Show More
@@ -1,891 +1,879 b'' | |||||
1 | use crate::utils::hg_path::HgPath; |
|
1 | use crate::utils::hg_path::HgPath; | |
2 | use crate::utils::hg_path::HgPathBuf; |
|
2 | use crate::utils::hg_path::HgPathBuf; | |
3 | use crate::Revision; |
|
3 | use crate::Revision; | |
4 | use crate::NULL_REVISION; |
|
4 | use crate::NULL_REVISION; | |
5 |
|
5 | |||
6 | use im_rc::ordmap::DiffItem; |
|
6 | use im_rc::ordmap::DiffItem; | |
7 | use im_rc::ordmap::Entry; |
|
7 | use im_rc::ordmap::Entry; | |
8 | use im_rc::ordmap::OrdMap; |
|
8 | use im_rc::ordmap::OrdMap; | |
9 |
|
9 | |||
10 | use std::cmp::Ordering; |
|
10 | use std::cmp::Ordering; | |
11 | use std::collections::HashMap; |
|
11 | use std::collections::HashMap; | |
12 | use std::collections::HashSet; |
|
12 | use std::collections::HashSet; | |
13 | use std::convert::TryInto; |
|
13 | use std::convert::TryInto; | |
14 |
|
14 | |||
15 | pub type PathCopies = HashMap<HgPathBuf, HgPathBuf>; |
|
15 | pub type PathCopies = HashMap<HgPathBuf, HgPathBuf>; | |
16 |
|
16 | |||
17 | type PathToken = usize; |
|
17 | type PathToken = usize; | |
18 |
|
18 | |||
19 | #[derive(Clone, Debug, PartialEq)] |
|
19 | #[derive(Clone, Debug, PartialEq)] | |
20 | struct CopySource { |
|
20 | struct CopySource { | |
21 | /// revision at which the copy information was added |
|
21 | /// revision at which the copy information was added | |
22 | rev: Revision, |
|
22 | rev: Revision, | |
23 | /// the copy source, (Set to None in case of deletion of the associated |
|
23 | /// the copy source, (Set to None in case of deletion of the associated | |
24 | /// key) |
|
24 | /// key) | |
25 | path: Option<PathToken>, |
|
25 | path: Option<PathToken>, | |
26 | /// a set of previous `CopySource.rev` value directly or indirectly |
|
26 | /// a set of previous `CopySource.rev` value directly or indirectly | |
27 | /// overwritten by this one. |
|
27 | /// overwritten by this one. | |
28 | overwritten: HashSet<Revision>, |
|
28 | overwritten: HashSet<Revision>, | |
29 | } |
|
29 | } | |
30 |
|
30 | |||
31 | impl CopySource { |
|
31 | impl CopySource { | |
32 | /// create a new CopySource |
|
32 | /// create a new CopySource | |
33 | /// |
|
33 | /// | |
34 | /// Use this when no previous copy source existed. |
|
34 | /// Use this when no previous copy source existed. | |
35 | fn new(rev: Revision, path: Option<PathToken>) -> Self { |
|
35 | fn new(rev: Revision, path: Option<PathToken>) -> Self { | |
36 | Self { |
|
36 | Self { | |
37 | rev, |
|
37 | rev, | |
38 | path, |
|
38 | path, | |
39 | overwritten: HashSet::new(), |
|
39 | overwritten: HashSet::new(), | |
40 | } |
|
40 | } | |
41 | } |
|
41 | } | |
42 |
|
42 | |||
43 | /// create a new CopySource from merging two others |
|
43 | /// create a new CopySource from merging two others | |
44 | /// |
|
44 | /// | |
45 | /// Use this when merging two InternalPathCopies requires active merging of |
|
45 | /// Use this when merging two InternalPathCopies requires active merging of | |
46 | /// some entries. |
|
46 | /// some entries. | |
47 | fn new_from_merge(rev: Revision, winner: &Self, loser: &Self) -> Self { |
|
47 | fn new_from_merge(rev: Revision, winner: &Self, loser: &Self) -> Self { | |
48 | let mut overwritten = HashSet::new(); |
|
48 | let mut overwritten = HashSet::new(); | |
49 | overwritten.extend(winner.overwritten.iter().copied()); |
|
49 | overwritten.extend(winner.overwritten.iter().copied()); | |
50 | overwritten.extend(loser.overwritten.iter().copied()); |
|
50 | overwritten.extend(loser.overwritten.iter().copied()); | |
51 | overwritten.insert(winner.rev); |
|
51 | overwritten.insert(winner.rev); | |
52 | overwritten.insert(loser.rev); |
|
52 | overwritten.insert(loser.rev); | |
53 | Self { |
|
53 | Self { | |
54 | rev, |
|
54 | rev, | |
55 | path: winner.path, |
|
55 | path: winner.path, | |
56 | overwritten: overwritten, |
|
56 | overwritten: overwritten, | |
57 | } |
|
57 | } | |
58 | } |
|
58 | } | |
59 |
|
59 | |||
60 | /// Update the value of a pre-existing CopySource |
|
60 | /// Update the value of a pre-existing CopySource | |
61 | /// |
|
61 | /// | |
62 | /// Use this when recording copy information from parent β child edges |
|
62 | /// Use this when recording copy information from parent β child edges | |
63 | fn overwrite(&mut self, rev: Revision, path: Option<PathToken>) { |
|
63 | fn overwrite(&mut self, rev: Revision, path: Option<PathToken>) { | |
64 | self.overwritten.insert(self.rev); |
|
64 | self.overwritten.insert(self.rev); | |
65 | self.rev = rev; |
|
65 | self.rev = rev; | |
66 | self.path = path; |
|
66 | self.path = path; | |
67 | } |
|
67 | } | |
68 |
|
68 | |||
69 | /// Mark pre-existing copy information as "dropped" by a file deletion |
|
69 | /// Mark pre-existing copy information as "dropped" by a file deletion | |
70 | /// |
|
70 | /// | |
71 | /// Use this when recording copy information from parent β child edges |
|
71 | /// Use this when recording copy information from parent β child edges | |
72 | fn mark_delete(&mut self, rev: Revision) { |
|
72 | fn mark_delete(&mut self, rev: Revision) { | |
73 | self.overwritten.insert(self.rev); |
|
73 | self.overwritten.insert(self.rev); | |
74 | self.rev = rev; |
|
74 | self.rev = rev; | |
75 | self.path = None; |
|
75 | self.path = None; | |
76 | } |
|
76 | } | |
77 |
|
77 | |||
78 | fn is_overwritten_by(&self, other: &Self) -> bool { |
|
78 | fn is_overwritten_by(&self, other: &Self) -> bool { | |
79 | other.overwritten.contains(&self.rev) |
|
79 | other.overwritten.contains(&self.rev) | |
80 | } |
|
80 | } | |
81 | } |
|
81 | } | |
82 |
|
82 | |||
83 | /// maps CopyDestination to Copy Source (+ a "timestamp" for the operation) |
|
83 | /// maps CopyDestination to Copy Source (+ a "timestamp" for the operation) | |
84 | type InternalPathCopies = OrdMap<PathToken, CopySource>; |
|
84 | type InternalPathCopies = OrdMap<PathToken, CopySource>; | |
85 |
|
85 | |||
86 | /// hold parent 1, parent 2 and relevant files actions. |
|
86 | /// hold parent 1, parent 2 and relevant files actions. | |
87 | pub type RevInfo<'a> = (Revision, Revision, ChangedFiles<'a>); |
|
87 | pub type RevInfo<'a> = (Revision, Revision, ChangedFiles<'a>); | |
88 |
|
88 | |||
89 | /// represent the files affected by a changesets |
|
89 | /// represent the files affected by a changesets | |
90 | /// |
|
90 | /// | |
91 | /// This hold a subset of mercurial.metadata.ChangingFiles as we do not need |
|
91 | /// This hold a subset of mercurial.metadata.ChangingFiles as we do not need | |
92 | /// all the data categories tracked by it. |
|
92 | /// all the data categories tracked by it. | |
93 | /// This hold a subset of mercurial.metadata.ChangingFiles as we do not need |
|
93 | /// This hold a subset of mercurial.metadata.ChangingFiles as we do not need | |
94 | /// all the data categories tracked by it. |
|
94 | /// all the data categories tracked by it. | |
95 | pub struct ChangedFiles<'a> { |
|
95 | pub struct ChangedFiles<'a> { | |
96 | nb_items: u32, |
|
96 | nb_items: u32, | |
97 | index: &'a [u8], |
|
97 | index: &'a [u8], | |
98 | data: &'a [u8], |
|
98 | data: &'a [u8], | |
99 | } |
|
99 | } | |
100 |
|
100 | |||
101 | /// Represent active changes that affect the copy tracing. |
|
101 | /// Represent active changes that affect the copy tracing. | |
102 | enum Action<'a> { |
|
102 | enum Action<'a> { | |
103 | /// The parent ? children edge is removing a file |
|
103 | /// The parent ? children edge is removing a file | |
104 | /// |
|
104 | /// | |
105 | /// (actually, this could be the edge from the other parent, but it does |
|
105 | /// (actually, this could be the edge from the other parent, but it does | |
106 | /// not matters) |
|
106 | /// not matters) | |
107 | Removed(&'a HgPath), |
|
107 | Removed(&'a HgPath), | |
108 | /// The parent ? children edge introduce copy information between (dest, |
|
108 | /// The parent ? children edge introduce copy information between (dest, | |
109 | /// source) |
|
109 | /// source) | |
110 | Copied(&'a HgPath, &'a HgPath), |
|
110 | Copied(&'a HgPath, &'a HgPath), | |
111 | } |
|
111 | } | |
112 |
|
112 | |||
113 | /// This express the possible "special" case we can get in a merge |
|
113 | /// This express the possible "special" case we can get in a merge | |
114 | /// |
|
114 | /// | |
115 | /// See mercurial/metadata.py for details on these values. |
|
115 | /// See mercurial/metadata.py for details on these values. | |
116 | #[derive(PartialEq)] |
|
116 | #[derive(PartialEq)] | |
117 | enum MergeCase { |
|
117 | enum MergeCase { | |
118 | /// Merged: file had history on both side that needed to be merged |
|
118 | /// Merged: file had history on both side that needed to be merged | |
119 | Merged, |
|
119 | Merged, | |
120 | /// Salvaged: file was candidate for deletion, but survived the merge |
|
120 | /// Salvaged: file was candidate for deletion, but survived the merge | |
121 | Salvaged, |
|
121 | Salvaged, | |
122 | /// Normal: Not one of the two cases above |
|
122 | /// Normal: Not one of the two cases above | |
123 | Normal, |
|
123 | Normal, | |
124 | } |
|
124 | } | |
125 |
|
125 | |||
126 | type FileChange<'a> = (u8, &'a HgPath, &'a HgPath); |
|
126 | type FileChange<'a> = (u8, &'a HgPath, &'a HgPath); | |
127 |
|
127 | |||
128 | const EMPTY: &[u8] = b""; |
|
128 | const EMPTY: &[u8] = b""; | |
129 | const COPY_MASK: u8 = 3; |
|
129 | const COPY_MASK: u8 = 3; | |
130 | const P1_COPY: u8 = 2; |
|
130 | const P1_COPY: u8 = 2; | |
131 | const P2_COPY: u8 = 3; |
|
131 | const P2_COPY: u8 = 3; | |
132 | const ACTION_MASK: u8 = 28; |
|
132 | const ACTION_MASK: u8 = 28; | |
133 | const REMOVED: u8 = 12; |
|
133 | const REMOVED: u8 = 12; | |
134 | const MERGED: u8 = 8; |
|
134 | const MERGED: u8 = 8; | |
135 | const SALVAGED: u8 = 16; |
|
135 | const SALVAGED: u8 = 16; | |
136 |
|
136 | |||
137 | impl<'a> ChangedFiles<'a> { |
|
137 | impl<'a> ChangedFiles<'a> { | |
138 | const INDEX_START: usize = 4; |
|
138 | const INDEX_START: usize = 4; | |
139 | const ENTRY_SIZE: u32 = 9; |
|
139 | const ENTRY_SIZE: u32 = 9; | |
140 | const FILENAME_START: u32 = 1; |
|
140 | const FILENAME_START: u32 = 1; | |
141 | const COPY_SOURCE_START: u32 = 5; |
|
141 | const COPY_SOURCE_START: u32 = 5; | |
142 |
|
142 | |||
143 | pub fn new(data: &'a [u8]) -> Self { |
|
143 | pub fn new(data: &'a [u8]) -> Self { | |
144 | assert!( |
|
144 | assert!( | |
145 | data.len() >= 4, |
|
145 | data.len() >= 4, | |
146 | "data size ({}) is too small to contain the header (4)", |
|
146 | "data size ({}) is too small to contain the header (4)", | |
147 | data.len() |
|
147 | data.len() | |
148 | ); |
|
148 | ); | |
149 | let nb_items_raw: [u8; 4] = (&data[0..=3]) |
|
149 | let nb_items_raw: [u8; 4] = (&data[0..=3]) | |
150 | .try_into() |
|
150 | .try_into() | |
151 | .expect("failed to turn 4 bytes into 4 bytes"); |
|
151 | .expect("failed to turn 4 bytes into 4 bytes"); | |
152 | let nb_items = u32::from_be_bytes(nb_items_raw); |
|
152 | let nb_items = u32::from_be_bytes(nb_items_raw); | |
153 |
|
153 | |||
154 | let index_size = (nb_items * Self::ENTRY_SIZE) as usize; |
|
154 | let index_size = (nb_items * Self::ENTRY_SIZE) as usize; | |
155 | let index_end = Self::INDEX_START + index_size; |
|
155 | let index_end = Self::INDEX_START + index_size; | |
156 |
|
156 | |||
157 | assert!( |
|
157 | assert!( | |
158 | data.len() >= index_end, |
|
158 | data.len() >= index_end, | |
159 | "data size ({}) is too small to fit the index_data ({})", |
|
159 | "data size ({}) is too small to fit the index_data ({})", | |
160 | data.len(), |
|
160 | data.len(), | |
161 | index_end |
|
161 | index_end | |
162 | ); |
|
162 | ); | |
163 |
|
163 | |||
164 | let ret = ChangedFiles { |
|
164 | let ret = ChangedFiles { | |
165 | nb_items, |
|
165 | nb_items, | |
166 | index: &data[Self::INDEX_START..index_end], |
|
166 | index: &data[Self::INDEX_START..index_end], | |
167 | data: &data[index_end..], |
|
167 | data: &data[index_end..], | |
168 | }; |
|
168 | }; | |
169 | let max_data = ret.filename_end(nb_items - 1) as usize; |
|
169 | let max_data = ret.filename_end(nb_items - 1) as usize; | |
170 | assert!( |
|
170 | assert!( | |
171 | ret.data.len() >= max_data, |
|
171 | ret.data.len() >= max_data, | |
172 | "data size ({}) is too small to fit all data ({})", |
|
172 | "data size ({}) is too small to fit all data ({})", | |
173 | data.len(), |
|
173 | data.len(), | |
174 | index_end + max_data |
|
174 | index_end + max_data | |
175 | ); |
|
175 | ); | |
176 | ret |
|
176 | ret | |
177 | } |
|
177 | } | |
178 |
|
178 | |||
179 | pub fn new_empty() -> Self { |
|
179 | pub fn new_empty() -> Self { | |
180 | ChangedFiles { |
|
180 | ChangedFiles { | |
181 | nb_items: 0, |
|
181 | nb_items: 0, | |
182 | index: EMPTY, |
|
182 | index: EMPTY, | |
183 | data: EMPTY, |
|
183 | data: EMPTY, | |
184 | } |
|
184 | } | |
185 | } |
|
185 | } | |
186 |
|
186 | |||
187 | /// internal function to return an individual entry at a given index |
|
187 | /// internal function to return an individual entry at a given index | |
188 | fn entry(&'a self, idx: u32) -> FileChange<'a> { |
|
188 | fn entry(&'a self, idx: u32) -> FileChange<'a> { | |
189 | if idx >= self.nb_items { |
|
189 | if idx >= self.nb_items { | |
190 | panic!( |
|
190 | panic!( | |
191 | "index for entry is higher that the number of file {} >= {}", |
|
191 | "index for entry is higher that the number of file {} >= {}", | |
192 | idx, self.nb_items |
|
192 | idx, self.nb_items | |
193 | ) |
|
193 | ) | |
194 | } |
|
194 | } | |
195 | let flags = self.flags(idx); |
|
195 | let flags = self.flags(idx); | |
196 | let filename = self.filename(idx); |
|
196 | let filename = self.filename(idx); | |
197 | let copy_idx = self.copy_idx(idx); |
|
197 | let copy_idx = self.copy_idx(idx); | |
198 | let copy_source = self.filename(copy_idx); |
|
198 | let copy_source = self.filename(copy_idx); | |
199 | (flags, filename, copy_source) |
|
199 | (flags, filename, copy_source) | |
200 | } |
|
200 | } | |
201 |
|
201 | |||
202 | /// internal function to return the filename of the entry at a given index |
|
202 | /// internal function to return the filename of the entry at a given index | |
203 | fn filename(&self, idx: u32) -> &HgPath { |
|
203 | fn filename(&self, idx: u32) -> &HgPath { | |
204 | let filename_start; |
|
204 | let filename_start; | |
205 | if idx == 0 { |
|
205 | if idx == 0 { | |
206 | filename_start = 0; |
|
206 | filename_start = 0; | |
207 | } else { |
|
207 | } else { | |
208 | filename_start = self.filename_end(idx - 1) |
|
208 | filename_start = self.filename_end(idx - 1) | |
209 | } |
|
209 | } | |
210 | let filename_end = self.filename_end(idx); |
|
210 | let filename_end = self.filename_end(idx); | |
211 | let filename_start = filename_start as usize; |
|
211 | let filename_start = filename_start as usize; | |
212 | let filename_end = filename_end as usize; |
|
212 | let filename_end = filename_end as usize; | |
213 | HgPath::new(&self.data[filename_start..filename_end]) |
|
213 | HgPath::new(&self.data[filename_start..filename_end]) | |
214 | } |
|
214 | } | |
215 |
|
215 | |||
216 | /// internal function to return the flag field of the entry at a given |
|
216 | /// internal function to return the flag field of the entry at a given | |
217 | /// index |
|
217 | /// index | |
218 | fn flags(&self, idx: u32) -> u8 { |
|
218 | fn flags(&self, idx: u32) -> u8 { | |
219 | let idx = idx as usize; |
|
219 | let idx = idx as usize; | |
220 | self.index[idx * (Self::ENTRY_SIZE as usize)] |
|
220 | self.index[idx * (Self::ENTRY_SIZE as usize)] | |
221 | } |
|
221 | } | |
222 |
|
222 | |||
223 | /// internal function to return the end of a filename part at a given index |
|
223 | /// internal function to return the end of a filename part at a given index | |
224 | fn filename_end(&self, idx: u32) -> u32 { |
|
224 | fn filename_end(&self, idx: u32) -> u32 { | |
225 | let start = (idx * Self::ENTRY_SIZE) + Self::FILENAME_START; |
|
225 | let start = (idx * Self::ENTRY_SIZE) + Self::FILENAME_START; | |
226 | let end = (idx * Self::ENTRY_SIZE) + Self::COPY_SOURCE_START; |
|
226 | let end = (idx * Self::ENTRY_SIZE) + Self::COPY_SOURCE_START; | |
227 | let start = start as usize; |
|
227 | let start = start as usize; | |
228 | let end = end as usize; |
|
228 | let end = end as usize; | |
229 | let raw = (&self.index[start..end]) |
|
229 | let raw = (&self.index[start..end]) | |
230 | .try_into() |
|
230 | .try_into() | |
231 | .expect("failed to turn 4 bytes into 4 bytes"); |
|
231 | .expect("failed to turn 4 bytes into 4 bytes"); | |
232 | u32::from_be_bytes(raw) |
|
232 | u32::from_be_bytes(raw) | |
233 | } |
|
233 | } | |
234 |
|
234 | |||
235 | /// internal function to return index of the copy source of the entry at a |
|
235 | /// internal function to return index of the copy source of the entry at a | |
236 | /// given index |
|
236 | /// given index | |
237 | fn copy_idx(&self, idx: u32) -> u32 { |
|
237 | fn copy_idx(&self, idx: u32) -> u32 { | |
238 | let start = (idx * Self::ENTRY_SIZE) + Self::COPY_SOURCE_START; |
|
238 | let start = (idx * Self::ENTRY_SIZE) + Self::COPY_SOURCE_START; | |
239 | let end = (idx + 1) * Self::ENTRY_SIZE; |
|
239 | let end = (idx + 1) * Self::ENTRY_SIZE; | |
240 | let start = start as usize; |
|
240 | let start = start as usize; | |
241 | let end = end as usize; |
|
241 | let end = end as usize; | |
242 | let raw = (&self.index[start..end]) |
|
242 | let raw = (&self.index[start..end]) | |
243 | .try_into() |
|
243 | .try_into() | |
244 | .expect("failed to turn 4 bytes into 4 bytes"); |
|
244 | .expect("failed to turn 4 bytes into 4 bytes"); | |
245 | u32::from_be_bytes(raw) |
|
245 | u32::from_be_bytes(raw) | |
246 | } |
|
246 | } | |
247 |
|
247 | |||
248 | /// Return an iterator over all the `Action` in this instance. |
|
248 | /// Return an iterator over all the `Action` in this instance. | |
249 | fn iter_actions(&self, parent: Parent) -> ActionsIterator { |
|
249 | fn iter_actions(&self, parent: Parent) -> ActionsIterator { | |
250 | ActionsIterator { |
|
250 | ActionsIterator { | |
251 | changes: &self, |
|
251 | changes: &self, | |
252 | parent: parent, |
|
252 | parent: parent, | |
253 | current: 0, |
|
253 | current: 0, | |
254 | } |
|
254 | } | |
255 | } |
|
255 | } | |
256 |
|
256 | |||
257 | /// return the MergeCase value associated with a filename |
|
257 | /// return the MergeCase value associated with a filename | |
258 | fn get_merge_case(&self, path: &HgPath) -> MergeCase { |
|
258 | fn get_merge_case(&self, path: &HgPath) -> MergeCase { | |
259 | if self.nb_items == 0 { |
|
259 | if self.nb_items == 0 { | |
260 | return MergeCase::Normal; |
|
260 | return MergeCase::Normal; | |
261 | } |
|
261 | } | |
262 | let mut low_part = 0; |
|
262 | let mut low_part = 0; | |
263 | let mut high_part = self.nb_items; |
|
263 | let mut high_part = self.nb_items; | |
264 |
|
264 | |||
265 | while low_part < high_part { |
|
265 | while low_part < high_part { | |
266 | let cursor = (low_part + high_part - 1) / 2; |
|
266 | let cursor = (low_part + high_part - 1) / 2; | |
267 | let (flags, filename, _source) = self.entry(cursor); |
|
267 | let (flags, filename, _source) = self.entry(cursor); | |
268 | match path.cmp(filename) { |
|
268 | match path.cmp(filename) { | |
269 | Ordering::Less => low_part = cursor + 1, |
|
269 | Ordering::Less => low_part = cursor + 1, | |
270 | Ordering::Greater => high_part = cursor, |
|
270 | Ordering::Greater => high_part = cursor, | |
271 | Ordering::Equal => { |
|
271 | Ordering::Equal => { | |
272 | return match flags & ACTION_MASK { |
|
272 | return match flags & ACTION_MASK { | |
273 | MERGED => MergeCase::Merged, |
|
273 | MERGED => MergeCase::Merged, | |
274 | SALVAGED => MergeCase::Salvaged, |
|
274 | SALVAGED => MergeCase::Salvaged, | |
275 | _ => MergeCase::Normal, |
|
275 | _ => MergeCase::Normal, | |
276 | }; |
|
276 | }; | |
277 | } |
|
277 | } | |
278 | } |
|
278 | } | |
279 | } |
|
279 | } | |
280 | MergeCase::Normal |
|
280 | MergeCase::Normal | |
281 | } |
|
281 | } | |
282 | } |
|
282 | } | |
283 |
|
283 | |||
284 | struct ActionsIterator<'a> { |
|
284 | struct ActionsIterator<'a> { | |
285 | changes: &'a ChangedFiles<'a>, |
|
285 | changes: &'a ChangedFiles<'a>, | |
286 | parent: Parent, |
|
286 | parent: Parent, | |
287 | current: u32, |
|
287 | current: u32, | |
288 | } |
|
288 | } | |
289 |
|
289 | |||
290 | impl<'a> Iterator for ActionsIterator<'a> { |
|
290 | impl<'a> Iterator for ActionsIterator<'a> { | |
291 | type Item = Action<'a>; |
|
291 | type Item = Action<'a>; | |
292 |
|
292 | |||
293 | fn next(&mut self) -> Option<Action<'a>> { |
|
293 | fn next(&mut self) -> Option<Action<'a>> { | |
294 | let copy_flag = match self.parent { |
|
294 | let copy_flag = match self.parent { | |
295 | Parent::FirstParent => P1_COPY, |
|
295 | Parent::FirstParent => P1_COPY, | |
296 | Parent::SecondParent => P2_COPY, |
|
296 | Parent::SecondParent => P2_COPY, | |
297 | }; |
|
297 | }; | |
298 | while self.current < self.changes.nb_items { |
|
298 | while self.current < self.changes.nb_items { | |
299 | let (flags, file, source) = self.changes.entry(self.current); |
|
299 | let (flags, file, source) = self.changes.entry(self.current); | |
300 | self.current += 1; |
|
300 | self.current += 1; | |
301 | if (flags & ACTION_MASK) == REMOVED { |
|
301 | if (flags & ACTION_MASK) == REMOVED { | |
302 | return Some(Action::Removed(file)); |
|
302 | return Some(Action::Removed(file)); | |
303 | } |
|
303 | } | |
304 | let copy = flags & COPY_MASK; |
|
304 | let copy = flags & COPY_MASK; | |
305 | if copy == copy_flag { |
|
305 | if copy == copy_flag { | |
306 | return Some(Action::Copied(file, source)); |
|
306 | return Some(Action::Copied(file, source)); | |
307 | } |
|
307 | } | |
308 | } |
|
308 | } | |
309 | return None; |
|
309 | return None; | |
310 | } |
|
310 | } | |
311 | } |
|
311 | } | |
312 |
|
312 | |||
313 | /// A small struct whose purpose is to ensure lifetime of bytes referenced in |
|
313 | /// A small struct whose purpose is to ensure lifetime of bytes referenced in | |
314 | /// ChangedFiles |
|
314 | /// ChangedFiles | |
315 | /// |
|
315 | /// | |
316 | /// It is passed to the RevInfoMaker callback who can assign any necessary |
|
316 | /// It is passed to the RevInfoMaker callback who can assign any necessary | |
317 | /// content to the `data` attribute. The copy tracing code is responsible for |
|
317 | /// content to the `data` attribute. The copy tracing code is responsible for | |
318 | /// keeping the DataHolder alive at least as long as the ChangedFiles object. |
|
318 | /// keeping the DataHolder alive at least as long as the ChangedFiles object. | |
319 | pub struct DataHolder<D> { |
|
319 | pub struct DataHolder<D> { | |
320 | /// RevInfoMaker callback should assign data referenced by the |
|
320 | /// RevInfoMaker callback should assign data referenced by the | |
321 | /// ChangedFiles struct it return to this attribute. The DataHolder |
|
321 | /// ChangedFiles struct it return to this attribute. The DataHolder | |
322 | /// lifetime will be at least as long as the ChangedFiles one. |
|
322 | /// lifetime will be at least as long as the ChangedFiles one. | |
323 | pub data: Option<D>, |
|
323 | pub data: Option<D>, | |
324 | } |
|
324 | } | |
325 |
|
325 | |||
326 | pub type RevInfoMaker<'a, D> = |
|
326 | pub type RevInfoMaker<'a, D> = | |
327 | Box<dyn for<'r> Fn(Revision, &'r mut DataHolder<D>) -> RevInfo<'r> + 'a>; |
|
327 | Box<dyn for<'r> Fn(Revision, &'r mut DataHolder<D>) -> RevInfo<'r> + 'a>; | |
328 |
|
328 | |||
329 | /// enum used to carry information about the parent β child currently processed |
|
329 | /// enum used to carry information about the parent β child currently processed | |
330 | #[derive(Copy, Clone, Debug)] |
|
330 | #[derive(Copy, Clone, Debug)] | |
331 | enum Parent { |
|
331 | enum Parent { | |
332 | /// The `p1(x) β x` edge |
|
332 | /// The `p1(x) β x` edge | |
333 | FirstParent, |
|
333 | FirstParent, | |
334 | /// The `p2(x) β x` edge |
|
334 | /// The `p2(x) β x` edge | |
335 | SecondParent, |
|
335 | SecondParent, | |
336 | } |
|
336 | } | |
337 |
|
337 | |||
338 | /// A small "tokenizer" responsible of turning full HgPath into lighter |
|
338 | /// A small "tokenizer" responsible of turning full HgPath into lighter | |
339 | /// PathToken |
|
339 | /// PathToken | |
340 | /// |
|
340 | /// | |
341 | /// Dealing with small object, like integer is much faster, so HgPath input are |
|
341 | /// Dealing with small object, like integer is much faster, so HgPath input are | |
342 | /// turned into integer "PathToken" and converted back in the end. |
|
342 | /// turned into integer "PathToken" and converted back in the end. | |
343 | #[derive(Clone, Debug, Default)] |
|
343 | #[derive(Clone, Debug, Default)] | |
344 | struct TwoWayPathMap { |
|
344 | struct TwoWayPathMap { | |
345 | token: HashMap<HgPathBuf, PathToken>, |
|
345 | token: HashMap<HgPathBuf, PathToken>, | |
346 | path: Vec<HgPathBuf>, |
|
346 | path: Vec<HgPathBuf>, | |
347 | } |
|
347 | } | |
348 |
|
348 | |||
349 | impl TwoWayPathMap { |
|
349 | impl TwoWayPathMap { | |
350 | fn tokenize(&mut self, path: &HgPath) -> PathToken { |
|
350 | fn tokenize(&mut self, path: &HgPath) -> PathToken { | |
351 | match self.token.get(path) { |
|
351 | match self.token.get(path) { | |
352 | Some(a) => *a, |
|
352 | Some(a) => *a, | |
353 | None => { |
|
353 | None => { | |
354 | let a = self.token.len(); |
|
354 | let a = self.token.len(); | |
355 | let buf = path.to_owned(); |
|
355 | let buf = path.to_owned(); | |
356 | self.path.push(buf.clone()); |
|
356 | self.path.push(buf.clone()); | |
357 | self.token.insert(buf, a); |
|
357 | self.token.insert(buf, a); | |
358 | a |
|
358 | a | |
359 | } |
|
359 | } | |
360 | } |
|
360 | } | |
361 | } |
|
361 | } | |
362 |
|
362 | |||
363 | fn untokenize(&self, token: PathToken) -> &HgPathBuf { |
|
363 | fn untokenize(&self, token: PathToken) -> &HgPathBuf { | |
364 | assert!(token < self.path.len(), format!("Unknown token: {}", token)); |
|
364 | assert!(token < self.path.len(), format!("Unknown token: {}", token)); | |
365 | &self.path[token] |
|
365 | &self.path[token] | |
366 | } |
|
366 | } | |
367 | } |
|
367 | } | |
368 |
|
368 | |||
369 | /// Same as mercurial.copies._combine_changeset_copies, but in Rust. |
|
369 | /// Same as mercurial.copies._combine_changeset_copies, but in Rust. | |
370 | /// |
|
370 | /// | |
371 | /// Arguments are: |
|
371 | /// Arguments are: | |
372 | /// |
|
372 | /// | |
373 | /// revs: all revisions to be considered |
|
373 | /// revs: all revisions to be considered | |
374 | /// children: a {parent ? [childrens]} mapping |
|
374 | /// children: a {parent ? [childrens]} mapping | |
375 | /// target_rev: the final revision we are combining copies to |
|
375 | /// target_rev: the final revision we are combining copies to | |
376 | /// rev_info(rev): callback to get revision information: |
|
376 | /// rev_info(rev): callback to get revision information: | |
377 | /// * first parent |
|
377 | /// * first parent | |
378 | /// * second parent |
|
378 | /// * second parent | |
379 | /// * ChangedFiles |
|
379 | /// * ChangedFiles | |
380 | /// isancestors(low_rev, high_rev): callback to check if a revision is an |
|
380 | /// isancestors(low_rev, high_rev): callback to check if a revision is an | |
381 | /// ancestor of another |
|
381 | /// ancestor of another | |
382 | pub fn combine_changeset_copies<D>( |
|
382 | pub fn combine_changeset_copies<D>( | |
383 | revs: Vec<Revision>, |
|
383 | revs: Vec<Revision>, | |
384 | mut children_count: HashMap<Revision, usize>, |
|
384 | mut children_count: HashMap<Revision, usize>, | |
385 | target_rev: Revision, |
|
385 | target_rev: Revision, | |
386 | rev_info: RevInfoMaker<D>, |
|
386 | rev_info: RevInfoMaker<D>, | |
387 | ) -> PathCopies { |
|
387 | ) -> PathCopies { | |
388 | let mut all_copies = HashMap::new(); |
|
388 | let mut all_copies = HashMap::new(); | |
389 |
|
389 | |||
390 | let mut path_map = TwoWayPathMap::default(); |
|
390 | let mut path_map = TwoWayPathMap::default(); | |
391 |
|
391 | |||
392 | for rev in revs { |
|
392 | for rev in revs { | |
393 | let mut d: DataHolder<D> = DataHolder { data: None }; |
|
393 | let mut d: DataHolder<D> = DataHolder { data: None }; | |
394 | let (p1, p2, changes) = rev_info(rev, &mut d); |
|
394 | let (p1, p2, changes) = rev_info(rev, &mut d); | |
395 |
|
395 | |||
396 | // We will chain the copies information accumulated for the parent with |
|
396 | // We will chain the copies information accumulated for the parent with | |
397 | // the individual copies information the curent revision. Creating a |
|
397 | // the individual copies information the curent revision. Creating a | |
398 | // new TimeStampedPath for each `rev` β `children` vertex. |
|
398 | // new TimeStampedPath for each `rev` β `children` vertex. | |
399 | let mut copies: Option<InternalPathCopies> = None; |
|
399 | let mut copies: Option<InternalPathCopies> = None; | |
400 | // Retrieve data computed in a previous iteration |
|
400 | // Retrieve data computed in a previous iteration | |
401 | let p1_copies = match p1 { |
|
401 | let p1_copies = match p1 { | |
402 | NULL_REVISION => None, |
|
402 | NULL_REVISION => None, | |
403 | _ => get_and_clean_parent_copies( |
|
403 | _ => get_and_clean_parent_copies( | |
404 | &mut all_copies, |
|
404 | &mut all_copies, | |
405 | &mut children_count, |
|
405 | &mut children_count, | |
406 | p1, |
|
406 | p1, | |
407 | ), // will be None if the vertex is not to be traversed |
|
407 | ), // will be None if the vertex is not to be traversed | |
408 | }; |
|
408 | }; | |
409 | let p2_copies = match p2 { |
|
409 | let p2_copies = match p2 { | |
410 | NULL_REVISION => None, |
|
410 | NULL_REVISION => None, | |
411 | _ => get_and_clean_parent_copies( |
|
411 | _ => get_and_clean_parent_copies( | |
412 | &mut all_copies, |
|
412 | &mut all_copies, | |
413 | &mut children_count, |
|
413 | &mut children_count, | |
414 | p2, |
|
414 | p2, | |
415 | ), // will be None if the vertex is not to be traversed |
|
415 | ), // will be None if the vertex is not to be traversed | |
416 | }; |
|
416 | }; | |
417 | if let Some(parent_copies) = p1_copies { |
|
417 | // combine it with data for that revision | |
418 | // combine it with data for that revision |
|
418 | let p1_copies = match p1_copies { | |
419 | let vertex_copies = add_from_changes( |
|
419 | None => None, | |
|
420 | Some(parent_copies) => Some(add_from_changes( | |||
420 | &mut path_map, |
|
421 | &mut path_map, | |
421 | &parent_copies, |
|
422 | &parent_copies, | |
422 | &changes, |
|
423 | &changes, | |
423 | Parent::FirstParent, |
|
424 | Parent::FirstParent, | |
424 | rev, |
|
425 | rev, | |
425 |
) |
|
426 | )), | |
426 | // keep that data around for potential later combination |
|
427 | }; | |
427 |
|
|
428 | let p2_copies = match p2_copies { | |
428 | } |
|
429 | None => None, | |
429 |
|
|
430 | Some(parent_copies) => Some(add_from_changes( | |
430 | // combine it with data for that revision |
|
|||
431 | let vertex_copies = add_from_changes( |
|
|||
432 | &mut path_map, |
|
431 | &mut path_map, | |
433 | &parent_copies, |
|
432 | &parent_copies, | |
434 | &changes, |
|
433 | &changes, | |
435 | Parent::SecondParent, |
|
434 | Parent::SecondParent, | |
436 | rev, |
|
435 | rev, | |
437 |
) |
|
436 | )), | |
438 |
|
437 | }; | ||
439 |
|
|
438 | let copies = match (p1_copies, p2_copies) { | |
440 |
|
|
439 | (None, None) => None, | |
441 | // Merge has two parents needs to combines their copy |
|
440 | (c, None) => c, | |
442 | // information. |
|
441 | (None, c) => c, | |
443 | // |
|
442 | (Some(p1_copies), Some(p2_copies)) => Some(merge_copies_dict( | |
444 | // If we got data from both parents, We need to combine |
|
443 | &path_map, rev, p2_copies, p1_copies, &changes, | |
445 |
|
|
444 | )), | |
446 | Some(copies) => Some(merge_copies_dict( |
|
445 | }; | |
447 | &path_map, |
|
446 | if let Some(c) = copies { | |
448 | rev, |
|
447 | all_copies.insert(rev, c); | |
449 | vertex_copies, |
|
|||
450 | copies, |
|
|||
451 | &changes, |
|
|||
452 | )), |
|
|||
453 | }; |
|
|||
454 | } |
|
|||
455 | match copies { |
|
|||
456 | Some(copies) => { |
|
|||
457 | all_copies.insert(rev, copies); |
|
|||
458 | } |
|
|||
459 | _ => {} |
|
|||
460 | } |
|
448 | } | |
461 | } |
|
449 | } | |
462 |
|
450 | |||
463 | // Drop internal information (like the timestamp) and return the final |
|
451 | // Drop internal information (like the timestamp) and return the final | |
464 | // mapping. |
|
452 | // mapping. | |
465 | let tt_result = all_copies |
|
453 | let tt_result = all_copies | |
466 | .remove(&target_rev) |
|
454 | .remove(&target_rev) | |
467 | .expect("target revision was not processed"); |
|
455 | .expect("target revision was not processed"); | |
468 | let mut result = PathCopies::default(); |
|
456 | let mut result = PathCopies::default(); | |
469 | for (dest, tt_source) in tt_result { |
|
457 | for (dest, tt_source) in tt_result { | |
470 | if let Some(path) = tt_source.path { |
|
458 | if let Some(path) = tt_source.path { | |
471 | let path_dest = path_map.untokenize(dest).to_owned(); |
|
459 | let path_dest = path_map.untokenize(dest).to_owned(); | |
472 | let path_path = path_map.untokenize(path).to_owned(); |
|
460 | let path_path = path_map.untokenize(path).to_owned(); | |
473 | result.insert(path_dest, path_path); |
|
461 | result.insert(path_dest, path_path); | |
474 | } |
|
462 | } | |
475 | } |
|
463 | } | |
476 | result |
|
464 | result | |
477 | } |
|
465 | } | |
478 |
|
466 | |||
479 | /// fetch previous computed information |
|
467 | /// fetch previous computed information | |
480 | /// |
|
468 | /// | |
481 | /// If no other children are expected to need this information, we drop it from |
|
469 | /// If no other children are expected to need this information, we drop it from | |
482 | /// the cache. |
|
470 | /// the cache. | |
483 | /// |
|
471 | /// | |
484 | /// If parent is not part of the set we are expected to walk, return None. |
|
472 | /// If parent is not part of the set we are expected to walk, return None. | |
485 | fn get_and_clean_parent_copies( |
|
473 | fn get_and_clean_parent_copies( | |
486 | all_copies: &mut HashMap<Revision, InternalPathCopies>, |
|
474 | all_copies: &mut HashMap<Revision, InternalPathCopies>, | |
487 | children_count: &mut HashMap<Revision, usize>, |
|
475 | children_count: &mut HashMap<Revision, usize>, | |
488 | parent_rev: Revision, |
|
476 | parent_rev: Revision, | |
489 | ) -> Option<InternalPathCopies> { |
|
477 | ) -> Option<InternalPathCopies> { | |
490 | let count = children_count.get_mut(&parent_rev)?; |
|
478 | let count = children_count.get_mut(&parent_rev)?; | |
491 | *count -= 1; |
|
479 | *count -= 1; | |
492 | if *count == 0 { |
|
480 | if *count == 0 { | |
493 | match all_copies.remove(&parent_rev) { |
|
481 | match all_copies.remove(&parent_rev) { | |
494 | Some(c) => Some(c), |
|
482 | Some(c) => Some(c), | |
495 | None => Some(InternalPathCopies::default()), |
|
483 | None => Some(InternalPathCopies::default()), | |
496 | } |
|
484 | } | |
497 | } else { |
|
485 | } else { | |
498 | match all_copies.get(&parent_rev) { |
|
486 | match all_copies.get(&parent_rev) { | |
499 | Some(c) => Some(c.clone()), |
|
487 | Some(c) => Some(c.clone()), | |
500 | None => Some(InternalPathCopies::default()), |
|
488 | None => Some(InternalPathCopies::default()), | |
501 | } |
|
489 | } | |
502 | } |
|
490 | } | |
503 | } |
|
491 | } | |
504 |
|
492 | |||
505 | /// Combine ChangedFiles with some existing PathCopies information and return |
|
493 | /// Combine ChangedFiles with some existing PathCopies information and return | |
506 | /// the result |
|
494 | /// the result | |
507 | fn add_from_changes( |
|
495 | fn add_from_changes( | |
508 | path_map: &mut TwoWayPathMap, |
|
496 | path_map: &mut TwoWayPathMap, | |
509 | base_copies: &InternalPathCopies, |
|
497 | base_copies: &InternalPathCopies, | |
510 | changes: &ChangedFiles, |
|
498 | changes: &ChangedFiles, | |
511 | parent: Parent, |
|
499 | parent: Parent, | |
512 | current_rev: Revision, |
|
500 | current_rev: Revision, | |
513 | ) -> InternalPathCopies { |
|
501 | ) -> InternalPathCopies { | |
514 | let mut copies = base_copies.clone(); |
|
502 | let mut copies = base_copies.clone(); | |
515 | for action in changes.iter_actions(parent) { |
|
503 | for action in changes.iter_actions(parent) { | |
516 | match action { |
|
504 | match action { | |
517 | Action::Copied(path_dest, path_source) => { |
|
505 | Action::Copied(path_dest, path_source) => { | |
518 | let dest = path_map.tokenize(path_dest); |
|
506 | let dest = path_map.tokenize(path_dest); | |
519 | let source = path_map.tokenize(path_source); |
|
507 | let source = path_map.tokenize(path_source); | |
520 | let entry; |
|
508 | let entry; | |
521 | if let Some(v) = base_copies.get(&source) { |
|
509 | if let Some(v) = base_copies.get(&source) { | |
522 | entry = match &v.path { |
|
510 | entry = match &v.path { | |
523 | Some(path) => Some((*(path)).to_owned()), |
|
511 | Some(path) => Some((*(path)).to_owned()), | |
524 | None => Some(source.to_owned()), |
|
512 | None => Some(source.to_owned()), | |
525 | } |
|
513 | } | |
526 | } else { |
|
514 | } else { | |
527 | entry = Some(source.to_owned()); |
|
515 | entry = Some(source.to_owned()); | |
528 | } |
|
516 | } | |
529 | // Each new entry is introduced by the children, we |
|
517 | // Each new entry is introduced by the children, we | |
530 | // record this information as we will need it to take |
|
518 | // record this information as we will need it to take | |
531 | // the right decision when merging conflicting copy |
|
519 | // the right decision when merging conflicting copy | |
532 | // information. See merge_copies_dict for details. |
|
520 | // information. See merge_copies_dict for details. | |
533 | match copies.entry(dest) { |
|
521 | match copies.entry(dest) { | |
534 | Entry::Vacant(slot) => { |
|
522 | Entry::Vacant(slot) => { | |
535 | let ttpc = CopySource::new(current_rev, entry); |
|
523 | let ttpc = CopySource::new(current_rev, entry); | |
536 | slot.insert(ttpc); |
|
524 | slot.insert(ttpc); | |
537 | } |
|
525 | } | |
538 | Entry::Occupied(mut slot) => { |
|
526 | Entry::Occupied(mut slot) => { | |
539 | let ttpc = slot.get_mut(); |
|
527 | let ttpc = slot.get_mut(); | |
540 | ttpc.overwrite(current_rev, entry); |
|
528 | ttpc.overwrite(current_rev, entry); | |
541 | } |
|
529 | } | |
542 | } |
|
530 | } | |
543 | } |
|
531 | } | |
544 | Action::Removed(deleted_path) => { |
|
532 | Action::Removed(deleted_path) => { | |
545 | // We must drop copy information for removed file. |
|
533 | // We must drop copy information for removed file. | |
546 | // |
|
534 | // | |
547 | // We need to explicitly record them as dropped to |
|
535 | // We need to explicitly record them as dropped to | |
548 | // propagate this information when merging two |
|
536 | // propagate this information when merging two | |
549 | // InternalPathCopies object. |
|
537 | // InternalPathCopies object. | |
550 | let deleted = path_map.tokenize(deleted_path); |
|
538 | let deleted = path_map.tokenize(deleted_path); | |
551 | copies.entry(deleted).and_modify(|old| { |
|
539 | copies.entry(deleted).and_modify(|old| { | |
552 | old.mark_delete(current_rev); |
|
540 | old.mark_delete(current_rev); | |
553 | }); |
|
541 | }); | |
554 | } |
|
542 | } | |
555 | } |
|
543 | } | |
556 | } |
|
544 | } | |
557 | copies |
|
545 | copies | |
558 | } |
|
546 | } | |
559 |
|
547 | |||
560 | /// merge two copies-mapping together, minor and major |
|
548 | /// merge two copies-mapping together, minor and major | |
561 | /// |
|
549 | /// | |
562 | /// In case of conflict, value from "major" will be picked, unless in some |
|
550 | /// In case of conflict, value from "major" will be picked, unless in some | |
563 | /// cases. See inline documentation for details. |
|
551 | /// cases. See inline documentation for details. | |
564 | fn merge_copies_dict( |
|
552 | fn merge_copies_dict( | |
565 | path_map: &TwoWayPathMap, |
|
553 | path_map: &TwoWayPathMap, | |
566 | current_merge: Revision, |
|
554 | current_merge: Revision, | |
567 | mut minor: InternalPathCopies, |
|
555 | mut minor: InternalPathCopies, | |
568 | mut major: InternalPathCopies, |
|
556 | mut major: InternalPathCopies, | |
569 | changes: &ChangedFiles, |
|
557 | changes: &ChangedFiles, | |
570 | ) -> InternalPathCopies { |
|
558 | ) -> InternalPathCopies { | |
571 | // This closure exist as temporary help while multiple developper are |
|
559 | // This closure exist as temporary help while multiple developper are | |
572 | // actively working on this code. Feel free to re-inline it once this |
|
560 | // actively working on this code. Feel free to re-inline it once this | |
573 | // code is more settled. |
|
561 | // code is more settled. | |
574 | let cmp_value = |
|
562 | let cmp_value = | |
575 | |dest: &PathToken, src_minor: &CopySource, src_major: &CopySource| { |
|
563 | |dest: &PathToken, src_minor: &CopySource, src_major: &CopySource| { | |
576 | compare_value( |
|
564 | compare_value( | |
577 | path_map, |
|
565 | path_map, | |
578 | current_merge, |
|
566 | current_merge, | |
579 | changes, |
|
567 | changes, | |
580 | dest, |
|
568 | dest, | |
581 | src_minor, |
|
569 | src_minor, | |
582 | src_major, |
|
570 | src_major, | |
583 | ) |
|
571 | ) | |
584 | }; |
|
572 | }; | |
585 | if minor.is_empty() { |
|
573 | if minor.is_empty() { | |
586 | major |
|
574 | major | |
587 | } else if major.is_empty() { |
|
575 | } else if major.is_empty() { | |
588 | minor |
|
576 | minor | |
589 | } else if minor.len() * 2 < major.len() { |
|
577 | } else if minor.len() * 2 < major.len() { | |
590 | // Lets says we are merging two InternalPathCopies instance A and B. |
|
578 | // Lets says we are merging two InternalPathCopies instance A and B. | |
591 | // |
|
579 | // | |
592 | // If A contains N items, the merge result will never contains more |
|
580 | // If A contains N items, the merge result will never contains more | |
593 | // than N values differents than the one in A |
|
581 | // than N values differents than the one in A | |
594 | // |
|
582 | // | |
595 | // If B contains M items, with M > N, the merge result will always |
|
583 | // If B contains M items, with M > N, the merge result will always | |
596 | // result in a minimum of M - N value differents than the on in |
|
584 | // result in a minimum of M - N value differents than the on in | |
597 | // A |
|
585 | // A | |
598 | // |
|
586 | // | |
599 | // As a result, if N < (M-N), we know that simply iterating over A will |
|
587 | // As a result, if N < (M-N), we know that simply iterating over A will | |
600 | // yield less difference than iterating over the difference |
|
588 | // yield less difference than iterating over the difference | |
601 | // between A and B. |
|
589 | // between A and B. | |
602 | // |
|
590 | // | |
603 | // This help performance a lot in case were a tiny |
|
591 | // This help performance a lot in case were a tiny | |
604 | // InternalPathCopies is merged with a much larger one. |
|
592 | // InternalPathCopies is merged with a much larger one. | |
605 | for (dest, src_minor) in minor { |
|
593 | for (dest, src_minor) in minor { | |
606 | let src_major = major.get(&dest); |
|
594 | let src_major = major.get(&dest); | |
607 | match src_major { |
|
595 | match src_major { | |
608 | None => { |
|
596 | None => { | |
609 | major.insert(dest, src_minor); |
|
597 | major.insert(dest, src_minor); | |
610 | } |
|
598 | } | |
611 | Some(src_major) => { |
|
599 | Some(src_major) => { | |
612 | let (pick, overwrite) = |
|
600 | let (pick, overwrite) = | |
613 | cmp_value(&dest, &src_minor, src_major); |
|
601 | cmp_value(&dest, &src_minor, src_major); | |
614 | if overwrite { |
|
602 | if overwrite { | |
615 | let src = match pick { |
|
603 | let src = match pick { | |
616 | MergePick::Major => CopySource::new_from_merge( |
|
604 | MergePick::Major => CopySource::new_from_merge( | |
617 | current_merge, |
|
605 | current_merge, | |
618 | src_major, |
|
606 | src_major, | |
619 | &src_minor, |
|
607 | &src_minor, | |
620 | ), |
|
608 | ), | |
621 | MergePick::Minor => CopySource::new_from_merge( |
|
609 | MergePick::Minor => CopySource::new_from_merge( | |
622 | current_merge, |
|
610 | current_merge, | |
623 | &src_minor, |
|
611 | &src_minor, | |
624 | src_major, |
|
612 | src_major, | |
625 | ), |
|
613 | ), | |
626 | MergePick::Any => CopySource::new_from_merge( |
|
614 | MergePick::Any => CopySource::new_from_merge( | |
627 | current_merge, |
|
615 | current_merge, | |
628 | src_major, |
|
616 | src_major, | |
629 | &src_minor, |
|
617 | &src_minor, | |
630 | ), |
|
618 | ), | |
631 | }; |
|
619 | }; | |
632 | major.insert(dest, src); |
|
620 | major.insert(dest, src); | |
633 | } else { |
|
621 | } else { | |
634 | match pick { |
|
622 | match pick { | |
635 | MergePick::Any | MergePick::Major => None, |
|
623 | MergePick::Any | MergePick::Major => None, | |
636 | MergePick::Minor => major.insert(dest, src_minor), |
|
624 | MergePick::Minor => major.insert(dest, src_minor), | |
637 | }; |
|
625 | }; | |
638 | } |
|
626 | } | |
639 | } |
|
627 | } | |
640 | }; |
|
628 | }; | |
641 | } |
|
629 | } | |
642 | major |
|
630 | major | |
643 | } else if major.len() * 2 < minor.len() { |
|
631 | } else if major.len() * 2 < minor.len() { | |
644 | // This use the same rational than the previous block. |
|
632 | // This use the same rational than the previous block. | |
645 | // (Check previous block documentation for details.) |
|
633 | // (Check previous block documentation for details.) | |
646 | for (dest, src_major) in major { |
|
634 | for (dest, src_major) in major { | |
647 | let src_minor = minor.get(&dest); |
|
635 | let src_minor = minor.get(&dest); | |
648 | match src_minor { |
|
636 | match src_minor { | |
649 | None => { |
|
637 | None => { | |
650 | minor.insert(dest, src_major); |
|
638 | minor.insert(dest, src_major); | |
651 | } |
|
639 | } | |
652 | Some(src_minor) => { |
|
640 | Some(src_minor) => { | |
653 | let (pick, overwrite) = |
|
641 | let (pick, overwrite) = | |
654 | cmp_value(&dest, src_minor, &src_major); |
|
642 | cmp_value(&dest, src_minor, &src_major); | |
655 | if overwrite { |
|
643 | if overwrite { | |
656 | let src = match pick { |
|
644 | let src = match pick { | |
657 | MergePick::Major => CopySource::new_from_merge( |
|
645 | MergePick::Major => CopySource::new_from_merge( | |
658 | current_merge, |
|
646 | current_merge, | |
659 | &src_major, |
|
647 | &src_major, | |
660 | src_minor, |
|
648 | src_minor, | |
661 | ), |
|
649 | ), | |
662 | MergePick::Minor => CopySource::new_from_merge( |
|
650 | MergePick::Minor => CopySource::new_from_merge( | |
663 | current_merge, |
|
651 | current_merge, | |
664 | src_minor, |
|
652 | src_minor, | |
665 | &src_major, |
|
653 | &src_major, | |
666 | ), |
|
654 | ), | |
667 | MergePick::Any => CopySource::new_from_merge( |
|
655 | MergePick::Any => CopySource::new_from_merge( | |
668 | current_merge, |
|
656 | current_merge, | |
669 | &src_major, |
|
657 | &src_major, | |
670 | src_minor, |
|
658 | src_minor, | |
671 | ), |
|
659 | ), | |
672 | }; |
|
660 | }; | |
673 | minor.insert(dest, src); |
|
661 | minor.insert(dest, src); | |
674 | } else { |
|
662 | } else { | |
675 | match pick { |
|
663 | match pick { | |
676 | MergePick::Any | MergePick::Minor => None, |
|
664 | MergePick::Any | MergePick::Minor => None, | |
677 | MergePick::Major => minor.insert(dest, src_major), |
|
665 | MergePick::Major => minor.insert(dest, src_major), | |
678 | }; |
|
666 | }; | |
679 | } |
|
667 | } | |
680 | } |
|
668 | } | |
681 | }; |
|
669 | }; | |
682 | } |
|
670 | } | |
683 | minor |
|
671 | minor | |
684 | } else { |
|
672 | } else { | |
685 | let mut override_minor = Vec::new(); |
|
673 | let mut override_minor = Vec::new(); | |
686 | let mut override_major = Vec::new(); |
|
674 | let mut override_major = Vec::new(); | |
687 |
|
675 | |||
688 | let mut to_major = |k: &PathToken, v: &CopySource| { |
|
676 | let mut to_major = |k: &PathToken, v: &CopySource| { | |
689 | override_major.push((k.clone(), v.clone())) |
|
677 | override_major.push((k.clone(), v.clone())) | |
690 | }; |
|
678 | }; | |
691 | let mut to_minor = |k: &PathToken, v: &CopySource| { |
|
679 | let mut to_minor = |k: &PathToken, v: &CopySource| { | |
692 | override_minor.push((k.clone(), v.clone())) |
|
680 | override_minor.push((k.clone(), v.clone())) | |
693 | }; |
|
681 | }; | |
694 |
|
682 | |||
695 | // The diff function leverage detection of the identical subpart if |
|
683 | // The diff function leverage detection of the identical subpart if | |
696 | // minor and major has some common ancestors. This make it very |
|
684 | // minor and major has some common ancestors. This make it very | |
697 | // fast is most case. |
|
685 | // fast is most case. | |
698 | // |
|
686 | // | |
699 | // In case where the two map are vastly different in size, the current |
|
687 | // In case where the two map are vastly different in size, the current | |
700 | // approach is still slowish because the iteration will iterate over |
|
688 | // approach is still slowish because the iteration will iterate over | |
701 | // all the "exclusive" content of the larger on. This situation can be |
|
689 | // all the "exclusive" content of the larger on. This situation can be | |
702 | // frequent when the subgraph of revision we are processing has a lot |
|
690 | // frequent when the subgraph of revision we are processing has a lot | |
703 | // of roots. Each roots adding they own fully new map to the mix (and |
|
691 | // of roots. Each roots adding they own fully new map to the mix (and | |
704 | // likely a small map, if the path from the root to the "main path" is |
|
692 | // likely a small map, if the path from the root to the "main path" is | |
705 | // small. |
|
693 | // small. | |
706 | // |
|
694 | // | |
707 | // We could do better by detecting such situation and processing them |
|
695 | // We could do better by detecting such situation and processing them | |
708 | // differently. |
|
696 | // differently. | |
709 | for d in minor.diff(&major) { |
|
697 | for d in minor.diff(&major) { | |
710 | match d { |
|
698 | match d { | |
711 | DiffItem::Add(k, v) => to_minor(k, v), |
|
699 | DiffItem::Add(k, v) => to_minor(k, v), | |
712 | DiffItem::Remove(k, v) => to_major(k, v), |
|
700 | DiffItem::Remove(k, v) => to_major(k, v), | |
713 | DiffItem::Update { old, new } => { |
|
701 | DiffItem::Update { old, new } => { | |
714 | let (dest, src_major) = new; |
|
702 | let (dest, src_major) = new; | |
715 | let (_, src_minor) = old; |
|
703 | let (_, src_minor) = old; | |
716 | let (pick, overwrite) = |
|
704 | let (pick, overwrite) = | |
717 | cmp_value(dest, src_minor, src_major); |
|
705 | cmp_value(dest, src_minor, src_major); | |
718 | if overwrite { |
|
706 | if overwrite { | |
719 | let src = match pick { |
|
707 | let src = match pick { | |
720 | MergePick::Major => CopySource::new_from_merge( |
|
708 | MergePick::Major => CopySource::new_from_merge( | |
721 | current_merge, |
|
709 | current_merge, | |
722 | src_major, |
|
710 | src_major, | |
723 | src_minor, |
|
711 | src_minor, | |
724 | ), |
|
712 | ), | |
725 | MergePick::Minor => CopySource::new_from_merge( |
|
713 | MergePick::Minor => CopySource::new_from_merge( | |
726 | current_merge, |
|
714 | current_merge, | |
727 | src_minor, |
|
715 | src_minor, | |
728 | src_major, |
|
716 | src_major, | |
729 | ), |
|
717 | ), | |
730 | MergePick::Any => CopySource::new_from_merge( |
|
718 | MergePick::Any => CopySource::new_from_merge( | |
731 | current_merge, |
|
719 | current_merge, | |
732 | src_major, |
|
720 | src_major, | |
733 | src_minor, |
|
721 | src_minor, | |
734 | ), |
|
722 | ), | |
735 | }; |
|
723 | }; | |
736 | to_minor(dest, &src); |
|
724 | to_minor(dest, &src); | |
737 | to_major(dest, &src); |
|
725 | to_major(dest, &src); | |
738 | } else { |
|
726 | } else { | |
739 | match pick { |
|
727 | match pick { | |
740 | MergePick::Major => to_minor(dest, src_major), |
|
728 | MergePick::Major => to_minor(dest, src_major), | |
741 | MergePick::Minor => to_major(dest, src_minor), |
|
729 | MergePick::Minor => to_major(dest, src_minor), | |
742 | // If the two entry are identical, no need to do |
|
730 | // If the two entry are identical, no need to do | |
743 | // anything (but diff should not have yield them) |
|
731 | // anything (but diff should not have yield them) | |
744 | MergePick::Any => unreachable!(), |
|
732 | MergePick::Any => unreachable!(), | |
745 | } |
|
733 | } | |
746 | } |
|
734 | } | |
747 | } |
|
735 | } | |
748 | }; |
|
736 | }; | |
749 | } |
|
737 | } | |
750 |
|
738 | |||
751 | let updates; |
|
739 | let updates; | |
752 | let mut result; |
|
740 | let mut result; | |
753 | if override_major.is_empty() { |
|
741 | if override_major.is_empty() { | |
754 | result = major |
|
742 | result = major | |
755 | } else if override_minor.is_empty() { |
|
743 | } else if override_minor.is_empty() { | |
756 | result = minor |
|
744 | result = minor | |
757 | } else { |
|
745 | } else { | |
758 | if override_minor.len() < override_major.len() { |
|
746 | if override_minor.len() < override_major.len() { | |
759 | updates = override_minor; |
|
747 | updates = override_minor; | |
760 | result = minor; |
|
748 | result = minor; | |
761 | } else { |
|
749 | } else { | |
762 | updates = override_major; |
|
750 | updates = override_major; | |
763 | result = major; |
|
751 | result = major; | |
764 | } |
|
752 | } | |
765 | for (k, v) in updates { |
|
753 | for (k, v) in updates { | |
766 | result.insert(k, v); |
|
754 | result.insert(k, v); | |
767 | } |
|
755 | } | |
768 | } |
|
756 | } | |
769 | result |
|
757 | result | |
770 | } |
|
758 | } | |
771 | } |
|
759 | } | |
772 |
|
760 | |||
773 | /// represent the side that should prevail when merging two |
|
761 | /// represent the side that should prevail when merging two | |
774 | /// InternalPathCopies |
|
762 | /// InternalPathCopies | |
775 | enum MergePick { |
|
763 | enum MergePick { | |
776 | /// The "major" (p1) side prevails |
|
764 | /// The "major" (p1) side prevails | |
777 | Major, |
|
765 | Major, | |
778 | /// The "minor" (p2) side prevails |
|
766 | /// The "minor" (p2) side prevails | |
779 | Minor, |
|
767 | Minor, | |
780 | /// Any side could be used (because they are the same) |
|
768 | /// Any side could be used (because they are the same) | |
781 | Any, |
|
769 | Any, | |
782 | } |
|
770 | } | |
783 |
|
771 | |||
784 | /// decide which side prevails in case of conflicting values |
|
772 | /// decide which side prevails in case of conflicting values | |
785 | #[allow(clippy::if_same_then_else)] |
|
773 | #[allow(clippy::if_same_then_else)] | |
786 | fn compare_value( |
|
774 | fn compare_value( | |
787 | path_map: &TwoWayPathMap, |
|
775 | path_map: &TwoWayPathMap, | |
788 | current_merge: Revision, |
|
776 | current_merge: Revision, | |
789 | changes: &ChangedFiles, |
|
777 | changes: &ChangedFiles, | |
790 | dest: &PathToken, |
|
778 | dest: &PathToken, | |
791 | src_minor: &CopySource, |
|
779 | src_minor: &CopySource, | |
792 | src_major: &CopySource, |
|
780 | src_major: &CopySource, | |
793 | ) -> (MergePick, bool) { |
|
781 | ) -> (MergePick, bool) { | |
794 | if src_major.rev == current_merge { |
|
782 | if src_major.rev == current_merge { | |
795 | if src_minor.rev == current_merge { |
|
783 | if src_minor.rev == current_merge { | |
796 | if src_major.path.is_none() { |
|
784 | if src_major.path.is_none() { | |
797 | // We cannot get different copy information for both p1 and p2 |
|
785 | // We cannot get different copy information for both p1 and p2 | |
798 | // from the same revision. Unless this was a |
|
786 | // from the same revision. Unless this was a | |
799 | // deletion. |
|
787 | // deletion. | |
800 | // |
|
788 | // | |
801 | // However the deletion might come over different data on each |
|
789 | // However the deletion might come over different data on each | |
802 | // branch. |
|
790 | // branch. | |
803 | let need_over = src_major.overwritten != src_minor.overwritten; |
|
791 | let need_over = src_major.overwritten != src_minor.overwritten; | |
804 | (MergePick::Any, need_over) |
|
792 | (MergePick::Any, need_over) | |
805 | } else { |
|
793 | } else { | |
806 | unreachable!(); |
|
794 | unreachable!(); | |
807 | } |
|
795 | } | |
808 | } else { |
|
796 | } else { | |
809 | // The last value comes the current merge, this value -will- win |
|
797 | // The last value comes the current merge, this value -will- win | |
810 | // eventually. |
|
798 | // eventually. | |
811 | (MergePick::Major, true) |
|
799 | (MergePick::Major, true) | |
812 | } |
|
800 | } | |
813 | } else if src_minor.rev == current_merge { |
|
801 | } else if src_minor.rev == current_merge { | |
814 | // The last value comes the current merge, this value -will- win |
|
802 | // The last value comes the current merge, this value -will- win | |
815 | // eventually. |
|
803 | // eventually. | |
816 | (MergePick::Minor, true) |
|
804 | (MergePick::Minor, true) | |
817 | } else if src_major.path == src_minor.path { |
|
805 | } else if src_major.path == src_minor.path { | |
818 | // we have the same value, but from other source; |
|
806 | // we have the same value, but from other source; | |
819 | if src_major.rev == src_minor.rev { |
|
807 | if src_major.rev == src_minor.rev { | |
820 | // If the two entry are identical, they are both valid |
|
808 | // If the two entry are identical, they are both valid | |
821 | debug_assert!(src_minor.overwritten == src_minor.overwritten); |
|
809 | debug_assert!(src_minor.overwritten == src_minor.overwritten); | |
822 | (MergePick::Any, false) |
|
810 | (MergePick::Any, false) | |
823 | } else if src_major.is_overwritten_by(src_minor) { |
|
811 | } else if src_major.is_overwritten_by(src_minor) { | |
824 | (MergePick::Minor, false) |
|
812 | (MergePick::Minor, false) | |
825 | } else if src_minor.is_overwritten_by(src_major) { |
|
813 | } else if src_minor.is_overwritten_by(src_major) { | |
826 | (MergePick::Major, false) |
|
814 | (MergePick::Major, false) | |
827 | } else { |
|
815 | } else { | |
828 | (MergePick::Any, true) |
|
816 | (MergePick::Any, true) | |
829 | } |
|
817 | } | |
830 | } else if src_major.rev == src_minor.rev { |
|
818 | } else if src_major.rev == src_minor.rev { | |
831 | // We cannot get copy information for both p1 and p2 in the |
|
819 | // We cannot get copy information for both p1 and p2 in the | |
832 | // same rev. So this is the same value. |
|
820 | // same rev. So this is the same value. | |
833 | unreachable!( |
|
821 | unreachable!( | |
834 | "conflicting information from p1 and p2 in the same revision" |
|
822 | "conflicting information from p1 and p2 in the same revision" | |
835 | ); |
|
823 | ); | |
836 | } else { |
|
824 | } else { | |
837 | let dest_path = path_map.untokenize(*dest); |
|
825 | let dest_path = path_map.untokenize(*dest); | |
838 | let action = changes.get_merge_case(dest_path); |
|
826 | let action = changes.get_merge_case(dest_path); | |
839 | if src_minor.path.is_some() |
|
827 | if src_minor.path.is_some() | |
840 | && src_major.path.is_none() |
|
828 | && src_major.path.is_none() | |
841 | && action == MergeCase::Salvaged |
|
829 | && action == MergeCase::Salvaged | |
842 | { |
|
830 | { | |
843 | // If the file is "deleted" in the major side but was |
|
831 | // If the file is "deleted" in the major side but was | |
844 | // salvaged by the merge, we keep the minor side alive |
|
832 | // salvaged by the merge, we keep the minor side alive | |
845 | (MergePick::Minor, true) |
|
833 | (MergePick::Minor, true) | |
846 | } else if src_major.path.is_some() |
|
834 | } else if src_major.path.is_some() | |
847 | && src_minor.path.is_none() |
|
835 | && src_minor.path.is_none() | |
848 | && action == MergeCase::Salvaged |
|
836 | && action == MergeCase::Salvaged | |
849 | { |
|
837 | { | |
850 | // If the file is "deleted" in the minor side but was |
|
838 | // If the file is "deleted" in the minor side but was | |
851 | // salvaged by the merge, unconditionnaly preserve the |
|
839 | // salvaged by the merge, unconditionnaly preserve the | |
852 | // major side. |
|
840 | // major side. | |
853 | (MergePick::Major, true) |
|
841 | (MergePick::Major, true) | |
854 | } else if src_minor.is_overwritten_by(src_major) { |
|
842 | } else if src_minor.is_overwritten_by(src_major) { | |
855 | // The information from the minor version are strictly older than |
|
843 | // The information from the minor version are strictly older than | |
856 | // the major version |
|
844 | // the major version | |
857 | if action == MergeCase::Merged { |
|
845 | if action == MergeCase::Merged { | |
858 | // If the file was actively merged, its means some non-copy |
|
846 | // If the file was actively merged, its means some non-copy | |
859 | // activity happened on the other branch. It |
|
847 | // activity happened on the other branch. It | |
860 | // mean the older copy information are still relevant. |
|
848 | // mean the older copy information are still relevant. | |
861 | // |
|
849 | // | |
862 | // The major side wins such conflict. |
|
850 | // The major side wins such conflict. | |
863 | (MergePick::Major, true) |
|
851 | (MergePick::Major, true) | |
864 | } else { |
|
852 | } else { | |
865 | // No activity on the minor branch, pick the newer one. |
|
853 | // No activity on the minor branch, pick the newer one. | |
866 | (MergePick::Major, false) |
|
854 | (MergePick::Major, false) | |
867 | } |
|
855 | } | |
868 | } else if src_major.is_overwritten_by(src_minor) { |
|
856 | } else if src_major.is_overwritten_by(src_minor) { | |
869 | if action == MergeCase::Merged { |
|
857 | if action == MergeCase::Merged { | |
870 | // If the file was actively merged, its means some non-copy |
|
858 | // If the file was actively merged, its means some non-copy | |
871 | // activity happened on the other branch. It |
|
859 | // activity happened on the other branch. It | |
872 | // mean the older copy information are still relevant. |
|
860 | // mean the older copy information are still relevant. | |
873 | // |
|
861 | // | |
874 | // The major side wins such conflict. |
|
862 | // The major side wins such conflict. | |
875 | (MergePick::Major, true) |
|
863 | (MergePick::Major, true) | |
876 | } else { |
|
864 | } else { | |
877 | // No activity on the minor branch, pick the newer one. |
|
865 | // No activity on the minor branch, pick the newer one. | |
878 | (MergePick::Minor, false) |
|
866 | (MergePick::Minor, false) | |
879 | } |
|
867 | } | |
880 | } else if src_minor.path.is_none() { |
|
868 | } else if src_minor.path.is_none() { | |
881 | // the minor side has no relevant information, pick the alive one |
|
869 | // the minor side has no relevant information, pick the alive one | |
882 | (MergePick::Major, true) |
|
870 | (MergePick::Major, true) | |
883 | } else if src_major.path.is_none() { |
|
871 | } else if src_major.path.is_none() { | |
884 | // the major side has no relevant information, pick the alive one |
|
872 | // the major side has no relevant information, pick the alive one | |
885 | (MergePick::Minor, true) |
|
873 | (MergePick::Minor, true) | |
886 | } else { |
|
874 | } else { | |
887 | // by default the major side wins |
|
875 | // by default the major side wins | |
888 | (MergePick::Major, true) |
|
876 | (MergePick::Major, true) | |
889 | } |
|
877 | } | |
890 | } |
|
878 | } | |
891 | } |
|
879 | } |
General Comments 0
You need to be logged in to leave comments.
Login now