .with_offset(INDEX_ENTRY_SIZE) | |
|
656 | .with_node(node0) | |
|
657 | .build(); | |
|
658 | let entry1_bytes = IndexEntryBuilder::new() | |
|
659 | .with_offset(INDEX_ENTRY_SIZE) | |
|
660 | .with_node(node1) | |
|
661 | .build(); | |
|
662 | let entry2_bytes = IndexEntryBuilder::new() | |
|
663 | .with_offset(INDEX_ENTRY_SIZE) | |
|
664 | .with_p1(0) | |
|
665 | .with_p2(1) | |
|
666 | .with_node(node2) | |
|
667 | .build(); | |
|
668 | let contents = vec![entry0_bytes, entry1_bytes, entry2_bytes] | |
|
669 | .into_iter() | |
|
670 | .flatten() | |
|
671 | .collect_vec(); | |
|
672 | std::fs::write(temp.path().join("foo.i"), contents).unwrap(); | |
|
673 | let revlog = Revlog::open(&vfs, "foo.i", None, false).unwrap(); | |
|
674 | ||
|
675 | let entry0 = revlog.get_entry(0).ok().unwrap(); | |
|
676 | assert_eq!(entry0.revision(), 0); | |
|
677 | assert_eq!(*entry0.node(), node0); | |
|
678 | assert!(!entry0.has_p1()); | |
|
679 | assert_eq!(entry0.p1(), None); | |
|
680 | assert_eq!(entry0.p2(), None); | |
|
681 | let p1_entry = entry0.p1_entry().unwrap(); | |
|
682 | assert!(p1_entry.is_none()); | |
|
683 | let p2_entry = entry0.p2_entry().unwrap(); | |
|
684 | assert!(p2_entry.is_none()); | |
|
685 | ||
|
686 | let entry1 = revlog.get_entry(1).ok().unwrap(); | |
|
687 | assert_eq!(entry1.revision(), 1); | |
|
688 | assert_eq!(*entry1.node(), node1); | |
|
689 | assert!(!entry1.has_p1()); | |
|
690 | assert_eq!(entry1.p1(), None); | |
|
691 | assert_eq!(entry1.p2(), None); | |
|
692 | let p1_entry = entry1.p1_entry().unwrap(); | |
|
693 | assert!(p1_entry.is_none()); | |
|
694 | let p2_entry = entry1.p2_entry().unwrap(); | |
|
695 | assert!(p2_entry.is_none()); | |
|
696 | ||
|
697 | let entry2 = revlog.get_entry(2).ok().unwrap(); | |
|
698 | assert_eq!(entry2.revision(), 2); | |
|
699 | assert_eq!(*entry2.node(), node2); | |
|
700 | assert!(entry2.has_p1()); | |
|
701 | assert_eq!(entry2.p1(), Some(0)); | |
|
702 | assert_eq!(entry2.p2(), Some(1)); | |
|
703 | let p1_entry = entry2.p1_entry().unwrap(); | |
|
704 | assert!(p1_entry.is_some()); | |
|
705 | assert_eq!(p1_entry.unwrap().revision(), 0); | |
|
706 | let p2_entry = entry2.p2_entry().unwrap(); | |
|
707 | assert!(p2_entry.is_some()); | |
|
708 | assert_eq!(p2_entry.unwrap().revision(), 1); | |
|
709 | } | |
|
710 | } |
@@ -4,9 +4,9 b'' | |||
|
4 | 4 | |
|
5 | 5 | use crate::errors::HgError; |
|
6 | 6 | use crate::repo::Repo; |
|
7 | use crate::revlog::revlog::{Revlog, RevlogError}; | |
|
8 | 7 | use crate::revlog::NodePrefix; |
|
9 | 8 | use crate::revlog::{Revision, NULL_REVISION, WORKING_DIRECTORY_HEX}; |
|
9 | use crate::revlog::{Revlog, RevlogError}; | |
|
10 | 10 | use crate::Node; |
|
11 | 11 | |
|
12 | 12 | /// Resolve a query string into a single revision. |
@@ -7,7 +7,7 b' use hg::dirstate_tree::on_disk::Dirstate' | |||
|
7 | 7 | use hg::errors::HgError; |
|
8 | 8 | use hg::exit_codes; |
|
9 | 9 | use hg::repo::RepoError; |
|
10 |
use hg::revlog:: |
|
|
10 | use hg::revlog::RevlogError; | |
|
11 | 11 | use hg::sparse::SparseConfigError; |
|
12 | 12 | use hg::utils::files::get_bytes_from_path; |
|
13 | 13 | use hg::{DirstateError, DirstateMapError, StatusError}; |
General Comments 0
You need to be logged in to leave comments.
Login now