OXYGENHUB
# 2b95bc24d7d6 oxygen / crates / oxy-core / src / repo / snapshot.rs
118.2 KiB · Rust · last changed Snapshots: large new files left out and named; past working copies expire 25 days ago source blame history raw
98286406 Andy McCall ↶ 1 
//! Working-copy snapshots, commits, and the change builder (spec §12.3).
2
3 
use super::*;
957fee30 Andy McCall ↶ 4 
use crate::objects::AtomRef;
98286406 Andy McCall ↶ 5
6 
#[derive(Clone, Copy, Debug, PartialEq, Eq)]
7 
pub enum ChangeStatus {
8 
    Added,
9 
    Modified,
10 
    Removed,
6e723b81 Andy McCall ↶ 11 
    /// Moved from another path (detected by similarity, tracked as a placement change).
12 
    Renamed,
98286406 Andy McCall ↶ 13 
    ModeChanged,
14 
    /// The file is conflicted in the base state and still contains conflict markers.
15 
    ConflictEdited,
16 
    /// The file was conflicted in the base state and the working copy resolves it.
17 
    Resolved,
18 
}
19
20 
/// Lines that mark a conflict region in a materialized file (spec §9.5).
21 
pub fn has_conflict_markers(content: &[u8]) -> bool {
22 
    split_lines(content).iter().any(|l| {
75167672 Andy McCall 23 
        l.starts_with(b"<<<<<<< oxy conflict")
24 
            || l.starts_with(b"======= oxy conflict")
25 
            || l.starts_with(b">>>>>>> oxy conflict")
98286406 Andy McCall ↶ 26 
    })
27 
}
28
29 
#[derive(Clone, Debug, PartialEq, Eq)]
30 
pub struct PathStatus {
31 
    pub path: String,
32 
    pub status: ChangeStatus,
33 
}
34
614541c3 Andy McCall ↶ 35 
/// Operations of a change built from the working directory against a base.
36 
pub(crate) struct BuiltChange {
37 
    pub tree_ops: Vec<TreeOp>,
38 
    pub edits: Vec<Edit>,
39 
    pub contributors: Vec<ChangeHash>,
40 
    pub resolves: bool,
41 
    pub statuses: Vec<PathStatus>,
b7f102fc Andy McCall ↶ 42 
    /// The stored blob of every opaque path, for the manifest's stat cache.
43 
    pub blobs: HashMap<String, ObjectRef>,
614541c3 Andy McCall ↶ 44 
}
45
98286406 Andy McCall ↶ 46 
/// Result of snapshotting the working copy.
47 
pub struct SnapshotResult {
5b4c5130 Andy McCall ↶ 48 
    /// The working-copy change's hash; for a commit's own snapshot, which
49 
    /// does not encode it, the hash it had before.
98286406 Andy McCall ↶ 50 
    pub wc_hash: ChangeHash,
51 
    pub wc: Change,
52 
    /// The base state (closure of the working-copy change's dependencies).
53 
    pub base: State,
54 
    pub base_snapshot: Snapshot,
55 
    pub statuses: Vec<PathStatus>,
56 
    /// Whether the snapshot rewrote the working-copy change.
57 
    pub changed: bool,
58 
    /// Whether the working copy resolves conflicts of its base (spec §9.6).
59 
    pub resolves: bool,
60 
    /// Changes that contributed to the conflicts being resolved.
61 
    pub contributors: Vec<ChangeHash>,
f9466021 Andy McCall ↶ 62 
    /// The `.oxyattributes` in force on disk, from the files the walk read.
63 
    pub(crate) attributes: crate::attributes::Attributes,
d4dbab97 Andy McCall ↶ 64 
    /// Folders holding a repository of their own (`.git` or `.oxy`) that
65 
    /// are not subrepos, so nothing in them is tracked (oxygen#43).
66 
    pub untracked_nested: Vec<String>,
236f35c0 Andy McCall ↶ 67 
    /// A commit's snapshot leaves the manifest to the commit, which writes
68 
    /// it once with the new base's graph keys.
69 
    pub(crate) manifest: Option<PendingManifest>,
98286406 Andy McCall ↶ 70 
}
71
236f35c0 Andy McCall ↶ 72 
/// The manifest a commit's snapshot would have written, as of the moment
73 
/// it would have written it (that moment is the racy-window stamp).
74 
pub(crate) struct PendingManifest {
75 
    entries: BTreeMap<String, FsEntry>,
76 
    blobs: HashMap<String, ObjectRef>,
77 
    dirs: Option<super::WalkedDirs>,
78 
    written: u128,
79 
}
80
98286406 Andy McCall ↶ 81 
pub fn is_binary(content: &[u8]) -> bool {
82 
    content[..content.len().min(BINARY_SNIFF)].contains(&0)
83 
}
84
85 
impl Repo {
614541c3 Andy McCall ↶ 86 
    /// Diff the files on disk against `base` and build the operations of a
87 
    /// change from `base` to the working directory. `reuse` maps paths to the
88 
    /// NodeIds an earlier version of the same change assigned to new paths.
75167672 Andy McCall 89 
    pub(crate) fn build_against(
90 
        &self,
91 
        base: &State,
92 
        base_snapshot: &Snapshot,
93 
        reuse: HashMap<String, (NodeId, NodeKind)>,
94 
    ) -> Result<BuiltChange> {
c863c7a7 Andy McCall ↶ 95 
        let mut fs_entries =
696bbf29 Andy McCall ↶ 96 
            self.walk_working_copy_with(None, &Self::tracked_kinds(base_snapshot), None)?;
c863c7a7 Andy McCall ↶ 97 
        self.platform_normalize(&mut fs_entries, base_snapshot);
c3069537 Andy McCall ↶ 98 
        self.build_from_entries(base, base_snapshot, reuse, &fs_entries)
99 
    }
100
c863c7a7 Andy McCall ↶ 101 
    /// Read the working copy as the platform left it (spec §7.7): without
102 
    /// executable bits a file keeps the bit it has in the base, and without
103 
    /// symbolic links a plain file where the base has a symlink is that
104 
    /// symlink, written out as its target, not a new file.
105 
    fn platform_normalize(&self, entries: &mut BTreeMap<String, FsEntry>, base: &Snapshot) {
106 
        let p = self.platform;
107 
        if p.exec_bits && p.symlinks {
108 
            return;
109 
        }
110 
        for (path, e) in entries.iter_mut() {
111 
            let Some(f) = base.files.get(path) else {
112 
                continue;
113 
            };
114 
            if !p.symlinks && f.kind == NodeKind::Symlink && e.kind == NodeKind::File {
115 
                e.kind = NodeKind::Symlink;
116 
                e.exec = false;
117 
            }
118 
            if !p.exec_bits && f.kind == NodeKind::File && e.kind == NodeKind::File {
119 
                e.exec = f.mode.exec;
120 
            }
121 
        }
122 
    }
123
c3069537 Andy McCall ↶ 124 
    /// Like `build_against`, for files given in memory (imports).
75167672 Andy McCall 125 
    pub(crate) fn build_from_entries(
126 
        &self,
127 
        base: &State,
128 
        base_snapshot: &Snapshot,
129 
        reuse: HashMap<String, (NodeId, NodeKind)>,
130 
        fs_entries: &BTreeMap<String, FsEntry>,
131 
    ) -> Result<BuiltChange> {
f7510b63 Andy McCall ↶ 132 
        self.build_from_entries_resolving(base, base_snapshot, reuse, fs_entries, &HashSet::new())
133 
    }
134
135 
    /// Like `build_from_entries`, settling the replace conflicts of the
136 
    /// opaque files in `resolve` to their content (spec §9.3, §9.6): a
137 
    /// caller that knows the user chose it, such as the hub's resolver. A
138 
    /// snapshot never settles one on its own, since a checkout writes one
139 
    /// candidate at the path.
140 
    pub(crate) fn build_from_entries_resolving(
141 
        &self,
142 
        base: &State,
143 
        base_snapshot: &Snapshot,
144 
        reuse: HashMap<String, (NodeId, NodeKind)>,
145 
        fs_entries: &BTreeMap<String, FsEntry>,
146 
        resolve: &HashSet<String>,
147 
    ) -> Result<BuiltChange> {
23ff1288 Andy McCall ↶ 148 
        self.build_from_entries_with(
149 
            base,
150 
            base_snapshot,
151 
            reuse,
152 
            fs_entries,
153 
            resolve,
154 
            Scope::default(),
155 
        )
dfb23a3a Andy McCall ↶ 156 
    }
157
158 
    /// `build_from_entries` whose new nodes take identifiers derived from
159 
    /// `seed` and their paths instead of random ones, so the same input
160 
    /// always builds the same change (a Git import, spec §23).
2306a4d2 Andy McCall ↶ 161 
    ///
162 
    /// With `dir_needed`, the snapshot and entries hold only the paths the
163 
    /// import touches (and the directories a removal may leave empty); a
164 
    /// directory stays while `dir_needed` says the new tree has something
165 
    /// under it.
dfb23a3a Andy McCall ↶ 166 
    pub(crate) fn build_from_entries_seeded(
167 
        &self,
168 
        base: &State,
169 
        base_snapshot: &Snapshot,
170 
        fs_entries: &BTreeMap<String, FsEntry>,
171 
        seed: [u8; 32],
2306a4d2 Andy McCall ↶ 172 
        dir_needed: Option<&dyn Fn(&str) -> bool>,
dfb23a3a Andy McCall ↶ 173 
    ) -> Result<BuiltChange> {
174 
        self.build_from_entries_with(
175 
            base,
176 
            base_snapshot,
177 
            HashMap::new(),
178 
            fs_entries,
179 
            &HashSet::new(),
23ff1288 Andy McCall ↶ 180 
            Scope {
181 
                seed: Some(seed),
182 
                dir_needed,
183 
            },
dfb23a3a Andy McCall ↶ 184 
        )
185 
    }
186
23ff1288 Andy McCall ↶ 187 
    /// `build_from_entries` for a snapshot most of whose files are known
188 
    /// unchanged (oxygen#41): only the paths that may differ are built
189 
    /// (files read rather than known unchanged, base files gone from disk,
190 
    /// and the directories those may leave empty), against a snapshot of
191 
    /// just those, so a large tree with a few edits costs what they touch.
192 
    fn build_changed_only(
193 
        &self,
194 
        base: &State,
195 
        base_snapshot: &Snapshot,
196 
        reuse: HashMap<String, (NodeId, NodeKind)>,
197 
        fs_entries: &BTreeMap<String, FsEntry>,
198 
    ) -> Result<BuiltChange> {
d616faf5 Andy McCall ↶ 199 
        let pick = crate::timing::phase("snapshot:   of which picking the changed");
23ff1288 Andy McCall ↶ 200 
        // A file known unchanged may still have changed mode: that alters
201 
        // neither its size nor its time.
d616faf5 Andy McCall ↶ 202 
        let changed: BTreeMap<String, FsEntry> = join(fs_entries, &base_snapshot.files)
203 
            .filter_map(|(p, e, f)| {
204 
                let e = e?;
205 
                (!e.unchanged || f.is_none_or(|f| f.kind != e.kind || f.mode.exec != e.exec))
206 
                    .then(|| (p.clone(), e.clone()))
23ff1288 Andy McCall ↶ 207 
            })
208 
            .collect();
209 
        let mut files: BTreeMap<String, crate::state::FileState> = BTreeMap::new();
210 
        for p in changed.keys() {
211 
            if let Some(f) = base_snapshot.files.get(p) {
212 
                files.insert(p.clone(), f.clone());
213 
            }
214 
        }
d616faf5 Andy McCall ↶ 215 
        // Paths the base has and the disk does not.
216 
        let gone = join(&base_snapshot.files, fs_entries)
217 
            .filter_map(|(p, f, e)| (e.is_none()).then_some((p, f?)));
218 
        for (p, f) in gone {
219 
            if f.kind == NodeKind::Dir {
23ff1288 Andy McCall ↶ 220 
                continue;
221 
            }
222 
            files.insert(p.clone(), f.clone());
223 
            let mut end = p.len();
224 
            while let Some(i) = p[..end].rfind('/') {
225 
                let dir = &p[..i];
226 
                if files.contains_key(dir) {
227 
                    break;
228 
                }
229 
                if let Some(d) = base_snapshot.files.get(dir) {
230 
                    files.insert(dir.to_string(), d.clone());
231 
                }
232 
                end = i;
233 
            }
234 
        }
235 
        let part = Snapshot {
236 
            tree: base_snapshot.tree.clone(),
237 
            files,
238 
            unrendered: BTreeSet::new(),
239 
        };
d616faf5 Andy McCall ↶ 240 
        drop(pick);
241 
        let _build = crate::timing::phase("snapshot:   of which building the change");
23ff1288 Andy McCall ↶ 242 
        // Whether anything on disk is under `dir`.
243 
        let dir_needed = |dir: &str| {
244 
            let prefix = format!("{dir}/");
245 
            fs_entries
246 
                .range(prefix.clone()..)
247 
                .next()
248 
                .is_some_and(|(k, _)| k.starts_with(&prefix))
249 
        };
250 
        self.build_from_entries_with(
251 
            base,
252 
            &part,
253 
            reuse,
254 
            &changed,
255 
            &HashSet::new(),
256 
            Scope {
257 
                seed: None,
258 
                dir_needed: Some(&dir_needed),
259 
            },
260 
        )
261 
    }
262
dfb23a3a Andy McCall ↶ 263 
    fn build_from_entries_with(
264 
        &self,
265 
        base: &State,
266 
        base_snapshot: &Snapshot,
267 
        reuse: HashMap<String, (NodeId, NodeKind)>,
268 
        fs_entries: &BTreeMap<String, FsEntry>,
269 
        resolve: &HashSet<String>,
23ff1288 Andy McCall ↶ 270 
        scope: Scope,
dfb23a3a Andy McCall ↶ 271 
    ) -> Result<BuiltChange> {
23ff1288 Andy McCall ↶ 272 
        let Scope { seed, dir_needed } = scope;
2306a4d2 Andy McCall ↶ 273 
        let mut laps = crate::timing::Laps::start();
cdb5beba Andy McCall ↶ 274 
        // Attributes come from the files being committed, so a new
275 
        // `.oxyattributes` applies to the same change that adds it.
276 
        let attrs = crate::attributes::Attributes::from_files(
75167672 Andy McCall 277 
            fs_entries
278 
                .iter()
279 
                .filter(|(p, _)| p.as_str() == ".oxyattributes" || p.ends_with("/.oxyattributes"))
280 
                .map(|(p, e)| (p.as_str(), e.content.as_slice())),
cdb5beba Andy McCall ↶ 281 
        );
282 
        let mut b = ChangeBuilder::new(base, base_snapshot, reuse).with_attributes(attrs);
f7510b63 Andy McCall ↶ 283 
        b.resolve = resolve.clone();
dfb23a3a Andy McCall ↶ 284 
        b.node_seed = seed;
2306a4d2 Andy McCall ↶ 285 
        laps.lap("build: builder");
6e723b81 Andy McCall ↶ 286 
        // Rename detection (spec §7.1, §23): a removed file and an added file
287 
        // of the same kind whose contents are at least half alike become one
288 
        // move of the same node, so identity survives across the rename.
289 
        let renames = detect_renames(base_snapshot, fs_entries, |p| self.in_scope(p));
75167672 Andy McCall 290 
        let renamed_from: HashMap<&str, &str> = renames
291 
            .iter()
292 
            .map(|(old, new)| (new.as_str(), old.as_str()))
293 
            .collect();
294 
        let renamed_to: HashMap<&str, &str> = renames
295 
            .iter()
296 
            .map(|(old, new)| (old.as_str(), new.as_str()))
297 
            .collect();
2306a4d2 Andy McCall ↶ 298 
        laps.lap("build: renames");
360aa74a Andy McCall ↶ 299 
        b.diff_in_parallel(fs_entries);
300 
        laps.lap("build: diffs");
c3069537 Andy McCall ↶ 301 
        for (path, fs_entry) in fs_entries {
98286406 Andy McCall ↶ 302 
            match base_snapshot.files.get(path) {
6e723b81 Andy McCall ↶ 303 
                None => match renamed_from.get(path.as_str()) {
75167672 Andy McCall 304 
                    Some(old) => {
305 
                        b.move_path(self, old, path, fs_entry, &base_snapshot.files[*old])?
306 
                    }
6e723b81 Andy McCall ↶ 307 
                    None => b.add_path(self, path, fs_entry)?,
308 
                },
98286406 Andy McCall ↶ 309 
                Some(expected) => b.update_path(self, path, fs_entry, expected)?,
310 
            }
311 
        }
2306a4d2 Andy McCall ↶ 312 
        laps.lap("build: paths");
958fb27d Andy McCall ↶ 313 
        // A directory stays while anything under it exists or is out of
314 
        // scope: the directories above each such path, gathered once.
315 
        let mut needed_dirs: HashSet<&str> = HashSet::new();
316 
        fn parents_of<'a>(p: &'a str, set: &mut HashSet<&'a str>) {
317 
            let mut end = p.len();
318 
            while let Some(i) = p[..end].rfind('/') {
319 
                if !set.insert(&p[..i]) {
320 
                    break;
321 
                }
322 
                end = i;
323 
            }
324 
        }
325 
        for p in fs_entries.keys() {
326 
            parents_of(p, &mut needed_dirs);
327 
        }
328 
        for p in base_snapshot.files.keys() {
329 
            if !self.in_scope(p) {
330 
                parents_of(p, &mut needed_dirs);
331 
            }
332 
        }
98286406 Andy McCall ↶ 333 
        for (path, expected) in &base_snapshot.files {
75167672 Andy McCall 334 
            if fs_entries.contains_key(path)
335 
                || !self.in_scope(path)
336 
                || renamed_to.contains_key(path.as_str())
337 
            {
98286406 Andy McCall ↶ 338 
                continue;
339 
            }
2306a4d2 Andy McCall ↶ 340 
            let still_needed = expected.kind == NodeKind::Dir
341 
                && match dir_needed {
342 
                    Some(f) => f(path),
343 
                    None => needed_dirs.contains(path.as_str()),
344 
                };
98286406 Andy McCall ↶ 345 
            if !still_needed {
346 
                b.remove_path(path, expected);
347 
            }
348 
        }
2306a4d2 Andy McCall ↶ 349 
        laps.lap("build: removals");
a87c7b9d Andy McCall ↶ 350 
        if seed.is_some() {
351 
            b.retire_stray_placements();
352 
        }
2306a4d2 Andy McCall ↶ 353 
        laps.lap("build: stray placements");
614541c3 Andy McCall ↶ 354 
        let mut statuses = b.statuses;
355 
        statuses.sort_by(|a, b| a.path.cmp(&b.path));
75167672 Andy McCall 356 
        Ok(BuiltChange {
357 
            tree_ops: b.tree_ops,
358 
            edits: b.edits,
359 
            contributors: b.contributors.into_iter().collect(),
360 
            resolves: b.resolves,
361 
            statuses,
362 
            blobs: b.blobs,
363 
        })
614541c3 Andy McCall ↶ 364 
    }
98286406 Andy McCall ↶ 365
c07e0e7d Andy McCall ↶ 366 
    /// Store a change and, when `core.sign = change` and we are its author,
367 
    /// sign it. A rewritten change by another author is stored unsigned: only
368 
    /// its author (or a delegate) can vouch for it.
614541c3 Andy McCall ↶ 369 
    pub(crate) fn store_signed(&self, change: &Change) -> Result<ChangeHash> {
c07e0e7d Andy McCall ↶ 370 
        let ours = change.author == self.author();
371 
        if self.config.sign_changes && ours && self.secret_key()?.is_none() {
614541c3 Andy McCall ↶ 372 
            return fail("core.sign is \"change\" but no secret key is available for author.key");
373 
        }
5b4c5130 Andy McCall ↶ 374 
        let storing = crate::timing::phase("store: the change");
614541c3 Andy McCall ↶ 375 
        let hash = self.store.put_change(change)?;
5b4c5130 Andy McCall ↶ 376 
        drop(storing);
c07e0e7d Andy McCall ↶ 377 
        if self.config.sign_changes && ours {
5b4c5130 Andy McCall ↶ 378 
            let _t = crate::timing::phase("store: sign it");
614541c3 Andy McCall ↶ 379 
            self.sign_change(&hash)?;
380 
        }
381 
        Ok(hash)
382 
    }
383
384 
    /// Snapshot the working copy into the working-copy change (spec §12.3).
cc7a0056 Andy McCall ↶ 385 
    /// The result carries the full base state and its whole rendering.
614541c3 Andy McCall ↶ 386 
    pub fn snapshot(&self, command: &str) -> Result<SnapshotResult> {
f9466021 Andy McCall ↶ 387 
        self.snapshot_as(command, false, false)
cc7a0056 Andy McCall ↶ 388 
    }
389
390 
    /// Snapshot the working copy as [`Repo::snapshot`] does, for commands
8ab265dc Andy McCall ↶ 391 
    /// that need its statuses, conflicts and base history, and the graphs
392 
    /// of the files it changes: status, log, show, commit, diff, hunks and
393 
    /// discard. Text files known unchanged since the last manifest are
cc7a0056 Andy McCall ↶ 394 
    /// neither read nor rendered: the result's `base` is a history-only
395 
    /// state without their graphs, and its `base_snapshot` lists them in
8ab265dc Andy McCall ↶ 396 
    /// `unrendered` with empty bytes. They are never conflicted, and no
397 
    /// change the working copy makes touches them.
cc7a0056 Andy McCall ↶ 398 
    pub fn snapshot_status(&self, command: &str) -> Result<SnapshotResult> {
f9466021 Andy McCall ↶ 399 
        self.snapshot_as(command, true, false)
cc7a0056 Andy McCall ↶ 400 
    }
401
cd4eb55e Andy McCall ↶ 402 
    /// `snapshot_as`; `for_commit` writes the manifest even where a status
403 
    /// would not bother (an edited file is read next time whatever it
404 
    /// says), since after the commit that file is unchanged, and records no
405 
    /// operation: the commit's own, recorded straight after, replaces the
406 
    /// working copy, so the snapshot's would cost a durable write for
407 
    /// nothing. Should the commit fail, the files are still on disk and the
408 
    /// next snapshot finds them.
f9466021 Andy McCall ↶ 409 
    fn snapshot_as(
410 
        &self,
411 
        command: &str,
412 
        partial: bool,
cd4eb55e Andy McCall ↶ 413 
        for_commit: bool,
f9466021 Andy McCall ↶ 414 
    ) -> Result<SnapshotResult> {
4e88e774 Andy McCall ↶ 415 
        let _lock = self.lock_snapshot()?;
8f10a07a Andy McCall ↶ 416 
        let batch = self.store.begin_batch()?;
cd4eb55e Andy McCall ↶ 417 
        let result = self.snapshot_locked(command, partial, for_commit);
8f10a07a Andy McCall ↶ 418 
        batch.finish()?;
419 
        result
420 
    }
421
f9466021 Andy McCall ↶ 422 
    fn snapshot_locked(
423 
        &self,
424 
        command: &str,
425 
        partial: bool,
cd4eb55e Andy McCall ↶ 426 
        for_commit: bool,
f9466021 Andy McCall ↶ 427 
    ) -> Result<SnapshotResult> {
0bdc3f91 Andy McCall ↶ 428 
        let _all = crate::timing::phase("snapshot");
614541c3 Andy McCall ↶ 429 
        let (op_ref, op) = self.head_op()?;
430 
        let (old_hash, old_wc) = self.working_copy()?;
cc7a0056 Andy McCall ↶ 431 
        // The base's history first: its graph keys say which files the
432 
        // manifest already knows equal to it.
433 
        let mut base = {
434 
            let _t = crate::timing::phase("snapshot: load base history");
f9575918 Andy McCall ↶ 435 
            self.load_state_indexed(&old_wc.deps)?
0bdc3f91 Andy McCall ↶ 436 
        };
d0d52107 Andy McCall ↶ 437 
        let view = {
438 
            let _t = crate::timing::phase("snapshot: tree view");
439 
            base.tree().view()
440 
        };
8b276e01 Andy McCall ↶ 441 
        let manifest = {
442 
            let _t = crate::timing::phase("snapshot: read manifest");
443 
            self.read_manifest()?.map(|(_, _, m)| m)
444 
        };
cc7a0056 Andy McCall ↶ 445 
        let tracked: BTreeMap<String, (NodeKind, Option<ObjectRef>)> = {
446 
            let _t = crate::timing::phase("snapshot: tracked paths");
36e51aa8 Andy McCall ↶ 447 
            // In the view's path order, so the map is built without searching.
448 
            view.entries
449 
                .iter()
450 
                .filter(|(_, e)| e.kind != NodeKind::Dir)
451 
                .map(|(p, e)| {
452 
                    let blob = if e.kind != NodeKind::File || e.mode.opaque {
453 
                        base.live_blobs(&e.node).first().copied()
454 
                    } else {
455 
                        None
456 
                    };
457 
                    (p.clone(), (e.kind, blob))
458 
                })
459 
                .collect()
cc7a0056 Andy McCall ↶ 460 
        };
992467d3 Andy McCall ↶ 461 
        // A text file's graph key, worked out when the walk asks for it.
1966f5db Andy McCall ↶ 462 
        let key_of = |p: &str| -> Option<[u8; 32]> {
992467d3 Andy McCall ↶ 463 
            let e = view.entries.get(p)?;
464 
            if e.kind != NodeKind::File || e.mode.opaque {
465 
                return None;
466 
            }
1966f5db Andy McCall ↶ 467 
            base.graph_key(&e.node)
992467d3 Andy McCall ↶ 468 
        };
cc7a0056 Andy McCall ↶ 469 
        let mut fs_entries = {
470 
            let _t = crate::timing::phase("snapshot: read working copy");
992467d3 Andy McCall ↶ 471 
            self.walk_working_copy_with(manifest.as_deref(), &tracked, Some(&key_of))?
cc7a0056 Andy McCall ↶ 472 
        };
992467d3 Andy McCall ↶ 473 
        // The keys of the files read, before building graphs drops them: a
474 
        // file found equal to its base rendering keeps its key.
1966f5db Andy McCall ↶ 475 
        let keys: HashMap<String, [u8; 32]> = fs_entries
992467d3 Andy McCall ↶ 476 
            .iter()
477 
            .filter(|(_, e)| !e.unchanged && e.kind == NodeKind::File)
478 
            .filter_map(|(p, _)| Some((p.clone(), key_of(p)?)))
479 
            .collect();
0bdc3f91 Andy McCall ↶ 480 
        let base_snapshot = {
cc7a0056 Andy McCall ↶ 481 
            let _t = crate::timing::phase("snapshot: load and render base");
482 
            if partial {
d616faf5 Andy McCall ↶ 483 
                let sets = crate::timing::phase("snapshot:   of which choosing graphs");
484 
                // Every file but those on disk known unchanged.
485 
                let need: Vec<(&str, NodeId)> = join(&view.entries, &fs_entries)
486 
                    .filter_map(|(p, v, f)| {
487 
                        let v = v?;
488 
                        (v.kind == NodeKind::File && !f.is_some_and(|f| f.unchanged))
489 
                            .then_some((p.as_str(), v.node))
490 
                    })
cc7a0056 Andy McCall ↶ 491 
                    .collect();
d616faf5 Andy McCall ↶ 492 
                let render: std::collections::HashSet<&str> =
493 
                    need.iter().map(|(p, _)| *p).collect();
494 
                let need: Vec<NodeId> = need.iter().map(|(_, n)| *n).collect();
495 
                drop(sets);
d0990c2b Andy McCall ↶ 496 
                {
497 
                    let _t = crate::timing::phase("snapshot:   of which building graphs");
498 
                    self.load_graphs_of(&mut base, &need)?;
499 
                }
500 
                let _t = crate::timing::phase("snapshot:   of which rendering");
d616faf5 Andy McCall ↶ 501 
                base.materialize_only(&render)
cc7a0056 Andy McCall ↶ 502 
            } else {
503 
                self.load_all_graphs(&mut base)?;
504 
                base.materialize()
505 
            }
0bdc3f91 Andy McCall ↶ 506 
        };
614541c3 Andy McCall ↶ 507
508 
        // Reuse the NodeIds the previous working-copy version assigned to new
509 
        // paths, so an unchanged working copy hashes identically (spec §12.3).
510 
        let reuse = added_node_paths(&old_wc, &base_snapshot);
c863c7a7 Andy McCall ↶ 511 
        self.platform_normalize(&mut fs_entries, &base_snapshot);
0bdc3f91 Andy McCall ↶ 512 
        let built = {
513 
            let _t = crate::timing::phase("snapshot: diff against base");
23ff1288 Andy McCall ↶ 514 
            if partial && self.config.sparse_paths.is_empty() {
515 
                #[cfg(test)]
516 
                let full =
517 
                    self.build_from_entries(&base, &base_snapshot, reuse.clone(), &fs_entries)?;
518 
                let built = self.build_changed_only(&base, &base_snapshot, reuse, &fs_entries)?;
519 
                #[cfg(test)]
520 
                {
521 
                    // New nodes take random identifiers: compare them by
522 
                    // order of appearance.
523 
                    let norm = |b: &BuiltChange| {
524 
                        let text = format!("{:?} {:?} {:?}", b.tree_ops, b.edits, b.statuses);
525 
                        let mut seen: Vec<String> = Vec::new();
526 
                        let mut out = String::new();
527 
                        let mut rest = text.as_str();
528 
                        while let Some(i) = rest.find("NodeId(") {
529 
                            out.push_str(&rest[..i]);
530 
                            let end = i + rest[i..].find(')').unwrap() + 1;
531 
                            let id = rest[i..end].to_string();
532 
                            let n = seen.iter().position(|x| *x == id).unwrap_or_else(|| {
533 
                                seen.push(id);
534 
                                seen.len() - 1
535 
                            });
536 
                            out.push_str(&format!("N{n}"));
537 
                            rest = &rest[end..];
538 
                        }
539 
                        out.push_str(rest);
540 
                        out
541 
                    };
542 
                    assert_eq!(
543 
                        norm(&built),
544 
                        norm(&full),
545 
                        "a changed-only build is the full one"
546 
                    );
547 
                }
548 
                built
549 
            } else {
550 
                self.build_from_entries(&base, &base_snapshot, reuse, &fs_entries)?
551 
            }
0bdc3f91 Andy McCall ↶ 552 
        };
cc7a0056 Andy McCall ↶ 553 
        // Text files found equal to an unconflicted base rendering keep its
554 
        // graph key, so the next snapshot need not read or render them.
555 
        {
556 
            let changed: HashSet<&str> = built.statuses.iter().map(|s| s.path.as_str()).collect();
557 
            for (p, e) in fs_entries.iter_mut() {
558 
                if e.unchanged || e.kind != NodeKind::File || changed.contains(p.as_str()) {
559 
                    continue;
560 
                }
561 
                e.graph_key = match base_snapshot.files.get(p).map(|f| &f.content) {
562 
                    Some(FileContent::Text {
563 
                        bytes,
564 
                        materialized,
40a40bb7 Andy McCall ↶ 565 
                    }) if !materialized.is_conflicted() && *bytes == *e.content => {
1966f5db Andy McCall ↶ 566 
                        keys.get(p).copied()
cc7a0056 Andy McCall ↶ 567 
                    }
568 
                    _ => None,
569 
                };
570 
            }
571 
        }
614541c3 Andy McCall ↶ 572
98286406 Andy McCall ↶ 573 
        let mut wc = Change {
6f8aeb99 Andy McCall ↶ 574 
            version: crate::objects::CHANGE_VERSION,
98286406 Andy McCall ↶ 575 
            change_id: old_wc.change_id,
576 
            deps: vec![],
577 
            author: self.author(),
578 
            hlc: old_wc.hlc,
579 
            message: old_wc.message.clone(),
614541c3 Andy McCall ↶ 580 
            tree_ops: built.tree_ops,
581 
            edits: built.edits,
98286406 Andy McCall ↶ 582 
            kind: ChangeKind::WorkingCopy,
583 
        };
584 
        let base_heads = base.maximal(&old_wc.deps);
585 
        let mut deps: BTreeSet<ChangeHash> = base_heads.into_iter().collect();
586 
        deps.extend(referenced_changes(&wc, &base));
614541c3 Andy McCall ↶ 587 
        deps.extend(built.contributors.iter().copied());
98286406 Andy McCall ↶ 588 
        wc.deps = deps.into_iter().collect();
589 
        wc.normalize();
614541c3 Andy McCall ↶ 590 
        let resolves = built.resolves;
591 
        let contributors = built.contributors;
592 
        let statuses = built.statuses;
98286406 Andy McCall ↶ 593 
        // Empty snapshot: hash the candidate before storing; only record an op on change.
8b276e01 Andy McCall ↶ 594 
        let record = crate::timing::phase("snapshot: hash and record");
17f94b13 Andy McCall ↶ 595 
        // Encoded once, its texts stored a batch at a time as they are
596 
        // made (a stored blob is not stored again); the change itself only
597 
        // when it changed.
5b4c5130 Andy McCall ↶ 598 
        // For a commit the change is neither encoded, stored nor recorded:
599 
        // the commit stores its own, built from this one's operations, and
600 
        // its texts with it. The hash it would have is not needed.
601 
        let (new_hash, changed, encoded) = if for_commit {
602 
            (old_hash, true, Vec::new())
603 
        } else {
604 
            let encoded = wc.encode_streaming(|batch| self.store.put_texts(&batch))?;
605 
            let new_hash = crate::hash::hash_object(crate::hash::ObjectType::Change, &encoded);
606 
            (new_hash, new_hash != old_hash, encoded)
607 
        };
cd4eb55e Andy McCall ↶ 608 
        if changed && !for_commit {
17f94b13 Andy McCall ↶ 609 
            self.store.put(crate::hash::ObjectType::Change, &encoded)?;
98286406 Andy McCall ↶ 610 
            let mut view = op.view.clone();
611 
            view.workspaces.insert(self.workspace, new_hash);
612 
            self.record_op(&[op_ref], &format!("{command}{SNAPSHOT_SUFFIX}"), view)?;
613 
        }
8b276e01 Andy McCall ↶ 614 
        drop(record);
135212de Andy McCall ↶ 615 
        // The prompt compares against this without recording anything.
0bdc3f91 Andy McCall ↶ 616 
        let _t = crate::timing::phase("snapshot: write manifest");
992467d3 Andy McCall ↶ 617 
        let conflicted = base_snapshot.is_conflicted();
618 
        let uncommitted = !wc.tree_ops.is_empty() || !wc.edits.is_empty();
619 
        // Rewriting the manifest pays only when it would teach the next
620 
        // snapshot something: other flags, a file found equal to the base
621 
        // (it need not be read again), a new blob, a path added or gone.
622 
        // An edited file is read next time whatever it says.
623 
        let worth = match self.read_manifest()? {
cd4eb55e Andy McCall ↶ 624 
            _ if for_commit => true,
992467d3 Andy McCall ↶ 625 
            None => true,
626 
            Some((c, u, m)) => {
627 
                c != conflicted
628 
                    || u != uncommitted
629 
                    || m.entries.len() != fs_entries.len()
630 
                    || fs_entries.iter().any(|(p, e)| {
631 
                        !e.unchanged && (e.graph_key.is_some() || !m.entries.contains_key(p))
632 
                    })
633 
                    || built
634 
                        .blobs
635 
                        .iter()
636 
                        .any(|(p, b)| m.entries.get(p).and_then(|e| e.blob) != Some(*b))
36041796 Andy McCall ↶ 637 
                    // With uncommitted work, the status cache below holds
638 
                    // only while the manifest describes the edited files as
639 
                    // they are now.
640 
                    || (uncommitted && statuses.iter().any(|st| {
641 
                        let Some(e) = fs_entries.get(&st.path) else {
642 
                            return m.entries.contains_key(&st.path);
643 
                        };
644 
                        let Some(me) = m.entries.get(&st.path) else {
645 
                            return true;
646 
                        };
647 
                        let hash = e
648 
                            .content_hash
649 
                            .unwrap_or_else(|| *blake3::hash(&e.content).as_bytes());
650 
                        let stat = e.stat.or_else(|| {
651 
                            fs::symlink_metadata(self.root.join(&st.path))
652 
                                .ok()
653 
                                .map(|md| (md.len(), super::prompt::mtime_ns(&md)))
654 
                        });
e226a6cc Andy McCall ↶ 655 
                        // The executable bit too: a mode change moves
656 
                        // neither size nor time.
657 
                        me.hash != hash || me.exec != e.exec || stat != Some((me.size, me.mtime_ns))
36041796 Andy McCall ↶ 658 
                    }))
992467d3 Andy McCall ↶ 659 
            }
660 
        };
f9466021 Andy McCall ↶ 661 
        let attributes = crate::attributes::Attributes::from_files(
662 
            fs_entries
663 
                .iter()
664 
                .filter(|(p, _)| is_attributes(p))
665 
                .map(|(p, e)| (p.as_str(), e.content.as_slice())),
666 
        );
236f35c0 Andy McCall ↶ 667 
        let mut manifest = None;
668 
        if for_commit {
669 
            manifest = Some(PendingManifest {
670 
                entries: fs_entries,
671 
                blobs: built.blobs.clone(),
672 
                dirs: self.walked_dirs.take(),
673 
                written: super::prompt::now_ns(),
674 
            });
675 
        } else {
676 
            if worth {
677 
                self.write_manifest_entries(conflicted, uncommitted, &fs_entries, &built.blobs)?;
678 
            }
679 
            self.write_status_cache((uncommitted && !conflicted).then_some(&new_hash), &statuses);
680 
        }
75167672 Andy McCall 681 
        Ok(SnapshotResult {
682 
            wc_hash: new_hash,
683 
            wc,
684 
            base,
685 
            base_snapshot,
686 
            statuses,
687 
            changed,
688 
            resolves,
689 
            contributors,
f9466021 Andy McCall ↶ 690 
            attributes,
d4dbab97 Andy McCall ↶ 691 
            untracked_nested: self.untracked_nested.borrow().clone(),
236f35c0 Andy McCall ↶ 692 
            manifest,
75167672 Andy McCall 693 
        })
98286406 Andy McCall ↶ 694 
    }
695
50a046e4 Andy McCall ↶ 696 
    /// `oxy status --json` (spec §9.5): the changed paths and every conflict
697 
    /// of the working copy's base with its kind, path, and contributing
698 
    /// ChangeIds.
699 
    pub fn status_json(&self, snap: &SnapshotResult) -> Result<String> {
700 
        let heads = self.bookmark_heads()?;
701 
        let base_snapshot = &snap.base_snapshot;
702 
        let by_node = base_snapshot.tree.by_node();
703 
        let path_of = |n: &NodeId| by_node.get(n).map(|e| e.path.clone());
75167672 Andy McCall 704 
        let id_of = |h: &ChangeHash| {
705 
            snap.base
706 
                .get(h)
707 
                .map(|m| m.change_id.to_hex())
708 
                .unwrap_or_else(|| h.to_hex())
709 
        };
710 
        let paths = json_array(snap.statuses.iter().map(|s| {
711 
            format!(
712 
                "{{\"path\":{},\"status\":{}}}",
713 
                json_str(&s.path),
714 
                json_str(&format!("{:?}", s.status))
715 
            )
716 
        }));
50a046e4 Andy McCall ↶ 717 
        let mut conflicts: Vec<String> = Vec::new();
718 
        for (p, f) in &base_snapshot.files {
719 
            if !f.is_conflicted() {
720 
                continue;
721 
            }
722 
            let (kind, changes) = match &f.content {
75167672 Andy McCall 723 
                FileContent::Text { .. } => {
724 
                    ("text", snap.base.conflict_regions(&f.node).contributors)
725 
                }
726 
                FileContent::Opaque { blobs } => {
727 
                    ("replace", snap.base.blob_contributors(&f.node, blobs))
728 
                }
50a046e4 Andy McCall ↶ 729 
                FileContent::Dir => continue,
730 
            };
75167672 Andy McCall 731 
            let resolved = match snap
732 
                .statuses
733 
                .iter()
734 
                .find(|s| s.path == *p)
735 
                .map(|s| s.status)
736 
            {
50a046e4 Andy McCall ↶ 737 
                Some(ChangeStatus::Resolved) => "true",
738 
                Some(ChangeStatus::ConflictEdited) => "\"markers\"",
739 
                _ => "false",
740 
            };
741 
            conflicts.push(format!(
742 
                "{{\"kind\":{},\"path\":{},\"node\":{},\"changes\":{},\"resolved\":{resolved}}}",
743 
                json_str(kind),
744 
                json_str(p),
745 
                json_str(&f.node.to_hex()),
746 
                json_array(changes.iter().map(|h| json_str(&id_of(h))))
747 
            ));
748 
        }
749 
        for c in &base_snapshot.tree.conflicts {
750 
            let (kind, nodes): (&str, Vec<NodeId>) = match c {
751 
                crate::tree::TreeConflict::Name { nodes, .. } => ("name", nodes.clone()),
75167672 Andy McCall 752 
                crate::tree::TreeConflict::MultiPlacement { node, .. } => {
753 
                    ("multi-placement", vec![*node])
754 
                }
50a046e4 Andy McCall ↶ 755 
                crate::tree::TreeConflict::Orphan { node, .. } => ("orphan", vec![*node]),
756 
                crate::tree::TreeConflict::Cycle { node } => ("cycle", vec![*node]),
757 
                crate::tree::TreeConflict::Mode { node, .. } => ("mode", vec![*node]),
758 
            };
759 
            let detail = match c {
760 
                crate::tree::TreeConflict::Name { name, parent, .. } => {
75167672 Andy McCall 761 
                    let dir = if *parent == NodeId::ROOT {
762 
                        String::new()
763 
                    } else {
764 
                        path_of(parent).unwrap_or_default()
765 
                    };
766 
                    format!(
767 
                        ",\"path\":{}",
768 
                        json_str(&if dir.is_empty() {
769 
                            name.clone()
770 
                        } else {
771 
                            format!("{dir}/{name}")
772 
                        })
773 
                    )
50a046e4 Andy McCall ↶ 774 
                }
75167672 Andy McCall 775 
                _ => nodes
776 
                    .first()
777 
                    .and_then(path_of)
778 
                    .map(|p| format!(",\"path\":{}", json_str(&p)))
779 
                    .unwrap_or_default(),
50a046e4 Andy McCall ↶ 780 
            };
781 
            conflicts.push(format!(
782 
                "{{\"kind\":{},\"nodes\":{}{detail}}}",
783 
                json_str(kind),
784 
                json_array(nodes.iter().map(|n| json_str(&n.to_hex())))
785 
            ));
786 
        }
787 
        Ok(format!(
b689779e Andy McCall ↶ 788 
            "{{\"bookmark\":{},\"working_copy\":{},\"heads\":{},\"clean\":{},\"conflicted\":{},\"paths\":{paths},\"conflicts\":{},\"untracked_nested\":{},\"skipped_large\":{}}}",
50a046e4 Andy McCall ↶ 789 
            json_str(&self.config.bookmark),
790 
            json_str(&snap.wc.change_id.to_hex()),
791 
            heads.len(),
792 
            snap.statuses.is_empty(),
793 
            base_snapshot.is_conflicted(),
d4dbab97 Andy McCall ↶ 794 
            json_array(conflicts),
b689779e Andy McCall ↶ 795 
            json_array(snap.untracked_nested.iter().map(|p| json_str(p))),
796 
            json_array(
797 
                self.skipped_large
798 
                    .borrow()
799 
                    .iter()
800 
                    .map(|(p, n)| format!("{{\"path\":{},\"size\":{n}}}", json_str(p)))
801 
            )
50a046e4 Andy McCall ↶ 802 
        ))
803 
    }
804
98286406 Andy McCall ↶ 805 
    /// Turn the working-copy change into a normal change on the bookmark (spec §12.3).
806 
    pub fn commit(&self, message: &str, extra_deps: &[ChangeHash]) -> Result<ChangeHash> {
957fee30 Andy McCall ↶ 807 
        self.commit_selection(message, extra_deps, &Selection::All)
808 
    }
809
810 
    /// The hunks of the working copy, for `oxy commit --interactive`: each
811 
    /// tree operation and each edit, in the order `select_hunks` numbers them.
812 
    pub fn hunks(&self) -> Result<(SnapshotResult, Vec<Hunk>)> {
8ab265dc Andy McCall ↶ 813 
        let snap = self.snapshot_status("oxy commit")?;
957fee30 Andy McCall ↶ 814 
        let hunks = render_hunks(&snap.wc, &snap.base, &snap.base_snapshot);
815 
        Ok((snap, hunks))
816 
    }
817
818 
    /// Commit part of the working copy (spec §12.3). What is not selected
819 
    /// stays on disk as uncommitted work: the next snapshot rebuilds it
820 
    /// against the new base, so the working copy always describes the
821 
    /// directory.
f9466021 Andy McCall ↶ 822 
    /// The graph key in `st` of each text file at `paths`, the paths a
823 
    /// commit changed: only their graphs changed, so the keys the manifest
824 
    /// holds for the others still stand (and any that does not is simply
825 
    /// not trusted by the next snapshot, which compares keys). Files with
826 
    /// an `eol` attribute are left out: their bytes on disk are not the
827 
    /// rendering.
828 
    fn committed_graph_keys(
829 
        &self,
830 
        st: &State,
831 
        paths: &[String],
832 
        attrs: &crate::attributes::Attributes,
833 
    ) -> HashMap<String, [u8; 32]> {
86e1824f Andy McCall ↶ 834 
        let _t = crate::timing::phase("commit: graph keys");
f9466021 Andy McCall ↶ 835 
        let view = st.tree().view();
836 
        paths
86e1824f Andy McCall ↶ 837 
            .iter()
f9466021 Andy McCall ↶ 838 
            .filter_map(|p| {
839 
                let e = view.entries.get(p)?;
840 
                (e.kind == NodeKind::File && !e.mode.opaque && attrs.lookup(p).eol.is_none())
841 
                    .then_some(())?;
842 
                Some((p.clone(), st.graph_key(&e.node)?))
86e1824f Andy McCall ↶ 843 
            })
8457a6cc Andy McCall ↶ 844 
            .collect()
86e1824f Andy McCall ↶ 845 
    }
846
75167672 Andy McCall 847 
    pub fn commit_selection(
848 
        &self,
849 
        message: &str,
850 
        extra_deps: &[ChangeHash],
851 
        selection: &Selection,
852 
    ) -> Result<ChangeHash> {
75051941 Andy McCall ↶ 853 
        Ok(self
854 
            .commit_selection_partial(message, extra_deps, selection)?
855 
            .0)
856 
    }
857
858 
    /// As [`Repo::commit_selection`], and whether the commit left part of
859 
    /// the working copy uncommitted: when it did not, nothing is left, and
860 
    /// no snapshot is needed to say so.
861 
    pub fn commit_selection_partial(
862 
        &self,
863 
        message: &str,
864 
        extra_deps: &[ChangeHash],
865 
        selection: &Selection,
866 
    ) -> Result<(ChangeHash, bool)> {
b30a880f Andy McCall ↶ 867 
        // The commit's objects are written together, made durable at once
868 
        // before the operation that names them (spec §13.6).
869 
        let batch = self.store.begin_batch()?;
870 
        let out = self.commit_selection_batched(message, extra_deps, selection);
871 
        batch.finish()?;
872 
        out
873 
    }
874
236f35c0 Andy McCall ↶ 875 
    /// Write the manifest a commit's snapshot left to the commit, with the
876 
    /// graph keys `keys` sets.
877 
    fn write_pending_manifest(
878 
        &self,
879 
        mut m: PendingManifest,
880 
        conflicted: bool,
881 
        keys: HashMap<String, [u8; 32]>,
882 
    ) -> Result<()> {
883 
        for (path, k) in keys {
884 
            if let Some(e) = m.entries.get_mut(&path) {
885 
                e.graph_key = Some(k);
886 
            }
887 
        }
888 
        self.write_manifest_entries_as_of(
889 
            conflicted,
890 
            false,
891 
            &m.entries,
892 
            &m.blobs,
893 
            m.written,
894 
            m.dirs.unwrap_or_default(),
895 
        )
896 
    }
897
b30a880f Andy McCall ↶ 898 
    fn commit_selection_batched(
899 
        &self,
900 
        message: &str,
901 
        extra_deps: &[ChangeHash],
902 
        selection: &Selection,
75051941 Andy McCall ↶ 903 
    ) -> Result<(ChangeHash, bool)> {
6caacfff Andy McCall ↶ 904 
        let _wc = self.lock_working_copy()?;
e226a6cc Andy McCall ↶ 905 
        let mut snap = match self.snapshot_from_cache()? {
906 
            Some(snap) => snap,
907 
            None => self.snapshot_as("oxy commit", true, true)?,
908 
        };
236f35c0 Andy McCall ↶ 909 
        let pending = snap.manifest.take();
98286406 Andy McCall ↶ 910 
        if snap.wc.tree_ops.is_empty() && snap.wc.edits.is_empty() {
236f35c0 Andy McCall ↶ 911 
            if let Some(m) = pending {
912 
                let conflicted = snap.base_snapshot.is_conflicted();
913 
                self.write_pending_manifest(m, conflicted, HashMap::new())?;
914 
            }
98286406 Andy McCall ↶ 915 
            return fail("nothing to commit: the working copy matches its base");
916 
        }
75167672 Andy McCall 917 
        if let Some(s) = snap
918 
            .statuses
919 
            .iter()
920 
            .find(|s| s.status == ChangeStatus::ConflictEdited)
921 
        {
98286406 Andy McCall ↶ 922 
            return fail(format!("{} still contains conflict markers", s.path));
923 
        }
924 
        if message.trim().is_empty() {
925 
            return fail("a commit message is required");
926 
        }
192fad1e Andy McCall ↶ 927 
        // What the whole working copy references, before a whole commit
928 
        // takes its operations over (they are moved, not copied: for a
929 
        // large tree they are most of the memory).
930 
        let wc_refs = referenced_changes(&snap.wc, &snap.base);
f9466021 Andy McCall ↶ 931 
        let changed_paths: Vec<String> = snap.statuses.iter().map(|s| s.path.clone()).collect();
957fee30 Andy McCall ↶ 932 
        let (tree_ops, edits, partial) = match selection {
192fad1e Andy McCall ↶ 933 
            Selection::All => (
934 
                std::mem::take(&mut snap.wc.tree_ops),
935 
                std::mem::take(&mut snap.wc.edits),
936 
                false,
937 
            ),
957fee30 Andy McCall ↶ 938 
            Selection::Paths(paths) => {
939 
                let (o, e) = select_paths(&snap.wc, &snap.base_snapshot, paths);
940 
                let partial = o.len() != snap.wc.tree_ops.len() || e.len() != snap.wc.edits.len();
941 
                (o, e, partial)
942 
            }
943 
            Selection::Hunks(indices) => {
944 
                let (o, e) = select_hunks(&snap.wc, indices);
945 
                let partial = o.len() != snap.wc.tree_ops.len() || e.len() != snap.wc.edits.len();
946 
                (o, e, partial)
947 
            }
948 
        };
949 
        if tree_ops.is_empty() && edits.is_empty() {
950 
            return fail("nothing selected to commit");
951 
        }
952 
        if partial && snap.resolves {
953 
            return fail("a conflict resolution must be committed whole");
954 
        }
6d3eefbc Andy McCall ↶ 955 
        // pre-commit hook: may veto.
956 
        self.run_hook(
957 
            "pre-commit",
958 
            &format!(
959 
                "{{\"event\":\"pre-commit\",\"message\":{},\"paths\":{}}}",
960 
                json_str(message),
75167672 Andy McCall 961 
                json_array(snap.statuses.iter().map(|s| format!(
962 
                    "{{\"path\":{},\"status\":{}}}",
963 
                    json_str(&s.path),
964 
                    json_str(&format!("{:?}", s.status))
965 
                )))
6d3eefbc Andy McCall ↶ 966 
            ),
967 
        )?;
98286406 Andy McCall ↶ 968 
        let (op_ref, op) = self.head_op()?;
192fad1e Andy McCall ↶ 969 
        let mut deps: BTreeSet<ChangeHash> = wc_refs;
6d3eefbc Andy McCall ↶ 970 
        // semantic-deps hook: may add explicit dependencies (spec §8.1).
971 
        if self.config.semantic_deps {
972 
            let touched: Vec<String> = snap.statuses.iter().map(|s| s.path.clone()).collect();
973 
            let by_node = snap.base_snapshot.tree.by_node();
974 
            let mut candidates: Vec<(ChangeId, ChangeHash, Vec<String>)> = Vec::new();
975 
            let mut seen: BTreeSet<ChangeHash> = BTreeSet::new();
976 
            for h in deps.iter().chain(snap.base.maximal(&snap.wc.deps).iter()) {
977 
                if !seen.insert(*h) {
978 
                    continue;
979 
                }
980 
                let c = self.store.get_change(h)?;
981 
                let mut paths: BTreeSet<String> = BTreeSet::new();
75167672 Andy McCall 982 
                for n in c
983 
                    .tree_ops
984 
                    .iter()
985 
                    .map(|o| o.node())
986 
                    .chain(c.edits.iter().map(|e| e.node()))
987 
                {
6d3eefbc Andy McCall ↶ 988 
                    if let Some(e) = by_node.get(&n) {
989 
                        paths.insert(e.path.clone());
990 
                    }
991 
                }
992 
                candidates.push((c.change_id, *h, paths.into_iter().collect()));
993 
            }
994 
            for d in self.semantic_deps(&snap.wc.change_id, &touched, &candidates)? {
995 
                if snap.base.contains(&d) {
996 
                    deps.insert(d);
997 
                }
998 
            }
999 
        }
98286406 Andy McCall ↶ 1000 
        // A resolution depends on every change that contributed to the
1001 
        // conflict (spec §8.1). Tree and opaque conflicts are resolved through
1002 
        // ordinary operations whose references already cover the contributors.
1003 
        deps.extend(snap.contributors.iter().copied());
1004 
        for d in extra_deps {
1005 
            if !snap.base.contains(d) {
75167672 Andy McCall 1006 
                return fail(format!(
1007 
                    "--depends-on {d}: not in the working copy's base state"
1008 
                ));
98286406 Andy McCall ↶ 1009 
            }
1010 
            deps.insert(*d);
1011 
        }
1012 
        let mut change = Change {
6f8aeb99 Andy McCall ↶ 1013 
            version: crate::objects::CHANGE_VERSION,
75167672 Andy McCall 1014 
            change_id: if partial {
1015 
                ChangeId::random()
1016 
            } else {
1017 
                snap.wc.change_id
1018 
            },
957fee30 Andy McCall ↶ 1019 
            deps: vec![],
98286406 Andy McCall ↶ 1020 
            author: self.author(),
1021 
            hlc: Hlc::now(),
1022 
            message: message.trim_end().to_string(),
957fee30 Andy McCall ↶ 1023 
            tree_ops,
1024 
            edits,
75167672 Andy McCall 1025 
            kind: if snap.resolves {
1026 
                ChangeKind::Resolution
1027 
            } else {
1028 
                ChangeKind::Normal
1029 
            },
98286406 Andy McCall ↶ 1030 
        };
957fee30 Andy McCall ↶ 1031 
        if partial {
1032 
            // Dependencies come from what the selected part references, not
1033 
            // from the whole working copy.
1034 
            deps = referenced_changes(&change, &snap.base);
1035 
            for d in extra_deps {
1036 
                deps.insert(*d);
1037 
            }
1038 
        }
1039 
        change.deps = deps.into_iter().collect();
98286406 Andy McCall ↶ 1040 
        change.normalize();
f9466021 Andy McCall ↶ 1041 
        let signing = crate::timing::phase("commit: sign and store");
614541c3 Andy McCall ↶ 1042 
        let hash = self.store_signed(&change)?;
f9466021 Andy McCall ↶ 1043 
        drop(signing);
1044 
        let extending = crate::timing::phase("commit: extend the state");
98286406 Andy McCall ↶ 1045
1046 
        // New bookmark heads: maximal elements of old heads plus the new change.
1047 
        // The working copy continues on its own base plus the new change, which
1048 
        // is exactly what the files on disk contain; the bookmark may hold
1049 
        // other heads (and conflicts) the working copy has not checked out.
1050 
        let old_heads = self.heads_of(&op.view, &self.config.bookmark)?;
1051 
        let old_base = snap.base.maximal(&snap.wc.deps);
1052 
        let mut all: Vec<ChangeHash> = old_heads.clone();
1053 
        all.extend(old_base.iter().copied());
1054 
        all.push(hash);
07eb99be Andy McCall ↶ 1055 
        // The snapshot already loaded the working copy's base; when that also
1056 
        // holds every bookmark head, it is the state to extend.
1057 
        let mut st = if old_heads.iter().all(|h| snap.base.contains(h)) {
1058 
            snap.base
1059 
        } else {
1060 
            self.load_state(&all)?
1061 
        };
192fad1e Andy McCall ↶ 1062 
        st.add_hashed(hash, change)?;
98286406 Andy McCall ↶ 1063 
        let mut with_new = old_heads.clone();
1064 
        with_new.push(hash);
1065 
        let new_heads = st.maximal(&with_new);
252e8cb2 Andy McCall ↶ 1066 
        // Automatic checkpoint once enough changes accumulated (spec §11).
75167672 Andy McCall 1067 
        let since = self
1068 
            .latest_checkpoint(&self.config.bookmark)?
1069 
            .map(|(_, cp)| cp.count)
1070 
            .unwrap_or(0);
1071 
        if self.config.checkpoint_interval > 0
1072 
            && (st.len() as u64).saturating_sub(since) >= self.config.checkpoint_interval
1073 
        {
252e8cb2 Andy McCall ↶ 1074 
            let cp_state = self.load_state_full(&new_heads)?;
1075 
            self.write_checkpoint(&self.config.bookmark, &cp_state, &new_heads)?;
1076 
        }
98286406 Andy McCall ↶ 1077 
        let mut wc_base = old_base.clone();
1078 
        wc_base.push(hash);
1079 
        let wc_base = st.maximal(&wc_base);
f9466021 Andy McCall ↶ 1080 
        drop(extending);
98286406 Andy McCall ↶ 1081
8457a6cc Andy McCall ↶ 1082 
        // The graph keys of the files as they now are, from the state the
1083 
        // commit just extended: a file whose graph it built has none, and
1084 
        // is read by the next snapshot; a key the next base does not share
1085 
        // (the bookmark held other heads) is simply not trusted then.
236f35c0 Andy McCall ↶ 1086 
        let keying = crate::timing::phase("commit: graph keys of the files");
f9466021 Andy McCall ↶ 1087 
        let committed_keys =
1088 
            (!partial).then(|| self.committed_graph_keys(&st, &changed_paths, &snap.attributes));
236f35c0 Andy McCall ↶ 1089 
        drop(keying);
f9466021 Andy McCall ↶ 1090 
        let recording = crate::timing::phase("commit: record");
98286406 Andy McCall ↶ 1091 
        let wc = self.new_working_copy(wc_base, "");
1092 
        let wc_hash = self.store.put_change(&wc)?;
1093 
        let mut view = op.view.clone();
75167672 Andy McCall 1094 
        view.bookmarks.insert(
1095 
            self.config.bookmark.clone(),
1096 
            self.store.put_state(&StateObject::new(new_heads))?,
1097 
        );
98286406 Andy McCall ↶ 1098 
        view.workspaces.insert(self.workspace, wc_hash);
1099 
        self.record_op(&[op_ref], &format!("oxy commit -m {message:?}"), view)?;
f9466021 Andy McCall ↶ 1100 
        drop(recording);
236f35c0 Andy McCall ↶ 1101 
        let rewriting = crate::timing::phase("commit: manifest");
957fee30 Andy McCall ↶ 1102 
        if partial {
1103 
            // The rest is rebuilt from disk by the next snapshot.
1104 
            self.drop_manifest();
1105 
        } else {
86e1824f Andy McCall ↶ 1106 
            let conflicted = snap.base_snapshot.is_conflicted() && !snap.resolves;
1107 
            // The files on disk are now exactly the new base's renderings:
1108 
            // record their graph keys, so the next snapshot neither reads nor
1109 
            // renders them (oxygen#35).
1110 
            let keys = if conflicted {
1111 
                HashMap::new()
1112 
            } else {
8457a6cc Andy McCall ↶ 1113 
                committed_keys.unwrap_or_default()
86e1824f Andy McCall ↶ 1114 
            };
236f35c0 Andy McCall ↶ 1115 
            match pending {
1116 
                Some(m) => self.write_pending_manifest(m, conflicted, keys)?,
1117 
                None => self.rewrite_manifest(conflicted, false, &keys)?,
1118 
            }
957fee30 Andy McCall ↶ 1119 
        }
236f35c0 Andy McCall ↶ 1120 
        drop(rewriting);
1121 
        let packing = crate::timing::phase("commit: autopack");
d02e545d Andy McCall ↶ 1122 
        self.maybe_autopack()?;
236f35c0 Andy McCall ↶ 1123 
        drop(packing);
1124 
        let _hook = crate::timing::phase("commit: hook");
75167672 Andy McCall 1125 
        let _ = self.run_hook(
1126 
            "post-commit",
1127 
            &format!(
1128 
                "{{\"event\":\"post-commit\",\"change_id\":{},\"hash\":{},\"message\":{}}}",
1129 
                json_str(&snap.wc.change_id.to_hex()),
1130 
                json_str(&hash.to_hex()),
1131 
                json_str(message)
1132 
            ),
1133 
        );
75051941 Andy McCall ↶ 1134 
        Ok((hash, partial))
98286406 Andy McCall ↶ 1135 
    }
1136 
}
1137
957fee30 Andy McCall ↶ 1138 
/// What part of the working copy a commit takes (spec §12.3).
1139 
#[derive(Clone, Debug)]
1140 
pub enum Selection {
1141 
    All,
1142 
    /// Every operation and edit on nodes under these paths.
1143 
    Paths(Vec<String>),
1144 
    /// Hunk indices as `Repo::hunks` numbers them; dependent hunks are added.
1145 
    Hunks(Vec<usize>),
1146 
}
1147
1148 
/// One selectable piece of the working copy: a tree operation or an edit.
1149 
#[derive(Clone, Debug)]
1150 
pub struct Hunk {
1151 
    pub index: usize,
1152 
    pub path: String,
1153 
    /// Rendered like `oxy diff`.
1154 
    pub text: String,
1155 
}
1156
1157 
/// The tree operations and edits on nodes under `paths`, plus the
1158 
/// additions of any new parent directories they need.
75167672 Andy McCall 1159 
fn select_paths(
1160 
    wc: &Change,
1161 
    base_snapshot: &Snapshot,
1162 
    paths: &[String],
1163 
) -> (Vec<TreeOp>, Vec<Edit>) {
1164 
    let selected = |p: &str| {
1165 
        paths
1166 
            .iter()
1167 
            .any(|x| p == x || p.starts_with(&format!("{}/", x.trim_end_matches('/'))))
1168 
    };
957fee30 Andy McCall ↶ 1169 
    let by_node = base_snapshot.tree.by_node();
1170 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
1171 
    for op in &wc.tree_ops {
75167672 Andy McCall 1172 
        if let TreeOp::AddNode {
1173 
            node,
1174 
            parent,
1175 
            name,
1176 
            kind,
1177 
            ..
1178 
        } = op
1179 
        {
957fee30 Andy McCall ↶ 1180 
            new_names.insert(*node, (*parent, name.clone(), *kind));
1181 
        }
1182 
    }
1183 
    let mut nodes: BTreeSet<NodeId> = BTreeSet::new();
75167672 Andy McCall 1184 
    for n in wc
1185 
        .tree_ops
1186 
        .iter()
1187 
        .map(|o| o.node())
1188 
        .chain(wc.edits.iter().map(|e| e.node()))
1189 
    {
957fee30 Andy McCall ↶ 1190 
        if path_of_node(&n, &by_node, &new_names).is_some_and(|p| selected(&p)) {
1191 
            nodes.insert(n);
1192 
        }
1193 
    }
1194 
    // New parents of selected new nodes come along.
1195 
    let mut changed = true;
1196 
    while changed {
1197 
        changed = false;
1198 
        for (n, (parent, _, _)) in &new_names {
1199 
            if nodes.contains(n) && new_names.contains_key(parent) && nodes.insert(*parent) {
1200 
                changed = true;
1201 
            }
1202 
        }
1203 
    }
75167672 Andy McCall 1204 
    let ops = wc
1205 
        .tree_ops
1206 
        .iter()
1207 
        .filter(|o| nodes.contains(&o.node()))
1208 
        .cloned()
1209 
        .collect();
1210 
    let edits = wc
1211 
        .edits
1212 
        .iter()
1213 
        .filter(|e| nodes.contains(&e.node()))
1214 
        .cloned()
1215 
        .collect();
957fee30 Andy McCall ↶ 1216 
    (ops, edits)
1217 
}
1218
1219 
/// The hunks `indices` name plus whatever they refer to within the working
1220 
/// copy (the parent directory of an added node, the insert that introduced
1221 
/// an anchor line), renumbered so the result is a well-formed change.
1222 
fn select_hunks(wc: &Change, indices: &[usize]) -> (Vec<TreeOp>, Vec<Edit>) {
1223 
    let n_ops = wc.tree_ops.len();
1224 
    let mut ops_sel = vec![false; n_ops];
1225 
    let mut edits_sel = vec![false; wc.edits.len()];
1226 
    for &i in indices {
1227 
        if i < n_ops {
1228 
            ops_sel[i] = true;
1229 
        } else if i - n_ops < wc.edits.len() {
1230 
            edits_sel[i - n_ops] = true;
1231 
        }
1232 
    }
1233 
    // Atom ranges each insert introduces, per node, in the working copy's numbering.
1234 
    let mut per_node: HashMap<NodeId, u32> = HashMap::new();
1235 
    let mut insert_ranges: Vec<Option<(NodeId, u32, u32)>> = Vec::new();
1236 
    for e in &wc.edits {
1237 
        insert_ranges.push(match e {
1238 
            Edit::Insert { node, lines, .. } => {
1239 
                let start = per_node.entry(*node).or_default();
1240 
                let r = (*node, *start, *start + lines.len() as u32);
1241 
                *start += lines.len() as u32;
1242 
                Some(r)
1243 
            }
1244 
            _ => None,
1245 
        });
1246 
    }
75167672 Andy McCall 1247 
    let add_index: HashMap<NodeId, usize> = wc
1248 
        .tree_ops
1249 
        .iter()
1250 
        .enumerate()
1251 
        .filter_map(|(i, o)| {
1252 
            if let TreeOp::AddNode { node, .. } = o {
1253 
                Some((*node, i))
1254 
            } else {
1255 
                None
1256 
            }
1257 
        })
1258 
        .collect();
1259 
    let insert_for = |node: &NodeId, index: u32| -> Option<usize> {
1260 
        insert_ranges
1261 
            .iter()
1262 
            .position(|r| matches!(r, Some((n, a, b)) if n == node && *a <= index && index < *b))
1263 
    };
957fee30 Andy McCall ↶ 1264 
    // Closure.
1265 
    let mut changed = true;
1266 
    while changed {
1267 
        changed = false;
1268 
        for i in 0..n_ops {
1269 
            if !ops_sel[i] {
1270 
                continue;
1271 
            }
d8c086ca Andy McCall ↶ 1272 
            if let TreeOp::AddNode { node, parent, .. } = &wc.tree_ops[i] {
957fee30 Andy McCall ↶ 1273 
                if let Some(&p) = add_index.get(parent) {
1274 
                    if !ops_sel[p] {
1275 
                        ops_sel[p] = true;
1276 
                        changed = true;
1277 
                    }
1278 
                }
d8c086ca Andy McCall ↶ 1279 
                // A new node's content belongs with its addition: nobody
1280 
                // wants an empty file committed apart from its lines.
1281 
                for (j, e) in wc.edits.iter().enumerate() {
1282 
                    if e.node() == *node && !edits_sel[j] {
1283 
                        edits_sel[j] = true;
1284 
                        changed = true;
1285 
                    }
1286 
                }
957fee30 Andy McCall ↶ 1287 
            }
1288 
        }
1289 
        for j in 0..wc.edits.len() {
1290 
            if !edits_sel[j] {
1291 
                continue;
1292 
            }
1293 
            let e = &wc.edits[j];
1294 
            if let Some(&a) = add_index.get(&e.node()) {
1295 
                if !ops_sel[a] {
1296 
                    ops_sel[a] = true;
1297 
                    changed = true;
1298 
                }
1299 
            }
75167672 Andy McCall 1300 
            if let Edit::Insert {
1301 
                node,
1302 
                after,
1303 
                before,
1304 
                ..
1305 
            } = e
1306 
            {
957fee30 Andy McCall ↶ 1307 
                for r in [after, before] {
1308 
                    if let AtomRef::Atom { change, index } = r {
1309 
                        if change.is_zero() {
1310 
                            if let Some(k) = insert_for(node, *index) {
1311 
                                if !edits_sel[k] {
1312 
                                    edits_sel[k] = true;
1313 
                                    changed = true;
1314 
                                }
1315 
                            }
1316 
                        }
1317 
                    }
1318 
                }
1319 
            }
1320 
        }
1321 
    }
1322 
    // Renumber self-references: old atom index → new index among selected inserts.
1323 
    let mut remap: HashMap<(NodeId, u32), u32> = HashMap::new();
1324 
    let mut new_counts: HashMap<NodeId, u32> = HashMap::new();
1325 
    for (j, r) in insert_ranges.iter().enumerate() {
1326 
        if let (true, Some((node, a, b))) = (edits_sel[j], r) {
1327 
            let next = new_counts.entry(*node).or_default();
1328 
            for old in *a..*b {
1329 
                remap.insert((*node, old), *next);
1330 
                *next += 1;
1331 
            }
1332 
        }
1333 
    }
1334 
    let map_ref = |r: AtomRef, node: &NodeId| -> AtomRef {
1335 
        match r {
75167672 Andy McCall 1336 
            AtomRef::Atom { change, index } if change.is_zero() => AtomRef::Atom {
1337 
                change,
1338 
                index: remap.get(&(*node, index)).copied().unwrap_or(index),
1339 
            },
957fee30 Andy McCall ↶ 1340 
            other => other,
1341 
        }
1342 
    };
75167672 Andy McCall 1343 
    let ops = wc
1344 
        .tree_ops
1345 
        .iter()
1346 
        .enumerate()
1347 
        .filter(|(i, _)| ops_sel[*i])
1348 
        .map(|(_, o)| o.clone())
1349 
        .collect();
957fee30 Andy McCall ↶ 1350 
    let edits = wc
1351 
        .edits
1352 
        .iter()
1353 
        .enumerate()
1354 
        .filter(|(j, _)| edits_sel[*j])
1355 
        .map(|(_, e)| match e {
75167672 Andy McCall 1356 
            Edit::Insert {
1357 
                node,
1358 
                after,
1359 
                before,
1360 
                lines,
1361 
            } => Edit::Insert {
1362 
                node: *node,
1363 
                after: map_ref(*after, node),
1364 
                before: map_ref(*before, node),
1365 
                lines: lines.clone(),
1366 
            },
957fee30 Andy McCall ↶ 1367 
            other => other.clone(),
1368 
        })
1369 
        .collect();
1370 
    (ops, edits)
1371 
}
1372
1373 
/// Render each tree operation and edit of `change` on its own, numbered.
1374 
pub fn render_hunks(change: &Change, base: &State, base_snapshot: &Snapshot) -> Vec<Hunk> {
1375 
    let by_node = base_snapshot.tree.by_node();
1376 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
1377 
    for op in &change.tree_ops {
75167672 Andy McCall 1378 
        if let TreeOp::AddNode {
1379 
            node,
1380 
            parent,
1381 
            name,
1382 
            kind,
1383 
            ..
1384 
        } = op
1385 
        {
957fee30 Andy McCall ↶ 1386 
            new_names.insert(*node, (*parent, name.clone(), *kind));
1387 
        }
1388 
    }
1389 
    let mut out = Vec::new();
1390 
    for (i, op) in change.tree_ops.iter().enumerate() {
75167672 Andy McCall 1391 
        let one = Change {
1392 
            tree_ops: vec![op.clone()],
1393 
            edits: vec![],
1394 
            ..change.clone()
1395 
        };
957fee30 Andy McCall ↶ 1396 
        let path = path_of_node(&op.node(), &by_node, &new_names).unwrap_or_default();
75167672 Andy McCall 1397 
        out.push(Hunk {
1398 
            index: i,
1399 
            path,
1400 
            text: render_change(&one, base, base_snapshot),
1401 
        });
957fee30 Andy McCall ↶ 1402 
    }
1403 
    for (j, e) in change.edits.iter().enumerate() {
75167672 Andy McCall 1404 
        let one = Change {
1405 
            tree_ops: vec![],
1406 
            edits: vec![e.clone()],
1407 
            ..change.clone()
1408 
        };
957fee30 Andy McCall ↶ 1409 
        let path = path_of_node(&e.node(), &by_node, &new_names).unwrap_or_default();
75167672 Andy McCall 1410 
        out.push(Hunk {
1411 
            index: change.tree_ops.len() + j,
1412 
            path,
1413 
            text: render_change(&one, base, base_snapshot),
1414 
        });
957fee30 Andy McCall ↶ 1415 
    }
1416 
    out
1417 
}
1418
c6aa5bd9 Andy McCall ↶ 1419 
/// One file's part of a diff, for `oxy diff --json`.
1420 
#[derive(Clone, Debug)]
1421 
pub struct FileDiff {
1422 
    pub path: String,
1423 
    /// `added`, `removed`, `modified`, `renamed`, `mode`, `replaced`.
1424 
    pub kind: String,
1425 
    pub hunks: Vec<DiffHunk>,
1426 
    /// For renames, the previous path.
1427 
    pub from: Option<String>,
1428 
}
1429
1430 
/// Lines inserted after `base_line` (0 = before the first line) and lines
1431 
/// removed, with their base line numbers.
1432 
#[derive(Clone, Debug, Default)]
1433 
pub struct DiffHunk {
1434 
    pub base_line: u32,
1435 
    pub added: Vec<Vec<u8>>,
1436 
    pub removed: Vec<(u32, Vec<u8>)>,
1437 
}
1438
1439 
impl Repo {
1440 
    /// The content of `path` in a state: `heads`, or the working copy's base.
1441 
    pub fn file_at(&self, path: &str, heads: Option<&[ChangeHash]>) -> Result<Vec<u8>> {
19303b4a Andy McCall ↶ 1442 
        // The state's history and tree, then the one file's graph: nothing
1443 
        // else is built or rendered (oxygen#92).
1444 
        // The working copy's base, and a state named by its heads (`-r
1445 
        // main` on main), come from the workspace's state index.
1446 
        let base = if self.is_bare() {
1447 
            None
1448 
        } else {
1449 
            Some(self.working_copy()?.1.deps)
1450 
        };
1451 
        let same = |a: &[ChangeHash], b: &[ChangeHash]| {
1452 
            let (mut a, mut b) = (a.to_vec(), b.to_vec());
1453 
            a.sort();
1454 
            b.sort();
1455 
            a == b
1456 
        };
1457 
        let mut state = match (heads, base) {
1458 
            (Some(h), Some(b)) if same(h, &b) => self.load_state_indexed(&b)?,
1459 
            (Some(h), _) => self.load_state_history(h)?,
1460 
            (None, Some(b)) => self.load_state_indexed(&b)?,
1461 
            (None, None) => return fail("this is a bare repository: it has no working copy"),
1462 
        };
1463 
        let view = state.tree().view();
1464 
        let Some(node) = view.entries.get(path).map(|e| e.node) else {
1465 
            return fail(format!("{path} is not in that state"));
c6aa5bd9 Andy McCall ↶ 1466 
        };
19303b4a Andy McCall ↶ 1467 
        self.load_graphs_of(&mut state, &[node])?;
1468 
        let snapshot = state.materialize_paths(view, [path]);
75167672 Andy McCall 1469 
        let Some(f) = snapshot.files.get(path) else {
1470 
            return fail(format!("{path} is not in that state"));
1471 
        };
c6aa5bd9 Andy McCall ↶ 1472 
        match &f.content {
1473 
            FileContent::Dir => fail(format!("{path} is a directory")),
1474 
            FileContent::Text { bytes, .. } => Ok(bytes.clone()),
1475 
            FileContent::Opaque { blobs } => match blobs.first() {
1476 
                Some(b) => self.read_blob(b),
1477 
                None => Ok(vec![]),
1478 
            },
1479 
        }
1480 
    }
1481
1482 
    /// Restore `paths` (files or directories) to the working copy's base:
1483 
    /// edits are undone, additions removed, removals put back. Returns the
1484 
    /// number of paths touched.
1485 
    pub fn discard(&self, paths: &[String]) -> Result<usize> {
6caacfff Andy McCall ↶ 1486 
        let _wc = self.lock_working_copy()?;
8ab265dc Andy McCall ↶ 1487 
        let snap = self.snapshot_status("oxy discard")?;
75167672 Andy McCall 1488 
        let under = |p: &str| {
1489 
            paths
1490 
                .iter()
1491 
                .any(|x| p == x || p.starts_with(&format!("{}/", x.trim_end_matches('/'))))
1492 
        };
c6aa5bd9 Andy McCall ↶ 1493 
        let attrs = Repo::state_attributes(&snap.base_snapshot);
1494 
        let mut n = 0;
1495 
        for s in &snap.statuses {
1496 
            if !under(&s.path) {
1497 
                continue;
1498 
            }
1499 
            match snap.base_snapshot.files.get(&s.path) {
1500 
                None => {
1501 
                    // Added (or the new name of a rename): remove it.
1502 
                    let full = self.root.join(&s.path);
1503 
                    if full.is_dir() {
1504 
                        let _ = fs::remove_dir_all(&full);
1505 
                    } else {
1506 
                        let _ = fs::remove_file(&full);
1507 
                    }
1508 
                    n += 1;
1509 
                }
1510 
                Some(f) => {
1511 
                    match &f.content {
1512 
                        FileContent::Dir => fs::create_dir_all(self.root.join(&s.path))?,
1513 
                        FileContent::Text { bytes, .. } => {
75167672 Andy McCall 1514 
                            let on_disk: std::borrow::Cow<[u8]> = if attrs.lookup(&s.path).eol
1515 
                                == Some(crate::attributes::Eol::Crlf)
1516 
                            {
1517 
                                crate::attributes::to_crlf(bytes).into()
1518 
                            } else {
1519 
                                bytes.as_slice().into()
1520 
                            };
c6aa5bd9 Andy McCall ↶ 1521 
                            self.write_file(&s.path, &on_disk, f.mode.exec)?;
1522 
                        }
1523 
                        FileContent::Opaque { blobs } => {
1524 
                            let bytes = match blobs.first() {
1525 
                                Some(b) => self.read_blob(b)?,
1526 
                                None => vec![],
1527 
                            };
1528 
                            if f.kind == NodeKind::Symlink {
1529 
                                self.write_symlink(&s.path, &bytes)?;
1530 
                            } else if f.kind == NodeKind::File {
1531 
                                self.write_file(&s.path, &bytes, f.mode.exec)?;
1532 
                            }
1533 
                        }
1534 
                    }
1535 
                    n += 1;
1536 
                }
1537 
            }
1538 
        }
1539 
        // A rename's old name is not in the statuses; put it back too.
1540 
        for (p, f) in &snap.base_snapshot.files {
75167672 Andy McCall 1541 
            if under(p)
1542 
                && !self.root.join(p).exists()
1543 
                && f.kind != NodeKind::Dir
1544 
                && f.kind != NodeKind::Subrepo
1545 
                && !snap.statuses.iter().any(|s| s.path == *p)
1546 
            {
c6aa5bd9 Andy McCall ↶ 1547 
                let bytes = match &f.content {
1548 
                    FileContent::Text { bytes, .. } => bytes.clone(),
75167672 Andy McCall 1549 
                    FileContent::Opaque { blobs } => blobs
1550 
                        .first()
1551 
                        .map(|b| self.read_blob(b))
1552 
                        .transpose()?
1553 
                        .unwrap_or_default(),
c6aa5bd9 Andy McCall ↶ 1554 
                    FileContent::Dir => continue,
1555 
                };
75167672 Andy McCall 1556 
                if f.kind == NodeKind::Symlink {
1557 
                    self.write_symlink(p, &bytes)?
1558 
                } else {
1559 
                    self.write_file(p, &bytes, f.mode.exec)?
1560 
                }
c6aa5bd9 Andy McCall ↶ 1561 
                n += 1;
1562 
            }
1563 
        }
1564 
        self.drop_manifest();
1565 
        Ok(n)
1566 
    }
1567
1568 
    /// Undo the working-copy hunks `indices` (as `hunks` numbers them) on
1569 
    /// disk, keeping the others: the files are rewritten from the base plus
1570 
    /// the remaining hunks.
1571 
    pub fn discard_hunks(&self, indices: &[usize]) -> Result<usize> {
6caacfff Andy McCall ↶ 1572 
        let _wc = self.lock_working_copy()?;
c6aa5bd9 Andy McCall ↶ 1573 
        let (snap, hunks) = self.hunks()?;
75167672 Andy McCall 1574 
        let keep: Vec<usize> = hunks
1575 
            .iter()
1576 
            .map(|h| h.index)
1577 
            .filter(|i| !indices.contains(i))
1578 
            .collect();
c6aa5bd9 Andy McCall ↶ 1579 
        let (ops, edits) = select_hunks(&snap.wc, &keep);
1580 
        let dropped_nodes: BTreeSet<NodeId> = hunks
1581 
            .iter()
1582 
            .filter(|h| indices.contains(&h.index))
75167672 Andy McCall 1583 
            .map(|h| {
1584 
                if h.index < snap.wc.tree_ops.len() {
1585 
                    snap.wc.tree_ops[h.index].node()
1586 
                } else {
1587 
                    snap.wc.edits[h.index - snap.wc.tree_ops.len()].node()
1588 
                }
1589 
            })
c6aa5bd9 Andy McCall ↶ 1590 
            .collect();
1591 
        let mut partial = snap.wc.clone();
1592 
        partial.tree_ops = ops;
1593 
        partial.edits = edits;
1594 
        partial.normalize();
1595 
        let mut st = snap.base.clone();
1596 
        st.add_unchecked(partial);
8ab265dc Andy McCall ↶ 1597 
        // Every file of the target is rendered: build the graphs the
1598 
        // partial snapshot left out.
1599 
        self.load_all_graphs(&mut st)?;
c6aa5bd9 Andy McCall ↶ 1600 
        let target = st.materialize();
1601 
        let by_node_target = target.tree.by_node();
1602 
        let by_node_base = snap.base_snapshot.tree.by_node();
1603 
        let attrs = Repo::state_attributes(&target);
1604 
        let mut n = 0;
1605 
        for node in dropped_nodes {
1606 
            match by_node_target.get(&node) {
1607 
                Some(e) => {
1608 
                    let f = &target.files[&e.path];
1609 
                    match &f.content {
1610 
                        FileContent::Dir => {}
1611 
                        FileContent::Text { bytes, .. } => {
75167672 Andy McCall 1612 
                            let on_disk: std::borrow::Cow<[u8]> = if attrs.lookup(&e.path).eol
1613 
                                == Some(crate::attributes::Eol::Crlf)
1614 
                            {
1615 
                                crate::attributes::to_crlf(bytes).into()
1616 
                            } else {
1617 
                                bytes.as_slice().into()
1618 
                            };
c6aa5bd9 Andy McCall ↶ 1619 
                            self.write_file(&e.path, &on_disk, f.mode.exec)?;
1620 
                        }
1621 
                        FileContent::Opaque { blobs } => {
75167672 Andy McCall 1622 
                            let bytes = blobs
1623 
                                .first()
1624 
                                .map(|b| self.read_blob(b))
1625 
                                .transpose()?
1626 
                                .unwrap_or_default();
1627 
                            if f.kind == NodeKind::Symlink {
1628 
                                self.write_symlink(&e.path, &bytes)?
1629 
                            } else {
1630 
                                self.write_file(&e.path, &bytes, f.mode.exec)?
1631 
                            }
c6aa5bd9 Andy McCall ↶ 1632 
                        }
1633 
                    }
1634 
                    // A discarded rename: the new name goes.
75167672 Andy McCall 1635 
                    if let Some(old_path) = snap
1636 
                        .statuses
1637 
                        .iter()
1638 
                        .find(|s| {
1639 
                            s.status == ChangeStatus::Renamed && !target.files.contains_key(&s.path)
1640 
                        })
1641 
                        .map(|s| s.path.clone())
1642 
                    {
c6aa5bd9 Andy McCall ↶ 1643 
                        let _ = fs::remove_file(self.root.join(old_path));
1644 
                    }
1645 
                    n += 1;
1646 
                }
1647 
                None => {
1648 
                    // The node exists in neither target nor base with a live path: an addition to remove.
75167672 Andy McCall 1649 
                    if let Some(added) = snap.statuses.iter().find(|s| {
1650 
                        s.status == ChangeStatus::Added
1651 
                            && !target.files.contains_key(&s.path)
1652 
                            && !by_node_base.contains_key(&node)
1653 
                    }) {
c6aa5bd9 Andy McCall ↶ 1654 
                        let full = self.root.join(&added.path);
75167672 Andy McCall 1655 
                        if full.is_dir() {
1656 
                            let _ = fs::remove_dir_all(&full);
1657 
                        } else {
1658 
                            let _ = fs::remove_file(&full);
1659 
                        }
c6aa5bd9 Andy McCall ↶ 1660 
                        n += 1;
1661 
                    }
1662 
                }
1663 
            }
1664 
        }
1665 
        self.drop_manifest();
1666 
        Ok(n)
1667 
    }
1668
1669 
    /// The diff of the working copy (`None`) or of a change, per file with
1670 
    /// line-numbered hunks (spec: `oxy diff --json`).
1671 
    pub fn diff_files(&self, change: Option<&ChangeHash>) -> Result<Vec<FileDiff>> {
8ab265dc Andy McCall ↶ 1672 
        let (c, base, base_snapshot) = match change {
c6aa5bd9 Andy McCall ↶ 1673 
            Some(h) => {
1674 
                let c = self.store.get_change(h)?;
8ab265dc Andy McCall ↶ 1675 
                let (base, snapshot) = self.change_base(&c)?;
1676 
                (c, base, snapshot)
c6aa5bd9 Andy McCall ↶ 1677 
            }
77445673 Andy McCall ↶ 1678 
            None => self.working_copy_with_base()?,
c6aa5bd9 Andy McCall ↶ 1679 
        };
1680 
        Ok(diff_files_of(&c, &base, &base_snapshot))
e226a6cc Andy McCall ↶ 1681 
    }
1682
1683 
    /// The snapshot a command would take, without taking it, when the
1684 
    /// status cache holds: every file is as the last snapshot left it, so
1685 
    /// that snapshot's working-copy change is the answer. Its base comes
1686 
    /// from the state index, the `.oxyattributes` files from disk. `None`
1687 
    /// when a snapshot must be taken, and when there is nothing to commit
1688 
    /// (the snapshot then says so). A conflicted base never gets here: the
1689 
    /// cache does not hold for one.
1690 
    pub(crate) fn snapshot_from_cache(&self) -> Result<Option<SnapshotResult>> {
1691 
        let Some(statuses) = self.status_by_manifest()? else {
1692 
            return Ok(None);
1693 
        };
1694 
        if statuses.is_empty() {
1695 
            return Ok(None);
1696 
        }
1697 
        let _t = crate::timing::phase("snapshot: from the status cache");
1698 
        let (wc_hash, wc) = self.working_copy()?;
1699 
        let base = self.load_state_indexed(&wc.deps)?;
1700 
        let (base, base_snapshot) = self.render_base_for(&wc, base)?;
1701 
        let attribute_files: Vec<(String, Vec<u8>)> = match self.read_manifest()? {
1702 
            Some((_, _, m)) => m
1703 
                .entries
1704 
                .keys()
1705 
                .filter(|p| is_attributes(p))
1706 
                .filter_map(|p| fs::read(self.root.join(p)).ok().map(|b| (p.clone(), b)))
1707 
                .collect(),
1708 
            None => Vec::new(),
1709 
        };
1710 
        let attributes = crate::attributes::Attributes::from_files(
1711 
            attribute_files
1712 
                .iter()
1713 
                .map(|(p, b)| (p.as_str(), b.as_slice())),
1714 
        );
1715 
        Ok(Some(SnapshotResult {
1716 
            wc_hash,
1717 
            wc,
1718 
            base,
1719 
            base_snapshot,
1720 
            statuses,
1721 
            changed: false,
1722 
            resolves: false,
1723 
            contributors: Vec::new(),
1724 
            attributes,
1725 
            untracked_nested: self.untracked_nested.borrow().clone(),
236f35c0 Andy McCall ↶ 1726 
            manifest: None,
e226a6cc Andy McCall ↶ 1727 
        }))
77445673 Andy McCall ↶ 1728 
    }
1729
1730 
    /// The working-copy change with its base state and the base's
1731 
    /// rendering, for showing what the working copy changes. While the
1732 
    /// status cache holds (every file as the last snapshot left it), the
1733 
    /// stored working-copy change is the answer and no snapshot is taken;
1734 
    /// otherwise a snapshot makes it.
1735 
    pub fn working_copy_with_base(&self) -> Result<(Change, State, Snapshot)> {
1736 
        if self.status_by_manifest()?.is_some() {
1737 
            let (_, wc) = self.working_copy()?;
1738 
            let base = self.load_state_indexed(&wc.deps)?;
1739 
            let (base, snapshot) = self.render_base_for(&wc, base)?;
1740 
            return Ok((wc, base, snapshot));
1741 
        }
1742 
        let snap = self.snapshot_status("oxy diff")?;
1743 
        Ok((snap.wc, snap.base, snap.base_snapshot))
c6aa5bd9 Andy McCall ↶ 1744 
    }
1745
1746 
    /// The log restricted to changes touching `path` (its node, so renames
1747 
    /// are followed), on the current bookmark.
1748 
    pub fn log_for_path(&self, path: &str) -> Result<Vec<LogEntry>> {
e02bf0cd Andy McCall ↶ 1749 
        let heads = self.bookmark_heads()?;
1750 
        if self.is_bare() {
1751 
            return self.log_for_path_in(path, &heads);
1752 
        }
c6aa5bd9 Andy McCall ↶ 1753 
        let entries = self.log()?;
1754 
        let (_, wc) = self.working_copy()?;
170bb436 Andy McCall ↶ 1755 
        // The path's node from the trees alone: the base's (from the
1756 
        // workspace's state index), else the bookmark's. Nothing is
1757 
        // rendered.
1758 
        let node_in = |state: State| state.tree().view().entries.get(path).map(|e| e.node);
1759 
        let node = match node_in(self.load_state_indexed(&wc.deps)?) {
1760 
            Some(n) => Some(n),
1761 
            None => node_in(self.load_state_history(&heads)?),
1762 
        };
75167672 Andy McCall 1763 
        let Some(node) = node else {
1764 
            return fail(format!("{path} is not tracked"));
1765 
        };
e02bf0cd Andy McCall ↶ 1766 
        Ok(self.entries_touching(entries, node))
1767 
    }
1768
1769 
    /// The log of the state with these heads restricted to changes touching
1770 
    /// `path` (its node, so renames are followed).
1771 
    pub fn log_for_path_in(&self, path: &str, heads: &[ChangeHash]) -> Result<Vec<LogEntry>> {
e8d38e47 Andy McCall ↶ 1772 
        self.log_for_path_of(&self.load_state_history(heads)?, path, None)
bc489b6c Andy McCall ↶ 1773 
    }
1774
1775 
    /// The changes of a loaded state touching `path`, newest first, at most
1776 
    /// `limit` of them. The path is found in the state's tree alone, so no
1777 
    /// file is rendered, and the walk stops once `limit` are found.
1778 
    pub fn log_for_path_of(
1779 
        &self,
1780 
        state: &State,
1781 
        path: &str,
1782 
        limit: Option<usize>,
1783 
    ) -> Result<Vec<LogEntry>> {
1784 
        let Some(node) = state.tree().view().entries.get(path).map(|e| e.node) else {
75167672 Andy McCall 1785 
            return fail(format!("{path} is not in that state"));
1786 
        };
bc489b6c Andy McCall ↶ 1787 
        Ok(self.entries_touching_up_to(self.log_of(state), node, limit))
e02bf0cd Andy McCall ↶ 1788 
    }
1789
1790 
    fn entries_touching(&self, entries: Vec<LogEntry>, node: NodeId) -> Vec<LogEntry> {
bc489b6c Andy McCall ↶ 1791 
        self.entries_touching_up_to(entries, node, None)
1792 
    }
1793
1794 
    fn entries_touching_up_to(
1795 
        &self,
1796 
        entries: Vec<LogEntry>,
1797 
        node: NodeId,
1798 
        limit: Option<usize>,
1799 
    ) -> Vec<LogEntry> {
170bb436 Andy McCall ↶ 1800 
        let mut touched = self.touched();
1801 
        if limit.is_none() {
1802 
            let hashes: Vec<ChangeHash> = entries.iter().map(|e| e.hash).collect();
1803 
            touched.fill(self, &hashes);
1804 
        }
1805 
        let wanted = HashSet::from([node]);
c6aa5bd9 Andy McCall ↶ 1806 
        let mut out = Vec::new();
1807 
        for e in entries {
bc489b6c Andy McCall ↶ 1808 
            if limit.is_some_and(|n| out.len() >= n) {
1809 
                break;
1810 
            }
170bb436 Andy McCall ↶ 1811 
            if touched.touches(self, &e.hash, &wanted) {
c6aa5bd9 Andy McCall ↶ 1812 
                out.push(e);
1813 
            }
1814 
        }
170bb436 Andy McCall ↶ 1815 
        touched.save();
e02bf0cd Andy McCall ↶ 1816 
        out
c6aa5bd9 Andy McCall ↶ 1817 
    }
1818 
}
1819
1820 
/// Per-file hunks of `change` against `base`.
1821 
pub fn diff_files_of(change: &Change, base: &State, base_snapshot: &Snapshot) -> Vec<FileDiff> {
1822 
    let by_node = base_snapshot.tree.by_node();
1823 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
1824 
    for op in &change.tree_ops {
75167672 Andy McCall 1825 
        if let TreeOp::AddNode {
1826 
            node,
1827 
            parent,
1828 
            name,
1829 
            kind,
1830 
            ..
1831 
        } = op
1832 
        {
c6aa5bd9 Andy McCall ↶ 1833 
            new_names.insert(*node, (*parent, name.clone(), *kind));
1834 
        }
1835 
    }
75167672 Andy McCall 1836 
    let path_of = |n: &NodeId| -> String {
1837 
        path_of_node(n, &by_node, &new_names).unwrap_or_else(|| format!("<node {}>", n.short(8)))
1838 
    };
c6aa5bd9 Andy McCall ↶ 1839 
    let mut files: BTreeMap<NodeId, FileDiff> = BTreeMap::new();
1840 
    let mut order: Vec<NodeId> = Vec::new();
1841 
    let mut entry = |files: &mut BTreeMap<NodeId, FileDiff>, n: NodeId, kind: &str| {
1842 
        if let std::collections::btree_map::Entry::Vacant(v) = files.entry(n) {
1843 
            order.push(n);
75167672 Andy McCall 1844 
            v.insert(FileDiff {
1845 
                path: path_of(&n),
1846 
                kind: kind.to_string(),
1847 
                hunks: vec![],
1848 
                from: None,
1849 
            });
c6aa5bd9 Andy McCall ↶ 1850 
        }
1851 
    };
1852 
    for op in &change.tree_ops {
1853 
        match op {
1854 
            TreeOp::AddNode { node, kind, .. } => {
1855 
                if *kind != NodeKind::Dir {
1856 
                    entry(&mut files, *node, "added");
1857 
                }
1858 
            }
1859 
            TreeOp::RemoveNode { node, .. } => {
1860 
                entry(&mut files, *node, "removed");
1861 
                if let Some(e) = by_node.get(node) {
1862 
                    files.get_mut(node).unwrap().path = e.path.clone();
1863 
                }
1864 
            }
75167672 Andy McCall 1865 
            TreeOp::Move {
1866 
                node, parent, name, ..
1867 
            } => {
c6aa5bd9 Andy McCall ↶ 1868 
                entry(&mut files, *node, "renamed");
1869 
                let f = files.get_mut(node).unwrap();
1870 
                f.from = by_node.get(node).map(|e| e.path.clone());
75167672 Andy McCall 1871 
                let dir = if parent.is_root() {
1872 
                    String::new()
1873 
                } else {
1874 
                    path_of_node(parent, &by_node, &new_names).unwrap_or_default()
1875 
                };
1876 
                f.path = if dir.is_empty() {
1877 
                    name.clone()
1878 
                } else {
1879 
                    format!("{dir}/{name}")
1880 
                };
c6aa5bd9 Andy McCall ↶ 1881 
            }
1882 
            TreeOp::SetMode { node, .. } => entry(&mut files, *node, "mode"),
1883 
        }
1884 
    }
1885 
    // Base line numbers per node, 1-based.
1886 
    let mut line_of: HashMap<NodeId, HashMap<(ChangeHash, u32), u32>> = HashMap::new();
1887 
    let mut lines_for = |n: &NodeId| -> HashMap<(ChangeHash, u32), u32> {
75167672 Andy McCall 1888 
        line_of
1889 
            .entry(*n)
1890 
            .or_insert_with(|| {
1891 
                base.live_lines(n)
1892 
                    .into_iter()
1893 
                    .enumerate()
1894 
                    .map(|(i, (k, _))| (k, i as u32 + 1))
1895 
                    .collect()
1896 
            })
1897 
            .clone()
c6aa5bd9 Andy McCall ↶ 1898 
    };
1899 
    for e in &change.edits {
1900 
        let n = e.node();
1901 
        let is_new = new_names.contains_key(&n);
1902 
        entry(&mut files, n, if is_new { "added" } else { "modified" });
1903 
        let f = files.get_mut(&n).unwrap();
1904 
        match e {
1905 
            Edit::Insert { after, lines, .. } => {
1906 
                let map = lines_for(&n);
1907 
                let base_line = match after {
1908 
                    AtomRef::Start => 0,
1909 
                    AtomRef::End => map.len() as u32,
75167672 Andy McCall 1910 
                    AtomRef::Atom { change, index } if change.is_zero() => {
1911 
                        f.hunks.last().map(|h| h.base_line).unwrap_or(0)
1912 
                    }
1913 
                    AtomRef::Atom { change, index } => {
1914 
                        map.get(&(*change, *index)).copied().unwrap_or(0)
1915 
                    }
c6aa5bd9 Andy McCall ↶ 1916 
                };
75167672 Andy McCall 1917 
                f.hunks.push(DiffHunk {
1918 
                    base_line,
01a47cd7 Andy McCall ↶ 1919 
                    added: lines.to_vec(),
75167672 Andy McCall 1920 
                    removed: vec![],
1921 
                });
c6aa5bd9 Andy McCall ↶ 1922 
            }
1923 
            Edit::Delete { atoms, .. } => {
1924 
                let map = lines_for(&n);
1925 
                let mut removed = Vec::new();
1926 
                for r in atoms {
1927 
                    for i in r.from..r.to {
1928 
                        let line = map.get(&(r.change, i)).copied().unwrap_or(0);
75167672 Andy McCall 1929 
                        let text = base
1930 
                            .file(&n)
1931 
                            .and_then(|g| g.content(&(r.change, i)))
1932 
                            .unwrap_or(b"")
1933 
                            .to_vec();
c6aa5bd9 Andy McCall ↶ 1934 
                        removed.push((line, text));
1935 
                    }
1936 
                }
75167672 Andy McCall 1937 
                let base_line = removed
1938 
                    .first()
1939 
                    .map(|(l, _)| l.saturating_sub(1))
1940 
                    .unwrap_or(0);
1941 
                f.hunks.push(DiffHunk {
1942 
                    base_line,
1943 
                    added: vec![],
1944 
                    removed,
1945 
                });
c6aa5bd9 Andy McCall ↶ 1946 
            }
1947 
            Edit::Replace { .. } => {
1948 
                if !is_new {
1949 
                    f.kind = "replaced".into();
1950 
                }
1951 
            }
1952 
        }
1953 
    }
1954 
    order.into_iter().filter_map(|n| files.remove(&n)).collect()
1955 
}
1956
98286406 Andy McCall ↶ 1957 
/// Builds the tree ops and edits of a working-copy change against a base.
dfb23a3a Andy McCall ↶ 1958 
/// `BLAKE3("oxy node 1" 0 seed kind path)`, truncated: a node identifier
1959 
/// that the same seed and path always give (never the root's).
1960 
fn seeded_node(seed: &[u8; 32], path: &str, kind: NodeKind) -> NodeId {
1961 
    let mut h = blake3::Hasher::new();
1962 
    h.update(b"oxy node 1\0");
1963 
    h.update(seed);
1964 
    h.update(&[match kind {
1965 
        NodeKind::File => 1,
1966 
        NodeKind::Dir => 2,
1967 
        NodeKind::Symlink => 3,
1968 
        NodeKind::Subrepo => 4,
1969 
    }]);
1970 
    h.update(path.as_bytes());
1971 
    let mut id = [0u8; 16];
1972 
    id.copy_from_slice(&h.finalize().as_bytes()[..16]);
1973 
    if id == [0; 16] {
1974 
        id[15] = 1;
1975 
    }
1976 
    NodeId(id)
1977 
}
1978
98286406 Andy McCall ↶ 1979 
struct ChangeBuilder<'a> {
1980 
    base: &'a State,
1981 
    base_snapshot: &'a Snapshot,
1982 
    tree_ops: Vec<TreeOp>,
1983 
    edits: Vec<Edit>,
1984 
    statuses: Vec<PathStatus>,
1985 
    dir_nodes: HashMap<String, NodeId>,
1986 
    reuse: HashMap<String, (NodeId, NodeKind)>,
1987 
    /// Changes that contributed to conflicts this working copy resolves.
1988 
    contributors: BTreeSet<ChangeHash>,
1989 
    resolves: bool,
cdb5beba Andy McCall ↶ 1990 
    attrs: crate::attributes::Attributes,
b7f102fc Andy McCall ↶ 1991 
    blobs: HashMap<String, ObjectRef>,
f7510b63 Andy McCall ↶ 1992 
    /// Opaque files whose replace conflict the content settles.
1993 
    resolve: HashSet<String>,
dfb23a3a Andy McCall ↶ 1994 
    /// When set, new nodes' identifiers derive from it and their paths.
1995 
    node_seed: Option<[u8; 32]>,
a87c7b9d Andy McCall ↶ 1996 
    /// For a Git import, the base's dead directories by where they were
1997 
    /// (see `ensure_dir`), worked out when first needed.
1998 
    dead_dirs: Option<HashMap<(NodeId, String), (NodeId, crate::tree::OpKey)>>,
360aa74a Andy McCall ↶ 1999 
    /// Text edits of edited files, worked out in parallel beforehand
2000 
    /// (`diff_in_parallel`); a file not here is diffed when it is updated.
2001 
    diffs: HashMap<String, Vec<Edit>>,
98286406 Andy McCall ↶ 2002 
}
2003
2004 
impl<'a> ChangeBuilder<'a> {
75167672 Andy McCall 2005 
    fn new(
2006 
        base: &'a State,
2007 
        base_snapshot: &'a Snapshot,
2008 
        reuse: HashMap<String, (NodeId, NodeKind)>,
2009 
    ) -> Self {
98286406 Andy McCall ↶ 2010 
        let mut dir_nodes: HashMap<String, NodeId> = HashMap::new();
2011 
        dir_nodes.insert(String::new(), NodeId::ROOT);
e83354cd Andy McCall ↶ 2012 
        for (p, e) in base_snapshot.tree.entries.iter() {
2306a4d2 Andy McCall ↶ 2013 
            if e.kind == NodeKind::Dir {
2014 
                dir_nodes.insert(p.clone(), e.node);
98286406 Andy McCall ↶ 2015 
            }
2016 
        }
75167672 Andy McCall 2017 
        ChangeBuilder {
2018 
            base,
2019 
            base_snapshot,
2020 
            tree_ops: vec![],
2021 
            edits: vec![],
2022 
            statuses: vec![],
2023 
            dir_nodes,
2024 
            reuse,
2025 
            contributors: BTreeSet::new(),
2026 
            resolves: false,
2027 
            attrs: crate::attributes::Attributes::default(),
2028 
            blobs: HashMap::new(),
f7510b63 Andy McCall ↶ 2029 
            resolve: HashSet::new(),
dfb23a3a Andy McCall ↶ 2030 
            node_seed: None,
a87c7b9d Andy McCall ↶ 2031 
            dead_dirs: None,
360aa74a Andy McCall ↶ 2032 
            diffs: HashMap::new(),
75167672 Andy McCall 2033 
        }
98286406 Andy McCall ↶ 2034 
    }
2035
360aa74a Andy McCall ↶ 2036 
    /// Diff the edited text files among `entries` against their base
2037 
    /// renderings at once, on every core, for `update_path` to take: the
2038 
    /// same diff it would make, only not one file after another.
2039 
    fn diff_in_parallel(&mut self, entries: &BTreeMap<String, FsEntry>) {
2040 
        use rayon::prelude::*;
2041 
        let files = &self.base_snapshot.files;
2042 
        let edited: Vec<(&String, &FsEntry, &crate::state::FileState)> = entries
2043 
            .iter()
2044 
            .filter_map(|(p, e)| {
2045 
                let x = files.get(p)?;
2046 
                (x.kind == e.kind && !e.unchanged && !self.opaque_for(p, e)).then_some((p, e, x))
2047 
            })
2048 
            .collect();
2049 
        // Below this, starting the threads costs more than it saves.
2050 
        if edited.len() < 16 {
2051 
            return;
2052 
        }
2053 
        let this = &*self;
2054 
        let attrs = &this.attrs;
2055 
        let diffs: Vec<(String, Vec<Edit>)> = edited
2056 
            .par_iter()
2057 
            .filter_map(|(p, e, x)| {
2058 
                let FileContent::Text {
2059 
                    bytes,
2060 
                    materialized,
2061 
                } = &x.content
2062 
                else {
2063 
                    return None;
2064 
                };
2065 
                if materialized.is_conflicted() {
2066 
                    return None;
2067 
                }
2068 
                let disk_text = if attrs.lookup(p).eol.is_some() {
2069 
                    std::sync::Arc::new(crate::attributes::normalize_eol(&e.content))
2070 
                } else {
2071 
                    e.content.clone()
2072 
                };
2073 
                if *bytes == *disk_text {
2074 
                    return None;
2075 
                }
2076 
                let keys = materialized.live_atoms();
2077 
                let lines = crate::diff::split_lines(bytes);
2078 
                (keys.len() == lines.len()).then(|| {
2079 
                    let edits = crate::diff::text_edits_of(x.node, &keys, &lines, &disk_text);
2080 
                    ((*p).clone(), edits)
2081 
                })
2082 
            })
2083 
            .collect();
2084 
        self.diffs = diffs.into_iter().collect();
2085 
    }
2086
cdb5beba Andy McCall ↶ 2087 
    fn with_attributes(mut self, attrs: crate::attributes::Attributes) -> Self {
2088 
        self.attrs = attrs;
2089 
        self
2090 
    }
2091
2092 
    /// Whether a file is opaque: forced by `binary`, forced text by `text`
2093 
    /// only when it has no NUL bytes, else sniffed.
b7f102fc Andy McCall ↶ 2094 
    /// The blob for an opaque entry: the cached one, else stored now.
75167672 Andy McCall 2095 
    fn blob_of(
2096 
        &mut self,
2097 
        repo: &Repo,
2098 
        path: &str,
2099 
        e: &FsEntry,
2100 
        nocompress: bool,
2101 
    ) -> Result<ObjectRef> {
b7f102fc Andy McCall ↶ 2102 
        let b = match e.blob {
2103 
            Some(b) => b,
2104 
            None => repo.store_blob_for(&e.content, nocompress)?,
2105 
        };
2106 
        self.blobs.insert(path.to_string(), b);
2107 
        Ok(b)
2108 
    }
2109
cdb5beba Andy McCall ↶ 2110 
    fn opaque_for(&self, path: &str, e: &FsEntry) -> bool {
b7f102fc Andy McCall ↶ 2111 
        if e.blob.is_some() {
2112 
            return true;
2113 
        }
4dee277c Andy McCall ↶ 2114 
        if e.kind == NodeKind::Symlink || e.kind == NodeKind::Subrepo {
cdb5beba Andy McCall ↶ 2115 
            return true;
2116 
        }
2117 
        match self.attrs.lookup(path).binary {
2118 
            Some(true) => true,
2119 
            Some(false) => is_binary(&e.content),
2120 
            None => is_binary(&e.content),
2121 
        }
2122 
    }
2123
2124 
    /// Text content as stored: `\r\n` normalized for `eol=` files.
40a40bb7 Andy McCall ↶ 2125 
    fn stored_text(
2126 
        &self,
2127 
        path: &str,
2128 
        content: &std::sync::Arc<Vec<u8>>,
2129 
    ) -> std::sync::Arc<Vec<u8>> {
75167672 Andy McCall 2130 
        if self.attrs.lookup(path).eol.is_some() {
40a40bb7 Andy McCall ↶ 2131 
            std::sync::Arc::new(crate::attributes::normalize_eol(content))
75167672 Andy McCall 2132 
        } else {
40a40bb7 Andy McCall ↶ 2133 
            content.clone()
dfb23a3a Andy McCall ↶ 2134 
        }
2135 
    }
2136
2137 
    /// The identifier of a node added at `path`: reused when asked, derived
2138 
    /// from the seed when there is one, random otherwise.
2139 
    fn new_node(&self, path: &str, kind: NodeKind) -> NodeId {
2140 
        if let Some(n) = self.reused_node(path, kind) {
2141 
            return n;
75167672 Andy McCall 2142 
        }
dfb23a3a Andy McCall ↶ 2143 
        match &self.node_seed {
2144 
            Some(seed) => seeded_node(seed, path, kind),
2145 
            None => NodeId::random(),
2146 
        }
cdb5beba Andy McCall ↶ 2147 
    }
2148
98286406 Andy McCall ↶ 2149 
    fn reused_node(&self, path: &str, kind: NodeKind) -> Option<NodeId> {
75167672 Andy McCall 2150 
        self.reuse
2151 
            .get(path)
2152 
            .filter(|(_, k)| *k == kind)
2153 
            .map(|(n, _)| *n)
98286406 Andy McCall ↶ 2154 
    }
2155
a87c7b9d Andy McCall ↶ 2156 
    /// A dead directory last placed at (`parent`, `name`) in the base.
2157 
    fn revivable_dir(
2158 
        &mut self,
2159 
        parent: NodeId,
2160 
        name: &str,
2161 
    ) -> Option<(NodeId, crate::tree::OpKey)> {
2162 
        let base = self.base;
2163 
        self.dead_dirs
2164 
            .get_or_insert_with(|| base.tree().dead_dirs())
2165 
            .get(&(parent, name.to_string()))
2166 
            .copied()
2167 
    }
2168
98286406 Andy McCall ↶ 2169 
    fn ensure_dir(&mut self, path: &str) -> NodeId {
2170 
        if let Some(n) = self.dir_nodes.get(path) {
2171 
            return *n;
2172 
        }
2173 
        let mut parent = NodeId::ROOT;
2174 
        let mut prefix = String::new();
2175 
        for comp in path.split('/') {
2176 
            if !prefix.is_empty() {
2177 
                prefix.push('/');
2178 
            }
2179 
            prefix.push_str(comp);
a87c7b9d Andy McCall ↶ 2180 
            let known = self.dir_nodes.get(&prefix).copied();
2181 
            let revive = match known {
2182 
                Some(_) => None,
2183 
                None if self.node_seed.is_some() => self.revivable_dir(parent, comp),
2184 
                None => None,
2185 
            };
2186 
            parent = match (known, revive) {
2187 
                (Some(n), _) => n,
2188 
                (None, Some((node, old))) => {
2189 
                    // Git has no identity for a directory: one that comes
2190 
                    // back where it was is the same directory, brought back
2191 
                    // with a move, so lines of history that placed files in
2192 
                    // it before and after agree (oxygen#30).
2193 
                    self.tree_ops.push(TreeOp::Move {
2194 
                        node,
2195 
                        old: OpRef {
2196 
                            change: old.0,
2197 
                            index: old.1,
2198 
                        },
2199 
                        parent,
2200 
                        name: comp.to_string(),
2201 
                    });
2202 
                    self.dir_nodes.insert(prefix.clone(), node);
2203 
                    node
2204 
                }
2205 
                (None, None) => {
dfb23a3a Andy McCall ↶ 2206 
                    let node = self.new_node(&prefix, NodeKind::Dir);
75167672 Andy McCall 2207 
                    self.tree_ops.push(TreeOp::AddNode {
2208 
                        node,
2209 
                        kind: NodeKind::Dir,
2210 
                        mode: Mode::default(),
2211 
                        parent,
2212 
                        name: comp.to_string(),
2213 
                    });
98286406 Andy McCall ↶ 2214 
                    self.dir_nodes.insert(prefix.clone(), node);
2215 
                    node
2216 
                }
2217 
            };
2218 
        }
2219 
        parent
2220 
    }
2221
2222 
    fn add_path(&mut self, repo: &Repo, path: &str, e: &FsEntry) -> Result<()> {
2223 
        let (dir, name) = split_path(path);
2224 
        let parent = self.ensure_dir(dir);
dfb23a3a Andy McCall ↶ 2225 
        let node = self.new_node(path, e.kind);
cdb5beba Andy McCall ↶ 2226 
        let opaque = self.opaque_for(path, e);
2227 
        let nocompress = self.attrs.lookup(path).nocompress;
75167672 Andy McCall 2228 
        self.tree_ops.push(TreeOp::AddNode {
2229 
            node,
2230 
            kind: e.kind,
2231 
            mode: Mode {
2232 
                exec: e.exec,
2233 
                opaque,
2234 
            },
2235 
            parent,
2236 
            name: name.to_string(),
2237 
        });
98286406 Andy McCall ↶ 2238 
        if opaque {
b7f102fc Andy McCall ↶ 2239 
            let blob = self.blob_of(repo, path, e, nocompress)?;
75167672 Andy McCall 2240 
            self.edits.push(Edit::Replace {
2241 
                node,
2242 
                old: None,
2243 
                new: Some(blob),
2244 
            });
98286406 Andy McCall ↶ 2245 
        } else if !e.content.is_empty() {
cdb5beba Andy McCall ↶ 2246 
            let content = self.stored_text(path, &e.content);
98286406 Andy McCall ↶ 2247 
            self.edits.push(Edit::Insert {
2248 
                node,
2249 
                after: crate::objects::AtomRef::Start,
2250 
                before: crate::objects::AtomRef::End,
40a40bb7 Andy McCall ↶ 2251 
                lines: crate::lines::Lines::split_shared(content),
98286406 Andy McCall ↶ 2252 
            });
2253 
        }
75167672 Andy McCall 2254 
        self.statuses.push(PathStatus {
2255 
            path: path.to_string(),
2256 
            status: ChangeStatus::Added,
2257 
        });
98286406 Andy McCall ↶ 2258 
        Ok(())
2259 
    }
2260
6e723b81 Andy McCall ↶ 2261 
    /// Move the node at `old` to `new` and apply `e`'s content to it.
75167672 Andy McCall 2262 
    fn move_path(
2263 
        &mut self,
2264 
        repo: &Repo,
2265 
        old: &str,
2266 
        new: &str,
2267 
        e: &FsEntry,
2268 
        expected: &crate::state::FileState,
2269 
    ) -> Result<()> {
e8e6e0ca Andy McCall ↶ 2270 
        let placement = self.base_snapshot.tree.entries[old].placement;
6e723b81 Andy McCall ↶ 2271 
        let (dir, name) = split_path(new);
2272 
        let parent = self.ensure_dir(dir);
75167672 Andy McCall 2273 
        self.tree_ops.push(TreeOp::Move {
2274 
            node: expected.node,
e8e6e0ca Andy McCall ↶ 2275 
            old: op_ref(placement),
75167672 Andy McCall 2276 
            parent,
2277 
            name: name.to_string(),
2278 
        });
e8e6e0ca Andy McCall ↶ 2279 
        for k in self.same_place(&expected.node, placement) {
2280 
            self.tree_ops.push(TreeOp::RemoveNode {
2281 
                node: expected.node,
2282 
                placement: op_ref(k),
2283 
            });
2284 
        }
6e723b81 Andy McCall ↶ 2285 
        self.update_path(repo, old, e, expected)?;
2286 
        // update_path reports under the old path; the rename is what the user sees.
2287 
        self.statuses.retain(|s| s.path != old);
75167672 Andy McCall 2288 
        self.statuses.push(PathStatus {
2289 
            path: new.to_string(),
2290 
            status: ChangeStatus::Renamed,
2291 
        });
6e723b81 Andy McCall ↶ 2292 
        Ok(())
a87c7b9d Andy McCall ↶ 2293 
    }
2294
2295 
    /// For a Git import: a node the commit keeps is in one place, the one
2296 
    /// its tree names, so any other live placement it has in the base goes;
2297 
    /// and a node the tree cannot show at all (an orphan) goes too. Such a
2298 
    /// placement is out of sight, under a directory another line of history
2299 
    /// removed, where no diff of the tree would ever reach it (oxygen#30).
2300 
    fn retire_stray_placements(&mut self) {
2301 
        let mut retired: HashSet<(ChangeHash, u32)> = HashSet::new();
2302 
        let mut placed: BTreeSet<NodeId> = BTreeSet::new();
2303 
        for op in &self.tree_ops {
2304 
            match op {
2305 
                TreeOp::Move { node, old, .. } => {
2306 
                    retired.insert((old.change, old.index));
2307 
                    placed.insert(*node);
2308 
                }
2309 
                TreeOp::RemoveNode { placement, .. } => {
2310 
                    retired.insert((placement.change, placement.index));
2311 
                }
2312 
                TreeOp::AddNode { .. } | TreeOp::SetMode { .. } => {}
2313 
            }
2314 
        }
2315 
        // Nodes kept where they were: what the base's tree view shows and
2316 
        // this change neither moves nor removes.
2317 
        let mut kept: HashMap<NodeId, (ChangeHash, u32)> = HashMap::new();
2318 
        for e in self.base_snapshot.tree.entries.values() {
2319 
            if !retired.contains(&e.placement) {
2320 
                kept.insert(e.node, e.placement);
2321 
            }
2322 
        }
2323 
        // Only a node with several live placements can have a stray one,
2324 
        // and the tree view names each as a conflict.
2325 
        let multi: BTreeSet<NodeId> = self
2326 
            .base_snapshot
2327 
            .tree
2328 
            .conflicts
2329 
            .iter()
2330 
            .filter_map(|c| match c {
2331 
                crate::tree::TreeConflict::MultiPlacement { node, .. } => Some(*node),
2332 
                _ => None,
2333 
            })
2334 
            .collect();
2335 
        let mut extra = Vec::new();
2336 
        // A node whose only placements are under a directory another line
2337 
        // removed (an orphan) is nowhere the tree can show, and Git's tree
2338 
        // has no such path: it goes.
2339 
        for c in &self.base_snapshot.tree.conflicts {
2340 
            let crate::tree::TreeConflict::Orphan { node, .. } = c else {
2341 
                continue;
2342 
            };
2343 
            if kept.contains_key(node) || placed.contains(node) {
2344 
                continue;
2345 
            }
2346 
            for (k, _) in self.base.tree().live_placements(node) {
2347 
                if retired.insert(k) {
2348 
                    extra.push(TreeOp::RemoveNode {
2349 
                        node: *node,
2350 
                        placement: op_ref(k),
2351 
                    });
2352 
                }
2353 
            }
2354 
        }
2355 
        for node in placed
2356 
            .iter()
2357 
            .chain(multi.iter().filter(|n| kept.contains_key(*n)))
2358 
        {
2359 
            for (k, _) in self.base.tree().live_placements(node) {
2360 
                if retired.contains(&k) || kept.get(node) == Some(&k) {
2361 
                    continue;
2362 
                }
2363 
                retired.insert(k);
2364 
                extra.push(TreeOp::RemoveNode {
2365 
                    node: *node,
2366 
                    placement: op_ref(k),
2367 
                });
2368 
            }
2369 
        }
2370 
        if !extra.is_empty() {
2371 
            self.resolves = true;
2372 
            self.tree_ops.extend(extra);
2373 
        }
6e723b81 Andy McCall ↶ 2374 
    }
2375
e8e6e0ca Andy McCall ↶ 2376 
    /// The node's other live placements at the same place as `k`, the one
2377 
    /// the tree shows: two lines of history that put a node back where it
2378 
    /// was each leave a placement, and the tree shows them as one file. A
2379 
    /// change that removes or moves it retires them together, or the copy
2380 
    /// left behind outlives its directory (oxygen#88).
2381 
    fn same_place(&self, node: &NodeId, k: (ChangeHash, u32)) -> Vec<(ChangeHash, u32)> {
2382 
        let tree = self.base.tree();
2383 
        let Some(shown) = tree.placement(&k) else {
2384 
            return Vec::new();
2385 
        };
2386 
        tree.live_placements(node)
2387 
            .into_iter()
2388 
            .filter(|(o, p)| *o != k && p.parent == shown.parent && p.name == shown.name)
2389 
            .map(|(o, _)| o)
2390 
            .collect()
2391 
    }
2392
98286406 Andy McCall ↶ 2393 
    fn remove_path(&mut self, path: &str, expected: &crate::state::FileState) {
2394 
        let entry = &self.base_snapshot.tree.entries[path];
e8e6e0ca Andy McCall ↶ 2395 
        let placement = entry.placement;
75167672 Andy McCall 2396 
        self.tree_ops.push(TreeOp::RemoveNode {
2397 
            node: expected.node,
e8e6e0ca Andy McCall ↶ 2398 
            placement: op_ref(placement),
75167672 Andy McCall 2399 
        });
e8e6e0ca Andy McCall ↶ 2400 
        for k in self.same_place(&expected.node, placement) {
2401 
            self.tree_ops.push(TreeOp::RemoveNode {
2402 
                node: expected.node,
2403 
                placement: op_ref(k),
2404 
            });
2405 
        }
98286406 Andy McCall ↶ 2406 
        if path.contains(".oxy-conflict-") || expected.is_conflicted() {
2407 
            self.resolves = true;
2408 
        }
75167672 Andy McCall 2409 
        self.statuses.push(PathStatus {
2410 
            path: path.to_string(),
2411 
            status: ChangeStatus::Removed,
2412 
        });
98286406 Andy McCall ↶ 2413 
    }
2414
75167672 Andy McCall 2415 
    fn update_path(
2416 
        &mut self,
2417 
        repo: &Repo,
2418 
        path: &str,
2419 
        e: &FsEntry,
2420 
        expected: &crate::state::FileState,
2421 
    ) -> Result<()> {
98286406 Andy McCall ↶ 2422 
        if expected.kind != e.kind {
2423 
            self.remove_path(path, expected);
2424 
            self.statuses.pop();
2425 
            self.add_path(repo, path, e)?;
2426 
            self.statuses.pop();
75167672 Andy McCall 2427 
            self.statuses.push(PathStatus {
2428 
                path: path.to_string(),
2429 
                status: ChangeStatus::Modified,
2430 
            });
98286406 Andy McCall ↶ 2431 
            return Ok(());
2432 
        }
2433 
        let entry = &self.base_snapshot.tree.entries[path];
2434 
        let node = expected.node;
2435 
        let mode_atom = entry.mode_atom.map(op_ref);
cc7a0056 Andy McCall ↶ 2436 
        if e.unchanged {
2437 
            // Known equal to the base's unconflicted rendering (spec
2438 
            // §12.3); only the executable bit can differ.
2439 
            if e.exec != expected.mode.exec {
2440 
                if let Some(old) = mode_atom {
2441 
                    self.tree_ops.push(TreeOp::SetMode {
2442 
                        node,
2443 
                        old,
2444 
                        mode: Mode {
2445 
                            exec: e.exec,
2446 
                            opaque: false,
2447 
                        },
2448 
                    });
2449 
                    self.statuses.push(PathStatus {
2450 
                        path: path.to_string(),
2451 
                        status: ChangeStatus::ModeChanged,
2452 
                    });
2453 
                }
2454 
            }
2455 
            return Ok(());
2456 
        }
98286406 Andy McCall ↶ 2457 
        let mut status: Option<ChangeStatus> = None;
cdb5beba Andy McCall ↶ 2458 
        let wants_opaque = self.opaque_for(path, e);
2459 
        let nocompress = self.attrs.lookup(path).nocompress;
2460 
        let disk_text = self.stored_text(path, &e.content);
98286406 Andy McCall ↶ 2461 
        match &expected.content {
2462 
            FileContent::Dir => {}
75167672 Andy McCall 2463 
            FileContent::Text {
2464 
                bytes,
2465 
                materialized,
2466 
            } => {
98286406 Andy McCall ↶ 2467 
                if materialized.is_conflicted() {
40a40bb7 Andy McCall ↶ 2468 
                    if bytes == &*disk_text {
98286406 Andy McCall ↶ 2469 
                        return Ok(());
2470 
                    }
2471 
                    if has_conflict_markers(&e.content) {
75167672 Andy McCall 2472 
                        self.statuses.push(PathStatus {
2473 
                            path: path.to_string(),
2474 
                            status: ChangeStatus::ConflictEdited,
2475 
                        });
98286406 Andy McCall ↶ 2476 
                        return Ok(());
2477 
                    }
2478 
                    // Resolution (spec §9.6): diff against the unconflicted lines
2479 
                    // with each region as a gap, then retire every region atom.
2480 
                    // Whatever the author kept from a region is re-inserted as
2481 
                    // new atoms, which leaves the graph totally ordered.
2482 
                    let regions = self.base.conflict_regions(&node);
cdb5beba Andy McCall ↶ 2483 
                    let mut edits = text_edits(node, &regions.plain, &disk_text);
98286406 Andy McCall ↶ 2484 
                    if !regions.region_atoms.is_empty() {
75167672 Andy McCall 2485 
                        edits.insert(
2486 
                            0,
2487 
                            Edit::Delete {
2488 
                                node,
2489 
                                atoms: ranges(&regions.region_atoms),
2490 
                            },
2491 
                        );
98286406 Andy McCall ↶ 2492 
                    }
2493 
                    self.edits.extend(edits);
2494 
                    self.contributors.extend(regions.contributors);
2495 
                    self.resolves = true;
2496 
                    if e.exec != expected.mode.exec {
2497 
                        if let Some(old) = mode_atom {
75167672 Andy McCall 2498 
                            self.tree_ops.push(TreeOp::SetMode {
2499 
                                node,
2500 
                                old,
2501 
                                mode: Mode {
2502 
                                    exec: e.exec,
2503 
                                    opaque: false,
2504 
                                },
2505 
                            });
98286406 Andy McCall ↶ 2506 
                        }
2507 
                    }
75167672 Andy McCall 2508 
                    self.statuses.push(PathStatus {
2509 
                        path: path.to_string(),
2510 
                        status: ChangeStatus::Resolved,
2511 
                    });
98286406 Andy McCall ↶ 2512 
                    return Ok(());
2513 
                }
cdb5beba Andy McCall ↶ 2514 
                if wants_opaque {
2515 
                    // Text → opaque (spec §7.5), by content or by attribute.
2516 
                    if let Some(old) = mode_atom {
75167672 Andy McCall 2517 
                        self.tree_ops.push(TreeOp::SetMode {
2518 
                            node,
2519 
                            old,
2520 
                            mode: Mode {
2521 
                                exec: e.exec,
2522 
                                opaque: true,
2523 
                            },
2524 
                        });
cdb5beba Andy McCall ↶ 2525 
                    }
2526 
                    let live: Vec<_> = repo_live_keys(self.base, &node);
2527 
                    if !live.is_empty() {
75167672 Andy McCall 2528 
                        self.edits.push(Edit::Delete {
2529 
                            node,
2530 
                            atoms: ranges(&live),
2531 
                        });
cdb5beba Andy McCall ↶ 2532 
                    }
b7f102fc Andy McCall ↶ 2533 
                    let blob = self.blob_of(repo, path, e, nocompress)?;
75167672 Andy McCall 2534 
                    self.edits.push(Edit::Replace {
2535 
                        node,
2536 
                        old: None,
2537 
                        new: Some(blob),
2538 
                    });
cdb5beba Andy McCall ↶ 2539 
                    status = Some(ChangeStatus::Modified);
40a40bb7 Andy McCall ↶ 2540 
                } else if bytes != &*disk_text {
2306a4d2 Andy McCall ↶ 2541 
                    let mut laps = crate::timing::Laps::start();
2542 
                    // The base's rendering holds its live atoms in order,
2543 
                    // one per line (it is not conflicted here).
360aa74a Andy McCall ↶ 2544 
                    let edits = match self.diffs.remove(path) {
2545 
                        Some(edits) => edits,
2546 
                        None => {
2547 
                            let keys = materialized.live_atoms();
2548 
                            let lines = crate::diff::split_lines(bytes);
2549 
                            if keys.len() == lines.len() {
2550 
                                crate::diff::text_edits_of(node, &keys, &lines, &disk_text)
2551 
                            } else {
2552 
                                text_edits(node, &self.base.live_lines(&node), &disk_text)
2553 
                            }
2554 
                        }
2306a4d2 Andy McCall ↶ 2555 
                    };
2556 
                    laps.lap("build: text diff");
cdb5beba Andy McCall ↶ 2557 
                    if !edits.is_empty() {
2558 
                        self.edits.extend(edits);
98286406 Andy McCall ↶ 2559 
                        status = Some(ChangeStatus::Modified);
2560 
                    }
2561 
                }
cdb5beba Andy McCall ↶ 2562 
                let switched_to_opaque = wants_opaque;
98286406 Andy McCall ↶ 2563 
                if !switched_to_opaque && e.exec != expected.mode.exec {
2564 
                    if let Some(old) = mode_atom {
75167672 Andy McCall 2565 
                        self.tree_ops.push(TreeOp::SetMode {
2566 
                            node,
2567 
                            old,
2568 
                            mode: Mode {
2569 
                                exec: e.exec,
2570 
                                opaque: false,
2571 
                            },
2572 
                        });
98286406 Andy McCall ↶ 2573 
                        status.get_or_insert(ChangeStatus::ModeChanged);
2574 
                    }
2575 
                }
2576 
            }
2577 
            FileContent::Opaque { blobs } => {
b7f102fc Andy McCall ↶ 2578 
                let new_ref = e.blob.unwrap_or_else(|| Repo::blob_ref(&e.content));
2579 
                if e.kind == NodeKind::File && wants_opaque {
2580 
                    // Unchanged or replaced, the file stays opaque: remember its blob.
2581 
                    self.blobs.insert(path.to_string(), new_ref);
f7510b63 Andy McCall ↶ 2582 
                }
2583 
                if blobs.len() > 1 && self.resolve.contains(path) && e.kind == expected.kind {
2584 
                    // Settle the replace conflict (spec §9.3): replace every
2585 
                    // live candidate with the chosen blob, depending on each
2586 
                    // change that made one live, so all are superseded.
2587 
                    let chosen = self.blob_of(repo, path, e, nocompress)?;
2588 
                    for b in blobs {
2589 
                        self.edits.push(Edit::Replace {
2590 
                            node,
2591 
                            old: Some(*b),
2592 
                            new: Some(chosen),
2593 
                        });
2594 
                    }
2595 
                    self.contributors
2596 
                        .extend(self.base.blob_contributors(&node, blobs));
2597 
                    self.resolves = true;
2598 
                    self.statuses.push(PathStatus {
2599 
                        path: path.to_string(),
2600 
                        status: ChangeStatus::Resolved,
2601 
                    });
2602 
                    return Ok(());
b7f102fc Andy McCall ↶ 2603 
                }
98286406 Andy McCall ↶ 2604 
                if blobs.len() > 1 {
2605 
                    if !blobs.contains(&new_ref) {
75167672 Andy McCall 2606 
                        self.statuses.push(PathStatus {
2607 
                            path: path.to_string(),
2608 
                            status: ChangeStatus::ConflictEdited,
2609 
                        });
98286406 Andy McCall ↶ 2610 
                    }
2611 
                    return Ok(());
2612 
                }
2613 
                let current = blobs.first().copied();
2614 
                if current != Some(new_ref) {
cdb5beba Andy McCall ↶ 2615 
                    if e.kind == NodeKind::File && !wants_opaque {
98286406 Andy McCall ↶ 2616 
                        // Opaque → text (spec §7.5).
2617 
                        if let Some(old) = mode_atom {
75167672 Andy McCall 2618 
                            self.tree_ops.push(TreeOp::SetMode {
2619 
                                node,
2620 
                                old,
2621 
                                mode: Mode {
2622 
                                    exec: e.exec,
2623 
                                    opaque: false,
2624 
                                },
2625 
                            });
98286406 Andy McCall ↶ 2626 
                        }
75167672 Andy McCall 2627 
                        self.edits.push(Edit::Replace {
2628 
                            node,
2629 
                            old: current,
2630 
                            new: None,
2631 
                        });
98286406 Andy McCall ↶ 2632 
                        if !e.content.is_empty() {
2633 
                            self.edits.push(Edit::Insert {
2634 
                                node,
2635 
                                after: crate::objects::AtomRef::Start,
2636 
                                before: crate::objects::AtomRef::End,
01a47cd7 Andy McCall ↶ 2637 
                                lines: crate::lines::Lines::split(&e.content),
98286406 Andy McCall ↶ 2638 
                            });
2639 
                        }
2640 
                    } else {
b7f102fc Andy McCall ↶ 2641 
                        let blob = self.blob_of(repo, path, e, nocompress)?;
75167672 Andy McCall 2642 
                        self.edits.push(Edit::Replace {
2643 
                            node,
2644 
                            old: current,
2645 
                            new: Some(blob),
2646 
                        });
6eae06b6 Andy McCall ↶ 2647 
                        // New contents and a new executable bit at once: the
2648 
                        // bit was dropped (VS Code's inno_updater.exe,
2649 
                        // 2018-02-28, lost +x for 49 commits; oxygen#87).
2650 
                        if e.kind == NodeKind::File && e.exec != expected.mode.exec {
2651 
                            if let Some(old) = mode_atom {
2652 
                                self.tree_ops.push(TreeOp::SetMode {
2653 
                                    node,
2654 
                                    old,
2655 
                                    mode: Mode {
2656 
                                        exec: e.exec,
2657 
                                        opaque: true,
2658 
                                    },
2659 
                                });
2660 
                            }
2661 
                        }
98286406 Andy McCall ↶ 2662 
                    }
2663 
                    status = Some(ChangeStatus::Modified);
2664 
                } else if e.kind == NodeKind::File && e.exec != expected.mode.exec {
2665 
                    if let Some(old) = mode_atom {
75167672 Andy McCall 2666 
                        self.tree_ops.push(TreeOp::SetMode {
2667 
                            node,
2668 
                            old,
2669 
                            mode: Mode {
2670 
                                exec: e.exec,
2671 
                                opaque: true,
2672 
                            },
2673 
                        });
98286406 Andy McCall ↶ 2674 
                        status = Some(ChangeStatus::ModeChanged);
2675 
                    }
2676 
                }
2677 
            }
2678 
        }
2679 
        if let Some(s) = status {
75167672 Andy McCall 2680 
            self.statuses.push(PathStatus {
2681 
                path: path.to_string(),
2682 
                status: s,
2683 
            });
98286406 Andy McCall ↶ 2684 
        }
2685 
        Ok(())
2686 
    }
2687 
}
2688
2689 
fn repo_live_keys(base: &State, node: &NodeId) -> Vec<crate::graph::AtomKey> {
2690 
    base.live_lines(node).into_iter().map(|(k, _)| k).collect()
2691 
}
2692
2693 
/// The changes a change's scope references (spec §8.1): introducers of every
2694 
/// atom, placement, and mode it touches, and of every node it operates on.
2695 
pub fn referenced_changes(change: &Change, base: &State) -> BTreeSet<ChangeHash> {
2696 
    let mut out = BTreeSet::new();
2697 
    let opref = |r: &OpRef, out: &mut BTreeSet<ChangeHash>| {
2698 
        if !r.change.is_zero() {
2699 
            out.insert(r.change);
2700 
        }
2701 
    };
2702 
    let mut nodes: Vec<NodeId> = Vec::new();
2703 
    for op in &change.tree_ops {
2704 
        match op {
2705 
            TreeOp::AddNode { parent, .. } => nodes.push(*parent),
2706 
            TreeOp::RemoveNode { node, placement } => {
2707 
                nodes.push(*node);
2708 
                opref(placement, &mut out);
2709 
            }
75167672 Andy McCall 2710 
            TreeOp::Move {
2711 
                node, old, parent, ..
2712 
            } => {
98286406 Andy McCall ↶ 2713 
                nodes.push(*node);
2714 
                nodes.push(*parent);
2715 
                opref(old, &mut out);
2716 
            }
2717 
            TreeOp::SetMode { node, old, .. } => {
2718 
                nodes.push(*node);
2719 
                opref(old, &mut out);
2720 
            }
2721 
        }
2722 
    }
2723 
    for e in &change.edits {
2724 
        nodes.push(e.node());
2725 
        match e {
2726 
            Edit::Insert { after, before, .. } => {
2727 
                for r in [after, before] {
2728 
                    if let crate::objects::AtomRef::Atom { change, .. } = r {
2729 
                        if !change.is_zero() {
2730 
                            out.insert(*change);
2731 
                        }
2732 
                    }
2733 
                }
2734 
            }
2735 
            Edit::Delete { atoms, .. } => {
62794ecc Andy McCall ↶ 2736 
                // Deleting an atom also depends on whoever inserted the live
2737 
                // atoms right after it: the author saw those insertions, and
2738 
                // without the dependency spec §9.1 would call the deletion
2739 
                // concurrent with them (a delete/insert conflict).
2740 
                let graph = base.file(&e.node());
98286406 Andy McCall ↶ 2741 
                for r in atoms {
2742 
                    if !r.change.is_zero() {
2743 
                        out.insert(r.change);
2744 
                    }
62794ecc Andy McCall ↶ 2745 
                    if let Some(g) = graph {
2746 
                        for i in r.from..r.to {
2747 
                            out.extend(
2748 
                                g.live_successor_changes(&(r.change, i))
2749 
                                    .into_iter()
2750 
                                    .filter(|c| !c.is_zero()),
2751 
                            );
2752 
                        }
2753 
                    }
98286406 Andy McCall ↶ 2754 
                }
2755 
            }
62794ecc Andy McCall ↶ 2756 
            // Replacing a blob depends on the change that made it current,
2757 
            // as deleting a line depends on the line's introducer.
2758 
            Edit::Replace {
2759 
                node,
2760 
                old: Some(old),
2761 
                ..
2762 
            } => {
2763 
                out.extend(
2764 
                    base.live_replaces(node)
2765 
                        .into_iter()
2766 
                        .filter(|(c, b)| b == old && !c.is_zero())
2767 
                        .map(|(c, _)| c),
2768 
                );
2769 
            }
98286406 Andy McCall ↶ 2770 
            Edit::Replace { .. } => {}
2771 
        }
2772 
    }
2773 
    for n in nodes {
2774 
        if let Some(h) = base.tree().introducer(&n) {
2775 
            out.insert(h);
2776 
        }
2777 
    }
2778 
    out
2779 
}
2780
2781 
/// Paths of the nodes a change adds on top of `base_snapshot`.
75167672 Andy McCall 2782 
pub fn added_node_paths(
2783 
    change: &Change,
2784 
    base_snapshot: &Snapshot,
2785 
) -> HashMap<String, (NodeId, NodeKind)> {
98286406 Andy McCall ↶ 2786 
    let by_node = base_snapshot.tree.by_node();
2787 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
2788 
    for op in &change.tree_ops {
75167672 Andy McCall 2789 
        if let TreeOp::AddNode {
2790 
            node,
2791 
            parent,
2792 
            name,
2793 
            kind,
2794 
            ..
2795 
        } = op
2796 
        {
98286406 Andy McCall ↶ 2797 
            new_names.insert(*node, (*parent, name.clone(), *kind));
2798 
        }
2799 
    }
2800 
    let mut out = HashMap::new();
2801 
    for (node, (_, _, kind)) in &new_names {
2802 
        if let Some(path) = path_of_node(node, &by_node, &new_names) {
2803 
            out.insert(path, (*node, *kind));
2804 
        }
2805 
    }
2806 
    out
2807 
}
2808
2809 
fn path_of_node(
2810 
    n: &NodeId,
2811 
    by_node: &HashMap<NodeId, &crate::tree::TreeEntry>,
2812 
    new_names: &HashMap<NodeId, (NodeId, String, NodeKind)>,
2813 
) -> Option<String> {
2814 
    if let Some(e) = by_node.get(n) {
2815 
        return Some(e.path.clone());
2816 
    }
2817 
    let mut parts = Vec::new();
2818 
    let mut cur = *n;
2819 
    loop {
2820 
        if cur.is_root() {
2821 
            break;
2822 
        }
2823 
        if let Some(e) = by_node.get(&cur) {
2824 
            if !e.path.is_empty() {
2825 
                parts.push(e.path.clone());
2826 
            }
2827 
            break;
2828 
        }
2829 
        let (parent, name, _) = new_names.get(&cur)?;
2830 
        parts.push(name.clone());
2831 
        cur = *parent;
2832 
    }
2833 
    parts.reverse();
2834 
    Some(parts.join("/"))
2835 
}
2836
2837 
/// Render a change's operations with line content, using `base` for deleted lines.
2838 
pub fn render_change(change: &Change, base: &State, base_snapshot: &Snapshot) -> String {
2839 
    let by_node = base_snapshot.tree.by_node();
2840 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
2841 
    for op in &change.tree_ops {
75167672 Andy McCall 2842 
        if let TreeOp::AddNode {
2843 
            node,
2844 
            parent,
2845 
            name,
2846 
            kind,
2847 
            ..
2848 
        } = op
2849 
        {
98286406 Andy McCall ↶ 2850 
            new_names.insert(*node, (*parent, name.clone(), *kind));
2851 
        }
2852 
    }
75167672 Andy McCall 2853 
    let path_of = |n: &NodeId| -> String {
2854 
        path_of_node(n, &by_node, &new_names).unwrap_or_else(|| format!("<node {}>", n.short(8)))
2855 
    };
98286406 Andy McCall ↶ 2856
2857 
    let mut out = String::new();
2858 
    for op in &change.tree_ops {
2859 
        match op {
75167672 Andy McCall 2860 
            TreeOp::AddNode {
2861 
                node, kind, mode, ..
2862 
            } => {
2863 
                out.push_str(&format!(
2864 
                    "A {} ({:?}{}{})\n",
2865 
                    path_of(node),
2866 
                    kind,
2867 
                    if mode.exec { ", exec" } else { "" },
2868 
                    if mode.opaque { ", opaque" } else { "" }
2869 
                ));
98286406 Andy McCall ↶ 2870 
            }
2871 
            TreeOp::RemoveNode { node, .. } => out.push_str(&format!("D {}\n", path_of(node))),
75167672 Andy McCall 2872 
            TreeOp::Move { node, name, .. } => {
2873 
                out.push_str(&format!("R {} -> {}\n", path_of(node), name))
2874 
            }
98286406 Andy McCall ↶ 2875 
            TreeOp::SetMode { node, mode, .. } => {
75167672 Andy McCall 2876 
                out.push_str(&format!(
2877 
                    "M {} mode: exec={} opaque={}\n",
2878 
                    path_of(node),
2879 
                    mode.exec,
2880 
                    mode.opaque
2881 
                ));
98286406 Andy McCall ↶ 2882 
            }
2883 
        }
2884 
    }
2885 
    let mut last_node: Option<NodeId> = None;
2886 
    for e in &change.edits {
2887 
        let n = e.node();
2888 
        if last_node != Some(n) {
2889 
            out.push_str(&format!("--- {}\n", path_of(&n)));
2890 
            last_node = Some(n);
2891 
        }
2892 
        match e {
2893 
            Edit::Insert { lines, .. } => {
01a47cd7 Andy McCall ↶ 2894 
                for l in lines.iter() {
98286406 Andy McCall ↶ 2895 
                    out.push('+');
2896 
                    out.push_str(&String::from_utf8_lossy(l));
2897 
                    if !l.ends_with(b"\n") {
2898 
                        out.push('\n');
2899 
                    }
2900 
                }
2901 
            }
2902 
            Edit::Delete { atoms, .. } => {
2903 
                for r in atoms {
2904 
                    for i in r.from..r.to {
75167672 Andy McCall 2905 
                        let content = base
2906 
                            .file(&n)
2907 
                            .and_then(|g| g.content(&(r.change, i)))
2908 
                            .unwrap_or(b"<unknown>\n");
98286406 Andy McCall ↶ 2909 
                        out.push('-');
2910 
                        out.push_str(&String::from_utf8_lossy(content));
2911 
                        if !content.ends_with(b"\n") {
2912 
                            out.push('\n');
2913 
                        }
2914 
                    }
2915 
                }
2916 
            }
2917 
            Edit::Replace { old, new, .. } => {
2918 
                out.push_str(&format!(
2919 
                    "  opaque content {} -> {}\n",
2920 
                    old.map(|o| o.short(12)).unwrap_or_else(|| "none".into()),
2921 
                    new.map(|o| o.short(12)).unwrap_or_else(|| "none".into())
2922 
                ));
2923 
            }
2924 
        }
2925 
    }
2926 
    out
2927 
}
6e723b81 Andy McCall ↶ 2928
2929 
/// Pair removed files with added files of the same kind whose contents are
2930 
/// at least `RENAME_THRESHOLD` percent alike, best matches first. Beyond
2931 
/// `RENAME_LIMIT` candidate pairs only identical contents are paired.
2932 
const RENAME_THRESHOLD: u32 = 50;
2933 
const RENAME_LIMIT: usize = 10_000;
2306a4d2 Andy McCall ↶ 2934
23ff1288 Andy McCall ↶ 2935 
/// How a change is built beyond its entries: for a Git import, the seed
2936 
/// its new nodes' identifiers derive from; and, when it is built from part
2937 
/// of the tree, whether the new tree has anything under a directory.
2938 
#[derive(Default)]
2939 
struct Scope<'a> {
2940 
    seed: Option<[u8; 32]>,
2306a4d2 Andy McCall ↶ 2941 
    dir_needed: Option<&'a dyn Fn(&str) -> bool>,
2942 
}
6e723b81 Andy McCall ↶ 2943
75167672 Andy McCall 2944 
fn detect_renames(
2945 
    base_snapshot: &Snapshot,
2946 
    fs_entries: &BTreeMap<String, FsEntry>,
2947 
    in_scope: impl Fn(&str) -> bool,
2948 
) -> Vec<(String, String)> {
6e723b81 Andy McCall ↶ 2949 
    let removed: Vec<(&String, &crate::state::FileState)> = base_snapshot
2950 
        .files
2951 
        .iter()
75167672 Andy McCall 2952 
        .filter(|(p, f)| {
2953 
            f.kind != NodeKind::Dir
2954 
                && !fs_entries.contains_key(*p)
2955 
                && in_scope(p)
2956 
                && !f.is_conflicted()
2957 
        })
2958 
        .collect();
2959 
    let added: Vec<(&String, &FsEntry)> = fs_entries
2960 
        .iter()
2961 
        .filter(|(p, e)| e.kind != NodeKind::Dir && !base_snapshot.files.contains_key(*p))
6e723b81 Andy McCall ↶ 2962 
        .collect();
2963 
    if removed.is_empty() || added.is_empty() {
2964 
        return vec![];
2965 
    }
2966 
    let exact_only = removed.len() * added.len() > RENAME_LIMIT;
2967 
    let old_content = |f: &crate::state::FileState| -> Option<Vec<u8>> {
2968 
        match &f.content {
2969 
            FileContent::Text { bytes, .. } => Some(bytes.clone()),
2970 
            _ => None,
2971 
        }
2972 
    };
2973 
    let mut scored: Vec<(u32, usize, usize)> = Vec::new();
2974 
    for (ri, (_, f)) in removed.iter().enumerate() {
2975 
        let old = old_content(f);
2976 
        let old_blob = match &f.content {
2977 
            FileContent::Opaque { blobs } if blobs.len() == 1 => Some(blobs[0]),
2978 
            _ => None,
2979 
        };
2980 
        for (ai, (_, e)) in added.iter().enumerate() {
2981 
            if e.kind != f.kind {
2982 
                continue;
2983 
            }
2984 
            let score = match (&old, old_blob) {
2985 
                (Some(o), _) => {
2986 
                    if exact_only {
40a40bb7 Andy McCall ↶ 2987 
                        if *o == *e.content {
75167672 Andy McCall 2988 
                            100
2989 
                        } else {
2990 
                            0
2991 
                        }
6e723b81 Andy McCall ↶ 2992 
                    } else {
2993 
                        similarity(o, &e.content)
2994 
                    }
2995 
                }
b7f102fc Andy McCall ↶ 2996 
                (None, Some(b)) if e.blob.unwrap_or_else(|| Repo::blob_ref(&e.content)) == b => 100,
6e723b81 Andy McCall ↶ 2997 
                _ => 0,
2998 
            };
2999 
            if score >= RENAME_THRESHOLD {
3000 
                scored.push((score, ri, ai));
3001 
            }
3002 
        }
3003 
    }
75167672 Andy McCall 3004 
    scored.sort_by(|a, b| {
3005 
        b.0.cmp(&a.0)
3006 
            .then(removed[a.1].0.cmp(removed[b.1].0))
3007 
            .then(added[a.2].0.cmp(added[b.2].0))
3008 
    });
6e723b81 Andy McCall ↶ 3009 
    let mut used_r = vec![false; removed.len()];
3010 
    let mut used_a = vec![false; added.len()];
3011 
    let mut out = Vec::new();
3012 
    for (_, ri, ai) in scored {
3013 
        if used_r[ri] || used_a[ai] {
3014 
            continue;
3015 
        }
3016 
        used_r[ri] = true;
3017 
        used_a[ai] = true;
3018 
        out.push((removed[ri].0.clone(), added[ai].0.clone()));
3019 
    }
3020 
    out
3021 
}
3022
3023 
/// Percentage of the larger file's bytes found in lines the two share
3024 
/// (multiset intersection of lines), like Git's similarity index.
3025 
pub fn similarity(a: &[u8], b: &[u8]) -> u32 {
3026 
    if a == b {
3027 
        return 100;
3028 
    }
3029 
    if a.is_empty() || b.is_empty() {
3030 
        return 0;
3031 
    }
3032 
    let mut counts: HashMap<&[u8], usize> = HashMap::new();
3033 
    for l in split_lines(a) {
3034 
        *counts.entry(l).or_default() += 1;
3035 
    }
3036 
    let mut shared = 0usize;
3037 
    for l in split_lines(b) {
3038 
        if let Some(n) = counts.get_mut(l) {
3039 
            if *n > 0 {
3040 
                *n -= 1;
3041 
                shared += l.len();
3042 
            }
3043 
        }
3044 
    }
3045 
    (shared * 100 / a.len().max(b.len())) as u32
d616faf5 Andy McCall ↶ 3046 
}
3047
3048 
/// Two maps side by side in key order: each key of either, with its value
3049 
/// in each. Linear, with no lookups, for maps of a whole tree.
3050 
fn join<'a, A, B>(
3051 
    a: &'a BTreeMap<String, A>,
3052 
    b: &'a BTreeMap<String, B>,
3053 
) -> impl Iterator<Item = (&'a String, Option<&'a A>, Option<&'a B>)> {
3054 
    let (mut a, mut b) = (a.iter().peekable(), b.iter().peekable());
3055 
    std::iter::from_fn(move || match (a.peek(), b.peek()) {
3056 
        (None, None) => None,
3057 
        (Some(_), None) => a.next().map(|(k, v)| (k, Some(v), None)),
3058 
        (None, Some(_)) => b.next().map(|(k, v)| (k, None, Some(v))),
3059 
        (Some((ka, _)), Some((kb, _))) => match ka.cmp(kb) {
3060 
            std::cmp::Ordering::Less => a.next().map(|(k, v)| (k, Some(v), None)),
3061 
            std::cmp::Ordering::Greater => b.next().map(|(k, v)| (k, None, Some(v))),
3062 
            std::cmp::Ordering::Equal => {
3063 
                let (k, va) = a.next()?;
3064 
                let (_, vb) = b.next()?;
3065 
                Some((k, Some(va), Some(vb)))
3066 
            }
3067 
        },
3068 
    })
3069 
}
3070
3071 
#[cfg(test)]
3072 
mod join_tests {
3073 
    use super::*;
3074
3075 
    #[test]
3076 
    fn join_pairs_every_key_in_order() {
3077 
        let a: BTreeMap<String, u8> = [("a", 1), ("b/c", 2), ("d", 3)]
3078 
            .into_iter()
3079 
            .map(|(k, v)| (k.to_string(), v))
3080 
            .collect();
3081 
        let b: BTreeMap<String, char> = [("b", 'x'), ("b/c", 'y'), ("e", 'z')]
3082 
            .into_iter()
3083 
            .map(|(k, v)| (k.to_string(), v))
3084 
            .collect();
3085 
        let got: Vec<(String, Option<u8>, Option<char>)> = join(&a, &b)
3086 
            .map(|(k, x, y)| (k.clone(), x.copied(), y.copied()))
3087 
            .collect();
3088 
        let want = vec![
3089 
            ("a".to_string(), Some(1), None),
3090 
            ("b".to_string(), None, Some('x')),
3091 
            ("b/c".to_string(), Some(2), Some('y')),
3092 
            ("d".to_string(), Some(3), None),
3093 
            ("e".to_string(), None, Some('z')),
3094 
        ];
3095 
        assert_eq!(got, want);
3096 
    }
6e723b81 Andy McCall ↶ 3097 
}