OXYGENHUB
# 236f35c0baba oxygen / crates / oxy-core / src / repo / snapshot.rs
111.3 KiB · Rust · last changed A commit writes the manifest once (refs #41) 1 month ago source blame history raw
98286406 Andy McCall ↶ 1 
//! Working-copy snapshots, commits, and the change builder (spec §12.3).
2
3 
use super::*;
957fee30 Andy McCall ↶ 4 
use crate::objects::AtomRef;
98286406 Andy McCall ↶ 5
6 
#[derive(Clone, Copy, Debug, PartialEq, Eq)]
7 
pub enum ChangeStatus {
8 
    Added,
9 
    Modified,
10 
    Removed,
6e723b81 Andy McCall ↶ 11 
    /// Moved from another path (detected by similarity, tracked as a placement change).
12 
    Renamed,
98286406 Andy McCall ↶ 13 
    ModeChanged,
14 
    /// The file is conflicted in the base state and still contains conflict markers.
15 
    ConflictEdited,
16 
    /// The file was conflicted in the base state and the working copy resolves it.
17 
    Resolved,
18 
}
19
20 
/// Lines that mark a conflict region in a materialized file (spec §9.5).
21 
pub fn has_conflict_markers(content: &[u8]) -> bool {
22 
    split_lines(content).iter().any(|l| {
75167672 Andy McCall 23 
        l.starts_with(b"<<<<<<< oxy conflict")
24 
            || l.starts_with(b"======= oxy conflict")
25 
            || l.starts_with(b">>>>>>> oxy conflict")
98286406 Andy McCall ↶ 26 
    })
27 
}
28
29 
#[derive(Clone, Debug, PartialEq, Eq)]
30 
pub struct PathStatus {
31 
    pub path: String,
32 
    pub status: ChangeStatus,
33 
}
34
614541c3 Andy McCall ↶ 35 
/// Operations of a change built from the working directory against a base.
36 
pub(crate) struct BuiltChange {
37 
    pub tree_ops: Vec<TreeOp>,
38 
    pub edits: Vec<Edit>,
39 
    pub contributors: Vec<ChangeHash>,
40 
    pub resolves: bool,
41 
    pub statuses: Vec<PathStatus>,
b7f102fc Andy McCall ↶ 42 
    /// The stored blob of every opaque path, for the manifest's stat cache.
43 
    pub blobs: HashMap<String, ObjectRef>,
614541c3 Andy McCall ↶ 44 
}
45
98286406 Andy McCall ↶ 46 
/// Result of snapshotting the working copy.
47 
pub struct SnapshotResult {
48 
    pub wc_hash: ChangeHash,
49 
    pub wc: Change,
50 
    /// The base state (closure of the working-copy change's dependencies).
51 
    pub base: State,
52 
    pub base_snapshot: Snapshot,
53 
    pub statuses: Vec<PathStatus>,
54 
    /// Whether the snapshot rewrote the working-copy change.
55 
    pub changed: bool,
56 
    /// Whether the working copy resolves conflicts of its base (spec §9.6).
57 
    pub resolves: bool,
58 
    /// Changes that contributed to the conflicts being resolved.
59 
    pub contributors: Vec<ChangeHash>,
f9466021 Andy McCall ↶ 60 
    /// The `.oxyattributes` in force on disk, from the files the walk read.
61 
    pub(crate) attributes: crate::attributes::Attributes,
d4dbab97 Andy McCall ↶ 62 
    /// Folders holding a repository of their own (`.git` or `.oxy`) that
63 
    /// are not subrepos, so nothing in them is tracked (oxygen#43).
64 
    pub untracked_nested: Vec<String>,
236f35c0 Andy McCall ↶ 65 
    /// A commit's snapshot leaves the manifest to the commit, which writes
66 
    /// it once with the new base's graph keys.
67 
    pub(crate) manifest: Option<PendingManifest>,
98286406 Andy McCall ↶ 68 
}
69
236f35c0 Andy McCall ↶ 70 
/// The manifest a commit's snapshot would have written, as of the moment
71 
/// it would have written it (that moment is the racy-window stamp).
72 
pub(crate) struct PendingManifest {
73 
    entries: BTreeMap<String, FsEntry>,
74 
    blobs: HashMap<String, ObjectRef>,
75 
    dirs: Option<super::WalkedDirs>,
76 
    written: u128,
77 
}
78
98286406 Andy McCall ↶ 79 
pub fn is_binary(content: &[u8]) -> bool {
80 
    content[..content.len().min(BINARY_SNIFF)].contains(&0)
81 
}
82
83 
impl Repo {
614541c3 Andy McCall ↶ 84 
    /// Diff the files on disk against `base` and build the operations of a
85 
    /// change from `base` to the working directory. `reuse` maps paths to the
86 
    /// NodeIds an earlier version of the same change assigned to new paths.
75167672 Andy McCall 87 
    pub(crate) fn build_against(
88 
        &self,
89 
        base: &State,
90 
        base_snapshot: &Snapshot,
91 
        reuse: HashMap<String, (NodeId, NodeKind)>,
92 
    ) -> Result<BuiltChange> {
c863c7a7 Andy McCall ↶ 93 
        let mut fs_entries =
696bbf29 Andy McCall ↶ 94 
            self.walk_working_copy_with(None, &Self::tracked_kinds(base_snapshot), None)?;
c863c7a7 Andy McCall ↶ 95 
        self.platform_normalize(&mut fs_entries, base_snapshot);
c3069537 Andy McCall ↶ 96 
        self.build_from_entries(base, base_snapshot, reuse, &fs_entries)
97 
    }
98
c863c7a7 Andy McCall ↶ 99 
    /// Read the working copy as the platform left it (spec §7.7): without
100 
    /// executable bits a file keeps the bit it has in the base, and without
101 
    /// symbolic links a plain file where the base has a symlink is that
102 
    /// symlink, written out as its target, not a new file.
103 
    fn platform_normalize(&self, entries: &mut BTreeMap<String, FsEntry>, base: &Snapshot) {
104 
        let p = self.platform;
105 
        if p.exec_bits && p.symlinks {
106 
            return;
107 
        }
108 
        for (path, e) in entries.iter_mut() {
109 
            let Some(f) = base.files.get(path) else {
110 
                continue;
111 
            };
112 
            if !p.symlinks && f.kind == NodeKind::Symlink && e.kind == NodeKind::File {
113 
                e.kind = NodeKind::Symlink;
114 
                e.exec = false;
115 
            }
116 
            if !p.exec_bits && f.kind == NodeKind::File && e.kind == NodeKind::File {
117 
                e.exec = f.mode.exec;
118 
            }
119 
        }
120 
    }
121
c3069537 Andy McCall ↶ 122 
    /// Like `build_against`, for files given in memory (imports).
75167672 Andy McCall 123 
    pub(crate) fn build_from_entries(
124 
        &self,
125 
        base: &State,
126 
        base_snapshot: &Snapshot,
127 
        reuse: HashMap<String, (NodeId, NodeKind)>,
128 
        fs_entries: &BTreeMap<String, FsEntry>,
129 
    ) -> Result<BuiltChange> {
f7510b63 Andy McCall ↶ 130 
        self.build_from_entries_resolving(base, base_snapshot, reuse, fs_entries, &HashSet::new())
131 
    }
132
133 
    /// Like `build_from_entries`, settling the replace conflicts of the
134 
    /// opaque files in `resolve` to their content (spec §9.3, §9.6): a
135 
    /// caller that knows the user chose it, such as the hub's resolver. A
136 
    /// snapshot never settles one on its own, since a checkout writes one
137 
    /// candidate at the path.
138 
    pub(crate) fn build_from_entries_resolving(
139 
        &self,
140 
        base: &State,
141 
        base_snapshot: &Snapshot,
142 
        reuse: HashMap<String, (NodeId, NodeKind)>,
143 
        fs_entries: &BTreeMap<String, FsEntry>,
144 
        resolve: &HashSet<String>,
145 
    ) -> Result<BuiltChange> {
23ff1288 Andy McCall ↶ 146 
        self.build_from_entries_with(
147 
            base,
148 
            base_snapshot,
149 
            reuse,
150 
            fs_entries,
151 
            resolve,
152 
            Scope::default(),
153 
        )
dfb23a3a Andy McCall ↶ 154 
    }
155
156 
    /// `build_from_entries` whose new nodes take identifiers derived from
157 
    /// `seed` and their paths instead of random ones, so the same input
158 
    /// always builds the same change (a Git import, spec §23).
2306a4d2 Andy McCall ↶ 159 
    ///
160 
    /// With `dir_needed`, the snapshot and entries hold only the paths the
161 
    /// import touches (and the directories a removal may leave empty); a
162 
    /// directory stays while `dir_needed` says the new tree has something
163 
    /// under it.
dfb23a3a Andy McCall ↶ 164 
    pub(crate) fn build_from_entries_seeded(
165 
        &self,
166 
        base: &State,
167 
        base_snapshot: &Snapshot,
168 
        fs_entries: &BTreeMap<String, FsEntry>,
169 
        seed: [u8; 32],
2306a4d2 Andy McCall ↶ 170 
        dir_needed: Option<&dyn Fn(&str) -> bool>,
dfb23a3a Andy McCall ↶ 171 
    ) -> Result<BuiltChange> {
172 
        self.build_from_entries_with(
173 
            base,
174 
            base_snapshot,
175 
            HashMap::new(),
176 
            fs_entries,
177 
            &HashSet::new(),
23ff1288 Andy McCall ↶ 178 
            Scope {
179 
                seed: Some(seed),
180 
                dir_needed,
181 
            },
dfb23a3a Andy McCall ↶ 182 
        )
183 
    }
184
23ff1288 Andy McCall ↶ 185 
    /// `build_from_entries` for a snapshot most of whose files are known
186 
    /// unchanged (oxygen#41): only the paths that may differ are built
187 
    /// (files read rather than known unchanged, base files gone from disk,
188 
    /// and the directories those may leave empty), against a snapshot of
189 
    /// just those, so a large tree with a few edits costs what they touch.
190 
    fn build_changed_only(
191 
        &self,
192 
        base: &State,
193 
        base_snapshot: &Snapshot,
194 
        reuse: HashMap<String, (NodeId, NodeKind)>,
195 
        fs_entries: &BTreeMap<String, FsEntry>,
196 
    ) -> Result<BuiltChange> {
d616faf5 Andy McCall ↶ 197 
        let pick = crate::timing::phase("snapshot:   of which picking the changed");
23ff1288 Andy McCall ↶ 198 
        // A file known unchanged may still have changed mode: that alters
199 
        // neither its size nor its time.
d616faf5 Andy McCall ↶ 200 
        let changed: BTreeMap<String, FsEntry> = join(fs_entries, &base_snapshot.files)
201 
            .filter_map(|(p, e, f)| {
202 
                let e = e?;
203 
                (!e.unchanged || f.is_none_or(|f| f.kind != e.kind || f.mode.exec != e.exec))
204 
                    .then(|| (p.clone(), e.clone()))
23ff1288 Andy McCall ↶ 205 
            })
206 
            .collect();
207 
        let mut files: BTreeMap<String, crate::state::FileState> = BTreeMap::new();
208 
        for p in changed.keys() {
209 
            if let Some(f) = base_snapshot.files.get(p) {
210 
                files.insert(p.clone(), f.clone());
211 
            }
212 
        }
d616faf5 Andy McCall ↶ 213 
        // Paths the base has and the disk does not.
214 
        let gone = join(&base_snapshot.files, fs_entries)
215 
            .filter_map(|(p, f, e)| (e.is_none()).then_some((p, f?)));
216 
        for (p, f) in gone {
217 
            if f.kind == NodeKind::Dir {
23ff1288 Andy McCall ↶ 218 
                continue;
219 
            }
220 
            files.insert(p.clone(), f.clone());
221 
            let mut end = p.len();
222 
            while let Some(i) = p[..end].rfind('/') {
223 
                let dir = &p[..i];
224 
                if files.contains_key(dir) {
225 
                    break;
226 
                }
227 
                if let Some(d) = base_snapshot.files.get(dir) {
228 
                    files.insert(dir.to_string(), d.clone());
229 
                }
230 
                end = i;
231 
            }
232 
        }
233 
        let part = Snapshot {
234 
            tree: base_snapshot.tree.clone(),
235 
            files,
236 
            unrendered: BTreeSet::new(),
237 
        };
d616faf5 Andy McCall ↶ 238 
        drop(pick);
239 
        let _build = crate::timing::phase("snapshot:   of which building the change");
23ff1288 Andy McCall ↶ 240 
        // Whether anything on disk is under `dir`.
241 
        let dir_needed = |dir: &str| {
242 
            let prefix = format!("{dir}/");
243 
            fs_entries
244 
                .range(prefix.clone()..)
245 
                .next()
246 
                .is_some_and(|(k, _)| k.starts_with(&prefix))
247 
        };
248 
        self.build_from_entries_with(
249 
            base,
250 
            &part,
251 
            reuse,
252 
            &changed,
253 
            &HashSet::new(),
254 
            Scope {
255 
                seed: None,
256 
                dir_needed: Some(&dir_needed),
257 
            },
258 
        )
259 
    }
260
dfb23a3a Andy McCall ↶ 261 
    fn build_from_entries_with(
262 
        &self,
263 
        base: &State,
264 
        base_snapshot: &Snapshot,
265 
        reuse: HashMap<String, (NodeId, NodeKind)>,
266 
        fs_entries: &BTreeMap<String, FsEntry>,
267 
        resolve: &HashSet<String>,
23ff1288 Andy McCall ↶ 268 
        scope: Scope,
dfb23a3a Andy McCall ↶ 269 
    ) -> Result<BuiltChange> {
23ff1288 Andy McCall ↶ 270 
        let Scope { seed, dir_needed } = scope;
2306a4d2 Andy McCall ↶ 271 
        let mut laps = crate::timing::Laps::start();
cdb5beba Andy McCall ↶ 272 
        // Attributes come from the files being committed, so a new
273 
        // `.oxyattributes` applies to the same change that adds it.
274 
        let attrs = crate::attributes::Attributes::from_files(
75167672 Andy McCall 275 
            fs_entries
276 
                .iter()
277 
                .filter(|(p, _)| p.as_str() == ".oxyattributes" || p.ends_with("/.oxyattributes"))
278 
                .map(|(p, e)| (p.as_str(), e.content.as_slice())),
cdb5beba Andy McCall ↶ 279 
        );
280 
        let mut b = ChangeBuilder::new(base, base_snapshot, reuse).with_attributes(attrs);
f7510b63 Andy McCall ↶ 281 
        b.resolve = resolve.clone();
dfb23a3a Andy McCall ↶ 282 
        b.node_seed = seed;
2306a4d2 Andy McCall ↶ 283 
        laps.lap("build: builder");
6e723b81 Andy McCall ↶ 284 
        // Rename detection (spec §7.1, §23): a removed file and an added file
285 
        // of the same kind whose contents are at least half alike become one
286 
        // move of the same node, so identity survives across the rename.
287 
        let renames = detect_renames(base_snapshot, fs_entries, |p| self.in_scope(p));
75167672 Andy McCall 288 
        let renamed_from: HashMap<&str, &str> = renames
289 
            .iter()
290 
            .map(|(old, new)| (new.as_str(), old.as_str()))
291 
            .collect();
292 
        let renamed_to: HashMap<&str, &str> = renames
293 
            .iter()
294 
            .map(|(old, new)| (old.as_str(), new.as_str()))
295 
            .collect();
2306a4d2 Andy McCall ↶ 296 
        laps.lap("build: renames");
c3069537 Andy McCall ↶ 297 
        for (path, fs_entry) in fs_entries {
98286406 Andy McCall ↶ 298 
            match base_snapshot.files.get(path) {
6e723b81 Andy McCall ↶ 299 
                None => match renamed_from.get(path.as_str()) {
75167672 Andy McCall 300 
                    Some(old) => {
301 
                        b.move_path(self, old, path, fs_entry, &base_snapshot.files[*old])?
302 
                    }
6e723b81 Andy McCall ↶ 303 
                    None => b.add_path(self, path, fs_entry)?,
304 
                },
98286406 Andy McCall ↶ 305 
                Some(expected) => b.update_path(self, path, fs_entry, expected)?,
306 
            }
307 
        }
2306a4d2 Andy McCall ↶ 308 
        laps.lap("build: paths");
958fb27d Andy McCall ↶ 309 
        // A directory stays while anything under it exists or is out of
310 
        // scope: the directories above each such path, gathered once.
311 
        let mut needed_dirs: HashSet<&str> = HashSet::new();
312 
        fn parents_of<'a>(p: &'a str, set: &mut HashSet<&'a str>) {
313 
            let mut end = p.len();
314 
            while let Some(i) = p[..end].rfind('/') {
315 
                if !set.insert(&p[..i]) {
316 
                    break;
317 
                }
318 
                end = i;
319 
            }
320 
        }
321 
        for p in fs_entries.keys() {
322 
            parents_of(p, &mut needed_dirs);
323 
        }
324 
        for p in base_snapshot.files.keys() {
325 
            if !self.in_scope(p) {
326 
                parents_of(p, &mut needed_dirs);
327 
            }
328 
        }
98286406 Andy McCall ↶ 329 
        for (path, expected) in &base_snapshot.files {
75167672 Andy McCall 330 
            if fs_entries.contains_key(path)
331 
                || !self.in_scope(path)
332 
                || renamed_to.contains_key(path.as_str())
333 
            {
98286406 Andy McCall ↶ 334 
                continue;
335 
            }
2306a4d2 Andy McCall ↶ 336 
            let still_needed = expected.kind == NodeKind::Dir
337 
                && match dir_needed {
338 
                    Some(f) => f(path),
339 
                    None => needed_dirs.contains(path.as_str()),
340 
                };
98286406 Andy McCall ↶ 341 
            if !still_needed {
342 
                b.remove_path(path, expected);
343 
            }
344 
        }
2306a4d2 Andy McCall ↶ 345 
        laps.lap("build: removals");
a87c7b9d Andy McCall ↶ 346 
        if seed.is_some() {
347 
            b.retire_stray_placements();
348 
        }
2306a4d2 Andy McCall ↶ 349 
        laps.lap("build: stray placements");
614541c3 Andy McCall ↶ 350 
        let mut statuses = b.statuses;
351 
        statuses.sort_by(|a, b| a.path.cmp(&b.path));
75167672 Andy McCall 352 
        Ok(BuiltChange {
353 
            tree_ops: b.tree_ops,
354 
            edits: b.edits,
355 
            contributors: b.contributors.into_iter().collect(),
356 
            resolves: b.resolves,
357 
            statuses,
358 
            blobs: b.blobs,
359 
        })
614541c3 Andy McCall ↶ 360 
    }
98286406 Andy McCall ↶ 361
c07e0e7d Andy McCall ↶ 362 
    /// Store a change and, when `core.sign = change` and we are its author,
363 
    /// sign it. A rewritten change by another author is stored unsigned: only
364 
    /// its author (or a delegate) can vouch for it.
614541c3 Andy McCall ↶ 365 
    pub(crate) fn store_signed(&self, change: &Change) -> Result<ChangeHash> {
c07e0e7d Andy McCall ↶ 366 
        let ours = change.author == self.author();
367 
        if self.config.sign_changes && ours && self.secret_key()?.is_none() {
614541c3 Andy McCall ↶ 368 
            return fail("core.sign is \"change\" but no secret key is available for author.key");
369 
        }
370 
        let hash = self.store.put_change(change)?;
c07e0e7d Andy McCall ↶ 371 
        if self.config.sign_changes && ours {
614541c3 Andy McCall ↶ 372 
            self.sign_change(&hash)?;
373 
        }
374 
        Ok(hash)
375 
    }
376
377 
    /// Snapshot the working copy into the working-copy change (spec §12.3).
cc7a0056 Andy McCall ↶ 378 
    /// The result carries the full base state and its whole rendering.
614541c3 Andy McCall ↶ 379 
    pub fn snapshot(&self, command: &str) -> Result<SnapshotResult> {
f9466021 Andy McCall ↶ 380 
        self.snapshot_as(command, false, false)
cc7a0056 Andy McCall ↶ 381 
    }
382
383 
    /// Snapshot the working copy as [`Repo::snapshot`] does, for commands
8ab265dc Andy McCall ↶ 384 
    /// that need its statuses, conflicts and base history, and the graphs
385 
    /// of the files it changes: status, log, show, commit, diff, hunks and
386 
    /// discard. Text files known unchanged since the last manifest are
cc7a0056 Andy McCall ↶ 387 
    /// neither read nor rendered: the result's `base` is a history-only
388 
    /// state without their graphs, and its `base_snapshot` lists them in
8ab265dc Andy McCall ↶ 389 
    /// `unrendered` with empty bytes. They are never conflicted, and no
390 
    /// change the working copy makes touches them.
cc7a0056 Andy McCall ↶ 391 
    pub fn snapshot_status(&self, command: &str) -> Result<SnapshotResult> {
f9466021 Andy McCall ↶ 392 
        self.snapshot_as(command, true, false)
cc7a0056 Andy McCall ↶ 393 
    }
394
cd4eb55e Andy McCall ↶ 395 
    /// `snapshot_as`; `for_commit` writes the manifest even where a status
396 
    /// would not bother (an edited file is read next time whatever it
397 
    /// says), since after the commit that file is unchanged, and records no
398 
    /// operation: the commit's own, recorded straight after, replaces the
399 
    /// working copy, so the snapshot's would cost a durable write for
400 
    /// nothing. Should the commit fail, the files are still on disk and the
401 
    /// next snapshot finds them.
f9466021 Andy McCall ↶ 402 
    fn snapshot_as(
403 
        &self,
404 
        command: &str,
405 
        partial: bool,
cd4eb55e Andy McCall ↶ 406 
        for_commit: bool,
f9466021 Andy McCall ↶ 407 
    ) -> Result<SnapshotResult> {
4e88e774 Andy McCall ↶ 408 
        let _lock = self.lock_snapshot()?;
8f10a07a Andy McCall ↶ 409 
        let batch = self.store.begin_batch()?;
cd4eb55e Andy McCall ↶ 410 
        let result = self.snapshot_locked(command, partial, for_commit);
8f10a07a Andy McCall ↶ 411 
        batch.finish()?;
412 
        result
413 
    }
414
f9466021 Andy McCall ↶ 415 
    fn snapshot_locked(
416 
        &self,
417 
        command: &str,
418 
        partial: bool,
cd4eb55e Andy McCall ↶ 419 
        for_commit: bool,
f9466021 Andy McCall ↶ 420 
    ) -> Result<SnapshotResult> {
0bdc3f91 Andy McCall ↶ 421 
        let _all = crate::timing::phase("snapshot");
614541c3 Andy McCall ↶ 422 
        let (op_ref, op) = self.head_op()?;
423 
        let (old_hash, old_wc) = self.working_copy()?;
cc7a0056 Andy McCall ↶ 424 
        // The base's history first: its graph keys say which files the
425 
        // manifest already knows equal to it.
426 
        let mut base = {
427 
            let _t = crate::timing::phase("snapshot: load base history");
f9575918 Andy McCall ↶ 428 
            self.load_state_indexed(&old_wc.deps)?
0bdc3f91 Andy McCall ↶ 429 
        };
d0d52107 Andy McCall ↶ 430 
        let view = {
431 
            let _t = crate::timing::phase("snapshot: tree view");
432 
            base.tree().view()
433 
        };
8b276e01 Andy McCall ↶ 434 
        let manifest = {
435 
            let _t = crate::timing::phase("snapshot: read manifest");
436 
            self.read_manifest()?.map(|(_, _, m)| m)
437 
        };
cc7a0056 Andy McCall ↶ 438 
        let tracked: BTreeMap<String, (NodeKind, Option<ObjectRef>)> = {
439 
            let _t = crate::timing::phase("snapshot: tracked paths");
36e51aa8 Andy McCall ↶ 440 
            // In the view's path order, so the map is built without searching.
441 
            view.entries
442 
                .iter()
443 
                .filter(|(_, e)| e.kind != NodeKind::Dir)
444 
                .map(|(p, e)| {
445 
                    let blob = if e.kind != NodeKind::File || e.mode.opaque {
446 
                        base.live_blobs(&e.node).first().copied()
447 
                    } else {
448 
                        None
449 
                    };
450 
                    (p.clone(), (e.kind, blob))
451 
                })
452 
                .collect()
cc7a0056 Andy McCall ↶ 453 
        };
992467d3 Andy McCall ↶ 454 
        // A text file's graph key, worked out when the walk asks for it.
1966f5db Andy McCall ↶ 455 
        let key_of = |p: &str| -> Option<[u8; 32]> {
992467d3 Andy McCall ↶ 456 
            let e = view.entries.get(p)?;
457 
            if e.kind != NodeKind::File || e.mode.opaque {
458 
                return None;
459 
            }
1966f5db Andy McCall ↶ 460 
            base.graph_key(&e.node)
992467d3 Andy McCall ↶ 461 
        };
cc7a0056 Andy McCall ↶ 462 
        let mut fs_entries = {
463 
            let _t = crate::timing::phase("snapshot: read working copy");
992467d3 Andy McCall ↶ 464 
            self.walk_working_copy_with(manifest.as_deref(), &tracked, Some(&key_of))?
cc7a0056 Andy McCall ↶ 465 
        };
992467d3 Andy McCall ↶ 466 
        // The keys of the files read, before building graphs drops them: a
467 
        // file found equal to its base rendering keeps its key.
1966f5db Andy McCall ↶ 468 
        let keys: HashMap<String, [u8; 32]> = fs_entries
992467d3 Andy McCall ↶ 469 
            .iter()
470 
            .filter(|(_, e)| !e.unchanged && e.kind == NodeKind::File)
471 
            .filter_map(|(p, _)| Some((p.clone(), key_of(p)?)))
472 
            .collect();
0bdc3f91 Andy McCall ↶ 473 
        let base_snapshot = {
cc7a0056 Andy McCall ↶ 474 
            let _t = crate::timing::phase("snapshot: load and render base");
475 
            if partial {
d616faf5 Andy McCall ↶ 476 
                let sets = crate::timing::phase("snapshot:   of which choosing graphs");
477 
                // Every file but those on disk known unchanged.
478 
                let need: Vec<(&str, NodeId)> = join(&view.entries, &fs_entries)
479 
                    .filter_map(|(p, v, f)| {
480 
                        let v = v?;
481 
                        (v.kind == NodeKind::File && !f.is_some_and(|f| f.unchanged))
482 
                            .then_some((p.as_str(), v.node))
483 
                    })
cc7a0056 Andy McCall ↶ 484 
                    .collect();
d616faf5 Andy McCall ↶ 485 
                let render: std::collections::HashSet<&str> =
486 
                    need.iter().map(|(p, _)| *p).collect();
487 
                let need: Vec<NodeId> = need.iter().map(|(_, n)| *n).collect();
488 
                drop(sets);
d0990c2b Andy McCall ↶ 489 
                {
490 
                    let _t = crate::timing::phase("snapshot:   of which building graphs");
491 
                    self.load_graphs_of(&mut base, &need)?;
492 
                }
493 
                let _t = crate::timing::phase("snapshot:   of which rendering");
d616faf5 Andy McCall ↶ 494 
                base.materialize_only(&render)
cc7a0056 Andy McCall ↶ 495 
            } else {
496 
                self.load_all_graphs(&mut base)?;
497 
                base.materialize()
498 
            }
0bdc3f91 Andy McCall ↶ 499 
        };
614541c3 Andy McCall ↶ 500
501 
        // Reuse the NodeIds the previous working-copy version assigned to new
502 
        // paths, so an unchanged working copy hashes identically (spec §12.3).
503 
        let reuse = added_node_paths(&old_wc, &base_snapshot);
c863c7a7 Andy McCall ↶ 504 
        self.platform_normalize(&mut fs_entries, &base_snapshot);
0bdc3f91 Andy McCall ↶ 505 
        let built = {
506 
            let _t = crate::timing::phase("snapshot: diff against base");
23ff1288 Andy McCall ↶ 507 
            if partial && self.config.sparse_paths.is_empty() {
508 
                #[cfg(test)]
509 
                let full =
510 
                    self.build_from_entries(&base, &base_snapshot, reuse.clone(), &fs_entries)?;
511 
                let built = self.build_changed_only(&base, &base_snapshot, reuse, &fs_entries)?;
512 
                #[cfg(test)]
513 
                {
514 
                    // New nodes take random identifiers: compare them by
515 
                    // order of appearance.
516 
                    let norm = |b: &BuiltChange| {
517 
                        let text = format!("{:?} {:?} {:?}", b.tree_ops, b.edits, b.statuses);
518 
                        let mut seen: Vec<String> = Vec::new();
519 
                        let mut out = String::new();
520 
                        let mut rest = text.as_str();
521 
                        while let Some(i) = rest.find("NodeId(") {
522 
                            out.push_str(&rest[..i]);
523 
                            let end = i + rest[i..].find(')').unwrap() + 1;
524 
                            let id = rest[i..end].to_string();
525 
                            let n = seen.iter().position(|x| *x == id).unwrap_or_else(|| {
526 
                                seen.push(id);
527 
                                seen.len() - 1
528 
                            });
529 
                            out.push_str(&format!("N{n}"));
530 
                            rest = &rest[end..];
531 
                        }
532 
                        out.push_str(rest);
533 
                        out
534 
                    };
535 
                    assert_eq!(
536 
                        norm(&built),
537 
                        norm(&full),
538 
                        "a changed-only build is the full one"
539 
                    );
540 
                }
541 
                built
542 
            } else {
543 
                self.build_from_entries(&base, &base_snapshot, reuse, &fs_entries)?
544 
            }
0bdc3f91 Andy McCall ↶ 545 
        };
cc7a0056 Andy McCall ↶ 546 
        // Text files found equal to an unconflicted base rendering keep its
547 
        // graph key, so the next snapshot need not read or render them.
548 
        {
549 
            let changed: HashSet<&str> = built.statuses.iter().map(|s| s.path.as_str()).collect();
550 
            for (p, e) in fs_entries.iter_mut() {
551 
                if e.unchanged || e.kind != NodeKind::File || changed.contains(p.as_str()) {
552 
                    continue;
553 
                }
554 
                e.graph_key = match base_snapshot.files.get(p).map(|f| &f.content) {
555 
                    Some(FileContent::Text {
556 
                        bytes,
557 
                        materialized,
40a40bb7 Andy McCall ↶ 558 
                    }) if !materialized.is_conflicted() && *bytes == *e.content => {
1966f5db Andy McCall ↶ 559 
                        keys.get(p).copied()
cc7a0056 Andy McCall ↶ 560 
                    }
561 
                    _ => None,
562 
                };
563 
            }
564 
        }
614541c3 Andy McCall ↶ 565
98286406 Andy McCall ↶ 566 
        let mut wc = Change {
6f8aeb99 Andy McCall ↶ 567 
            version: crate::objects::CHANGE_VERSION,
98286406 Andy McCall ↶ 568 
            change_id: old_wc.change_id,
569 
            deps: vec![],
570 
            author: self.author(),
571 
            hlc: old_wc.hlc,
572 
            message: old_wc.message.clone(),
614541c3 Andy McCall ↶ 573 
            tree_ops: built.tree_ops,
574 
            edits: built.edits,
98286406 Andy McCall ↶ 575 
            kind: ChangeKind::WorkingCopy,
576 
        };
577 
        let base_heads = base.maximal(&old_wc.deps);
578 
        let mut deps: BTreeSet<ChangeHash> = base_heads.into_iter().collect();
579 
        deps.extend(referenced_changes(&wc, &base));
614541c3 Andy McCall ↶ 580 
        deps.extend(built.contributors.iter().copied());
98286406 Andy McCall ↶ 581 
        wc.deps = deps.into_iter().collect();
582 
        wc.normalize();
614541c3 Andy McCall ↶ 583 
        let resolves = built.resolves;
584 
        let contributors = built.contributors;
585 
        let statuses = built.statuses;
98286406 Andy McCall ↶ 586 
        // Empty snapshot: hash the candidate before storing; only record an op on change.
8b276e01 Andy McCall ↶ 587 
        let record = crate::timing::phase("snapshot: hash and record");
17f94b13 Andy McCall ↶ 588 
        // Encoded once, its texts stored a batch at a time as they are
589 
        // made (a stored blob is not stored again); the change itself only
590 
        // when it changed.
591 
        let encoded = wc.encode_streaming(|batch| self.store.put_texts(&batch))?;
192fad1e Andy McCall ↶ 592 
        let new_hash = crate::hash::hash_object(crate::hash::ObjectType::Change, &encoded);
98286406 Andy McCall ↶ 593 
        let changed = new_hash != old_hash;
cd4eb55e Andy McCall ↶ 594 
        // For a commit the change is neither stored nor recorded: the commit
595 
        // stores its own, built from this one's operations.
596 
        if changed && !for_commit {
17f94b13 Andy McCall ↶ 597 
            self.store.put(crate::hash::ObjectType::Change, &encoded)?;
98286406 Andy McCall ↶ 598 
            let mut view = op.view.clone();
599 
            view.workspaces.insert(self.workspace, new_hash);
600 
            self.record_op(&[op_ref], &format!("{command}{SNAPSHOT_SUFFIX}"), view)?;
601 
        }
8b276e01 Andy McCall ↶ 602 
        drop(record);
135212de Andy McCall ↶ 603 
        // The prompt compares against this without recording anything.
0bdc3f91 Andy McCall ↶ 604 
        let _t = crate::timing::phase("snapshot: write manifest");
992467d3 Andy McCall ↶ 605 
        let conflicted = base_snapshot.is_conflicted();
606 
        let uncommitted = !wc.tree_ops.is_empty() || !wc.edits.is_empty();
607 
        // Rewriting the manifest pays only when it would teach the next
608 
        // snapshot something: other flags, a file found equal to the base
609 
        // (it need not be read again), a new blob, a path added or gone.
610 
        // An edited file is read next time whatever it says.
611 
        let worth = match self.read_manifest()? {
cd4eb55e Andy McCall ↶ 612 
            _ if for_commit => true,
992467d3 Andy McCall ↶ 613 
            None => true,
614 
            Some((c, u, m)) => {
615 
                c != conflicted
616 
                    || u != uncommitted
617 
                    || m.entries.len() != fs_entries.len()
618 
                    || fs_entries.iter().any(|(p, e)| {
619 
                        !e.unchanged && (e.graph_key.is_some() || !m.entries.contains_key(p))
620 
                    })
621 
                    || built
622 
                        .blobs
623 
                        .iter()
624 
                        .any(|(p, b)| m.entries.get(p).and_then(|e| e.blob) != Some(*b))
36041796 Andy McCall ↶ 625 
                    // With uncommitted work, the status cache below holds
626 
                    // only while the manifest describes the edited files as
627 
                    // they are now.
628 
                    || (uncommitted && statuses.iter().any(|st| {
629 
                        let Some(e) = fs_entries.get(&st.path) else {
630 
                            return m.entries.contains_key(&st.path);
631 
                        };
632 
                        let Some(me) = m.entries.get(&st.path) else {
633 
                            return true;
634 
                        };
635 
                        let hash = e
636 
                            .content_hash
637 
                            .unwrap_or_else(|| *blake3::hash(&e.content).as_bytes());
638 
                        let stat = e.stat.or_else(|| {
639 
                            fs::symlink_metadata(self.root.join(&st.path))
640 
                                .ok()
641 
                                .map(|md| (md.len(), super::prompt::mtime_ns(&md)))
642 
                        });
e226a6cc Andy McCall ↶ 643 
                        // The executable bit too: a mode change moves
644 
                        // neither size nor time.
645 
                        me.hash != hash || me.exec != e.exec || stat != Some((me.size, me.mtime_ns))
36041796 Andy McCall ↶ 646 
                    }))
992467d3 Andy McCall ↶ 647 
            }
648 
        };
f9466021 Andy McCall ↶ 649 
        let attributes = crate::attributes::Attributes::from_files(
650 
            fs_entries
651 
                .iter()
652 
                .filter(|(p, _)| is_attributes(p))
653 
                .map(|(p, e)| (p.as_str(), e.content.as_slice())),
654 
        );
236f35c0 Andy McCall ↶ 655 
        let mut manifest = None;
656 
        if for_commit {
657 
            manifest = Some(PendingManifest {
658 
                entries: fs_entries,
659 
                blobs: built.blobs.clone(),
660 
                dirs: self.walked_dirs.take(),
661 
                written: super::prompt::now_ns(),
662 
            });
663 
        } else {
664 
            if worth {
665 
                self.write_manifest_entries(conflicted, uncommitted, &fs_entries, &built.blobs)?;
666 
            }
667 
            self.write_status_cache((uncommitted && !conflicted).then_some(&new_hash), &statuses);
668 
        }
75167672 Andy McCall 669 
        Ok(SnapshotResult {
670 
            wc_hash: new_hash,
671 
            wc,
672 
            base,
673 
            base_snapshot,
674 
            statuses,
675 
            changed,
676 
            resolves,
677 
            contributors,
f9466021 Andy McCall ↶ 678 
            attributes,
d4dbab97 Andy McCall ↶ 679 
            untracked_nested: self.untracked_nested.borrow().clone(),
236f35c0 Andy McCall ↶ 680 
            manifest,
75167672 Andy McCall 681 
        })
98286406 Andy McCall ↶ 682 
    }
683
50a046e4 Andy McCall ↶ 684 
    /// `oxy status --json` (spec §9.5): the changed paths and every conflict
685 
    /// of the working copy's base with its kind, path, and contributing
686 
    /// ChangeIds.
687 
    pub fn status_json(&self, snap: &SnapshotResult) -> Result<String> {
688 
        let heads = self.bookmark_heads()?;
689 
        let base_snapshot = &snap.base_snapshot;
690 
        let by_node = base_snapshot.tree.by_node();
691 
        let path_of = |n: &NodeId| by_node.get(n).map(|e| e.path.clone());
75167672 Andy McCall 692 
        let id_of = |h: &ChangeHash| {
693 
            snap.base
694 
                .get(h)
695 
                .map(|m| m.change_id.to_hex())
696 
                .unwrap_or_else(|| h.to_hex())
697 
        };
698 
        let paths = json_array(snap.statuses.iter().map(|s| {
699 
            format!(
700 
                "{{\"path\":{},\"status\":{}}}",
701 
                json_str(&s.path),
702 
                json_str(&format!("{:?}", s.status))
703 
            )
704 
        }));
50a046e4 Andy McCall ↶ 705 
        let mut conflicts: Vec<String> = Vec::new();
706 
        for (p, f) in &base_snapshot.files {
707 
            if !f.is_conflicted() {
708 
                continue;
709 
            }
710 
            let (kind, changes) = match &f.content {
75167672 Andy McCall 711 
                FileContent::Text { .. } => {
712 
                    ("text", snap.base.conflict_regions(&f.node).contributors)
713 
                }
714 
                FileContent::Opaque { blobs } => {
715 
                    ("replace", snap.base.blob_contributors(&f.node, blobs))
716 
                }
50a046e4 Andy McCall ↶ 717 
                FileContent::Dir => continue,
718 
            };
75167672 Andy McCall 719 
            let resolved = match snap
720 
                .statuses
721 
                .iter()
722 
                .find(|s| s.path == *p)
723 
                .map(|s| s.status)
724 
            {
50a046e4 Andy McCall ↶ 725 
                Some(ChangeStatus::Resolved) => "true",
726 
                Some(ChangeStatus::ConflictEdited) => "\"markers\"",
727 
                _ => "false",
728 
            };
729 
            conflicts.push(format!(
730 
                "{{\"kind\":{},\"path\":{},\"node\":{},\"changes\":{},\"resolved\":{resolved}}}",
731 
                json_str(kind),
732 
                json_str(p),
733 
                json_str(&f.node.to_hex()),
734 
                json_array(changes.iter().map(|h| json_str(&id_of(h))))
735 
            ));
736 
        }
737 
        for c in &base_snapshot.tree.conflicts {
738 
            let (kind, nodes): (&str, Vec<NodeId>) = match c {
739 
                crate::tree::TreeConflict::Name { nodes, .. } => ("name", nodes.clone()),
75167672 Andy McCall 740 
                crate::tree::TreeConflict::MultiPlacement { node, .. } => {
741 
                    ("multi-placement", vec![*node])
742 
                }
50a046e4 Andy McCall ↶ 743 
                crate::tree::TreeConflict::Orphan { node, .. } => ("orphan", vec![*node]),
744 
                crate::tree::TreeConflict::Cycle { node } => ("cycle", vec![*node]),
745 
                crate::tree::TreeConflict::Mode { node, .. } => ("mode", vec![*node]),
746 
            };
747 
            let detail = match c {
748 
                crate::tree::TreeConflict::Name { name, parent, .. } => {
75167672 Andy McCall 749 
                    let dir = if *parent == NodeId::ROOT {
750 
                        String::new()
751 
                    } else {
752 
                        path_of(parent).unwrap_or_default()
753 
                    };
754 
                    format!(
755 
                        ",\"path\":{}",
756 
                        json_str(&if dir.is_empty() {
757 
                            name.clone()
758 
                        } else {
759 
                            format!("{dir}/{name}")
760 
                        })
761 
                    )
50a046e4 Andy McCall ↶ 762 
                }
75167672 Andy McCall 763 
                _ => nodes
764 
                    .first()
765 
                    .and_then(path_of)
766 
                    .map(|p| format!(",\"path\":{}", json_str(&p)))
767 
                    .unwrap_or_default(),
50a046e4 Andy McCall ↶ 768 
            };
769 
            conflicts.push(format!(
770 
                "{{\"kind\":{},\"nodes\":{}{detail}}}",
771 
                json_str(kind),
772 
                json_array(nodes.iter().map(|n| json_str(&n.to_hex())))
773 
            ));
774 
        }
775 
        Ok(format!(
d4dbab97 Andy McCall ↶ 776 
            "{{\"bookmark\":{},\"working_copy\":{},\"heads\":{},\"clean\":{},\"conflicted\":{},\"paths\":{paths},\"conflicts\":{},\"untracked_nested\":{}}}",
50a046e4 Andy McCall ↶ 777 
            json_str(&self.config.bookmark),
778 
            json_str(&snap.wc.change_id.to_hex()),
779 
            heads.len(),
780 
            snap.statuses.is_empty(),
781 
            base_snapshot.is_conflicted(),
d4dbab97 Andy McCall ↶ 782 
            json_array(conflicts),
783 
            json_array(snap.untracked_nested.iter().map(|p| json_str(p)))
50a046e4 Andy McCall ↶ 784 
        ))
785 
    }
786
98286406 Andy McCall ↶ 787 
    /// Turn the working-copy change into a normal change on the bookmark (spec §12.3).
788 
    pub fn commit(&self, message: &str, extra_deps: &[ChangeHash]) -> Result<ChangeHash> {
957fee30 Andy McCall ↶ 789 
        self.commit_selection(message, extra_deps, &Selection::All)
790 
    }
791
792 
    /// The hunks of the working copy, for `oxy commit --interactive`: each
793 
    /// tree operation and each edit, in the order `select_hunks` numbers them.
794 
    pub fn hunks(&self) -> Result<(SnapshotResult, Vec<Hunk>)> {
8ab265dc Andy McCall ↶ 795 
        let snap = self.snapshot_status("oxy commit")?;
957fee30 Andy McCall ↶ 796 
        let hunks = render_hunks(&snap.wc, &snap.base, &snap.base_snapshot);
797 
        Ok((snap, hunks))
798 
    }
799
800 
    /// Commit part of the working copy (spec §12.3). What is not selected
801 
    /// stays on disk as uncommitted work: the next snapshot rebuilds it
802 
    /// against the new base, so the working copy always describes the
803 
    /// directory.
f9466021 Andy McCall ↶ 804 
    /// The graph key in `st` of each text file at `paths`, the paths a
805 
    /// commit changed: only their graphs changed, so the keys the manifest
806 
    /// holds for the others still stand (and any that does not is simply
807 
    /// not trusted by the next snapshot, which compares keys). Files with
808 
    /// an `eol` attribute are left out: their bytes on disk are not the
809 
    /// rendering.
810 
    fn committed_graph_keys(
811 
        &self,
812 
        st: &State,
813 
        paths: &[String],
814 
        attrs: &crate::attributes::Attributes,
815 
    ) -> HashMap<String, [u8; 32]> {
86e1824f Andy McCall ↶ 816 
        let _t = crate::timing::phase("commit: graph keys");
f9466021 Andy McCall ↶ 817 
        let view = st.tree().view();
818 
        paths
86e1824f Andy McCall ↶ 819 
            .iter()
f9466021 Andy McCall ↶ 820 
            .filter_map(|p| {
821 
                let e = view.entries.get(p)?;
822 
                (e.kind == NodeKind::File && !e.mode.opaque && attrs.lookup(p).eol.is_none())
823 
                    .then_some(())?;
824 
                Some((p.clone(), st.graph_key(&e.node)?))
86e1824f Andy McCall ↶ 825 
            })
8457a6cc Andy McCall ↶ 826 
            .collect()
86e1824f Andy McCall ↶ 827 
    }
828
75167672 Andy McCall 829 
    pub fn commit_selection(
830 
        &self,
831 
        message: &str,
832 
        extra_deps: &[ChangeHash],
833 
        selection: &Selection,
834 
    ) -> Result<ChangeHash> {
b30a880f Andy McCall ↶ 835 
        // The commit's objects are written together, made durable at once
836 
        // before the operation that names them (spec §13.6).
837 
        let batch = self.store.begin_batch()?;
838 
        let out = self.commit_selection_batched(message, extra_deps, selection);
839 
        batch.finish()?;
840 
        out
841 
    }
842
236f35c0 Andy McCall ↶ 843 
    /// Write the manifest a commit's snapshot left to the commit, with the
844 
    /// graph keys `keys` sets.
845 
    fn write_pending_manifest(
846 
        &self,
847 
        mut m: PendingManifest,
848 
        conflicted: bool,
849 
        keys: HashMap<String, [u8; 32]>,
850 
    ) -> Result<()> {
851 
        for (path, k) in keys {
852 
            if let Some(e) = m.entries.get_mut(&path) {
853 
                e.graph_key = Some(k);
854 
            }
855 
        }
856 
        self.write_manifest_entries_as_of(
857 
            conflicted,
858 
            false,
859 
            &m.entries,
860 
            &m.blobs,
861 
            m.written,
862 
            m.dirs.unwrap_or_default(),
863 
        )
864 
    }
865
b30a880f Andy McCall ↶ 866 
    fn commit_selection_batched(
867 
        &self,
868 
        message: &str,
869 
        extra_deps: &[ChangeHash],
870 
        selection: &Selection,
871 
    ) -> Result<ChangeHash> {
6caacfff Andy McCall ↶ 872 
        let _wc = self.lock_working_copy()?;
e226a6cc Andy McCall ↶ 873 
        let mut snap = match self.snapshot_from_cache()? {
874 
            Some(snap) => snap,
875 
            None => self.snapshot_as("oxy commit", true, true)?,
876 
        };
236f35c0 Andy McCall ↶ 877 
        let pending = snap.manifest.take();
98286406 Andy McCall ↶ 878 
        if snap.wc.tree_ops.is_empty() && snap.wc.edits.is_empty() {
236f35c0 Andy McCall ↶ 879 
            if let Some(m) = pending {
880 
                let conflicted = snap.base_snapshot.is_conflicted();
881 
                self.write_pending_manifest(m, conflicted, HashMap::new())?;
882 
            }
98286406 Andy McCall ↶ 883 
            return fail("nothing to commit: the working copy matches its base");
884 
        }
75167672 Andy McCall 885 
        if let Some(s) = snap
886 
            .statuses
887 
            .iter()
888 
            .find(|s| s.status == ChangeStatus::ConflictEdited)
889 
        {
98286406 Andy McCall ↶ 890 
            return fail(format!("{} still contains conflict markers", s.path));
891 
        }
892 
        if message.trim().is_empty() {
893 
            return fail("a commit message is required");
894 
        }
192fad1e Andy McCall ↶ 895 
        // What the whole working copy references, before a whole commit
896 
        // takes its operations over (they are moved, not copied: for a
897 
        // large tree they are most of the memory).
898 
        let wc_refs = referenced_changes(&snap.wc, &snap.base);
f9466021 Andy McCall ↶ 899 
        let changed_paths: Vec<String> = snap.statuses.iter().map(|s| s.path.clone()).collect();
957fee30 Andy McCall ↶ 900 
        let (tree_ops, edits, partial) = match selection {
192fad1e Andy McCall ↶ 901 
            Selection::All => (
902 
                std::mem::take(&mut snap.wc.tree_ops),
903 
                std::mem::take(&mut snap.wc.edits),
904 
                false,
905 
            ),
957fee30 Andy McCall ↶ 906 
            Selection::Paths(paths) => {
907 
                let (o, e) = select_paths(&snap.wc, &snap.base_snapshot, paths);
908 
                let partial = o.len() != snap.wc.tree_ops.len() || e.len() != snap.wc.edits.len();
909 
                (o, e, partial)
910 
            }
911 
            Selection::Hunks(indices) => {
912 
                let (o, e) = select_hunks(&snap.wc, indices);
913 
                let partial = o.len() != snap.wc.tree_ops.len() || e.len() != snap.wc.edits.len();
914 
                (o, e, partial)
915 
            }
916 
        };
917 
        if tree_ops.is_empty() && edits.is_empty() {
918 
            return fail("nothing selected to commit");
919 
        }
920 
        if partial && snap.resolves {
921 
            return fail("a conflict resolution must be committed whole");
922 
        }
6d3eefbc Andy McCall ↶ 923 
        // pre-commit hook: may veto.
924 
        self.run_hook(
925 
            "pre-commit",
926 
            &format!(
927 
                "{{\"event\":\"pre-commit\",\"message\":{},\"paths\":{}}}",
928 
                json_str(message),
75167672 Andy McCall 929 
                json_array(snap.statuses.iter().map(|s| format!(
930 
                    "{{\"path\":{},\"status\":{}}}",
931 
                    json_str(&s.path),
932 
                    json_str(&format!("{:?}", s.status))
933 
                )))
6d3eefbc Andy McCall ↶ 934 
            ),
935 
        )?;
98286406 Andy McCall ↶ 936 
        let (op_ref, op) = self.head_op()?;
192fad1e Andy McCall ↶ 937 
        let mut deps: BTreeSet<ChangeHash> = wc_refs;
6d3eefbc Andy McCall ↶ 938 
        // semantic-deps hook: may add explicit dependencies (spec §8.1).
939 
        if self.config.semantic_deps {
940 
            let touched: Vec<String> = snap.statuses.iter().map(|s| s.path.clone()).collect();
941 
            let by_node = snap.base_snapshot.tree.by_node();
942 
            let mut candidates: Vec<(ChangeId, ChangeHash, Vec<String>)> = Vec::new();
943 
            let mut seen: BTreeSet<ChangeHash> = BTreeSet::new();
944 
            for h in deps.iter().chain(snap.base.maximal(&snap.wc.deps).iter()) {
945 
                if !seen.insert(*h) {
946 
                    continue;
947 
                }
948 
                let c = self.store.get_change(h)?;
949 
                let mut paths: BTreeSet<String> = BTreeSet::new();
75167672 Andy McCall 950 
                for n in c
951 
                    .tree_ops
952 
                    .iter()
953 
                    .map(|o| o.node())
954 
                    .chain(c.edits.iter().map(|e| e.node()))
955 
                {
6d3eefbc Andy McCall ↶ 956 
                    if let Some(e) = by_node.get(&n) {
957 
                        paths.insert(e.path.clone());
958 
                    }
959 
                }
960 
                candidates.push((c.change_id, *h, paths.into_iter().collect()));
961 
            }
962 
            for d in self.semantic_deps(&snap.wc.change_id, &touched, &candidates)? {
963 
                if snap.base.contains(&d) {
964 
                    deps.insert(d);
965 
                }
966 
            }
967 
        }
98286406 Andy McCall ↶ 968 
        // A resolution depends on every change that contributed to the
969 
        // conflict (spec §8.1). Tree and opaque conflicts are resolved through
970 
        // ordinary operations whose references already cover the contributors.
971 
        deps.extend(snap.contributors.iter().copied());
972 
        for d in extra_deps {
973 
            if !snap.base.contains(d) {
75167672 Andy McCall 974 
                return fail(format!(
975 
                    "--depends-on {d}: not in the working copy's base state"
976 
                ));
98286406 Andy McCall ↶ 977 
            }
978 
            deps.insert(*d);
979 
        }
980 
        let mut change = Change {
6f8aeb99 Andy McCall ↶ 981 
            version: crate::objects::CHANGE_VERSION,
75167672 Andy McCall 982 
            change_id: if partial {
983 
                ChangeId::random()
984 
            } else {
985 
                snap.wc.change_id
986 
            },
957fee30 Andy McCall ↶ 987 
            deps: vec![],
98286406 Andy McCall ↶ 988 
            author: self.author(),
989 
            hlc: Hlc::now(),
990 
            message: message.trim_end().to_string(),
957fee30 Andy McCall ↶ 991 
            tree_ops,
992 
            edits,
75167672 Andy McCall 993 
            kind: if snap.resolves {
994 
                ChangeKind::Resolution
995 
            } else {
996 
                ChangeKind::Normal
997 
            },
98286406 Andy McCall ↶ 998 
        };
957fee30 Andy McCall ↶ 999 
        if partial {
1000 
            // Dependencies come from what the selected part references, not
1001 
            // from the whole working copy.
1002 
            deps = referenced_changes(&change, &snap.base);
1003 
            for d in extra_deps {
1004 
                deps.insert(*d);
1005 
            }
1006 
        }
1007 
        change.deps = deps.into_iter().collect();
98286406 Andy McCall ↶ 1008 
        change.normalize();
f9466021 Andy McCall ↶ 1009 
        let signing = crate::timing::phase("commit: sign and store");
614541c3 Andy McCall ↶ 1010 
        let hash = self.store_signed(&change)?;
f9466021 Andy McCall ↶ 1011 
        drop(signing);
1012 
        let extending = crate::timing::phase("commit: extend the state");
98286406 Andy McCall ↶ 1013
1014 
        // New bookmark heads: maximal elements of old heads plus the new change.
1015 
        // The working copy continues on its own base plus the new change, which
1016 
        // is exactly what the files on disk contain; the bookmark may hold
1017 
        // other heads (and conflicts) the working copy has not checked out.
1018 
        let old_heads = self.heads_of(&op.view, &self.config.bookmark)?;
1019 
        let old_base = snap.base.maximal(&snap.wc.deps);
1020 
        let mut all: Vec<ChangeHash> = old_heads.clone();
1021 
        all.extend(old_base.iter().copied());
1022 
        all.push(hash);
07eb99be Andy McCall ↶ 1023 
        // The snapshot already loaded the working copy's base; when that also
1024 
        // holds every bookmark head, it is the state to extend.
1025 
        let mut st = if old_heads.iter().all(|h| snap.base.contains(h)) {
1026 
            snap.base
1027 
        } else {
1028 
            self.load_state(&all)?
1029 
        };
192fad1e Andy McCall ↶ 1030 
        st.add_hashed(hash, change)?;
98286406 Andy McCall ↶ 1031 
        let mut with_new = old_heads.clone();
1032 
        with_new.push(hash);
1033 
        let new_heads = st.maximal(&with_new);
252e8cb2 Andy McCall ↶ 1034 
        // Automatic checkpoint once enough changes accumulated (spec §11).
75167672 Andy McCall 1035 
        let since = self
1036 
            .latest_checkpoint(&self.config.bookmark)?
1037 
            .map(|(_, cp)| cp.count)
1038 
            .unwrap_or(0);
1039 
        if self.config.checkpoint_interval > 0
1040 
            && (st.len() as u64).saturating_sub(since) >= self.config.checkpoint_interval
1041 
        {
252e8cb2 Andy McCall ↶ 1042 
            let cp_state = self.load_state_full(&new_heads)?;
1043 
            self.write_checkpoint(&self.config.bookmark, &cp_state, &new_heads)?;
1044 
        }
98286406 Andy McCall ↶ 1045 
        let mut wc_base = old_base.clone();
1046 
        wc_base.push(hash);
1047 
        let wc_base = st.maximal(&wc_base);
f9466021 Andy McCall ↶ 1048 
        drop(extending);
98286406 Andy McCall ↶ 1049
8457a6cc Andy McCall ↶ 1050 
        // The graph keys of the files as they now are, from the state the
1051 
        // commit just extended: a file whose graph it built has none, and
1052 
        // is read by the next snapshot; a key the next base does not share
1053 
        // (the bookmark held other heads) is simply not trusted then.
236f35c0 Andy McCall ↶ 1054 
        let keying = crate::timing::phase("commit: graph keys of the files");
f9466021 Andy McCall ↶ 1055 
        let committed_keys =
1056 
            (!partial).then(|| self.committed_graph_keys(&st, &changed_paths, &snap.attributes));
236f35c0 Andy McCall ↶ 1057 
        drop(keying);
f9466021 Andy McCall ↶ 1058 
        let recording = crate::timing::phase("commit: record");
98286406 Andy McCall ↶ 1059 
        let wc = self.new_working_copy(wc_base, "");
1060 
        let wc_hash = self.store.put_change(&wc)?;
1061 
        let mut view = op.view.clone();
75167672 Andy McCall 1062 
        view.bookmarks.insert(
1063 
            self.config.bookmark.clone(),
1064 
            self.store.put_state(&StateObject::new(new_heads))?,
1065 
        );
98286406 Andy McCall ↶ 1066 
        view.workspaces.insert(self.workspace, wc_hash);
1067 
        self.record_op(&[op_ref], &format!("oxy commit -m {message:?}"), view)?;
f9466021 Andy McCall ↶ 1068 
        drop(recording);
236f35c0 Andy McCall ↶ 1069 
        let rewriting = crate::timing::phase("commit: manifest");
957fee30 Andy McCall ↶ 1070 
        if partial {
1071 
            // The rest is rebuilt from disk by the next snapshot.
1072 
            self.drop_manifest();
1073 
        } else {
86e1824f Andy McCall ↶ 1074 
            let conflicted = snap.base_snapshot.is_conflicted() && !snap.resolves;
1075 
            // The files on disk are now exactly the new base's renderings:
1076 
            // record their graph keys, so the next snapshot neither reads nor
1077 
            // renders them (oxygen#35).
1078 
            let keys = if conflicted {
1079 
                HashMap::new()
1080 
            } else {
8457a6cc Andy McCall ↶ 1081 
                committed_keys.unwrap_or_default()
86e1824f Andy McCall ↶ 1082 
            };
236f35c0 Andy McCall ↶ 1083 
            match pending {
1084 
                Some(m) => self.write_pending_manifest(m, conflicted, keys)?,
1085 
                None => self.rewrite_manifest(conflicted, false, &keys)?,
1086 
            }
957fee30 Andy McCall ↶ 1087 
        }
236f35c0 Andy McCall ↶ 1088 
        drop(rewriting);
1089 
        let packing = crate::timing::phase("commit: autopack");
d02e545d Andy McCall ↶ 1090 
        self.maybe_autopack()?;
236f35c0 Andy McCall ↶ 1091 
        drop(packing);
1092 
        let _hook = crate::timing::phase("commit: hook");
75167672 Andy McCall 1093 
        let _ = self.run_hook(
1094 
            "post-commit",
1095 
            &format!(
1096 
                "{{\"event\":\"post-commit\",\"change_id\":{},\"hash\":{},\"message\":{}}}",
1097 
                json_str(&snap.wc.change_id.to_hex()),
1098 
                json_str(&hash.to_hex()),
1099 
                json_str(message)
1100 
            ),
1101 
        );
98286406 Andy McCall ↶ 1102 
        Ok(hash)
1103 
    }
1104 
}
1105
957fee30 Andy McCall ↶ 1106 
/// What part of the working copy a commit takes (spec §12.3).
1107 
#[derive(Clone, Debug)]
1108 
pub enum Selection {
1109 
    All,
1110 
    /// Every operation and edit on nodes under these paths.
1111 
    Paths(Vec<String>),
1112 
    /// Hunk indices as `Repo::hunks` numbers them; dependent hunks are added.
1113 
    Hunks(Vec<usize>),
1114 
}
1115
1116 
/// One selectable piece of the working copy: a tree operation or an edit.
1117 
#[derive(Clone, Debug)]
1118 
pub struct Hunk {
1119 
    pub index: usize,
1120 
    pub path: String,
1121 
    /// Rendered like `oxy diff`.
1122 
    pub text: String,
1123 
}
1124
1125 
/// The tree operations and edits on nodes under `paths`, plus the
1126 
/// additions of any new parent directories they need.
75167672 Andy McCall 1127 
fn select_paths(
1128 
    wc: &Change,
1129 
    base_snapshot: &Snapshot,
1130 
    paths: &[String],
1131 
) -> (Vec<TreeOp>, Vec<Edit>) {
1132 
    let selected = |p: &str| {
1133 
        paths
1134 
            .iter()
1135 
            .any(|x| p == x || p.starts_with(&format!("{}/", x.trim_end_matches('/'))))
1136 
    };
957fee30 Andy McCall ↶ 1137 
    let by_node = base_snapshot.tree.by_node();
1138 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
1139 
    for op in &wc.tree_ops {
75167672 Andy McCall 1140 
        if let TreeOp::AddNode {
1141 
            node,
1142 
            parent,
1143 
            name,
1144 
            kind,
1145 
            ..
1146 
        } = op
1147 
        {
957fee30 Andy McCall ↶ 1148 
            new_names.insert(*node, (*parent, name.clone(), *kind));
1149 
        }
1150 
    }
1151 
    let mut nodes: BTreeSet<NodeId> = BTreeSet::new();
75167672 Andy McCall 1152 
    for n in wc
1153 
        .tree_ops
1154 
        .iter()
1155 
        .map(|o| o.node())
1156 
        .chain(wc.edits.iter().map(|e| e.node()))
1157 
    {
957fee30 Andy McCall ↶ 1158 
        if path_of_node(&n, &by_node, &new_names).is_some_and(|p| selected(&p)) {
1159 
            nodes.insert(n);
1160 
        }
1161 
    }
1162 
    // New parents of selected new nodes come along.
1163 
    let mut changed = true;
1164 
    while changed {
1165 
        changed = false;
1166 
        for (n, (parent, _, _)) in &new_names {
1167 
            if nodes.contains(n) && new_names.contains_key(parent) && nodes.insert(*parent) {
1168 
                changed = true;
1169 
            }
1170 
        }
1171 
    }
75167672 Andy McCall 1172 
    let ops = wc
1173 
        .tree_ops
1174 
        .iter()
1175 
        .filter(|o| nodes.contains(&o.node()))
1176 
        .cloned()
1177 
        .collect();
1178 
    let edits = wc
1179 
        .edits
1180 
        .iter()
1181 
        .filter(|e| nodes.contains(&e.node()))
1182 
        .cloned()
1183 
        .collect();
957fee30 Andy McCall ↶ 1184 
    (ops, edits)
1185 
}
1186
1187 
/// The hunks `indices` name plus whatever they refer to within the working
1188 
/// copy (the parent directory of an added node, the insert that introduced
1189 
/// an anchor line), renumbered so the result is a well-formed change.
1190 
fn select_hunks(wc: &Change, indices: &[usize]) -> (Vec<TreeOp>, Vec<Edit>) {
1191 
    let n_ops = wc.tree_ops.len();
1192 
    let mut ops_sel = vec![false; n_ops];
1193 
    let mut edits_sel = vec![false; wc.edits.len()];
1194 
    for &i in indices {
1195 
        if i < n_ops {
1196 
            ops_sel[i] = true;
1197 
        } else if i - n_ops < wc.edits.len() {
1198 
            edits_sel[i - n_ops] = true;
1199 
        }
1200 
    }
1201 
    // Atom ranges each insert introduces, per node, in the working copy's numbering.
1202 
    let mut per_node: HashMap<NodeId, u32> = HashMap::new();
1203 
    let mut insert_ranges: Vec<Option<(NodeId, u32, u32)>> = Vec::new();
1204 
    for e in &wc.edits {
1205 
        insert_ranges.push(match e {
1206 
            Edit::Insert { node, lines, .. } => {
1207 
                let start = per_node.entry(*node).or_default();
1208 
                let r = (*node, *start, *start + lines.len() as u32);
1209 
                *start += lines.len() as u32;
1210 
                Some(r)
1211 
            }
1212 
            _ => None,
1213 
        });
1214 
    }
75167672 Andy McCall 1215 
    let add_index: HashMap<NodeId, usize> = wc
1216 
        .tree_ops
1217 
        .iter()
1218 
        .enumerate()
1219 
        .filter_map(|(i, o)| {
1220 
            if let TreeOp::AddNode { node, .. } = o {
1221 
                Some((*node, i))
1222 
            } else {
1223 
                None
1224 
            }
1225 
        })
1226 
        .collect();
1227 
    let insert_for = |node: &NodeId, index: u32| -> Option<usize> {
1228 
        insert_ranges
1229 
            .iter()
1230 
            .position(|r| matches!(r, Some((n, a, b)) if n == node && *a <= index && index < *b))
1231 
    };
957fee30 Andy McCall ↶ 1232 
    // Closure.
1233 
    let mut changed = true;
1234 
    while changed {
1235 
        changed = false;
1236 
        for i in 0..n_ops {
1237 
            if !ops_sel[i] {
1238 
                continue;
1239 
            }
d8c086ca Andy McCall ↶ 1240 
            if let TreeOp::AddNode { node, parent, .. } = &wc.tree_ops[i] {
957fee30 Andy McCall ↶ 1241 
                if let Some(&p) = add_index.get(parent) {
1242 
                    if !ops_sel[p] {
1243 
                        ops_sel[p] = true;
1244 
                        changed = true;
1245 
                    }
1246 
                }
d8c086ca Andy McCall ↶ 1247 
                // A new node's content belongs with its addition: nobody
1248 
                // wants an empty file committed apart from its lines.
1249 
                for (j, e) in wc.edits.iter().enumerate() {
1250 
                    if e.node() == *node && !edits_sel[j] {
1251 
                        edits_sel[j] = true;
1252 
                        changed = true;
1253 
                    }
1254 
                }
957fee30 Andy McCall ↶ 1255 
            }
1256 
        }
1257 
        for j in 0..wc.edits.len() {
1258 
            if !edits_sel[j] {
1259 
                continue;
1260 
            }
1261 
            let e = &wc.edits[j];
1262 
            if let Some(&a) = add_index.get(&e.node()) {
1263 
                if !ops_sel[a] {
1264 
                    ops_sel[a] = true;
1265 
                    changed = true;
1266 
                }
1267 
            }
75167672 Andy McCall 1268 
            if let Edit::Insert {
1269 
                node,
1270 
                after,
1271 
                before,
1272 
                ..
1273 
            } = e
1274 
            {
957fee30 Andy McCall ↶ 1275 
                for r in [after, before] {
1276 
                    if let AtomRef::Atom { change, index } = r {
1277 
                        if change.is_zero() {
1278 
                            if let Some(k) = insert_for(node, *index) {
1279 
                                if !edits_sel[k] {
1280 
                                    edits_sel[k] = true;
1281 
                                    changed = true;
1282 
                                }
1283 
                            }
1284 
                        }
1285 
                    }
1286 
                }
1287 
            }
1288 
        }
1289 
    }
1290 
    // Renumber self-references: old atom index → new index among selected inserts.
1291 
    let mut remap: HashMap<(NodeId, u32), u32> = HashMap::new();
1292 
    let mut new_counts: HashMap<NodeId, u32> = HashMap::new();
1293 
    for (j, r) in insert_ranges.iter().enumerate() {
1294 
        if let (true, Some((node, a, b))) = (edits_sel[j], r) {
1295 
            let next = new_counts.entry(*node).or_default();
1296 
            for old in *a..*b {
1297 
                remap.insert((*node, old), *next);
1298 
                *next += 1;
1299 
            }
1300 
        }
1301 
    }
1302 
    let map_ref = |r: AtomRef, node: &NodeId| -> AtomRef {
1303 
        match r {
75167672 Andy McCall 1304 
            AtomRef::Atom { change, index } if change.is_zero() => AtomRef::Atom {
1305 
                change,
1306 
                index: remap.get(&(*node, index)).copied().unwrap_or(index),
1307 
            },
957fee30 Andy McCall ↶ 1308 
            other => other,
1309 
        }
1310 
    };
75167672 Andy McCall 1311 
    let ops = wc
1312 
        .tree_ops
1313 
        .iter()
1314 
        .enumerate()
1315 
        .filter(|(i, _)| ops_sel[*i])
1316 
        .map(|(_, o)| o.clone())
1317 
        .collect();
957fee30 Andy McCall ↶ 1318 
    let edits = wc
1319 
        .edits
1320 
        .iter()
1321 
        .enumerate()
1322 
        .filter(|(j, _)| edits_sel[*j])
1323 
        .map(|(_, e)| match e {
75167672 Andy McCall 1324 
            Edit::Insert {
1325 
                node,
1326 
                after,
1327 
                before,
1328 
                lines,
1329 
            } => Edit::Insert {
1330 
                node: *node,
1331 
                after: map_ref(*after, node),
1332 
                before: map_ref(*before, node),
1333 
                lines: lines.clone(),
1334 
            },
957fee30 Andy McCall ↶ 1335 
            other => other.clone(),
1336 
        })
1337 
        .collect();
1338 
    (ops, edits)
1339 
}
1340
1341 
/// Render each tree operation and edit of `change` on its own, numbered.
1342 
pub fn render_hunks(change: &Change, base: &State, base_snapshot: &Snapshot) -> Vec<Hunk> {
1343 
    let by_node = base_snapshot.tree.by_node();
1344 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
1345 
    for op in &change.tree_ops {
75167672 Andy McCall 1346 
        if let TreeOp::AddNode {
1347 
            node,
1348 
            parent,
1349 
            name,
1350 
            kind,
1351 
            ..
1352 
        } = op
1353 
        {
957fee30 Andy McCall ↶ 1354 
            new_names.insert(*node, (*parent, name.clone(), *kind));
1355 
        }
1356 
    }
1357 
    let mut out = Vec::new();
1358 
    for (i, op) in change.tree_ops.iter().enumerate() {
75167672 Andy McCall 1359 
        let one = Change {
1360 
            tree_ops: vec![op.clone()],
1361 
            edits: vec![],
1362 
            ..change.clone()
1363 
        };
957fee30 Andy McCall ↶ 1364 
        let path = path_of_node(&op.node(), &by_node, &new_names).unwrap_or_default();
75167672 Andy McCall 1365 
        out.push(Hunk {
1366 
            index: i,
1367 
            path,
1368 
            text: render_change(&one, base, base_snapshot),
1369 
        });
957fee30 Andy McCall ↶ 1370 
    }
1371 
    for (j, e) in change.edits.iter().enumerate() {
75167672 Andy McCall 1372 
        let one = Change {
1373 
            tree_ops: vec![],
1374 
            edits: vec![e.clone()],
1375 
            ..change.clone()
1376 
        };
957fee30 Andy McCall ↶ 1377 
        let path = path_of_node(&e.node(), &by_node, &new_names).unwrap_or_default();
75167672 Andy McCall 1378 
        out.push(Hunk {
1379 
            index: change.tree_ops.len() + j,
1380 
            path,
1381 
            text: render_change(&one, base, base_snapshot),
1382 
        });
957fee30 Andy McCall ↶ 1383 
    }
1384 
    out
1385 
}
1386
c6aa5bd9 Andy McCall ↶ 1387 
/// One file's part of a diff, for `oxy diff --json`.
1388 
#[derive(Clone, Debug)]
1389 
pub struct FileDiff {
1390 
    pub path: String,
1391 
    /// `added`, `removed`, `modified`, `renamed`, `mode`, `replaced`.
1392 
    pub kind: String,
1393 
    pub hunks: Vec<DiffHunk>,
1394 
    /// For renames, the previous path.
1395 
    pub from: Option<String>,
1396 
}
1397
1398 
/// Lines inserted after `base_line` (0 = before the first line) and lines
1399 
/// removed, with their base line numbers.
1400 
#[derive(Clone, Debug, Default)]
1401 
pub struct DiffHunk {
1402 
    pub base_line: u32,
1403 
    pub added: Vec<Vec<u8>>,
1404 
    pub removed: Vec<(u32, Vec<u8>)>,
1405 
}
1406
1407 
impl Repo {
1408 
    /// The content of `path` in a state: `heads`, or the working copy's base.
1409 
    pub fn file_at(&self, path: &str, heads: Option<&[ChangeHash]>) -> Result<Vec<u8>> {
1410 
        let snapshot = match heads {
1411 
            Some(h) => self.load_state(h)?.materialize(),
1412 
            None => {
1413 
                let (_, wc) = self.working_copy()?;
1414 
                self.load_state(&wc.deps)?.materialize()
1415 
            }
1416 
        };
75167672 Andy McCall 1417 
        let Some(f) = snapshot.files.get(path) else {
1418 
            return fail(format!("{path} is not in that state"));
1419 
        };
c6aa5bd9 Andy McCall ↶ 1420 
        match &f.content {
1421 
            FileContent::Dir => fail(format!("{path} is a directory")),
1422 
            FileContent::Text { bytes, .. } => Ok(bytes.clone()),
1423 
            FileContent::Opaque { blobs } => match blobs.first() {
1424 
                Some(b) => self.read_blob(b),
1425 
                None => Ok(vec![]),
1426 
            },
1427 
        }
1428 
    }
1429
1430 
    /// Restore `paths` (files or directories) to the working copy's base:
1431 
    /// edits are undone, additions removed, removals put back. Returns the
1432 
    /// number of paths touched.
1433 
    pub fn discard(&self, paths: &[String]) -> Result<usize> {
6caacfff Andy McCall ↶ 1434 
        let _wc = self.lock_working_copy()?;
8ab265dc Andy McCall ↶ 1435 
        let snap = self.snapshot_status("oxy discard")?;
75167672 Andy McCall 1436 
        let under = |p: &str| {
1437 
            paths
1438 
                .iter()
1439 
                .any(|x| p == x || p.starts_with(&format!("{}/", x.trim_end_matches('/'))))
1440 
        };
c6aa5bd9 Andy McCall ↶ 1441 
        let attrs = Repo::state_attributes(&snap.base_snapshot);
1442 
        let mut n = 0;
1443 
        for s in &snap.statuses {
1444 
            if !under(&s.path) {
1445 
                continue;
1446 
            }
1447 
            match snap.base_snapshot.files.get(&s.path) {
1448 
                None => {
1449 
                    // Added (or the new name of a rename): remove it.
1450 
                    let full = self.root.join(&s.path);
1451 
                    if full.is_dir() {
1452 
                        let _ = fs::remove_dir_all(&full);
1453 
                    } else {
1454 
                        let _ = fs::remove_file(&full);
1455 
                    }
1456 
                    n += 1;
1457 
                }
1458 
                Some(f) => {
1459 
                    match &f.content {
1460 
                        FileContent::Dir => fs::create_dir_all(self.root.join(&s.path))?,
1461 
                        FileContent::Text { bytes, .. } => {
75167672 Andy McCall 1462 
                            let on_disk: std::borrow::Cow<[u8]> = if attrs.lookup(&s.path).eol
1463 
                                == Some(crate::attributes::Eol::Crlf)
1464 
                            {
1465 
                                crate::attributes::to_crlf(bytes).into()
1466 
                            } else {
1467 
                                bytes.as_slice().into()
1468 
                            };
c6aa5bd9 Andy McCall ↶ 1469 
                            self.write_file(&s.path, &on_disk, f.mode.exec)?;
1470 
                        }
1471 
                        FileContent::Opaque { blobs } => {
1472 
                            let bytes = match blobs.first() {
1473 
                                Some(b) => self.read_blob(b)?,
1474 
                                None => vec![],
1475 
                            };
1476 
                            if f.kind == NodeKind::Symlink {
1477 
                                self.write_symlink(&s.path, &bytes)?;
1478 
                            } else if f.kind == NodeKind::File {
1479 
                                self.write_file(&s.path, &bytes, f.mode.exec)?;
1480 
                            }
1481 
                        }
1482 
                    }
1483 
                    n += 1;
1484 
                }
1485 
            }
1486 
        }
1487 
        // A rename's old name is not in the statuses; put it back too.
1488 
        for (p, f) in &snap.base_snapshot.files {
75167672 Andy McCall 1489 
            if under(p)
1490 
                && !self.root.join(p).exists()
1491 
                && f.kind != NodeKind::Dir
1492 
                && f.kind != NodeKind::Subrepo
1493 
                && !snap.statuses.iter().any(|s| s.path == *p)
1494 
            {
c6aa5bd9 Andy McCall ↶ 1495 
                let bytes = match &f.content {
1496 
                    FileContent::Text { bytes, .. } => bytes.clone(),
75167672 Andy McCall 1497 
                    FileContent::Opaque { blobs } => blobs
1498 
                        .first()
1499 
                        .map(|b| self.read_blob(b))
1500 
                        .transpose()?
1501 
                        .unwrap_or_default(),
c6aa5bd9 Andy McCall ↶ 1502 
                    FileContent::Dir => continue,
1503 
                };
75167672 Andy McCall 1504 
                if f.kind == NodeKind::Symlink {
1505 
                    self.write_symlink(p, &bytes)?
1506 
                } else {
1507 
                    self.write_file(p, &bytes, f.mode.exec)?
1508 
                }
c6aa5bd9 Andy McCall ↶ 1509 
                n += 1;
1510 
            }
1511 
        }
1512 
        self.drop_manifest();
1513 
        Ok(n)
1514 
    }
1515
1516 
    /// Undo the working-copy hunks `indices` (as `hunks` numbers them) on
1517 
    /// disk, keeping the others: the files are rewritten from the base plus
1518 
    /// the remaining hunks.
1519 
    pub fn discard_hunks(&self, indices: &[usize]) -> Result<usize> {
6caacfff Andy McCall ↶ 1520 
        let _wc = self.lock_working_copy()?;
c6aa5bd9 Andy McCall ↶ 1521 
        let (snap, hunks) = self.hunks()?;
75167672 Andy McCall 1522 
        let keep: Vec<usize> = hunks
1523 
            .iter()
1524 
            .map(|h| h.index)
1525 
            .filter(|i| !indices.contains(i))
1526 
            .collect();
c6aa5bd9 Andy McCall ↶ 1527 
        let (ops, edits) = select_hunks(&snap.wc, &keep);
1528 
        let dropped_nodes: BTreeSet<NodeId> = hunks
1529 
            .iter()
1530 
            .filter(|h| indices.contains(&h.index))
75167672 Andy McCall 1531 
            .map(|h| {
1532 
                if h.index < snap.wc.tree_ops.len() {
1533 
                    snap.wc.tree_ops[h.index].node()
1534 
                } else {
1535 
                    snap.wc.edits[h.index - snap.wc.tree_ops.len()].node()
1536 
                }
1537 
            })
c6aa5bd9 Andy McCall ↶ 1538 
            .collect();
1539 
        let mut partial = snap.wc.clone();
1540 
        partial.tree_ops = ops;
1541 
        partial.edits = edits;
1542 
        partial.normalize();
1543 
        let mut st = snap.base.clone();
1544 
        st.add_unchecked(partial);
8ab265dc Andy McCall ↶ 1545 
        // Every file of the target is rendered: build the graphs the
1546 
        // partial snapshot left out.
1547 
        self.load_all_graphs(&mut st)?;
c6aa5bd9 Andy McCall ↶ 1548 
        let target = st.materialize();
1549 
        let by_node_target = target.tree.by_node();
1550 
        let by_node_base = snap.base_snapshot.tree.by_node();
1551 
        let attrs = Repo::state_attributes(&target);
1552 
        let mut n = 0;
1553 
        for node in dropped_nodes {
1554 
            match by_node_target.get(&node) {
1555 
                Some(e) => {
1556 
                    let f = &target.files[&e.path];
1557 
                    match &f.content {
1558 
                        FileContent::Dir => {}
1559 
                        FileContent::Text { bytes, .. } => {
75167672 Andy McCall 1560 
                            let on_disk: std::borrow::Cow<[u8]> = if attrs.lookup(&e.path).eol
1561 
                                == Some(crate::attributes::Eol::Crlf)
1562 
                            {
1563 
                                crate::attributes::to_crlf(bytes).into()
1564 
                            } else {
1565 
                                bytes.as_slice().into()
1566 
                            };
c6aa5bd9 Andy McCall ↶ 1567 
                            self.write_file(&e.path, &on_disk, f.mode.exec)?;
1568 
                        }
1569 
                        FileContent::Opaque { blobs } => {
75167672 Andy McCall 1570 
                            let bytes = blobs
1571 
                                .first()
1572 
                                .map(|b| self.read_blob(b))
1573 
                                .transpose()?
1574 
                                .unwrap_or_default();
1575 
                            if f.kind == NodeKind::Symlink {
1576 
                                self.write_symlink(&e.path, &bytes)?
1577 
                            } else {
1578 
                                self.write_file(&e.path, &bytes, f.mode.exec)?
1579 
                            }
c6aa5bd9 Andy McCall ↶ 1580 
                        }
1581 
                    }
1582 
                    // A discarded rename: the new name goes.
75167672 Andy McCall 1583 
                    if let Some(old_path) = snap
1584 
                        .statuses
1585 
                        .iter()
1586 
                        .find(|s| {
1587 
                            s.status == ChangeStatus::Renamed && !target.files.contains_key(&s.path)
1588 
                        })
1589 
                        .map(|s| s.path.clone())
1590 
                    {
c6aa5bd9 Andy McCall ↶ 1591 
                        let _ = fs::remove_file(self.root.join(old_path));
1592 
                    }
1593 
                    n += 1;
1594 
                }
1595 
                None => {
1596 
                    // The node exists in neither target nor base with a live path: an addition to remove.
75167672 Andy McCall 1597 
                    if let Some(added) = snap.statuses.iter().find(|s| {
1598 
                        s.status == ChangeStatus::Added
1599 
                            && !target.files.contains_key(&s.path)
1600 
                            && !by_node_base.contains_key(&node)
1601 
                    }) {
c6aa5bd9 Andy McCall ↶ 1602 
                        let full = self.root.join(&added.path);
75167672 Andy McCall 1603 
                        if full.is_dir() {
1604 
                            let _ = fs::remove_dir_all(&full);
1605 
                        } else {
1606 
                            let _ = fs::remove_file(&full);
1607 
                        }
c6aa5bd9 Andy McCall ↶ 1608 
                        n += 1;
1609 
                    }
1610 
                }
1611 
            }
1612 
        }
1613 
        self.drop_manifest();
1614 
        Ok(n)
1615 
    }
1616
1617 
    /// The diff of the working copy (`None`) or of a change, per file with
1618 
    /// line-numbered hunks (spec: `oxy diff --json`).
1619 
    pub fn diff_files(&self, change: Option<&ChangeHash>) -> Result<Vec<FileDiff>> {
8ab265dc Andy McCall ↶ 1620 
        let (c, base, base_snapshot) = match change {
c6aa5bd9 Andy McCall ↶ 1621 
            Some(h) => {
1622 
                let c = self.store.get_change(h)?;
8ab265dc Andy McCall ↶ 1623 
                let (base, snapshot) = self.change_base(&c)?;
1624 
                (c, base, snapshot)
c6aa5bd9 Andy McCall ↶ 1625 
            }
77445673 Andy McCall ↶ 1626 
            None => self.working_copy_with_base()?,
c6aa5bd9 Andy McCall ↶ 1627 
        };
1628 
        Ok(diff_files_of(&c, &base, &base_snapshot))
e226a6cc Andy McCall ↶ 1629 
    }
1630
1631 
    /// The snapshot a command would take, without taking it, when the
1632 
    /// status cache holds: every file is as the last snapshot left it, so
1633 
    /// that snapshot's working-copy change is the answer. Its base comes
1634 
    /// from the state index, the `.oxyattributes` files from disk. `None`
1635 
    /// when a snapshot must be taken, and when there is nothing to commit
1636 
    /// (the snapshot then says so). A conflicted base never gets here: the
1637 
    /// cache does not hold for one.
1638 
    pub(crate) fn snapshot_from_cache(&self) -> Result<Option<SnapshotResult>> {
1639 
        let Some(statuses) = self.status_by_manifest()? else {
1640 
            return Ok(None);
1641 
        };
1642 
        if statuses.is_empty() {
1643 
            return Ok(None);
1644 
        }
1645 
        let _t = crate::timing::phase("snapshot: from the status cache");
1646 
        let (wc_hash, wc) = self.working_copy()?;
1647 
        let base = self.load_state_indexed(&wc.deps)?;
1648 
        let (base, base_snapshot) = self.render_base_for(&wc, base)?;
1649 
        let attribute_files: Vec<(String, Vec<u8>)> = match self.read_manifest()? {
1650 
            Some((_, _, m)) => m
1651 
                .entries
1652 
                .keys()
1653 
                .filter(|p| is_attributes(p))
1654 
                .filter_map(|p| fs::read(self.root.join(p)).ok().map(|b| (p.clone(), b)))
1655 
                .collect(),
1656 
            None => Vec::new(),
1657 
        };
1658 
        let attributes = crate::attributes::Attributes::from_files(
1659 
            attribute_files
1660 
                .iter()
1661 
                .map(|(p, b)| (p.as_str(), b.as_slice())),
1662 
        );
1663 
        Ok(Some(SnapshotResult {
1664 
            wc_hash,
1665 
            wc,
1666 
            base,
1667 
            base_snapshot,
1668 
            statuses,
1669 
            changed: false,
1670 
            resolves: false,
1671 
            contributors: Vec::new(),
1672 
            attributes,
1673 
            untracked_nested: self.untracked_nested.borrow().clone(),
236f35c0 Andy McCall ↶ 1674 
            manifest: None,
e226a6cc Andy McCall ↶ 1675 
        }))
77445673 Andy McCall ↶ 1676 
    }
1677
1678 
    /// The working-copy change with its base state and the base's
1679 
    /// rendering, for showing what the working copy changes. While the
1680 
    /// status cache holds (every file as the last snapshot left it), the
1681 
    /// stored working-copy change is the answer and no snapshot is taken;
1682 
    /// otherwise a snapshot makes it.
1683 
    pub fn working_copy_with_base(&self) -> Result<(Change, State, Snapshot)> {
1684 
        if self.status_by_manifest()?.is_some() {
1685 
            let (_, wc) = self.working_copy()?;
1686 
            let base = self.load_state_indexed(&wc.deps)?;
1687 
            let (base, snapshot) = self.render_base_for(&wc, base)?;
1688 
            return Ok((wc, base, snapshot));
1689 
        }
1690 
        let snap = self.snapshot_status("oxy diff")?;
1691 
        Ok((snap.wc, snap.base, snap.base_snapshot))
c6aa5bd9 Andy McCall ↶ 1692 
    }
1693
1694 
    /// The log restricted to changes touching `path` (its node, so renames
1695 
    /// are followed), on the current bookmark.
1696 
    pub fn log_for_path(&self, path: &str) -> Result<Vec<LogEntry>> {
e02bf0cd Andy McCall ↶ 1697 
        let heads = self.bookmark_heads()?;
1698 
        if self.is_bare() {
1699 
            return self.log_for_path_in(path, &heads);
1700 
        }
c6aa5bd9 Andy McCall ↶ 1701 
        let entries = self.log()?;
1702 
        let (_, wc) = self.working_copy()?;
1703 
        let base_snapshot = self.load_state(&wc.deps)?.materialize();
1704 
        let state_snapshot = self.load_state(&heads)?.materialize();
75167672 Andy McCall 1705 
        let node = base_snapshot
1706 
            .files
1707 
            .get(path)
1708 
            .or_else(|| state_snapshot.files.get(path))
1709 
            .map(|f| f.node);
1710 
        let Some(node) = node else {
1711 
            return fail(format!("{path} is not tracked"));
1712 
        };
e02bf0cd Andy McCall ↶ 1713 
        Ok(self.entries_touching(entries, node))
1714 
    }
1715
1716 
    /// The log of the state with these heads restricted to changes touching
1717 
    /// `path` (its node, so renames are followed).
1718 
    pub fn log_for_path_in(&self, path: &str, heads: &[ChangeHash]) -> Result<Vec<LogEntry>> {
e8d38e47 Andy McCall ↶ 1719 
        self.log_for_path_of(&self.load_state_history(heads)?, path, None)
bc489b6c Andy McCall ↶ 1720 
    }
1721
1722 
    /// The changes of a loaded state touching `path`, newest first, at most
1723 
    /// `limit` of them. The path is found in the state's tree alone, so no
1724 
    /// file is rendered, and the walk stops once `limit` are found.
1725 
    pub fn log_for_path_of(
1726 
        &self,
1727 
        state: &State,
1728 
        path: &str,
1729 
        limit: Option<usize>,
1730 
    ) -> Result<Vec<LogEntry>> {
1731 
        let Some(node) = state.tree().view().entries.get(path).map(|e| e.node) else {
75167672 Andy McCall 1732 
            return fail(format!("{path} is not in that state"));
1733 
        };
bc489b6c Andy McCall ↶ 1734 
        Ok(self.entries_touching_up_to(self.log_of(state), node, limit))
e02bf0cd Andy McCall ↶ 1735 
    }
1736
1737 
    fn entries_touching(&self, entries: Vec<LogEntry>, node: NodeId) -> Vec<LogEntry> {
bc489b6c Andy McCall ↶ 1738 
        self.entries_touching_up_to(entries, node, None)
1739 
    }
1740
1741 
    fn entries_touching_up_to(
1742 
        &self,
1743 
        entries: Vec<LogEntry>,
1744 
        node: NodeId,
1745 
        limit: Option<usize>,
1746 
    ) -> Vec<LogEntry> {
c6aa5bd9 Andy McCall ↶ 1747 
        let mut out = Vec::new();
1748 
        for e in entries {
bc489b6c Andy McCall ↶ 1749 
            if limit.is_some_and(|n| out.len() >= n) {
1750 
                break;
1751 
            }
75167672 Andy McCall 1752 
            let Ok(c) = self.store.get_change(&e.hash) else {
1753 
                continue;
1754 
            };
1755 
            if c.tree_ops.iter().any(|o| o.node() == node)
1756 
                || c.edits.iter().any(|x| x.node() == node)
1757 
            {
c6aa5bd9 Andy McCall ↶ 1758 
                out.push(e);
1759 
            }
1760 
        }
e02bf0cd Andy McCall ↶ 1761 
        out
c6aa5bd9 Andy McCall ↶ 1762 
    }
1763 
}
1764
1765 
/// Per-file hunks of `change` against `base`.
1766 
pub fn diff_files_of(change: &Change, base: &State, base_snapshot: &Snapshot) -> Vec<FileDiff> {
1767 
    let by_node = base_snapshot.tree.by_node();
1768 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
1769 
    for op in &change.tree_ops {
75167672 Andy McCall 1770 
        if let TreeOp::AddNode {
1771 
            node,
1772 
            parent,
1773 
            name,
1774 
            kind,
1775 
            ..
1776 
        } = op
1777 
        {
c6aa5bd9 Andy McCall ↶ 1778 
            new_names.insert(*node, (*parent, name.clone(), *kind));
1779 
        }
1780 
    }
75167672 Andy McCall 1781 
    let path_of = |n: &NodeId| -> String {
1782 
        path_of_node(n, &by_node, &new_names).unwrap_or_else(|| format!("<node {}>", n.short(8)))
1783 
    };
c6aa5bd9 Andy McCall ↶ 1784 
    let mut files: BTreeMap<NodeId, FileDiff> = BTreeMap::new();
1785 
    let mut order: Vec<NodeId> = Vec::new();
1786 
    let mut entry = |files: &mut BTreeMap<NodeId, FileDiff>, n: NodeId, kind: &str| {
1787 
        if let std::collections::btree_map::Entry::Vacant(v) = files.entry(n) {
1788 
            order.push(n);
75167672 Andy McCall 1789 
            v.insert(FileDiff {
1790 
                path: path_of(&n),
1791 
                kind: kind.to_string(),
1792 
                hunks: vec![],
1793 
                from: None,
1794 
            });
c6aa5bd9 Andy McCall ↶ 1795 
        }
1796 
    };
1797 
    for op in &change.tree_ops {
1798 
        match op {
1799 
            TreeOp::AddNode { node, kind, .. } => {
1800 
                if *kind != NodeKind::Dir {
1801 
                    entry(&mut files, *node, "added");
1802 
                }
1803 
            }
1804 
            TreeOp::RemoveNode { node, .. } => {
1805 
                entry(&mut files, *node, "removed");
1806 
                if let Some(e) = by_node.get(node) {
1807 
                    files.get_mut(node).unwrap().path = e.path.clone();
1808 
                }
1809 
            }
75167672 Andy McCall 1810 
            TreeOp::Move {
1811 
                node, parent, name, ..
1812 
            } => {
c6aa5bd9 Andy McCall ↶ 1813 
                entry(&mut files, *node, "renamed");
1814 
                let f = files.get_mut(node).unwrap();
1815 
                f.from = by_node.get(node).map(|e| e.path.clone());
75167672 Andy McCall 1816 
                let dir = if parent.is_root() {
1817 
                    String::new()
1818 
                } else {
1819 
                    path_of_node(parent, &by_node, &new_names).unwrap_or_default()
1820 
                };
1821 
                f.path = if dir.is_empty() {
1822 
                    name.clone()
1823 
                } else {
1824 
                    format!("{dir}/{name}")
1825 
                };
c6aa5bd9 Andy McCall ↶ 1826 
            }
1827 
            TreeOp::SetMode { node, .. } => entry(&mut files, *node, "mode"),
1828 
        }
1829 
    }
1830 
    // Base line numbers per node, 1-based.
1831 
    let mut line_of: HashMap<NodeId, HashMap<(ChangeHash, u32), u32>> = HashMap::new();
1832 
    let mut lines_for = |n: &NodeId| -> HashMap<(ChangeHash, u32), u32> {
75167672 Andy McCall 1833 
        line_of
1834 
            .entry(*n)
1835 
            .or_insert_with(|| {
1836 
                base.live_lines(n)
1837 
                    .into_iter()
1838 
                    .enumerate()
1839 
                    .map(|(i, (k, _))| (k, i as u32 + 1))
1840 
                    .collect()
1841 
            })
1842 
            .clone()
c6aa5bd9 Andy McCall ↶ 1843 
    };
1844 
    for e in &change.edits {
1845 
        let n = e.node();
1846 
        let is_new = new_names.contains_key(&n);
1847 
        entry(&mut files, n, if is_new { "added" } else { "modified" });
1848 
        let f = files.get_mut(&n).unwrap();
1849 
        match e {
1850 
            Edit::Insert { after, lines, .. } => {
1851 
                let map = lines_for(&n);
1852 
                let base_line = match after {
1853 
                    AtomRef::Start => 0,
1854 
                    AtomRef::End => map.len() as u32,
75167672 Andy McCall 1855 
                    AtomRef::Atom { change, index } if change.is_zero() => {
1856 
                        f.hunks.last().map(|h| h.base_line).unwrap_or(0)
1857 
                    }
1858 
                    AtomRef::Atom { change, index } => {
1859 
                        map.get(&(*change, *index)).copied().unwrap_or(0)
1860 
                    }
c6aa5bd9 Andy McCall ↶ 1861 
                };
75167672 Andy McCall 1862 
                f.hunks.push(DiffHunk {
1863 
                    base_line,
01a47cd7 Andy McCall ↶ 1864 
                    added: lines.to_vec(),
75167672 Andy McCall 1865 
                    removed: vec![],
1866 
                });
c6aa5bd9 Andy McCall ↶ 1867 
            }
1868 
            Edit::Delete { atoms, .. } => {
1869 
                let map = lines_for(&n);
1870 
                let mut removed = Vec::new();
1871 
                for r in atoms {
1872 
                    for i in r.from..r.to {
1873 
                        let line = map.get(&(r.change, i)).copied().unwrap_or(0);
75167672 Andy McCall 1874 
                        let text = base
1875 
                            .file(&n)
1876 
                            .and_then(|g| g.content(&(r.change, i)))
1877 
                            .unwrap_or(b"")
1878 
                            .to_vec();
c6aa5bd9 Andy McCall ↶ 1879 
                        removed.push((line, text));
1880 
                    }
1881 
                }
75167672 Andy McCall 1882 
                let base_line = removed
1883 
                    .first()
1884 
                    .map(|(l, _)| l.saturating_sub(1))
1885 
                    .unwrap_or(0);
1886 
                f.hunks.push(DiffHunk {
1887 
                    base_line,
1888 
                    added: vec![],
1889 
                    removed,
1890 
                });
c6aa5bd9 Andy McCall ↶ 1891 
            }
1892 
            Edit::Replace { .. } => {
1893 
                if !is_new {
1894 
                    f.kind = "replaced".into();
1895 
                }
1896 
            }
1897 
        }
1898 
    }
1899 
    order.into_iter().filter_map(|n| files.remove(&n)).collect()
1900 
}
1901
98286406 Andy McCall ↶ 1902 
/// Builds the tree ops and edits of a working-copy change against a base.
dfb23a3a Andy McCall ↶ 1903 
/// `BLAKE3("oxy node 1" 0 seed kind path)`, truncated: a node identifier
1904 
/// that the same seed and path always give (never the root's).
1905 
fn seeded_node(seed: &[u8; 32], path: &str, kind: NodeKind) -> NodeId {
1906 
    let mut h = blake3::Hasher::new();
1907 
    h.update(b"oxy node 1\0");
1908 
    h.update(seed);
1909 
    h.update(&[match kind {
1910 
        NodeKind::File => 1,
1911 
        NodeKind::Dir => 2,
1912 
        NodeKind::Symlink => 3,
1913 
        NodeKind::Subrepo => 4,
1914 
    }]);
1915 
    h.update(path.as_bytes());
1916 
    let mut id = [0u8; 16];
1917 
    id.copy_from_slice(&h.finalize().as_bytes()[..16]);
1918 
    if id == [0; 16] {
1919 
        id[15] = 1;
1920 
    }
1921 
    NodeId(id)
1922 
}
1923
98286406 Andy McCall ↶ 1924 
struct ChangeBuilder<'a> {
1925 
    base: &'a State,
1926 
    base_snapshot: &'a Snapshot,
1927 
    tree_ops: Vec<TreeOp>,
1928 
    edits: Vec<Edit>,
1929 
    statuses: Vec<PathStatus>,
1930 
    dir_nodes: HashMap<String, NodeId>,
1931 
    reuse: HashMap<String, (NodeId, NodeKind)>,
1932 
    /// Changes that contributed to conflicts this working copy resolves.
1933 
    contributors: BTreeSet<ChangeHash>,
1934 
    resolves: bool,
cdb5beba Andy McCall ↶ 1935 
    attrs: crate::attributes::Attributes,
b7f102fc Andy McCall ↶ 1936 
    blobs: HashMap<String, ObjectRef>,
f7510b63 Andy McCall ↶ 1937 
    /// Opaque files whose replace conflict the content settles.
1938 
    resolve: HashSet<String>,
dfb23a3a Andy McCall ↶ 1939 
    /// When set, new nodes' identifiers derive from it and their paths.
1940 
    node_seed: Option<[u8; 32]>,
a87c7b9d Andy McCall ↶ 1941 
    /// For a Git import, the base's dead directories by where they were
1942 
    /// (see `ensure_dir`), worked out when first needed.
1943 
    dead_dirs: Option<HashMap<(NodeId, String), (NodeId, crate::tree::OpKey)>>,
98286406 Andy McCall ↶ 1944 
}
1945
1946 
impl<'a> ChangeBuilder<'a> {
75167672 Andy McCall 1947 
    fn new(
1948 
        base: &'a State,
1949 
        base_snapshot: &'a Snapshot,
1950 
        reuse: HashMap<String, (NodeId, NodeKind)>,
1951 
    ) -> Self {
98286406 Andy McCall ↶ 1952 
        let mut dir_nodes: HashMap<String, NodeId> = HashMap::new();
1953 
        dir_nodes.insert(String::new(), NodeId::ROOT);
e83354cd Andy McCall ↶ 1954 
        for (p, e) in base_snapshot.tree.entries.iter() {
2306a4d2 Andy McCall ↶ 1955 
            if e.kind == NodeKind::Dir {
1956 
                dir_nodes.insert(p.clone(), e.node);
98286406 Andy McCall ↶ 1957 
            }
1958 
        }
75167672 Andy McCall 1959 
        ChangeBuilder {
1960 
            base,
1961 
            base_snapshot,
1962 
            tree_ops: vec![],
1963 
            edits: vec![],
1964 
            statuses: vec![],
1965 
            dir_nodes,
1966 
            reuse,
1967 
            contributors: BTreeSet::new(),
1968 
            resolves: false,
1969 
            attrs: crate::attributes::Attributes::default(),
1970 
            blobs: HashMap::new(),
f7510b63 Andy McCall ↶ 1971 
            resolve: HashSet::new(),
dfb23a3a Andy McCall ↶ 1972 
            node_seed: None,
a87c7b9d Andy McCall ↶ 1973 
            dead_dirs: None,
75167672 Andy McCall 1974 
        }
98286406 Andy McCall ↶ 1975 
    }
1976
cdb5beba Andy McCall ↶ 1977 
    fn with_attributes(mut self, attrs: crate::attributes::Attributes) -> Self {
1978 
        self.attrs = attrs;
1979 
        self
1980 
    }
1981
1982 
    /// Whether a file is opaque: forced by `binary`, forced text by `text`
1983 
    /// only when it has no NUL bytes, else sniffed.
b7f102fc Andy McCall ↶ 1984 
    /// The blob for an opaque entry: the cached one, else stored now.
75167672 Andy McCall 1985 
    fn blob_of(
1986 
        &mut self,
1987 
        repo: &Repo,
1988 
        path: &str,
1989 
        e: &FsEntry,
1990 
        nocompress: bool,
1991 
    ) -> Result<ObjectRef> {
b7f102fc Andy McCall ↶ 1992 
        let b = match e.blob {
1993 
            Some(b) => b,
1994 
            None => repo.store_blob_for(&e.content, nocompress)?,
1995 
        };
1996 
        self.blobs.insert(path.to_string(), b);
1997 
        Ok(b)
1998 
    }
1999
cdb5beba Andy McCall ↶ 2000 
    fn opaque_for(&self, path: &str, e: &FsEntry) -> bool {
b7f102fc Andy McCall ↶ 2001 
        if e.blob.is_some() {
2002 
            return true;
2003 
        }
4dee277c Andy McCall ↶ 2004 
        if e.kind == NodeKind::Symlink || e.kind == NodeKind::Subrepo {
cdb5beba Andy McCall ↶ 2005 
            return true;
2006 
        }
2007 
        match self.attrs.lookup(path).binary {
2008 
            Some(true) => true,
2009 
            Some(false) => is_binary(&e.content),
2010 
            None => is_binary(&e.content),
2011 
        }
2012 
    }
2013
2014 
    /// Text content as stored: `\r\n` normalized for `eol=` files.
40a40bb7 Andy McCall ↶ 2015 
    fn stored_text(
2016 
        &self,
2017 
        path: &str,
2018 
        content: &std::sync::Arc<Vec<u8>>,
2019 
    ) -> std::sync::Arc<Vec<u8>> {
75167672 Andy McCall 2020 
        if self.attrs.lookup(path).eol.is_some() {
40a40bb7 Andy McCall ↶ 2021 
            std::sync::Arc::new(crate::attributes::normalize_eol(content))
75167672 Andy McCall 2022 
        } else {
40a40bb7 Andy McCall ↶ 2023 
            content.clone()
dfb23a3a Andy McCall ↶ 2024 
        }
2025 
    }
2026
2027 
    /// The identifier of a node added at `path`: reused when asked, derived
2028 
    /// from the seed when there is one, random otherwise.
2029 
    fn new_node(&self, path: &str, kind: NodeKind) -> NodeId {
2030 
        if let Some(n) = self.reused_node(path, kind) {
2031 
            return n;
75167672 Andy McCall 2032 
        }
dfb23a3a Andy McCall ↶ 2033 
        match &self.node_seed {
2034 
            Some(seed) => seeded_node(seed, path, kind),
2035 
            None => NodeId::random(),
2036 
        }
cdb5beba Andy McCall ↶ 2037 
    }
2038
98286406 Andy McCall ↶ 2039 
    fn reused_node(&self, path: &str, kind: NodeKind) -> Option<NodeId> {
75167672 Andy McCall 2040 
        self.reuse
2041 
            .get(path)
2042 
            .filter(|(_, k)| *k == kind)
2043 
            .map(|(n, _)| *n)
98286406 Andy McCall ↶ 2044 
    }
2045
a87c7b9d Andy McCall ↶ 2046 
    /// A dead directory last placed at (`parent`, `name`) in the base.
2047 
    fn revivable_dir(
2048 
        &mut self,
2049 
        parent: NodeId,
2050 
        name: &str,
2051 
    ) -> Option<(NodeId, crate::tree::OpKey)> {
2052 
        let base = self.base;
2053 
        self.dead_dirs
2054 
            .get_or_insert_with(|| base.tree().dead_dirs())
2055 
            .get(&(parent, name.to_string()))
2056 
            .copied()
2057 
    }
2058
98286406 Andy McCall ↶ 2059 
    fn ensure_dir(&mut self, path: &str) -> NodeId {
2060 
        if let Some(n) = self.dir_nodes.get(path) {
2061 
            return *n;
2062 
        }
2063 
        let mut parent = NodeId::ROOT;
2064 
        let mut prefix = String::new();
2065 
        for comp in path.split('/') {
2066 
            if !prefix.is_empty() {
2067 
                prefix.push('/');
2068 
            }
2069 
            prefix.push_str(comp);
a87c7b9d Andy McCall ↶ 2070 
            let known = self.dir_nodes.get(&prefix).copied();
2071 
            let revive = match known {
2072 
                Some(_) => None,
2073 
                None if self.node_seed.is_some() => self.revivable_dir(parent, comp),
2074 
                None => None,
2075 
            };
2076 
            parent = match (known, revive) {
2077 
                (Some(n), _) => n,
2078 
                (None, Some((node, old))) => {
2079 
                    // Git has no identity for a directory: one that comes
2080 
                    // back where it was is the same directory, brought back
2081 
                    // with a move, so lines of history that placed files in
2082 
                    // it before and after agree (oxygen#30).
2083 
                    self.tree_ops.push(TreeOp::Move {
2084 
                        node,
2085 
                        old: OpRef {
2086 
                            change: old.0,
2087 
                            index: old.1,
2088 
                        },
2089 
                        parent,
2090 
                        name: comp.to_string(),
2091 
                    });
2092 
                    self.dir_nodes.insert(prefix.clone(), node);
2093 
                    node
2094 
                }
2095 
                (None, None) => {
dfb23a3a Andy McCall ↶ 2096 
                    let node = self.new_node(&prefix, NodeKind::Dir);
75167672 Andy McCall 2097 
                    self.tree_ops.push(TreeOp::AddNode {
2098 
                        node,
2099 
                        kind: NodeKind::Dir,
2100 
                        mode: Mode::default(),
2101 
                        parent,
2102 
                        name: comp.to_string(),
2103 
                    });
98286406 Andy McCall ↶ 2104 
                    self.dir_nodes.insert(prefix.clone(), node);
2105 
                    node
2106 
                }
2107 
            };
2108 
        }
2109 
        parent
2110 
    }
2111
2112 
    fn add_path(&mut self, repo: &Repo, path: &str, e: &FsEntry) -> Result<()> {
2113 
        let (dir, name) = split_path(path);
2114 
        let parent = self.ensure_dir(dir);
dfb23a3a Andy McCall ↶ 2115 
        let node = self.new_node(path, e.kind);
cdb5beba Andy McCall ↶ 2116 
        let opaque = self.opaque_for(path, e);
2117 
        let nocompress = self.attrs.lookup(path).nocompress;
75167672 Andy McCall 2118 
        self.tree_ops.push(TreeOp::AddNode {
2119 
            node,
2120 
            kind: e.kind,
2121 
            mode: Mode {
2122 
                exec: e.exec,
2123 
                opaque,
2124 
            },
2125 
            parent,
2126 
            name: name.to_string(),
2127 
        });
98286406 Andy McCall ↶ 2128 
        if opaque {
b7f102fc Andy McCall ↶ 2129 
            let blob = self.blob_of(repo, path, e, nocompress)?;
75167672 Andy McCall 2130 
            self.edits.push(Edit::Replace {
2131 
                node,
2132 
                old: None,
2133 
                new: Some(blob),
2134 
            });
98286406 Andy McCall ↶ 2135 
        } else if !e.content.is_empty() {
cdb5beba Andy McCall ↶ 2136 
            let content = self.stored_text(path, &e.content);
98286406 Andy McCall ↶ 2137 
            self.edits.push(Edit::Insert {
2138 
                node,
2139 
                after: crate::objects::AtomRef::Start,
2140 
                before: crate::objects::AtomRef::End,
40a40bb7 Andy McCall ↶ 2141 
                lines: crate::lines::Lines::split_shared(content),
98286406 Andy McCall ↶ 2142 
            });
2143 
        }
75167672 Andy McCall 2144 
        self.statuses.push(PathStatus {
2145 
            path: path.to_string(),
2146 
            status: ChangeStatus::Added,
2147 
        });
98286406 Andy McCall ↶ 2148 
        Ok(())
2149 
    }
2150
6e723b81 Andy McCall ↶ 2151 
    /// Move the node at `old` to `new` and apply `e`'s content to it.
75167672 Andy McCall 2152 
    fn move_path(
2153 
        &mut self,
2154 
        repo: &Repo,
2155 
        old: &str,
2156 
        new: &str,
2157 
        e: &FsEntry,
2158 
        expected: &crate::state::FileState,
2159 
    ) -> Result<()> {
6e723b81 Andy McCall ↶ 2160 
        let entry = &self.base_snapshot.tree.entries[old];
2161 
        let (dir, name) = split_path(new);
2162 
        let parent = self.ensure_dir(dir);
75167672 Andy McCall 2163 
        self.tree_ops.push(TreeOp::Move {
2164 
            node: expected.node,
2165 
            old: op_ref(entry.placement),
2166 
            parent,
2167 
            name: name.to_string(),
2168 
        });
6e723b81 Andy McCall ↶ 2169 
        self.update_path(repo, old, e, expected)?;
2170 
        // update_path reports under the old path; the rename is what the user sees.
2171 
        self.statuses.retain(|s| s.path != old);
75167672 Andy McCall 2172 
        self.statuses.push(PathStatus {
2173 
            path: new.to_string(),
2174 
            status: ChangeStatus::Renamed,
2175 
        });
6e723b81 Andy McCall ↶ 2176 
        Ok(())
a87c7b9d Andy McCall ↶ 2177 
    }
2178
2179 
    /// For a Git import: a node the commit keeps is in one place, the one
2180 
    /// its tree names, so any other live placement it has in the base goes;
2181 
    /// and a node the tree cannot show at all (an orphan) goes too. Such a
2182 
    /// placement is out of sight, under a directory another line of history
2183 
    /// removed, where no diff of the tree would ever reach it (oxygen#30).
2184 
    fn retire_stray_placements(&mut self) {
2185 
        let mut retired: HashSet<(ChangeHash, u32)> = HashSet::new();
2186 
        let mut placed: BTreeSet<NodeId> = BTreeSet::new();
2187 
        for op in &self.tree_ops {
2188 
            match op {
2189 
                TreeOp::Move { node, old, .. } => {
2190 
                    retired.insert((old.change, old.index));
2191 
                    placed.insert(*node);
2192 
                }
2193 
                TreeOp::RemoveNode { placement, .. } => {
2194 
                    retired.insert((placement.change, placement.index));
2195 
                }
2196 
                TreeOp::AddNode { .. } | TreeOp::SetMode { .. } => {}
2197 
            }
2198 
        }
2199 
        // Nodes kept where they were: what the base's tree view shows and
2200 
        // this change neither moves nor removes.
2201 
        let mut kept: HashMap<NodeId, (ChangeHash, u32)> = HashMap::new();
2202 
        for e in self.base_snapshot.tree.entries.values() {
2203 
            if !retired.contains(&e.placement) {
2204 
                kept.insert(e.node, e.placement);
2205 
            }
2206 
        }
2207 
        // Only a node with several live placements can have a stray one,
2208 
        // and the tree view names each as a conflict.
2209 
        let multi: BTreeSet<NodeId> = self
2210 
            .base_snapshot
2211 
            .tree
2212 
            .conflicts
2213 
            .iter()
2214 
            .filter_map(|c| match c {
2215 
                crate::tree::TreeConflict::MultiPlacement { node, .. } => Some(*node),
2216 
                _ => None,
2217 
            })
2218 
            .collect();
2219 
        let mut extra = Vec::new();
2220 
        // A node whose only placements are under a directory another line
2221 
        // removed (an orphan) is nowhere the tree can show, and Git's tree
2222 
        // has no such path: it goes.
2223 
        for c in &self.base_snapshot.tree.conflicts {
2224 
            let crate::tree::TreeConflict::Orphan { node, .. } = c else {
2225 
                continue;
2226 
            };
2227 
            if kept.contains_key(node) || placed.contains(node) {
2228 
                continue;
2229 
            }
2230 
            for (k, _) in self.base.tree().live_placements(node) {
2231 
                if retired.insert(k) {
2232 
                    extra.push(TreeOp::RemoveNode {
2233 
                        node: *node,
2234 
                        placement: op_ref(k),
2235 
                    });
2236 
                }
2237 
            }
2238 
        }
2239 
        for node in placed
2240 
            .iter()
2241 
            .chain(multi.iter().filter(|n| kept.contains_key(*n)))
2242 
        {
2243 
            for (k, _) in self.base.tree().live_placements(node) {
2244 
                if retired.contains(&k) || kept.get(node) == Some(&k) {
2245 
                    continue;
2246 
                }
2247 
                retired.insert(k);
2248 
                extra.push(TreeOp::RemoveNode {
2249 
                    node: *node,
2250 
                    placement: op_ref(k),
2251 
                });
2252 
            }
2253 
        }
2254 
        if !extra.is_empty() {
2255 
            self.resolves = true;
2256 
            self.tree_ops.extend(extra);
2257 
        }
6e723b81 Andy McCall ↶ 2258 
    }
2259
98286406 Andy McCall ↶ 2260 
    fn remove_path(&mut self, path: &str, expected: &crate::state::FileState) {
2261 
        let entry = &self.base_snapshot.tree.entries[path];
75167672 Andy McCall 2262 
        self.tree_ops.push(TreeOp::RemoveNode {
2263 
            node: expected.node,
2264 
            placement: op_ref(entry.placement),
2265 
        });
98286406 Andy McCall ↶ 2266 
        if path.contains(".oxy-conflict-") || expected.is_conflicted() {
2267 
            self.resolves = true;
2268 
        }
75167672 Andy McCall 2269 
        self.statuses.push(PathStatus {
2270 
            path: path.to_string(),
2271 
            status: ChangeStatus::Removed,
2272 
        });
98286406 Andy McCall ↶ 2273 
    }
2274
75167672 Andy McCall 2275 
    fn update_path(
2276 
        &mut self,
2277 
        repo: &Repo,
2278 
        path: &str,
2279 
        e: &FsEntry,
2280 
        expected: &crate::state::FileState,
2281 
    ) -> Result<()> {
98286406 Andy McCall ↶ 2282 
        if expected.kind != e.kind {
2283 
            self.remove_path(path, expected);
2284 
            self.statuses.pop();
2285 
            self.add_path(repo, path, e)?;
2286 
            self.statuses.pop();
75167672 Andy McCall 2287 
            self.statuses.push(PathStatus {
2288 
                path: path.to_string(),
2289 
                status: ChangeStatus::Modified,
2290 
            });
98286406 Andy McCall ↶ 2291 
            return Ok(());
2292 
        }
2293 
        let entry = &self.base_snapshot.tree.entries[path];
2294 
        let node = expected.node;
2295 
        let mode_atom = entry.mode_atom.map(op_ref);
cc7a0056 Andy McCall ↶ 2296 
        if e.unchanged {
2297 
            // Known equal to the base's unconflicted rendering (spec
2298 
            // §12.3); only the executable bit can differ.
2299 
            if e.exec != expected.mode.exec {
2300 
                if let Some(old) = mode_atom {
2301 
                    self.tree_ops.push(TreeOp::SetMode {
2302 
                        node,
2303 
                        old,
2304 
                        mode: Mode {
2305 
                            exec: e.exec,
2306 
                            opaque: false,
2307 
                        },
2308 
                    });
2309 
                    self.statuses.push(PathStatus {
2310 
                        path: path.to_string(),
2311 
                        status: ChangeStatus::ModeChanged,
2312 
                    });
2313 
                }
2314 
            }
2315 
            return Ok(());
2316 
        }
98286406 Andy McCall ↶ 2317 
        let mut status: Option<ChangeStatus> = None;
cdb5beba Andy McCall ↶ 2318 
        let wants_opaque = self.opaque_for(path, e);
2319 
        let nocompress = self.attrs.lookup(path).nocompress;
2320 
        let disk_text = self.stored_text(path, &e.content);
98286406 Andy McCall ↶ 2321 
        match &expected.content {
2322 
            FileContent::Dir => {}
75167672 Andy McCall 2323 
            FileContent::Text {
2324 
                bytes,
2325 
                materialized,
2326 
            } => {
98286406 Andy McCall ↶ 2327 
                if materialized.is_conflicted() {
40a40bb7 Andy McCall ↶ 2328 
                    if bytes == &*disk_text {
98286406 Andy McCall ↶ 2329 
                        return Ok(());
2330 
                    }
2331 
                    if has_conflict_markers(&e.content) {
75167672 Andy McCall 2332 
                        self.statuses.push(PathStatus {
2333 
                            path: path.to_string(),
2334 
                            status: ChangeStatus::ConflictEdited,
2335 
                        });
98286406 Andy McCall ↶ 2336 
                        return Ok(());
2337 
                    }
2338 
                    // Resolution (spec §9.6): diff against the unconflicted lines
2339 
                    // with each region as a gap, then retire every region atom.
2340 
                    // Whatever the author kept from a region is re-inserted as
2341 
                    // new atoms, which leaves the graph totally ordered.
2342 
                    let regions = self.base.conflict_regions(&node);
cdb5beba Andy McCall ↶ 2343 
                    let mut edits = text_edits(node, &regions.plain, &disk_text);
98286406 Andy McCall ↶ 2344 
                    if !regions.region_atoms.is_empty() {
75167672 Andy McCall 2345 
                        edits.insert(
2346 
                            0,
2347 
                            Edit::Delete {
2348 
                                node,
2349 
                                atoms: ranges(&regions.region_atoms),
2350 
                            },
2351 
                        );
98286406 Andy McCall ↶ 2352 
                    }
2353 
                    self.edits.extend(edits);
2354 
                    self.contributors.extend(regions.contributors);
2355 
                    self.resolves = true;
2356 
                    if e.exec != expected.mode.exec {
2357 
                        if let Some(old) = mode_atom {
75167672 Andy McCall 2358 
                            self.tree_ops.push(TreeOp::SetMode {
2359 
                                node,
2360 
                                old,
2361 
                                mode: Mode {
2362 
                                    exec: e.exec,
2363 
                                    opaque: false,
2364 
                                },
2365 
                            });
98286406 Andy McCall ↶ 2366 
                        }
2367 
                    }
75167672 Andy McCall 2368 
                    self.statuses.push(PathStatus {
2369 
                        path: path.to_string(),
2370 
                        status: ChangeStatus::Resolved,
2371 
                    });
98286406 Andy McCall ↶ 2372 
                    return Ok(());
2373 
                }
cdb5beba Andy McCall ↶ 2374 
                if wants_opaque {
2375 
                    // Text → opaque (spec §7.5), by content or by attribute.
2376 
                    if let Some(old) = mode_atom {
75167672 Andy McCall 2377 
                        self.tree_ops.push(TreeOp::SetMode {
2378 
                            node,
2379 
                            old,
2380 
                            mode: Mode {
2381 
                                exec: e.exec,
2382 
                                opaque: true,
2383 
                            },
2384 
                        });
cdb5beba Andy McCall ↶ 2385 
                    }
2386 
                    let live: Vec<_> = repo_live_keys(self.base, &node);
2387 
                    if !live.is_empty() {
75167672 Andy McCall 2388 
                        self.edits.push(Edit::Delete {
2389 
                            node,
2390 
                            atoms: ranges(&live),
2391 
                        });
cdb5beba Andy McCall ↶ 2392 
                    }
b7f102fc Andy McCall ↶ 2393 
                    let blob = self.blob_of(repo, path, e, nocompress)?;
75167672 Andy McCall 2394 
                    self.edits.push(Edit::Replace {
2395 
                        node,
2396 
                        old: None,
2397 
                        new: Some(blob),
2398 
                    });
cdb5beba Andy McCall ↶ 2399 
                    status = Some(ChangeStatus::Modified);
40a40bb7 Andy McCall ↶ 2400 
                } else if bytes != &*disk_text {
2306a4d2 Andy McCall ↶ 2401 
                    let mut laps = crate::timing::Laps::start();
2402 
                    // The base's rendering holds its live atoms in order,
2403 
                    // one per line (it is not conflicted here).
2404 
                    let keys = materialized.live_atoms();
2405 
                    let lines = crate::diff::split_lines(bytes);
2406 
                    let edits = if keys.len() == lines.len() {
2407 
                        crate::diff::text_edits_of(node, &keys, &lines, &disk_text)
2408 
                    } else {
2409 
                        text_edits(node, &self.base.live_lines(&node), &disk_text)
2410 
                    };
2411 
                    laps.lap("build: text diff");
cdb5beba Andy McCall ↶ 2412 
                    if !edits.is_empty() {
2413 
                        self.edits.extend(edits);
98286406 Andy McCall ↶ 2414 
                        status = Some(ChangeStatus::Modified);
2415 
                    }
2416 
                }
cdb5beba Andy McCall ↶ 2417 
                let switched_to_opaque = wants_opaque;
98286406 Andy McCall ↶ 2418 
                if !switched_to_opaque && e.exec != expected.mode.exec {
2419 
                    if let Some(old) = mode_atom {
75167672 Andy McCall 2420 
                        self.tree_ops.push(TreeOp::SetMode {
2421 
                            node,
2422 
                            old,
2423 
                            mode: Mode {
2424 
                                exec: e.exec,
2425 
                                opaque: false,
2426 
                            },
2427 
                        });
98286406 Andy McCall ↶ 2428 
                        status.get_or_insert(ChangeStatus::ModeChanged);
2429 
                    }
2430 
                }
2431 
            }
2432 
            FileContent::Opaque { blobs } => {
b7f102fc Andy McCall ↶ 2433 
                let new_ref = e.blob.unwrap_or_else(|| Repo::blob_ref(&e.content));
2434 
                if e.kind == NodeKind::File && wants_opaque {
2435 
                    // Unchanged or replaced, the file stays opaque: remember its blob.
2436 
                    self.blobs.insert(path.to_string(), new_ref);
f7510b63 Andy McCall ↶ 2437 
                }
2438 
                if blobs.len() > 1 && self.resolve.contains(path) && e.kind == expected.kind {
2439 
                    // Settle the replace conflict (spec §9.3): replace every
2440 
                    // live candidate with the chosen blob, depending on each
2441 
                    // change that made one live, so all are superseded.
2442 
                    let chosen = self.blob_of(repo, path, e, nocompress)?;
2443 
                    for b in blobs {
2444 
                        self.edits.push(Edit::Replace {
2445 
                            node,
2446 
                            old: Some(*b),
2447 
                            new: Some(chosen),
2448 
                        });
2449 
                    }
2450 
                    self.contributors
2451 
                        .extend(self.base.blob_contributors(&node, blobs));
2452 
                    self.resolves = true;
2453 
                    self.statuses.push(PathStatus {
2454 
                        path: path.to_string(),
2455 
                        status: ChangeStatus::Resolved,
2456 
                    });
2457 
                    return Ok(());
b7f102fc Andy McCall ↶ 2458 
                }
98286406 Andy McCall ↶ 2459 
                if blobs.len() > 1 {
2460 
                    if !blobs.contains(&new_ref) {
75167672 Andy McCall 2461 
                        self.statuses.push(PathStatus {
2462 
                            path: path.to_string(),
2463 
                            status: ChangeStatus::ConflictEdited,
2464 
                        });
98286406 Andy McCall ↶ 2465 
                    }
2466 
                    return Ok(());
2467 
                }
2468 
                let current = blobs.first().copied();
2469 
                if current != Some(new_ref) {
cdb5beba Andy McCall ↶ 2470 
                    if e.kind == NodeKind::File && !wants_opaque {
98286406 Andy McCall ↶ 2471 
                        // Opaque → text (spec §7.5).
2472 
                        if let Some(old) = mode_atom {
75167672 Andy McCall 2473 
                            self.tree_ops.push(TreeOp::SetMode {
2474 
                                node,
2475 
                                old,
2476 
                                mode: Mode {
2477 
                                    exec: e.exec,
2478 
                                    opaque: false,
2479 
                                },
2480 
                            });
98286406 Andy McCall ↶ 2481 
                        }
75167672 Andy McCall 2482 
                        self.edits.push(Edit::Replace {
2483 
                            node,
2484 
                            old: current,
2485 
                            new: None,
2486 
                        });
98286406 Andy McCall ↶ 2487 
                        if !e.content.is_empty() {
2488 
                            self.edits.push(Edit::Insert {
2489 
                                node,
2490 
                                after: crate::objects::AtomRef::Start,
2491 
                                before: crate::objects::AtomRef::End,
01a47cd7 Andy McCall ↶ 2492 
                                lines: crate::lines::Lines::split(&e.content),
98286406 Andy McCall ↶ 2493 
                            });
2494 
                        }
2495 
                    } else {
b7f102fc Andy McCall ↶ 2496 
                        let blob = self.blob_of(repo, path, e, nocompress)?;
75167672 Andy McCall 2497 
                        self.edits.push(Edit::Replace {
2498 
                            node,
2499 
                            old: current,
2500 
                            new: Some(blob),
2501 
                        });
98286406 Andy McCall ↶ 2502 
                    }
2503 
                    status = Some(ChangeStatus::Modified);
2504 
                } else if e.kind == NodeKind::File && e.exec != expected.mode.exec {
2505 
                    if let Some(old) = mode_atom {
75167672 Andy McCall 2506 
                        self.tree_ops.push(TreeOp::SetMode {
2507 
                            node,
2508 
                            old,
2509 
                            mode: Mode {
2510 
                                exec: e.exec,
2511 
                                opaque: true,
2512 
                            },
2513 
                        });
98286406 Andy McCall ↶ 2514 
                        status = Some(ChangeStatus::ModeChanged);
2515 
                    }
2516 
                }
2517 
            }
2518 
        }
2519 
        if let Some(s) = status {
75167672 Andy McCall 2520 
            self.statuses.push(PathStatus {
2521 
                path: path.to_string(),
2522 
                status: s,
2523 
            });
98286406 Andy McCall ↶ 2524 
        }
2525 
        Ok(())
2526 
    }
2527 
}
2528
2529 
fn repo_live_keys(base: &State, node: &NodeId) -> Vec<crate::graph::AtomKey> {
2530 
    base.live_lines(node).into_iter().map(|(k, _)| k).collect()
2531 
}
2532
2533 
/// The changes a change's scope references (spec §8.1): introducers of every
2534 
/// atom, placement, and mode it touches, and of every node it operates on.
2535 
pub fn referenced_changes(change: &Change, base: &State) -> BTreeSet<ChangeHash> {
2536 
    let mut out = BTreeSet::new();
2537 
    let opref = |r: &OpRef, out: &mut BTreeSet<ChangeHash>| {
2538 
        if !r.change.is_zero() {
2539 
            out.insert(r.change);
2540 
        }
2541 
    };
2542 
    let mut nodes: Vec<NodeId> = Vec::new();
2543 
    for op in &change.tree_ops {
2544 
        match op {
2545 
            TreeOp::AddNode { parent, .. } => nodes.push(*parent),
2546 
            TreeOp::RemoveNode { node, placement } => {
2547 
                nodes.push(*node);
2548 
                opref(placement, &mut out);
2549 
            }
75167672 Andy McCall 2550 
            TreeOp::Move {
2551 
                node, old, parent, ..
2552 
            } => {
98286406 Andy McCall ↶ 2553 
                nodes.push(*node);
2554 
                nodes.push(*parent);
2555 
                opref(old, &mut out);
2556 
            }
2557 
            TreeOp::SetMode { node, old, .. } => {
2558 
                nodes.push(*node);
2559 
                opref(old, &mut out);
2560 
            }
2561 
        }
2562 
    }
2563 
    for e in &change.edits {
2564 
        nodes.push(e.node());
2565 
        match e {
2566 
            Edit::Insert { after, before, .. } => {
2567 
                for r in [after, before] {
2568 
                    if let crate::objects::AtomRef::Atom { change, .. } = r {
2569 
                        if !change.is_zero() {
2570 
                            out.insert(*change);
2571 
                        }
2572 
                    }
2573 
                }
2574 
            }
2575 
            Edit::Delete { atoms, .. } => {
62794ecc Andy McCall ↶ 2576 
                // Deleting an atom also depends on whoever inserted the live
2577 
                // atoms right after it: the author saw those insertions, and
2578 
                // without the dependency spec §9.1 would call the deletion
2579 
                // concurrent with them (a delete/insert conflict).
2580 
                let graph = base.file(&e.node());
98286406 Andy McCall ↶ 2581 
                for r in atoms {
2582 
                    if !r.change.is_zero() {
2583 
                        out.insert(r.change);
2584 
                    }
62794ecc Andy McCall ↶ 2585 
                    if let Some(g) = graph {
2586 
                        for i in r.from..r.to {
2587 
                            out.extend(
2588 
                                g.live_successor_changes(&(r.change, i))
2589 
                                    .into_iter()
2590 
                                    .filter(|c| !c.is_zero()),
2591 
                            );
2592 
                        }
2593 
                    }
98286406 Andy McCall ↶ 2594 
                }
2595 
            }
62794ecc Andy McCall ↶ 2596 
            // Replacing a blob depends on the change that made it current,
2597 
            // as deleting a line depends on the line's introducer.
2598 
            Edit::Replace {
2599 
                node,
2600 
                old: Some(old),
2601 
                ..
2602 
            } => {
2603 
                out.extend(
2604 
                    base.live_replaces(node)
2605 
                        .into_iter()
2606 
                        .filter(|(c, b)| b == old && !c.is_zero())
2607 
                        .map(|(c, _)| c),
2608 
                );
2609 
            }
98286406 Andy McCall ↶ 2610 
            Edit::Replace { .. } => {}
2611 
        }
2612 
    }
2613 
    for n in nodes {
2614 
        if let Some(h) = base.tree().introducer(&n) {
2615 
            out.insert(h);
2616 
        }
2617 
    }
2618 
    out
2619 
}
2620
2621 
/// Paths of the nodes a change adds on top of `base_snapshot`.
75167672 Andy McCall 2622 
pub fn added_node_paths(
2623 
    change: &Change,
2624 
    base_snapshot: &Snapshot,
2625 
) -> HashMap<String, (NodeId, NodeKind)> {
98286406 Andy McCall ↶ 2626 
    let by_node = base_snapshot.tree.by_node();
2627 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
2628 
    for op in &change.tree_ops {
75167672 Andy McCall 2629 
        if let TreeOp::AddNode {
2630 
            node,
2631 
            parent,
2632 
            name,
2633 
            kind,
2634 
            ..
2635 
        } = op
2636 
        {
98286406 Andy McCall ↶ 2637 
            new_names.insert(*node, (*parent, name.clone(), *kind));
2638 
        }
2639 
    }
2640 
    let mut out = HashMap::new();
2641 
    for (node, (_, _, kind)) in &new_names {
2642 
        if let Some(path) = path_of_node(node, &by_node, &new_names) {
2643 
            out.insert(path, (*node, *kind));
2644 
        }
2645 
    }
2646 
    out
2647 
}
2648
2649 
fn path_of_node(
2650 
    n: &NodeId,
2651 
    by_node: &HashMap<NodeId, &crate::tree::TreeEntry>,
2652 
    new_names: &HashMap<NodeId, (NodeId, String, NodeKind)>,
2653 
) -> Option<String> {
2654 
    if let Some(e) = by_node.get(n) {
2655 
        return Some(e.path.clone());
2656 
    }
2657 
    let mut parts = Vec::new();
2658 
    let mut cur = *n;
2659 
    loop {
2660 
        if cur.is_root() {
2661 
            break;
2662 
        }
2663 
        if let Some(e) = by_node.get(&cur) {
2664 
            if !e.path.is_empty() {
2665 
                parts.push(e.path.clone());
2666 
            }
2667 
            break;
2668 
        }
2669 
        let (parent, name, _) = new_names.get(&cur)?;
2670 
        parts.push(name.clone());
2671 
        cur = *parent;
2672 
    }
2673 
    parts.reverse();
2674 
    Some(parts.join("/"))
2675 
}
2676
2677 
/// Render a change's operations with line content, using `base` for deleted lines.
2678 
pub fn render_change(change: &Change, base: &State, base_snapshot: &Snapshot) -> String {
2679 
    let by_node = base_snapshot.tree.by_node();
2680 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
2681 
    for op in &change.tree_ops {
75167672 Andy McCall 2682 
        if let TreeOp::AddNode {
2683 
            node,
2684 
            parent,
2685 
            name,
2686 
            kind,
2687 
            ..
2688 
        } = op
2689 
        {
98286406 Andy McCall ↶ 2690 
            new_names.insert(*node, (*parent, name.clone(), *kind));
2691 
        }
2692 
    }
75167672 Andy McCall 2693 
    let path_of = |n: &NodeId| -> String {
2694 
        path_of_node(n, &by_node, &new_names).unwrap_or_else(|| format!("<node {}>", n.short(8)))
2695 
    };
98286406 Andy McCall ↶ 2696
2697 
    let mut out = String::new();
2698 
    for op in &change.tree_ops {
2699 
        match op {
75167672 Andy McCall 2700 
            TreeOp::AddNode {
2701 
                node, kind, mode, ..
2702 
            } => {
2703 
                out.push_str(&format!(
2704 
                    "A {} ({:?}{}{})\n",
2705 
                    path_of(node),
2706 
                    kind,
2707 
                    if mode.exec { ", exec" } else { "" },
2708 
                    if mode.opaque { ", opaque" } else { "" }
2709 
                ));
98286406 Andy McCall ↶ 2710 
            }
2711 
            TreeOp::RemoveNode { node, .. } => out.push_str(&format!("D {}\n", path_of(node))),
75167672 Andy McCall 2712 
            TreeOp::Move { node, name, .. } => {
2713 
                out.push_str(&format!("R {} -> {}\n", path_of(node), name))
2714 
            }
98286406 Andy McCall ↶ 2715 
            TreeOp::SetMode { node, mode, .. } => {
75167672 Andy McCall 2716 
                out.push_str(&format!(
2717 
                    "M {} mode: exec={} opaque={}\n",
2718 
                    path_of(node),
2719 
                    mode.exec,
2720 
                    mode.opaque
2721 
                ));
98286406 Andy McCall ↶ 2722 
            }
2723 
        }
2724 
    }
2725 
    let mut last_node: Option<NodeId> = None;
2726 
    for e in &change.edits {
2727 
        let n = e.node();
2728 
        if last_node != Some(n) {
2729 
            out.push_str(&format!("--- {}\n", path_of(&n)));
2730 
            last_node = Some(n);
2731 
        }
2732 
        match e {
2733 
            Edit::Insert { lines, .. } => {
01a47cd7 Andy McCall ↶ 2734 
                for l in lines.iter() {
98286406 Andy McCall ↶ 2735 
                    out.push('+');
2736 
                    out.push_str(&String::from_utf8_lossy(l));
2737 
                    if !l.ends_with(b"\n") {
2738 
                        out.push('\n');
2739 
                    }
2740 
                }
2741 
            }
2742 
            Edit::Delete { atoms, .. } => {
2743 
                for r in atoms {
2744 
                    for i in r.from..r.to {
75167672 Andy McCall 2745 
                        let content = base
2746 
                            .file(&n)
2747 
                            .and_then(|g| g.content(&(r.change, i)))
2748 
                            .unwrap_or(b"<unknown>\n");
98286406 Andy McCall ↶ 2749 
                        out.push('-');
2750 
                        out.push_str(&String::from_utf8_lossy(content));
2751 
                        if !content.ends_with(b"\n") {
2752 
                            out.push('\n');
2753 
                        }
2754 
                    }
2755 
                }
2756 
            }
2757 
            Edit::Replace { old, new, .. } => {
2758 
                out.push_str(&format!(
2759 
                    "  opaque content {} -> {}\n",
2760 
                    old.map(|o| o.short(12)).unwrap_or_else(|| "none".into()),
2761 
                    new.map(|o| o.short(12)).unwrap_or_else(|| "none".into())
2762 
                ));
2763 
            }
2764 
        }
2765 
    }
2766 
    out
2767 
}
6e723b81 Andy McCall ↶ 2768
2769 
/// Pair removed files with added files of the same kind whose contents are
2770 
/// at least `RENAME_THRESHOLD` percent alike, best matches first. Beyond
2771 
/// `RENAME_LIMIT` candidate pairs only identical contents are paired.
2772 
const RENAME_THRESHOLD: u32 = 50;
2773 
const RENAME_LIMIT: usize = 10_000;
2306a4d2 Andy McCall ↶ 2774
23ff1288 Andy McCall ↶ 2775 
/// How a change is built beyond its entries: for a Git import, the seed
2776 
/// its new nodes' identifiers derive from; and, when it is built from part
2777 
/// of the tree, whether the new tree has anything under a directory.
2778 
#[derive(Default)]
2779 
struct Scope<'a> {
2780 
    seed: Option<[u8; 32]>,
2306a4d2 Andy McCall ↶ 2781 
    dir_needed: Option<&'a dyn Fn(&str) -> bool>,
2782 
}
6e723b81 Andy McCall ↶ 2783
75167672 Andy McCall 2784 
fn detect_renames(
2785 
    base_snapshot: &Snapshot,
2786 
    fs_entries: &BTreeMap<String, FsEntry>,
2787 
    in_scope: impl Fn(&str) -> bool,
2788 
) -> Vec<(String, String)> {
6e723b81 Andy McCall ↶ 2789 
    let removed: Vec<(&String, &crate::state::FileState)> = base_snapshot
2790 
        .files
2791 
        .iter()
75167672 Andy McCall 2792 
        .filter(|(p, f)| {
2793 
            f.kind != NodeKind::Dir
2794 
                && !fs_entries.contains_key(*p)
2795 
                && in_scope(p)
2796 
                && !f.is_conflicted()
2797 
        })
2798 
        .collect();
2799 
    let added: Vec<(&String, &FsEntry)> = fs_entries
2800 
        .iter()
2801 
        .filter(|(p, e)| e.kind != NodeKind::Dir && !base_snapshot.files.contains_key(*p))
6e723b81 Andy McCall ↶ 2802 
        .collect();
2803 
    if removed.is_empty() || added.is_empty() {
2804 
        return vec![];
2805 
    }
2806 
    let exact_only = removed.len() * added.len() > RENAME_LIMIT;
2807 
    let old_content = |f: &crate::state::FileState| -> Option<Vec<u8>> {
2808 
        match &f.content {
2809 
            FileContent::Text { bytes, .. } => Some(bytes.clone()),
2810 
            _ => None,
2811 
        }
2812 
    };
2813 
    let mut scored: Vec<(u32, usize, usize)> = Vec::new();
2814 
    for (ri, (_, f)) in removed.iter().enumerate() {
2815 
        let old = old_content(f);
2816 
        let old_blob = match &f.content {
2817 
            FileContent::Opaque { blobs } if blobs.len() == 1 => Some(blobs[0]),
2818 
            _ => None,
2819 
        };
2820 
        for (ai, (_, e)) in added.iter().enumerate() {
2821 
            if e.kind != f.kind {
2822 
                continue;
2823 
            }
2824 
            let score = match (&old, old_blob) {
2825 
                (Some(o), _) => {
2826 
                    if exact_only {
40a40bb7 Andy McCall ↶ 2827 
                        if *o == *e.content {
75167672 Andy McCall 2828 
                            100
2829 
                        } else {
2830 
                            0
2831 
                        }
6e723b81 Andy McCall ↶ 2832 
                    } else {
2833 
                        similarity(o, &e.content)
2834 
                    }
2835 
                }
b7f102fc Andy McCall ↶ 2836 
                (None, Some(b)) if e.blob.unwrap_or_else(|| Repo::blob_ref(&e.content)) == b => 100,
6e723b81 Andy McCall ↶ 2837 
                _ => 0,
2838 
            };
2839 
            if score >= RENAME_THRESHOLD {
2840 
                scored.push((score, ri, ai));
2841 
            }
2842 
        }
2843 
    }
75167672 Andy McCall 2844 
    scored.sort_by(|a, b| {
2845 
        b.0.cmp(&a.0)
2846 
            .then(removed[a.1].0.cmp(removed[b.1].0))
2847 
            .then(added[a.2].0.cmp(added[b.2].0))
2848 
    });
6e723b81 Andy McCall ↶ 2849 
    let mut used_r = vec![false; removed.len()];
2850 
    let mut used_a = vec![false; added.len()];
2851 
    let mut out = Vec::new();
2852 
    for (_, ri, ai) in scored {
2853 
        if used_r[ri] || used_a[ai] {
2854 
            continue;
2855 
        }
2856 
        used_r[ri] = true;
2857 
        used_a[ai] = true;
2858 
        out.push((removed[ri].0.clone(), added[ai].0.clone()));
2859 
    }
2860 
    out
2861 
}
2862
2863 
/// Percentage of the larger file's bytes found in lines the two share
2864 
/// (multiset intersection of lines), like Git's similarity index.
2865 
pub fn similarity(a: &[u8], b: &[u8]) -> u32 {
2866 
    if a == b {
2867 
        return 100;
2868 
    }
2869 
    if a.is_empty() || b.is_empty() {
2870 
        return 0;
2871 
    }
2872 
    let mut counts: HashMap<&[u8], usize> = HashMap::new();
2873 
    for l in split_lines(a) {
2874 
        *counts.entry(l).or_default() += 1;
2875 
    }
2876 
    let mut shared = 0usize;
2877 
    for l in split_lines(b) {
2878 
        if let Some(n) = counts.get_mut(l) {
2879 
            if *n > 0 {
2880 
                *n -= 1;
2881 
                shared += l.len();
2882 
            }
2883 
        }
2884 
    }
2885 
    (shared * 100 / a.len().max(b.len())) as u32
d616faf5 Andy McCall ↶ 2886 
}
2887
2888 
/// Two maps side by side in key order: each key of either, with its value
2889 
/// in each. Linear, with no lookups, for maps of a whole tree.
2890 
fn join<'a, A, B>(
2891 
    a: &'a BTreeMap<String, A>,
2892 
    b: &'a BTreeMap<String, B>,
2893 
) -> impl Iterator<Item = (&'a String, Option<&'a A>, Option<&'a B>)> {
2894 
    let (mut a, mut b) = (a.iter().peekable(), b.iter().peekable());
2895 
    std::iter::from_fn(move || match (a.peek(), b.peek()) {
2896 
        (None, None) => None,
2897 
        (Some(_), None) => a.next().map(|(k, v)| (k, Some(v), None)),
2898 
        (None, Some(_)) => b.next().map(|(k, v)| (k, None, Some(v))),
2899 
        (Some((ka, _)), Some((kb, _))) => match ka.cmp(kb) {
2900 
            std::cmp::Ordering::Less => a.next().map(|(k, v)| (k, Some(v), None)),
2901 
            std::cmp::Ordering::Greater => b.next().map(|(k, v)| (k, None, Some(v))),
2902 
            std::cmp::Ordering::Equal => {
2903 
                let (k, va) = a.next()?;
2904 
                let (_, vb) = b.next()?;
2905 
                Some((k, Some(va), Some(vb)))
2906 
            }
2907 
        },
2908 
    })
2909 
}
2910
2911 
#[cfg(test)]
2912 
mod join_tests {
2913 
    use super::*;
2914
2915 
    #[test]
2916 
    fn join_pairs_every_key_in_order() {
2917 
        let a: BTreeMap<String, u8> = [("a", 1), ("b/c", 2), ("d", 3)]
2918 
            .into_iter()
2919 
            .map(|(k, v)| (k.to_string(), v))
2920 
            .collect();
2921 
        let b: BTreeMap<String, char> = [("b", 'x'), ("b/c", 'y'), ("e", 'z')]
2922 
            .into_iter()
2923 
            .map(|(k, v)| (k.to_string(), v))
2924 
            .collect();
2925 
        let got: Vec<(String, Option<u8>, Option<char>)> = join(&a, &b)
2926 
            .map(|(k, x, y)| (k.clone(), x.copied(), y.copied()))
2927 
            .collect();
2928 
        let want = vec![
2929 
            ("a".to_string(), Some(1), None),
2930 
            ("b".to_string(), None, Some('x')),
2931 
            ("b/c".to_string(), Some(2), Some('y')),
2932 
            ("d".to_string(), Some(3), None),
2933 
            ("e".to_string(), None, Some('z')),
2934 
        ];
2935 
        assert_eq!(got, want);
2936 
    }
6e723b81 Andy McCall ↶ 2937 
}