OXYGENHUB
◆ v0.6.0 oxygen / crates / oxy-core / src / repo / snapshot.rs
125.5 KiB · Rust · last changed Marker-like text no longer blocks a conflict resolution 17 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
3652104a Andy McCall ↶ 29 
/// Whether `content` still holds a conflict marker oxy wrote for this file
30 
/// (oxygen#126): a marker-like line of the materialized conflicted file
31 
/// that is not one of its unconflicted lines (`plain`). Text that merely
32 
/// looks like a marker, as oxy's own documentation or a test fixture has,
33 
/// is in the unconflicted lines, so it is not taken for one.
34 
pub fn has_own_conflict_markers(content: &[u8], materialized: &[u8], plain: &[&[u8]]) -> bool {
35 
    let line = |l: &[u8]| -> Vec<u8> {
36 
        let l = l.strip_suffix(b"\n").unwrap_or(l);
37 
        l.strip_suffix(b"\r").unwrap_or(l).to_vec()
38 
    };
39 
    let marker = |l: &[u8]| {
40 
        l.starts_with(b"<<<<<<< oxy conflict")
41 
            || l.starts_with(b"======= oxy conflict")
42 
            || l.starts_with(b">>>>>>> oxy conflict")
43 
    };
44 
    let unconflicted: HashSet<Vec<u8>> = plain.iter().map(|l| line(l)).collect();
45 
    let own: HashSet<Vec<u8>> = split_lines(materialized)
46 
        .into_iter()
47 
        .map(line)
48 
        .filter(|l| marker(l) && !unconflicted.contains(l))
49 
        .collect();
50 
    !own.is_empty()
51 
        && split_lines(content)
52 
            .into_iter()
53 
            .any(|l| own.contains(&line(l)))
54 
}
55
98286406 Andy McCall ↶ 56 
#[derive(Clone, Debug, PartialEq, Eq)]
57 
pub struct PathStatus {
58 
    pub path: String,
59 
    pub status: ChangeStatus,
60 
}
61
614541c3 Andy McCall ↶ 62 
/// Operations of a change built from the working directory against a base.
63 
pub(crate) struct BuiltChange {
64 
    pub tree_ops: Vec<TreeOp>,
65 
    pub edits: Vec<Edit>,
66 
    pub contributors: Vec<ChangeHash>,
67 
    pub resolves: bool,
68 
    pub statuses: Vec<PathStatus>,
b7f102fc Andy McCall ↶ 69 
    /// The stored blob of every opaque path, for the manifest's stat cache.
70 
    pub blobs: HashMap<String, ObjectRef>,
614541c3 Andy McCall ↶ 71 
}
72
98286406 Andy McCall ↶ 73 
/// Result of snapshotting the working copy.
74 
pub struct SnapshotResult {
5b4c5130 Andy McCall ↶ 75 
    /// The working-copy change's hash; for a commit's own snapshot, which
76 
    /// does not encode it, the hash it had before.
98286406 Andy McCall ↶ 77 
    pub wc_hash: ChangeHash,
78 
    pub wc: Change,
79 
    /// The base state (closure of the working-copy change's dependencies).
80 
    pub base: State,
81 
    pub base_snapshot: Snapshot,
82 
    pub statuses: Vec<PathStatus>,
83 
    /// Whether the snapshot rewrote the working-copy change.
84 
    pub changed: bool,
85 
    /// Whether the working copy resolves conflicts of its base (spec §9.6).
86 
    pub resolves: bool,
87 
    /// Changes that contributed to the conflicts being resolved.
88 
    pub contributors: Vec<ChangeHash>,
f9466021 Andy McCall ↶ 89 
    /// The `.oxyattributes` in force on disk, from the files the walk read.
90 
    pub(crate) attributes: crate::attributes::Attributes,
d4dbab97 Andy McCall ↶ 91 
    /// Folders holding a repository of their own (`.git` or `.oxy`) that
92 
    /// are not subrepos, so nothing in them is tracked (oxygen#43).
93 
    pub untracked_nested: Vec<String>,
236f35c0 Andy McCall ↶ 94 
    /// A commit's snapshot leaves the manifest to the commit, which writes
95 
    /// it once with the new base's graph keys.
96 
    pub(crate) manifest: Option<PendingManifest>,
98286406 Andy McCall ↶ 97 
}
98
236f35c0 Andy McCall ↶ 99 
/// The manifest a commit's snapshot would have written, as of the moment
100 
/// it would have written it (that moment is the racy-window stamp).
101 
pub(crate) struct PendingManifest {
102 
    entries: BTreeMap<String, FsEntry>,
103 
    blobs: HashMap<String, ObjectRef>,
104 
    dirs: Option<super::WalkedDirs>,
105 
    written: u128,
106 
}
107
98286406 Andy McCall ↶ 108 
pub fn is_binary(content: &[u8]) -> bool {
109 
    content[..content.len().min(BINARY_SNIFF)].contains(&0)
110 
}
111
112 
impl Repo {
614541c3 Andy McCall ↶ 113 
    /// Diff the files on disk against `base` and build the operations of a
114 
    /// change from `base` to the working directory. `reuse` maps paths to the
115 
    /// NodeIds an earlier version of the same change assigned to new paths.
75167672 Andy McCall 116 
    pub(crate) fn build_against(
117 
        &self,
118 
        base: &State,
119 
        base_snapshot: &Snapshot,
120 
        reuse: HashMap<String, (NodeId, NodeKind)>,
121 
    ) -> Result<BuiltChange> {
c863c7a7 Andy McCall ↶ 122 
        let mut fs_entries =
696bbf29 Andy McCall ↶ 123 
            self.walk_working_copy_with(None, &Self::tracked_kinds(base_snapshot), None)?;
c863c7a7 Andy McCall ↶ 124 
        self.platform_normalize(&mut fs_entries, base_snapshot);
c3069537 Andy McCall ↶ 125 
        self.build_from_entries(base, base_snapshot, reuse, &fs_entries)
126 
    }
127
c863c7a7 Andy McCall ↶ 128 
    /// Read the working copy as the platform left it (spec §7.7): without
129 
    /// executable bits a file keeps the bit it has in the base, and without
130 
    /// symbolic links a plain file where the base has a symlink is that
131 
    /// symlink, written out as its target, not a new file.
132 
    fn platform_normalize(&self, entries: &mut BTreeMap<String, FsEntry>, base: &Snapshot) {
133 
        let p = self.platform;
134 
        if p.exec_bits && p.symlinks {
135 
            return;
136 
        }
137 
        for (path, e) in entries.iter_mut() {
138 
            let Some(f) = base.files.get(path) else {
139 
                continue;
140 
            };
141 
            if !p.symlinks && f.kind == NodeKind::Symlink && e.kind == NodeKind::File {
142 
                e.kind = NodeKind::Symlink;
143 
                e.exec = false;
144 
            }
145 
            if !p.exec_bits && f.kind == NodeKind::File && e.kind == NodeKind::File {
146 
                e.exec = f.mode.exec;
147 
            }
148 
        }
149 
    }
150
c3069537 Andy McCall ↶ 151 
    /// Like `build_against`, for files given in memory (imports).
75167672 Andy McCall 152 
    pub(crate) fn build_from_entries(
153 
        &self,
154 
        base: &State,
155 
        base_snapshot: &Snapshot,
156 
        reuse: HashMap<String, (NodeId, NodeKind)>,
157 
        fs_entries: &BTreeMap<String, FsEntry>,
158 
    ) -> Result<BuiltChange> {
f7510b63 Andy McCall ↶ 159 
        self.build_from_entries_resolving(base, base_snapshot, reuse, fs_entries, &HashSet::new())
160 
    }
161
162 
    /// Like `build_from_entries`, settling the replace conflicts of the
163 
    /// opaque files in `resolve` to their content (spec §9.3, §9.6): a
164 
    /// caller that knows the user chose it, such as the hub's resolver. A
165 
    /// snapshot never settles one on its own, since a checkout writes one
166 
    /// candidate at the path.
167 
    pub(crate) fn build_from_entries_resolving(
168 
        &self,
169 
        base: &State,
170 
        base_snapshot: &Snapshot,
171 
        reuse: HashMap<String, (NodeId, NodeKind)>,
172 
        fs_entries: &BTreeMap<String, FsEntry>,
173 
        resolve: &HashSet<String>,
174 
    ) -> Result<BuiltChange> {
23ff1288 Andy McCall ↶ 175 
        self.build_from_entries_with(
176 
            base,
177 
            base_snapshot,
178 
            reuse,
179 
            fs_entries,
180 
            resolve,
181 
            Scope::default(),
182 
        )
dfb23a3a Andy McCall ↶ 183 
    }
184
185 
    /// `build_from_entries` whose new nodes take identifiers derived from
186 
    /// `seed` and their paths instead of random ones, so the same input
187 
    /// always builds the same change (a Git import, spec §23).
2306a4d2 Andy McCall ↶ 188 
    ///
189 
    /// With `dir_needed`, the snapshot and entries hold only the paths the
190 
    /// import touches (and the directories a removal may leave empty); a
191 
    /// directory stays while `dir_needed` says the new tree has something
192 
    /// under it.
dfb23a3a Andy McCall ↶ 193 
    pub(crate) fn build_from_entries_seeded(
194 
        &self,
195 
        base: &State,
196 
        base_snapshot: &Snapshot,
197 
        fs_entries: &BTreeMap<String, FsEntry>,
198 
        seed: [u8; 32],
2306a4d2 Andy McCall ↶ 199 
        dir_needed: Option<&dyn Fn(&str) -> bool>,
dfb23a3a Andy McCall ↶ 200 
    ) -> Result<BuiltChange> {
201 
        self.build_from_entries_with(
202 
            base,
203 
            base_snapshot,
204 
            HashMap::new(),
205 
            fs_entries,
206 
            &HashSet::new(),
23ff1288 Andy McCall ↶ 207 
            Scope {
208 
                seed: Some(seed),
209 
                dir_needed,
210 
            },
dfb23a3a Andy McCall ↶ 211 
        )
212 
    }
213
23ff1288 Andy McCall ↶ 214 
    /// `build_from_entries` for a snapshot most of whose files are known
215 
    /// unchanged (oxygen#41): only the paths that may differ are built
216 
    /// (files read rather than known unchanged, base files gone from disk,
217 
    /// and the directories those may leave empty), against a snapshot of
218 
    /// just those, so a large tree with a few edits costs what they touch.
219 
    fn build_changed_only(
220 
        &self,
221 
        base: &State,
222 
        base_snapshot: &Snapshot,
223 
        reuse: HashMap<String, (NodeId, NodeKind)>,
224 
        fs_entries: &BTreeMap<String, FsEntry>,
225 
    ) -> Result<BuiltChange> {
d616faf5 Andy McCall ↶ 226 
        let pick = crate::timing::phase("snapshot:   of which picking the changed");
23ff1288 Andy McCall ↶ 227 
        // A file known unchanged may still have changed mode: that alters
228 
        // neither its size nor its time.
d616faf5 Andy McCall ↶ 229 
        let changed: BTreeMap<String, FsEntry> = join(fs_entries, &base_snapshot.files)
230 
            .filter_map(|(p, e, f)| {
231 
                let e = e?;
232 
                (!e.unchanged || f.is_none_or(|f| f.kind != e.kind || f.mode.exec != e.exec))
233 
                    .then(|| (p.clone(), e.clone()))
23ff1288 Andy McCall ↶ 234 
            })
235 
            .collect();
236 
        let mut files: BTreeMap<String, crate::state::FileState> = BTreeMap::new();
237 
        for p in changed.keys() {
238 
            if let Some(f) = base_snapshot.files.get(p) {
239 
                files.insert(p.clone(), f.clone());
240 
            }
241 
        }
d616faf5 Andy McCall ↶ 242 
        // Paths the base has and the disk does not.
243 
        let gone = join(&base_snapshot.files, fs_entries)
244 
            .filter_map(|(p, f, e)| (e.is_none()).then_some((p, f?)));
245 
        for (p, f) in gone {
246 
            if f.kind == NodeKind::Dir {
23ff1288 Andy McCall ↶ 247 
                continue;
248 
            }
249 
            files.insert(p.clone(), f.clone());
250 
            let mut end = p.len();
251 
            while let Some(i) = p[..end].rfind('/') {
252 
                let dir = &p[..i];
253 
                if files.contains_key(dir) {
254 
                    break;
255 
                }
256 
                if let Some(d) = base_snapshot.files.get(dir) {
257 
                    files.insert(dir.to_string(), d.clone());
258 
                }
259 
                end = i;
260 
            }
261 
        }
262 
        let part = Snapshot {
263 
            tree: base_snapshot.tree.clone(),
264 
            files,
265 
            unrendered: BTreeSet::new(),
266 
        };
d616faf5 Andy McCall ↶ 267 
        drop(pick);
268 
        let _build = crate::timing::phase("snapshot:   of which building the change");
23ff1288 Andy McCall ↶ 269 
        // Whether anything on disk is under `dir`.
270 
        let dir_needed = |dir: &str| {
271 
            let prefix = format!("{dir}/");
272 
            fs_entries
273 
                .range(prefix.clone()..)
274 
                .next()
275 
                .is_some_and(|(k, _)| k.starts_with(&prefix))
276 
        };
277 
        self.build_from_entries_with(
278 
            base,
279 
            &part,
280 
            reuse,
281 
            &changed,
282 
            &HashSet::new(),
283 
            Scope {
284 
                seed: None,
285 
                dir_needed: Some(&dir_needed),
286 
            },
287 
        )
288 
    }
289
dfb23a3a Andy McCall ↶ 290 
    fn build_from_entries_with(
291 
        &self,
292 
        base: &State,
293 
        base_snapshot: &Snapshot,
294 
        reuse: HashMap<String, (NodeId, NodeKind)>,
295 
        fs_entries: &BTreeMap<String, FsEntry>,
296 
        resolve: &HashSet<String>,
23ff1288 Andy McCall ↶ 297 
        scope: Scope,
dfb23a3a Andy McCall ↶ 298 
    ) -> Result<BuiltChange> {
23ff1288 Andy McCall ↶ 299 
        let Scope { seed, dir_needed } = scope;
2306a4d2 Andy McCall ↶ 300 
        let mut laps = crate::timing::Laps::start();
cdb5beba Andy McCall ↶ 301 
        // Attributes come from the files being committed, so a new
302 
        // `.oxyattributes` applies to the same change that adds it.
303 
        let attrs = crate::attributes::Attributes::from_files(
75167672 Andy McCall 304 
            fs_entries
305 
                .iter()
306 
                .filter(|(p, _)| p.as_str() == ".oxyattributes" || p.ends_with("/.oxyattributes"))
307 
                .map(|(p, e)| (p.as_str(), e.content.as_slice())),
cdb5beba Andy McCall ↶ 308 
        );
309 
        let mut b = ChangeBuilder::new(base, base_snapshot, reuse).with_attributes(attrs);
f7510b63 Andy McCall ↶ 310 
        b.resolve = resolve.clone();
dfb23a3a Andy McCall ↶ 311 
        b.node_seed = seed;
2306a4d2 Andy McCall ↶ 312 
        laps.lap("build: builder");
6e723b81 Andy McCall ↶ 313 
        // Rename detection (spec §7.1, §23): a removed file and an added file
314 
        // of the same kind whose contents are at least half alike become one
315 
        // move of the same node, so identity survives across the rename.
316 
        let renames = detect_renames(base_snapshot, fs_entries, |p| self.in_scope(p));
75167672 Andy McCall 317 
        let renamed_from: HashMap<&str, &str> = renames
318 
            .iter()
319 
            .map(|(old, new)| (new.as_str(), old.as_str()))
320 
            .collect();
321 
        let renamed_to: HashMap<&str, &str> = renames
322 
            .iter()
323 
            .map(|(old, new)| (old.as_str(), new.as_str()))
324 
            .collect();
2306a4d2 Andy McCall ↶ 325 
        laps.lap("build: renames");
360aa74a Andy McCall ↶ 326 
        b.diff_in_parallel(fs_entries);
327 
        laps.lap("build: diffs");
c3069537 Andy McCall ↶ 328 
        for (path, fs_entry) in fs_entries {
98286406 Andy McCall ↶ 329 
            match base_snapshot.files.get(path) {
6e723b81 Andy McCall ↶ 330 
                None => match renamed_from.get(path.as_str()) {
75167672 Andy McCall 331 
                    Some(old) => {
332 
                        b.move_path(self, old, path, fs_entry, &base_snapshot.files[*old])?
333 
                    }
6e723b81 Andy McCall ↶ 334 
                    None => b.add_path(self, path, fs_entry)?,
335 
                },
98286406 Andy McCall ↶ 336 
                Some(expected) => b.update_path(self, path, fs_entry, expected)?,
337 
            }
338 
        }
2306a4d2 Andy McCall ↶ 339 
        laps.lap("build: paths");
958fb27d Andy McCall ↶ 340 
        // A directory stays while anything under it exists or is out of
341 
        // scope: the directories above each such path, gathered once.
342 
        let mut needed_dirs: HashSet<&str> = HashSet::new();
343 
        fn parents_of<'a>(p: &'a str, set: &mut HashSet<&'a str>) {
344 
            let mut end = p.len();
345 
            while let Some(i) = p[..end].rfind('/') {
346 
                if !set.insert(&p[..i]) {
347 
                    break;
348 
                }
349 
                end = i;
350 
            }
351 
        }
352 
        for p in fs_entries.keys() {
353 
            parents_of(p, &mut needed_dirs);
354 
        }
355 
        for p in base_snapshot.files.keys() {
356 
            if !self.in_scope(p) {
357 
                parents_of(p, &mut needed_dirs);
358 
            }
359 
        }
98286406 Andy McCall ↶ 360 
        for (path, expected) in &base_snapshot.files {
75167672 Andy McCall 361 
            if fs_entries.contains_key(path)
362 
                || !self.in_scope(path)
363 
                || renamed_to.contains_key(path.as_str())
364 
            {
98286406 Andy McCall ↶ 365 
                continue;
366 
            }
2306a4d2 Andy McCall ↶ 367 
            let still_needed = expected.kind == NodeKind::Dir
368 
                && match dir_needed {
369 
                    Some(f) => f(path),
370 
                    None => needed_dirs.contains(path.as_str()),
371 
                };
98286406 Andy McCall ↶ 372 
            if !still_needed {
373 
                b.remove_path(path, expected);
374 
            }
375 
        }
2306a4d2 Andy McCall ↶ 376 
        laps.lap("build: removals");
a87c7b9d Andy McCall ↶ 377 
        if seed.is_some() {
378 
            b.retire_stray_placements();
379 
        }
c2e1d090 Andy McCall 380 
        b.retire_unified();
2306a4d2 Andy McCall ↶ 381 
        laps.lap("build: stray placements");
614541c3 Andy McCall ↶ 382 
        let mut statuses = b.statuses;
383 
        statuses.sort_by(|a, b| a.path.cmp(&b.path));
75167672 Andy McCall 384 
        Ok(BuiltChange {
385 
            tree_ops: b.tree_ops,
386 
            edits: b.edits,
387 
            contributors: b.contributors.into_iter().collect(),
388 
            resolves: b.resolves,
389 
            statuses,
390 
            blobs: b.blobs,
391 
        })
614541c3 Andy McCall ↶ 392 
    }
98286406 Andy McCall ↶ 393
c07e0e7d Andy McCall ↶ 394 
    /// Store a change and, when `core.sign = change` and we are its author,
395 
    /// sign it. A rewritten change by another author is stored unsigned: only
396 
    /// its author (or a delegate) can vouch for it.
614541c3 Andy McCall ↶ 397 
    pub(crate) fn store_signed(&self, change: &Change) -> Result<ChangeHash> {
c07e0e7d Andy McCall ↶ 398 
        let ours = change.author == self.author();
399 
        if self.config.sign_changes && ours && self.secret_key()?.is_none() {
614541c3 Andy McCall ↶ 400 
            return fail("core.sign is \"change\" but no secret key is available for author.key");
401 
        }
5b4c5130 Andy McCall ↶ 402 
        let storing = crate::timing::phase("store: the change");
614541c3 Andy McCall ↶ 403 
        let hash = self.store.put_change(change)?;
5b4c5130 Andy McCall ↶ 404 
        drop(storing);
c07e0e7d Andy McCall ↶ 405 
        if self.config.sign_changes && ours {
5b4c5130 Andy McCall ↶ 406 
            let _t = crate::timing::phase("store: sign it");
614541c3 Andy McCall ↶ 407 
            self.sign_change(&hash)?;
408 
        }
409 
        Ok(hash)
410 
    }
411
412 
    /// Snapshot the working copy into the working-copy change (spec §12.3).
cc7a0056 Andy McCall ↶ 413 
    /// The result carries the full base state and its whole rendering.
614541c3 Andy McCall ↶ 414 
    pub fn snapshot(&self, command: &str) -> Result<SnapshotResult> {
f9466021 Andy McCall ↶ 415 
        self.snapshot_as(command, false, false)
cc7a0056 Andy McCall ↶ 416 
    }
417
418 
    /// Snapshot the working copy as [`Repo::snapshot`] does, for commands
8ab265dc Andy McCall ↶ 419 
    /// that need its statuses, conflicts and base history, and the graphs
420 
    /// of the files it changes: status, log, show, commit, diff, hunks and
421 
    /// discard. Text files known unchanged since the last manifest are
cc7a0056 Andy McCall ↶ 422 
    /// neither read nor rendered: the result's `base` is a history-only
423 
    /// state without their graphs, and its `base_snapshot` lists them in
8ab265dc Andy McCall ↶ 424 
    /// `unrendered` with empty bytes. They are never conflicted, and no
425 
    /// change the working copy makes touches them.
cc7a0056 Andy McCall ↶ 426 
    pub fn snapshot_status(&self, command: &str) -> Result<SnapshotResult> {
f9466021 Andy McCall ↶ 427 
        self.snapshot_as(command, true, false)
cc7a0056 Andy McCall ↶ 428 
    }
429
cd4eb55e Andy McCall ↶ 430 
    /// `snapshot_as`; `for_commit` writes the manifest even where a status
431 
    /// would not bother (an edited file is read next time whatever it
432 
    /// says), since after the commit that file is unchanged, and records no
433 
    /// operation: the commit's own, recorded straight after, replaces the
434 
    /// working copy, so the snapshot's would cost a durable write for
435 
    /// nothing. Should the commit fail, the files are still on disk and the
436 
    /// next snapshot finds them.
f9466021 Andy McCall ↶ 437 
    fn snapshot_as(
438 
        &self,
439 
        command: &str,
440 
        partial: bool,
cd4eb55e Andy McCall ↶ 441 
        for_commit: bool,
f9466021 Andy McCall ↶ 442 
    ) -> Result<SnapshotResult> {
4e88e774 Andy McCall ↶ 443 
        let _lock = self.lock_snapshot()?;
8f10a07a Andy McCall ↶ 444 
        let batch = self.store.begin_batch()?;
cd4eb55e Andy McCall ↶ 445 
        let result = self.snapshot_locked(command, partial, for_commit);
8f10a07a Andy McCall ↶ 446 
        batch.finish()?;
447 
        result
448 
    }
449
f9466021 Andy McCall ↶ 450 
    fn snapshot_locked(
451 
        &self,
452 
        command: &str,
453 
        partial: bool,
cd4eb55e Andy McCall ↶ 454 
        for_commit: bool,
f9466021 Andy McCall ↶ 455 
    ) -> Result<SnapshotResult> {
0bdc3f91 Andy McCall ↶ 456 
        let _all = crate::timing::phase("snapshot");
614541c3 Andy McCall ↶ 457 
        let (op_ref, op) = self.head_op()?;
458 
        let (old_hash, old_wc) = self.working_copy()?;
cc7a0056 Andy McCall ↶ 459 
        // The base's history first: its graph keys say which files the
460 
        // manifest already knows equal to it.
461 
        let mut base = {
462 
            let _t = crate::timing::phase("snapshot: load base history");
f9575918 Andy McCall ↶ 463 
            self.load_state_indexed(&old_wc.deps)?
0bdc3f91 Andy McCall ↶ 464 
        };
d0d52107 Andy McCall ↶ 465 
        let view = {
466 
            let _t = crate::timing::phase("snapshot: tree view");
c2e1d090 Andy McCall 467 
            self.tree_view(&mut base)?
d0d52107 Andy McCall ↶ 468 
        };
8b276e01 Andy McCall ↶ 469 
        let manifest = {
470 
            let _t = crate::timing::phase("snapshot: read manifest");
471 
            self.read_manifest()?.map(|(_, _, m)| m)
472 
        };
cc7a0056 Andy McCall ↶ 473 
        let tracked: BTreeMap<String, (NodeKind, Option<ObjectRef>)> = {
474 
            let _t = crate::timing::phase("snapshot: tracked paths");
36e51aa8 Andy McCall ↶ 475 
            // In the view's path order, so the map is built without searching.
476 
            view.entries
477 
                .iter()
478 
                .filter(|(_, e)| e.kind != NodeKind::Dir)
479 
                .map(|(p, e)| {
480 
                    let blob = if e.kind != NodeKind::File || e.mode.opaque {
481 
                        base.live_blobs(&e.node).first().copied()
482 
                    } else {
483 
                        None
484 
                    };
485 
                    (p.clone(), (e.kind, blob))
486 
                })
487 
                .collect()
cc7a0056 Andy McCall ↶ 488 
        };
992467d3 Andy McCall ↶ 489 
        // A text file's graph key, worked out when the walk asks for it.
1966f5db Andy McCall ↶ 490 
        let key_of = |p: &str| -> Option<[u8; 32]> {
992467d3 Andy McCall ↶ 491 
            let e = view.entries.get(p)?;
492 
            if e.kind != NodeKind::File || e.mode.opaque {
493 
                return None;
494 
            }
1966f5db Andy McCall ↶ 495 
            base.graph_key(&e.node)
992467d3 Andy McCall ↶ 496 
        };
cc7a0056 Andy McCall ↶ 497 
        let mut fs_entries = {
498 
            let _t = crate::timing::phase("snapshot: read working copy");
992467d3 Andy McCall ↶ 499 
            self.walk_working_copy_with(manifest.as_deref(), &tracked, Some(&key_of))?
cc7a0056 Andy McCall ↶ 500 
        };
992467d3 Andy McCall ↶ 501 
        // The keys of the files read, before building graphs drops them: a
502 
        // file found equal to its base rendering keeps its key.
1966f5db Andy McCall ↶ 503 
        let keys: HashMap<String, [u8; 32]> = fs_entries
992467d3 Andy McCall ↶ 504 
            .iter()
505 
            .filter(|(_, e)| !e.unchanged && e.kind == NodeKind::File)
506 
            .filter_map(|(p, _)| Some((p.clone(), key_of(p)?)))
507 
            .collect();
0bdc3f91 Andy McCall ↶ 508 
        let base_snapshot = {
cc7a0056 Andy McCall ↶ 509 
            let _t = crate::timing::phase("snapshot: load and render base");
510 
            if partial {
d616faf5 Andy McCall ↶ 511 
                let sets = crate::timing::phase("snapshot:   of which choosing graphs");
512 
                // Every file but those on disk known unchanged.
513 
                let need: Vec<(&str, NodeId)> = join(&view.entries, &fs_entries)
514 
                    .filter_map(|(p, v, f)| {
515 
                        let v = v?;
516 
                        (v.kind == NodeKind::File && !f.is_some_and(|f| f.unchanged))
517 
                            .then_some((p.as_str(), v.node))
518 
                    })
cc7a0056 Andy McCall ↶ 519 
                    .collect();
d616faf5 Andy McCall ↶ 520 
                let render: std::collections::HashSet<&str> =
521 
                    need.iter().map(|(p, _)| *p).collect();
522 
                let need: Vec<NodeId> = need.iter().map(|(_, n)| *n).collect();
523 
                drop(sets);
d0990c2b Andy McCall ↶ 524 
                {
525 
                    let _t = crate::timing::phase("snapshot:   of which building graphs");
526 
                    self.load_graphs_of(&mut base, &need)?;
527 
                }
528 
                let _t = crate::timing::phase("snapshot:   of which rendering");
d616faf5 Andy McCall ↶ 529 
                base.materialize_only(&render)
cc7a0056 Andy McCall ↶ 530 
            } else {
531 
                self.load_all_graphs(&mut base)?;
532 
                base.materialize()
533 
            }
0bdc3f91 Andy McCall ↶ 534 
        };
614541c3 Andy McCall ↶ 535
536 
        // Reuse the NodeIds the previous working-copy version assigned to new
537 
        // paths, so an unchanged working copy hashes identically (spec §12.3).
538 
        let reuse = added_node_paths(&old_wc, &base_snapshot);
c863c7a7 Andy McCall ↶ 539 
        self.platform_normalize(&mut fs_entries, &base_snapshot);
0bdc3f91 Andy McCall ↶ 540 
        let built = {
541 
            let _t = crate::timing::phase("snapshot: diff against base");
23ff1288 Andy McCall ↶ 542 
            if partial && self.config.sparse_paths.is_empty() {
543 
                #[cfg(test)]
544 
                let full =
545 
                    self.build_from_entries(&base, &base_snapshot, reuse.clone(), &fs_entries)?;
546 
                let built = self.build_changed_only(&base, &base_snapshot, reuse, &fs_entries)?;
547 
                #[cfg(test)]
548 
                {
549 
                    // New nodes take random identifiers: compare them by
550 
                    // order of appearance.
551 
                    let norm = |b: &BuiltChange| {
552 
                        let text = format!("{:?} {:?} {:?}", b.tree_ops, b.edits, b.statuses);
553 
                        let mut seen: Vec<String> = Vec::new();
554 
                        let mut out = String::new();
555 
                        let mut rest = text.as_str();
556 
                        while let Some(i) = rest.find("NodeId(") {
557 
                            out.push_str(&rest[..i]);
558 
                            let end = i + rest[i..].find(')').unwrap() + 1;
559 
                            let id = rest[i..end].to_string();
560 
                            let n = seen.iter().position(|x| *x == id).unwrap_or_else(|| {
561 
                                seen.push(id);
562 
                                seen.len() - 1
563 
                            });
564 
                            out.push_str(&format!("N{n}"));
565 
                            rest = &rest[end..];
566 
                        }
567 
                        out.push_str(rest);
568 
                        out
569 
                    };
570 
                    assert_eq!(
571 
                        norm(&built),
572 
                        norm(&full),
573 
                        "a changed-only build is the full one"
574 
                    );
575 
                }
576 
                built
577 
            } else {
578 
                self.build_from_entries(&base, &base_snapshot, reuse, &fs_entries)?
579 
            }
0bdc3f91 Andy McCall ↶ 580 
        };
cc7a0056 Andy McCall ↶ 581 
        // Text files found equal to an unconflicted base rendering keep its
582 
        // graph key, so the next snapshot need not read or render them.
583 
        {
584 
            let changed: HashSet<&str> = built.statuses.iter().map(|s| s.path.as_str()).collect();
585 
            for (p, e) in fs_entries.iter_mut() {
586 
                if e.unchanged || e.kind != NodeKind::File || changed.contains(p.as_str()) {
587 
                    continue;
588 
                }
589 
                e.graph_key = match base_snapshot.files.get(p).map(|f| &f.content) {
590 
                    Some(FileContent::Text {
591 
                        bytes,
592 
                        materialized,
40a40bb7 Andy McCall ↶ 593 
                    }) if !materialized.is_conflicted() && *bytes == *e.content => {
1966f5db Andy McCall ↶ 594 
                        keys.get(p).copied()
cc7a0056 Andy McCall ↶ 595 
                    }
596 
                    _ => None,
597 
                };
598 
            }
599 
        }
614541c3 Andy McCall ↶ 600
98286406 Andy McCall ↶ 601 
        let mut wc = Change {
6f8aeb99 Andy McCall ↶ 602 
            version: crate::objects::CHANGE_VERSION,
98286406 Andy McCall ↶ 603 
            change_id: old_wc.change_id,
604 
            deps: vec![],
605 
            author: self.author(),
606 
            hlc: old_wc.hlc,
607 
            message: old_wc.message.clone(),
614541c3 Andy McCall ↶ 608 
            tree_ops: built.tree_ops,
609 
            edits: built.edits,
98286406 Andy McCall ↶ 610 
            kind: ChangeKind::WorkingCopy,
611 
        };
612 
        let base_heads = base.maximal(&old_wc.deps);
613 
        let mut deps: BTreeSet<ChangeHash> = base_heads.into_iter().collect();
614 
        deps.extend(referenced_changes(&wc, &base));
614541c3 Andy McCall ↶ 615 
        deps.extend(built.contributors.iter().copied());
98286406 Andy McCall ↶ 616 
        wc.deps = deps.into_iter().collect();
617 
        wc.normalize();
614541c3 Andy McCall ↶ 618 
        let resolves = built.resolves;
619 
        let contributors = built.contributors;
620 
        let statuses = built.statuses;
98286406 Andy McCall ↶ 621 
        // Empty snapshot: hash the candidate before storing; only record an op on change.
8b276e01 Andy McCall ↶ 622 
        let record = crate::timing::phase("snapshot: hash and record");
17f94b13 Andy McCall ↶ 623 
        // Encoded once, its texts stored a batch at a time as they are
624 
        // made (a stored blob is not stored again); the change itself only
625 
        // when it changed.
5b4c5130 Andy McCall ↶ 626 
        // For a commit the change is neither encoded, stored nor recorded:
627 
        // the commit stores its own, built from this one's operations, and
628 
        // its texts with it. The hash it would have is not needed.
629 
        let (new_hash, changed, encoded) = if for_commit {
630 
            (old_hash, true, Vec::new())
631 
        } else {
632 
            let encoded = wc.encode_streaming(|batch| self.store.put_texts(&batch))?;
633 
            let new_hash = crate::hash::hash_object(crate::hash::ObjectType::Change, &encoded);
634 
            (new_hash, new_hash != old_hash, encoded)
635 
        };
cd4eb55e Andy McCall ↶ 636 
        if changed && !for_commit {
17f94b13 Andy McCall ↶ 637 
            self.store.put(crate::hash::ObjectType::Change, &encoded)?;
98286406 Andy McCall ↶ 638 
            let mut view = op.view.clone();
639 
            view.workspaces.insert(self.workspace, new_hash);
640 
            self.record_op(&[op_ref], &format!("{command}{SNAPSHOT_SUFFIX}"), view)?;
641 
        }
8b276e01 Andy McCall ↶ 642 
        drop(record);
135212de Andy McCall ↶ 643 
        // The prompt compares against this without recording anything.
0bdc3f91 Andy McCall ↶ 644 
        let _t = crate::timing::phase("snapshot: write manifest");
992467d3 Andy McCall ↶ 645 
        let conflicted = base_snapshot.is_conflicted();
646 
        let uncommitted = !wc.tree_ops.is_empty() || !wc.edits.is_empty();
647 
        // Rewriting the manifest pays only when it would teach the next
648 
        // snapshot something: other flags, a file found equal to the base
649 
        // (it need not be read again), a new blob, a path added or gone.
650 
        // An edited file is read next time whatever it says.
651 
        let worth = match self.read_manifest()? {
cd4eb55e Andy McCall ↶ 652 
            _ if for_commit => true,
992467d3 Andy McCall ↶ 653 
            None => true,
654 
            Some((c, u, m)) => {
655 
                c != conflicted
656 
                    || u != uncommitted
657 
                    || m.entries.len() != fs_entries.len()
658 
                    || fs_entries.iter().any(|(p, e)| {
659 
                        !e.unchanged && (e.graph_key.is_some() || !m.entries.contains_key(p))
660 
                    })
661 
                    || built
662 
                        .blobs
663 
                        .iter()
664 
                        .any(|(p, b)| m.entries.get(p).and_then(|e| e.blob) != Some(*b))
36041796 Andy McCall ↶ 665 
                    // With uncommitted work, the status cache below holds
666 
                    // only while the manifest describes the edited files as
667 
                    // they are now.
668 
                    || (uncommitted && statuses.iter().any(|st| {
669 
                        let Some(e) = fs_entries.get(&st.path) else {
670 
                            return m.entries.contains_key(&st.path);
671 
                        };
672 
                        let Some(me) = m.entries.get(&st.path) else {
673 
                            return true;
674 
                        };
675 
                        let hash = e
676 
                            .content_hash
677 
                            .unwrap_or_else(|| *blake3::hash(&e.content).as_bytes());
678 
                        let stat = e.stat.or_else(|| {
679 
                            fs::symlink_metadata(self.root.join(&st.path))
680 
                                .ok()
681 
                                .map(|md| (md.len(), super::prompt::mtime_ns(&md)))
682 
                        });
e226a6cc Andy McCall ↶ 683 
                        // The executable bit too: a mode change moves
684 
                        // neither size nor time.
685 
                        me.hash != hash || me.exec != e.exec || stat != Some((me.size, me.mtime_ns))
36041796 Andy McCall ↶ 686 
                    }))
992467d3 Andy McCall ↶ 687 
            }
688 
        };
f9466021 Andy McCall ↶ 689 
        let attributes = crate::attributes::Attributes::from_files(
690 
            fs_entries
691 
                .iter()
692 
                .filter(|(p, _)| is_attributes(p))
693 
                .map(|(p, e)| (p.as_str(), e.content.as_slice())),
694 
        );
236f35c0 Andy McCall ↶ 695 
        let mut manifest = None;
696 
        if for_commit {
697 
            manifest = Some(PendingManifest {
698 
                entries: fs_entries,
699 
                blobs: built.blobs.clone(),
700 
                dirs: self.walked_dirs.take(),
701 
                written: super::prompt::now_ns(),
702 
            });
703 
        } else {
704 
            if worth {
705 
                self.write_manifest_entries(conflicted, uncommitted, &fs_entries, &built.blobs)?;
706 
            }
707 
            self.write_status_cache((uncommitted && !conflicted).then_some(&new_hash), &statuses);
708 
        }
75167672 Andy McCall 709 
        Ok(SnapshotResult {
710 
            wc_hash: new_hash,
711 
            wc,
712 
            base,
713 
            base_snapshot,
714 
            statuses,
715 
            changed,
716 
            resolves,
717 
            contributors,
f9466021 Andy McCall ↶ 718 
            attributes,
d4dbab97 Andy McCall ↶ 719 
            untracked_nested: self.untracked_nested.borrow().clone(),
236f35c0 Andy McCall ↶ 720 
            manifest,
75167672 Andy McCall 721 
        })
98286406 Andy McCall ↶ 722 
    }
723
50a046e4 Andy McCall ↶ 724 
    /// `oxy status --json` (spec §9.5): the changed paths and every conflict
725 
    /// of the working copy's base with its kind, path, and contributing
726 
    /// ChangeIds.
727 
    pub fn status_json(&self, snap: &SnapshotResult) -> Result<String> {
728 
        let heads = self.bookmark_heads()?;
729 
        let base_snapshot = &snap.base_snapshot;
730 
        let by_node = base_snapshot.tree.by_node();
731 
        let path_of = |n: &NodeId| by_node.get(n).map(|e| e.path.clone());
75167672 Andy McCall 732 
        let id_of = |h: &ChangeHash| {
733 
            snap.base
734 
                .get(h)
735 
                .map(|m| m.change_id.to_hex())
736 
                .unwrap_or_else(|| h.to_hex())
737 
        };
738 
        let paths = json_array(snap.statuses.iter().map(|s| {
739 
            format!(
740 
                "{{\"path\":{},\"status\":{}}}",
741 
                json_str(&s.path),
742 
                json_str(&format!("{:?}", s.status))
743 
            )
744 
        }));
50a046e4 Andy McCall ↶ 745 
        let mut conflicts: Vec<String> = Vec::new();
746 
        for (p, f) in &base_snapshot.files {
747 
            if !f.is_conflicted() {
748 
                continue;
749 
            }
750 
            let (kind, changes) = match &f.content {
75167672 Andy McCall 751 
                FileContent::Text { .. } => {
752 
                    ("text", snap.base.conflict_regions(&f.node).contributors)
753 
                }
754 
                FileContent::Opaque { blobs } => {
755 
                    ("replace", snap.base.blob_contributors(&f.node, blobs))
756 
                }
50a046e4 Andy McCall ↶ 757 
                FileContent::Dir => continue,
758 
            };
75167672 Andy McCall 759 
            let resolved = match snap
760 
                .statuses
761 
                .iter()
762 
                .find(|s| s.path == *p)
763 
                .map(|s| s.status)
764 
            {
50a046e4 Andy McCall ↶ 765 
                Some(ChangeStatus::Resolved) => "true",
766 
                Some(ChangeStatus::ConflictEdited) => "\"markers\"",
767 
                _ => "false",
768 
            };
769 
            conflicts.push(format!(
770 
                "{{\"kind\":{},\"path\":{},\"node\":{},\"changes\":{},\"resolved\":{resolved}}}",
771 
                json_str(kind),
772 
                json_str(p),
773 
                json_str(&f.node.to_hex()),
774 
                json_array(changes.iter().map(|h| json_str(&id_of(h))))
775 
            ));
776 
        }
777 
        for c in &base_snapshot.tree.conflicts {
778 
            let (kind, nodes): (&str, Vec<NodeId>) = match c {
779 
                crate::tree::TreeConflict::Name { nodes, .. } => ("name", nodes.clone()),
75167672 Andy McCall 780 
                crate::tree::TreeConflict::MultiPlacement { node, .. } => {
781 
                    ("multi-placement", vec![*node])
782 
                }
50a046e4 Andy McCall ↶ 783 
                crate::tree::TreeConflict::Orphan { node, .. } => ("orphan", vec![*node]),
784 
                crate::tree::TreeConflict::Cycle { node } => ("cycle", vec![*node]),
785 
                crate::tree::TreeConflict::Mode { node, .. } => ("mode", vec![*node]),
786 
            };
787 
            let detail = match c {
788 
                crate::tree::TreeConflict::Name { name, parent, .. } => {
75167672 Andy McCall 789 
                    let dir = if *parent == NodeId::ROOT {
790 
                        String::new()
791 
                    } else {
792 
                        path_of(parent).unwrap_or_default()
793 
                    };
794 
                    format!(
795 
                        ",\"path\":{}",
796 
                        json_str(&if dir.is_empty() {
797 
                            name.clone()
798 
                        } else {
799 
                            format!("{dir}/{name}")
800 
                        })
801 
                    )
50a046e4 Andy McCall ↶ 802 
                }
75167672 Andy McCall 803 
                _ => nodes
804 
                    .first()
805 
                    .and_then(path_of)
806 
                    .map(|p| format!(",\"path\":{}", json_str(&p)))
807 
                    .unwrap_or_default(),
50a046e4 Andy McCall ↶ 808 
            };
809 
            conflicts.push(format!(
810 
                "{{\"kind\":{},\"nodes\":{}{detail}}}",
811 
                json_str(kind),
812 
                json_array(nodes.iter().map(|n| json_str(&n.to_hex())))
813 
            ));
814 
        }
815 
        Ok(format!(
b689779e Andy McCall ↶ 816 
            "{{\"bookmark\":{},\"working_copy\":{},\"heads\":{},\"clean\":{},\"conflicted\":{},\"paths\":{paths},\"conflicts\":{},\"untracked_nested\":{},\"skipped_large\":{}}}",
50a046e4 Andy McCall ↶ 817 
            json_str(&self.config.bookmark),
818 
            json_str(&snap.wc.change_id.to_hex()),
819 
            heads.len(),
820 
            snap.statuses.is_empty(),
821 
            base_snapshot.is_conflicted(),
d4dbab97 Andy McCall ↶ 822 
            json_array(conflicts),
b689779e Andy McCall ↶ 823 
            json_array(snap.untracked_nested.iter().map(|p| json_str(p))),
824 
            json_array(
825 
                self.skipped_large
826 
                    .borrow()
827 
                    .iter()
828 
                    .map(|(p, n)| format!("{{\"path\":{},\"size\":{n}}}", json_str(p)))
829 
            )
50a046e4 Andy McCall ↶ 830 
        ))
831 
    }
832
98286406 Andy McCall ↶ 833 
    /// Turn the working-copy change into a normal change on the bookmark (spec §12.3).
834 
    pub fn commit(&self, message: &str, extra_deps: &[ChangeHash]) -> Result<ChangeHash> {
957fee30 Andy McCall ↶ 835 
        self.commit_selection(message, extra_deps, &Selection::All)
836 
    }
837
838 
    /// The hunks of the working copy, for `oxy commit --interactive`: each
839 
    /// tree operation and each edit, in the order `select_hunks` numbers them.
840 
    pub fn hunks(&self) -> Result<(SnapshotResult, Vec<Hunk>)> {
8ab265dc Andy McCall ↶ 841 
        let snap = self.snapshot_status("oxy commit")?;
957fee30 Andy McCall ↶ 842 
        let hunks = render_hunks(&snap.wc, &snap.base, &snap.base_snapshot);
843 
        Ok((snap, hunks))
844 
    }
845
846 
    /// Commit part of the working copy (spec §12.3). What is not selected
847 
    /// stays on disk as uncommitted work: the next snapshot rebuilds it
848 
    /// against the new base, so the working copy always describes the
849 
    /// directory.
f9466021 Andy McCall ↶ 850 
    /// The graph key in `st` of each text file at `paths`, the paths a
851 
    /// commit changed: only their graphs changed, so the keys the manifest
852 
    /// holds for the others still stand (and any that does not is simply
853 
    /// not trusted by the next snapshot, which compares keys). Files with
854 
    /// an `eol` attribute are left out: their bytes on disk are not the
855 
    /// rendering.
856 
    fn committed_graph_keys(
857 
        &self,
858 
        st: &State,
859 
        paths: &[String],
860 
        attrs: &crate::attributes::Attributes,
861 
    ) -> HashMap<String, [u8; 32]> {
86e1824f Andy McCall ↶ 862 
        let _t = crate::timing::phase("commit: graph keys");
c2e1d090 Andy McCall 863 
        let view = st.view();
f9466021 Andy McCall ↶ 864 
        paths
86e1824f Andy McCall ↶ 865 
            .iter()
f9466021 Andy McCall ↶ 866 
            .filter_map(|p| {
867 
                let e = view.entries.get(p)?;
868 
                (e.kind == NodeKind::File && !e.mode.opaque && attrs.lookup(p).eol.is_none())
869 
                    .then_some(())?;
870 
                Some((p.clone(), st.graph_key(&e.node)?))
86e1824f Andy McCall ↶ 871 
            })
8457a6cc Andy McCall ↶ 872 
            .collect()
86e1824f Andy McCall ↶ 873 
    }
874
75167672 Andy McCall 875 
    pub fn commit_selection(
876 
        &self,
877 
        message: &str,
878 
        extra_deps: &[ChangeHash],
879 
        selection: &Selection,
880 
    ) -> Result<ChangeHash> {
75051941 Andy McCall ↶ 881 
        Ok(self
882 
            .commit_selection_partial(message, extra_deps, selection)?
883 
            .0)
884 
    }
885
886 
    /// As [`Repo::commit_selection`], and whether the commit left part of
887 
    /// the working copy uncommitted: when it did not, nothing is left, and
888 
    /// no snapshot is needed to say so.
889 
    pub fn commit_selection_partial(
890 
        &self,
891 
        message: &str,
892 
        extra_deps: &[ChangeHash],
893 
        selection: &Selection,
894 
    ) -> Result<(ChangeHash, bool)> {
b30a880f Andy McCall ↶ 895 
        // The commit's objects are written together, made durable at once
896 
        // before the operation that names them (spec §13.6).
897 
        let batch = self.store.begin_batch()?;
898 
        let out = self.commit_selection_batched(message, extra_deps, selection);
899 
        batch.finish()?;
900 
        out
901 
    }
902
236f35c0 Andy McCall ↶ 903 
    /// Write the manifest a commit's snapshot left to the commit, with the
904 
    /// graph keys `keys` sets.
905 
    fn write_pending_manifest(
906 
        &self,
907 
        mut m: PendingManifest,
908 
        conflicted: bool,
909 
        keys: HashMap<String, [u8; 32]>,
910 
    ) -> Result<()> {
911 
        for (path, k) in keys {
912 
            if let Some(e) = m.entries.get_mut(&path) {
913 
                e.graph_key = Some(k);
914 
            }
915 
        }
916 
        self.write_manifest_entries_as_of(
917 
            conflicted,
918 
            false,
919 
            &m.entries,
920 
            &m.blobs,
921 
            m.written,
922 
            m.dirs.unwrap_or_default(),
923 
        )
924 
    }
925
b30a880f Andy McCall ↶ 926 
    fn commit_selection_batched(
927 
        &self,
928 
        message: &str,
929 
        extra_deps: &[ChangeHash],
930 
        selection: &Selection,
75051941 Andy McCall ↶ 931 
    ) -> Result<(ChangeHash, bool)> {
6caacfff Andy McCall ↶ 932 
        let _wc = self.lock_working_copy()?;
e226a6cc Andy McCall ↶ 933 
        let mut snap = match self.snapshot_from_cache()? {
934 
            Some(snap) => snap,
935 
            None => self.snapshot_as("oxy commit", true, true)?,
936 
        };
236f35c0 Andy McCall ↶ 937 
        let pending = snap.manifest.take();
98286406 Andy McCall ↶ 938 
        if snap.wc.tree_ops.is_empty() && snap.wc.edits.is_empty() {
236f35c0 Andy McCall ↶ 939 
            if let Some(m) = pending {
940 
                let conflicted = snap.base_snapshot.is_conflicted();
941 
                self.write_pending_manifest(m, conflicted, HashMap::new())?;
942 
            }
98286406 Andy McCall ↶ 943 
            return fail("nothing to commit: the working copy matches its base");
944 
        }
75167672 Andy McCall 945 
        if let Some(s) = snap
946 
            .statuses
947 
            .iter()
948 
            .find(|s| s.status == ChangeStatus::ConflictEdited)
949 
        {
98286406 Andy McCall ↶ 950 
            return fail(format!("{} still contains conflict markers", s.path));
951 
        }
952 
        if message.trim().is_empty() {
953 
            return fail("a commit message is required");
954 
        }
192fad1e Andy McCall ↶ 955 
        // What the whole working copy references, before a whole commit
956 
        // takes its operations over (they are moved, not copied: for a
957 
        // large tree they are most of the memory).
958 
        let wc_refs = referenced_changes(&snap.wc, &snap.base);
f9466021 Andy McCall ↶ 959 
        let changed_paths: Vec<String> = snap.statuses.iter().map(|s| s.path.clone()).collect();
957fee30 Andy McCall ↶ 960 
        let (tree_ops, edits, partial) = match selection {
192fad1e Andy McCall ↶ 961 
            Selection::All => (
962 
                std::mem::take(&mut snap.wc.tree_ops),
963 
                std::mem::take(&mut snap.wc.edits),
964 
                false,
965 
            ),
957fee30 Andy McCall ↶ 966 
            Selection::Paths(paths) => {
967 
                let (o, e) = select_paths(&snap.wc, &snap.base_snapshot, paths);
968 
                let partial = o.len() != snap.wc.tree_ops.len() || e.len() != snap.wc.edits.len();
969 
                (o, e, partial)
970 
            }
971 
            Selection::Hunks(indices) => {
c2e1d090 Andy McCall 972 
                let (o, e) = select_hunks(&snap.wc, indices, &snap.base_snapshot.tree.unified);
957fee30 Andy McCall ↶ 973 
                let partial = o.len() != snap.wc.tree_ops.len() || e.len() != snap.wc.edits.len();
974 
                (o, e, partial)
975 
            }
976 
        };
977 
        if tree_ops.is_empty() && edits.is_empty() {
978 
            return fail("nothing selected to commit");
979 
        }
69953b50 Andy McCall ↶ 980 
        // A resolution may leave other work uncommitted (oxygen#118), but
981 
        // takes every file it resolves, whole.
957fee30 Andy McCall ↶ 982 
        if partial && snap.resolves {
69953b50 Andy McCall ↶ 983 
            let resolved = snap
984 
                .statuses
985 
                .iter()
986 
                .filter(|s| s.status == ChangeStatus::Resolved)
987 
                .map(|s| s.path.as_str());
988 
            let missing: Vec<&str> = match selection {
989 
                Selection::Paths(paths) => resolved
990 
                    .filter(|p| {
991 
                        !paths.iter().any(|x| {
992 
                            p == x || p.starts_with(&format!("{}/", x.trim_end_matches('/')))
993 
                        })
994 
                    })
995 
                    .collect(),
996 
                _ => resolved.collect(),
997 
            };
998 
            if !missing.is_empty() {
999 
                return fail(format!(
1000 
                    "a conflict resolution takes every file it resolves, whole: commit by paths, including {}",
1001 
                    missing.join(", ")
1002 
                ));
1003 
            }
957fee30 Andy McCall ↶ 1004 
        }
6d3eefbc Andy McCall ↶ 1005 
        // pre-commit hook: may veto.
1006 
        self.run_hook(
1007 
            "pre-commit",
1008 
            &format!(
1009 
                "{{\"event\":\"pre-commit\",\"message\":{},\"paths\":{}}}",
1010 
                json_str(message),
75167672 Andy McCall 1011 
                json_array(snap.statuses.iter().map(|s| format!(
1012 
                    "{{\"path\":{},\"status\":{}}}",
1013 
                    json_str(&s.path),
1014 
                    json_str(&format!("{:?}", s.status))
1015 
                )))
6d3eefbc Andy McCall ↶ 1016 
            ),
1017 
        )?;
98286406 Andy McCall ↶ 1018 
        let (op_ref, op) = self.head_op()?;
192fad1e Andy McCall ↶ 1019 
        let mut deps: BTreeSet<ChangeHash> = wc_refs;
6d3eefbc Andy McCall ↶ 1020 
        // semantic-deps hook: may add explicit dependencies (spec §8.1).
1021 
        if self.config.semantic_deps {
1022 
            let touched: Vec<String> = snap.statuses.iter().map(|s| s.path.clone()).collect();
1023 
            let by_node = snap.base_snapshot.tree.by_node();
1024 
            let mut candidates: Vec<(ChangeId, ChangeHash, Vec<String>)> = Vec::new();
1025 
            let mut seen: BTreeSet<ChangeHash> = BTreeSet::new();
1026 
            for h in deps.iter().chain(snap.base.maximal(&snap.wc.deps).iter()) {
1027 
                if !seen.insert(*h) {
1028 
                    continue;
1029 
                }
1030 
                let c = self.store.get_change(h)?;
1031 
                let mut paths: BTreeSet<String> = BTreeSet::new();
75167672 Andy McCall 1032 
                for n in c
1033 
                    .tree_ops
1034 
                    .iter()
1035 
                    .map(|o| o.node())
1036 
                    .chain(c.edits.iter().map(|e| e.node()))
1037 
                {
6d3eefbc Andy McCall ↶ 1038 
                    if let Some(e) = by_node.get(&n) {
1039 
                        paths.insert(e.path.clone());
1040 
                    }
1041 
                }
1042 
                candidates.push((c.change_id, *h, paths.into_iter().collect()));
1043 
            }
1044 
            for d in self.semantic_deps(&snap.wc.change_id, &touched, &candidates)? {
1045 
                if snap.base.contains(&d) {
1046 
                    deps.insert(d);
1047 
                }
1048 
            }
1049 
        }
98286406 Andy McCall ↶ 1050 
        // A resolution depends on every change that contributed to the
1051 
        // conflict (spec §8.1). Tree and opaque conflicts are resolved through
1052 
        // ordinary operations whose references already cover the contributors.
1053 
        deps.extend(snap.contributors.iter().copied());
1054 
        for d in extra_deps {
1055 
            if !snap.base.contains(d) {
75167672 Andy McCall 1056 
                return fail(format!(
1057 
                    "--depends-on {d}: not in the working copy's base state"
1058 
                ));
98286406 Andy McCall ↶ 1059 
            }
1060 
            deps.insert(*d);
1061 
        }
1062 
        let mut change = Change {
6f8aeb99 Andy McCall ↶ 1063 
            version: crate::objects::CHANGE_VERSION,
75167672 Andy McCall 1064 
            change_id: if partial {
1065 
                ChangeId::random()
1066 
            } else {
1067 
                snap.wc.change_id
1068 
            },
957fee30 Andy McCall ↶ 1069 
            deps: vec![],
98286406 Andy McCall ↶ 1070 
            author: self.author(),
1071 
            hlc: Hlc::now(),
1072 
            message: message.trim_end().to_string(),
957fee30 Andy McCall ↶ 1073 
            tree_ops,
1074 
            edits,
75167672 Andy McCall 1075 
            kind: if snap.resolves {
1076 
                ChangeKind::Resolution
1077 
            } else {
1078 
                ChangeKind::Normal
1079 
            },
98286406 Andy McCall ↶ 1080 
        };
957fee30 Andy McCall ↶ 1081 
        if partial {
1082 
            // Dependencies come from what the selected part references, not
1083 
            // from the whole working copy.
1084 
            deps = referenced_changes(&change, &snap.base);
1085 
            for d in extra_deps {
1086 
                deps.insert(*d);
1087 
            }
1088 
        }
1089 
        change.deps = deps.into_iter().collect();
98286406 Andy McCall ↶ 1090 
        change.normalize();
f9466021 Andy McCall ↶ 1091 
        let signing = crate::timing::phase("commit: sign and store");
614541c3 Andy McCall ↶ 1092 
        let hash = self.store_signed(&change)?;
f9466021 Andy McCall ↶ 1093 
        drop(signing);
1094 
        let extending = crate::timing::phase("commit: extend the state");
98286406 Andy McCall ↶ 1095
1096 
        // New bookmark heads: maximal elements of old heads plus the new change.
1097 
        // The working copy continues on its own base plus the new change, which
1098 
        // is exactly what the files on disk contain; the bookmark may hold
1099 
        // other heads (and conflicts) the working copy has not checked out.
1100 
        let old_heads = self.heads_of(&op.view, &self.config.bookmark)?;
1101 
        let old_base = snap.base.maximal(&snap.wc.deps);
1102 
        let mut all: Vec<ChangeHash> = old_heads.clone();
1103 
        all.extend(old_base.iter().copied());
1104 
        all.push(hash);
07eb99be Andy McCall ↶ 1105 
        // The snapshot already loaded the working copy's base; when that also
1106 
        // holds every bookmark head, it is the state to extend.
1107 
        let mut st = if old_heads.iter().all(|h| snap.base.contains(h)) {
1108 
            snap.base
1109 
        } else {
1110 
            self.load_state(&all)?
1111 
        };
192fad1e Andy McCall ↶ 1112 
        st.add_hashed(hash, change)?;
98286406 Andy McCall ↶ 1113 
        let mut with_new = old_heads.clone();
1114 
        with_new.push(hash);
1115 
        let new_heads = st.maximal(&with_new);
252e8cb2 Andy McCall ↶ 1116 
        // Automatic checkpoint once enough changes accumulated (spec §11).
75167672 Andy McCall 1117 
        let since = self
1118 
            .latest_checkpoint(&self.config.bookmark)?
1119 
            .map(|(_, cp)| cp.count)
1120 
            .unwrap_or(0);
1121 
        if self.config.checkpoint_interval > 0
1122 
            && (st.len() as u64).saturating_sub(since) >= self.config.checkpoint_interval
1123 
        {
252e8cb2 Andy McCall ↶ 1124 
            let cp_state = self.load_state_full(&new_heads)?;
1125 
            self.write_checkpoint(&self.config.bookmark, &cp_state, &new_heads)?;
1126 
        }
98286406 Andy McCall ↶ 1127 
        let mut wc_base = old_base.clone();
1128 
        wc_base.push(hash);
1129 
        let wc_base = st.maximal(&wc_base);
f9466021 Andy McCall ↶ 1130 
        drop(extending);
98286406 Andy McCall ↶ 1131
8457a6cc Andy McCall ↶ 1132 
        // The graph keys of the files as they now are, from the state the
1133 
        // commit just extended: a file whose graph it built has none, and
1134 
        // is read by the next snapshot; a key the next base does not share
1135 
        // (the bookmark held other heads) is simply not trusted then.
236f35c0 Andy McCall ↶ 1136 
        let keying = crate::timing::phase("commit: graph keys of the files");
f9466021 Andy McCall ↶ 1137 
        let committed_keys =
1138 
            (!partial).then(|| self.committed_graph_keys(&st, &changed_paths, &snap.attributes));
236f35c0 Andy McCall ↶ 1139 
        drop(keying);
f9466021 Andy McCall ↶ 1140 
        let recording = crate::timing::phase("commit: record");
98286406 Andy McCall ↶ 1141 
        let wc = self.new_working_copy(wc_base, "");
1142 
        let wc_hash = self.store.put_change(&wc)?;
1143 
        let mut view = op.view.clone();
75167672 Andy McCall 1144 
        view.bookmarks.insert(
1145 
            self.config.bookmark.clone(),
1146 
            self.store.put_state(&StateObject::new(new_heads))?,
1147 
        );
98286406 Andy McCall ↶ 1148 
        view.workspaces.insert(self.workspace, wc_hash);
1149 
        self.record_op(&[op_ref], &format!("oxy commit -m {message:?}"), view)?;
f9466021 Andy McCall ↶ 1150 
        drop(recording);
236f35c0 Andy McCall ↶ 1151 
        let rewriting = crate::timing::phase("commit: manifest");
957fee30 Andy McCall ↶ 1152 
        if partial {
1153 
            // The rest is rebuilt from disk by the next snapshot.
1154 
            self.drop_manifest();
1155 
        } else {
86e1824f Andy McCall ↶ 1156 
            let conflicted = snap.base_snapshot.is_conflicted() && !snap.resolves;
1157 
            // The files on disk are now exactly the new base's renderings:
1158 
            // record their graph keys, so the next snapshot neither reads nor
1159 
            // renders them (oxygen#35).
1160 
            let keys = if conflicted {
1161 
                HashMap::new()
1162 
            } else {
8457a6cc Andy McCall ↶ 1163 
                committed_keys.unwrap_or_default()
86e1824f Andy McCall ↶ 1164 
            };
236f35c0 Andy McCall ↶ 1165 
            match pending {
1166 
                Some(m) => self.write_pending_manifest(m, conflicted, keys)?,
1167 
                None => self.rewrite_manifest(conflicted, false, &keys)?,
1168 
            }
957fee30 Andy McCall ↶ 1169 
        }
236f35c0 Andy McCall ↶ 1170 
        drop(rewriting);
1171 
        let packing = crate::timing::phase("commit: autopack");
d02e545d Andy McCall ↶ 1172 
        self.maybe_autopack()?;
236f35c0 Andy McCall ↶ 1173 
        drop(packing);
1174 
        let _hook = crate::timing::phase("commit: hook");
75167672 Andy McCall 1175 
        let _ = self.run_hook(
1176 
            "post-commit",
1177 
            &format!(
1178 
                "{{\"event\":\"post-commit\",\"change_id\":{},\"hash\":{},\"message\":{}}}",
1179 
                json_str(&snap.wc.change_id.to_hex()),
1180 
                json_str(&hash.to_hex()),
1181 
                json_str(message)
1182 
            ),
1183 
        );
75051941 Andy McCall ↶ 1184 
        Ok((hash, partial))
98286406 Andy McCall ↶ 1185 
    }
1186 
}
1187
957fee30 Andy McCall ↶ 1188 
/// What part of the working copy a commit takes (spec §12.3).
1189 
#[derive(Clone, Debug)]
1190 
pub enum Selection {
1191 
    All,
1192 
    /// Every operation and edit on nodes under these paths.
1193 
    Paths(Vec<String>),
1194 
    /// Hunk indices as `Repo::hunks` numbers them; dependent hunks are added.
1195 
    Hunks(Vec<usize>),
1196 
}
1197
1198 
/// One selectable piece of the working copy: a tree operation or an edit.
1199 
#[derive(Clone, Debug)]
1200 
pub struct Hunk {
1201 
    pub index: usize,
1202 
    pub path: String,
1203 
    /// Rendered like `oxy diff`.
1204 
    pub text: String,
1205 
}
1206
1207 
/// The tree operations and edits on nodes under `paths`, plus the
1208 
/// additions of any new parent directories they need.
75167672 Andy McCall 1209 
fn select_paths(
1210 
    wc: &Change,
1211 
    base_snapshot: &Snapshot,
1212 
    paths: &[String],
1213 
) -> (Vec<TreeOp>, Vec<Edit>) {
1214 
    let selected = |p: &str| {
1215 
        paths
1216 
            .iter()
1217 
            .any(|x| p == x || p.starts_with(&format!("{}/", x.trim_end_matches('/'))))
1218 
    };
957fee30 Andy McCall ↶ 1219 
    let by_node = base_snapshot.tree.by_node();
1220 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
1221 
    for op in &wc.tree_ops {
75167672 Andy McCall 1222 
        if let TreeOp::AddNode {
1223 
            node,
1224 
            parent,
1225 
            name,
1226 
            kind,
1227 
            ..
1228 
        } = op
1229 
        {
957fee30 Andy McCall ↶ 1230 
            new_names.insert(*node, (*parent, name.clone(), *kind));
1231 
        }
1232 
    }
1233 
    let mut nodes: BTreeSet<NodeId> = BTreeSet::new();
75167672 Andy McCall 1234 
    for n in wc
1235 
        .tree_ops
1236 
        .iter()
1237 
        .map(|o| o.node())
1238 
        .chain(wc.edits.iter().map(|e| e.node()))
1239 
    {
957fee30 Andy McCall ↶ 1240 
        if path_of_node(&n, &by_node, &new_names).is_some_and(|p| selected(&p)) {
1241 
            nodes.insert(n);
1242 
        }
1243 
    }
c2e1d090 Andy McCall 1244 
    // The nodes shown as one with a selected node are retired with it.
1245 
    for (dup, shown) in &base_snapshot.tree.unified {
1246 
        if nodes.contains(shown) {
1247 
            nodes.insert(*dup);
1248 
        }
1249 
    }
957fee30 Andy McCall ↶ 1250 
    // New parents of selected new nodes come along.
1251 
    let mut changed = true;
1252 
    while changed {
1253 
        changed = false;
1254 
        for (n, (parent, _, _)) in &new_names {
1255 
            if nodes.contains(n) && new_names.contains_key(parent) && nodes.insert(*parent) {
1256 
                changed = true;
1257 
            }
1258 
        }
1259 
    }
75167672 Andy McCall 1260 
    let ops = wc
1261 
        .tree_ops
1262 
        .iter()
1263 
        .filter(|o| nodes.contains(&o.node()))
1264 
        .cloned()
1265 
        .collect();
1266 
    let edits = wc
1267 
        .edits
1268 
        .iter()
1269 
        .filter(|e| nodes.contains(&e.node()))
1270 
        .cloned()
1271 
        .collect();
957fee30 Andy McCall ↶ 1272 
    (ops, edits)
1273 
}
1274
1275 
/// The hunks `indices` name plus whatever they refer to within the working
1276 
/// copy (the parent directory of an added node, the insert that introduced
1277 
/// an anchor line), renumbered so the result is a well-formed change.
c2e1d090 Andy McCall 1278 
/// The nodes `unified` shows as one with another (a base's
1279 
/// [`TreeView::unified`](crate::tree::TreeView::unified)) are retired with
1280 
/// any hunk of the node shown.
1281 
fn select_hunks(
1282 
    wc: &Change,
1283 
    indices: &[usize],
1284 
    unified: &BTreeMap<NodeId, NodeId>,
1285 
) -> (Vec<TreeOp>, Vec<Edit>) {
957fee30 Andy McCall ↶ 1286 
    let n_ops = wc.tree_ops.len();
1287 
    let mut ops_sel = vec![false; n_ops];
1288 
    let mut edits_sel = vec![false; wc.edits.len()];
1289 
    for &i in indices {
1290 
        if i < n_ops {
1291 
            ops_sel[i] = true;
1292 
        } else if i - n_ops < wc.edits.len() {
1293 
            edits_sel[i - n_ops] = true;
1294 
        }
1295 
    }
1296 
    // Atom ranges each insert introduces, per node, in the working copy's numbering.
1297 
    let mut per_node: HashMap<NodeId, u32> = HashMap::new();
1298 
    let mut insert_ranges: Vec<Option<(NodeId, u32, u32)>> = Vec::new();
1299 
    for e in &wc.edits {
1300 
        insert_ranges.push(match e {
1301 
            Edit::Insert { node, lines, .. } => {
1302 
                let start = per_node.entry(*node).or_default();
1303 
                let r = (*node, *start, *start + lines.len() as u32);
1304 
                *start += lines.len() as u32;
1305 
                Some(r)
1306 
            }
1307 
            _ => None,
1308 
        });
1309 
    }
75167672 Andy McCall 1310 
    let add_index: HashMap<NodeId, usize> = wc
1311 
        .tree_ops
1312 
        .iter()
1313 
        .enumerate()
1314 
        .filter_map(|(i, o)| {
1315 
            if let TreeOp::AddNode { node, .. } = o {
1316 
                Some((*node, i))
1317 
            } else {
1318 
                None
1319 
            }
1320 
        })
1321 
        .collect();
1322 
    let insert_for = |node: &NodeId, index: u32| -> Option<usize> {
1323 
        insert_ranges
1324 
            .iter()
1325 
            .position(|r| matches!(r, Some((n, a, b)) if n == node && *a <= index && index < *b))
1326 
    };
957fee30 Andy McCall ↶ 1327 
    // Closure.
1328 
    let mut changed = true;
1329 
    while changed {
1330 
        changed = false;
1331 
        for i in 0..n_ops {
1332 
            if !ops_sel[i] {
1333 
                continue;
1334 
            }
d8c086ca Andy McCall ↶ 1335 
            if let TreeOp::AddNode { node, parent, .. } = &wc.tree_ops[i] {
957fee30 Andy McCall ↶ 1336 
                if let Some(&p) = add_index.get(parent) {
1337 
                    if !ops_sel[p] {
1338 
                        ops_sel[p] = true;
1339 
                        changed = true;
1340 
                    }
1341 
                }
d8c086ca Andy McCall ↶ 1342 
                // A new node's content belongs with its addition: nobody
1343 
                // wants an empty file committed apart from its lines.
1344 
                for (j, e) in wc.edits.iter().enumerate() {
1345 
                    if e.node() == *node && !edits_sel[j] {
1346 
                        edits_sel[j] = true;
1347 
                        changed = true;
1348 
                    }
1349 
                }
957fee30 Andy McCall ↶ 1350 
            }
1351 
        }
1352 
        for j in 0..wc.edits.len() {
1353 
            if !edits_sel[j] {
1354 
                continue;
1355 
            }
1356 
            let e = &wc.edits[j];
1357 
            if let Some(&a) = add_index.get(&e.node()) {
1358 
                if !ops_sel[a] {
1359 
                    ops_sel[a] = true;
1360 
                    changed = true;
1361 
                }
1362 
            }
75167672 Andy McCall 1363 
            if let Edit::Insert {
1364 
                node,
1365 
                after,
1366 
                before,
1367 
                ..
1368 
            } = e
1369 
            {
957fee30 Andy McCall ↶ 1370 
                for r in [after, before] {
1371 
                    if let AtomRef::Atom { change, index } = r {
1372 
                        if change.is_zero() {
1373 
                            if let Some(k) = insert_for(node, *index) {
1374 
                                if !edits_sel[k] {
1375 
                                    edits_sel[k] = true;
1376 
                                    changed = true;
1377 
                                }
1378 
                            }
1379 
                        }
1380 
                    }
c2e1d090 Andy McCall 1381 
                }
1382 
            }
1383 
        }
1384 
    }
1385 
    // The retirement of a node shown as one with a selected node.
1386 
    if !unified.is_empty() {
1387 
        let chosen: HashSet<NodeId> = (0..n_ops)
1388 
            .filter(|i| ops_sel[*i])
1389 
            .map(|i| wc.tree_ops[i].node())
1390 
            .chain(
1391 
                (0..wc.edits.len())
1392 
                    .filter(|j| edits_sel[*j])
1393 
                    .map(|j| wc.edits[j].node()),
1394 
            )
1395 
            .collect();
1396 
        for (i, o) in wc.tree_ops.iter().enumerate() {
1397 
            if let TreeOp::RemoveNode { node, .. } = o {
1398 
                if unified.get(node).is_some_and(|s| chosen.contains(s)) {
1399 
                    ops_sel[i] = true;
957fee30 Andy McCall ↶ 1400 
                }
1401 
            }
1402 
        }
1403 
    }
1404 
    // Renumber self-references: old atom index → new index among selected inserts.
1405 
    let mut remap: HashMap<(NodeId, u32), u32> = HashMap::new();
1406 
    let mut new_counts: HashMap<NodeId, u32> = HashMap::new();
1407 
    for (j, r) in insert_ranges.iter().enumerate() {
1408 
        if let (true, Some((node, a, b))) = (edits_sel[j], r) {
1409 
            let next = new_counts.entry(*node).or_default();
1410 
            for old in *a..*b {
1411 
                remap.insert((*node, old), *next);
1412 
                *next += 1;
1413 
            }
1414 
        }
1415 
    }
1416 
    let map_ref = |r: AtomRef, node: &NodeId| -> AtomRef {
1417 
        match r {
75167672 Andy McCall 1418 
            AtomRef::Atom { change, index } if change.is_zero() => AtomRef::Atom {
1419 
                change,
1420 
                index: remap.get(&(*node, index)).copied().unwrap_or(index),
1421 
            },
957fee30 Andy McCall ↶ 1422 
            other => other,
1423 
        }
1424 
    };
75167672 Andy McCall 1425 
    let ops = wc
1426 
        .tree_ops
1427 
        .iter()
1428 
        .enumerate()
1429 
        .filter(|(i, _)| ops_sel[*i])
1430 
        .map(|(_, o)| o.clone())
1431 
        .collect();
957fee30 Andy McCall ↶ 1432 
    let edits = wc
1433 
        .edits
1434 
        .iter()
1435 
        .enumerate()
1436 
        .filter(|(j, _)| edits_sel[*j])
1437 
        .map(|(_, e)| match e {
75167672 Andy McCall 1438 
            Edit::Insert {
1439 
                node,
1440 
                after,
1441 
                before,
1442 
                lines,
1443 
            } => Edit::Insert {
1444 
                node: *node,
1445 
                after: map_ref(*after, node),
1446 
                before: map_ref(*before, node),
1447 
                lines: lines.clone(),
1448 
            },
957fee30 Andy McCall ↶ 1449 
            other => other.clone(),
1450 
        })
1451 
        .collect();
1452 
    (ops, edits)
1453 
}
1454
1455 
/// Render each tree operation and edit of `change` on its own, numbered.
1456 
pub fn render_hunks(change: &Change, base: &State, base_snapshot: &Snapshot) -> Vec<Hunk> {
1457 
    let by_node = base_snapshot.tree.by_node();
1458 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
1459 
    for op in &change.tree_ops {
75167672 Andy McCall 1460 
        if let TreeOp::AddNode {
1461 
            node,
1462 
            parent,
1463 
            name,
1464 
            kind,
1465 
            ..
1466 
        } = op
1467 
        {
957fee30 Andy McCall ↶ 1468 
            new_names.insert(*node, (*parent, name.clone(), *kind));
1469 
        }
1470 
    }
1471 
    let mut out = Vec::new();
1472 
    for (i, op) in change.tree_ops.iter().enumerate() {
c2e1d090 Andy McCall 1473 
        // A retired duplicate goes with the node shown: no hunk of its own.
1474 
        if is_unified_retirement(op, base_snapshot) {
1475 
            continue;
1476 
        }
75167672 Andy McCall 1477 
        let one = Change {
1478 
            tree_ops: vec![op.clone()],
1479 
            edits: vec![],
1480 
            ..change.clone()
1481 
        };
957fee30 Andy McCall ↶ 1482 
        let path = path_of_node(&op.node(), &by_node, &new_names).unwrap_or_default();
75167672 Andy McCall 1483 
        out.push(Hunk {
1484 
            index: i,
1485 
            path,
1486 
            text: render_change(&one, base, base_snapshot),
1487 
        });
957fee30 Andy McCall ↶ 1488 
    }
1489 
    for (j, e) in change.edits.iter().enumerate() {
75167672 Andy McCall 1490 
        let one = Change {
1491 
            tree_ops: vec![],
1492 
            edits: vec![e.clone()],
1493 
            ..change.clone()
1494 
        };
957fee30 Andy McCall ↶ 1495 
        let path = path_of_node(&e.node(), &by_node, &new_names).unwrap_or_default();
75167672 Andy McCall 1496 
        out.push(Hunk {
1497 
            index: change.tree_ops.len() + j,
1498 
            path,
1499 
            text: render_change(&one, base, base_snapshot),
1500 
        });
957fee30 Andy McCall ↶ 1501 
    }
1502 
    out
1503 
}
1504
c6aa5bd9 Andy McCall ↶ 1505 
/// One file's part of a diff, for `oxy diff --json`.
1506 
#[derive(Clone, Debug)]
1507 
pub struct FileDiff {
1508 
    pub path: String,
1509 
    /// `added`, `removed`, `modified`, `renamed`, `mode`, `replaced`.
1510 
    pub kind: String,
1511 
    pub hunks: Vec<DiffHunk>,
1512 
    /// For renames, the previous path.
1513 
    pub from: Option<String>,
1514 
}
1515
1516 
/// Lines inserted after `base_line` (0 = before the first line) and lines
1517 
/// removed, with their base line numbers.
1518 
#[derive(Clone, Debug, Default)]
1519 
pub struct DiffHunk {
1520 
    pub base_line: u32,
1521 
    pub added: Vec<Vec<u8>>,
1522 
    pub removed: Vec<(u32, Vec<u8>)>,
1523 
}
1524
1525 
impl Repo {
1526 
    /// The content of `path` in a state: `heads`, or the working copy's base.
1527 
    pub fn file_at(&self, path: &str, heads: Option<&[ChangeHash]>) -> Result<Vec<u8>> {
19303b4a Andy McCall ↶ 1528 
        // The state's history and tree, then the one file's graph: nothing
1529 
        // else is built or rendered (oxygen#92).
1530 
        // The working copy's base, and a state named by its heads (`-r
1531 
        // main` on main), come from the workspace's state index.
1532 
        let base = if self.is_bare() {
1533 
            None
1534 
        } else {
1535 
            Some(self.working_copy()?.1.deps)
1536 
        };
1537 
        let same = |a: &[ChangeHash], b: &[ChangeHash]| {
1538 
            let (mut a, mut b) = (a.to_vec(), b.to_vec());
1539 
            a.sort();
1540 
            b.sort();
1541 
            a == b
1542 
        };
1543 
        let mut state = match (heads, base) {
1544 
            (Some(h), Some(b)) if same(h, &b) => self.load_state_indexed(&b)?,
1545 
            (Some(h), _) => self.load_state_history(h)?,
1546 
            (None, Some(b)) => self.load_state_indexed(&b)?,
1547 
            (None, None) => return fail("this is a bare repository: it has no working copy"),
1548 
        };
c2e1d090 Andy McCall 1549 
        let view = self.tree_view(&mut state)?;
19303b4a Andy McCall ↶ 1550 
        let Some(node) = view.entries.get(path).map(|e| e.node) else {
1551 
            return fail(format!("{path} is not in that state"));
c6aa5bd9 Andy McCall ↶ 1552 
        };
19303b4a Andy McCall ↶ 1553 
        self.load_graphs_of(&mut state, &[node])?;
1554 
        let snapshot = state.materialize_paths(view, [path]);
75167672 Andy McCall 1555 
        let Some(f) = snapshot.files.get(path) else {
1556 
            return fail(format!("{path} is not in that state"));
1557 
        };
c6aa5bd9 Andy McCall ↶ 1558 
        match &f.content {
1559 
            FileContent::Dir => fail(format!("{path} is a directory")),
1560 
            FileContent::Text { bytes, .. } => Ok(bytes.clone()),
1561 
            FileContent::Opaque { blobs } => match blobs.first() {
1562 
                Some(b) => self.read_blob(b),
1563 
                None => Ok(vec![]),
1564 
            },
1565 
        }
1566 
    }
1567
1568 
    /// Restore `paths` (files or directories) to the working copy's base:
1569 
    /// edits are undone, additions removed, removals put back. Returns the
1570 
    /// number of paths touched.
1571 
    pub fn discard(&self, paths: &[String]) -> Result<usize> {
6caacfff Andy McCall ↶ 1572 
        let _wc = self.lock_working_copy()?;
8ab265dc Andy McCall ↶ 1573 
        let snap = self.snapshot_status("oxy discard")?;
75167672 Andy McCall 1574 
        let under = |p: &str| {
1575 
            paths
1576 
                .iter()
1577 
                .any(|x| p == x || p.starts_with(&format!("{}/", x.trim_end_matches('/'))))
1578 
        };
c6aa5bd9 Andy McCall ↶ 1579 
        let attrs = Repo::state_attributes(&snap.base_snapshot);
1580 
        let mut n = 0;
1581 
        for s in &snap.statuses {
1582 
            if !under(&s.path) {
1583 
                continue;
1584 
            }
1585 
            match snap.base_snapshot.files.get(&s.path) {
1586 
                None => {
1587 
                    // Added (or the new name of a rename): remove it.
1588 
                    let full = self.root.join(&s.path);
1589 
                    if full.is_dir() {
1590 
                        let _ = fs::remove_dir_all(&full);
1591 
                    } else {
1592 
                        let _ = fs::remove_file(&full);
1593 
                    }
1594 
                    n += 1;
1595 
                }
1596 
                Some(f) => {
1597 
                    match &f.content {
1598 
                        FileContent::Dir => fs::create_dir_all(self.root.join(&s.path))?,
1599 
                        FileContent::Text { bytes, .. } => {
75167672 Andy McCall 1600 
                            let on_disk: std::borrow::Cow<[u8]> = if attrs.lookup(&s.path).eol
1601 
                                == Some(crate::attributes::Eol::Crlf)
1602 
                            {
1603 
                                crate::attributes::to_crlf(bytes).into()
1604 
                            } else {
1605 
                                bytes.as_slice().into()
1606 
                            };
c6aa5bd9 Andy McCall ↶ 1607 
                            self.write_file(&s.path, &on_disk, f.mode.exec)?;
1608 
                        }
1609 
                        FileContent::Opaque { blobs } => {
1610 
                            let bytes = match blobs.first() {
1611 
                                Some(b) => self.read_blob(b)?,
1612 
                                None => vec![],
1613 
                            };
1614 
                            if f.kind == NodeKind::Symlink {
1615 
                                self.write_symlink(&s.path, &bytes)?;
1616 
                            } else if f.kind == NodeKind::File {
1617 
                                self.write_file(&s.path, &bytes, f.mode.exec)?;
1618 
                            }
1619 
                        }
1620 
                    }
1621 
                    n += 1;
1622 
                }
1623 
            }
1624 
        }
1625 
        // A rename's old name is not in the statuses; put it back too.
1626 
        for (p, f) in &snap.base_snapshot.files {
75167672 Andy McCall 1627 
            if under(p)
1628 
                && !self.root.join(p).exists()
1629 
                && f.kind != NodeKind::Dir
1630 
                && f.kind != NodeKind::Subrepo
1631 
                && !snap.statuses.iter().any(|s| s.path == *p)
1632 
            {
c6aa5bd9 Andy McCall ↶ 1633 
                let bytes = match &f.content {
1634 
                    FileContent::Text { bytes, .. } => bytes.clone(),
75167672 Andy McCall 1635 
                    FileContent::Opaque { blobs } => blobs
1636 
                        .first()
1637 
                        .map(|b| self.read_blob(b))
1638 
                        .transpose()?
1639 
                        .unwrap_or_default(),
c6aa5bd9 Andy McCall ↶ 1640 
                    FileContent::Dir => continue,
1641 
                };
75167672 Andy McCall 1642 
                if f.kind == NodeKind::Symlink {
1643 
                    self.write_symlink(p, &bytes)?
1644 
                } else {
1645 
                    self.write_file(p, &bytes, f.mode.exec)?
1646 
                }
c6aa5bd9 Andy McCall ↶ 1647 
                n += 1;
1648 
            }
1649 
        }
1650 
        self.drop_manifest();
1651 
        Ok(n)
1652 
    }
1653
1654 
    /// Undo the working-copy hunks `indices` (as `hunks` numbers them) on
1655 
    /// disk, keeping the others: the files are rewritten from the base plus
1656 
    /// the remaining hunks.
1657 
    pub fn discard_hunks(&self, indices: &[usize]) -> Result<usize> {
6caacfff Andy McCall ↶ 1658 
        let _wc = self.lock_working_copy()?;
c6aa5bd9 Andy McCall ↶ 1659 
        let (snap, hunks) = self.hunks()?;
75167672 Andy McCall 1660 
        let keep: Vec<usize> = hunks
1661 
            .iter()
1662 
            .map(|h| h.index)
1663 
            .filter(|i| !indices.contains(i))
1664 
            .collect();
c2e1d090 Andy McCall 1665 
        let (ops, edits) = select_hunks(&snap.wc, &keep, &snap.base_snapshot.tree.unified);
c6aa5bd9 Andy McCall ↶ 1666 
        let dropped_nodes: BTreeSet<NodeId> = hunks
1667 
            .iter()
1668 
            .filter(|h| indices.contains(&h.index))
75167672 Andy McCall 1669 
            .map(|h| {
1670 
                if h.index < snap.wc.tree_ops.len() {
1671 
                    snap.wc.tree_ops[h.index].node()
1672 
                } else {
1673 
                    snap.wc.edits[h.index - snap.wc.tree_ops.len()].node()
1674 
                }
1675 
            })
c6aa5bd9 Andy McCall ↶ 1676 
            .collect();
1677 
        let mut partial = snap.wc.clone();
1678 
        partial.tree_ops = ops;
1679 
        partial.edits = edits;
1680 
        partial.normalize();
1681 
        let mut st = snap.base.clone();
1682 
        st.add_unchecked(partial);
8ab265dc Andy McCall ↶ 1683 
        // Every file of the target is rendered: build the graphs the
1684 
        // partial snapshot left out.
1685 
        self.load_all_graphs(&mut st)?;
c6aa5bd9 Andy McCall ↶ 1686 
        let target = st.materialize();
1687 
        let by_node_target = target.tree.by_node();
1688 
        let by_node_base = snap.base_snapshot.tree.by_node();
1689 
        let attrs = Repo::state_attributes(&target);
1690 
        let mut n = 0;
1691 
        for node in dropped_nodes {
1692 
            match by_node_target.get(&node) {
1693 
                Some(e) => {
1694 
                    let f = &target.files[&e.path];
1695 
                    match &f.content {
1696 
                        FileContent::Dir => {}
1697 
                        FileContent::Text { bytes, .. } => {
75167672 Andy McCall 1698 
                            let on_disk: std::borrow::Cow<[u8]> = if attrs.lookup(&e.path).eol
1699 
                                == Some(crate::attributes::Eol::Crlf)
1700 
                            {
1701 
                                crate::attributes::to_crlf(bytes).into()
1702 
                            } else {
1703 
                                bytes.as_slice().into()
1704 
                            };
c6aa5bd9 Andy McCall ↶ 1705 
                            self.write_file(&e.path, &on_disk, f.mode.exec)?;
1706 
                        }
1707 
                        FileContent::Opaque { blobs } => {
75167672 Andy McCall 1708 
                            let bytes = blobs
1709 
                                .first()
1710 
                                .map(|b| self.read_blob(b))
1711 
                                .transpose()?
1712 
                                .unwrap_or_default();
1713 
                            if f.kind == NodeKind::Symlink {
1714 
                                self.write_symlink(&e.path, &bytes)?
1715 
                            } else {
1716 
                                self.write_file(&e.path, &bytes, f.mode.exec)?
1717 
                            }
c6aa5bd9 Andy McCall ↶ 1718 
                        }
1719 
                    }
1720 
                    // A discarded rename: the new name goes.
75167672 Andy McCall 1721 
                    if let Some(old_path) = snap
1722 
                        .statuses
1723 
                        .iter()
1724 
                        .find(|s| {
1725 
                            s.status == ChangeStatus::Renamed && !target.files.contains_key(&s.path)
1726 
                        })
1727 
                        .map(|s| s.path.clone())
1728 
                    {
c6aa5bd9 Andy McCall ↶ 1729 
                        let _ = fs::remove_file(self.root.join(old_path));
1730 
                    }
1731 
                    n += 1;
1732 
                }
1733 
                None => {
1734 
                    // The node exists in neither target nor base with a live path: an addition to remove.
75167672 Andy McCall 1735 
                    if let Some(added) = snap.statuses.iter().find(|s| {
1736 
                        s.status == ChangeStatus::Added
1737 
                            && !target.files.contains_key(&s.path)
1738 
                            && !by_node_base.contains_key(&node)
1739 
                    }) {
c6aa5bd9 Andy McCall ↶ 1740 
                        let full = self.root.join(&added.path);
75167672 Andy McCall 1741 
                        if full.is_dir() {
1742 
                            let _ = fs::remove_dir_all(&full);
1743 
                        } else {
1744 
                            let _ = fs::remove_file(&full);
1745 
                        }
c6aa5bd9 Andy McCall ↶ 1746 
                        n += 1;
1747 
                    }
1748 
                }
1749 
            }
1750 
        }
1751 
        self.drop_manifest();
1752 
        Ok(n)
1753 
    }
1754
1755 
    /// The diff of the working copy (`None`) or of a change, per file with
1756 
    /// line-numbered hunks (spec: `oxy diff --json`).
1757 
    pub fn diff_files(&self, change: Option<&ChangeHash>) -> Result<Vec<FileDiff>> {
8ab265dc Andy McCall ↶ 1758 
        let (c, base, base_snapshot) = match change {
c6aa5bd9 Andy McCall ↶ 1759 
            Some(h) => {
1760 
                let c = self.store.get_change(h)?;
8ab265dc Andy McCall ↶ 1761 
                let (base, snapshot) = self.change_base(&c)?;
1762 
                (c, base, snapshot)
c6aa5bd9 Andy McCall ↶ 1763 
            }
74b4a3c6 Andy McCall ↶ 1764 
            None => match self.working_copy_with_base_unless_unchanged()? {
1765 
                Some(all) => all,
1766 
                None => return Ok(Vec::new()),
1767 
            },
c6aa5bd9 Andy McCall ↶ 1768 
        };
1769 
        Ok(diff_files_of(&c, &base, &base_snapshot))
e226a6cc Andy McCall ↶ 1770 
    }
1771
1772 
    /// The snapshot a command would take, without taking it, when the
1773 
    /// status cache holds: every file is as the last snapshot left it, so
1774 
    /// that snapshot's working-copy change is the answer. Its base comes
1775 
    /// from the state index, the `.oxyattributes` files from disk. `None`
1776 
    /// when a snapshot must be taken, and when there is nothing to commit
1777 
    /// (the snapshot then says so). A conflicted base never gets here: the
1778 
    /// cache does not hold for one.
1779 
    pub(crate) fn snapshot_from_cache(&self) -> Result<Option<SnapshotResult>> {
1780 
        let Some(statuses) = self.status_by_manifest()? else {
1781 
            return Ok(None);
1782 
        };
1783 
        if statuses.is_empty() {
1784 
            return Ok(None);
1785 
        }
1786 
        let _t = crate::timing::phase("snapshot: from the status cache");
1787 
        let (wc_hash, wc) = self.working_copy()?;
1788 
        let base = self.load_state_indexed(&wc.deps)?;
1789 
        let (base, base_snapshot) = self.render_base_for(&wc, base)?;
1790 
        let attribute_files: Vec<(String, Vec<u8>)> = match self.read_manifest()? {
1791 
            Some((_, _, m)) => m
1792 
                .entries
1793 
                .keys()
1794 
                .filter(|p| is_attributes(p))
1795 
                .filter_map(|p| fs::read(self.root.join(p)).ok().map(|b| (p.clone(), b)))
1796 
                .collect(),
1797 
            None => Vec::new(),
1798 
        };
1799 
        let attributes = crate::attributes::Attributes::from_files(
1800 
            attribute_files
1801 
                .iter()
1802 
                .map(|(p, b)| (p.as_str(), b.as_slice())),
1803 
        );
1804 
        Ok(Some(SnapshotResult {
1805 
            wc_hash,
1806 
            wc,
1807 
            base,
1808 
            base_snapshot,
1809 
            statuses,
1810 
            changed: false,
1811 
            resolves: false,
1812 
            contributors: Vec::new(),
1813 
            attributes,
1814 
            untracked_nested: self.untracked_nested.borrow().clone(),
236f35c0 Andy McCall ↶ 1815 
            manifest: None,
e226a6cc Andy McCall ↶ 1816 
        }))
77445673 Andy McCall ↶ 1817 
    }
1818
dd6147dc Andy McCall ↶ 1819 
    /// The working-copy change with its base state and a snapshot of the
1820 
    /// base, for showing what the working copy changes ([`render_change`],
1821 
    /// [`diff_files_of`]). While the status cache holds (every file as the
1822 
    /// last snapshot left it), the stored working-copy change is the
1823 
    /// answer, no snapshot is taken, and the base's snapshot holds its tree
1824 
    /// but no file; otherwise a snapshot makes all three.
77445673 Andy McCall ↶ 1825 
    pub fn working_copy_with_base(&self) -> Result<(Change, State, Snapshot)> {
74b4a3c6 Andy McCall ↶ 1826 
        match self.working_copy_with_base_unless_unchanged()? {
1827 
            Some(all) => Ok(all),
1828 
            None => {
1829 
                let (_, wc) = self.working_copy()?;
1830 
                let base = self.load_state_indexed(&wc.deps)?;
1831 
                let (base, snapshot) = self.tree_base_for(&wc, base)?;
1832 
                Ok((wc, base, snapshot))
1833 
            }
1834 
        }
1835 
    }
1836
1837 
    /// [`Repo::working_copy_with_base`], or `None` when the status cache
1838 
    /// holds and the working-copy change changes nothing: its diff is
1839 
    /// empty, and the base is not loaded for it (oxygen#94).
1840 
    pub fn working_copy_with_base_unless_unchanged(
1841 
        &self,
1842 
    ) -> Result<Option<(Change, State, Snapshot)>> {
77445673 Andy McCall ↶ 1843 
        if self.status_by_manifest()?.is_some() {
1844 
            let (_, wc) = self.working_copy()?;
74b4a3c6 Andy McCall ↶ 1845 
            if wc.tree_ops.is_empty() && wc.edits.is_empty() {
1846 
                return Ok(None);
1847 
            }
77445673 Andy McCall ↶ 1848 
            let base = self.load_state_indexed(&wc.deps)?;
dd6147dc Andy McCall ↶ 1849 
            let (base, snapshot) = self.tree_base_for(&wc, base)?;
74b4a3c6 Andy McCall ↶ 1850 
            return Ok(Some((wc, base, snapshot)));
77445673 Andy McCall ↶ 1851 
        }
1852 
        let snap = self.snapshot_status("oxy diff")?;
74b4a3c6 Andy McCall ↶ 1853 
        Ok(Some((snap.wc, snap.base, snap.base_snapshot)))
c6aa5bd9 Andy McCall ↶ 1854 
    }
1855
1856 
    /// The log restricted to changes touching `path` (its node, so renames
1857 
    /// are followed), on the current bookmark.
1858 
    pub fn log_for_path(&self, path: &str) -> Result<Vec<LogEntry>> {
e02bf0cd Andy McCall ↶ 1859 
        let heads = self.bookmark_heads()?;
1860 
        if self.is_bare() {
1861 
            return self.log_for_path_in(path, &heads);
1862 
        }
c6aa5bd9 Andy McCall ↶ 1863 
        let entries = self.log()?;
1864 
        let (_, wc) = self.working_copy()?;
170bb436 Andy McCall ↶ 1865 
        // The path's node from the trees alone: the base's (from the
1866 
        // workspace's state index), else the bookmark's. Nothing is
1867 
        // rendered.
c2e1d090 Andy McCall 1868 
        let node_in = |mut state: State| -> Result<Option<HashSet<NodeId>>> {
1869 
            Ok(nodes_at(&self.tree_view(&mut state)?, path))
1870 
        };
1871 
        let nodes = match node_in(self.load_state_indexed(&wc.deps)?)? {
170bb436 Andy McCall ↶ 1872 
            Some(n) => Some(n),
c2e1d090 Andy McCall 1873 
            None => node_in(self.load_state_history(&heads)?)?,
170bb436 Andy McCall ↶ 1874 
        };
c2e1d090 Andy McCall 1875 
        let Some(nodes) = nodes else {
75167672 Andy McCall 1876 
            return fail(format!("{path} is not tracked"));
1877 
        };
c2e1d090 Andy McCall 1878 
        Ok(self.entries_touching_up_to(entries, nodes, None))
e02bf0cd Andy McCall ↶ 1879 
    }
1880
1881 
    /// The log of the state with these heads restricted to changes touching
1882 
    /// `path` (its node, so renames are followed).
1883 
    pub fn log_for_path_in(&self, path: &str, heads: &[ChangeHash]) -> Result<Vec<LogEntry>> {
e8d38e47 Andy McCall ↶ 1884 
        self.log_for_path_of(&self.load_state_history(heads)?, path, None)
bc489b6c Andy McCall ↶ 1885 
    }
1886
1887 
    /// The changes of a loaded state touching `path`, newest first, at most
1888 
    /// `limit` of them. The path is found in the state's tree alone, so no
1889 
    /// file is rendered, and the walk stops once `limit` are found.
1890 
    pub fn log_for_path_of(
1891 
        &self,
1892 
        state: &State,
1893 
        path: &str,
1894 
        limit: Option<usize>,
1895 
    ) -> Result<Vec<LogEntry>> {
c2e1d090 Andy McCall 1896 
        let Some(nodes) = nodes_at(&state.view(), path) else {
75167672 Andy McCall 1897 
            return fail(format!("{path} is not in that state"));
1898 
        };
c2e1d090 Andy McCall 1899 
        Ok(self.entries_touching_up_to(self.log_of(state), nodes, limit))
bc489b6c Andy McCall ↶ 1900 
    }
1901
1902 
    fn entries_touching_up_to(
1903 
        &self,
1904 
        entries: Vec<LogEntry>,
c2e1d090 Andy McCall 1905 
        wanted: HashSet<NodeId>,
bc489b6c Andy McCall ↶ 1906 
        limit: Option<usize>,
1907 
    ) -> Vec<LogEntry> {
170bb436 Andy McCall ↶ 1908 
        let mut touched = self.touched();
1909 
        if limit.is_none() {
1910 
            let hashes: Vec<ChangeHash> = entries.iter().map(|e| e.hash).collect();
1911 
            touched.fill(self, &hashes);
1912 
        }
c6aa5bd9 Andy McCall ↶ 1913 
        let mut out = Vec::new();
1914 
        for e in entries {
bc489b6c Andy McCall ↶ 1915 
            if limit.is_some_and(|n| out.len() >= n) {
1916 
                break;
1917 
            }
170bb436 Andy McCall ↶ 1918 
            if touched.touches(self, &e.hash, &wanted) {
c6aa5bd9 Andy McCall ↶ 1919 
                out.push(e);
1920 
            }
1921 
        }
170bb436 Andy McCall ↶ 1922 
        touched.save();
e02bf0cd Andy McCall ↶ 1923 
        out
c6aa5bd9 Andy McCall ↶ 1924 
    }
c2e1d090 Andy McCall 1925 
}
1926
1927 
/// The node at `path` in `view`, with the nodes it shows as one with it
1928 
/// (identical additions, spec §9.2): the file's history is all of theirs.
1929 
fn nodes_at(view: &crate::tree::TreeView, path: &str) -> Option<HashSet<NodeId>> {
1930 
    let node = view.entries.get(path)?.node;
1931 
    let mut out = HashSet::from([node]);
1932 
    out.extend(
1933 
        view.unified
1934 
            .iter()
1935 
            .filter(|(_, shown)| **shown == node)
1936 
            .map(|(dup, _)| *dup),
1937 
    );
1938 
    Some(out)
c6aa5bd9 Andy McCall ↶ 1939 
}
1940
1941 
/// Per-file hunks of `change` against `base`.
1942 
pub fn diff_files_of(change: &Change, base: &State, base_snapshot: &Snapshot) -> Vec<FileDiff> {
1943 
    let by_node = base_snapshot.tree.by_node();
1944 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
1945 
    for op in &change.tree_ops {
75167672 Andy McCall 1946 
        if let TreeOp::AddNode {
1947 
            node,
1948 
            parent,
1949 
            name,
1950 
            kind,
1951 
            ..
1952 
        } = op
1953 
        {
c6aa5bd9 Andy McCall ↶ 1954 
            new_names.insert(*node, (*parent, name.clone(), *kind));
1955 
        }
1956 
    }
75167672 Andy McCall 1957 
    let path_of = |n: &NodeId| -> String {
1958 
        path_of_node(n, &by_node, &new_names).unwrap_or_else(|| format!("<node {}>", n.short(8)))
1959 
    };
c6aa5bd9 Andy McCall ↶ 1960 
    let mut files: BTreeMap<NodeId, FileDiff> = BTreeMap::new();
1961 
    let mut order: Vec<NodeId> = Vec::new();
1962 
    let mut entry = |files: &mut BTreeMap<NodeId, FileDiff>, n: NodeId, kind: &str| {
1963 
        if let std::collections::btree_map::Entry::Vacant(v) = files.entry(n) {
1964 
            order.push(n);
75167672 Andy McCall 1965 
            v.insert(FileDiff {
1966 
                path: path_of(&n),
1967 
                kind: kind.to_string(),
1968 
                hunks: vec![],
1969 
                from: None,
1970 
            });
c6aa5bd9 Andy McCall ↶ 1971 
        }
1972 
    };
c2e1d090 Andy McCall 1973 
    for op in change
1974 
        .tree_ops
1975 
        .iter()
1976 
        .filter(|o| !is_unified_retirement(o, base_snapshot))
1977 
    {
c6aa5bd9 Andy McCall ↶ 1978 
        match op {
1979 
            TreeOp::AddNode { node, kind, .. } => {
1980 
                if *kind != NodeKind::Dir {
1981 
                    entry(&mut files, *node, "added");
1982 
                }
1983 
            }
1984 
            TreeOp::RemoveNode { node, .. } => {
1985 
                entry(&mut files, *node, "removed");
1986 
                if let Some(e) = by_node.get(node) {
1987 
                    files.get_mut(node).unwrap().path = e.path.clone();
1988 
                }
1989 
            }
75167672 Andy McCall 1990 
            TreeOp::Move {
1991 
                node, parent, name, ..
1992 
            } => {
c6aa5bd9 Andy McCall ↶ 1993 
                entry(&mut files, *node, "renamed");
1994 
                let f = files.get_mut(node).unwrap();
1995 
                f.from = by_node.get(node).map(|e| e.path.clone());
75167672 Andy McCall 1996 
                let dir = if parent.is_root() {
1997 
                    String::new()
1998 
                } else {
1999 
                    path_of_node(parent, &by_node, &new_names).unwrap_or_default()
2000 
                };
2001 
                f.path = if dir.is_empty() {
2002 
                    name.clone()
2003 
                } else {
2004 
                    format!("{dir}/{name}")
2005 
                };
c6aa5bd9 Andy McCall ↶ 2006 
            }
2007 
            TreeOp::SetMode { node, .. } => entry(&mut files, *node, "mode"),
2008 
        }
2009 
    }
2010 
    // Base line numbers per node, 1-based.
2011 
    let mut line_of: HashMap<NodeId, HashMap<(ChangeHash, u32), u32>> = HashMap::new();
2012 
    let mut lines_for = |n: &NodeId| -> HashMap<(ChangeHash, u32), u32> {
75167672 Andy McCall 2013 
        line_of
2014 
            .entry(*n)
2015 
            .or_insert_with(|| {
2016 
                base.live_lines(n)
2017 
                    .into_iter()
2018 
                    .enumerate()
2019 
                    .map(|(i, (k, _))| (k, i as u32 + 1))
2020 
                    .collect()
2021 
            })
2022 
            .clone()
c6aa5bd9 Andy McCall ↶ 2023 
    };
2024 
    for e in &change.edits {
2025 
        let n = e.node();
2026 
        let is_new = new_names.contains_key(&n);
2027 
        entry(&mut files, n, if is_new { "added" } else { "modified" });
2028 
        let f = files.get_mut(&n).unwrap();
2029 
        match e {
2030 
            Edit::Insert { after, lines, .. } => {
2031 
                let map = lines_for(&n);
2032 
                let base_line = match after {
2033 
                    AtomRef::Start => 0,
2034 
                    AtomRef::End => map.len() as u32,
75167672 Andy McCall 2035 
                    AtomRef::Atom { change, index } if change.is_zero() => {
2036 
                        f.hunks.last().map(|h| h.base_line).unwrap_or(0)
2037 
                    }
2038 
                    AtomRef::Atom { change, index } => {
2039 
                        map.get(&(*change, *index)).copied().unwrap_or(0)
2040 
                    }
c6aa5bd9 Andy McCall ↶ 2041 
                };
75167672 Andy McCall 2042 
                f.hunks.push(DiffHunk {
2043 
                    base_line,
01a47cd7 Andy McCall ↶ 2044 
                    added: lines.to_vec(),
75167672 Andy McCall 2045 
                    removed: vec![],
2046 
                });
c6aa5bd9 Andy McCall ↶ 2047 
            }
2048 
            Edit::Delete { atoms, .. } => {
2049 
                let map = lines_for(&n);
2050 
                let mut removed = Vec::new();
2051 
                for r in atoms {
2052 
                    for i in r.from..r.to {
2053 
                        let line = map.get(&(r.change, i)).copied().unwrap_or(0);
75167672 Andy McCall 2054 
                        let text = base
2055 
                            .file(&n)
2056 
                            .and_then(|g| g.content(&(r.change, i)))
2057 
                            .unwrap_or(b"")
2058 
                            .to_vec();
c6aa5bd9 Andy McCall ↶ 2059 
                        removed.push((line, text));
2060 
                    }
2061 
                }
75167672 Andy McCall 2062 
                let base_line = removed
2063 
                    .first()
2064 
                    .map(|(l, _)| l.saturating_sub(1))
2065 
                    .unwrap_or(0);
2066 
                f.hunks.push(DiffHunk {
2067 
                    base_line,
2068 
                    added: vec![],
2069 
                    removed,
2070 
                });
c6aa5bd9 Andy McCall ↶ 2071 
            }
2072 
            Edit::Replace { .. } => {
2073 
                if !is_new {
2074 
                    f.kind = "replaced".into();
2075 
                }
2076 
            }
2077 
        }
2078 
    }
2079 
    order.into_iter().filter_map(|n| files.remove(&n)).collect()
2080 
}
2081
98286406 Andy McCall ↶ 2082 
/// Builds the tree ops and edits of a working-copy change against a base.
dfb23a3a Andy McCall ↶ 2083 
/// `BLAKE3("oxy node 1" 0 seed kind path)`, truncated: a node identifier
2084 
/// that the same seed and path always give (never the root's).
2085 
fn seeded_node(seed: &[u8; 32], path: &str, kind: NodeKind) -> NodeId {
2086 
    let mut h = blake3::Hasher::new();
2087 
    h.update(b"oxy node 1\0");
2088 
    h.update(seed);
2089 
    h.update(&[match kind {
2090 
        NodeKind::File => 1,
2091 
        NodeKind::Dir => 2,
2092 
        NodeKind::Symlink => 3,
2093 
        NodeKind::Subrepo => 4,
2094 
    }]);
2095 
    h.update(path.as_bytes());
2096 
    let mut id = [0u8; 16];
2097 
    id.copy_from_slice(&h.finalize().as_bytes()[..16]);
2098 
    if id == [0; 16] {
2099 
        id[15] = 1;
2100 
    }
2101 
    NodeId(id)
2102 
}
2103
98286406 Andy McCall ↶ 2104 
struct ChangeBuilder<'a> {
2105 
    base: &'a State,
2106 
    base_snapshot: &'a Snapshot,
2107 
    tree_ops: Vec<TreeOp>,
2108 
    edits: Vec<Edit>,
2109 
    statuses: Vec<PathStatus>,
2110 
    dir_nodes: HashMap<String, NodeId>,
2111 
    reuse: HashMap<String, (NodeId, NodeKind)>,
2112 
    /// Changes that contributed to conflicts this working copy resolves.
2113 
    contributors: BTreeSet<ChangeHash>,
2114 
    resolves: bool,
cdb5beba Andy McCall ↶ 2115 
    attrs: crate::attributes::Attributes,
b7f102fc Andy McCall ↶ 2116 
    blobs: HashMap<String, ObjectRef>,
f7510b63 Andy McCall ↶ 2117 
    /// Opaque files whose replace conflict the content settles.
2118 
    resolve: HashSet<String>,
dfb23a3a Andy McCall ↶ 2119 
    /// When set, new nodes' identifiers derive from it and their paths.
2120 
    node_seed: Option<[u8; 32]>,
a87c7b9d Andy McCall ↶ 2121 
    /// For a Git import, the base's dead directories by where they were
2122 
    /// (see `ensure_dir`), worked out when first needed.
2123 
    dead_dirs: Option<HashMap<(NodeId, String), (NodeId, crate::tree::OpKey)>>,
360aa74a Andy McCall ↶ 2124 
    /// Text edits of edited files, worked out in parallel beforehand
2125 
    /// (`diff_in_parallel`); a file not here is diffed when it is updated.
2126 
    diffs: HashMap<String, Vec<Edit>>,
98286406 Andy McCall ↶ 2127 
}
2128
2129 
impl<'a> ChangeBuilder<'a> {
75167672 Andy McCall 2130 
    fn new(
2131 
        base: &'a State,
2132 
        base_snapshot: &'a Snapshot,
2133 
        reuse: HashMap<String, (NodeId, NodeKind)>,
2134 
    ) -> Self {
98286406 Andy McCall ↶ 2135 
        let mut dir_nodes: HashMap<String, NodeId> = HashMap::new();
2136 
        dir_nodes.insert(String::new(), NodeId::ROOT);
e83354cd Andy McCall ↶ 2137 
        for (p, e) in base_snapshot.tree.entries.iter() {
2306a4d2 Andy McCall ↶ 2138 
            if e.kind == NodeKind::Dir {
2139 
                dir_nodes.insert(p.clone(), e.node);
98286406 Andy McCall ↶ 2140 
            }
2141 
        }
75167672 Andy McCall 2142 
        ChangeBuilder {
2143 
            base,
2144 
            base_snapshot,
2145 
            tree_ops: vec![],
2146 
            edits: vec![],
2147 
            statuses: vec![],
2148 
            dir_nodes,
2149 
            reuse,
2150 
            contributors: BTreeSet::new(),
2151 
            resolves: false,
2152 
            attrs: crate::attributes::Attributes::default(),
2153 
            blobs: HashMap::new(),
f7510b63 Andy McCall ↶ 2154 
            resolve: HashSet::new(),
dfb23a3a Andy McCall ↶ 2155 
            node_seed: None,
a87c7b9d Andy McCall ↶ 2156 
            dead_dirs: None,
360aa74a Andy McCall ↶ 2157 
            diffs: HashMap::new(),
75167672 Andy McCall 2158 
        }
98286406 Andy McCall ↶ 2159 
    }
2160
360aa74a Andy McCall ↶ 2161 
    /// Diff the edited text files among `entries` against their base
2162 
    /// renderings at once, on every core, for `update_path` to take: the
2163 
    /// same diff it would make, only not one file after another.
2164 
    fn diff_in_parallel(&mut self, entries: &BTreeMap<String, FsEntry>) {
2165 
        use rayon::prelude::*;
2166 
        let files = &self.base_snapshot.files;
2167 
        let edited: Vec<(&String, &FsEntry, &crate::state::FileState)> = entries
2168 
            .iter()
2169 
            .filter_map(|(p, e)| {
2170 
                let x = files.get(p)?;
2171 
                (x.kind == e.kind && !e.unchanged && !self.opaque_for(p, e)).then_some((p, e, x))
2172 
            })
2173 
            .collect();
2174 
        // Below this, starting the threads costs more than it saves.
2175 
        if edited.len() < 16 {
2176 
            return;
2177 
        }
2178 
        let this = &*self;
2179 
        let attrs = &this.attrs;
2180 
        let diffs: Vec<(String, Vec<Edit>)> = edited
2181 
            .par_iter()
2182 
            .filter_map(|(p, e, x)| {
2183 
                let FileContent::Text {
2184 
                    bytes,
2185 
                    materialized,
2186 
                } = &x.content
2187 
                else {
2188 
                    return None;
2189 
                };
2190 
                if materialized.is_conflicted() {
2191 
                    return None;
2192 
                }
2193 
                let disk_text = if attrs.lookup(p).eol.is_some() {
2194 
                    std::sync::Arc::new(crate::attributes::normalize_eol(&e.content))
2195 
                } else {
2196 
                    e.content.clone()
2197 
                };
2198 
                if *bytes == *disk_text {
2199 
                    return None;
2200 
                }
2201 
                let keys = materialized.live_atoms();
2202 
                let lines = crate::diff::split_lines(bytes);
2203 
                (keys.len() == lines.len()).then(|| {
2204 
                    let edits = crate::diff::text_edits_of(x.node, &keys, &lines, &disk_text);
2205 
                    ((*p).clone(), edits)
2206 
                })
2207 
            })
2208 
            .collect();
2209 
        self.diffs = diffs.into_iter().collect();
2210 
    }
2211
cdb5beba Andy McCall ↶ 2212 
    fn with_attributes(mut self, attrs: crate::attributes::Attributes) -> Self {
2213 
        self.attrs = attrs;
2214 
        self
2215 
    }
2216
2217 
    /// Whether a file is opaque: forced by `binary`, forced text by `text`
2218 
    /// only when it has no NUL bytes, else sniffed.
b7f102fc Andy McCall ↶ 2219 
    /// The blob for an opaque entry: the cached one, else stored now.
75167672 Andy McCall 2220 
    fn blob_of(
2221 
        &mut self,
2222 
        repo: &Repo,
2223 
        path: &str,
2224 
        e: &FsEntry,
2225 
        nocompress: bool,
2226 
    ) -> Result<ObjectRef> {
b7f102fc Andy McCall ↶ 2227 
        let b = match e.blob {
2228 
            Some(b) => b,
2229 
            None => repo.store_blob_for(&e.content, nocompress)?,
2230 
        };
2231 
        self.blobs.insert(path.to_string(), b);
2232 
        Ok(b)
2233 
    }
2234
cdb5beba Andy McCall ↶ 2235 
    fn opaque_for(&self, path: &str, e: &FsEntry) -> bool {
b7f102fc Andy McCall ↶ 2236 
        if e.blob.is_some() {
2237 
            return true;
2238 
        }
4dee277c Andy McCall ↶ 2239 
        if e.kind == NodeKind::Symlink || e.kind == NodeKind::Subrepo {
cdb5beba Andy McCall ↶ 2240 
            return true;
2241 
        }
2242 
        match self.attrs.lookup(path).binary {
2243 
            Some(true) => true,
2244 
            Some(false) => is_binary(&e.content),
2245 
            None => is_binary(&e.content),
2246 
        }
2247 
    }
2248
2249 
    /// Text content as stored: `\r\n` normalized for `eol=` files.
40a40bb7 Andy McCall ↶ 2250 
    fn stored_text(
2251 
        &self,
2252 
        path: &str,
2253 
        content: &std::sync::Arc<Vec<u8>>,
2254 
    ) -> std::sync::Arc<Vec<u8>> {
75167672 Andy McCall 2255 
        if self.attrs.lookup(path).eol.is_some() {
40a40bb7 Andy McCall ↶ 2256 
            std::sync::Arc::new(crate::attributes::normalize_eol(content))
75167672 Andy McCall 2257 
        } else {
40a40bb7 Andy McCall ↶ 2258 
            content.clone()
dfb23a3a Andy McCall ↶ 2259 
        }
2260 
    }
2261
2262 
    /// The identifier of a node added at `path`: reused when asked, derived
2263 
    /// from the seed when there is one, random otherwise.
2264 
    fn new_node(&self, path: &str, kind: NodeKind) -> NodeId {
2265 
        if let Some(n) = self.reused_node(path, kind) {
2266 
            return n;
75167672 Andy McCall 2267 
        }
dfb23a3a Andy McCall ↶ 2268 
        match &self.node_seed {
2269 
            Some(seed) => seeded_node(seed, path, kind),
2270 
            None => NodeId::random(),
2271 
        }
cdb5beba Andy McCall ↶ 2272 
    }
2273
98286406 Andy McCall ↶ 2274 
    fn reused_node(&self, path: &str, kind: NodeKind) -> Option<NodeId> {
75167672 Andy McCall 2275 
        self.reuse
2276 
            .get(path)
2277 
            .filter(|(_, k)| *k == kind)
2278 
            .map(|(n, _)| *n)
98286406 Andy McCall ↶ 2279 
    }
2280
a87c7b9d Andy McCall ↶ 2281 
    /// A dead directory last placed at (`parent`, `name`) in the base.
2282 
    fn revivable_dir(
2283 
        &mut self,
2284 
        parent: NodeId,
2285 
        name: &str,
2286 
    ) -> Option<(NodeId, crate::tree::OpKey)> {
2287 
        let base = self.base;
2288 
        self.dead_dirs
2289 
            .get_or_insert_with(|| base.tree().dead_dirs())
2290 
            .get(&(parent, name.to_string()))
2291 
            .copied()
2292 
    }
2293
98286406 Andy McCall ↶ 2294 
    fn ensure_dir(&mut self, path: &str) -> NodeId {
2295 
        if let Some(n) = self.dir_nodes.get(path) {
2296 
            return *n;
2297 
        }
2298 
        let mut parent = NodeId::ROOT;
2299 
        let mut prefix = String::new();
2300 
        for comp in path.split('/') {
2301 
            if !prefix.is_empty() {
2302 
                prefix.push('/');
2303 
            }
2304 
            prefix.push_str(comp);
a87c7b9d Andy McCall ↶ 2305 
            let known = self.dir_nodes.get(&prefix).copied();
2306 
            let revive = match known {
2307 
                Some(_) => None,
2308 
                None if self.node_seed.is_some() => self.revivable_dir(parent, comp),
2309 
                None => None,
2310 
            };
2311 
            parent = match (known, revive) {
2312 
                (Some(n), _) => n,
2313 
                (None, Some((node, old))) => {
2314 
                    // Git has no identity for a directory: one that comes
2315 
                    // back where it was is the same directory, brought back
2316 
                    // with a move, so lines of history that placed files in
2317 
                    // it before and after agree (oxygen#30).
2318 
                    self.tree_ops.push(TreeOp::Move {
2319 
                        node,
2320 
                        old: OpRef {
2321 
                            change: old.0,
2322 
                            index: old.1,
2323 
                        },
2324 
                        parent,
2325 
                        name: comp.to_string(),
2326 
                    });
2327 
                    self.dir_nodes.insert(prefix.clone(), node);
2328 
                    node
2329 
                }
2330 
                (None, None) => {
dfb23a3a Andy McCall ↶ 2331 
                    let node = self.new_node(&prefix, NodeKind::Dir);
75167672 Andy McCall 2332 
                    self.tree_ops.push(TreeOp::AddNode {
2333 
                        node,
2334 
                        kind: NodeKind::Dir,
2335 
                        mode: Mode::default(),
2336 
                        parent,
2337 
                        name: comp.to_string(),
2338 
                    });
98286406 Andy McCall ↶ 2339 
                    self.dir_nodes.insert(prefix.clone(), node);
2340 
                    node
2341 
                }
2342 
            };
2343 
        }
2344 
        parent
2345 
    }
2346
2347 
    fn add_path(&mut self, repo: &Repo, path: &str, e: &FsEntry) -> Result<()> {
2348 
        let (dir, name) = split_path(path);
2349 
        let parent = self.ensure_dir(dir);
dfb23a3a Andy McCall ↶ 2350 
        let node = self.new_node(path, e.kind);
cdb5beba Andy McCall ↶ 2351 
        let opaque = self.opaque_for(path, e);
2352 
        let nocompress = self.attrs.lookup(path).nocompress;
75167672 Andy McCall 2353 
        self.tree_ops.push(TreeOp::AddNode {
2354 
            node,
2355 
            kind: e.kind,
2356 
            mode: Mode {
2357 
                exec: e.exec,
2358 
                opaque,
2359 
            },
2360 
            parent,
2361 
            name: name.to_string(),
2362 
        });
98286406 Andy McCall ↶ 2363 
        if opaque {
b7f102fc Andy McCall ↶ 2364 
            let blob = self.blob_of(repo, path, e, nocompress)?;
75167672 Andy McCall 2365 
            self.edits.push(Edit::Replace {
2366 
                node,
2367 
                old: None,
2368 
                new: Some(blob),
2369 
            });
98286406 Andy McCall ↶ 2370 
        } else if !e.content.is_empty() {
cdb5beba Andy McCall ↶ 2371 
            let content = self.stored_text(path, &e.content);
98286406 Andy McCall ↶ 2372 
            self.edits.push(Edit::Insert {
2373 
                node,
2374 
                after: crate::objects::AtomRef::Start,
2375 
                before: crate::objects::AtomRef::End,
40a40bb7 Andy McCall ↶ 2376 
                lines: crate::lines::Lines::split_shared(content),
98286406 Andy McCall ↶ 2377 
            });
2378 
        }
75167672 Andy McCall 2379 
        self.statuses.push(PathStatus {
2380 
            path: path.to_string(),
2381 
            status: ChangeStatus::Added,
2382 
        });
98286406 Andy McCall ↶ 2383 
        Ok(())
2384 
    }
2385
6e723b81 Andy McCall ↶ 2386 
    /// Move the node at `old` to `new` and apply `e`'s content to it.
75167672 Andy McCall 2387 
    fn move_path(
2388 
        &mut self,
2389 
        repo: &Repo,
2390 
        old: &str,
2391 
        new: &str,
2392 
        e: &FsEntry,
2393 
        expected: &crate::state::FileState,
2394 
    ) -> Result<()> {
e8e6e0ca Andy McCall ↶ 2395 
        let placement = self.base_snapshot.tree.entries[old].placement;
6e723b81 Andy McCall ↶ 2396 
        let (dir, name) = split_path(new);
2397 
        let parent = self.ensure_dir(dir);
75167672 Andy McCall 2398 
        self.tree_ops.push(TreeOp::Move {
2399 
            node: expected.node,
e8e6e0ca Andy McCall ↶ 2400 
            old: op_ref(placement),
75167672 Andy McCall 2401 
            parent,
2402 
            name: name.to_string(),
2403 
        });
e8e6e0ca Andy McCall ↶ 2404 
        for k in self.same_place(&expected.node, placement) {
2405 
            self.tree_ops.push(TreeOp::RemoveNode {
2406 
                node: expected.node,
2407 
                placement: op_ref(k),
2408 
            });
2409 
        }
6e723b81 Andy McCall ↶ 2410 
        self.update_path(repo, old, e, expected)?;
2411 
        // update_path reports under the old path; the rename is what the user sees.
2412 
        self.statuses.retain(|s| s.path != old);
75167672 Andy McCall 2413 
        self.statuses.push(PathStatus {
2414 
            path: new.to_string(),
2415 
            status: ChangeStatus::Renamed,
2416 
        });
6e723b81 Andy McCall ↶ 2417 
        Ok(())
a87c7b9d Andy McCall ↶ 2418 
    }
2419
2420 
    /// For a Git import: a node the commit keeps is in one place, the one
2421 
    /// its tree names, so any other live placement it has in the base goes;
2422 
    /// and a node the tree cannot show at all (an orphan) goes too. Such a
2423 
    /// placement is out of sight, under a directory another line of history
2424 
    /// removed, where no diff of the tree would ever reach it (oxygen#30).
2425 
    fn retire_stray_placements(&mut self) {
2426 
        let mut retired: HashSet<(ChangeHash, u32)> = HashSet::new();
2427 
        let mut placed: BTreeSet<NodeId> = BTreeSet::new();
2428 
        for op in &self.tree_ops {
2429 
            match op {
2430 
                TreeOp::Move { node, old, .. } => {
2431 
                    retired.insert((old.change, old.index));
2432 
                    placed.insert(*node);
2433 
                }
2434 
                TreeOp::RemoveNode { placement, .. } => {
2435 
                    retired.insert((placement.change, placement.index));
2436 
                }
2437 
                TreeOp::AddNode { .. } | TreeOp::SetMode { .. } => {}
2438 
            }
2439 
        }
2440 
        // Nodes kept where they were: what the base's tree view shows and
2441 
        // this change neither moves nor removes.
2442 
        let mut kept: HashMap<NodeId, (ChangeHash, u32)> = HashMap::new();
2443 
        for e in self.base_snapshot.tree.entries.values() {
2444 
            if !retired.contains(&e.placement) {
2445 
                kept.insert(e.node, e.placement);
2446 
            }
2447 
        }
2448 
        // Only a node with several live placements can have a stray one,
2449 
        // and the tree view names each as a conflict.
2450 
        let multi: BTreeSet<NodeId> = self
2451 
            .base_snapshot
2452 
            .tree
2453 
            .conflicts
2454 
            .iter()
2455 
            .filter_map(|c| match c {
2456 
                crate::tree::TreeConflict::MultiPlacement { node, .. } => Some(*node),
2457 
                _ => None,
2458 
            })
2459 
            .collect();
2460 
        let mut extra = Vec::new();
2461 
        // A node whose only placements are under a directory another line
2462 
        // removed (an orphan) is nowhere the tree can show, and Git's tree
2463 
        // has no such path: it goes.
2464 
        for c in &self.base_snapshot.tree.conflicts {
2465 
            let crate::tree::TreeConflict::Orphan { node, .. } = c else {
2466 
                continue;
2467 
            };
2468 
            if kept.contains_key(node) || placed.contains(node) {
2469 
                continue;
2470 
            }
2471 
            for (k, _) in self.base.tree().live_placements(node) {
2472 
                if retired.insert(k) {
2473 
                    extra.push(TreeOp::RemoveNode {
2474 
                        node: *node,
2475 
                        placement: op_ref(k),
2476 
                    });
2477 
                }
2478 
            }
2479 
        }
2480 
        for node in placed
2481 
            .iter()
2482 
            .chain(multi.iter().filter(|n| kept.contains_key(*n)))
2483 
        {
2484 
            for (k, _) in self.base.tree().live_placements(node) {
2485 
                if retired.contains(&k) || kept.get(node) == Some(&k) {
2486 
                    continue;
2487 
                }
2488 
                retired.insert(k);
2489 
                extra.push(TreeOp::RemoveNode {
2490 
                    node: *node,
2491 
                    placement: op_ref(k),
2492 
                });
2493 
            }
2494 
        }
2495 
        if !extra.is_empty() {
2496 
            self.resolves = true;
2497 
            self.tree_ops.extend(extra);
2498 
        }
6e723b81 Andy McCall ↶ 2499 
    }
2500
c2e1d090 Andy McCall 2501 
    /// Nodes the base shows as one with another (identical additions at one
2502 
    /// path, spec §9.2) are retired by the change that first touches the
2503 
    /// node shown: edited, moved, removed or given a mode alone, it would
2504 
    /// differ from them and the name conflict would come back. Nothing
2505 
    /// else changes for the user, so no status reports it.
2506 
    fn retire_unified(&mut self) {
2507 
        let unified = &self.base_snapshot.tree.unified;
2508 
        if unified.is_empty() {
2509 
            return;
2510 
        }
2511 
        let touched: HashSet<NodeId> = self
2512 
            .tree_ops
2513 
            .iter()
2514 
            .map(|o| o.node())
2515 
            .chain(self.edits.iter().map(|e| e.node()))
2516 
            .collect();
2517 
        let mut retired: HashSet<(ChangeHash, u32)> = HashSet::new();
2518 
        for op in &self.tree_ops {
2519 
            match op {
2520 
                TreeOp::Move { old, .. } => {
2521 
                    retired.insert((old.change, old.index));
2522 
                }
2523 
                TreeOp::RemoveNode { placement, .. } => {
2524 
                    retired.insert((placement.change, placement.index));
2525 
                }
2526 
                TreeOp::AddNode { .. } | TreeOp::SetMode { .. } => {}
2527 
            }
2528 
        }
2529 
        let mut extra = Vec::new();
2530 
        for (dup, shown) in unified {
2531 
            if !touched.contains(shown) {
2532 
                continue;
2533 
            }
2534 
            for (k, _) in self.base.tree().live_placements(dup) {
2535 
                if retired.insert(k) {
2536 
                    extra.push(TreeOp::RemoveNode {
2537 
                        node: *dup,
2538 
                        placement: op_ref(k),
2539 
                    });
2540 
                }
2541 
            }
2542 
        }
2543 
        self.tree_ops.extend(extra);
2544 
    }
2545
e8e6e0ca Andy McCall ↶ 2546 
    /// The node's other live placements at the same place as `k`, the one
2547 
    /// the tree shows: two lines of history that put a node back where it
2548 
    /// was each leave a placement, and the tree shows them as one file. A
2549 
    /// change that removes or moves it retires them together, or the copy
2550 
    /// left behind outlives its directory (oxygen#88).
2551 
    fn same_place(&self, node: &NodeId, k: (ChangeHash, u32)) -> Vec<(ChangeHash, u32)> {
2552 
        let tree = self.base.tree();
2553 
        let Some(shown) = tree.placement(&k) else {
2554 
            return Vec::new();
2555 
        };
2556 
        tree.live_placements(node)
2557 
            .into_iter()
2558 
            .filter(|(o, p)| *o != k && p.parent == shown.parent && p.name == shown.name)
2559 
            .map(|(o, _)| o)
2560 
            .collect()
2561 
    }
2562
98286406 Andy McCall ↶ 2563 
    fn remove_path(&mut self, path: &str, expected: &crate::state::FileState) {
2564 
        let entry = &self.base_snapshot.tree.entries[path];
e8e6e0ca Andy McCall ↶ 2565 
        let placement = entry.placement;
75167672 Andy McCall 2566 
        self.tree_ops.push(TreeOp::RemoveNode {
2567 
            node: expected.node,
e8e6e0ca Andy McCall ↶ 2568 
            placement: op_ref(placement),
75167672 Andy McCall 2569 
        });
e8e6e0ca Andy McCall ↶ 2570 
        for k in self.same_place(&expected.node, placement) {
2571 
            self.tree_ops.push(TreeOp::RemoveNode {
2572 
                node: expected.node,
2573 
                placement: op_ref(k),
2574 
            });
2575 
        }
98286406 Andy McCall ↶ 2576 
        if path.contains(".oxy-conflict-") || expected.is_conflicted() {
2577 
            self.resolves = true;
2578 
        }
75167672 Andy McCall 2579 
        self.statuses.push(PathStatus {
2580 
            path: path.to_string(),
2581 
            status: ChangeStatus::Removed,
2582 
        });
98286406 Andy McCall ↶ 2583 
    }
2584
75167672 Andy McCall 2585 
    fn update_path(
2586 
        &mut self,
2587 
        repo: &Repo,
2588 
        path: &str,
2589 
        e: &FsEntry,
2590 
        expected: &crate::state::FileState,
2591 
    ) -> Result<()> {
98286406 Andy McCall ↶ 2592 
        if expected.kind != e.kind {
2593 
            self.remove_path(path, expected);
2594 
            self.statuses.pop();
2595 
            self.add_path(repo, path, e)?;
2596 
            self.statuses.pop();
75167672 Andy McCall 2597 
            self.statuses.push(PathStatus {
2598 
                path: path.to_string(),
2599 
                status: ChangeStatus::Modified,
2600 
            });
98286406 Andy McCall ↶ 2601 
            return Ok(());
2602 
        }
2603 
        let entry = &self.base_snapshot.tree.entries[path];
2604 
        let node = expected.node;
2605 
        let mode_atom = entry.mode_atom.map(op_ref);
cc7a0056 Andy McCall ↶ 2606 
        if e.unchanged {
2607 
            // Known equal to the base's unconflicted rendering (spec
2608 
            // §12.3); only the executable bit can differ.
2609 
            if e.exec != expected.mode.exec {
2610 
                if let Some(old) = mode_atom {
2611 
                    self.tree_ops.push(TreeOp::SetMode {
2612 
                        node,
2613 
                        old,
2614 
                        mode: Mode {
2615 
                            exec: e.exec,
2616 
                            opaque: false,
2617 
                        },
2618 
                    });
2619 
                    self.statuses.push(PathStatus {
2620 
                        path: path.to_string(),
2621 
                        status: ChangeStatus::ModeChanged,
2622 
                    });
2623 
                }
2624 
            }
2625 
            return Ok(());
2626 
        }
98286406 Andy McCall ↶ 2627 
        let mut status: Option<ChangeStatus> = None;
cdb5beba Andy McCall ↶ 2628 
        let wants_opaque = self.opaque_for(path, e);
2629 
        let nocompress = self.attrs.lookup(path).nocompress;
2630 
        let disk_text = self.stored_text(path, &e.content);
98286406 Andy McCall ↶ 2631 
        match &expected.content {
2632 
            FileContent::Dir => {}
75167672 Andy McCall 2633 
            FileContent::Text {
2634 
                bytes,
2635 
                materialized,
2636 
            } => {
98286406 Andy McCall ↶ 2637 
                if materialized.is_conflicted() {
40a40bb7 Andy McCall ↶ 2638 
                    if bytes == &*disk_text {
98286406 Andy McCall ↶ 2639 
                        return Ok(());
2640 
                    }
3652104a Andy McCall ↶ 2641 
                    let regions = self.base.conflict_regions(&node);
2642 
                    let plain: Vec<&[u8]> =
2643 
                        regions.plain.iter().map(|(_, l)| l.as_slice()).collect();
2644 
                    if has_own_conflict_markers(&e.content, bytes, &plain) {
75167672 Andy McCall 2645 
                        self.statuses.push(PathStatus {
2646 
                            path: path.to_string(),
2647 
                            status: ChangeStatus::ConflictEdited,
2648 
                        });
98286406 Andy McCall ↶ 2649 
                        return Ok(());
2650 
                    }
2651 
                    // Resolution (spec §9.6): diff against the unconflicted lines
2652 
                    // with each region as a gap, then retire every region atom.
2653 
                    // Whatever the author kept from a region is re-inserted as
2654 
                    // new atoms, which leaves the graph totally ordered.
cdb5beba Andy McCall ↶ 2655 
                    let mut edits = text_edits(node, &regions.plain, &disk_text);
98286406 Andy McCall ↶ 2656 
                    if !regions.region_atoms.is_empty() {
75167672 Andy McCall 2657 
                        edits.insert(
2658 
                            0,
2659 
                            Edit::Delete {
2660 
                                node,
2661 
                                atoms: ranges(&regions.region_atoms),
2662 
                            },
2663 
                        );
98286406 Andy McCall ↶ 2664 
                    }
2665 
                    self.edits.extend(edits);
2666 
                    self.contributors.extend(regions.contributors);
2667 
                    self.resolves = true;
2668 
                    if e.exec != expected.mode.exec {
2669 
                        if let Some(old) = mode_atom {
75167672 Andy McCall 2670 
                            self.tree_ops.push(TreeOp::SetMode {
2671 
                                node,
2672 
                                old,
2673 
                                mode: Mode {
2674 
                                    exec: e.exec,
2675 
                                    opaque: false,
2676 
                                },
2677 
                            });
98286406 Andy McCall ↶ 2678 
                        }
2679 
                    }
75167672 Andy McCall 2680 
                    self.statuses.push(PathStatus {
2681 
                        path: path.to_string(),
2682 
                        status: ChangeStatus::Resolved,
2683 
                    });
98286406 Andy McCall ↶ 2684 
                    return Ok(());
2685 
                }
cdb5beba Andy McCall ↶ 2686 
                if wants_opaque {
2687 
                    // Text → opaque (spec §7.5), by content or by attribute.
2688 
                    if let Some(old) = mode_atom {
75167672 Andy McCall 2689 
                        self.tree_ops.push(TreeOp::SetMode {
2690 
                            node,
2691 
                            old,
2692 
                            mode: Mode {
2693 
                                exec: e.exec,
2694 
                                opaque: true,
2695 
                            },
2696 
                        });
cdb5beba Andy McCall ↶ 2697 
                    }
2698 
                    let live: Vec<_> = repo_live_keys(self.base, &node);
2699 
                    if !live.is_empty() {
75167672 Andy McCall 2700 
                        self.edits.push(Edit::Delete {
2701 
                            node,
2702 
                            atoms: ranges(&live),
2703 
                        });
cdb5beba Andy McCall ↶ 2704 
                    }
b7f102fc Andy McCall ↶ 2705 
                    let blob = self.blob_of(repo, path, e, nocompress)?;
75167672 Andy McCall 2706 
                    self.edits.push(Edit::Replace {
2707 
                        node,
2708 
                        old: None,
2709 
                        new: Some(blob),
2710 
                    });
cdb5beba Andy McCall ↶ 2711 
                    status = Some(ChangeStatus::Modified);
40a40bb7 Andy McCall ↶ 2712 
                } else if bytes != &*disk_text {
2306a4d2 Andy McCall ↶ 2713 
                    let mut laps = crate::timing::Laps::start();
2714 
                    // The base's rendering holds its live atoms in order,
2715 
                    // one per line (it is not conflicted here).
360aa74a Andy McCall ↶ 2716 
                    let edits = match self.diffs.remove(path) {
2717 
                        Some(edits) => edits,
2718 
                        None => {
2719 
                            let keys = materialized.live_atoms();
2720 
                            let lines = crate::diff::split_lines(bytes);
2721 
                            if keys.len() == lines.len() {
2722 
                                crate::diff::text_edits_of(node, &keys, &lines, &disk_text)
2723 
                            } else {
2724 
                                text_edits(node, &self.base.live_lines(&node), &disk_text)
2725 
                            }
2726 
                        }
2306a4d2 Andy McCall ↶ 2727 
                    };
2728 
                    laps.lap("build: text diff");
cdb5beba Andy McCall ↶ 2729 
                    if !edits.is_empty() {
2730 
                        self.edits.extend(edits);
98286406 Andy McCall ↶ 2731 
                        status = Some(ChangeStatus::Modified);
2732 
                    }
2733 
                }
cdb5beba Andy McCall ↶ 2734 
                let switched_to_opaque = wants_opaque;
98286406 Andy McCall ↶ 2735 
                if !switched_to_opaque && e.exec != expected.mode.exec {
2736 
                    if let Some(old) = mode_atom {
75167672 Andy McCall 2737 
                        self.tree_ops.push(TreeOp::SetMode {
2738 
                            node,
2739 
                            old,
2740 
                            mode: Mode {
2741 
                                exec: e.exec,
2742 
                                opaque: false,
2743 
                            },
2744 
                        });
98286406 Andy McCall ↶ 2745 
                        status.get_or_insert(ChangeStatus::ModeChanged);
2746 
                    }
2747 
                }
2748 
            }
2749 
            FileContent::Opaque { blobs } => {
b7f102fc Andy McCall ↶ 2750 
                let new_ref = e.blob.unwrap_or_else(|| Repo::blob_ref(&e.content));
2751 
                if e.kind == NodeKind::File && wants_opaque {
2752 
                    // Unchanged or replaced, the file stays opaque: remember its blob.
2753 
                    self.blobs.insert(path.to_string(), new_ref);
f7510b63 Andy McCall ↶ 2754 
                }
2755 
                if blobs.len() > 1 && self.resolve.contains(path) && e.kind == expected.kind {
2756 
                    // Settle the replace conflict (spec §9.3): replace every
2757 
                    // live candidate with the chosen blob, depending on each
2758 
                    // change that made one live, so all are superseded.
2759 
                    let chosen = self.blob_of(repo, path, e, nocompress)?;
2760 
                    for b in blobs {
2761 
                        self.edits.push(Edit::Replace {
2762 
                            node,
2763 
                            old: Some(*b),
2764 
                            new: Some(chosen),
2765 
                        });
2766 
                    }
2767 
                    self.contributors
2768 
                        .extend(self.base.blob_contributors(&node, blobs));
2769 
                    self.resolves = true;
2770 
                    self.statuses.push(PathStatus {
2771 
                        path: path.to_string(),
2772 
                        status: ChangeStatus::Resolved,
2773 
                    });
2774 
                    return Ok(());
b7f102fc Andy McCall ↶ 2775 
                }
98286406 Andy McCall ↶ 2776 
                if blobs.len() > 1 {
2777 
                    if !blobs.contains(&new_ref) {
75167672 Andy McCall 2778 
                        self.statuses.push(PathStatus {
2779 
                            path: path.to_string(),
2780 
                            status: ChangeStatus::ConflictEdited,
2781 
                        });
98286406 Andy McCall ↶ 2782 
                    }
2783 
                    return Ok(());
2784 
                }
2785 
                let current = blobs.first().copied();
2786 
                if current != Some(new_ref) {
cdb5beba Andy McCall ↶ 2787 
                    if e.kind == NodeKind::File && !wants_opaque {
98286406 Andy McCall ↶ 2788 
                        // Opaque → text (spec §7.5).
2789 
                        if let Some(old) = mode_atom {
75167672 Andy McCall 2790 
                            self.tree_ops.push(TreeOp::SetMode {
2791 
                                node,
2792 
                                old,
2793 
                                mode: Mode {
2794 
                                    exec: e.exec,
2795 
                                    opaque: false,
2796 
                                },
2797 
                            });
98286406 Andy McCall ↶ 2798 
                        }
75167672 Andy McCall 2799 
                        self.edits.push(Edit::Replace {
2800 
                            node,
2801 
                            old: current,
2802 
                            new: None,
2803 
                        });
98286406 Andy McCall ↶ 2804 
                        if !e.content.is_empty() {
2805 
                            self.edits.push(Edit::Insert {
2806 
                                node,
2807 
                                after: crate::objects::AtomRef::Start,
2808 
                                before: crate::objects::AtomRef::End,
01a47cd7 Andy McCall ↶ 2809 
                                lines: crate::lines::Lines::split(&e.content),
98286406 Andy McCall ↶ 2810 
                            });
2811 
                        }
2812 
                    } else {
b7f102fc Andy McCall ↶ 2813 
                        let blob = self.blob_of(repo, path, e, nocompress)?;
75167672 Andy McCall 2814 
                        self.edits.push(Edit::Replace {
2815 
                            node,
2816 
                            old: current,
2817 
                            new: Some(blob),
2818 
                        });
6eae06b6 Andy McCall ↶ 2819 
                        // New contents and a new executable bit at once: the
2820 
                        // bit was dropped (VS Code's inno_updater.exe,
2821 
                        // 2018-02-28, lost +x for 49 commits; oxygen#87).
2822 
                        if e.kind == NodeKind::File && e.exec != expected.mode.exec {
2823 
                            if let Some(old) = mode_atom {
2824 
                                self.tree_ops.push(TreeOp::SetMode {
2825 
                                    node,
2826 
                                    old,
2827 
                                    mode: Mode {
2828 
                                        exec: e.exec,
2829 
                                        opaque: true,
2830 
                                    },
2831 
                                });
2832 
                            }
2833 
                        }
98286406 Andy McCall ↶ 2834 
                    }
2835 
                    status = Some(ChangeStatus::Modified);
2836 
                } else if e.kind == NodeKind::File && e.exec != expected.mode.exec {
2837 
                    if let Some(old) = mode_atom {
75167672 Andy McCall 2838 
                        self.tree_ops.push(TreeOp::SetMode {
2839 
                            node,
2840 
                            old,
2841 
                            mode: Mode {
2842 
                                exec: e.exec,
2843 
                                opaque: true,
2844 
                            },
2845 
                        });
98286406 Andy McCall ↶ 2846 
                        status = Some(ChangeStatus::ModeChanged);
2847 
                    }
2848 
                }
2849 
            }
2850 
        }
2851 
        if let Some(s) = status {
75167672 Andy McCall 2852 
            self.statuses.push(PathStatus {
2853 
                path: path.to_string(),
2854 
                status: s,
2855 
            });
98286406 Andy McCall ↶ 2856 
        }
2857 
        Ok(())
2858 
    }
2859 
}
2860
2861 
fn repo_live_keys(base: &State, node: &NodeId) -> Vec<crate::graph::AtomKey> {
2862 
    base.live_lines(node).into_iter().map(|(k, _)| k).collect()
2863 
}
2864
2865 
/// The changes a change's scope references (spec §8.1): introducers of every
2866 
/// atom, placement, and mode it touches, and of every node it operates on.
2867 
pub fn referenced_changes(change: &Change, base: &State) -> BTreeSet<ChangeHash> {
2868 
    let mut out = BTreeSet::new();
2869 
    let opref = |r: &OpRef, out: &mut BTreeSet<ChangeHash>| {
2870 
        if !r.change.is_zero() {
2871 
            out.insert(r.change);
2872 
        }
2873 
    };
2874 
    let mut nodes: Vec<NodeId> = Vec::new();
2875 
    for op in &change.tree_ops {
2876 
        match op {
2877 
            TreeOp::AddNode { parent, .. } => nodes.push(*parent),
2878 
            TreeOp::RemoveNode { node, placement } => {
2879 
                nodes.push(*node);
2880 
                opref(placement, &mut out);
2881 
            }
75167672 Andy McCall 2882 
            TreeOp::Move {
2883 
                node, old, parent, ..
2884 
            } => {
98286406 Andy McCall ↶ 2885 
                nodes.push(*node);
2886 
                nodes.push(*parent);
2887 
                opref(old, &mut out);
2888 
            }
2889 
            TreeOp::SetMode { node, old, .. } => {
2890 
                nodes.push(*node);
2891 
                opref(old, &mut out);
2892 
            }
2893 
        }
2894 
    }
2895 
    for e in &change.edits {
2896 
        nodes.push(e.node());
2897 
        match e {
2898 
            Edit::Insert { after, before, .. } => {
2899 
                for r in [after, before] {
2900 
                    if let crate::objects::AtomRef::Atom { change, .. } = r {
2901 
                        if !change.is_zero() {
2902 
                            out.insert(*change);
2903 
                        }
2904 
                    }
2905 
                }
2906 
            }
2907 
            Edit::Delete { atoms, .. } => {
62794ecc Andy McCall ↶ 2908 
                // Deleting an atom also depends on whoever inserted the live
2909 
                // atoms right after it: the author saw those insertions, and
2910 
                // without the dependency spec §9.1 would call the deletion
2911 
                // concurrent with them (a delete/insert conflict).
2912 
                let graph = base.file(&e.node());
98286406 Andy McCall ↶ 2913 
                for r in atoms {
2914 
                    if !r.change.is_zero() {
2915 
                        out.insert(r.change);
2916 
                    }
62794ecc Andy McCall ↶ 2917 
                    if let Some(g) = graph {
2918 
                        for i in r.from..r.to {
2919 
                            out.extend(
2920 
                                g.live_successor_changes(&(r.change, i))
2921 
                                    .into_iter()
2922 
                                    .filter(|c| !c.is_zero()),
2923 
                            );
2924 
                        }
2925 
                    }
98286406 Andy McCall ↶ 2926 
                }
2927 
            }
62794ecc Andy McCall ↶ 2928 
            // Replacing a blob depends on the change that made it current,
2929 
            // as deleting a line depends on the line's introducer.
2930 
            Edit::Replace {
2931 
                node,
2932 
                old: Some(old),
2933 
                ..
2934 
            } => {
2935 
                out.extend(
2936 
                    base.live_replaces(node)
2937 
                        .into_iter()
2938 
                        .filter(|(c, b)| b == old && !c.is_zero())
2939 
                        .map(|(c, _)| c),
2940 
                );
2941 
            }
98286406 Andy McCall ↶ 2942 
            Edit::Replace { .. } => {}
2943 
        }
2944 
    }
2945 
    for n in nodes {
2946 
        if let Some(h) = base.tree().introducer(&n) {
2947 
            out.insert(h);
2948 
        }
2949 
    }
2950 
    out
2951 
}
2952
2953 
/// Paths of the nodes a change adds on top of `base_snapshot`.
75167672 Andy McCall 2954 
pub fn added_node_paths(
2955 
    change: &Change,
2956 
    base_snapshot: &Snapshot,
2957 
) -> HashMap<String, (NodeId, NodeKind)> {
98286406 Andy McCall ↶ 2958 
    let by_node = base_snapshot.tree.by_node();
2959 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
2960 
    for op in &change.tree_ops {
75167672 Andy McCall 2961 
        if let TreeOp::AddNode {
2962 
            node,
2963 
            parent,
2964 
            name,
2965 
            kind,
2966 
            ..
2967 
        } = op
2968 
        {
98286406 Andy McCall ↶ 2969 
            new_names.insert(*node, (*parent, name.clone(), *kind));
2970 
        }
2971 
    }
2972 
    let mut out = HashMap::new();
2973 
    for (node, (_, _, kind)) in &new_names {
2974 
        if let Some(path) = path_of_node(node, &by_node, &new_names) {
2975 
            out.insert(path, (*node, *kind));
2976 
        }
2977 
    }
2978 
    out
2979 
}
2980
c2e1d090 Andy McCall 2981 
/// Whether `op` retires a node `base_snapshot` shows as one with another
2982 
/// (spec §9.2): part of touching that node, never a removal the user made.
2983 
fn is_unified_retirement(op: &TreeOp, base_snapshot: &Snapshot) -> bool {
2984 
    matches!(op, TreeOp::RemoveNode { node, .. } if base_snapshot.tree.unified.contains_key(node))
2985 
}
2986
98286406 Andy McCall ↶ 2987 
fn path_of_node(
2988 
    n: &NodeId,
2989 
    by_node: &HashMap<NodeId, &crate::tree::TreeEntry>,
2990 
    new_names: &HashMap<NodeId, (NodeId, String, NodeKind)>,
2991 
) -> Option<String> {
2992 
    if let Some(e) = by_node.get(n) {
2993 
        return Some(e.path.clone());
2994 
    }
2995 
    let mut parts = Vec::new();
2996 
    let mut cur = *n;
2997 
    loop {
2998 
        if cur.is_root() {
2999 
            break;
3000 
        }
3001 
        if let Some(e) = by_node.get(&cur) {
3002 
            if !e.path.is_empty() {
3003 
                parts.push(e.path.clone());
3004 
            }
3005 
            break;
3006 
        }
3007 
        let (parent, name, _) = new_names.get(&cur)?;
3008 
        parts.push(name.clone());
3009 
        cur = *parent;
3010 
    }
3011 
    parts.reverse();
3012 
    Some(parts.join("/"))
3013 
}
3014
3015 
/// Render a change's operations with line content, using `base` for deleted lines.
3016 
pub fn render_change(change: &Change, base: &State, base_snapshot: &Snapshot) -> String {
3017 
    let by_node = base_snapshot.tree.by_node();
3018 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
3019 
    for op in &change.tree_ops {
75167672 Andy McCall 3020 
        if let TreeOp::AddNode {
3021 
            node,
3022 
            parent,
3023 
            name,
3024 
            kind,
3025 
            ..
3026 
        } = op
3027 
        {
98286406 Andy McCall ↶ 3028 
            new_names.insert(*node, (*parent, name.clone(), *kind));
3029 
        }
3030 
    }
75167672 Andy McCall 3031 
    let path_of = |n: &NodeId| -> String {
3032 
        path_of_node(n, &by_node, &new_names).unwrap_or_else(|| format!("<node {}>", n.short(8)))
3033 
    };
98286406 Andy McCall ↶ 3034
3035 
    let mut out = String::new();
c2e1d090 Andy McCall 3036 
    for op in change
3037 
        .tree_ops
3038 
        .iter()
3039 
        .filter(|o| !is_unified_retirement(o, base_snapshot))
3040 
    {
98286406 Andy McCall ↶ 3041 
        match op {
75167672 Andy McCall 3042 
            TreeOp::AddNode {
3043 
                node, kind, mode, ..
3044 
            } => {
3045 
                out.push_str(&format!(
3046 
                    "A {} ({:?}{}{})\n",
3047 
                    path_of(node),
3048 
                    kind,
3049 
                    if mode.exec { ", exec" } else { "" },
3050 
                    if mode.opaque { ", opaque" } else { "" }
3051 
                ));
98286406 Andy McCall ↶ 3052 
            }
3053 
            TreeOp::RemoveNode { node, .. } => out.push_str(&format!("D {}\n", path_of(node))),
0912a033 Andy McCall ↶ 3054 
            TreeOp::Move {
3055 
                node, parent, name, ..
3056 
            } => {
3057 
                let to = if parent.is_root() {
3058 
                    name.clone()
3059 
                } else {
3060 
                    format!("{}/{name}", path_of(parent))
3061 
                };
3062 
                out.push_str(&format!("R {} -> {to}\n", path_of(node)))
75167672 Andy McCall 3063 
            }
98286406 Andy McCall ↶ 3064 
            TreeOp::SetMode { node, mode, .. } => {
75167672 Andy McCall 3065 
                out.push_str(&format!(
3066 
                    "M {} mode: exec={} opaque={}\n",
3067 
                    path_of(node),
3068 
                    mode.exec,
3069 
                    mode.opaque
3070 
                ));
98286406 Andy McCall ↶ 3071 
            }
0912a033 Andy McCall ↶ 3072 
        }
3073 
    }
3074 
    // Every other file the change edits is modified: the summary names
3075 
    // every path it touches, not only those with a tree operation.
3076 
    let with_op: HashSet<NodeId> = change.tree_ops.iter().map(|o| o.node()).collect();
3077 
    let mut listed: HashSet<NodeId> = HashSet::new();
3078 
    for e in &change.edits {
3079 
        let n = e.node();
3080 
        if !with_op.contains(&n) && listed.insert(n) {
3081 
            out.push_str(&format!("M {}\n", path_of(&n)));
98286406 Andy McCall ↶ 3082 
        }
3083 
    }
3084 
    let mut last_node: Option<NodeId> = None;
3085 
    for e in &change.edits {
3086 
        let n = e.node();
3087 
        if last_node != Some(n) {
3088 
            out.push_str(&format!("--- {}\n", path_of(&n)));
3089 
            last_node = Some(n);
3090 
        }
3091 
        match e {
3092 
            Edit::Insert { lines, .. } => {
01a47cd7 Andy McCall ↶ 3093 
                for l in lines.iter() {
98286406 Andy McCall ↶ 3094 
                    out.push('+');
3095 
                    out.push_str(&String::from_utf8_lossy(l));
3096 
                    if !l.ends_with(b"\n") {
3097 
                        out.push('\n');
3098 
                    }
3099 
                }
3100 
            }
3101 
            Edit::Delete { atoms, .. } => {
3102 
                for r in atoms {
3103 
                    for i in r.from..r.to {
75167672 Andy McCall 3104 
                        let content = base
3105 
                            .file(&n)
3106 
                            .and_then(|g| g.content(&(r.change, i)))
3107 
                            .unwrap_or(b"<unknown>\n");
98286406 Andy McCall ↶ 3108 
                        out.push('-');
3109 
                        out.push_str(&String::from_utf8_lossy(content));
3110 
                        if !content.ends_with(b"\n") {
3111 
                            out.push('\n');
3112 
                        }
3113 
                    }
3114 
                }
3115 
            }
3116 
            Edit::Replace { old, new, .. } => {
3117 
                out.push_str(&format!(
3118 
                    "  opaque content {} -> {}\n",
3119 
                    old.map(|o| o.short(12)).unwrap_or_else(|| "none".into()),
3120 
                    new.map(|o| o.short(12)).unwrap_or_else(|| "none".into())
3121 
                ));
3122 
            }
3123 
        }
3124 
    }
3125 
    out
3126 
}
6e723b81 Andy McCall ↶ 3127
3128 
/// Pair removed files with added files of the same kind whose contents are
3129 
/// at least `RENAME_THRESHOLD` percent alike, best matches first. Beyond
3130 
/// `RENAME_LIMIT` candidate pairs only identical contents are paired.
3131 
const RENAME_THRESHOLD: u32 = 50;
3132 
const RENAME_LIMIT: usize = 10_000;
2306a4d2 Andy McCall ↶ 3133
23ff1288 Andy McCall ↶ 3134 
/// How a change is built beyond its entries: for a Git import, the seed
3135 
/// its new nodes' identifiers derive from; and, when it is built from part
3136 
/// of the tree, whether the new tree has anything under a directory.
3137 
#[derive(Default)]
3138 
struct Scope<'a> {
3139 
    seed: Option<[u8; 32]>,
2306a4d2 Andy McCall ↶ 3140 
    dir_needed: Option<&'a dyn Fn(&str) -> bool>,
3141 
}
6e723b81 Andy McCall ↶ 3142
75167672 Andy McCall 3143 
fn detect_renames(
3144 
    base_snapshot: &Snapshot,
3145 
    fs_entries: &BTreeMap<String, FsEntry>,
3146 
    in_scope: impl Fn(&str) -> bool,
3147 
) -> Vec<(String, String)> {
6e723b81 Andy McCall ↶ 3148 
    let removed: Vec<(&String, &crate::state::FileState)> = base_snapshot
3149 
        .files
3150 
        .iter()
75167672 Andy McCall 3151 
        .filter(|(p, f)| {
3152 
            f.kind != NodeKind::Dir
3153 
                && !fs_entries.contains_key(*p)
3154 
                && in_scope(p)
3155 
                && !f.is_conflicted()
3156 
        })
3157 
        .collect();
3158 
    let added: Vec<(&String, &FsEntry)> = fs_entries
3159 
        .iter()
3160 
        .filter(|(p, e)| e.kind != NodeKind::Dir && !base_snapshot.files.contains_key(*p))
6e723b81 Andy McCall ↶ 3161 
        .collect();
3162 
    if removed.is_empty() || added.is_empty() {
3163 
        return vec![];
3164 
    }
3165 
    let exact_only = removed.len() * added.len() > RENAME_LIMIT;
3166 
    let old_content = |f: &crate::state::FileState| -> Option<Vec<u8>> {
3167 
        match &f.content {
3168 
            FileContent::Text { bytes, .. } => Some(bytes.clone()),
3169 
            _ => None,
3170 
        }
3171 
    };
3172 
    let mut scored: Vec<(u32, usize, usize)> = Vec::new();
3173 
    for (ri, (_, f)) in removed.iter().enumerate() {
3174 
        let old = old_content(f);
3175 
        let old_blob = match &f.content {
3176 
            FileContent::Opaque { blobs } if blobs.len() == 1 => Some(blobs[0]),
3177 
            _ => None,
3178 
        };
3179 
        for (ai, (_, e)) in added.iter().enumerate() {
3180 
            if e.kind != f.kind {
3181 
                continue;
3182 
            }
3183 
            let score = match (&old, old_blob) {
3184 
                (Some(o), _) => {
3185 
                    if exact_only {
40a40bb7 Andy McCall ↶ 3186 
                        if *o == *e.content {
75167672 Andy McCall 3187 
                            100
3188 
                        } else {
3189 
                            0
3190 
                        }
6e723b81 Andy McCall ↶ 3191 
                    } else {
3192 
                        similarity(o, &e.content)
3193 
                    }
3194 
                }
b7f102fc Andy McCall ↶ 3195 
                (None, Some(b)) if e.blob.unwrap_or_else(|| Repo::blob_ref(&e.content)) == b => 100,
6e723b81 Andy McCall ↶ 3196 
                _ => 0,
3197 
            };
3198 
            if score >= RENAME_THRESHOLD {
3199 
                scored.push((score, ri, ai));
3200 
            }
3201 
        }
3202 
    }
75167672 Andy McCall 3203 
    scored.sort_by(|a, b| {
3204 
        b.0.cmp(&a.0)
3205 
            .then(removed[a.1].0.cmp(removed[b.1].0))
3206 
            .then(added[a.2].0.cmp(added[b.2].0))
3207 
    });
6e723b81 Andy McCall ↶ 3208 
    let mut used_r = vec![false; removed.len()];
3209 
    let mut used_a = vec![false; added.len()];
3210 
    let mut out = Vec::new();
3211 
    for (_, ri, ai) in scored {
3212 
        if used_r[ri] || used_a[ai] {
3213 
            continue;
3214 
        }
3215 
        used_r[ri] = true;
3216 
        used_a[ai] = true;
3217 
        out.push((removed[ri].0.clone(), added[ai].0.clone()));
3218 
    }
3219 
    out
3220 
}
3221
3222 
/// Percentage of the larger file's bytes found in lines the two share
3223 
/// (multiset intersection of lines), like Git's similarity index.
3224 
pub fn similarity(a: &[u8], b: &[u8]) -> u32 {
3225 
    if a == b {
3226 
        return 100;
3227 
    }
3228 
    if a.is_empty() || b.is_empty() {
3229 
        return 0;
3230 
    }
3231 
    let mut counts: HashMap<&[u8], usize> = HashMap::new();
3232 
    for l in split_lines(a) {
3233 
        *counts.entry(l).or_default() += 1;
3234 
    }
3235 
    let mut shared = 0usize;
3236 
    for l in split_lines(b) {
3237 
        if let Some(n) = counts.get_mut(l) {
3238 
            if *n > 0 {
3239 
                *n -= 1;
3240 
                shared += l.len();
3241 
            }
3242 
        }
3243 
    }
3244 
    (shared * 100 / a.len().max(b.len())) as u32
d616faf5 Andy McCall ↶ 3245 
}
3246
3247 
#[cfg(test)]
3248 
mod join_tests {
3249 
    use super::*;
f57bf670 Andy McCall ↶ 3250
3251 
    #[test]
3252 
    fn missing_from_finds_what_the_other_lacks() {
3253 
        let a: BTreeMap<String, u8> = [("a", 1), ("b", 2), ("c/d", 3), ("e", 4)]
3254 
            .into_iter()
3255 
            .map(|(k, v)| (k.to_string(), v))
3256 
            .collect();
3257 
        let b: BTreeMap<String, ()> = [("0", ()), ("b", ()), ("c", ()), ("e", ())]
3258 
            .into_iter()
3259 
            .map(|(k, v)| (k.to_string(), v))
3260 
            .collect();
3261 
        let got: Vec<&str> = missing_from(&a, &b)
3262 
            .iter()
3263 
            .map(|(k, _)| k.as_str())
3264 
            .collect();
3265 
        assert_eq!(got, vec!["a", "c/d"]);
3266 
    }
d616faf5 Andy McCall ↶ 3267
3268 
    #[test]
3269 
    fn join_pairs_every_key_in_order() {
3270 
        let a: BTreeMap<String, u8> = [("a", 1), ("b/c", 2), ("d", 3)]
3271 
            .into_iter()
3272 
            .map(|(k, v)| (k.to_string(), v))
3273 
            .collect();
3274 
        let b: BTreeMap<String, char> = [("b", 'x'), ("b/c", 'y'), ("e", 'z')]
3275 
            .into_iter()
3276 
            .map(|(k, v)| (k.to_string(), v))
3277 
            .collect();
3278 
        let got: Vec<(String, Option<u8>, Option<char>)> = join(&a, &b)
3279 
            .map(|(k, x, y)| (k.clone(), x.copied(), y.copied()))
3280 
            .collect();
3281 
        let want = vec![
3282 
            ("a".to_string(), Some(1), None),
3283 
            ("b".to_string(), None, Some('x')),
3284 
            ("b/c".to_string(), Some(2), Some('y')),
3285 
            ("d".to_string(), Some(3), None),
3286 
            ("e".to_string(), None, Some('z')),
3287 
        ];
3288 
        assert_eq!(got, want);
3289 
    }
6e723b81 Andy McCall ↶ 3290 
}