Show More
@@ -1,158 +1,169 b'' | |||
|
1 | 1 | // list_tracked_files.rs |
|
2 | 2 | // |
|
3 | 3 | // Copyright 2020 Antoine Cezar <antoine.cezar@octobus.net> |
|
4 | 4 | // |
|
5 | 5 | // This software may be used and distributed according to the terms of the |
|
6 | 6 | // GNU General Public License version 2 or any later version. |
|
7 | 7 | |
|
8 | 8 | use std::convert::From; |
|
9 | use std::path::PathBuf; | |
|
9 | use std::path::{Path, PathBuf}; | |
|
10 | 10 | |
|
11 | 11 | use crate::revlog::changelog::Changelog; |
|
12 | 12 | use crate::revlog::manifest::{Manifest, ManifestEntry}; |
|
13 | 13 | use crate::revlog::path_encode::path_encode; |
|
14 | 14 | use crate::revlog::revlog::Revlog; |
|
15 | 15 | use crate::revlog::revlog::RevlogError; |
|
16 | 16 | use crate::revlog::Revision; |
|
17 |
use crate::utils:: |
|
|
17 | use crate::utils::files::get_path_from_bytes; | |
|
18 | use crate::utils::hg_path::{HgPath, HgPathBuf}; | |
|
18 | 19 | |
|
19 | 20 | const METADATA_DELIMITER: [u8; 2] = [b'\x01', b'\n']; |
|
20 | 21 | |
|
21 | 22 | /// Kind of error encountered by `CatRev` |
|
22 | 23 | #[derive(Debug)] |
|
23 | 24 | pub enum CatRevErrorKind { |
|
24 | 25 | /// Error when reading a `revlog` file. |
|
25 | 26 | IoError(std::io::Error), |
|
26 | 27 | /// The revision has not been found. |
|
27 | 28 | InvalidRevision, |
|
28 | 29 | /// A `revlog` file is corrupted. |
|
29 | 30 | CorruptedRevlog, |
|
30 | 31 | /// The `revlog` format version is not supported. |
|
31 | 32 | UnsuportedRevlogVersion(u16), |
|
32 | 33 | /// The `revlog` data format is not supported. |
|
33 | 34 | UnknowRevlogDataFormat(u8), |
|
34 | 35 | } |
|
35 | 36 | |
|
36 | 37 | /// A `CatRev` error |
|
37 | 38 | #[derive(Debug)] |
|
38 | 39 | pub struct CatRevError { |
|
39 | 40 | /// Kind of error encountered by `CatRev` |
|
40 | 41 | pub kind: CatRevErrorKind, |
|
41 | 42 | } |
|
42 | 43 | |
|
43 | 44 | impl From<CatRevErrorKind> for CatRevError { |
|
44 | 45 | fn from(kind: CatRevErrorKind) -> Self { |
|
45 | 46 | CatRevError { kind } |
|
46 | 47 | } |
|
47 | 48 | } |
|
48 | 49 | |
|
49 | 50 | impl From<RevlogError> for CatRevError { |
|
50 | 51 | fn from(err: RevlogError) -> Self { |
|
51 | 52 | match err { |
|
52 | 53 | RevlogError::IoError(err) => CatRevErrorKind::IoError(err), |
|
53 | 54 | RevlogError::UnsuportedVersion(version) => { |
|
54 | 55 | CatRevErrorKind::UnsuportedRevlogVersion(version) |
|
55 | 56 | } |
|
56 | 57 | RevlogError::InvalidRevision => CatRevErrorKind::InvalidRevision, |
|
57 | 58 | RevlogError::Corrupted => CatRevErrorKind::CorruptedRevlog, |
|
58 | 59 | RevlogError::UnknowDataFormat(format) => { |
|
59 | 60 | CatRevErrorKind::UnknowRevlogDataFormat(format) |
|
60 | 61 | } |
|
61 | 62 | } |
|
62 | 63 | .into() |
|
63 | 64 | } |
|
64 | 65 | } |
|
65 | 66 | |
|
66 | 67 | /// List files under Mercurial control at a given revision. |
|
67 | 68 | pub struct CatRev<'a> { |
|
68 | 69 | root: &'a PathBuf, |
|
69 | 70 | /// The revision to cat the files from. |
|
70 | 71 | rev: &'a str, |
|
71 | 72 | /// The files to output. |
|
72 | 73 | files: &'a [HgPathBuf], |
|
73 | 74 | /// The changelog file |
|
74 | 75 | changelog: Changelog, |
|
75 | 76 | /// The manifest file |
|
76 | 77 | manifest: Manifest, |
|
77 | 78 | /// The manifest entry corresponding to the revision. |
|
78 | 79 | /// |
|
79 | 80 | /// Used to hold the owner of the returned references. |
|
80 | 81 | manifest_entry: Option<ManifestEntry>, |
|
81 | 82 | } |
|
82 | 83 | |
|
83 | 84 | impl<'a> CatRev<'a> { |
|
84 | 85 | pub fn new( |
|
85 | 86 | root: &'a PathBuf, |
|
86 | 87 | rev: &'a str, |
|
87 | 88 | files: &'a [HgPathBuf], |
|
88 | 89 | ) -> Result<Self, CatRevError> { |
|
89 | 90 | let changelog = Changelog::open(&root)?; |
|
90 | 91 | let manifest = Manifest::open(&root)?; |
|
91 | 92 | let manifest_entry = None; |
|
92 | 93 | |
|
93 | 94 | Ok(Self { |
|
94 | 95 | root, |
|
95 | 96 | rev, |
|
96 | 97 | files, |
|
97 | 98 | changelog, |
|
98 | 99 | manifest, |
|
99 | 100 | manifest_entry, |
|
100 | 101 | }) |
|
101 | 102 | } |
|
102 | 103 | |
|
103 | 104 | pub fn run(&mut self) -> Result<Vec<u8>, CatRevError> { |
|
104 | 105 | let changelog_entry = match self.rev.parse::<Revision>() { |
|
105 | 106 | Ok(rev) => self.changelog.get_rev(rev)?, |
|
106 | 107 | _ => { |
|
107 | 108 | let changelog_node = hex::decode(&self.rev) |
|
108 | 109 | .map_err(|_| CatRevErrorKind::InvalidRevision)?; |
|
109 | 110 | self.changelog.get_node(&changelog_node)? |
|
110 | 111 | } |
|
111 | 112 | }; |
|
112 | 113 | let manifest_node = hex::decode(&changelog_entry.manifest_node()?) |
|
113 | 114 | .map_err(|_| CatRevErrorKind::CorruptedRevlog)?; |
|
114 | 115 | |
|
115 | 116 | self.manifest_entry = Some(self.manifest.get_node(&manifest_node)?); |
|
116 | 117 | if let Some(ref manifest_entry) = self.manifest_entry { |
|
117 | 118 | let mut bytes = vec![]; |
|
118 | 119 | |
|
119 | 120 | for (manifest_file, node_bytes) in |
|
120 | 121 | manifest_entry.files_with_nodes() |
|
121 | 122 | { |
|
122 | 123 | for cat_file in self.files.iter() { |
|
123 | 124 | if cat_file.as_bytes() == manifest_file.as_bytes() { |
|
124 |
let |
|
|
125 |
path |
|
|
126 |
let |
|
|
127 |
" |
|
|
128 | String::from_utf8_lossy(&encoded_bytes), | |
|
129 |
|
|
|
130 |
let |
|
|
131 | self.root.join(&revlog_index_string); | |
|
132 | let file_log = Revlog::open(&revlog_index_path)?; | |
|
125 | let index_path = | |
|
126 | store_path(self.root, manifest_file, b".i"); | |
|
127 | let data_path = | |
|
128 | store_path(self.root, manifest_file, b".d"); | |
|
129 | ||
|
130 | let file_log = | |
|
131 | Revlog::open(&index_path, Some(&data_path))?; | |
|
133 | 132 | let file_node = hex::decode(&node_bytes) |
|
134 | 133 | .map_err(|_| CatRevErrorKind::CorruptedRevlog)?; |
|
135 | 134 | let file_rev = file_log.get_node_rev(&file_node)?; |
|
136 | 135 | let data = file_log.get_rev_data(file_rev)?; |
|
137 | 136 | if data.starts_with(&METADATA_DELIMITER) { |
|
138 | 137 | let end_delimiter_position = data |
|
139 | 138 | [METADATA_DELIMITER.len()..] |
|
140 | 139 | .windows(METADATA_DELIMITER.len()) |
|
141 | 140 | .position(|bytes| bytes == METADATA_DELIMITER); |
|
142 | 141 | if let Some(position) = end_delimiter_position { |
|
143 | 142 | let offset = METADATA_DELIMITER.len() * 2; |
|
144 | 143 | bytes.extend(data[position + offset..].iter()); |
|
145 | 144 | } |
|
146 | 145 | } else { |
|
147 | 146 | bytes.extend(data); |
|
148 | 147 | } |
|
149 | 148 | } |
|
150 | 149 | } |
|
151 | 150 | } |
|
152 | 151 | |
|
153 | 152 | Ok(bytes) |
|
154 | 153 | } else { |
|
155 | 154 | unreachable!("manifest_entry should have been stored"); |
|
156 | 155 | } |
|
157 | 156 | } |
|
158 | 157 | } |
|
158 | ||
|
159 | fn store_path(root: &Path, hg_path: &HgPath, suffix: &[u8]) -> PathBuf { | |
|
160 | let encoded_bytes = | |
|
161 | path_encode(&[b"data/", hg_path.as_bytes(), suffix].concat()); | |
|
162 | [ | |
|
163 | root, | |
|
164 | &Path::new(".hg/store/"), | |
|
165 | get_path_from_bytes(&encoded_bytes), | |
|
166 | ] | |
|
167 | .iter() | |
|
168 | .collect() | |
|
169 | } |
@@ -1,110 +1,110 b'' | |||
|
1 | 1 | // debugdata.rs |
|
2 | 2 | // |
|
3 | 3 | // Copyright 2020 Antoine Cezar <antoine.cezar@octobus.net> |
|
4 | 4 | // |
|
5 | 5 | // This software may be used and distributed according to the terms of the |
|
6 | 6 | // GNU General Public License version 2 or any later version. |
|
7 | 7 | |
|
8 | 8 | use super::find_root; |
|
9 | 9 | use crate::revlog::revlog::{Revlog, RevlogError}; |
|
10 | 10 | use crate::revlog::Revision; |
|
11 | 11 | |
|
12 | 12 | /// Kind of data to debug |
|
13 | 13 | #[derive(Debug, Copy, Clone)] |
|
14 | 14 | pub enum DebugDataKind { |
|
15 | 15 | Changelog, |
|
16 | 16 | Manifest, |
|
17 | 17 | } |
|
18 | 18 | |
|
19 | 19 | /// Kind of error encountered by DebugData |
|
20 | 20 | #[derive(Debug)] |
|
21 | 21 | pub enum DebugDataErrorKind { |
|
22 | 22 | FindRootError(find_root::FindRootError), |
|
23 | 23 | /// Error when reading a `revlog` file. |
|
24 | 24 | IoError(std::io::Error), |
|
25 | 25 | /// The revision has not been found. |
|
26 | 26 | InvalidRevision, |
|
27 | 27 | /// A `revlog` file is corrupted. |
|
28 | 28 | CorruptedRevlog, |
|
29 | 29 | /// The `revlog` format version is not supported. |
|
30 | 30 | UnsuportedRevlogVersion(u16), |
|
31 | 31 | /// The `revlog` data format is not supported. |
|
32 | 32 | UnknowRevlogDataFormat(u8), |
|
33 | 33 | } |
|
34 | 34 | |
|
35 | 35 | /// A DebugData error |
|
36 | 36 | #[derive(Debug)] |
|
37 | 37 | pub struct DebugDataError { |
|
38 | 38 | /// Kind of error encountered by DebugData |
|
39 | 39 | pub kind: DebugDataErrorKind, |
|
40 | 40 | } |
|
41 | 41 | |
|
42 | 42 | impl From<DebugDataErrorKind> for DebugDataError { |
|
43 | 43 | fn from(kind: DebugDataErrorKind) -> Self { |
|
44 | 44 | DebugDataError { kind } |
|
45 | 45 | } |
|
46 | 46 | } |
|
47 | 47 | |
|
48 | 48 | impl From<find_root::FindRootError> for DebugDataError { |
|
49 | 49 | fn from(err: find_root::FindRootError) -> Self { |
|
50 | 50 | let kind = DebugDataErrorKind::FindRootError(err); |
|
51 | 51 | DebugDataError { kind } |
|
52 | 52 | } |
|
53 | 53 | } |
|
54 | 54 | |
|
55 | 55 | impl From<std::io::Error> for DebugDataError { |
|
56 | 56 | fn from(err: std::io::Error) -> Self { |
|
57 | 57 | let kind = DebugDataErrorKind::IoError(err); |
|
58 | 58 | DebugDataError { kind } |
|
59 | 59 | } |
|
60 | 60 | } |
|
61 | 61 | |
|
62 | 62 | impl From<RevlogError> for DebugDataError { |
|
63 | 63 | fn from(err: RevlogError) -> Self { |
|
64 | 64 | match err { |
|
65 | 65 | RevlogError::IoError(err) => DebugDataErrorKind::IoError(err), |
|
66 | 66 | RevlogError::UnsuportedVersion(version) => { |
|
67 | 67 | DebugDataErrorKind::UnsuportedRevlogVersion(version) |
|
68 | 68 | } |
|
69 | 69 | RevlogError::InvalidRevision => { |
|
70 | 70 | DebugDataErrorKind::InvalidRevision |
|
71 | 71 | } |
|
72 | 72 | RevlogError::Corrupted => DebugDataErrorKind::CorruptedRevlog, |
|
73 | 73 | RevlogError::UnknowDataFormat(format) => { |
|
74 | 74 | DebugDataErrorKind::UnknowRevlogDataFormat(format) |
|
75 | 75 | } |
|
76 | 76 | } |
|
77 | 77 | .into() |
|
78 | 78 | } |
|
79 | 79 | } |
|
80 | 80 | |
|
81 | 81 | /// Dump the contents data of a revision. |
|
82 | 82 | pub struct DebugData<'a> { |
|
83 | 83 | /// Revision or hash of the revision. |
|
84 | 84 | rev: &'a str, |
|
85 | 85 | /// Kind of data to debug. |
|
86 | 86 | kind: DebugDataKind, |
|
87 | 87 | } |
|
88 | 88 | |
|
89 | 89 | impl<'a> DebugData<'a> { |
|
90 | 90 | pub fn new(rev: &'a str, kind: DebugDataKind) -> Self { |
|
91 | 91 | DebugData { rev, kind } |
|
92 | 92 | } |
|
93 | 93 | |
|
94 | 94 | pub fn run(&mut self) -> Result<Vec<u8>, DebugDataError> { |
|
95 | 95 | let rev = self |
|
96 | 96 | .rev |
|
97 | 97 | .parse::<Revision>() |
|
98 | 98 | .or(Err(DebugDataErrorKind::InvalidRevision))?; |
|
99 | 99 | |
|
100 | 100 | let root = find_root::FindRoot::new().run()?; |
|
101 | 101 | let index_file = match self.kind { |
|
102 | 102 | DebugDataKind::Changelog => root.join(".hg/store/00changelog.i"), |
|
103 | 103 | DebugDataKind::Manifest => root.join(".hg/store/00manifest.i"), |
|
104 | 104 | }; |
|
105 | let revlog = Revlog::open(&index_file)?; | |
|
105 | let revlog = Revlog::open(&index_file, None)?; | |
|
106 | 106 | let data = revlog.get_rev_data(rev)?; |
|
107 | 107 | |
|
108 | 108 | Ok(data) |
|
109 | 109 | } |
|
110 | 110 | } |
@@ -1,58 +1,58 b'' | |||
|
1 | 1 | use crate::revlog::revlog::{Revlog, RevlogError}; |
|
2 | 2 | use crate::revlog::Revision; |
|
3 | 3 | use std::path::PathBuf; |
|
4 | 4 | |
|
5 | 5 | /// A specialized `Revlog` to work with `changelog` data format. |
|
6 | 6 | pub struct Changelog { |
|
7 | 7 | /// The generic `revlog` format. |
|
8 | 8 | revlog: Revlog, |
|
9 | 9 | } |
|
10 | 10 | |
|
11 | 11 | impl Changelog { |
|
12 | 12 | /// Open the `changelog` of a repository given by its root. |
|
13 | 13 | pub fn open(root: &PathBuf) -> Result<Self, RevlogError> { |
|
14 | 14 | let index_file = root.join(".hg/store/00changelog.i"); |
|
15 | let revlog = Revlog::open(&index_file)?; | |
|
15 | let revlog = Revlog::open(&index_file, None)?; | |
|
16 | 16 | Ok(Self { revlog }) |
|
17 | 17 | } |
|
18 | 18 | |
|
19 | 19 | /// Return the `ChangelogEntry` a given node id. |
|
20 | 20 | pub fn get_node( |
|
21 | 21 | &self, |
|
22 | 22 | node: &[u8], |
|
23 | 23 | ) -> Result<ChangelogEntry, RevlogError> { |
|
24 | 24 | let rev = self.revlog.get_node_rev(node)?; |
|
25 | 25 | self.get_rev(rev) |
|
26 | 26 | } |
|
27 | 27 | |
|
28 | 28 | /// Return the `ChangelogEntry` of a given node revision. |
|
29 | 29 | pub fn get_rev( |
|
30 | 30 | &self, |
|
31 | 31 | rev: Revision, |
|
32 | 32 | ) -> Result<ChangelogEntry, RevlogError> { |
|
33 | 33 | let bytes = self.revlog.get_rev_data(rev)?; |
|
34 | 34 | Ok(ChangelogEntry { bytes }) |
|
35 | 35 | } |
|
36 | 36 | } |
|
37 | 37 | |
|
38 | 38 | /// `Changelog` entry which knows how to interpret the `changelog` data bytes. |
|
39 | 39 | #[derive(Debug)] |
|
40 | 40 | pub struct ChangelogEntry { |
|
41 | 41 | /// The data bytes of the `changelog` entry. |
|
42 | 42 | bytes: Vec<u8>, |
|
43 | 43 | } |
|
44 | 44 | |
|
45 | 45 | impl ChangelogEntry { |
|
46 | 46 | /// Return an iterator over the lines of the entry. |
|
47 | 47 | pub fn lines(&self) -> impl Iterator<Item = &[u8]> { |
|
48 | 48 | self.bytes |
|
49 | 49 | .split(|b| b == &b'\n') |
|
50 | 50 | .filter(|line| !line.is_empty()) |
|
51 | 51 | } |
|
52 | 52 | |
|
53 | 53 | /// Return the node id of the `manifest` referenced by this `changelog` |
|
54 | 54 | /// entry. |
|
55 | 55 | pub fn manifest_node(&self) -> Result<&[u8], RevlogError> { |
|
56 | 56 | self.lines().next().ok_or(RevlogError::Corrupted) |
|
57 | 57 | } |
|
58 | 58 | } |
@@ -1,73 +1,73 b'' | |||
|
1 | 1 | use crate::revlog::revlog::{Revlog, RevlogError}; |
|
2 | 2 | use crate::revlog::Revision; |
|
3 | 3 | use crate::utils::hg_path::HgPath; |
|
4 | 4 | use std::path::PathBuf; |
|
5 | 5 | |
|
6 | 6 | /// A specialized `Revlog` to work with `manifest` data format. |
|
7 | 7 | pub struct Manifest { |
|
8 | 8 | /// The generic `revlog` format. |
|
9 | 9 | revlog: Revlog, |
|
10 | 10 | } |
|
11 | 11 | |
|
12 | 12 | impl Manifest { |
|
13 | 13 | /// Open the `manifest` of a repository given by its root. |
|
14 | 14 | pub fn open(root: &PathBuf) -> Result<Self, RevlogError> { |
|
15 | 15 | let index_file = root.join(".hg/store/00manifest.i"); |
|
16 | let revlog = Revlog::open(&index_file)?; | |
|
16 | let revlog = Revlog::open(&index_file, None)?; | |
|
17 | 17 | Ok(Self { revlog }) |
|
18 | 18 | } |
|
19 | 19 | |
|
20 | 20 | /// Return the `ManifestEntry` of a given node id. |
|
21 | 21 | pub fn get_node(&self, node: &[u8]) -> Result<ManifestEntry, RevlogError> { |
|
22 | 22 | let rev = self.revlog.get_node_rev(node)?; |
|
23 | 23 | self.get_rev(rev) |
|
24 | 24 | } |
|
25 | 25 | |
|
26 | 26 | /// Return the `ManifestEntry` of a given node revision. |
|
27 | 27 | pub fn get_rev( |
|
28 | 28 | &self, |
|
29 | 29 | rev: Revision, |
|
30 | 30 | ) -> Result<ManifestEntry, RevlogError> { |
|
31 | 31 | let bytes = self.revlog.get_rev_data(rev)?; |
|
32 | 32 | Ok(ManifestEntry { bytes }) |
|
33 | 33 | } |
|
34 | 34 | } |
|
35 | 35 | |
|
36 | 36 | /// `Manifest` entry which knows how to interpret the `manifest` data bytes. |
|
37 | 37 | #[derive(Debug)] |
|
38 | 38 | pub struct ManifestEntry { |
|
39 | 39 | bytes: Vec<u8>, |
|
40 | 40 | } |
|
41 | 41 | |
|
42 | 42 | impl ManifestEntry { |
|
43 | 43 | /// Return an iterator over the lines of the entry. |
|
44 | 44 | pub fn lines(&self) -> impl Iterator<Item = &[u8]> { |
|
45 | 45 | self.bytes |
|
46 | 46 | .split(|b| b == &b'\n') |
|
47 | 47 | .filter(|line| !line.is_empty()) |
|
48 | 48 | } |
|
49 | 49 | |
|
50 | 50 | /// Return an iterator over the files of the entry. |
|
51 | 51 | pub fn files(&self) -> impl Iterator<Item = &HgPath> { |
|
52 | 52 | self.lines().filter(|line| !line.is_empty()).map(|line| { |
|
53 | 53 | let pos = line |
|
54 | 54 | .iter() |
|
55 | 55 | .position(|x| x == &b'\0') |
|
56 | 56 | .expect("manifest line should contain \\0"); |
|
57 | 57 | HgPath::new(&line[..pos]) |
|
58 | 58 | }) |
|
59 | 59 | } |
|
60 | 60 | |
|
61 | 61 | /// Return an iterator over the files of the entry. |
|
62 | 62 | pub fn files_with_nodes(&self) -> impl Iterator<Item = (&HgPath, &[u8])> { |
|
63 | 63 | self.lines().filter(|line| !line.is_empty()).map(|line| { |
|
64 | 64 | let pos = line |
|
65 | 65 | .iter() |
|
66 | 66 | .position(|x| x == &b'\0') |
|
67 | 67 | .expect("manifest line should contain \\0"); |
|
68 | 68 | let hash_start = pos + 1; |
|
69 | 69 | let hash_end = hash_start + 40; |
|
70 | 70 | (HgPath::new(&line[..pos]), &line[hash_start..hash_end]) |
|
71 | 71 | }) |
|
72 | 72 | } |
|
73 | 73 | } |
@@ -1,331 +1,335 b'' | |||
|
1 | 1 | use std::borrow::Cow; |
|
2 | 2 | use std::fs::File; |
|
3 | 3 | use std::io::Read; |
|
4 | 4 | use std::ops::Deref; |
|
5 | 5 | use std::path::Path; |
|
6 | 6 | |
|
7 | 7 | use byteorder::{BigEndian, ByteOrder}; |
|
8 | 8 | use crypto::digest::Digest; |
|
9 | 9 | use crypto::sha1::Sha1; |
|
10 | 10 | use flate2::read::ZlibDecoder; |
|
11 | 11 | use memmap::{Mmap, MmapOptions}; |
|
12 | 12 | use micro_timer::timed; |
|
13 | 13 | use zstd; |
|
14 | 14 | |
|
15 | 15 | use super::index::Index; |
|
16 | 16 | use super::node::{NODE_BYTES_LENGTH, NULL_NODE_ID}; |
|
17 | 17 | use super::patch; |
|
18 | 18 | use crate::revlog::Revision; |
|
19 | 19 | |
|
20 | 20 | pub enum RevlogError { |
|
21 | 21 | IoError(std::io::Error), |
|
22 | 22 | UnsuportedVersion(u16), |
|
23 | 23 | InvalidRevision, |
|
24 | 24 | Corrupted, |
|
25 | 25 | UnknowDataFormat(u8), |
|
26 | 26 | } |
|
27 | 27 | |
|
28 | 28 | fn mmap_open(path: &Path) -> Result<Mmap, std::io::Error> { |
|
29 | 29 | let file = File::open(path)?; |
|
30 | 30 | let mmap = unsafe { MmapOptions::new().map(&file) }?; |
|
31 | 31 | Ok(mmap) |
|
32 | 32 | } |
|
33 | 33 | |
|
34 | 34 | /// Read only implementation of revlog. |
|
35 | 35 | pub struct Revlog { |
|
36 | 36 | /// When index and data are not interleaved: bytes of the revlog index. |
|
37 | 37 | /// When index and data are interleaved: bytes of the revlog index and |
|
38 | 38 | /// data. |
|
39 | 39 | index: Index, |
|
40 | 40 | /// When index and data are not interleaved: bytes of the revlog data |
|
41 | 41 | data_bytes: Option<Box<dyn Deref<Target = [u8]> + Send>>, |
|
42 | 42 | } |
|
43 | 43 | |
|
44 | 44 | impl Revlog { |
|
45 | 45 | /// Open a revlog index file. |
|
46 | 46 | /// |
|
47 | 47 | /// It will also open the associated data file if index and data are not |
|
48 | 48 | /// interleaved. |
|
49 | 49 | #[timed] |
|
50 | pub fn open(index_path: &Path) -> Result<Self, RevlogError> { | |
|
50 | pub fn open( | |
|
51 | index_path: &Path, | |
|
52 | data_path: Option<&Path>, | |
|
53 | ) -> Result<Self, RevlogError> { | |
|
51 | 54 | let index_mmap = |
|
52 | 55 | mmap_open(&index_path).map_err(RevlogError::IoError)?; |
|
53 | 56 | |
|
54 | 57 | let version = get_version(&index_mmap); |
|
55 | 58 | if version != 1 { |
|
56 | 59 | return Err(RevlogError::UnsuportedVersion(version)); |
|
57 | 60 | } |
|
58 | 61 | |
|
59 | 62 | let index = Index::new(Box::new(index_mmap))?; |
|
60 | 63 | |
|
61 | // TODO load data only when needed // | |
|
64 | let default_data_path = index_path.with_extension("d"); | |
|
65 | ||
|
62 | 66 | // type annotation required |
|
63 | 67 | // won't recognize Mmap as Deref<Target = [u8]> |
|
64 | 68 | let data_bytes: Option<Box<dyn Deref<Target = [u8]> + Send>> = |
|
65 | 69 | if index.is_inline() { |
|
66 | 70 | None |
|
67 | 71 | } else { |
|
68 |
let data_path = |
|
|
72 | let data_path = data_path.unwrap_or(&default_data_path); | |
|
69 | 73 | let data_mmap = |
|
70 |
mmap_open( |
|
|
74 | mmap_open(data_path).map_err(RevlogError::IoError)?; | |
|
71 | 75 | Some(Box::new(data_mmap)) |
|
72 | 76 | }; |
|
73 | 77 | |
|
74 | 78 | Ok(Revlog { index, data_bytes }) |
|
75 | 79 | } |
|
76 | 80 | |
|
77 | 81 | /// Return number of entries of the `Revlog`. |
|
78 | 82 | pub fn len(&self) -> usize { |
|
79 | 83 | self.index.len() |
|
80 | 84 | } |
|
81 | 85 | |
|
82 | 86 | /// Returns `true` if the `Revlog` has zero `entries`. |
|
83 | 87 | pub fn is_empty(&self) -> bool { |
|
84 | 88 | self.index.is_empty() |
|
85 | 89 | } |
|
86 | 90 | |
|
87 | 91 | /// Return the full data associated to a node. |
|
88 | 92 | #[timed] |
|
89 | 93 | pub fn get_node_rev(&self, node: &[u8]) -> Result<Revision, RevlogError> { |
|
90 | 94 | // This is brute force. But it is fast enough for now. |
|
91 | 95 | // Optimization will come later. |
|
92 | 96 | for rev in (0..self.len() as Revision).rev() { |
|
93 | 97 | let index_entry = |
|
94 | 98 | self.index.get_entry(rev).ok_or(RevlogError::Corrupted)?; |
|
95 | 99 | if node == index_entry.hash() { |
|
96 | 100 | return Ok(rev); |
|
97 | 101 | } |
|
98 | 102 | } |
|
99 | 103 | Err(RevlogError::InvalidRevision) |
|
100 | 104 | } |
|
101 | 105 | |
|
102 | 106 | /// Return the full data associated to a revision. |
|
103 | 107 | /// |
|
104 | 108 | /// All entries required to build the final data out of deltas will be |
|
105 | 109 | /// retrieved as needed, and the deltas will be applied to the inital |
|
106 | 110 | /// snapshot to rebuild the final data. |
|
107 | 111 | #[timed] |
|
108 | 112 | pub fn get_rev_data(&self, rev: Revision) -> Result<Vec<u8>, RevlogError> { |
|
109 | 113 | // Todo return -> Cow |
|
110 | 114 | let mut entry = self.get_entry(rev)?; |
|
111 | 115 | let mut delta_chain = vec![]; |
|
112 | 116 | while let Some(base_rev) = entry.base_rev { |
|
113 | 117 | delta_chain.push(entry); |
|
114 | 118 | entry = |
|
115 | 119 | self.get_entry(base_rev).or(Err(RevlogError::Corrupted))?; |
|
116 | 120 | } |
|
117 | 121 | |
|
118 | 122 | // TODO do not look twice in the index |
|
119 | 123 | let index_entry = self |
|
120 | 124 | .index |
|
121 | 125 | .get_entry(rev) |
|
122 | 126 | .ok_or(RevlogError::InvalidRevision)?; |
|
123 | 127 | |
|
124 | 128 | let data: Vec<u8> = if delta_chain.is_empty() { |
|
125 | 129 | entry.data()?.into() |
|
126 | 130 | } else { |
|
127 | 131 | Revlog::build_data_from_deltas(entry, &delta_chain)? |
|
128 | 132 | }; |
|
129 | 133 | |
|
130 | 134 | if self.check_hash( |
|
131 | 135 | index_entry.p1(), |
|
132 | 136 | index_entry.p2(), |
|
133 | 137 | index_entry.hash(), |
|
134 | 138 | &data, |
|
135 | 139 | ) { |
|
136 | 140 | Ok(data) |
|
137 | 141 | } else { |
|
138 | 142 | Err(RevlogError::Corrupted) |
|
139 | 143 | } |
|
140 | 144 | } |
|
141 | 145 | |
|
142 | 146 | /// Check the hash of some given data against the recorded hash. |
|
143 | 147 | pub fn check_hash( |
|
144 | 148 | &self, |
|
145 | 149 | p1: Revision, |
|
146 | 150 | p2: Revision, |
|
147 | 151 | expected: &[u8], |
|
148 | 152 | data: &[u8], |
|
149 | 153 | ) -> bool { |
|
150 | 154 | let e1 = self.index.get_entry(p1); |
|
151 | 155 | let h1 = match e1 { |
|
152 | 156 | Some(ref entry) => entry.hash(), |
|
153 | 157 | None => &NULL_NODE_ID, |
|
154 | 158 | }; |
|
155 | 159 | let e2 = self.index.get_entry(p2); |
|
156 | 160 | let h2 = match e2 { |
|
157 | 161 | Some(ref entry) => entry.hash(), |
|
158 | 162 | None => &NULL_NODE_ID, |
|
159 | 163 | }; |
|
160 | 164 | |
|
161 | 165 | hash(data, &h1, &h2).as_slice() == expected |
|
162 | 166 | } |
|
163 | 167 | |
|
164 | 168 | /// Build the full data of a revision out its snapshot |
|
165 | 169 | /// and its deltas. |
|
166 | 170 | #[timed] |
|
167 | 171 | fn build_data_from_deltas( |
|
168 | 172 | snapshot: RevlogEntry, |
|
169 | 173 | deltas: &[RevlogEntry], |
|
170 | 174 | ) -> Result<Vec<u8>, RevlogError> { |
|
171 | 175 | let snapshot = snapshot.data()?; |
|
172 | 176 | let deltas = deltas |
|
173 | 177 | .iter() |
|
174 | 178 | .rev() |
|
175 | 179 | .map(RevlogEntry::data) |
|
176 | 180 | .collect::<Result<Vec<Cow<'_, [u8]>>, RevlogError>>()?; |
|
177 | 181 | let patches: Vec<_> = |
|
178 | 182 | deltas.iter().map(|d| patch::PatchList::new(d)).collect(); |
|
179 | 183 | let patch = patch::fold_patch_lists(&patches); |
|
180 | 184 | Ok(patch.apply(&snapshot)) |
|
181 | 185 | } |
|
182 | 186 | |
|
183 | 187 | /// Return the revlog data. |
|
184 | 188 | fn data(&self) -> &[u8] { |
|
185 | 189 | match self.data_bytes { |
|
186 | 190 | Some(ref data_bytes) => &data_bytes, |
|
187 | 191 | None => panic!( |
|
188 | 192 | "forgot to load the data or trying to access inline data" |
|
189 | 193 | ), |
|
190 | 194 | } |
|
191 | 195 | } |
|
192 | 196 | |
|
193 | 197 | /// Get an entry of the revlog. |
|
194 | 198 | fn get_entry(&self, rev: Revision) -> Result<RevlogEntry, RevlogError> { |
|
195 | 199 | let index_entry = self |
|
196 | 200 | .index |
|
197 | 201 | .get_entry(rev) |
|
198 | 202 | .ok_or(RevlogError::InvalidRevision)?; |
|
199 | 203 | let start = index_entry.offset(); |
|
200 | 204 | let end = start + index_entry.compressed_len(); |
|
201 | 205 | let data = if self.index.is_inline() { |
|
202 | 206 | self.index.data(start, end) |
|
203 | 207 | } else { |
|
204 | 208 | &self.data()[start..end] |
|
205 | 209 | }; |
|
206 | 210 | let entry = RevlogEntry { |
|
207 | 211 | rev, |
|
208 | 212 | bytes: data, |
|
209 | 213 | compressed_len: index_entry.compressed_len(), |
|
210 | 214 | uncompressed_len: index_entry.uncompressed_len(), |
|
211 | 215 | base_rev: if index_entry.base_revision() == rev { |
|
212 | 216 | None |
|
213 | 217 | } else { |
|
214 | 218 | Some(index_entry.base_revision()) |
|
215 | 219 | }, |
|
216 | 220 | }; |
|
217 | 221 | Ok(entry) |
|
218 | 222 | } |
|
219 | 223 | } |
|
220 | 224 | |
|
221 | 225 | /// The revlog entry's bytes and the necessary informations to extract |
|
222 | 226 | /// the entry's data. |
|
223 | 227 | #[derive(Debug)] |
|
224 | 228 | pub struct RevlogEntry<'a> { |
|
225 | 229 | rev: Revision, |
|
226 | 230 | bytes: &'a [u8], |
|
227 | 231 | compressed_len: usize, |
|
228 | 232 | uncompressed_len: usize, |
|
229 | 233 | base_rev: Option<Revision>, |
|
230 | 234 | } |
|
231 | 235 | |
|
232 | 236 | impl<'a> RevlogEntry<'a> { |
|
233 | 237 | /// Extract the data contained in the entry. |
|
234 | 238 | pub fn data(&self) -> Result<Cow<'_, [u8]>, RevlogError> { |
|
235 | 239 | if self.bytes.is_empty() { |
|
236 | 240 | return Ok(Cow::Borrowed(&[])); |
|
237 | 241 | } |
|
238 | 242 | match self.bytes[0] { |
|
239 | 243 | // Revision data is the entirety of the entry, including this |
|
240 | 244 | // header. |
|
241 | 245 | b'\0' => Ok(Cow::Borrowed(self.bytes)), |
|
242 | 246 | // Raw revision data follows. |
|
243 | 247 | b'u' => Ok(Cow::Borrowed(&self.bytes[1..])), |
|
244 | 248 | // zlib (RFC 1950) data. |
|
245 | 249 | b'x' => Ok(Cow::Owned(self.uncompressed_zlib_data()?)), |
|
246 | 250 | // zstd data. |
|
247 | 251 | b'\x28' => Ok(Cow::Owned(self.uncompressed_zstd_data()?)), |
|
248 | 252 | format_type => Err(RevlogError::UnknowDataFormat(format_type)), |
|
249 | 253 | } |
|
250 | 254 | } |
|
251 | 255 | |
|
252 | 256 | fn uncompressed_zlib_data(&self) -> Result<Vec<u8>, RevlogError> { |
|
253 | 257 | let mut decoder = ZlibDecoder::new(self.bytes); |
|
254 | 258 | if self.is_delta() { |
|
255 | 259 | let mut buf = Vec::with_capacity(self.compressed_len); |
|
256 | 260 | decoder |
|
257 | 261 | .read_to_end(&mut buf) |
|
258 | 262 | .or(Err(RevlogError::Corrupted))?; |
|
259 | 263 | Ok(buf) |
|
260 | 264 | } else { |
|
261 | 265 | let mut buf = vec![0; self.uncompressed_len]; |
|
262 | 266 | decoder |
|
263 | 267 | .read_exact(&mut buf) |
|
264 | 268 | .or(Err(RevlogError::Corrupted))?; |
|
265 | 269 | Ok(buf) |
|
266 | 270 | } |
|
267 | 271 | } |
|
268 | 272 | |
|
269 | 273 | fn uncompressed_zstd_data(&self) -> Result<Vec<u8>, RevlogError> { |
|
270 | 274 | if self.is_delta() { |
|
271 | 275 | let mut buf = Vec::with_capacity(self.compressed_len); |
|
272 | 276 | zstd::stream::copy_decode(self.bytes, &mut buf) |
|
273 | 277 | .or(Err(RevlogError::Corrupted))?; |
|
274 | 278 | Ok(buf) |
|
275 | 279 | } else { |
|
276 | 280 | let mut buf = vec![0; self.uncompressed_len]; |
|
277 | 281 | let len = zstd::block::decompress_to_buffer(self.bytes, &mut buf) |
|
278 | 282 | .or(Err(RevlogError::Corrupted))?; |
|
279 | 283 | if len != self.uncompressed_len { |
|
280 | 284 | Err(RevlogError::Corrupted) |
|
281 | 285 | } else { |
|
282 | 286 | Ok(buf) |
|
283 | 287 | } |
|
284 | 288 | } |
|
285 | 289 | } |
|
286 | 290 | |
|
287 | 291 | /// Tell if the entry is a snapshot or a delta |
|
288 | 292 | /// (influences on decompression). |
|
289 | 293 | fn is_delta(&self) -> bool { |
|
290 | 294 | self.base_rev.is_some() |
|
291 | 295 | } |
|
292 | 296 | } |
|
293 | 297 | |
|
294 | 298 | /// Format version of the revlog. |
|
295 | 299 | pub fn get_version(index_bytes: &[u8]) -> u16 { |
|
296 | 300 | BigEndian::read_u16(&index_bytes[2..=3]) |
|
297 | 301 | } |
|
298 | 302 | |
|
299 | 303 | /// Calculate the hash of a revision given its data and its parents. |
|
300 | 304 | fn hash(data: &[u8], p1_hash: &[u8], p2_hash: &[u8]) -> Vec<u8> { |
|
301 | 305 | let mut hasher = Sha1::new(); |
|
302 | 306 | let (a, b) = (p1_hash, p2_hash); |
|
303 | 307 | if a > b { |
|
304 | 308 | hasher.input(b); |
|
305 | 309 | hasher.input(a); |
|
306 | 310 | } else { |
|
307 | 311 | hasher.input(a); |
|
308 | 312 | hasher.input(b); |
|
309 | 313 | } |
|
310 | 314 | hasher.input(data); |
|
311 | 315 | let mut hash = vec![0; NODE_BYTES_LENGTH]; |
|
312 | 316 | hasher.result(&mut hash); |
|
313 | 317 | hash |
|
314 | 318 | } |
|
315 | 319 | |
|
316 | 320 | #[cfg(test)] |
|
317 | 321 | mod tests { |
|
318 | 322 | use super::*; |
|
319 | 323 | |
|
320 | 324 | use super::super::index::IndexEntryBuilder; |
|
321 | 325 | |
|
322 | 326 | #[test] |
|
323 | 327 | fn version_test() { |
|
324 | 328 | let bytes = IndexEntryBuilder::new() |
|
325 | 329 | .is_first(true) |
|
326 | 330 | .with_version(1) |
|
327 | 331 | .build(); |
|
328 | 332 | |
|
329 | 333 | assert_eq!(get_version(&bytes), 1) |
|
330 | 334 | } |
|
331 | 335 | } |
General Comments 0
You need to be logged in to leave comments.
Login now