Show More
@@ -1,154 +1,154 b'' | |||
|
1 | 1 | // status.rs |
|
2 | 2 | // |
|
3 | 3 | // Copyright 2019 Raphaël Gomès <rgomes@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 | //! Rust implementation of dirstate.status (dirstate.py). |
|
9 | 9 | //! It is currently missing a lot of functionality compared to the Python one |
|
10 | 10 | //! and will only be triggered in narrow cases. |
|
11 | 11 | |
|
12 | use crate::dirstate::entry::TruncatedTimestamp; | |
|
12 | 13 | use crate::dirstate_tree::on_disk::DirstateV2ParseError; |
|
13 | ||
|
14 | 14 | use crate::{ |
|
15 | 15 | utils::hg_path::{HgPath, HgPathError}, |
|
16 | 16 | PatternError, |
|
17 | 17 | }; |
|
18 | 18 | |
|
19 | 19 | use std::{borrow::Cow, fmt}; |
|
20 | 20 | |
|
21 | 21 | /// Wrong type of file from a `BadMatch` |
|
22 | 22 | /// Note: a lot of those don't exist on all platforms. |
|
23 | 23 | #[derive(Debug, Copy, Clone)] |
|
24 | 24 | pub enum BadType { |
|
25 | 25 | CharacterDevice, |
|
26 | 26 | BlockDevice, |
|
27 | 27 | FIFO, |
|
28 | 28 | Socket, |
|
29 | 29 | Directory, |
|
30 | 30 | Unknown, |
|
31 | 31 | } |
|
32 | 32 | |
|
33 | 33 | impl fmt::Display for BadType { |
|
34 | 34 | fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { |
|
35 | 35 | f.write_str(match self { |
|
36 | 36 | BadType::CharacterDevice => "character device", |
|
37 | 37 | BadType::BlockDevice => "block device", |
|
38 | 38 | BadType::FIFO => "fifo", |
|
39 | 39 | BadType::Socket => "socket", |
|
40 | 40 | BadType::Directory => "directory", |
|
41 | 41 | BadType::Unknown => "unknown", |
|
42 | 42 | }) |
|
43 | 43 | } |
|
44 | 44 | } |
|
45 | 45 | |
|
46 | 46 | /// Was explicitly matched but cannot be found/accessed |
|
47 | 47 | #[derive(Debug, Copy, Clone)] |
|
48 | 48 | pub enum BadMatch { |
|
49 | 49 | OsError(i32), |
|
50 | 50 | BadType(BadType), |
|
51 | 51 | } |
|
52 | 52 | |
|
53 | 53 | /// `Box<dyn Trait>` is syntactic sugar for `Box<dyn Trait + 'static>`, so add |
|
54 | 54 | /// an explicit lifetime here to not fight `'static` bounds "out of nowhere". |
|
55 | 55 | pub type IgnoreFnType<'a> = |
|
56 | 56 | Box<dyn for<'r> Fn(&'r HgPath) -> bool + Sync + 'a>; |
|
57 | 57 | |
|
58 | 58 | /// We have a good mix of owned (from directory traversal) and borrowed (from |
|
59 | 59 | /// the dirstate/explicit) paths, this comes up a lot. |
|
60 | 60 | pub type HgPathCow<'a> = Cow<'a, HgPath>; |
|
61 | 61 | |
|
62 | 62 | #[derive(Debug, Copy, Clone)] |
|
63 | 63 | pub struct StatusOptions { |
|
64 | 64 | /// Whether we are on a filesystem with UNIX-like exec flags |
|
65 | 65 | pub check_exec: bool, |
|
66 | 66 | pub list_clean: bool, |
|
67 | 67 | pub list_unknown: bool, |
|
68 | 68 | pub list_ignored: bool, |
|
69 | 69 | /// Whether to populate `StatusPath::copy_source` |
|
70 | 70 | pub list_copies: bool, |
|
71 | 71 | /// Whether to collect traversed dirs for applying a callback later. |
|
72 | 72 | /// Used by `hg purge` for example. |
|
73 | 73 | pub collect_traversed_dirs: bool, |
|
74 | 74 | } |
|
75 | 75 | |
|
76 | 76 | #[derive(Default)] |
|
77 | 77 | pub struct DirstateStatus<'a> { |
|
78 | 78 | /// The current time at the start of the `status()` algorithm, as measured |
|
79 | 79 | /// and possibly truncated by the filesystem. |
|
80 |
pub filesystem_time_at_status_start: Option< |
|
|
80 | pub filesystem_time_at_status_start: Option<TruncatedTimestamp>, | |
|
81 | 81 | |
|
82 | 82 | /// Tracked files whose contents have changed since the parent revision |
|
83 | 83 | pub modified: Vec<StatusPath<'a>>, |
|
84 | 84 | |
|
85 | 85 | /// Newly-tracked files that were not present in the parent |
|
86 | 86 | pub added: Vec<StatusPath<'a>>, |
|
87 | 87 | |
|
88 | 88 | /// Previously-tracked files that have been (re)moved with an hg command |
|
89 | 89 | pub removed: Vec<StatusPath<'a>>, |
|
90 | 90 | |
|
91 | 91 | /// (Still) tracked files that are missing, (re)moved with an non-hg |
|
92 | 92 | /// command |
|
93 | 93 | pub deleted: Vec<StatusPath<'a>>, |
|
94 | 94 | |
|
95 | 95 | /// Tracked files that are up to date with the parent. |
|
96 | 96 | /// Only pupulated if `StatusOptions::list_clean` is true. |
|
97 | 97 | pub clean: Vec<StatusPath<'a>>, |
|
98 | 98 | |
|
99 | 99 | /// Files in the working directory that are ignored with `.hgignore`. |
|
100 | 100 | /// Only pupulated if `StatusOptions::list_ignored` is true. |
|
101 | 101 | pub ignored: Vec<StatusPath<'a>>, |
|
102 | 102 | |
|
103 | 103 | /// Files in the working directory that are neither tracked nor ignored. |
|
104 | 104 | /// Only pupulated if `StatusOptions::list_unknown` is true. |
|
105 | 105 | pub unknown: Vec<StatusPath<'a>>, |
|
106 | 106 | |
|
107 | 107 | /// Was explicitly matched but cannot be found/accessed |
|
108 | 108 | pub bad: Vec<(HgPathCow<'a>, BadMatch)>, |
|
109 | 109 | |
|
110 | 110 | /// Either clean or modified, but we can’t tell from filesystem metadata |
|
111 | 111 | /// alone. The file contents need to be read and compared with that in |
|
112 | 112 | /// the parent. |
|
113 | 113 | pub unsure: Vec<StatusPath<'a>>, |
|
114 | 114 | |
|
115 | 115 | /// Only filled if `collect_traversed_dirs` is `true` |
|
116 | 116 | pub traversed: Vec<HgPathCow<'a>>, |
|
117 | 117 | |
|
118 | 118 | /// Whether `status()` made changed to the `DirstateMap` that should be |
|
119 | 119 | /// written back to disk |
|
120 | 120 | pub dirty: bool, |
|
121 | 121 | } |
|
122 | 122 | |
|
123 | 123 | #[derive(Clone, PartialEq, Eq, PartialOrd, Ord)] |
|
124 | 124 | pub struct StatusPath<'a> { |
|
125 | 125 | pub path: HgPathCow<'a>, |
|
126 | 126 | pub copy_source: Option<HgPathCow<'a>>, |
|
127 | 127 | } |
|
128 | 128 | |
|
129 | 129 | #[derive(Debug, derive_more::From)] |
|
130 | 130 | pub enum StatusError { |
|
131 | 131 | /// Generic IO error |
|
132 | 132 | IO(std::io::Error), |
|
133 | 133 | /// An invalid path that cannot be represented in Mercurial was found |
|
134 | 134 | Path(HgPathError), |
|
135 | 135 | /// An invalid "ignore" pattern was found |
|
136 | 136 | Pattern(PatternError), |
|
137 | 137 | /// Corrupted dirstate |
|
138 | 138 | DirstateV2ParseError(DirstateV2ParseError), |
|
139 | 139 | } |
|
140 | 140 | |
|
141 | 141 | pub type StatusResult<T> = Result<T, StatusError>; |
|
142 | 142 | |
|
143 | 143 | impl fmt::Display for StatusError { |
|
144 | 144 | fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { |
|
145 | 145 | match self { |
|
146 | 146 | StatusError::IO(error) => error.fmt(f), |
|
147 | 147 | StatusError::Path(error) => error.fmt(f), |
|
148 | 148 | StatusError::Pattern(error) => error.fmt(f), |
|
149 | 149 | StatusError::DirstateV2ParseError(_) => { |
|
150 | 150 | f.write_str("dirstate-v2 parse error") |
|
151 | 151 | } |
|
152 | 152 | } |
|
153 | 153 | } |
|
154 | 154 | } |
@@ -1,837 +1,843 b'' | |||
|
1 | 1 | //! The "version 2" disk representation of the dirstate |
|
2 | 2 | //! |
|
3 | 3 | //! See `mercurial/helptext/internals/dirstate-v2.txt` |
|
4 | 4 | |
|
5 | 5 | use crate::dirstate::TruncatedTimestamp; |
|
6 | 6 | use crate::dirstate_tree::dirstate_map::{self, DirstateMap, NodeRef}; |
|
7 | 7 | use crate::dirstate_tree::path_with_basename::WithBasename; |
|
8 | 8 | use crate::errors::HgError; |
|
9 | 9 | use crate::utils::hg_path::HgPath; |
|
10 | 10 | use crate::DirstateEntry; |
|
11 | 11 | use crate::DirstateError; |
|
12 | 12 | use crate::DirstateParents; |
|
13 | 13 | use bitflags::bitflags; |
|
14 | 14 | use bytes_cast::unaligned::{U16Be, U32Be}; |
|
15 | 15 | use bytes_cast::BytesCast; |
|
16 | 16 | use format_bytes::format_bytes; |
|
17 | 17 | use rand::Rng; |
|
18 | 18 | use std::borrow::Cow; |
|
19 | 19 | use std::convert::{TryFrom, TryInto}; |
|
20 | 20 | use std::fmt::Write; |
|
21 | 21 | |
|
22 | 22 | /// Added at the start of `.hg/dirstate` when the "v2" format is used. |
|
23 | 23 | /// This a redundant sanity check more than an actual "magic number" since |
|
24 | 24 | /// `.hg/requires` already governs which format should be used. |
|
25 | 25 | pub const V2_FORMAT_MARKER: &[u8; 12] = b"dirstate-v2\n"; |
|
26 | 26 | |
|
27 | 27 | /// Keep space for 256-bit hashes |
|
28 | 28 | const STORED_NODE_ID_BYTES: usize = 32; |
|
29 | 29 | |
|
30 | 30 | /// … even though only 160 bits are used for now, with SHA-1 |
|
31 | 31 | const USED_NODE_ID_BYTES: usize = 20; |
|
32 | 32 | |
|
33 | 33 | pub(super) const IGNORE_PATTERNS_HASH_LEN: usize = 20; |
|
34 | 34 | pub(super) type IgnorePatternsHash = [u8; IGNORE_PATTERNS_HASH_LEN]; |
|
35 | 35 | |
|
36 | 36 | /// Must match constants of the same names in `mercurial/dirstateutils/v2.py` |
|
37 | 37 | const TREE_METADATA_SIZE: usize = 44; |
|
38 | 38 | const NODE_SIZE: usize = 44; |
|
39 | 39 | |
|
40 | 40 | /// Make sure that size-affecting changes are made knowingly |
|
41 | 41 | #[allow(unused)] |
|
42 | 42 | fn static_assert_size_of() { |
|
43 | 43 | let _ = std::mem::transmute::<TreeMetadata, [u8; TREE_METADATA_SIZE]>; |
|
44 | 44 | let _ = std::mem::transmute::<DocketHeader, [u8; TREE_METADATA_SIZE + 81]>; |
|
45 | 45 | let _ = std::mem::transmute::<Node, [u8; NODE_SIZE]>; |
|
46 | 46 | } |
|
47 | 47 | |
|
48 | 48 | // Must match `HEADER` in `mercurial/dirstateutils/docket.py` |
|
49 | 49 | #[derive(BytesCast)] |
|
50 | 50 | #[repr(C)] |
|
51 | 51 | struct DocketHeader { |
|
52 | 52 | marker: [u8; V2_FORMAT_MARKER.len()], |
|
53 | 53 | parent_1: [u8; STORED_NODE_ID_BYTES], |
|
54 | 54 | parent_2: [u8; STORED_NODE_ID_BYTES], |
|
55 | 55 | |
|
56 | 56 | metadata: TreeMetadata, |
|
57 | 57 | |
|
58 | 58 | /// Counted in bytes |
|
59 | 59 | data_size: Size, |
|
60 | 60 | |
|
61 | 61 | uuid_size: u8, |
|
62 | 62 | } |
|
63 | 63 | |
|
64 | 64 | pub struct Docket<'on_disk> { |
|
65 | 65 | header: &'on_disk DocketHeader, |
|
66 | 66 | pub uuid: &'on_disk [u8], |
|
67 | 67 | } |
|
68 | 68 | |
|
69 | 69 | /// Fields are documented in the *Tree metadata in the docket file* |
|
70 | 70 | /// section of `mercurial/helptext/internals/dirstate-v2.txt` |
|
71 | 71 | #[derive(BytesCast)] |
|
72 | 72 | #[repr(C)] |
|
73 | 73 | pub struct TreeMetadata { |
|
74 | 74 | root_nodes: ChildNodes, |
|
75 | 75 | nodes_with_entry_count: Size, |
|
76 | 76 | nodes_with_copy_source_count: Size, |
|
77 | 77 | unreachable_bytes: Size, |
|
78 | 78 | unused: [u8; 4], |
|
79 | 79 | |
|
80 | 80 | /// See *Optional hash of ignore patterns* section of |
|
81 | 81 | /// `mercurial/helptext/internals/dirstate-v2.txt` |
|
82 | 82 | ignore_patterns_hash: IgnorePatternsHash, |
|
83 | 83 | } |
|
84 | 84 | |
|
85 | 85 | /// Fields are documented in the *The data file format* |
|
86 | 86 | /// section of `mercurial/helptext/internals/dirstate-v2.txt` |
|
87 | 87 | #[derive(BytesCast)] |
|
88 | 88 | #[repr(C)] |
|
89 | 89 | pub(super) struct Node { |
|
90 | 90 | full_path: PathSlice, |
|
91 | 91 | |
|
92 | 92 | /// In bytes from `self.full_path.start` |
|
93 | 93 | base_name_start: PathSize, |
|
94 | 94 | |
|
95 | 95 | copy_source: OptPathSlice, |
|
96 | 96 | children: ChildNodes, |
|
97 | 97 | pub(super) descendants_with_entry_count: Size, |
|
98 | 98 | pub(super) tracked_descendants_count: Size, |
|
99 | 99 | flags: U16Be, |
|
100 | 100 | size: U32Be, |
|
101 | 101 | mtime: PackedTruncatedTimestamp, |
|
102 | 102 | } |
|
103 | 103 | |
|
104 | 104 | bitflags! { |
|
105 | 105 | #[repr(C)] |
|
106 | 106 | struct Flags: u16 { |
|
107 | 107 | const WDIR_TRACKED = 1 << 0; |
|
108 | 108 | const P1_TRACKED = 1 << 1; |
|
109 | 109 | const P2_INFO = 1 << 2; |
|
110 | 110 | const MODE_EXEC_PERM = 1 << 3; |
|
111 | 111 | const MODE_IS_SYMLINK = 1 << 4; |
|
112 | 112 | const HAS_FALLBACK_EXEC = 1 << 5; |
|
113 | 113 | const FALLBACK_EXEC = 1 << 6; |
|
114 | 114 | const HAS_FALLBACK_SYMLINK = 1 << 7; |
|
115 | 115 | const FALLBACK_SYMLINK = 1 << 8; |
|
116 | 116 | const EXPECTED_STATE_IS_MODIFIED = 1 << 9; |
|
117 | 117 | const HAS_MODE_AND_SIZE = 1 <<10; |
|
118 | 118 | const HAS_MTIME = 1 <<11; |
|
119 | 119 | const MTIME_SECOND_AMBIGUOUS = 1 << 12; |
|
120 | 120 | const DIRECTORY = 1 <<13; |
|
121 | 121 | const ALL_UNKNOWN_RECORDED = 1 <<14; |
|
122 | 122 | const ALL_IGNORED_RECORDED = 1 <<15; |
|
123 | 123 | } |
|
124 | 124 | } |
|
125 | 125 | |
|
126 | 126 | /// Duration since the Unix epoch |
|
127 | 127 | #[derive(BytesCast, Copy, Clone)] |
|
128 | 128 | #[repr(C)] |
|
129 | 129 | struct PackedTruncatedTimestamp { |
|
130 | 130 | truncated_seconds: U32Be, |
|
131 | 131 | nanoseconds: U32Be, |
|
132 | 132 | } |
|
133 | 133 | |
|
134 | 134 | /// Counted in bytes from the start of the file |
|
135 | 135 | /// |
|
136 | 136 | /// NOTE: not supporting `.hg/dirstate` files larger than 4 GiB. |
|
137 | 137 | type Offset = U32Be; |
|
138 | 138 | |
|
139 | 139 | /// Counted in number of items |
|
140 | 140 | /// |
|
141 | 141 | /// NOTE: we choose not to support counting more than 4 billion nodes anywhere. |
|
142 | 142 | type Size = U32Be; |
|
143 | 143 | |
|
144 | 144 | /// Counted in bytes |
|
145 | 145 | /// |
|
146 | 146 | /// NOTE: we choose not to support file names/paths longer than 64 KiB. |
|
147 | 147 | type PathSize = U16Be; |
|
148 | 148 | |
|
149 | 149 | /// A contiguous sequence of `len` times `Node`, representing the child nodes |
|
150 | 150 | /// of either some other node or of the repository root. |
|
151 | 151 | /// |
|
152 | 152 | /// Always sorted by ascending `full_path`, to allow binary search. |
|
153 | 153 | /// Since nodes with the same parent nodes also have the same parent path, |
|
154 | 154 | /// only the `base_name`s need to be compared during binary search. |
|
155 | 155 | #[derive(BytesCast, Copy, Clone)] |
|
156 | 156 | #[repr(C)] |
|
157 | 157 | struct ChildNodes { |
|
158 | 158 | start: Offset, |
|
159 | 159 | len: Size, |
|
160 | 160 | } |
|
161 | 161 | |
|
162 | 162 | /// A `HgPath` of `len` bytes |
|
163 | 163 | #[derive(BytesCast, Copy, Clone)] |
|
164 | 164 | #[repr(C)] |
|
165 | 165 | struct PathSlice { |
|
166 | 166 | start: Offset, |
|
167 | 167 | len: PathSize, |
|
168 | 168 | } |
|
169 | 169 | |
|
170 | 170 | /// Either nothing if `start == 0`, or a `HgPath` of `len` bytes |
|
171 | 171 | type OptPathSlice = PathSlice; |
|
172 | 172 | |
|
173 | 173 | /// Unexpected file format found in `.hg/dirstate` with the "v2" format. |
|
174 | 174 | /// |
|
175 | 175 | /// This should only happen if Mercurial is buggy or a repository is corrupted. |
|
176 | 176 | #[derive(Debug)] |
|
177 | 177 | pub struct DirstateV2ParseError; |
|
178 | 178 | |
|
179 | 179 | impl From<DirstateV2ParseError> for HgError { |
|
180 | 180 | fn from(_: DirstateV2ParseError) -> Self { |
|
181 | 181 | HgError::corrupted("dirstate-v2 parse error") |
|
182 | 182 | } |
|
183 | 183 | } |
|
184 | 184 | |
|
185 | 185 | impl From<DirstateV2ParseError> for crate::DirstateError { |
|
186 | 186 | fn from(error: DirstateV2ParseError) -> Self { |
|
187 | 187 | HgError::from(error).into() |
|
188 | 188 | } |
|
189 | 189 | } |
|
190 | 190 | |
|
191 | 191 | impl TreeMetadata { |
|
192 | 192 | pub fn as_bytes(&self) -> &[u8] { |
|
193 | 193 | BytesCast::as_bytes(self) |
|
194 | 194 | } |
|
195 | 195 | } |
|
196 | 196 | |
|
197 | 197 | impl<'on_disk> Docket<'on_disk> { |
|
198 | 198 | /// Generate the identifier for a new data file |
|
199 | 199 | /// |
|
200 | 200 | /// TODO: support the `HGTEST_UUIDFILE` environment variable. |
|
201 | 201 | /// See `mercurial/revlogutils/docket.py` |
|
202 | 202 | pub fn new_uid() -> String { |
|
203 | 203 | const ID_LENGTH: usize = 8; |
|
204 | 204 | let mut id = String::with_capacity(ID_LENGTH); |
|
205 | 205 | let mut rng = rand::thread_rng(); |
|
206 | 206 | for _ in 0..ID_LENGTH { |
|
207 | 207 | // One random hexadecimal digit. |
|
208 | 208 | // `unwrap` never panics because `impl Write for String` |
|
209 | 209 | // never returns an error. |
|
210 | 210 | write!(&mut id, "{:x}", rng.gen_range(0, 16)).unwrap(); |
|
211 | 211 | } |
|
212 | 212 | id |
|
213 | 213 | } |
|
214 | 214 | |
|
215 | 215 | pub fn serialize( |
|
216 | 216 | parents: DirstateParents, |
|
217 | 217 | tree_metadata: TreeMetadata, |
|
218 | 218 | data_size: u64, |
|
219 | 219 | uuid: &[u8], |
|
220 | 220 | ) -> Result<Vec<u8>, std::num::TryFromIntError> { |
|
221 | 221 | let header = DocketHeader { |
|
222 | 222 | marker: *V2_FORMAT_MARKER, |
|
223 | 223 | parent_1: parents.p1.pad_to_256_bits(), |
|
224 | 224 | parent_2: parents.p2.pad_to_256_bits(), |
|
225 | 225 | metadata: tree_metadata, |
|
226 | 226 | data_size: u32::try_from(data_size)?.into(), |
|
227 | 227 | uuid_size: uuid.len().try_into()?, |
|
228 | 228 | }; |
|
229 | 229 | let header = header.as_bytes(); |
|
230 | 230 | let mut docket = Vec::with_capacity(header.len() + uuid.len()); |
|
231 | 231 | docket.extend_from_slice(header); |
|
232 | 232 | docket.extend_from_slice(uuid); |
|
233 | 233 | Ok(docket) |
|
234 | 234 | } |
|
235 | 235 | |
|
236 | 236 | pub fn parents(&self) -> DirstateParents { |
|
237 | 237 | use crate::Node; |
|
238 | 238 | let p1 = Node::try_from(&self.header.parent_1[..USED_NODE_ID_BYTES]) |
|
239 | 239 | .unwrap() |
|
240 | 240 | .clone(); |
|
241 | 241 | let p2 = Node::try_from(&self.header.parent_2[..USED_NODE_ID_BYTES]) |
|
242 | 242 | .unwrap() |
|
243 | 243 | .clone(); |
|
244 | 244 | DirstateParents { p1, p2 } |
|
245 | 245 | } |
|
246 | 246 | |
|
247 | 247 | pub fn tree_metadata(&self) -> &[u8] { |
|
248 | 248 | self.header.metadata.as_bytes() |
|
249 | 249 | } |
|
250 | 250 | |
|
251 | 251 | pub fn data_size(&self) -> usize { |
|
252 | 252 | // This `unwrap` could only panic on a 16-bit CPU |
|
253 | 253 | self.header.data_size.get().try_into().unwrap() |
|
254 | 254 | } |
|
255 | 255 | |
|
256 | 256 | pub fn data_filename(&self) -> String { |
|
257 | 257 | String::from_utf8(format_bytes!(b"dirstate.{}", self.uuid)).unwrap() |
|
258 | 258 | } |
|
259 | 259 | } |
|
260 | 260 | |
|
261 | 261 | pub fn read_docket( |
|
262 | 262 | on_disk: &[u8], |
|
263 | 263 | ) -> Result<Docket<'_>, DirstateV2ParseError> { |
|
264 | 264 | let (header, uuid) = |
|
265 | 265 | DocketHeader::from_bytes(on_disk).map_err(|_| DirstateV2ParseError)?; |
|
266 | 266 | let uuid_size = header.uuid_size as usize; |
|
267 | 267 | if header.marker == *V2_FORMAT_MARKER && uuid.len() == uuid_size { |
|
268 | 268 | Ok(Docket { header, uuid }) |
|
269 | 269 | } else { |
|
270 | 270 | Err(DirstateV2ParseError) |
|
271 | 271 | } |
|
272 | 272 | } |
|
273 | 273 | |
|
274 | 274 | pub(super) fn read<'on_disk>( |
|
275 | 275 | on_disk: &'on_disk [u8], |
|
276 | 276 | metadata: &[u8], |
|
277 | 277 | ) -> Result<DirstateMap<'on_disk>, DirstateV2ParseError> { |
|
278 | 278 | if on_disk.is_empty() { |
|
279 | 279 | return Ok(DirstateMap::empty(on_disk)); |
|
280 | 280 | } |
|
281 | 281 | let (meta, _) = TreeMetadata::from_bytes(metadata) |
|
282 | 282 | .map_err(|_| DirstateV2ParseError)?; |
|
283 | 283 | let dirstate_map = DirstateMap { |
|
284 | 284 | on_disk, |
|
285 | 285 | root: dirstate_map::ChildNodes::OnDisk(read_nodes( |
|
286 | 286 | on_disk, |
|
287 | 287 | meta.root_nodes, |
|
288 | 288 | )?), |
|
289 | 289 | nodes_with_entry_count: meta.nodes_with_entry_count.get(), |
|
290 | 290 | nodes_with_copy_source_count: meta.nodes_with_copy_source_count.get(), |
|
291 | 291 | ignore_patterns_hash: meta.ignore_patterns_hash, |
|
292 | 292 | unreachable_bytes: meta.unreachable_bytes.get(), |
|
293 | 293 | }; |
|
294 | 294 | Ok(dirstate_map) |
|
295 | 295 | } |
|
296 | 296 | |
|
297 | 297 | impl Node { |
|
298 | 298 | pub(super) fn full_path<'on_disk>( |
|
299 | 299 | &self, |
|
300 | 300 | on_disk: &'on_disk [u8], |
|
301 | 301 | ) -> Result<&'on_disk HgPath, DirstateV2ParseError> { |
|
302 | 302 | read_hg_path(on_disk, self.full_path) |
|
303 | 303 | } |
|
304 | 304 | |
|
305 | 305 | pub(super) fn base_name_start<'on_disk>( |
|
306 | 306 | &self, |
|
307 | 307 | ) -> Result<usize, DirstateV2ParseError> { |
|
308 | 308 | let start = self.base_name_start.get(); |
|
309 | 309 | if start < self.full_path.len.get() { |
|
310 | 310 | let start = usize::try_from(start) |
|
311 | 311 | // u32 -> usize, could only panic on a 16-bit CPU |
|
312 | 312 | .expect("dirstate-v2 base_name_start out of bounds"); |
|
313 | 313 | Ok(start) |
|
314 | 314 | } else { |
|
315 | 315 | Err(DirstateV2ParseError) |
|
316 | 316 | } |
|
317 | 317 | } |
|
318 | 318 | |
|
319 | 319 | pub(super) fn base_name<'on_disk>( |
|
320 | 320 | &self, |
|
321 | 321 | on_disk: &'on_disk [u8], |
|
322 | 322 | ) -> Result<&'on_disk HgPath, DirstateV2ParseError> { |
|
323 | 323 | let full_path = self.full_path(on_disk)?; |
|
324 | 324 | let base_name_start = self.base_name_start()?; |
|
325 | 325 | Ok(HgPath::new(&full_path.as_bytes()[base_name_start..])) |
|
326 | 326 | } |
|
327 | 327 | |
|
328 | 328 | pub(super) fn path<'on_disk>( |
|
329 | 329 | &self, |
|
330 | 330 | on_disk: &'on_disk [u8], |
|
331 | 331 | ) -> Result<dirstate_map::NodeKey<'on_disk>, DirstateV2ParseError> { |
|
332 | 332 | Ok(WithBasename::from_raw_parts( |
|
333 | 333 | Cow::Borrowed(self.full_path(on_disk)?), |
|
334 | 334 | self.base_name_start()?, |
|
335 | 335 | )) |
|
336 | 336 | } |
|
337 | 337 | |
|
338 | 338 | pub(super) fn has_copy_source<'on_disk>(&self) -> bool { |
|
339 | 339 | self.copy_source.start.get() != 0 |
|
340 | 340 | } |
|
341 | 341 | |
|
342 | 342 | pub(super) fn copy_source<'on_disk>( |
|
343 | 343 | &self, |
|
344 | 344 | on_disk: &'on_disk [u8], |
|
345 | 345 | ) -> Result<Option<&'on_disk HgPath>, DirstateV2ParseError> { |
|
346 | 346 | Ok(if self.has_copy_source() { |
|
347 | 347 | Some(read_hg_path(on_disk, self.copy_source)?) |
|
348 | 348 | } else { |
|
349 | 349 | None |
|
350 | 350 | }) |
|
351 | 351 | } |
|
352 | 352 | |
|
353 | 353 | fn flags(&self) -> Flags { |
|
354 | 354 | Flags::from_bits_truncate(self.flags.get()) |
|
355 | 355 | } |
|
356 | 356 | |
|
357 | 357 | fn has_entry(&self) -> bool { |
|
358 | 358 | self.flags().intersects( |
|
359 | 359 | Flags::WDIR_TRACKED | Flags::P1_TRACKED | Flags::P2_INFO, |
|
360 | 360 | ) |
|
361 | 361 | } |
|
362 | 362 | |
|
363 | 363 | pub(super) fn node_data( |
|
364 | 364 | &self, |
|
365 | 365 | ) -> Result<dirstate_map::NodeData, DirstateV2ParseError> { |
|
366 | 366 | if self.has_entry() { |
|
367 | 367 | Ok(dirstate_map::NodeData::Entry(self.assume_entry()?)) |
|
368 | 368 | } else if let Some(mtime) = self.cached_directory_mtime()? { |
|
369 | 369 | Ok(dirstate_map::NodeData::CachedDirectory { mtime }) |
|
370 | 370 | } else { |
|
371 | 371 | Ok(dirstate_map::NodeData::None) |
|
372 | 372 | } |
|
373 | 373 | } |
|
374 | 374 | |
|
375 | 375 | pub(super) fn cached_directory_mtime( |
|
376 | 376 | &self, |
|
377 | 377 | ) -> Result<Option<TruncatedTimestamp>, DirstateV2ParseError> { |
|
378 | 378 | // For now we do not have code to handle the absence of |
|
379 | 379 | // ALL_UNKNOWN_RECORDED, so we ignore the mtime if the flag is |
|
380 | 380 | // unset. |
|
381 | 381 | if self.flags().contains(Flags::DIRECTORY) |
|
382 | 382 | && self.flags().contains(Flags::HAS_MTIME) |
|
383 | 383 | && self.flags().contains(Flags::ALL_UNKNOWN_RECORDED) |
|
384 | 384 | { |
|
385 |
Ok(Some(self.mtime |
|
|
385 | Ok(Some(self.mtime()?)) | |
|
386 | 386 | } else { |
|
387 | 387 | Ok(None) |
|
388 | 388 | } |
|
389 | 389 | } |
|
390 | 390 | |
|
391 | 391 | fn synthesize_unix_mode(&self) -> u32 { |
|
392 | 392 | let file_type = if self.flags().contains(Flags::MODE_IS_SYMLINK) { |
|
393 | 393 | libc::S_IFLNK |
|
394 | 394 | } else { |
|
395 | 395 | libc::S_IFREG |
|
396 | 396 | }; |
|
397 | 397 | let permisions = if self.flags().contains(Flags::MODE_EXEC_PERM) { |
|
398 | 398 | 0o755 |
|
399 | 399 | } else { |
|
400 | 400 | 0o644 |
|
401 | 401 | }; |
|
402 | 402 | file_type | permisions |
|
403 | 403 | } |
|
404 | 404 | |
|
405 | fn mtime(&self) -> Result<TruncatedTimestamp, DirstateV2ParseError> { | |
|
406 | let mut m: TruncatedTimestamp = self.mtime.try_into()?; | |
|
407 | if self.flags().contains(Flags::MTIME_SECOND_AMBIGUOUS) { | |
|
408 | m.second_ambiguous = true; | |
|
409 | } | |
|
410 | Ok(m) | |
|
411 | } | |
|
412 | ||
|
405 | 413 | fn assume_entry(&self) -> Result<DirstateEntry, DirstateV2ParseError> { |
|
406 | 414 | // TODO: convert through raw bits instead? |
|
407 | 415 | let wdir_tracked = self.flags().contains(Flags::WDIR_TRACKED); |
|
408 | 416 | let p1_tracked = self.flags().contains(Flags::P1_TRACKED); |
|
409 | 417 | let p2_info = self.flags().contains(Flags::P2_INFO); |
|
410 | 418 | let mode_size = if self.flags().contains(Flags::HAS_MODE_AND_SIZE) |
|
411 | 419 | && !self.flags().contains(Flags::EXPECTED_STATE_IS_MODIFIED) |
|
412 | 420 | { |
|
413 | 421 | Some((self.synthesize_unix_mode(), self.size.into())) |
|
414 | 422 | } else { |
|
415 | 423 | None |
|
416 | 424 | }; |
|
417 | 425 | let mtime = if self.flags().contains(Flags::HAS_MTIME) |
|
418 | 426 | && !self.flags().contains(Flags::DIRECTORY) |
|
419 | 427 | && !self.flags().contains(Flags::EXPECTED_STATE_IS_MODIFIED) |
|
420 | 428 | { |
|
421 | let mut m: TruncatedTimestamp = self.mtime.try_into()?; | |
|
422 | if self.flags().contains(Flags::MTIME_SECOND_AMBIGUOUS) { | |
|
423 | m.second_ambiguous = true; | |
|
424 | } | |
|
425 | Some(m) | |
|
429 | Some(self.mtime()?) | |
|
426 | 430 | } else { |
|
427 | 431 | None |
|
428 | 432 | }; |
|
429 | 433 | let fallback_exec = if self.flags().contains(Flags::HAS_FALLBACK_EXEC) |
|
430 | 434 | { |
|
431 | 435 | Some(self.flags().contains(Flags::FALLBACK_EXEC)) |
|
432 | 436 | } else { |
|
433 | 437 | None |
|
434 | 438 | }; |
|
435 | 439 | let fallback_symlink = |
|
436 | 440 | if self.flags().contains(Flags::HAS_FALLBACK_SYMLINK) { |
|
437 | 441 | Some(self.flags().contains(Flags::FALLBACK_SYMLINK)) |
|
438 | 442 | } else { |
|
439 | 443 | None |
|
440 | 444 | }; |
|
441 | 445 | Ok(DirstateEntry::from_v2_data( |
|
442 | 446 | wdir_tracked, |
|
443 | 447 | p1_tracked, |
|
444 | 448 | p2_info, |
|
445 | 449 | mode_size, |
|
446 | 450 | mtime, |
|
447 | 451 | fallback_exec, |
|
448 | 452 | fallback_symlink, |
|
449 | 453 | )) |
|
450 | 454 | } |
|
451 | 455 | |
|
452 | 456 | pub(super) fn entry( |
|
453 | 457 | &self, |
|
454 | 458 | ) -> Result<Option<DirstateEntry>, DirstateV2ParseError> { |
|
455 | 459 | if self.has_entry() { |
|
456 | 460 | Ok(Some(self.assume_entry()?)) |
|
457 | 461 | } else { |
|
458 | 462 | Ok(None) |
|
459 | 463 | } |
|
460 | 464 | } |
|
461 | 465 | |
|
462 | 466 | pub(super) fn children<'on_disk>( |
|
463 | 467 | &self, |
|
464 | 468 | on_disk: &'on_disk [u8], |
|
465 | 469 | ) -> Result<&'on_disk [Node], DirstateV2ParseError> { |
|
466 | 470 | read_nodes(on_disk, self.children) |
|
467 | 471 | } |
|
468 | 472 | |
|
469 | 473 | pub(super) fn to_in_memory_node<'on_disk>( |
|
470 | 474 | &self, |
|
471 | 475 | on_disk: &'on_disk [u8], |
|
472 | 476 | ) -> Result<dirstate_map::Node<'on_disk>, DirstateV2ParseError> { |
|
473 | 477 | Ok(dirstate_map::Node { |
|
474 | 478 | children: dirstate_map::ChildNodes::OnDisk( |
|
475 | 479 | self.children(on_disk)?, |
|
476 | 480 | ), |
|
477 | 481 | copy_source: self.copy_source(on_disk)?.map(Cow::Borrowed), |
|
478 | 482 | data: self.node_data()?, |
|
479 | 483 | descendants_with_entry_count: self |
|
480 | 484 | .descendants_with_entry_count |
|
481 | 485 | .get(), |
|
482 | 486 | tracked_descendants_count: self.tracked_descendants_count.get(), |
|
483 | 487 | }) |
|
484 | 488 | } |
|
485 | 489 | |
|
486 | 490 | fn from_dirstate_entry( |
|
487 | 491 | entry: &DirstateEntry, |
|
488 | 492 | ) -> (Flags, U32Be, PackedTruncatedTimestamp) { |
|
489 | 493 | let ( |
|
490 | 494 | wdir_tracked, |
|
491 | 495 | p1_tracked, |
|
492 | 496 | p2_info, |
|
493 | 497 | mode_size_opt, |
|
494 | 498 | mtime_opt, |
|
495 | 499 | fallback_exec, |
|
496 | 500 | fallback_symlink, |
|
497 | 501 | ) = entry.v2_data(); |
|
498 | 502 | // TODO: convert throug raw flag bits instead? |
|
499 | 503 | let mut flags = Flags::empty(); |
|
500 | 504 | flags.set(Flags::WDIR_TRACKED, wdir_tracked); |
|
501 | 505 | flags.set(Flags::P1_TRACKED, p1_tracked); |
|
502 | 506 | flags.set(Flags::P2_INFO, p2_info); |
|
503 | 507 | let size = if let Some((m, s)) = mode_size_opt { |
|
504 | 508 | let exec_perm = m & libc::S_IXUSR != 0; |
|
505 | 509 | let is_symlink = m & libc::S_IFMT == libc::S_IFLNK; |
|
506 | 510 | flags.set(Flags::MODE_EXEC_PERM, exec_perm); |
|
507 | 511 | flags.set(Flags::MODE_IS_SYMLINK, is_symlink); |
|
508 | 512 | flags.insert(Flags::HAS_MODE_AND_SIZE); |
|
509 | 513 | s.into() |
|
510 | 514 | } else { |
|
511 | 515 | 0.into() |
|
512 | 516 | }; |
|
513 | 517 | let mtime = if let Some(m) = mtime_opt { |
|
514 | 518 | flags.insert(Flags::HAS_MTIME); |
|
515 | 519 | if m.second_ambiguous { |
|
516 | 520 | flags.insert(Flags::MTIME_SECOND_AMBIGUOUS); |
|
517 | 521 | }; |
|
518 | 522 | m.into() |
|
519 | 523 | } else { |
|
520 | 524 | PackedTruncatedTimestamp::null() |
|
521 | 525 | }; |
|
522 | 526 | if let Some(f_exec) = fallback_exec { |
|
523 | 527 | flags.insert(Flags::HAS_FALLBACK_EXEC); |
|
524 | 528 | if f_exec { |
|
525 | 529 | flags.insert(Flags::FALLBACK_EXEC); |
|
526 | 530 | } |
|
527 | 531 | } |
|
528 | 532 | if let Some(f_symlink) = fallback_symlink { |
|
529 | 533 | flags.insert(Flags::HAS_FALLBACK_SYMLINK); |
|
530 | 534 | if f_symlink { |
|
531 | 535 | flags.insert(Flags::FALLBACK_SYMLINK); |
|
532 | 536 | } |
|
533 | 537 | } |
|
534 | 538 | (flags, size, mtime) |
|
535 | 539 | } |
|
536 | 540 | } |
|
537 | 541 | |
|
538 | 542 | fn read_hg_path( |
|
539 | 543 | on_disk: &[u8], |
|
540 | 544 | slice: PathSlice, |
|
541 | 545 | ) -> Result<&HgPath, DirstateV2ParseError> { |
|
542 | 546 | read_slice(on_disk, slice.start, slice.len.get()).map(HgPath::new) |
|
543 | 547 | } |
|
544 | 548 | |
|
545 | 549 | fn read_nodes( |
|
546 | 550 | on_disk: &[u8], |
|
547 | 551 | slice: ChildNodes, |
|
548 | 552 | ) -> Result<&[Node], DirstateV2ParseError> { |
|
549 | 553 | read_slice(on_disk, slice.start, slice.len.get()) |
|
550 | 554 | } |
|
551 | 555 | |
|
552 | 556 | fn read_slice<T, Len>( |
|
553 | 557 | on_disk: &[u8], |
|
554 | 558 | start: Offset, |
|
555 | 559 | len: Len, |
|
556 | 560 | ) -> Result<&[T], DirstateV2ParseError> |
|
557 | 561 | where |
|
558 | 562 | T: BytesCast, |
|
559 | 563 | Len: TryInto<usize>, |
|
560 | 564 | { |
|
561 | 565 | // Either `usize::MAX` would result in "out of bounds" error since a single |
|
562 | 566 | // `&[u8]` cannot occupy the entire addess space. |
|
563 | 567 | let start = start.get().try_into().unwrap_or(std::usize::MAX); |
|
564 | 568 | let len = len.try_into().unwrap_or(std::usize::MAX); |
|
565 | 569 | on_disk |
|
566 | 570 | .get(start..) |
|
567 | 571 | .and_then(|bytes| T::slice_from_bytes(bytes, len).ok()) |
|
568 | 572 | .map(|(slice, _rest)| slice) |
|
569 | 573 | .ok_or_else(|| DirstateV2ParseError) |
|
570 | 574 | } |
|
571 | 575 | |
|
572 | 576 | pub(crate) fn for_each_tracked_path<'on_disk>( |
|
573 | 577 | on_disk: &'on_disk [u8], |
|
574 | 578 | metadata: &[u8], |
|
575 | 579 | mut f: impl FnMut(&'on_disk HgPath), |
|
576 | 580 | ) -> Result<(), DirstateV2ParseError> { |
|
577 | 581 | let (meta, _) = TreeMetadata::from_bytes(metadata) |
|
578 | 582 | .map_err(|_| DirstateV2ParseError)?; |
|
579 | 583 | fn recur<'on_disk>( |
|
580 | 584 | on_disk: &'on_disk [u8], |
|
581 | 585 | nodes: ChildNodes, |
|
582 | 586 | f: &mut impl FnMut(&'on_disk HgPath), |
|
583 | 587 | ) -> Result<(), DirstateV2ParseError> { |
|
584 | 588 | for node in read_nodes(on_disk, nodes)? { |
|
585 | 589 | if let Some(entry) = node.entry()? { |
|
586 | 590 | if entry.state().is_tracked() { |
|
587 | 591 | f(node.full_path(on_disk)?) |
|
588 | 592 | } |
|
589 | 593 | } |
|
590 | 594 | recur(on_disk, node.children, f)? |
|
591 | 595 | } |
|
592 | 596 | Ok(()) |
|
593 | 597 | } |
|
594 | 598 | recur(on_disk, meta.root_nodes, &mut f) |
|
595 | 599 | } |
|
596 | 600 | |
|
597 | 601 | /// Returns new data and metadata, together with whether that data should be |
|
598 | 602 | /// appended to the existing data file whose content is at |
|
599 | 603 | /// `dirstate_map.on_disk` (true), instead of written to a new data file |
|
600 | 604 | /// (false). |
|
601 | 605 | pub(super) fn write( |
|
602 | 606 | dirstate_map: &DirstateMap, |
|
603 | 607 | can_append: bool, |
|
604 | 608 | ) -> Result<(Vec<u8>, TreeMetadata, bool), DirstateError> { |
|
605 | 609 | let append = can_append && dirstate_map.write_should_append(); |
|
606 | 610 | |
|
607 | 611 | // This ignores the space for paths, and for nodes without an entry. |
|
608 | 612 | // TODO: better estimate? Skip the `Vec` and write to a file directly? |
|
609 | 613 | let size_guess = std::mem::size_of::<Node>() |
|
610 | 614 | * dirstate_map.nodes_with_entry_count as usize; |
|
611 | 615 | |
|
612 | 616 | let mut writer = Writer { |
|
613 | 617 | dirstate_map, |
|
614 | 618 | append, |
|
615 | 619 | out: Vec::with_capacity(size_guess), |
|
616 | 620 | }; |
|
617 | 621 | |
|
618 | 622 | let root_nodes = writer.write_nodes(dirstate_map.root.as_ref())?; |
|
619 | 623 | |
|
620 | 624 | let meta = TreeMetadata { |
|
621 | 625 | root_nodes, |
|
622 | 626 | nodes_with_entry_count: dirstate_map.nodes_with_entry_count.into(), |
|
623 | 627 | nodes_with_copy_source_count: dirstate_map |
|
624 | 628 | .nodes_with_copy_source_count |
|
625 | 629 | .into(), |
|
626 | 630 | unreachable_bytes: dirstate_map.unreachable_bytes.into(), |
|
627 | 631 | unused: [0; 4], |
|
628 | 632 | ignore_patterns_hash: dirstate_map.ignore_patterns_hash, |
|
629 | 633 | }; |
|
630 | 634 | Ok((writer.out, meta, append)) |
|
631 | 635 | } |
|
632 | 636 | |
|
633 | 637 | struct Writer<'dmap, 'on_disk> { |
|
634 | 638 | dirstate_map: &'dmap DirstateMap<'on_disk>, |
|
635 | 639 | append: bool, |
|
636 | 640 | out: Vec<u8>, |
|
637 | 641 | } |
|
638 | 642 | |
|
639 | 643 | impl Writer<'_, '_> { |
|
640 | 644 | fn write_nodes( |
|
641 | 645 | &mut self, |
|
642 | 646 | nodes: dirstate_map::ChildNodesRef, |
|
643 | 647 | ) -> Result<ChildNodes, DirstateError> { |
|
644 | 648 | // Reuse already-written nodes if possible |
|
645 | 649 | if self.append { |
|
646 | 650 | if let dirstate_map::ChildNodesRef::OnDisk(nodes_slice) = nodes { |
|
647 | 651 | let start = self.on_disk_offset_of(nodes_slice).expect( |
|
648 | 652 | "dirstate-v2 OnDisk nodes not found within on_disk", |
|
649 | 653 | ); |
|
650 | 654 | let len = child_nodes_len_from_usize(nodes_slice.len()); |
|
651 | 655 | return Ok(ChildNodes { start, len }); |
|
652 | 656 | } |
|
653 | 657 | } |
|
654 | 658 | |
|
655 | 659 | // `dirstate_map::ChildNodes::InMemory` contains a `HashMap` which has |
|
656 | 660 | // undefined iteration order. Sort to enable binary search in the |
|
657 | 661 | // written file. |
|
658 | 662 | let nodes = nodes.sorted(); |
|
659 | 663 | let nodes_len = nodes.len(); |
|
660 | 664 | |
|
661 | 665 | // First accumulate serialized nodes in a `Vec` |
|
662 | 666 | let mut on_disk_nodes = Vec::with_capacity(nodes_len); |
|
663 | 667 | for node in nodes { |
|
664 | 668 | let children = |
|
665 | 669 | self.write_nodes(node.children(self.dirstate_map.on_disk)?)?; |
|
666 | 670 | let full_path = node.full_path(self.dirstate_map.on_disk)?; |
|
667 | 671 | let full_path = self.write_path(full_path.as_bytes()); |
|
668 | 672 | let copy_source = if let Some(source) = |
|
669 | 673 | node.copy_source(self.dirstate_map.on_disk)? |
|
670 | 674 | { |
|
671 | 675 | self.write_path(source.as_bytes()) |
|
672 | 676 | } else { |
|
673 | 677 | PathSlice { |
|
674 | 678 | start: 0.into(), |
|
675 | 679 | len: 0.into(), |
|
676 | 680 | } |
|
677 | 681 | }; |
|
678 | 682 | on_disk_nodes.push(match node { |
|
679 | 683 | NodeRef::InMemory(path, node) => { |
|
680 | 684 | let (flags, size, mtime) = match &node.data { |
|
681 | 685 | dirstate_map::NodeData::Entry(entry) => { |
|
682 | 686 | Node::from_dirstate_entry(entry) |
|
683 | 687 | } |
|
684 |
dirstate_map::NodeData::CachedDirectory { mtime } => |
|
|
688 | dirstate_map::NodeData::CachedDirectory { mtime } => { | |
|
685 | 689 | // we currently never set a mtime if unknown file |
|
686 | 690 | // are present. |
|
687 | 691 | // So if we have a mtime for a directory, we know |
|
688 | 692 | // they are no unknown |
|
689 | 693 | // files and we |
|
690 | 694 | // blindly set ALL_UNKNOWN_RECORDED. |
|
691 | 695 | // |
|
692 | 696 | // We never set ALL_IGNORED_RECORDED since we |
|
693 | 697 | // don't track that case |
|
694 | 698 | // currently. |
|
695 | Flags::DIRECTORY | |
|
699 | let mut flags = Flags::DIRECTORY | |
|
696 | 700 | | Flags::HAS_MTIME |
|
697 |
| Flags::ALL_UNKNOWN_RECORDED |
|
|
698 |
|
|
|
699 | (*mtime).into(), | |
|
700 |
|
|
|
701 | | Flags::ALL_UNKNOWN_RECORDED; | |
|
702 | if mtime.second_ambiguous { | |
|
703 | flags.insert(Flags::MTIME_SECOND_AMBIGUOUS) | |
|
704 | } | |
|
705 | (flags, 0.into(), (*mtime).into()) | |
|
706 | } | |
|
701 | 707 | dirstate_map::NodeData::None => ( |
|
702 | 708 | Flags::DIRECTORY, |
|
703 | 709 | 0.into(), |
|
704 | 710 | PackedTruncatedTimestamp::null(), |
|
705 | 711 | ), |
|
706 | 712 | }; |
|
707 | 713 | Node { |
|
708 | 714 | children, |
|
709 | 715 | copy_source, |
|
710 | 716 | full_path, |
|
711 | 717 | base_name_start: u16::try_from(path.base_name_start()) |
|
712 | 718 | // Could only panic for paths over 64 KiB |
|
713 | 719 | .expect("dirstate-v2 path length overflow") |
|
714 | 720 | .into(), |
|
715 | 721 | descendants_with_entry_count: node |
|
716 | 722 | .descendants_with_entry_count |
|
717 | 723 | .into(), |
|
718 | 724 | tracked_descendants_count: node |
|
719 | 725 | .tracked_descendants_count |
|
720 | 726 | .into(), |
|
721 | 727 | flags: flags.bits().into(), |
|
722 | 728 | size, |
|
723 | 729 | mtime, |
|
724 | 730 | } |
|
725 | 731 | } |
|
726 | 732 | NodeRef::OnDisk(node) => Node { |
|
727 | 733 | children, |
|
728 | 734 | copy_source, |
|
729 | 735 | full_path, |
|
730 | 736 | ..*node |
|
731 | 737 | }, |
|
732 | 738 | }) |
|
733 | 739 | } |
|
734 | 740 | // … so we can write them contiguously, after writing everything else |
|
735 | 741 | // they refer to. |
|
736 | 742 | let start = self.current_offset(); |
|
737 | 743 | let len = child_nodes_len_from_usize(nodes_len); |
|
738 | 744 | self.out.extend(on_disk_nodes.as_bytes()); |
|
739 | 745 | Ok(ChildNodes { start, len }) |
|
740 | 746 | } |
|
741 | 747 | |
|
742 | 748 | /// If the given slice of items is within `on_disk`, returns its offset |
|
743 | 749 | /// from the start of `on_disk`. |
|
744 | 750 | fn on_disk_offset_of<T>(&self, slice: &[T]) -> Option<Offset> |
|
745 | 751 | where |
|
746 | 752 | T: BytesCast, |
|
747 | 753 | { |
|
748 | 754 | fn address_range(slice: &[u8]) -> std::ops::RangeInclusive<usize> { |
|
749 | 755 | let start = slice.as_ptr() as usize; |
|
750 | 756 | let end = start + slice.len(); |
|
751 | 757 | start..=end |
|
752 | 758 | } |
|
753 | 759 | let slice_addresses = address_range(slice.as_bytes()); |
|
754 | 760 | let on_disk_addresses = address_range(self.dirstate_map.on_disk); |
|
755 | 761 | if on_disk_addresses.contains(slice_addresses.start()) |
|
756 | 762 | && on_disk_addresses.contains(slice_addresses.end()) |
|
757 | 763 | { |
|
758 | 764 | let offset = slice_addresses.start() - on_disk_addresses.start(); |
|
759 | 765 | Some(offset_from_usize(offset)) |
|
760 | 766 | } else { |
|
761 | 767 | None |
|
762 | 768 | } |
|
763 | 769 | } |
|
764 | 770 | |
|
765 | 771 | fn current_offset(&mut self) -> Offset { |
|
766 | 772 | let mut offset = self.out.len(); |
|
767 | 773 | if self.append { |
|
768 | 774 | offset += self.dirstate_map.on_disk.len() |
|
769 | 775 | } |
|
770 | 776 | offset_from_usize(offset) |
|
771 | 777 | } |
|
772 | 778 | |
|
773 | 779 | fn write_path(&mut self, slice: &[u8]) -> PathSlice { |
|
774 | 780 | let len = path_len_from_usize(slice.len()); |
|
775 | 781 | // Reuse an already-written path if possible |
|
776 | 782 | if self.append { |
|
777 | 783 | if let Some(start) = self.on_disk_offset_of(slice) { |
|
778 | 784 | return PathSlice { start, len }; |
|
779 | 785 | } |
|
780 | 786 | } |
|
781 | 787 | let start = self.current_offset(); |
|
782 | 788 | self.out.extend(slice.as_bytes()); |
|
783 | 789 | PathSlice { start, len } |
|
784 | 790 | } |
|
785 | 791 | } |
|
786 | 792 | |
|
787 | 793 | fn offset_from_usize(x: usize) -> Offset { |
|
788 | 794 | u32::try_from(x) |
|
789 | 795 | // Could only panic for a dirstate file larger than 4 GiB |
|
790 | 796 | .expect("dirstate-v2 offset overflow") |
|
791 | 797 | .into() |
|
792 | 798 | } |
|
793 | 799 | |
|
794 | 800 | fn child_nodes_len_from_usize(x: usize) -> Size { |
|
795 | 801 | u32::try_from(x) |
|
796 | 802 | // Could only panic with over 4 billion nodes |
|
797 | 803 | .expect("dirstate-v2 slice length overflow") |
|
798 | 804 | .into() |
|
799 | 805 | } |
|
800 | 806 | |
|
801 | 807 | fn path_len_from_usize(x: usize) -> PathSize { |
|
802 | 808 | u16::try_from(x) |
|
803 | 809 | // Could only panic for paths over 64 KiB |
|
804 | 810 | .expect("dirstate-v2 path length overflow") |
|
805 | 811 | .into() |
|
806 | 812 | } |
|
807 | 813 | |
|
808 | 814 | impl From<TruncatedTimestamp> for PackedTruncatedTimestamp { |
|
809 | 815 | fn from(timestamp: TruncatedTimestamp) -> Self { |
|
810 | 816 | Self { |
|
811 | 817 | truncated_seconds: timestamp.truncated_seconds().into(), |
|
812 | 818 | nanoseconds: timestamp.nanoseconds().into(), |
|
813 | 819 | } |
|
814 | 820 | } |
|
815 | 821 | } |
|
816 | 822 | |
|
817 | 823 | impl TryFrom<PackedTruncatedTimestamp> for TruncatedTimestamp { |
|
818 | 824 | type Error = DirstateV2ParseError; |
|
819 | 825 | |
|
820 | 826 | fn try_from( |
|
821 | 827 | timestamp: PackedTruncatedTimestamp, |
|
822 | 828 | ) -> Result<Self, Self::Error> { |
|
823 | 829 | Self::from_already_truncated( |
|
824 | 830 | timestamp.truncated_seconds.get(), |
|
825 | 831 | timestamp.nanoseconds.get(), |
|
826 | 832 | false, |
|
827 | 833 | ) |
|
828 | 834 | } |
|
829 | 835 | } |
|
830 | 836 | impl PackedTruncatedTimestamp { |
|
831 | 837 | fn null() -> Self { |
|
832 | 838 | Self { |
|
833 | 839 | truncated_seconds: 0.into(), |
|
834 | 840 | nanoseconds: 0.into(), |
|
835 | 841 | } |
|
836 | 842 | } |
|
837 | 843 | } |
@@ -1,782 +1,798 b'' | |||
|
1 | 1 | use crate::dirstate::entry::TruncatedTimestamp; |
|
2 | 2 | use crate::dirstate::status::IgnoreFnType; |
|
3 | 3 | use crate::dirstate::status::StatusPath; |
|
4 | 4 | use crate::dirstate_tree::dirstate_map::BorrowedPath; |
|
5 | 5 | use crate::dirstate_tree::dirstate_map::ChildNodesRef; |
|
6 | 6 | use crate::dirstate_tree::dirstate_map::DirstateMap; |
|
7 | 7 | use crate::dirstate_tree::dirstate_map::NodeData; |
|
8 | 8 | use crate::dirstate_tree::dirstate_map::NodeRef; |
|
9 | 9 | use crate::dirstate_tree::on_disk::DirstateV2ParseError; |
|
10 | 10 | use crate::matchers::get_ignore_function; |
|
11 | 11 | use crate::matchers::Matcher; |
|
12 | 12 | use crate::utils::files::get_bytes_from_os_string; |
|
13 | 13 | use crate::utils::files::get_path_from_bytes; |
|
14 | 14 | use crate::utils::hg_path::HgPath; |
|
15 | 15 | use crate::BadMatch; |
|
16 | 16 | use crate::DirstateStatus; |
|
17 | 17 | use crate::EntryState; |
|
18 | 18 | use crate::HgPathBuf; |
|
19 | 19 | use crate::HgPathCow; |
|
20 | 20 | use crate::PatternFileWarning; |
|
21 | 21 | use crate::StatusError; |
|
22 | 22 | use crate::StatusOptions; |
|
23 | 23 | use micro_timer::timed; |
|
24 | 24 | use rayon::prelude::*; |
|
25 | 25 | use sha1::{Digest, Sha1}; |
|
26 | 26 | use std::borrow::Cow; |
|
27 | 27 | use std::io; |
|
28 | 28 | use std::path::Path; |
|
29 | 29 | use std::path::PathBuf; |
|
30 | 30 | use std::sync::Mutex; |
|
31 | 31 | use std::time::SystemTime; |
|
32 | 32 | |
|
33 | 33 | /// Returns the status of the working directory compared to its parent |
|
34 | 34 | /// changeset. |
|
35 | 35 | /// |
|
36 | 36 | /// This algorithm is based on traversing the filesystem tree (`fs` in function |
|
37 | 37 | /// and variable names) and dirstate tree at the same time. The core of this |
|
38 | 38 | /// traversal is the recursive `traverse_fs_directory_and_dirstate` function |
|
39 | 39 | /// and its use of `itertools::merge_join_by`. When reaching a path that only |
|
40 | 40 | /// exists in one of the two trees, depending on information requested by |
|
41 | 41 | /// `options` we may need to traverse the remaining subtree. |
|
42 | 42 | #[timed] |
|
43 | 43 | pub fn status<'tree, 'on_disk: 'tree>( |
|
44 | 44 | dmap: &'tree mut DirstateMap<'on_disk>, |
|
45 | 45 | matcher: &(dyn Matcher + Sync), |
|
46 | 46 | root_dir: PathBuf, |
|
47 | 47 | ignore_files: Vec<PathBuf>, |
|
48 | 48 | options: StatusOptions, |
|
49 | 49 | ) -> Result<(DirstateStatus<'on_disk>, Vec<PatternFileWarning>), StatusError> { |
|
50 | 50 | let (ignore_fn, warnings, patterns_changed): (IgnoreFnType, _, _) = |
|
51 | 51 | if options.list_ignored || options.list_unknown { |
|
52 | 52 | let mut hasher = Sha1::new(); |
|
53 | 53 | let (ignore_fn, warnings) = get_ignore_function( |
|
54 | 54 | ignore_files, |
|
55 | 55 | &root_dir, |
|
56 | 56 | &mut |pattern_bytes| hasher.update(pattern_bytes), |
|
57 | 57 | )?; |
|
58 | 58 | let new_hash = *hasher.finalize().as_ref(); |
|
59 | 59 | let changed = new_hash != dmap.ignore_patterns_hash; |
|
60 | 60 | dmap.ignore_patterns_hash = new_hash; |
|
61 | 61 | (ignore_fn, warnings, Some(changed)) |
|
62 | 62 | } else { |
|
63 | 63 | (Box::new(|&_| true), vec![], None) |
|
64 | 64 | }; |
|
65 | 65 | |
|
66 |
let filesystem_time_at_status_start = |
|
|
66 | let filesystem_time_at_status_start = | |
|
67 | filesystem_now(&root_dir).ok().map(TruncatedTimestamp::from); | |
|
67 | 68 | let outcome = DirstateStatus { |
|
68 | 69 | filesystem_time_at_status_start, |
|
69 | 70 | ..Default::default() |
|
70 | 71 | }; |
|
71 | 72 | let common = StatusCommon { |
|
72 | 73 | dmap, |
|
73 | 74 | options, |
|
74 | 75 | matcher, |
|
75 | 76 | ignore_fn, |
|
76 | 77 | outcome: Mutex::new(outcome), |
|
77 | 78 | ignore_patterns_have_changed: patterns_changed, |
|
78 | 79 | new_cachable_directories: Default::default(), |
|
79 | 80 | outated_cached_directories: Default::default(), |
|
80 | 81 | filesystem_time_at_status_start, |
|
81 | 82 | }; |
|
82 | 83 | let is_at_repo_root = true; |
|
83 | 84 | let hg_path = &BorrowedPath::OnDisk(HgPath::new("")); |
|
84 | 85 | let has_ignored_ancestor = false; |
|
85 | 86 | let root_cached_mtime = None; |
|
86 | 87 | let root_dir_metadata = None; |
|
87 | 88 | // If the path we have for the repository root is a symlink, do follow it. |
|
88 | 89 | // (As opposed to symlinks within the working directory which are not |
|
89 | 90 | // followed, using `std::fs::symlink_metadata`.) |
|
90 | 91 | common.traverse_fs_directory_and_dirstate( |
|
91 | 92 | has_ignored_ancestor, |
|
92 | 93 | dmap.root.as_ref(), |
|
93 | 94 | hg_path, |
|
94 | 95 | &root_dir, |
|
95 | 96 | root_dir_metadata, |
|
96 | 97 | root_cached_mtime, |
|
97 | 98 | is_at_repo_root, |
|
98 | 99 | )?; |
|
99 | 100 | let mut outcome = common.outcome.into_inner().unwrap(); |
|
100 | 101 | let new_cachable = common.new_cachable_directories.into_inner().unwrap(); |
|
101 | 102 | let outdated = common.outated_cached_directories.into_inner().unwrap(); |
|
102 | 103 | |
|
103 | 104 | outcome.dirty = common.ignore_patterns_have_changed == Some(true) |
|
104 | 105 | || !outdated.is_empty() |
|
105 | 106 | || !new_cachable.is_empty(); |
|
106 | 107 | |
|
107 | 108 | // Remove outdated mtimes before adding new mtimes, in case a given |
|
108 | 109 | // directory is both |
|
109 | 110 | for path in &outdated { |
|
110 | 111 | let node = dmap.get_or_insert(path)?; |
|
111 | 112 | if let NodeData::CachedDirectory { .. } = &node.data { |
|
112 | 113 | node.data = NodeData::None |
|
113 | 114 | } |
|
114 | 115 | } |
|
115 | 116 | for (path, mtime) in &new_cachable { |
|
116 | 117 | let node = dmap.get_or_insert(path)?; |
|
117 | 118 | match &node.data { |
|
118 | 119 | NodeData::Entry(_) => {} // Don’t overwrite an entry |
|
119 | 120 | NodeData::CachedDirectory { .. } | NodeData::None => { |
|
120 | 121 | node.data = NodeData::CachedDirectory { mtime: *mtime } |
|
121 | 122 | } |
|
122 | 123 | } |
|
123 | 124 | } |
|
124 | 125 | |
|
125 | 126 | Ok((outcome, warnings)) |
|
126 | 127 | } |
|
127 | 128 | |
|
128 | 129 | /// Bag of random things needed by various parts of the algorithm. Reduces the |
|
129 | 130 | /// number of parameters passed to functions. |
|
130 | 131 | struct StatusCommon<'a, 'tree, 'on_disk: 'tree> { |
|
131 | 132 | dmap: &'tree DirstateMap<'on_disk>, |
|
132 | 133 | options: StatusOptions, |
|
133 | 134 | matcher: &'a (dyn Matcher + Sync), |
|
134 | 135 | ignore_fn: IgnoreFnType<'a>, |
|
135 | 136 | outcome: Mutex<DirstateStatus<'on_disk>>, |
|
136 | 137 | new_cachable_directories: |
|
137 | 138 | Mutex<Vec<(Cow<'on_disk, HgPath>, TruncatedTimestamp)>>, |
|
138 | 139 | outated_cached_directories: Mutex<Vec<Cow<'on_disk, HgPath>>>, |
|
139 | 140 | |
|
140 | 141 | /// Whether ignore files like `.hgignore` have changed since the previous |
|
141 | 142 | /// time a `status()` call wrote their hash to the dirstate. `None` means |
|
142 | 143 | /// we don’t know as this run doesn’t list either ignored or uknown files |
|
143 | 144 | /// and therefore isn’t reading `.hgignore`. |
|
144 | 145 | ignore_patterns_have_changed: Option<bool>, |
|
145 | 146 | |
|
146 | 147 | /// The current time at the start of the `status()` algorithm, as measured |
|
147 | 148 | /// and possibly truncated by the filesystem. |
|
148 |
filesystem_time_at_status_start: Option< |
|
|
149 | filesystem_time_at_status_start: Option<TruncatedTimestamp>, | |
|
149 | 150 | } |
|
150 | 151 | |
|
151 | 152 | enum Outcome { |
|
152 | 153 | Modified, |
|
153 | 154 | Added, |
|
154 | 155 | Removed, |
|
155 | 156 | Deleted, |
|
156 | 157 | Clean, |
|
157 | 158 | Ignored, |
|
158 | 159 | Unknown, |
|
159 | 160 | Unsure, |
|
160 | 161 | } |
|
161 | 162 | |
|
162 | 163 | impl<'a, 'tree, 'on_disk> StatusCommon<'a, 'tree, 'on_disk> { |
|
163 | 164 | fn push_outcome( |
|
164 | 165 | &self, |
|
165 | 166 | which: Outcome, |
|
166 | 167 | dirstate_node: &NodeRef<'tree, 'on_disk>, |
|
167 | 168 | ) -> Result<(), DirstateV2ParseError> { |
|
168 | 169 | let path = dirstate_node |
|
169 | 170 | .full_path_borrowed(self.dmap.on_disk)? |
|
170 | 171 | .detach_from_tree(); |
|
171 | 172 | let copy_source = if self.options.list_copies { |
|
172 | 173 | dirstate_node |
|
173 | 174 | .copy_source_borrowed(self.dmap.on_disk)? |
|
174 | 175 | .map(|source| source.detach_from_tree()) |
|
175 | 176 | } else { |
|
176 | 177 | None |
|
177 | 178 | }; |
|
178 | 179 | self.push_outcome_common(which, path, copy_source); |
|
179 | 180 | Ok(()) |
|
180 | 181 | } |
|
181 | 182 | |
|
182 | 183 | fn push_outcome_without_copy_source( |
|
183 | 184 | &self, |
|
184 | 185 | which: Outcome, |
|
185 | 186 | path: &BorrowedPath<'_, 'on_disk>, |
|
186 | 187 | ) { |
|
187 | 188 | self.push_outcome_common(which, path.detach_from_tree(), None) |
|
188 | 189 | } |
|
189 | 190 | |
|
190 | 191 | fn push_outcome_common( |
|
191 | 192 | &self, |
|
192 | 193 | which: Outcome, |
|
193 | 194 | path: HgPathCow<'on_disk>, |
|
194 | 195 | copy_source: Option<HgPathCow<'on_disk>>, |
|
195 | 196 | ) { |
|
196 | 197 | let mut outcome = self.outcome.lock().unwrap(); |
|
197 | 198 | let vec = match which { |
|
198 | 199 | Outcome::Modified => &mut outcome.modified, |
|
199 | 200 | Outcome::Added => &mut outcome.added, |
|
200 | 201 | Outcome::Removed => &mut outcome.removed, |
|
201 | 202 | Outcome::Deleted => &mut outcome.deleted, |
|
202 | 203 | Outcome::Clean => &mut outcome.clean, |
|
203 | 204 | Outcome::Ignored => &mut outcome.ignored, |
|
204 | 205 | Outcome::Unknown => &mut outcome.unknown, |
|
205 | 206 | Outcome::Unsure => &mut outcome.unsure, |
|
206 | 207 | }; |
|
207 | 208 | vec.push(StatusPath { path, copy_source }); |
|
208 | 209 | } |
|
209 | 210 | |
|
210 | 211 | fn read_dir( |
|
211 | 212 | &self, |
|
212 | 213 | hg_path: &HgPath, |
|
213 | 214 | fs_path: &Path, |
|
214 | 215 | is_at_repo_root: bool, |
|
215 | 216 | ) -> Result<Vec<DirEntry>, ()> { |
|
216 | 217 | DirEntry::read_dir(fs_path, is_at_repo_root) |
|
217 | 218 | .map_err(|error| self.io_error(error, hg_path)) |
|
218 | 219 | } |
|
219 | 220 | |
|
220 | 221 | fn io_error(&self, error: std::io::Error, hg_path: &HgPath) { |
|
221 | 222 | let errno = error.raw_os_error().expect("expected real OS error"); |
|
222 | 223 | self.outcome |
|
223 | 224 | .lock() |
|
224 | 225 | .unwrap() |
|
225 | 226 | .bad |
|
226 | 227 | .push((hg_path.to_owned().into(), BadMatch::OsError(errno))) |
|
227 | 228 | } |
|
228 | 229 | |
|
229 | 230 | fn check_for_outdated_directory_cache( |
|
230 | 231 | &self, |
|
231 | 232 | dirstate_node: &NodeRef<'tree, 'on_disk>, |
|
232 | 233 | ) -> Result<(), DirstateV2ParseError> { |
|
233 | 234 | if self.ignore_patterns_have_changed == Some(true) |
|
234 | 235 | && dirstate_node.cached_directory_mtime()?.is_some() |
|
235 | 236 | { |
|
236 | 237 | self.outated_cached_directories.lock().unwrap().push( |
|
237 | 238 | dirstate_node |
|
238 | 239 | .full_path_borrowed(self.dmap.on_disk)? |
|
239 | 240 | .detach_from_tree(), |
|
240 | 241 | ) |
|
241 | 242 | } |
|
242 | 243 | Ok(()) |
|
243 | 244 | } |
|
244 | 245 | |
|
245 | 246 | /// If this returns true, we can get accurate results by only using |
|
246 | 247 | /// `symlink_metadata` for child nodes that exist in the dirstate and don’t |
|
247 | 248 | /// need to call `read_dir`. |
|
248 | 249 | fn can_skip_fs_readdir( |
|
249 | 250 | &self, |
|
250 | 251 | directory_metadata: Option<&std::fs::Metadata>, |
|
251 | 252 | cached_directory_mtime: Option<TruncatedTimestamp>, |
|
252 | 253 | ) -> bool { |
|
253 | 254 | if !self.options.list_unknown && !self.options.list_ignored { |
|
254 | 255 | // All states that we care about listing have corresponding |
|
255 | 256 | // dirstate entries. |
|
256 | 257 | // This happens for example with `hg status -mard`. |
|
257 | 258 | return true; |
|
258 | 259 | } |
|
259 | 260 | if !self.options.list_ignored |
|
260 | 261 | && self.ignore_patterns_have_changed == Some(false) |
|
261 | 262 | { |
|
262 | 263 | if let Some(cached_mtime) = cached_directory_mtime { |
|
263 | 264 | // The dirstate contains a cached mtime for this directory, set |
|
264 | 265 | // by a previous run of the `status` algorithm which found this |
|
265 | 266 | // directory eligible for `read_dir` caching. |
|
266 | 267 | if let Some(meta) = directory_metadata { |
|
267 | 268 | if cached_mtime |
|
268 | 269 | .likely_equal_to_mtime_of(meta) |
|
269 | 270 | .unwrap_or(false) |
|
270 | 271 | { |
|
271 | 272 | // The mtime of that directory has not changed |
|
272 | 273 | // since then, which means that the results of |
|
273 | 274 | // `read_dir` should also be unchanged. |
|
274 | 275 | return true; |
|
275 | 276 | } |
|
276 | 277 | } |
|
277 | 278 | } |
|
278 | 279 | } |
|
279 | 280 | false |
|
280 | 281 | } |
|
281 | 282 | |
|
282 | 283 | /// Returns whether all child entries of the filesystem directory have a |
|
283 | 284 | /// corresponding dirstate node or are ignored. |
|
284 | 285 | fn traverse_fs_directory_and_dirstate( |
|
285 | 286 | &self, |
|
286 | 287 | has_ignored_ancestor: bool, |
|
287 | 288 | dirstate_nodes: ChildNodesRef<'tree, 'on_disk>, |
|
288 | 289 | directory_hg_path: &BorrowedPath<'tree, 'on_disk>, |
|
289 | 290 | directory_fs_path: &Path, |
|
290 | 291 | directory_metadata: Option<&std::fs::Metadata>, |
|
291 | 292 | cached_directory_mtime: Option<TruncatedTimestamp>, |
|
292 | 293 | is_at_repo_root: bool, |
|
293 | 294 | ) -> Result<bool, DirstateV2ParseError> { |
|
294 | 295 | if self.can_skip_fs_readdir(directory_metadata, cached_directory_mtime) |
|
295 | 296 | { |
|
296 | 297 | dirstate_nodes |
|
297 | 298 | .par_iter() |
|
298 | 299 | .map(|dirstate_node| { |
|
299 | 300 | let fs_path = directory_fs_path.join(get_path_from_bytes( |
|
300 | 301 | dirstate_node.base_name(self.dmap.on_disk)?.as_bytes(), |
|
301 | 302 | )); |
|
302 | 303 | match std::fs::symlink_metadata(&fs_path) { |
|
303 | 304 | Ok(fs_metadata) => self.traverse_fs_and_dirstate( |
|
304 | 305 | &fs_path, |
|
305 | 306 | &fs_metadata, |
|
306 | 307 | dirstate_node, |
|
307 | 308 | has_ignored_ancestor, |
|
308 | 309 | ), |
|
309 | 310 | Err(e) if e.kind() == std::io::ErrorKind::NotFound => { |
|
310 | 311 | self.traverse_dirstate_only(dirstate_node) |
|
311 | 312 | } |
|
312 | 313 | Err(error) => { |
|
313 | 314 | let hg_path = |
|
314 | 315 | dirstate_node.full_path(self.dmap.on_disk)?; |
|
315 | 316 | Ok(self.io_error(error, hg_path)) |
|
316 | 317 | } |
|
317 | 318 | } |
|
318 | 319 | }) |
|
319 | 320 | .collect::<Result<_, _>>()?; |
|
320 | 321 | |
|
321 | 322 | // We don’t know, so conservatively say this isn’t the case |
|
322 | 323 | let children_all_have_dirstate_node_or_are_ignored = false; |
|
323 | 324 | |
|
324 | 325 | return Ok(children_all_have_dirstate_node_or_are_ignored); |
|
325 | 326 | } |
|
326 | 327 | |
|
327 | 328 | let mut fs_entries = if let Ok(entries) = self.read_dir( |
|
328 | 329 | directory_hg_path, |
|
329 | 330 | directory_fs_path, |
|
330 | 331 | is_at_repo_root, |
|
331 | 332 | ) { |
|
332 | 333 | entries |
|
333 | 334 | } else { |
|
334 | 335 | // Treat an unreadable directory (typically because of insufficient |
|
335 | 336 | // permissions) like an empty directory. `self.read_dir` has |
|
336 | 337 | // already called `self.io_error` so a warning will be emitted. |
|
337 | 338 | Vec::new() |
|
338 | 339 | }; |
|
339 | 340 | |
|
340 | 341 | // `merge_join_by` requires both its input iterators to be sorted: |
|
341 | 342 | |
|
342 | 343 | let dirstate_nodes = dirstate_nodes.sorted(); |
|
343 | 344 | // `sort_unstable_by_key` doesn’t allow keys borrowing from the value: |
|
344 | 345 | // https://github.com/rust-lang/rust/issues/34162 |
|
345 | 346 | fs_entries.sort_unstable_by(|e1, e2| e1.base_name.cmp(&e2.base_name)); |
|
346 | 347 | |
|
347 | 348 | // Propagate here any error that would happen inside the comparison |
|
348 | 349 | // callback below |
|
349 | 350 | for dirstate_node in &dirstate_nodes { |
|
350 | 351 | dirstate_node.base_name(self.dmap.on_disk)?; |
|
351 | 352 | } |
|
352 | 353 | itertools::merge_join_by( |
|
353 | 354 | dirstate_nodes, |
|
354 | 355 | &fs_entries, |
|
355 | 356 | |dirstate_node, fs_entry| { |
|
356 | 357 | // This `unwrap` never panics because we already propagated |
|
357 | 358 | // those errors above |
|
358 | 359 | dirstate_node |
|
359 | 360 | .base_name(self.dmap.on_disk) |
|
360 | 361 | .unwrap() |
|
361 | 362 | .cmp(&fs_entry.base_name) |
|
362 | 363 | }, |
|
363 | 364 | ) |
|
364 | 365 | .par_bridge() |
|
365 | 366 | .map(|pair| { |
|
366 | 367 | use itertools::EitherOrBoth::*; |
|
367 | 368 | let has_dirstate_node_or_is_ignored; |
|
368 | 369 | match pair { |
|
369 | 370 | Both(dirstate_node, fs_entry) => { |
|
370 | 371 | self.traverse_fs_and_dirstate( |
|
371 | 372 | &fs_entry.full_path, |
|
372 | 373 | &fs_entry.metadata, |
|
373 | 374 | dirstate_node, |
|
374 | 375 | has_ignored_ancestor, |
|
375 | 376 | )?; |
|
376 | 377 | has_dirstate_node_or_is_ignored = true |
|
377 | 378 | } |
|
378 | 379 | Left(dirstate_node) => { |
|
379 | 380 | self.traverse_dirstate_only(dirstate_node)?; |
|
380 | 381 | has_dirstate_node_or_is_ignored = true; |
|
381 | 382 | } |
|
382 | 383 | Right(fs_entry) => { |
|
383 | 384 | has_dirstate_node_or_is_ignored = self.traverse_fs_only( |
|
384 | 385 | has_ignored_ancestor, |
|
385 | 386 | directory_hg_path, |
|
386 | 387 | fs_entry, |
|
387 | 388 | ) |
|
388 | 389 | } |
|
389 | 390 | } |
|
390 | 391 | Ok(has_dirstate_node_or_is_ignored) |
|
391 | 392 | }) |
|
392 | 393 | .try_reduce(|| true, |a, b| Ok(a && b)) |
|
393 | 394 | } |
|
394 | 395 | |
|
395 | 396 | fn traverse_fs_and_dirstate( |
|
396 | 397 | &self, |
|
397 | 398 | fs_path: &Path, |
|
398 | 399 | fs_metadata: &std::fs::Metadata, |
|
399 | 400 | dirstate_node: NodeRef<'tree, 'on_disk>, |
|
400 | 401 | has_ignored_ancestor: bool, |
|
401 | 402 | ) -> Result<(), DirstateV2ParseError> { |
|
402 | 403 | self.check_for_outdated_directory_cache(&dirstate_node)?; |
|
403 | 404 | let hg_path = &dirstate_node.full_path_borrowed(self.dmap.on_disk)?; |
|
404 | 405 | let file_type = fs_metadata.file_type(); |
|
405 | 406 | let file_or_symlink = file_type.is_file() || file_type.is_symlink(); |
|
406 | 407 | if !file_or_symlink { |
|
407 | 408 | // If we previously had a file here, it was removed (with |
|
408 | 409 | // `hg rm` or similar) or deleted before it could be |
|
409 | 410 | // replaced by a directory or something else. |
|
410 | 411 | self.mark_removed_or_deleted_if_file(&dirstate_node)?; |
|
411 | 412 | } |
|
412 | 413 | if file_type.is_dir() { |
|
413 | 414 | if self.options.collect_traversed_dirs { |
|
414 | 415 | self.outcome |
|
415 | 416 | .lock() |
|
416 | 417 | .unwrap() |
|
417 | 418 | .traversed |
|
418 | 419 | .push(hg_path.detach_from_tree()) |
|
419 | 420 | } |
|
420 | 421 | let is_ignored = has_ignored_ancestor || (self.ignore_fn)(hg_path); |
|
421 | 422 | let is_at_repo_root = false; |
|
422 | 423 | let children_all_have_dirstate_node_or_are_ignored = self |
|
423 | 424 | .traverse_fs_directory_and_dirstate( |
|
424 | 425 | is_ignored, |
|
425 | 426 | dirstate_node.children(self.dmap.on_disk)?, |
|
426 | 427 | hg_path, |
|
427 | 428 | fs_path, |
|
428 | 429 | Some(fs_metadata), |
|
429 | 430 | dirstate_node.cached_directory_mtime()?, |
|
430 | 431 | is_at_repo_root, |
|
431 | 432 | )?; |
|
432 | 433 | self.maybe_save_directory_mtime( |
|
433 | 434 | children_all_have_dirstate_node_or_are_ignored, |
|
434 | 435 | fs_metadata, |
|
435 | 436 | dirstate_node, |
|
436 | 437 | )? |
|
437 | 438 | } else { |
|
438 | 439 | if file_or_symlink && self.matcher.matches(hg_path) { |
|
439 | 440 | if let Some(state) = dirstate_node.state()? { |
|
440 | 441 | match state { |
|
441 | 442 | EntryState::Added => { |
|
442 | 443 | self.push_outcome(Outcome::Added, &dirstate_node)? |
|
443 | 444 | } |
|
444 | 445 | EntryState::Removed => self |
|
445 | 446 | .push_outcome(Outcome::Removed, &dirstate_node)?, |
|
446 | 447 | EntryState::Merged => self |
|
447 | 448 | .push_outcome(Outcome::Modified, &dirstate_node)?, |
|
448 | 449 | EntryState::Normal => self |
|
449 | 450 | .handle_normal_file(&dirstate_node, fs_metadata)?, |
|
450 | 451 | } |
|
451 | 452 | } else { |
|
452 | 453 | // `node.entry.is_none()` indicates a "directory" |
|
453 | 454 | // node, but the filesystem has a file |
|
454 | 455 | self.mark_unknown_or_ignored( |
|
455 | 456 | has_ignored_ancestor, |
|
456 | 457 | hg_path, |
|
457 | 458 | ); |
|
458 | 459 | } |
|
459 | 460 | } |
|
460 | 461 | |
|
461 | 462 | for child_node in dirstate_node.children(self.dmap.on_disk)?.iter() |
|
462 | 463 | { |
|
463 | 464 | self.traverse_dirstate_only(child_node)? |
|
464 | 465 | } |
|
465 | 466 | } |
|
466 | 467 | Ok(()) |
|
467 | 468 | } |
|
468 | 469 | |
|
469 | 470 | fn maybe_save_directory_mtime( |
|
470 | 471 | &self, |
|
471 | 472 | children_all_have_dirstate_node_or_are_ignored: bool, |
|
472 | 473 | directory_metadata: &std::fs::Metadata, |
|
473 | 474 | dirstate_node: NodeRef<'tree, 'on_disk>, |
|
474 | 475 | ) -> Result<(), DirstateV2ParseError> { |
|
475 | if children_all_have_dirstate_node_or_are_ignored { | |
|
476 | // All filesystem directory entries from `read_dir` have a | |
|
477 | // corresponding node in the dirstate, so we can reconstitute the | |
|
478 | // names of those entries without calling `read_dir` again. | |
|
479 | if let (Some(status_start), Ok(directory_mtime)) = ( | |
|
480 | &self.filesystem_time_at_status_start, | |
|
481 | directory_metadata.modified(), | |
|
476 | if !children_all_have_dirstate_node_or_are_ignored { | |
|
477 | return Ok(()); | |
|
478 | } | |
|
479 | // All filesystem directory entries from `read_dir` have a | |
|
480 | // corresponding node in the dirstate, so we can reconstitute the | |
|
481 | // names of those entries without calling `read_dir` again. | |
|
482 | ||
|
483 | // TODO: use let-else here and below when available: | |
|
484 | // https://github.com/rust-lang/rust/issues/87335 | |
|
485 | let status_start = if let Some(status_start) = | |
|
486 | &self.filesystem_time_at_status_start | |
|
487 | { | |
|
488 | status_start | |
|
489 | } else { | |
|
490 | return Ok(()); | |
|
491 | }; | |
|
492 | ||
|
493 | // Although the Rust standard library’s `SystemTime` type | |
|
494 | // has nanosecond precision, the times reported for a | |
|
495 | // directory’s (or file’s) modified time may have lower | |
|
496 | // resolution based on the filesystem (for example ext3 | |
|
497 | // only stores integer seconds), kernel (see | |
|
498 | // https://stackoverflow.com/a/14393315/1162888), etc. | |
|
499 | let directory_mtime = if let Ok(option) = | |
|
500 | TruncatedTimestamp::for_reliable_mtime_of( | |
|
501 | directory_metadata, | |
|
502 | status_start, | |
|
482 | 503 | ) { |
|
483 | // Although the Rust standard library’s `SystemTime` type | |
|
484 | // has nanosecond precision, the times reported for a | |
|
485 | // directory’s (or file’s) modified time may have lower | |
|
486 | // resolution based on the filesystem (for example ext3 | |
|
487 | // only stores integer seconds), kernel (see | |
|
488 | // https://stackoverflow.com/a/14393315/1162888), etc. | |
|
489 | if &directory_mtime >= status_start { | |
|
490 | // The directory was modified too recently, don’t cache its | |
|
491 | // `read_dir` results. | |
|
492 | // | |
|
493 | // A timeline like this is possible: | |
|
494 | // | |
|
495 | // 1. A change to this directory (direct child was | |
|
496 |
|
|
|
497 | // (possibly truncated) to `directory_mtime` | |
|
498 | // 2. This `status` algorithm calls `read_dir` | |
|
499 | // 3. An other change is made to the same directory is | |
|
500 | // made so that calling `read_dir` agin would give | |
|
501 | // different results, but soon enough after 1. that | |
|
502 | // the mtime stays the same | |
|
503 | // | |
|
504 | // On a system where the time resolution poor, this | |
|
505 | // scenario is not unlikely if all three steps are caused | |
|
506 | // by the same script. | |
|
507 | } else { | |
|
508 | // We’ve observed (through `status_start`) that time has | |
|
509 | // “progressed” since `directory_mtime`, so any further | |
|
510 | // change to this directory is extremely likely to cause a | |
|
511 | // different mtime. | |
|
512 | // | |
|
513 | // Having the same mtime again is not entirely impossible | |
|
514 | // since the system clock is not monotonous. It could jump | |
|
515 | // backward to some point before `directory_mtime`, then a | |
|
516 | // directory change could potentially happen during exactly | |
|
517 | // the wrong tick. | |
|
518 | // | |
|
519 | // We deem this scenario (unlike the previous one) to be | |
|
520 | // unlikely enough in practice. | |
|
521 | let truncated = TruncatedTimestamp::from(directory_mtime); | |
|
522 | let is_up_to_date = if let Some(cached) = | |
|
523 | dirstate_node.cached_directory_mtime()? | |
|
524 | { | |
|
525 | cached.likely_equal(truncated) | |
|
526 | } else { | |
|
527 | false | |
|
528 | }; | |
|
529 | if !is_up_to_date { | |
|
530 | let hg_path = dirstate_node | |
|
531 | .full_path_borrowed(self.dmap.on_disk)? | |
|
532 | .detach_from_tree(); | |
|
533 | self.new_cachable_directories | |
|
534 | .lock() | |
|
535 | .unwrap() | |
|
536 | .push((hg_path, truncated)) | |
|
537 | } | |
|
538 | } | |
|
504 | if let Some(directory_mtime) = option { | |
|
505 | directory_mtime | |
|
506 | } else { | |
|
507 | // The directory was modified too recently, | |
|
508 | // don’t cache its `read_dir` results. | |
|
509 | // | |
|
510 | // 1. A change to this directory (direct child was | |
|
511 | // added or removed) cause its mtime to be set | |
|
512 | // (possibly truncated) to `directory_mtime` | |
|
513 | // 2. This `status` algorithm calls `read_dir` | |
|
514 | // 3. An other change is made to the same directory is | |
|
515 | // made so that calling `read_dir` agin would give | |
|
516 | // different results, but soon enough after 1. that | |
|
517 | // the mtime stays the same | |
|
518 | // | |
|
519 | // On a system where the time resolution poor, this | |
|
520 | // scenario is not unlikely if all three steps are caused | |
|
521 | // by the same script. | |
|
522 | return Ok(()); | |
|
539 | 523 | } |
|
524 | } else { | |
|
525 | // OS/libc does not support mtime? | |
|
526 | return Ok(()); | |
|
527 | }; | |
|
528 | // We’ve observed (through `status_start`) that time has | |
|
529 | // “progressed” since `directory_mtime`, so any further | |
|
530 | // change to this directory is extremely likely to cause a | |
|
531 | // different mtime. | |
|
532 | // | |
|
533 | // Having the same mtime again is not entirely impossible | |
|
534 | // since the system clock is not monotonous. It could jump | |
|
535 | // backward to some point before `directory_mtime`, then a | |
|
536 | // directory change could potentially happen during exactly | |
|
537 | // the wrong tick. | |
|
538 | // | |
|
539 | // We deem this scenario (unlike the previous one) to be | |
|
540 | // unlikely enough in practice. | |
|
541 | ||
|
542 | let is_up_to_date = | |
|
543 | if let Some(cached) = dirstate_node.cached_directory_mtime()? { | |
|
544 | cached.likely_equal(directory_mtime) | |
|
545 | } else { | |
|
546 | false | |
|
547 | }; | |
|
548 | if !is_up_to_date { | |
|
549 | let hg_path = dirstate_node | |
|
550 | .full_path_borrowed(self.dmap.on_disk)? | |
|
551 | .detach_from_tree(); | |
|
552 | self.new_cachable_directories | |
|
553 | .lock() | |
|
554 | .unwrap() | |
|
555 | .push((hg_path, directory_mtime)) | |
|
540 | 556 | } |
|
541 | 557 | Ok(()) |
|
542 | 558 | } |
|
543 | 559 | |
|
544 | 560 | /// A file with `EntryState::Normal` in the dirstate was found in the |
|
545 | 561 | /// filesystem |
|
546 | 562 | fn handle_normal_file( |
|
547 | 563 | &self, |
|
548 | 564 | dirstate_node: &NodeRef<'tree, 'on_disk>, |
|
549 | 565 | fs_metadata: &std::fs::Metadata, |
|
550 | 566 | ) -> Result<(), DirstateV2ParseError> { |
|
551 | 567 | // Keep the low 31 bits |
|
552 | 568 | fn truncate_u64(value: u64) -> i32 { |
|
553 | 569 | (value & 0x7FFF_FFFF) as i32 |
|
554 | 570 | } |
|
555 | 571 | |
|
556 | 572 | let entry = dirstate_node |
|
557 | 573 | .entry()? |
|
558 | 574 | .expect("handle_normal_file called with entry-less node"); |
|
559 | 575 | let mode_changed = |
|
560 | 576 | || self.options.check_exec && entry.mode_changed(fs_metadata); |
|
561 | 577 | let size = entry.size(); |
|
562 | 578 | let size_changed = size != truncate_u64(fs_metadata.len()); |
|
563 | 579 | if size >= 0 && size_changed && fs_metadata.file_type().is_symlink() { |
|
564 | 580 | // issue6456: Size returned may be longer due to encryption |
|
565 | 581 | // on EXT-4 fscrypt. TODO maybe only do it on EXT4? |
|
566 | 582 | self.push_outcome(Outcome::Unsure, dirstate_node)? |
|
567 | 583 | } else if dirstate_node.has_copy_source() |
|
568 | 584 | || entry.is_from_other_parent() |
|
569 | 585 | || (size >= 0 && (size_changed || mode_changed())) |
|
570 | 586 | { |
|
571 | 587 | self.push_outcome(Outcome::Modified, dirstate_node)? |
|
572 | 588 | } else { |
|
573 | 589 | let mtime_looks_clean; |
|
574 | 590 | if let Some(dirstate_mtime) = entry.truncated_mtime() { |
|
575 | 591 | let fs_mtime = TruncatedTimestamp::for_mtime_of(fs_metadata) |
|
576 | 592 | .expect("OS/libc does not support mtime?"); |
|
577 | 593 | // There might be a change in the future if for example the |
|
578 | 594 | // internal clock become off while process run, but this is a |
|
579 | 595 | // case where the issues the user would face |
|
580 | 596 | // would be a lot worse and there is nothing we |
|
581 | 597 | // can really do. |
|
582 | 598 | mtime_looks_clean = fs_mtime.likely_equal(dirstate_mtime) |
|
583 | 599 | } else { |
|
584 | 600 | // No mtime in the dirstate entry |
|
585 | 601 | mtime_looks_clean = false |
|
586 | 602 | }; |
|
587 | 603 | if !mtime_looks_clean { |
|
588 | 604 | self.push_outcome(Outcome::Unsure, dirstate_node)? |
|
589 | 605 | } else if self.options.list_clean { |
|
590 | 606 | self.push_outcome(Outcome::Clean, dirstate_node)? |
|
591 | 607 | } |
|
592 | 608 | } |
|
593 | 609 | Ok(()) |
|
594 | 610 | } |
|
595 | 611 | |
|
596 | 612 | /// A node in the dirstate tree has no corresponding filesystem entry |
|
597 | 613 | fn traverse_dirstate_only( |
|
598 | 614 | &self, |
|
599 | 615 | dirstate_node: NodeRef<'tree, 'on_disk>, |
|
600 | 616 | ) -> Result<(), DirstateV2ParseError> { |
|
601 | 617 | self.check_for_outdated_directory_cache(&dirstate_node)?; |
|
602 | 618 | self.mark_removed_or_deleted_if_file(&dirstate_node)?; |
|
603 | 619 | dirstate_node |
|
604 | 620 | .children(self.dmap.on_disk)? |
|
605 | 621 | .par_iter() |
|
606 | 622 | .map(|child_node| self.traverse_dirstate_only(child_node)) |
|
607 | 623 | .collect() |
|
608 | 624 | } |
|
609 | 625 | |
|
610 | 626 | /// A node in the dirstate tree has no corresponding *file* on the |
|
611 | 627 | /// filesystem |
|
612 | 628 | /// |
|
613 | 629 | /// Does nothing on a "directory" node |
|
614 | 630 | fn mark_removed_or_deleted_if_file( |
|
615 | 631 | &self, |
|
616 | 632 | dirstate_node: &NodeRef<'tree, 'on_disk>, |
|
617 | 633 | ) -> Result<(), DirstateV2ParseError> { |
|
618 | 634 | if let Some(state) = dirstate_node.state()? { |
|
619 | 635 | let path = dirstate_node.full_path(self.dmap.on_disk)?; |
|
620 | 636 | if self.matcher.matches(path) { |
|
621 | 637 | if let EntryState::Removed = state { |
|
622 | 638 | self.push_outcome(Outcome::Removed, dirstate_node)? |
|
623 | 639 | } else { |
|
624 | 640 | self.push_outcome(Outcome::Deleted, &dirstate_node)? |
|
625 | 641 | } |
|
626 | 642 | } |
|
627 | 643 | } |
|
628 | 644 | Ok(()) |
|
629 | 645 | } |
|
630 | 646 | |
|
631 | 647 | /// Something in the filesystem has no corresponding dirstate node |
|
632 | 648 | /// |
|
633 | 649 | /// Returns whether that path is ignored |
|
634 | 650 | fn traverse_fs_only( |
|
635 | 651 | &self, |
|
636 | 652 | has_ignored_ancestor: bool, |
|
637 | 653 | directory_hg_path: &HgPath, |
|
638 | 654 | fs_entry: &DirEntry, |
|
639 | 655 | ) -> bool { |
|
640 | 656 | let hg_path = directory_hg_path.join(&fs_entry.base_name); |
|
641 | 657 | let file_type = fs_entry.metadata.file_type(); |
|
642 | 658 | let file_or_symlink = file_type.is_file() || file_type.is_symlink(); |
|
643 | 659 | if file_type.is_dir() { |
|
644 | 660 | let is_ignored = |
|
645 | 661 | has_ignored_ancestor || (self.ignore_fn)(&hg_path); |
|
646 | 662 | let traverse_children = if is_ignored { |
|
647 | 663 | // Descendants of an ignored directory are all ignored |
|
648 | 664 | self.options.list_ignored |
|
649 | 665 | } else { |
|
650 | 666 | // Descendants of an unknown directory may be either unknown or |
|
651 | 667 | // ignored |
|
652 | 668 | self.options.list_unknown || self.options.list_ignored |
|
653 | 669 | }; |
|
654 | 670 | if traverse_children { |
|
655 | 671 | let is_at_repo_root = false; |
|
656 | 672 | if let Ok(children_fs_entries) = self.read_dir( |
|
657 | 673 | &hg_path, |
|
658 | 674 | &fs_entry.full_path, |
|
659 | 675 | is_at_repo_root, |
|
660 | 676 | ) { |
|
661 | 677 | children_fs_entries.par_iter().for_each(|child_fs_entry| { |
|
662 | 678 | self.traverse_fs_only( |
|
663 | 679 | is_ignored, |
|
664 | 680 | &hg_path, |
|
665 | 681 | child_fs_entry, |
|
666 | 682 | ); |
|
667 | 683 | }) |
|
668 | 684 | } |
|
669 | 685 | } |
|
670 | 686 | if self.options.collect_traversed_dirs { |
|
671 | 687 | self.outcome.lock().unwrap().traversed.push(hg_path.into()) |
|
672 | 688 | } |
|
673 | 689 | is_ignored |
|
674 | 690 | } else { |
|
675 | 691 | if file_or_symlink { |
|
676 | 692 | if self.matcher.matches(&hg_path) { |
|
677 | 693 | self.mark_unknown_or_ignored( |
|
678 | 694 | has_ignored_ancestor, |
|
679 | 695 | &BorrowedPath::InMemory(&hg_path), |
|
680 | 696 | ) |
|
681 | 697 | } else { |
|
682 | 698 | // We haven’t computed whether this path is ignored. It |
|
683 | 699 | // might not be, and a future run of status might have a |
|
684 | 700 | // different matcher that matches it. So treat it as not |
|
685 | 701 | // ignored. That is, inhibit readdir caching of the parent |
|
686 | 702 | // directory. |
|
687 | 703 | false |
|
688 | 704 | } |
|
689 | 705 | } else { |
|
690 | 706 | // This is neither a directory, a plain file, or a symlink. |
|
691 | 707 | // Treat it like an ignored file. |
|
692 | 708 | true |
|
693 | 709 | } |
|
694 | 710 | } |
|
695 | 711 | } |
|
696 | 712 | |
|
697 | 713 | /// Returns whether that path is ignored |
|
698 | 714 | fn mark_unknown_or_ignored( |
|
699 | 715 | &self, |
|
700 | 716 | has_ignored_ancestor: bool, |
|
701 | 717 | hg_path: &BorrowedPath<'_, 'on_disk>, |
|
702 | 718 | ) -> bool { |
|
703 | 719 | let is_ignored = has_ignored_ancestor || (self.ignore_fn)(&hg_path); |
|
704 | 720 | if is_ignored { |
|
705 | 721 | if self.options.list_ignored { |
|
706 | 722 | self.push_outcome_without_copy_source( |
|
707 | 723 | Outcome::Ignored, |
|
708 | 724 | hg_path, |
|
709 | 725 | ) |
|
710 | 726 | } |
|
711 | 727 | } else { |
|
712 | 728 | if self.options.list_unknown { |
|
713 | 729 | self.push_outcome_without_copy_source( |
|
714 | 730 | Outcome::Unknown, |
|
715 | 731 | hg_path, |
|
716 | 732 | ) |
|
717 | 733 | } |
|
718 | 734 | } |
|
719 | 735 | is_ignored |
|
720 | 736 | } |
|
721 | 737 | } |
|
722 | 738 | |
|
723 | 739 | struct DirEntry { |
|
724 | 740 | base_name: HgPathBuf, |
|
725 | 741 | full_path: PathBuf, |
|
726 | 742 | metadata: std::fs::Metadata, |
|
727 | 743 | } |
|
728 | 744 | |
|
729 | 745 | impl DirEntry { |
|
730 | 746 | /// Returns **unsorted** entries in the given directory, with name and |
|
731 | 747 | /// metadata. |
|
732 | 748 | /// |
|
733 | 749 | /// If a `.hg` sub-directory is encountered: |
|
734 | 750 | /// |
|
735 | 751 | /// * At the repository root, ignore that sub-directory |
|
736 | 752 | /// * Elsewhere, we’re listing the content of a sub-repo. Return an empty |
|
737 | 753 | /// list instead. |
|
738 | 754 | fn read_dir(path: &Path, is_at_repo_root: bool) -> io::Result<Vec<Self>> { |
|
739 | 755 | let mut results = Vec::new(); |
|
740 | 756 | for entry in path.read_dir()? { |
|
741 | 757 | let entry = entry?; |
|
742 | 758 | let metadata = entry.metadata()?; |
|
743 | 759 | let name = get_bytes_from_os_string(entry.file_name()); |
|
744 | 760 | // FIXME don't do this when cached |
|
745 | 761 | if name == b".hg" { |
|
746 | 762 | if is_at_repo_root { |
|
747 | 763 | // Skip the repo’s own .hg (might be a symlink) |
|
748 | 764 | continue; |
|
749 | 765 | } else if metadata.is_dir() { |
|
750 | 766 | // A .hg sub-directory at another location means a subrepo, |
|
751 | 767 | // skip it entirely. |
|
752 | 768 | return Ok(Vec::new()); |
|
753 | 769 | } |
|
754 | 770 | } |
|
755 | 771 | results.push(DirEntry { |
|
756 | 772 | base_name: name.into(), |
|
757 | 773 | full_path: entry.path(), |
|
758 | 774 | metadata, |
|
759 | 775 | }) |
|
760 | 776 | } |
|
761 | 777 | Ok(results) |
|
762 | 778 | } |
|
763 | 779 | } |
|
764 | 780 | |
|
765 | 781 | /// Return the `mtime` of a temporary file newly-created in the `.hg` directory |
|
766 | 782 | /// of the give repository. |
|
767 | 783 | /// |
|
768 | 784 | /// This is similar to `SystemTime::now()`, with the result truncated to the |
|
769 | 785 | /// same time resolution as other files’ modification times. Using `.hg` |
|
770 | 786 | /// instead of the system’s default temporary directory (such as `/tmp`) makes |
|
771 | 787 | /// it more likely the temporary file is in the same disk partition as contents |
|
772 | 788 | /// of the working directory, which can matter since different filesystems may |
|
773 | 789 | /// store timestamps with different resolutions. |
|
774 | 790 | /// |
|
775 | 791 | /// This may fail, typically if we lack write permissions. In that case we |
|
776 | 792 | /// should continue the `status()` algoritm anyway and consider the current |
|
777 | 793 | /// date/time to be unknown. |
|
778 | 794 | fn filesystem_now(repo_root: &Path) -> Result<SystemTime, io::Error> { |
|
779 | 795 | tempfile::tempfile_in(repo_root.join(".hg"))? |
|
780 | 796 | .metadata()? |
|
781 | 797 | .modified() |
|
782 | 798 | } |
@@ -1,506 +1,505 b'' | |||
|
1 | 1 | // status.rs |
|
2 | 2 | // |
|
3 | 3 | // Copyright 2020, Georges Racinet <georges.racinets@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 crate::error::CommandError; |
|
9 | 9 | use crate::ui::Ui; |
|
10 | 10 | use crate::utils::path_utils::RelativizePaths; |
|
11 | 11 | use clap::{Arg, SubCommand}; |
|
12 | 12 | use format_bytes::format_bytes; |
|
13 | 13 | use hg; |
|
14 | 14 | use hg::config::Config; |
|
15 | 15 | use hg::dirstate::has_exec_bit; |
|
16 | 16 | use hg::dirstate::status::StatusPath; |
|
17 | 17 | use hg::dirstate::TruncatedTimestamp; |
|
18 | 18 | use hg::dirstate::RANGE_MASK_31BIT; |
|
19 | 19 | use hg::errors::{HgError, IoResultExt}; |
|
20 | 20 | use hg::lock::LockError; |
|
21 | 21 | use hg::manifest::Manifest; |
|
22 | 22 | use hg::matchers::AlwaysMatcher; |
|
23 | 23 | use hg::repo::Repo; |
|
24 | 24 | use hg::utils::files::get_bytes_from_os_string; |
|
25 | 25 | use hg::utils::files::get_path_from_bytes; |
|
26 | 26 | use hg::utils::hg_path::{hg_path_to_path_buf, HgPath}; |
|
27 | 27 | use hg::StatusOptions; |
|
28 | 28 | use log::{info, warn}; |
|
29 | 29 | use std::io; |
|
30 | 30 | use std::path::PathBuf; |
|
31 | 31 | |
|
32 | 32 | pub const HELP_TEXT: &str = " |
|
33 | 33 | Show changed files in the working directory |
|
34 | 34 | |
|
35 | 35 | This is a pure Rust version of `hg status`. |
|
36 | 36 | |
|
37 | 37 | Some options might be missing, check the list below. |
|
38 | 38 | "; |
|
39 | 39 | |
|
40 | 40 | pub fn args() -> clap::App<'static, 'static> { |
|
41 | 41 | SubCommand::with_name("status") |
|
42 | 42 | .alias("st") |
|
43 | 43 | .about(HELP_TEXT) |
|
44 | 44 | .arg( |
|
45 | 45 | Arg::with_name("all") |
|
46 | 46 | .help("show status of all files") |
|
47 | 47 | .short("-A") |
|
48 | 48 | .long("--all"), |
|
49 | 49 | ) |
|
50 | 50 | .arg( |
|
51 | 51 | Arg::with_name("modified") |
|
52 | 52 | .help("show only modified files") |
|
53 | 53 | .short("-m") |
|
54 | 54 | .long("--modified"), |
|
55 | 55 | ) |
|
56 | 56 | .arg( |
|
57 | 57 | Arg::with_name("added") |
|
58 | 58 | .help("show only added files") |
|
59 | 59 | .short("-a") |
|
60 | 60 | .long("--added"), |
|
61 | 61 | ) |
|
62 | 62 | .arg( |
|
63 | 63 | Arg::with_name("removed") |
|
64 | 64 | .help("show only removed files") |
|
65 | 65 | .short("-r") |
|
66 | 66 | .long("--removed"), |
|
67 | 67 | ) |
|
68 | 68 | .arg( |
|
69 | 69 | Arg::with_name("clean") |
|
70 | 70 | .help("show only clean files") |
|
71 | 71 | .short("-c") |
|
72 | 72 | .long("--clean"), |
|
73 | 73 | ) |
|
74 | 74 | .arg( |
|
75 | 75 | Arg::with_name("deleted") |
|
76 | 76 | .help("show only deleted files") |
|
77 | 77 | .short("-d") |
|
78 | 78 | .long("--deleted"), |
|
79 | 79 | ) |
|
80 | 80 | .arg( |
|
81 | 81 | Arg::with_name("unknown") |
|
82 | 82 | .help("show only unknown (not tracked) files") |
|
83 | 83 | .short("-u") |
|
84 | 84 | .long("--unknown"), |
|
85 | 85 | ) |
|
86 | 86 | .arg( |
|
87 | 87 | Arg::with_name("ignored") |
|
88 | 88 | .help("show only ignored files") |
|
89 | 89 | .short("-i") |
|
90 | 90 | .long("--ignored"), |
|
91 | 91 | ) |
|
92 | 92 | .arg( |
|
93 | 93 | Arg::with_name("copies") |
|
94 | 94 | .help("show source of copied files (DEFAULT: ui.statuscopies)") |
|
95 | 95 | .short("-C") |
|
96 | 96 | .long("--copies"), |
|
97 | 97 | ) |
|
98 | 98 | .arg( |
|
99 | 99 | Arg::with_name("no-status") |
|
100 | 100 | .help("hide status prefix") |
|
101 | 101 | .short("-n") |
|
102 | 102 | .long("--no-status"), |
|
103 | 103 | ) |
|
104 | 104 | } |
|
105 | 105 | |
|
106 | 106 | /// Pure data type allowing the caller to specify file states to display |
|
107 | 107 | #[derive(Copy, Clone, Debug)] |
|
108 | 108 | pub struct DisplayStates { |
|
109 | 109 | pub modified: bool, |
|
110 | 110 | pub added: bool, |
|
111 | 111 | pub removed: bool, |
|
112 | 112 | pub clean: bool, |
|
113 | 113 | pub deleted: bool, |
|
114 | 114 | pub unknown: bool, |
|
115 | 115 | pub ignored: bool, |
|
116 | 116 | } |
|
117 | 117 | |
|
118 | 118 | pub const DEFAULT_DISPLAY_STATES: DisplayStates = DisplayStates { |
|
119 | 119 | modified: true, |
|
120 | 120 | added: true, |
|
121 | 121 | removed: true, |
|
122 | 122 | clean: false, |
|
123 | 123 | deleted: true, |
|
124 | 124 | unknown: true, |
|
125 | 125 | ignored: false, |
|
126 | 126 | }; |
|
127 | 127 | |
|
128 | 128 | pub const ALL_DISPLAY_STATES: DisplayStates = DisplayStates { |
|
129 | 129 | modified: true, |
|
130 | 130 | added: true, |
|
131 | 131 | removed: true, |
|
132 | 132 | clean: true, |
|
133 | 133 | deleted: true, |
|
134 | 134 | unknown: true, |
|
135 | 135 | ignored: true, |
|
136 | 136 | }; |
|
137 | 137 | |
|
138 | 138 | impl DisplayStates { |
|
139 | 139 | pub fn is_empty(&self) -> bool { |
|
140 | 140 | !(self.modified |
|
141 | 141 | || self.added |
|
142 | 142 | || self.removed |
|
143 | 143 | || self.clean |
|
144 | 144 | || self.deleted |
|
145 | 145 | || self.unknown |
|
146 | 146 | || self.ignored) |
|
147 | 147 | } |
|
148 | 148 | } |
|
149 | 149 | |
|
150 | 150 | pub fn run(invocation: &crate::CliInvocation) -> Result<(), CommandError> { |
|
151 | 151 | let status_enabled_default = false; |
|
152 | 152 | let status_enabled = invocation.config.get_option(b"rhg", b"status")?; |
|
153 | 153 | if !status_enabled.unwrap_or(status_enabled_default) { |
|
154 | 154 | return Err(CommandError::unsupported( |
|
155 | 155 | "status is experimental in rhg (enable it with 'rhg.status = true' \ |
|
156 | 156 | or enable fallback with 'rhg.on-unsupported = fallback')" |
|
157 | 157 | )); |
|
158 | 158 | } |
|
159 | 159 | |
|
160 | 160 | // TODO: lift these limitations |
|
161 | 161 | if invocation.config.get_bool(b"ui", b"tweakdefaults")? { |
|
162 | 162 | return Err(CommandError::unsupported( |
|
163 | 163 | "ui.tweakdefaults is not yet supported with rhg status", |
|
164 | 164 | )); |
|
165 | 165 | } |
|
166 | 166 | if invocation.config.get_bool(b"ui", b"statuscopies")? { |
|
167 | 167 | return Err(CommandError::unsupported( |
|
168 | 168 | "ui.statuscopies is not yet supported with rhg status", |
|
169 | 169 | )); |
|
170 | 170 | } |
|
171 | 171 | if invocation |
|
172 | 172 | .config |
|
173 | 173 | .get(b"commands", b"status.terse") |
|
174 | 174 | .is_some() |
|
175 | 175 | { |
|
176 | 176 | return Err(CommandError::unsupported( |
|
177 | 177 | "status.terse is not yet supported with rhg status", |
|
178 | 178 | )); |
|
179 | 179 | } |
|
180 | 180 | |
|
181 | 181 | let ui = invocation.ui; |
|
182 | 182 | let config = invocation.config; |
|
183 | 183 | let args = invocation.subcommand_args; |
|
184 | 184 | let all = args.is_present("all"); |
|
185 | 185 | let display_states = if all { |
|
186 | 186 | // TODO when implementing `--quiet`: it excludes clean files |
|
187 | 187 | // from `--all` |
|
188 | 188 | ALL_DISPLAY_STATES |
|
189 | 189 | } else { |
|
190 | 190 | let requested = DisplayStates { |
|
191 | 191 | modified: args.is_present("modified"), |
|
192 | 192 | added: args.is_present("added"), |
|
193 | 193 | removed: args.is_present("removed"), |
|
194 | 194 | clean: args.is_present("clean"), |
|
195 | 195 | deleted: args.is_present("deleted"), |
|
196 | 196 | unknown: args.is_present("unknown"), |
|
197 | 197 | ignored: args.is_present("ignored"), |
|
198 | 198 | }; |
|
199 | 199 | if requested.is_empty() { |
|
200 | 200 | DEFAULT_DISPLAY_STATES |
|
201 | 201 | } else { |
|
202 | 202 | requested |
|
203 | 203 | } |
|
204 | 204 | }; |
|
205 | 205 | let no_status = args.is_present("no-status"); |
|
206 | 206 | let list_copies = all |
|
207 | 207 | || args.is_present("copies") |
|
208 | 208 | || config.get_bool(b"ui", b"statuscopies")?; |
|
209 | 209 | |
|
210 | 210 | let repo = invocation.repo?; |
|
211 | 211 | |
|
212 | 212 | if repo.has_sparse() || repo.has_narrow() { |
|
213 | 213 | return Err(CommandError::unsupported( |
|
214 | 214 | "rhg status is not supported for sparse checkouts or narrow clones yet" |
|
215 | 215 | )); |
|
216 | 216 | } |
|
217 | 217 | |
|
218 | 218 | let mut dmap = repo.dirstate_map_mut()?; |
|
219 | 219 | |
|
220 | 220 | let options = StatusOptions { |
|
221 | 221 | // we're currently supporting file systems with exec flags only |
|
222 | 222 | // anyway |
|
223 | 223 | check_exec: true, |
|
224 | 224 | list_clean: display_states.clean, |
|
225 | 225 | list_unknown: display_states.unknown, |
|
226 | 226 | list_ignored: display_states.ignored, |
|
227 | 227 | list_copies, |
|
228 | 228 | collect_traversed_dirs: false, |
|
229 | 229 | }; |
|
230 | 230 | let (mut ds_status, pattern_warnings) = dmap.status( |
|
231 | 231 | &AlwaysMatcher, |
|
232 | 232 | repo.working_directory_path().to_owned(), |
|
233 | 233 | ignore_files(repo, config), |
|
234 | 234 | options, |
|
235 | 235 | )?; |
|
236 | 236 | if !pattern_warnings.is_empty() { |
|
237 | 237 | warn!("Pattern warnings: {:?}", &pattern_warnings); |
|
238 | 238 | } |
|
239 | 239 | |
|
240 | 240 | for (path, error) in ds_status.bad { |
|
241 | 241 | let error = match error { |
|
242 | 242 | hg::BadMatch::OsError(code) => { |
|
243 | 243 | std::io::Error::from_raw_os_error(code).to_string() |
|
244 | 244 | } |
|
245 | 245 | hg::BadMatch::BadType(ty) => { |
|
246 | 246 | format!("unsupported file type (type is {})", ty) |
|
247 | 247 | } |
|
248 | 248 | }; |
|
249 | 249 | ui.write_stderr(&format_bytes!( |
|
250 | 250 | b"{}: {}\n", |
|
251 | 251 | path.as_bytes(), |
|
252 | 252 | error.as_bytes() |
|
253 | 253 | ))? |
|
254 | 254 | } |
|
255 | 255 | if !ds_status.unsure.is_empty() { |
|
256 | 256 | info!( |
|
257 | 257 | "Files to be rechecked by retrieval from filelog: {:?}", |
|
258 | 258 | ds_status.unsure.iter().map(|s| &s.path).collect::<Vec<_>>() |
|
259 | 259 | ); |
|
260 | 260 | } |
|
261 | 261 | let mut fixup = Vec::new(); |
|
262 | 262 | if !ds_status.unsure.is_empty() |
|
263 | 263 | && (display_states.modified || display_states.clean) |
|
264 | 264 | { |
|
265 | 265 | let p1 = repo.dirstate_parents()?.p1; |
|
266 | 266 | let manifest = repo.manifest_for_node(p1).map_err(|e| { |
|
267 | 267 | CommandError::from((e, &*format!("{:x}", p1.short()))) |
|
268 | 268 | })?; |
|
269 | 269 | for to_check in ds_status.unsure { |
|
270 | 270 | if unsure_is_modified(repo, &manifest, &to_check.path)? { |
|
271 | 271 | if display_states.modified { |
|
272 | 272 | ds_status.modified.push(to_check); |
|
273 | 273 | } |
|
274 | 274 | } else { |
|
275 | 275 | if display_states.clean { |
|
276 | 276 | ds_status.clean.push(to_check.clone()); |
|
277 | 277 | } |
|
278 | 278 | fixup.push(to_check.path.into_owned()) |
|
279 | 279 | } |
|
280 | 280 | } |
|
281 | 281 | } |
|
282 | 282 | let relative_paths = (!ui.plain()) |
|
283 | 283 | && config |
|
284 | 284 | .get_option(b"commands", b"status.relative")? |
|
285 | 285 | .unwrap_or(config.get_bool(b"ui", b"relative-paths")?); |
|
286 | 286 | let output = DisplayStatusPaths { |
|
287 | 287 | ui, |
|
288 | 288 | no_status, |
|
289 | 289 | relativize: if relative_paths { |
|
290 | 290 | Some(RelativizePaths::new(repo)?) |
|
291 | 291 | } else { |
|
292 | 292 | None |
|
293 | 293 | }, |
|
294 | 294 | }; |
|
295 | 295 | if display_states.modified { |
|
296 | 296 | output.display(b"M", ds_status.modified)?; |
|
297 | 297 | } |
|
298 | 298 | if display_states.added { |
|
299 | 299 | output.display(b"A", ds_status.added)?; |
|
300 | 300 | } |
|
301 | 301 | if display_states.removed { |
|
302 | 302 | output.display(b"R", ds_status.removed)?; |
|
303 | 303 | } |
|
304 | 304 | if display_states.deleted { |
|
305 | 305 | output.display(b"!", ds_status.deleted)?; |
|
306 | 306 | } |
|
307 | 307 | if display_states.unknown { |
|
308 | 308 | output.display(b"?", ds_status.unknown)?; |
|
309 | 309 | } |
|
310 | 310 | if display_states.ignored { |
|
311 | 311 | output.display(b"I", ds_status.ignored)?; |
|
312 | 312 | } |
|
313 | 313 | if display_states.clean { |
|
314 | 314 | output.display(b"C", ds_status.clean)?; |
|
315 | 315 | } |
|
316 | 316 | |
|
317 | 317 | let mut dirstate_write_needed = ds_status.dirty; |
|
318 |
let filesystem_time_at_status_start = |
|
|
319 | .filesystem_time_at_status_start | |
|
320 | .map(TruncatedTimestamp::from); | |
|
318 | let filesystem_time_at_status_start = | |
|
319 | ds_status.filesystem_time_at_status_start; | |
|
321 | 320 | |
|
322 | 321 | if (fixup.is_empty() || filesystem_time_at_status_start.is_none()) |
|
323 | 322 | && !dirstate_write_needed |
|
324 | 323 | { |
|
325 | 324 | // Nothing to update |
|
326 | 325 | return Ok(()); |
|
327 | 326 | } |
|
328 | 327 | |
|
329 | 328 | // Update the dirstate on disk if we can |
|
330 | 329 | let with_lock_result = |
|
331 | 330 | repo.try_with_wlock_no_wait(|| -> Result<(), CommandError> { |
|
332 | 331 | if let Some(mtime_boundary) = filesystem_time_at_status_start { |
|
333 | 332 | for hg_path in fixup { |
|
334 | 333 | use std::os::unix::fs::MetadataExt; |
|
335 | 334 | let fs_path = hg_path_to_path_buf(&hg_path) |
|
336 | 335 | .expect("HgPath conversion"); |
|
337 | 336 | // Specifically do not reuse `fs_metadata` from |
|
338 | 337 | // `unsure_is_clean` which was needed before reading |
|
339 | 338 | // contents. Here we access metadata again after reading |
|
340 | 339 | // content, in case it changed in the meantime. |
|
341 | 340 | let fs_metadata = repo |
|
342 | 341 | .working_directory_vfs() |
|
343 | 342 | .symlink_metadata(&fs_path)?; |
|
344 | 343 | if let Some(mtime) = |
|
345 | 344 | TruncatedTimestamp::for_reliable_mtime_of( |
|
346 | 345 | &fs_metadata, |
|
347 | 346 | &mtime_boundary, |
|
348 | 347 | ) |
|
349 | 348 | .when_reading_file(&fs_path)? |
|
350 | 349 | { |
|
351 | 350 | let mode = fs_metadata.mode(); |
|
352 | 351 | let size = fs_metadata.len() as u32 & RANGE_MASK_31BIT; |
|
353 | 352 | let mut entry = dmap |
|
354 | 353 | .get(&hg_path)? |
|
355 | 354 | .expect("ambiguous file not in dirstate"); |
|
356 | 355 | entry.set_clean(mode, size, mtime); |
|
357 | 356 | dmap.add_file(&hg_path, entry)?; |
|
358 | 357 | dirstate_write_needed = true |
|
359 | 358 | } |
|
360 | 359 | } |
|
361 | 360 | } |
|
362 | 361 | drop(dmap); // Avoid "already mutably borrowed" RefCell panics |
|
363 | 362 | if dirstate_write_needed { |
|
364 | 363 | repo.write_dirstate()? |
|
365 | 364 | } |
|
366 | 365 | Ok(()) |
|
367 | 366 | }); |
|
368 | 367 | match with_lock_result { |
|
369 | 368 | Ok(closure_result) => closure_result?, |
|
370 | 369 | Err(LockError::AlreadyHeld) => { |
|
371 | 370 | // Not updating the dirstate is not ideal but not critical: |
|
372 | 371 | // don’t keep our caller waiting until some other Mercurial |
|
373 | 372 | // process releases the lock. |
|
374 | 373 | } |
|
375 | 374 | Err(LockError::Other(HgError::IoError { error, .. })) |
|
376 | 375 | if error.kind() == io::ErrorKind::PermissionDenied => |
|
377 | 376 | { |
|
378 | 377 | // `hg status` on a read-only repository is fine |
|
379 | 378 | } |
|
380 | 379 | Err(LockError::Other(error)) => { |
|
381 | 380 | // Report other I/O errors |
|
382 | 381 | Err(error)? |
|
383 | 382 | } |
|
384 | 383 | } |
|
385 | 384 | Ok(()) |
|
386 | 385 | } |
|
387 | 386 | |
|
388 | 387 | fn ignore_files(repo: &Repo, config: &Config) -> Vec<PathBuf> { |
|
389 | 388 | let mut ignore_files = Vec::new(); |
|
390 | 389 | let repo_ignore = repo.working_directory_vfs().join(".hgignore"); |
|
391 | 390 | if repo_ignore.exists() { |
|
392 | 391 | ignore_files.push(repo_ignore) |
|
393 | 392 | } |
|
394 | 393 | for (key, value) in config.iter_section(b"ui") { |
|
395 | 394 | if key == b"ignore" || key.starts_with(b"ignore.") { |
|
396 | 395 | let path = get_path_from_bytes(value); |
|
397 | 396 | // TODO: expand "~/" and environment variable here, like Python |
|
398 | 397 | // does with `os.path.expanduser` and `os.path.expandvars` |
|
399 | 398 | |
|
400 | 399 | let joined = repo.working_directory_path().join(path); |
|
401 | 400 | ignore_files.push(joined); |
|
402 | 401 | } |
|
403 | 402 | } |
|
404 | 403 | ignore_files |
|
405 | 404 | } |
|
406 | 405 | |
|
407 | 406 | struct DisplayStatusPaths<'a> { |
|
408 | 407 | ui: &'a Ui, |
|
409 | 408 | no_status: bool, |
|
410 | 409 | relativize: Option<RelativizePaths>, |
|
411 | 410 | } |
|
412 | 411 | |
|
413 | 412 | impl DisplayStatusPaths<'_> { |
|
414 | 413 | // Probably more elegant to use a Deref or Borrow trait rather than |
|
415 | 414 | // harcode HgPathBuf, but probably not really useful at this point |
|
416 | 415 | fn display( |
|
417 | 416 | &self, |
|
418 | 417 | status_prefix: &[u8], |
|
419 | 418 | mut paths: Vec<StatusPath<'_>>, |
|
420 | 419 | ) -> Result<(), CommandError> { |
|
421 | 420 | paths.sort_unstable(); |
|
422 | 421 | for StatusPath { path, copy_source } in paths { |
|
423 | 422 | let relative; |
|
424 | 423 | let path = if let Some(relativize) = &self.relativize { |
|
425 | 424 | relative = relativize.relativize(&path); |
|
426 | 425 | &*relative |
|
427 | 426 | } else { |
|
428 | 427 | path.as_bytes() |
|
429 | 428 | }; |
|
430 | 429 | // TODO optim, probably lots of unneeded copies here, especially |
|
431 | 430 | // if out stream is buffered |
|
432 | 431 | if self.no_status { |
|
433 | 432 | self.ui.write_stdout(&format_bytes!(b"{}\n", path))? |
|
434 | 433 | } else { |
|
435 | 434 | self.ui.write_stdout(&format_bytes!( |
|
436 | 435 | b"{} {}\n", |
|
437 | 436 | status_prefix, |
|
438 | 437 | path |
|
439 | 438 | ))? |
|
440 | 439 | } |
|
441 | 440 | if let Some(source) = copy_source { |
|
442 | 441 | self.ui.write_stdout(&format_bytes!( |
|
443 | 442 | b" {}\n", |
|
444 | 443 | source.as_bytes() |
|
445 | 444 | ))? |
|
446 | 445 | } |
|
447 | 446 | } |
|
448 | 447 | Ok(()) |
|
449 | 448 | } |
|
450 | 449 | } |
|
451 | 450 | |
|
452 | 451 | /// Check if a file is modified by comparing actual repo store and file system. |
|
453 | 452 | /// |
|
454 | 453 | /// This meant to be used for those that the dirstate cannot resolve, due |
|
455 | 454 | /// to time resolution limits. |
|
456 | 455 | fn unsure_is_modified( |
|
457 | 456 | repo: &Repo, |
|
458 | 457 | manifest: &Manifest, |
|
459 | 458 | hg_path: &HgPath, |
|
460 | 459 | ) -> Result<bool, HgError> { |
|
461 | 460 | let vfs = repo.working_directory_vfs(); |
|
462 | 461 | let fs_path = hg_path_to_path_buf(hg_path).expect("HgPath conversion"); |
|
463 | 462 | let fs_metadata = vfs.symlink_metadata(&fs_path)?; |
|
464 | 463 | let is_symlink = fs_metadata.file_type().is_symlink(); |
|
465 | 464 | // TODO: Also account for `FALLBACK_SYMLINK` and `FALLBACK_EXEC` from the |
|
466 | 465 | // dirstate |
|
467 | 466 | let fs_flags = if is_symlink { |
|
468 | 467 | Some(b'l') |
|
469 | 468 | } else if has_exec_bit(&fs_metadata) { |
|
470 | 469 | Some(b'x') |
|
471 | 470 | } else { |
|
472 | 471 | None |
|
473 | 472 | }; |
|
474 | 473 | |
|
475 | 474 | let entry = manifest |
|
476 | 475 | .find_by_path(hg_path)? |
|
477 | 476 | .expect("ambgious file not in p1"); |
|
478 | 477 | if entry.flags != fs_flags { |
|
479 | 478 | return Ok(true); |
|
480 | 479 | } |
|
481 | 480 | let filelog = repo.filelog(hg_path)?; |
|
482 | 481 | let fs_len = fs_metadata.len(); |
|
483 | 482 | // TODO: check `fs_len` here like below, but based on |
|
484 | 483 | // `RevlogEntry::uncompressed_len` without decompressing the full filelog |
|
485 | 484 | // contents where possible. This is only valid if the revlog data does not |
|
486 | 485 | // contain metadata. See how Python’s `revlog.rawsize` calls |
|
487 | 486 | // `storageutil.filerevisioncopied`. |
|
488 | 487 | // (Maybe also check for content-modifying flags? See `revlog.size`.) |
|
489 | 488 | let filelog_entry = |
|
490 | 489 | filelog.data_for_node(entry.node_id()?).map_err(|_| { |
|
491 | 490 | HgError::corrupted("filelog missing node from manifest") |
|
492 | 491 | })?; |
|
493 | 492 | let contents_in_p1 = filelog_entry.data()?; |
|
494 | 493 | if contents_in_p1.len() as u64 != fs_len { |
|
495 | 494 | // No need to read the file contents: |
|
496 | 495 | // it cannot be equal if it has a different length. |
|
497 | 496 | return Ok(true); |
|
498 | 497 | } |
|
499 | 498 | |
|
500 | 499 | let fs_contents = if is_symlink { |
|
501 | 500 | get_bytes_from_os_string(vfs.read_link(fs_path)?.into_os_string()) |
|
502 | 501 | } else { |
|
503 | 502 | vfs.read(fs_path)? |
|
504 | 503 | }; |
|
505 | 504 | Ok(contents_in_p1 != &*fs_contents) |
|
506 | 505 | } |
General Comments 0
You need to be logged in to leave comments.
Login now