OXYGENHUB
# 19303b4a06fb oxygen / crates / oxy-core / src / repo / snapshot.rs
115.0 KiB · Rust · last changed oxy cat builds only the file it prints 28 days ago source blame history raw
98286406 Andy McCall ↶ 1 
//! Working-copy snapshots, commits, and the change builder (spec §12.3).
2
3 
use super::*;
957fee30 Andy McCall ↶ 4 
use crate::objects::AtomRef;
98286406 Andy McCall ↶ 5
6 
#[derive(Clone, Copy, Debug, PartialEq, Eq)]
7 
pub enum ChangeStatus {
8 
    Added,
9 
    Modified,
10 
    Removed,
6e723b81 Andy McCall ↶ 11 
    /// Moved from another path (detected by similarity, tracked as a placement change).
12 
    Renamed,
98286406 Andy McCall ↶ 13 
    ModeChanged,
14 
    /// The file is conflicted in the base state and still contains conflict markers.
15 
    ConflictEdited,
16 
    /// The file was conflicted in the base state and the working copy resolves it.
17 
    Resolved,
18 
}
19
20 
/// Lines that mark a conflict region in a materialized file (spec §9.5).
21 
pub fn has_conflict_markers(content: &[u8]) -> bool {
22 
    split_lines(content).iter().any(|l| {
75167672 Andy McCall 23 
        l.starts_with(b"<<<<<<< oxy conflict")
24 
            || l.starts_with(b"======= oxy conflict")
25 
            || l.starts_with(b">>>>>>> oxy conflict")
98286406 Andy McCall ↶ 26 
    })
27 
}
28
29 
#[derive(Clone, Debug, PartialEq, Eq)]
30 
pub struct PathStatus {
31 
    pub path: String,
32 
    pub status: ChangeStatus,
33 
}
34
614541c3 Andy McCall ↶ 35 
/// Operations of a change built from the working directory against a base.
36 
pub(crate) struct BuiltChange {
37 
    pub tree_ops: Vec<TreeOp>,
38 
    pub edits: Vec<Edit>,
39 
    pub contributors: Vec<ChangeHash>,
40 
    pub resolves: bool,
41 
    pub statuses: Vec<PathStatus>,
b7f102fc Andy McCall ↶ 42 
    /// The stored blob of every opaque path, for the manifest's stat cache.
43 
    pub blobs: HashMap<String, ObjectRef>,
614541c3 Andy McCall ↶ 44 
}
45
98286406 Andy McCall ↶ 46 
/// Result of snapshotting the working copy.
47 
pub struct SnapshotResult {
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");
360aa74a Andy McCall ↶ 297 
        b.diff_in_parallel(fs_entries);
298 
        laps.lap("build: diffs");
c3069537 Andy McCall ↶ 299 
        for (path, fs_entry) in fs_entries {
98286406 Andy McCall ↶ 300 
            match base_snapshot.files.get(path) {
6e723b81 Andy McCall ↶ 301 
                None => match renamed_from.get(path.as_str()) {
75167672 Andy McCall 302 
                    Some(old) => {
303 
                        b.move_path(self, old, path, fs_entry, &base_snapshot.files[*old])?
304 
                    }
6e723b81 Andy McCall ↶ 305 
                    None => b.add_path(self, path, fs_entry)?,
306 
                },
98286406 Andy McCall ↶ 307 
                Some(expected) => b.update_path(self, path, fs_entry, expected)?,
308 
            }
309 
        }
2306a4d2 Andy McCall ↶ 310 
        laps.lap("build: paths");
958fb27d Andy McCall ↶ 311 
        // A directory stays while anything under it exists or is out of
312 
        // scope: the directories above each such path, gathered once.
313 
        let mut needed_dirs: HashSet<&str> = HashSet::new();
314 
        fn parents_of<'a>(p: &'a str, set: &mut HashSet<&'a str>) {
315 
            let mut end = p.len();
316 
            while let Some(i) = p[..end].rfind('/') {
317 
                if !set.insert(&p[..i]) {
318 
                    break;
319 
                }
320 
                end = i;
321 
            }
322 
        }
323 
        for p in fs_entries.keys() {
324 
            parents_of(p, &mut needed_dirs);
325 
        }
326 
        for p in base_snapshot.files.keys() {
327 
            if !self.in_scope(p) {
328 
                parents_of(p, &mut needed_dirs);
329 
            }
330 
        }
98286406 Andy McCall ↶ 331 
        for (path, expected) in &base_snapshot.files {
75167672 Andy McCall 332 
            if fs_entries.contains_key(path)
333 
                || !self.in_scope(path)
334 
                || renamed_to.contains_key(path.as_str())
335 
            {
98286406 Andy McCall ↶ 336 
                continue;
337 
            }
2306a4d2 Andy McCall ↶ 338 
            let still_needed = expected.kind == NodeKind::Dir
339 
                && match dir_needed {
340 
                    Some(f) => f(path),
341 
                    None => needed_dirs.contains(path.as_str()),
342 
                };
98286406 Andy McCall ↶ 343 
            if !still_needed {
344 
                b.remove_path(path, expected);
345 
            }
346 
        }
2306a4d2 Andy McCall ↶ 347 
        laps.lap("build: removals");
a87c7b9d Andy McCall ↶ 348 
        if seed.is_some() {
349 
            b.retire_stray_placements();
350 
        }
2306a4d2 Andy McCall ↶ 351 
        laps.lap("build: stray placements");
614541c3 Andy McCall ↶ 352 
        let mut statuses = b.statuses;
353 
        statuses.sort_by(|a, b| a.path.cmp(&b.path));
75167672 Andy McCall 354 
        Ok(BuiltChange {
355 
            tree_ops: b.tree_ops,
356 
            edits: b.edits,
357 
            contributors: b.contributors.into_iter().collect(),
358 
            resolves: b.resolves,
359 
            statuses,
360 
            blobs: b.blobs,
361 
        })
614541c3 Andy McCall ↶ 362 
    }
98286406 Andy McCall ↶ 363
c07e0e7d Andy McCall ↶ 364 
    /// Store a change and, when `core.sign = change` and we are its author,
365 
    /// sign it. A rewritten change by another author is stored unsigned: only
366 
    /// its author (or a delegate) can vouch for it.
614541c3 Andy McCall ↶ 367 
    pub(crate) fn store_signed(&self, change: &Change) -> Result<ChangeHash> {
c07e0e7d Andy McCall ↶ 368 
        let ours = change.author == self.author();
369 
        if self.config.sign_changes && ours && self.secret_key()?.is_none() {
614541c3 Andy McCall ↶ 370 
            return fail("core.sign is \"change\" but no secret key is available for author.key");
371 
        }
372 
        let hash = self.store.put_change(change)?;
c07e0e7d Andy McCall ↶ 373 
        if self.config.sign_changes && ours {
614541c3 Andy McCall ↶ 374 
            self.sign_change(&hash)?;
375 
        }
376 
        Ok(hash)
377 
    }
378
379 
    /// Snapshot the working copy into the working-copy change (spec §12.3).
cc7a0056 Andy McCall ↶ 380 
    /// The result carries the full base state and its whole rendering.
614541c3 Andy McCall ↶ 381 
    pub fn snapshot(&self, command: &str) -> Result<SnapshotResult> {
f9466021 Andy McCall ↶ 382 
        self.snapshot_as(command, false, false)
cc7a0056 Andy McCall ↶ 383 
    }
384
385 
    /// Snapshot the working copy as [`Repo::snapshot`] does, for commands
8ab265dc Andy McCall ↶ 386 
    /// that need its statuses, conflicts and base history, and the graphs
387 
    /// of the files it changes: status, log, show, commit, diff, hunks and
388 
    /// discard. Text files known unchanged since the last manifest are
cc7a0056 Andy McCall ↶ 389 
    /// neither read nor rendered: the result's `base` is a history-only
390 
    /// state without their graphs, and its `base_snapshot` lists them in
8ab265dc Andy McCall ↶ 391 
    /// `unrendered` with empty bytes. They are never conflicted, and no
392 
    /// change the working copy makes touches them.
cc7a0056 Andy McCall ↶ 393 
    pub fn snapshot_status(&self, command: &str) -> Result<SnapshotResult> {
f9466021 Andy McCall ↶ 394 
        self.snapshot_as(command, true, false)
cc7a0056 Andy McCall ↶ 395 
    }
396
cd4eb55e Andy McCall ↶ 397 
    /// `snapshot_as`; `for_commit` writes the manifest even where a status
398 
    /// would not bother (an edited file is read next time whatever it
399 
    /// says), since after the commit that file is unchanged, and records no
400 
    /// operation: the commit's own, recorded straight after, replaces the
401 
    /// working copy, so the snapshot's would cost a durable write for
402 
    /// nothing. Should the commit fail, the files are still on disk and the
403 
    /// next snapshot finds them.
f9466021 Andy McCall ↶ 404 
    fn snapshot_as(
405 
        &self,
406 
        command: &str,
407 
        partial: bool,
cd4eb55e Andy McCall ↶ 408 
        for_commit: bool,
f9466021 Andy McCall ↶ 409 
    ) -> Result<SnapshotResult> {
4e88e774 Andy McCall ↶ 410 
        let _lock = self.lock_snapshot()?;
8f10a07a Andy McCall ↶ 411 
        let batch = self.store.begin_batch()?;
cd4eb55e Andy McCall ↶ 412 
        let result = self.snapshot_locked(command, partial, for_commit);
8f10a07a Andy McCall ↶ 413 
        batch.finish()?;
414 
        result
415 
    }
416
f9466021 Andy McCall ↶ 417 
    fn snapshot_locked(
418 
        &self,
419 
        command: &str,
420 
        partial: bool,
cd4eb55e Andy McCall ↶ 421 
        for_commit: bool,
f9466021 Andy McCall ↶ 422 
    ) -> Result<SnapshotResult> {
0bdc3f91 Andy McCall ↶ 423 
        let _all = crate::timing::phase("snapshot");
614541c3 Andy McCall ↶ 424 
        let (op_ref, op) = self.head_op()?;
425 
        let (old_hash, old_wc) = self.working_copy()?;
cc7a0056 Andy McCall ↶ 426 
        // The base's history first: its graph keys say which files the
427 
        // manifest already knows equal to it.
428 
        let mut base = {
429 
            let _t = crate::timing::phase("snapshot: load base history");
f9575918 Andy McCall ↶ 430 
            self.load_state_indexed(&old_wc.deps)?
0bdc3f91 Andy McCall ↶ 431 
        };
d0d52107 Andy McCall ↶ 432 
        let view = {
433 
            let _t = crate::timing::phase("snapshot: tree view");
434 
            base.tree().view()
435 
        };
8b276e01 Andy McCall ↶ 436 
        let manifest = {
437 
            let _t = crate::timing::phase("snapshot: read manifest");
438 
            self.read_manifest()?.map(|(_, _, m)| m)
439 
        };
cc7a0056 Andy McCall ↶ 440 
        let tracked: BTreeMap<String, (NodeKind, Option<ObjectRef>)> = {
441 
            let _t = crate::timing::phase("snapshot: tracked paths");
36e51aa8 Andy McCall ↶ 442 
            // In the view's path order, so the map is built without searching.
443 
            view.entries
444 
                .iter()
445 
                .filter(|(_, e)| e.kind != NodeKind::Dir)
446 
                .map(|(p, e)| {
447 
                    let blob = if e.kind != NodeKind::File || e.mode.opaque {
448 
                        base.live_blobs(&e.node).first().copied()
449 
                    } else {
450 
                        None
451 
                    };
452 
                    (p.clone(), (e.kind, blob))
453 
                })
454 
                .collect()
cc7a0056 Andy McCall ↶ 455 
        };
992467d3 Andy McCall ↶ 456 
        // A text file's graph key, worked out when the walk asks for it.
1966f5db Andy McCall ↶ 457 
        let key_of = |p: &str| -> Option<[u8; 32]> {
992467d3 Andy McCall ↶ 458 
            let e = view.entries.get(p)?;
459 
            if e.kind != NodeKind::File || e.mode.opaque {
460 
                return None;
461 
            }
1966f5db Andy McCall ↶ 462 
            base.graph_key(&e.node)
992467d3 Andy McCall ↶ 463 
        };
cc7a0056 Andy McCall ↶ 464 
        let mut fs_entries = {
465 
            let _t = crate::timing::phase("snapshot: read working copy");
992467d3 Andy McCall ↶ 466 
            self.walk_working_copy_with(manifest.as_deref(), &tracked, Some(&key_of))?
cc7a0056 Andy McCall ↶ 467 
        };
992467d3 Andy McCall ↶ 468 
        // The keys of the files read, before building graphs drops them: a
469 
        // file found equal to its base rendering keeps its key.
1966f5db Andy McCall ↶ 470 
        let keys: HashMap<String, [u8; 32]> = fs_entries
992467d3 Andy McCall ↶ 471 
            .iter()
472 
            .filter(|(_, e)| !e.unchanged && e.kind == NodeKind::File)
473 
            .filter_map(|(p, _)| Some((p.clone(), key_of(p)?)))
474 
            .collect();
0bdc3f91 Andy McCall ↶ 475 
        let base_snapshot = {
cc7a0056 Andy McCall ↶ 476 
            let _t = crate::timing::phase("snapshot: load and render base");
477 
            if partial {
d616faf5 Andy McCall ↶ 478 
                let sets = crate::timing::phase("snapshot:   of which choosing graphs");
479 
                // Every file but those on disk known unchanged.
480 
                let need: Vec<(&str, NodeId)> = join(&view.entries, &fs_entries)
481 
                    .filter_map(|(p, v, f)| {
482 
                        let v = v?;
483 
                        (v.kind == NodeKind::File && !f.is_some_and(|f| f.unchanged))
484 
                            .then_some((p.as_str(), v.node))
485 
                    })
cc7a0056 Andy McCall ↶ 486 
                    .collect();
d616faf5 Andy McCall ↶ 487 
                let render: std::collections::HashSet<&str> =
488 
                    need.iter().map(|(p, _)| *p).collect();
489 
                let need: Vec<NodeId> = need.iter().map(|(_, n)| *n).collect();
490 
                drop(sets);
d0990c2b Andy McCall ↶ 491 
                {
492 
                    let _t = crate::timing::phase("snapshot:   of which building graphs");
493 
                    self.load_graphs_of(&mut base, &need)?;
494 
                }
495 
                let _t = crate::timing::phase("snapshot:   of which rendering");
d616faf5 Andy McCall ↶ 496 
                base.materialize_only(&render)
cc7a0056 Andy McCall ↶ 497 
            } else {
498 
                self.load_all_graphs(&mut base)?;
499 
                base.materialize()
500 
            }
0bdc3f91 Andy McCall ↶ 501 
        };
614541c3 Andy McCall ↶ 502
503 
        // Reuse the NodeIds the previous working-copy version assigned to new
504 
        // paths, so an unchanged working copy hashes identically (spec §12.3).
505 
        let reuse = added_node_paths(&old_wc, &base_snapshot);
c863c7a7 Andy McCall ↶ 506 
        self.platform_normalize(&mut fs_entries, &base_snapshot);
0bdc3f91 Andy McCall ↶ 507 
        let built = {
508 
            let _t = crate::timing::phase("snapshot: diff against base");
23ff1288 Andy McCall ↶ 509 
            if partial && self.config.sparse_paths.is_empty() {
510 
                #[cfg(test)]
511 
                let full =
512 
                    self.build_from_entries(&base, &base_snapshot, reuse.clone(), &fs_entries)?;
513 
                let built = self.build_changed_only(&base, &base_snapshot, reuse, &fs_entries)?;
514 
                #[cfg(test)]
515 
                {
516 
                    // New nodes take random identifiers: compare them by
517 
                    // order of appearance.
518 
                    let norm = |b: &BuiltChange| {
519 
                        let text = format!("{:?} {:?} {:?}", b.tree_ops, b.edits, b.statuses);
520 
                        let mut seen: Vec<String> = Vec::new();
521 
                        let mut out = String::new();
522 
                        let mut rest = text.as_str();
523 
                        while let Some(i) = rest.find("NodeId(") {
524 
                            out.push_str(&rest[..i]);
525 
                            let end = i + rest[i..].find(')').unwrap() + 1;
526 
                            let id = rest[i..end].to_string();
527 
                            let n = seen.iter().position(|x| *x == id).unwrap_or_else(|| {
528 
                                seen.push(id);
529 
                                seen.len() - 1
530 
                            });
531 
                            out.push_str(&format!("N{n}"));
532 
                            rest = &rest[end..];
533 
                        }
534 
                        out.push_str(rest);
535 
                        out
536 
                    };
537 
                    assert_eq!(
538 
                        norm(&built),
539 
                        norm(&full),
540 
                        "a changed-only build is the full one"
541 
                    );
542 
                }
543 
                built
544 
            } else {
545 
                self.build_from_entries(&base, &base_snapshot, reuse, &fs_entries)?
546 
            }
0bdc3f91 Andy McCall ↶ 547 
        };
cc7a0056 Andy McCall ↶ 548 
        // Text files found equal to an unconflicted base rendering keep its
549 
        // graph key, so the next snapshot need not read or render them.
550 
        {
551 
            let changed: HashSet<&str> = built.statuses.iter().map(|s| s.path.as_str()).collect();
552 
            for (p, e) in fs_entries.iter_mut() {
553 
                if e.unchanged || e.kind != NodeKind::File || changed.contains(p.as_str()) {
554 
                    continue;
555 
                }
556 
                e.graph_key = match base_snapshot.files.get(p).map(|f| &f.content) {
557 
                    Some(FileContent::Text {
558 
                        bytes,
559 
                        materialized,
40a40bb7 Andy McCall ↶ 560 
                    }) if !materialized.is_conflicted() && *bytes == *e.content => {
1966f5db Andy McCall ↶ 561 
                        keys.get(p).copied()
cc7a0056 Andy McCall ↶ 562 
                    }
563 
                    _ => None,
564 
                };
565 
            }
566 
        }
614541c3 Andy McCall ↶ 567
98286406 Andy McCall ↶ 568 
        let mut wc = Change {
6f8aeb99 Andy McCall ↶ 569 
            version: crate::objects::CHANGE_VERSION,
98286406 Andy McCall ↶ 570 
            change_id: old_wc.change_id,
571 
            deps: vec![],
572 
            author: self.author(),
573 
            hlc: old_wc.hlc,
574 
            message: old_wc.message.clone(),
614541c3 Andy McCall ↶ 575 
            tree_ops: built.tree_ops,
576 
            edits: built.edits,
98286406 Andy McCall ↶ 577 
            kind: ChangeKind::WorkingCopy,
578 
        };
579 
        let base_heads = base.maximal(&old_wc.deps);
580 
        let mut deps: BTreeSet<ChangeHash> = base_heads.into_iter().collect();
581 
        deps.extend(referenced_changes(&wc, &base));
614541c3 Andy McCall ↶ 582 
        deps.extend(built.contributors.iter().copied());
98286406 Andy McCall ↶ 583 
        wc.deps = deps.into_iter().collect();
584 
        wc.normalize();
614541c3 Andy McCall ↶ 585 
        let resolves = built.resolves;
586 
        let contributors = built.contributors;
587 
        let statuses = built.statuses;
98286406 Andy McCall ↶ 588 
        // Empty snapshot: hash the candidate before storing; only record an op on change.
8b276e01 Andy McCall ↶ 589 
        let record = crate::timing::phase("snapshot: hash and record");
17f94b13 Andy McCall ↶ 590 
        // Encoded once, its texts stored a batch at a time as they are
591 
        // made (a stored blob is not stored again); the change itself only
592 
        // when it changed.
593 
        let encoded = wc.encode_streaming(|batch| self.store.put_texts(&batch))?;
192fad1e Andy McCall ↶ 594 
        let new_hash = crate::hash::hash_object(crate::hash::ObjectType::Change, &encoded);
98286406 Andy McCall ↶ 595 
        let changed = new_hash != old_hash;
cd4eb55e Andy McCall ↶ 596 
        // For a commit the change is neither stored nor recorded: the commit
597 
        // stores its own, built from this one's operations.
598 
        if changed && !for_commit {
17f94b13 Andy McCall ↶ 599 
            self.store.put(crate::hash::ObjectType::Change, &encoded)?;
98286406 Andy McCall ↶ 600 
            let mut view = op.view.clone();
601 
            view.workspaces.insert(self.workspace, new_hash);
602 
            self.record_op(&[op_ref], &format!("{command}{SNAPSHOT_SUFFIX}"), view)?;
603 
        }
8b276e01 Andy McCall ↶ 604 
        drop(record);
135212de Andy McCall ↶ 605 
        // The prompt compares against this without recording anything.
0bdc3f91 Andy McCall ↶ 606 
        let _t = crate::timing::phase("snapshot: write manifest");
992467d3 Andy McCall ↶ 607 
        let conflicted = base_snapshot.is_conflicted();
608 
        let uncommitted = !wc.tree_ops.is_empty() || !wc.edits.is_empty();
609 
        // Rewriting the manifest pays only when it would teach the next
610 
        // snapshot something: other flags, a file found equal to the base
611 
        // (it need not be read again), a new blob, a path added or gone.
612 
        // An edited file is read next time whatever it says.
613 
        let worth = match self.read_manifest()? {
cd4eb55e Andy McCall ↶ 614 
            _ if for_commit => true,
992467d3 Andy McCall ↶ 615 
            None => true,
616 
            Some((c, u, m)) => {
617 
                c != conflicted
618 
                    || u != uncommitted
619 
                    || m.entries.len() != fs_entries.len()
620 
                    || fs_entries.iter().any(|(p, e)| {
621 
                        !e.unchanged && (e.graph_key.is_some() || !m.entries.contains_key(p))
622 
                    })
623 
                    || built
624 
                        .blobs
625 
                        .iter()
626 
                        .any(|(p, b)| m.entries.get(p).and_then(|e| e.blob) != Some(*b))
36041796 Andy McCall ↶ 627 
                    // With uncommitted work, the status cache below holds
628 
                    // only while the manifest describes the edited files as
629 
                    // they are now.
630 
                    || (uncommitted && statuses.iter().any(|st| {
631 
                        let Some(e) = fs_entries.get(&st.path) else {
632 
                            return m.entries.contains_key(&st.path);
633 
                        };
634 
                        let Some(me) = m.entries.get(&st.path) else {
635 
                            return true;
636 
                        };
637 
                        let hash = e
638 
                            .content_hash
639 
                            .unwrap_or_else(|| *blake3::hash(&e.content).as_bytes());
640 
                        let stat = e.stat.or_else(|| {
641 
                            fs::symlink_metadata(self.root.join(&st.path))
642 
                                .ok()
643 
                                .map(|md| (md.len(), super::prompt::mtime_ns(&md)))
644 
                        });
e226a6cc Andy McCall ↶ 645 
                        // The executable bit too: a mode change moves
646 
                        // neither size nor time.
647 
                        me.hash != hash || me.exec != e.exec || stat != Some((me.size, me.mtime_ns))
36041796 Andy McCall ↶ 648 
                    }))
992467d3 Andy McCall ↶ 649 
            }
650 
        };
f9466021 Andy McCall ↶ 651 
        let attributes = crate::attributes::Attributes::from_files(
652 
            fs_entries
653 
                .iter()
654 
                .filter(|(p, _)| is_attributes(p))
655 
                .map(|(p, e)| (p.as_str(), e.content.as_slice())),
656 
        );
236f35c0 Andy McCall ↶ 657 
        let mut manifest = None;
658 
        if for_commit {
659 
            manifest = Some(PendingManifest {
660 
                entries: fs_entries,
661 
                blobs: built.blobs.clone(),
662 
                dirs: self.walked_dirs.take(),
663 
                written: super::prompt::now_ns(),
664 
            });
665 
        } else {
666 
            if worth {
667 
                self.write_manifest_entries(conflicted, uncommitted, &fs_entries, &built.blobs)?;
668 
            }
669 
            self.write_status_cache((uncommitted && !conflicted).then_some(&new_hash), &statuses);
670 
        }
75167672 Andy McCall 671 
        Ok(SnapshotResult {
672 
            wc_hash: new_hash,
673 
            wc,
674 
            base,
675 
            base_snapshot,
676 
            statuses,
677 
            changed,
678 
            resolves,
679 
            contributors,
f9466021 Andy McCall ↶ 680 
            attributes,
d4dbab97 Andy McCall ↶ 681 
            untracked_nested: self.untracked_nested.borrow().clone(),
236f35c0 Andy McCall ↶ 682 
            manifest,
75167672 Andy McCall 683 
        })
98286406 Andy McCall ↶ 684 
    }
685
50a046e4 Andy McCall ↶ 686 
    /// `oxy status --json` (spec §9.5): the changed paths and every conflict
687 
    /// of the working copy's base with its kind, path, and contributing
688 
    /// ChangeIds.
689 
    pub fn status_json(&self, snap: &SnapshotResult) -> Result<String> {
690 
        let heads = self.bookmark_heads()?;
691 
        let base_snapshot = &snap.base_snapshot;
692 
        let by_node = base_snapshot.tree.by_node();
693 
        let path_of = |n: &NodeId| by_node.get(n).map(|e| e.path.clone());
75167672 Andy McCall 694 
        let id_of = |h: &ChangeHash| {
695 
            snap.base
696 
                .get(h)
697 
                .map(|m| m.change_id.to_hex())
698 
                .unwrap_or_else(|| h.to_hex())
699 
        };
700 
        let paths = json_array(snap.statuses.iter().map(|s| {
701 
            format!(
702 
                "{{\"path\":{},\"status\":{}}}",
703 
                json_str(&s.path),
704 
                json_str(&format!("{:?}", s.status))
705 
            )
706 
        }));
50a046e4 Andy McCall ↶ 707 
        let mut conflicts: Vec<String> = Vec::new();
708 
        for (p, f) in &base_snapshot.files {
709 
            if !f.is_conflicted() {
710 
                continue;
711 
            }
712 
            let (kind, changes) = match &f.content {
75167672 Andy McCall 713 
                FileContent::Text { .. } => {
714 
                    ("text", snap.base.conflict_regions(&f.node).contributors)
715 
                }
716 
                FileContent::Opaque { blobs } => {
717 
                    ("replace", snap.base.blob_contributors(&f.node, blobs))
718 
                }
50a046e4 Andy McCall ↶ 719 
                FileContent::Dir => continue,
720 
            };
75167672 Andy McCall 721 
            let resolved = match snap
722 
                .statuses
723 
                .iter()
724 
                .find(|s| s.path == *p)
725 
                .map(|s| s.status)
726 
            {
50a046e4 Andy McCall ↶ 727 
                Some(ChangeStatus::Resolved) => "true",
728 
                Some(ChangeStatus::ConflictEdited) => "\"markers\"",
729 
                _ => "false",
730 
            };
731 
            conflicts.push(format!(
732 
                "{{\"kind\":{},\"path\":{},\"node\":{},\"changes\":{},\"resolved\":{resolved}}}",
733 
                json_str(kind),
734 
                json_str(p),
735 
                json_str(&f.node.to_hex()),
736 
                json_array(changes.iter().map(|h| json_str(&id_of(h))))
737 
            ));
738 
        }
739 
        for c in &base_snapshot.tree.conflicts {
740 
            let (kind, nodes): (&str, Vec<NodeId>) = match c {
741 
                crate::tree::TreeConflict::Name { nodes, .. } => ("name", nodes.clone()),
75167672 Andy McCall 742 
                crate::tree::TreeConflict::MultiPlacement { node, .. } => {
743 
                    ("multi-placement", vec![*node])
744 
                }
50a046e4 Andy McCall ↶ 745 
                crate::tree::TreeConflict::Orphan { node, .. } => ("orphan", vec![*node]),
746 
                crate::tree::TreeConflict::Cycle { node } => ("cycle", vec![*node]),
747 
                crate::tree::TreeConflict::Mode { node, .. } => ("mode", vec![*node]),
748 
            };
749 
            let detail = match c {
750 
                crate::tree::TreeConflict::Name { name, parent, .. } => {
75167672 Andy McCall 751 
                    let dir = if *parent == NodeId::ROOT {
752 
                        String::new()
753 
                    } else {
754 
                        path_of(parent).unwrap_or_default()
755 
                    };
756 
                    format!(
757 
                        ",\"path\":{}",
758 
                        json_str(&if dir.is_empty() {
759 
                            name.clone()
760 
                        } else {
761 
                            format!("{dir}/{name}")
762 
                        })
763 
                    )
50a046e4 Andy McCall ↶ 764 
                }
75167672 Andy McCall 765 
                _ => nodes
766 
                    .first()
767 
                    .and_then(path_of)
768 
                    .map(|p| format!(",\"path\":{}", json_str(&p)))
769 
                    .unwrap_or_default(),
50a046e4 Andy McCall ↶ 770 
            };
771 
            conflicts.push(format!(
772 
                "{{\"kind\":{},\"nodes\":{}{detail}}}",
773 
                json_str(kind),
774 
                json_array(nodes.iter().map(|n| json_str(&n.to_hex())))
775 
            ));
776 
        }
777 
        Ok(format!(
d4dbab97 Andy McCall ↶ 778 
            "{{\"bookmark\":{},\"working_copy\":{},\"heads\":{},\"clean\":{},\"conflicted\":{},\"paths\":{paths},\"conflicts\":{},\"untracked_nested\":{}}}",
50a046e4 Andy McCall ↶ 779 
            json_str(&self.config.bookmark),
780 
            json_str(&snap.wc.change_id.to_hex()),
781 
            heads.len(),
782 
            snap.statuses.is_empty(),
783 
            base_snapshot.is_conflicted(),
d4dbab97 Andy McCall ↶ 784 
            json_array(conflicts),
785 
            json_array(snap.untracked_nested.iter().map(|p| json_str(p)))
50a046e4 Andy McCall ↶ 786 
        ))
787 
    }
788
98286406 Andy McCall ↶ 789 
    /// Turn the working-copy change into a normal change on the bookmark (spec §12.3).
790 
    pub fn commit(&self, message: &str, extra_deps: &[ChangeHash]) -> Result<ChangeHash> {
957fee30 Andy McCall ↶ 791 
        self.commit_selection(message, extra_deps, &Selection::All)
792 
    }
793
794 
    /// The hunks of the working copy, for `oxy commit --interactive`: each
795 
    /// tree operation and each edit, in the order `select_hunks` numbers them.
796 
    pub fn hunks(&self) -> Result<(SnapshotResult, Vec<Hunk>)> {
8ab265dc Andy McCall ↶ 797 
        let snap = self.snapshot_status("oxy commit")?;
957fee30 Andy McCall ↶ 798 
        let hunks = render_hunks(&snap.wc, &snap.base, &snap.base_snapshot);
799 
        Ok((snap, hunks))
800 
    }
801
802 
    /// Commit part of the working copy (spec §12.3). What is not selected
803 
    /// stays on disk as uncommitted work: the next snapshot rebuilds it
804 
    /// against the new base, so the working copy always describes the
805 
    /// directory.
f9466021 Andy McCall ↶ 806 
    /// The graph key in `st` of each text file at `paths`, the paths a
807 
    /// commit changed: only their graphs changed, so the keys the manifest
808 
    /// holds for the others still stand (and any that does not is simply
809 
    /// not trusted by the next snapshot, which compares keys). Files with
810 
    /// an `eol` attribute are left out: their bytes on disk are not the
811 
    /// rendering.
812 
    fn committed_graph_keys(
813 
        &self,
814 
        st: &State,
815 
        paths: &[String],
816 
        attrs: &crate::attributes::Attributes,
817 
    ) -> HashMap<String, [u8; 32]> {
86e1824f Andy McCall ↶ 818 
        let _t = crate::timing::phase("commit: graph keys");
f9466021 Andy McCall ↶ 819 
        let view = st.tree().view();
820 
        paths
86e1824f Andy McCall ↶ 821 
            .iter()
f9466021 Andy McCall ↶ 822 
            .filter_map(|p| {
823 
                let e = view.entries.get(p)?;
824 
                (e.kind == NodeKind::File && !e.mode.opaque && attrs.lookup(p).eol.is_none())
825 
                    .then_some(())?;
826 
                Some((p.clone(), st.graph_key(&e.node)?))
86e1824f Andy McCall ↶ 827 
            })
8457a6cc Andy McCall ↶ 828 
            .collect()
86e1824f Andy McCall ↶ 829 
    }
830
75167672 Andy McCall 831 
    pub fn commit_selection(
832 
        &self,
833 
        message: &str,
834 
        extra_deps: &[ChangeHash],
835 
        selection: &Selection,
836 
    ) -> Result<ChangeHash> {
b30a880f Andy McCall ↶ 837 
        // The commit's objects are written together, made durable at once
838 
        // before the operation that names them (spec §13.6).
839 
        let batch = self.store.begin_batch()?;
840 
        let out = self.commit_selection_batched(message, extra_deps, selection);
841 
        batch.finish()?;
842 
        out
843 
    }
844
236f35c0 Andy McCall ↶ 845 
    /// Write the manifest a commit's snapshot left to the commit, with the
846 
    /// graph keys `keys` sets.
847 
    fn write_pending_manifest(
848 
        &self,
849 
        mut m: PendingManifest,
850 
        conflicted: bool,
851 
        keys: HashMap<String, [u8; 32]>,
852 
    ) -> Result<()> {
853 
        for (path, k) in keys {
854 
            if let Some(e) = m.entries.get_mut(&path) {
855 
                e.graph_key = Some(k);
856 
            }
857 
        }
858 
        self.write_manifest_entries_as_of(
859 
            conflicted,
860 
            false,
861 
            &m.entries,
862 
            &m.blobs,
863 
            m.written,
864 
            m.dirs.unwrap_or_default(),
865 
        )
866 
    }
867
b30a880f Andy McCall ↶ 868 
    fn commit_selection_batched(
869 
        &self,
870 
        message: &str,
871 
        extra_deps: &[ChangeHash],
872 
        selection: &Selection,
873 
    ) -> Result<ChangeHash> {
6caacfff Andy McCall ↶ 874 
        let _wc = self.lock_working_copy()?;
e226a6cc Andy McCall ↶ 875 
        let mut snap = match self.snapshot_from_cache()? {
876 
            Some(snap) => snap,
877 
            None => self.snapshot_as("oxy commit", true, true)?,
878 
        };
236f35c0 Andy McCall ↶ 879 
        let pending = snap.manifest.take();
98286406 Andy McCall ↶ 880 
        if snap.wc.tree_ops.is_empty() && snap.wc.edits.is_empty() {
236f35c0 Andy McCall ↶ 881 
            if let Some(m) = pending {
882 
                let conflicted = snap.base_snapshot.is_conflicted();
883 
                self.write_pending_manifest(m, conflicted, HashMap::new())?;
884 
            }
98286406 Andy McCall ↶ 885 
            return fail("nothing to commit: the working copy matches its base");
886 
        }
75167672 Andy McCall 887 
        if let Some(s) = snap
888 
            .statuses
889 
            .iter()
890 
            .find(|s| s.status == ChangeStatus::ConflictEdited)
891 
        {
98286406 Andy McCall ↶ 892 
            return fail(format!("{} still contains conflict markers", s.path));
893 
        }
894 
        if message.trim().is_empty() {
895 
            return fail("a commit message is required");
896 
        }
192fad1e Andy McCall ↶ 897 
        // What the whole working copy references, before a whole commit
898 
        // takes its operations over (they are moved, not copied: for a
899 
        // large tree they are most of the memory).
900 
        let wc_refs = referenced_changes(&snap.wc, &snap.base);
f9466021 Andy McCall ↶ 901 
        let changed_paths: Vec<String> = snap.statuses.iter().map(|s| s.path.clone()).collect();
957fee30 Andy McCall ↶ 902 
        let (tree_ops, edits, partial) = match selection {
192fad1e Andy McCall ↶ 903 
            Selection::All => (
904 
                std::mem::take(&mut snap.wc.tree_ops),
905 
                std::mem::take(&mut snap.wc.edits),
906 
                false,
907 
            ),
957fee30 Andy McCall ↶ 908 
            Selection::Paths(paths) => {
909 
                let (o, e) = select_paths(&snap.wc, &snap.base_snapshot, paths);
910 
                let partial = o.len() != snap.wc.tree_ops.len() || e.len() != snap.wc.edits.len();
911 
                (o, e, partial)
912 
            }
913 
            Selection::Hunks(indices) => {
914 
                let (o, e) = select_hunks(&snap.wc, indices);
915 
                let partial = o.len() != snap.wc.tree_ops.len() || e.len() != snap.wc.edits.len();
916 
                (o, e, partial)
917 
            }
918 
        };
919 
        if tree_ops.is_empty() && edits.is_empty() {
920 
            return fail("nothing selected to commit");
921 
        }
922 
        if partial && snap.resolves {
923 
            return fail("a conflict resolution must be committed whole");
924 
        }
6d3eefbc Andy McCall ↶ 925 
        // pre-commit hook: may veto.
926 
        self.run_hook(
927 
            "pre-commit",
928 
            &format!(
929 
                "{{\"event\":\"pre-commit\",\"message\":{},\"paths\":{}}}",
930 
                json_str(message),
75167672 Andy McCall 931 
                json_array(snap.statuses.iter().map(|s| format!(
932 
                    "{{\"path\":{},\"status\":{}}}",
933 
                    json_str(&s.path),
934 
                    json_str(&format!("{:?}", s.status))
935 
                )))
6d3eefbc Andy McCall ↶ 936 
            ),
937 
        )?;
98286406 Andy McCall ↶ 938 
        let (op_ref, op) = self.head_op()?;
192fad1e Andy McCall ↶ 939 
        let mut deps: BTreeSet<ChangeHash> = wc_refs;
6d3eefbc Andy McCall ↶ 940 
        // semantic-deps hook: may add explicit dependencies (spec §8.1).
941 
        if self.config.semantic_deps {
942 
            let touched: Vec<String> = snap.statuses.iter().map(|s| s.path.clone()).collect();
943 
            let by_node = snap.base_snapshot.tree.by_node();
944 
            let mut candidates: Vec<(ChangeId, ChangeHash, Vec<String>)> = Vec::new();
945 
            let mut seen: BTreeSet<ChangeHash> = BTreeSet::new();
946 
            for h in deps.iter().chain(snap.base.maximal(&snap.wc.deps).iter()) {
947 
                if !seen.insert(*h) {
948 
                    continue;
949 
                }
950 
                let c = self.store.get_change(h)?;
951 
                let mut paths: BTreeSet<String> = BTreeSet::new();
75167672 Andy McCall 952 
                for n in c
953 
                    .tree_ops
954 
                    .iter()
955 
                    .map(|o| o.node())
956 
                    .chain(c.edits.iter().map(|e| e.node()))
957 
                {
6d3eefbc Andy McCall ↶ 958 
                    if let Some(e) = by_node.get(&n) {
959 
                        paths.insert(e.path.clone());
960 
                    }
961 
                }
962 
                candidates.push((c.change_id, *h, paths.into_iter().collect()));
963 
            }
964 
            for d in self.semantic_deps(&snap.wc.change_id, &touched, &candidates)? {
965 
                if snap.base.contains(&d) {
966 
                    deps.insert(d);
967 
                }
968 
            }
969 
        }
98286406 Andy McCall ↶ 970 
        // A resolution depends on every change that contributed to the
971 
        // conflict (spec §8.1). Tree and opaque conflicts are resolved through
972 
        // ordinary operations whose references already cover the contributors.
973 
        deps.extend(snap.contributors.iter().copied());
974 
        for d in extra_deps {
975 
            if !snap.base.contains(d) {
75167672 Andy McCall 976 
                return fail(format!(
977 
                    "--depends-on {d}: not in the working copy's base state"
978 
                ));
98286406 Andy McCall ↶ 979 
            }
980 
            deps.insert(*d);
981 
        }
982 
        let mut change = Change {
6f8aeb99 Andy McCall ↶ 983 
            version: crate::objects::CHANGE_VERSION,
75167672 Andy McCall 984 
            change_id: if partial {
985 
                ChangeId::random()
986 
            } else {
987 
                snap.wc.change_id
988 
            },
957fee30 Andy McCall ↶ 989 
            deps: vec![],
98286406 Andy McCall ↶ 990 
            author: self.author(),
991 
            hlc: Hlc::now(),
992 
            message: message.trim_end().to_string(),
957fee30 Andy McCall ↶ 993 
            tree_ops,
994 
            edits,
75167672 Andy McCall 995 
            kind: if snap.resolves {
996 
                ChangeKind::Resolution
997 
            } else {
998 
                ChangeKind::Normal
999 
            },
98286406 Andy McCall ↶ 1000 
        };
957fee30 Andy McCall ↶ 1001 
        if partial {
1002 
            // Dependencies come from what the selected part references, not
1003 
            // from the whole working copy.
1004 
            deps = referenced_changes(&change, &snap.base);
1005 
            for d in extra_deps {
1006 
                deps.insert(*d);
1007 
            }
1008 
        }
1009 
        change.deps = deps.into_iter().collect();
98286406 Andy McCall ↶ 1010 
        change.normalize();
f9466021 Andy McCall ↶ 1011 
        let signing = crate::timing::phase("commit: sign and store");
614541c3 Andy McCall ↶ 1012 
        let hash = self.store_signed(&change)?;
f9466021 Andy McCall ↶ 1013 
        drop(signing);
1014 
        let extending = crate::timing::phase("commit: extend the state");
98286406 Andy McCall ↶ 1015
1016 
        // New bookmark heads: maximal elements of old heads plus the new change.
1017 
        // The working copy continues on its own base plus the new change, which
1018 
        // is exactly what the files on disk contain; the bookmark may hold
1019 
        // other heads (and conflicts) the working copy has not checked out.
1020 
        let old_heads = self.heads_of(&op.view, &self.config.bookmark)?;
1021 
        let old_base = snap.base.maximal(&snap.wc.deps);
1022 
        let mut all: Vec<ChangeHash> = old_heads.clone();
1023 
        all.extend(old_base.iter().copied());
1024 
        all.push(hash);
07eb99be Andy McCall ↶ 1025 
        // The snapshot already loaded the working copy's base; when that also
1026 
        // holds every bookmark head, it is the state to extend.
1027 
        let mut st = if old_heads.iter().all(|h| snap.base.contains(h)) {
1028 
            snap.base
1029 
        } else {
1030 
            self.load_state(&all)?
1031 
        };
192fad1e Andy McCall ↶ 1032 
        st.add_hashed(hash, change)?;
98286406 Andy McCall ↶ 1033 
        let mut with_new = old_heads.clone();
1034 
        with_new.push(hash);
1035 
        let new_heads = st.maximal(&with_new);
252e8cb2 Andy McCall ↶ 1036 
        // Automatic checkpoint once enough changes accumulated (spec §11).
75167672 Andy McCall 1037 
        let since = self
1038 
            .latest_checkpoint(&self.config.bookmark)?
1039 
            .map(|(_, cp)| cp.count)
1040 
            .unwrap_or(0);
1041 
        if self.config.checkpoint_interval > 0
1042 
            && (st.len() as u64).saturating_sub(since) >= self.config.checkpoint_interval
1043 
        {
252e8cb2 Andy McCall ↶ 1044 
            let cp_state = self.load_state_full(&new_heads)?;
1045 
            self.write_checkpoint(&self.config.bookmark, &cp_state, &new_heads)?;
1046 
        }
98286406 Andy McCall ↶ 1047 
        let mut wc_base = old_base.clone();
1048 
        wc_base.push(hash);
1049 
        let wc_base = st.maximal(&wc_base);
f9466021 Andy McCall ↶ 1050 
        drop(extending);
98286406 Andy McCall ↶ 1051
8457a6cc Andy McCall ↶ 1052 
        // The graph keys of the files as they now are, from the state the
1053 
        // commit just extended: a file whose graph it built has none, and
1054 
        // is read by the next snapshot; a key the next base does not share
1055 
        // (the bookmark held other heads) is simply not trusted then.
236f35c0 Andy McCall ↶ 1056 
        let keying = crate::timing::phase("commit: graph keys of the files");
f9466021 Andy McCall ↶ 1057 
        let committed_keys =
1058 
            (!partial).then(|| self.committed_graph_keys(&st, &changed_paths, &snap.attributes));
236f35c0 Andy McCall ↶ 1059 
        drop(keying);
f9466021 Andy McCall ↶ 1060 
        let recording = crate::timing::phase("commit: record");
98286406 Andy McCall ↶ 1061 
        let wc = self.new_working_copy(wc_base, "");
1062 
        let wc_hash = self.store.put_change(&wc)?;
1063 
        let mut view = op.view.clone();
75167672 Andy McCall 1064 
        view.bookmarks.insert(
1065 
            self.config.bookmark.clone(),
1066 
            self.store.put_state(&StateObject::new(new_heads))?,
1067 
        );
98286406 Andy McCall ↶ 1068 
        view.workspaces.insert(self.workspace, wc_hash);
1069 
        self.record_op(&[op_ref], &format!("oxy commit -m {message:?}"), view)?;
f9466021 Andy McCall ↶ 1070 
        drop(recording);
236f35c0 Andy McCall ↶ 1071 
        let rewriting = crate::timing::phase("commit: manifest");
957fee30 Andy McCall ↶ 1072 
        if partial {
1073 
            // The rest is rebuilt from disk by the next snapshot.
1074 
            self.drop_manifest();
1075 
        } else {
86e1824f Andy McCall ↶ 1076 
            let conflicted = snap.base_snapshot.is_conflicted() && !snap.resolves;
1077 
            // The files on disk are now exactly the new base's renderings:
1078 
            // record their graph keys, so the next snapshot neither reads nor
1079 
            // renders them (oxygen#35).
1080 
            let keys = if conflicted {
1081 
                HashMap::new()
1082 
            } else {
8457a6cc Andy McCall ↶ 1083 
                committed_keys.unwrap_or_default()
86e1824f Andy McCall ↶ 1084 
            };
236f35c0 Andy McCall ↶ 1085 
            match pending {
1086 
                Some(m) => self.write_pending_manifest(m, conflicted, keys)?,
1087 
                None => self.rewrite_manifest(conflicted, false, &keys)?,
1088 
            }
957fee30 Andy McCall ↶ 1089 
        }
236f35c0 Andy McCall ↶ 1090 
        drop(rewriting);
1091 
        let packing = crate::timing::phase("commit: autopack");
d02e545d Andy McCall ↶ 1092 
        self.maybe_autopack()?;
236f35c0 Andy McCall ↶ 1093 
        drop(packing);
1094 
        let _hook = crate::timing::phase("commit: hook");
75167672 Andy McCall 1095 
        let _ = self.run_hook(
1096 
            "post-commit",
1097 
            &format!(
1098 
                "{{\"event\":\"post-commit\",\"change_id\":{},\"hash\":{},\"message\":{}}}",
1099 
                json_str(&snap.wc.change_id.to_hex()),
1100 
                json_str(&hash.to_hex()),
1101 
                json_str(message)
1102 
            ),
1103 
        );
98286406 Andy McCall ↶ 1104 
        Ok(hash)
1105 
    }
1106 
}
1107
957fee30 Andy McCall ↶ 1108 
/// What part of the working copy a commit takes (spec §12.3).
1109 
#[derive(Clone, Debug)]
1110 
pub enum Selection {
1111 
    All,
1112 
    /// Every operation and edit on nodes under these paths.
1113 
    Paths(Vec<String>),
1114 
    /// Hunk indices as `Repo::hunks` numbers them; dependent hunks are added.
1115 
    Hunks(Vec<usize>),
1116 
}
1117
1118 
/// One selectable piece of the working copy: a tree operation or an edit.
1119 
#[derive(Clone, Debug)]
1120 
pub struct Hunk {
1121 
    pub index: usize,
1122 
    pub path: String,
1123 
    /// Rendered like `oxy diff`.
1124 
    pub text: String,
1125 
}
1126
1127 
/// The tree operations and edits on nodes under `paths`, plus the
1128 
/// additions of any new parent directories they need.
75167672 Andy McCall 1129 
fn select_paths(
1130 
    wc: &Change,
1131 
    base_snapshot: &Snapshot,
1132 
    paths: &[String],
1133 
) -> (Vec<TreeOp>, Vec<Edit>) {
1134 
    let selected = |p: &str| {
1135 
        paths
1136 
            .iter()
1137 
            .any(|x| p == x || p.starts_with(&format!("{}/", x.trim_end_matches('/'))))
1138 
    };
957fee30 Andy McCall ↶ 1139 
    let by_node = base_snapshot.tree.by_node();
1140 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
1141 
    for op in &wc.tree_ops {
75167672 Andy McCall 1142 
        if let TreeOp::AddNode {
1143 
            node,
1144 
            parent,
1145 
            name,
1146 
            kind,
1147 
            ..
1148 
        } = op
1149 
        {
957fee30 Andy McCall ↶ 1150 
            new_names.insert(*node, (*parent, name.clone(), *kind));
1151 
        }
1152 
    }
1153 
    let mut nodes: BTreeSet<NodeId> = BTreeSet::new();
75167672 Andy McCall 1154 
    for n in wc
1155 
        .tree_ops
1156 
        .iter()
1157 
        .map(|o| o.node())
1158 
        .chain(wc.edits.iter().map(|e| e.node()))
1159 
    {
957fee30 Andy McCall ↶ 1160 
        if path_of_node(&n, &by_node, &new_names).is_some_and(|p| selected(&p)) {
1161 
            nodes.insert(n);
1162 
        }
1163 
    }
1164 
    // New parents of selected new nodes come along.
1165 
    let mut changed = true;
1166 
    while changed {
1167 
        changed = false;
1168 
        for (n, (parent, _, _)) in &new_names {
1169 
            if nodes.contains(n) && new_names.contains_key(parent) && nodes.insert(*parent) {
1170 
                changed = true;
1171 
            }
1172 
        }
1173 
    }
75167672 Andy McCall 1174 
    let ops = wc
1175 
        .tree_ops
1176 
        .iter()
1177 
        .filter(|o| nodes.contains(&o.node()))
1178 
        .cloned()
1179 
        .collect();
1180 
    let edits = wc
1181 
        .edits
1182 
        .iter()
1183 
        .filter(|e| nodes.contains(&e.node()))
1184 
        .cloned()
1185 
        .collect();
957fee30 Andy McCall ↶ 1186 
    (ops, edits)
1187 
}
1188
1189 
/// The hunks `indices` name plus whatever they refer to within the working
1190 
/// copy (the parent directory of an added node, the insert that introduced
1191 
/// an anchor line), renumbered so the result is a well-formed change.
1192 
fn select_hunks(wc: &Change, indices: &[usize]) -> (Vec<TreeOp>, Vec<Edit>) {
1193 
    let n_ops = wc.tree_ops.len();
1194 
    let mut ops_sel = vec![false; n_ops];
1195 
    let mut edits_sel = vec![false; wc.edits.len()];
1196 
    for &i in indices {
1197 
        if i < n_ops {
1198 
            ops_sel[i] = true;
1199 
        } else if i - n_ops < wc.edits.len() {
1200 
            edits_sel[i - n_ops] = true;
1201 
        }
1202 
    }
1203 
    // Atom ranges each insert introduces, per node, in the working copy's numbering.
1204 
    let mut per_node: HashMap<NodeId, u32> = HashMap::new();
1205 
    let mut insert_ranges: Vec<Option<(NodeId, u32, u32)>> = Vec::new();
1206 
    for e in &wc.edits {
1207 
        insert_ranges.push(match e {
1208 
            Edit::Insert { node, lines, .. } => {
1209 
                let start = per_node.entry(*node).or_default();
1210 
                let r = (*node, *start, *start + lines.len() as u32);
1211 
                *start += lines.len() as u32;
1212 
                Some(r)
1213 
            }
1214 
            _ => None,
1215 
        });
1216 
    }
75167672 Andy McCall 1217 
    let add_index: HashMap<NodeId, usize> = wc
1218 
        .tree_ops
1219 
        .iter()
1220 
        .enumerate()
1221 
        .filter_map(|(i, o)| {
1222 
            if let TreeOp::AddNode { node, .. } = o {
1223 
                Some((*node, i))
1224 
            } else {
1225 
                None
1226 
            }
1227 
        })
1228 
        .collect();
1229 
    let insert_for = |node: &NodeId, index: u32| -> Option<usize> {
1230 
        insert_ranges
1231 
            .iter()
1232 
            .position(|r| matches!(r, Some((n, a, b)) if n == node && *a <= index && index < *b))
1233 
    };
957fee30 Andy McCall ↶ 1234 
    // Closure.
1235 
    let mut changed = true;
1236 
    while changed {
1237 
        changed = false;
1238 
        for i in 0..n_ops {
1239 
            if !ops_sel[i] {
1240 
                continue;
1241 
            }
d8c086ca Andy McCall ↶ 1242 
            if let TreeOp::AddNode { node, parent, .. } = &wc.tree_ops[i] {
957fee30 Andy McCall ↶ 1243 
                if let Some(&p) = add_index.get(parent) {
1244 
                    if !ops_sel[p] {
1245 
                        ops_sel[p] = true;
1246 
                        changed = true;
1247 
                    }
1248 
                }
d8c086ca Andy McCall ↶ 1249 
                // A new node's content belongs with its addition: nobody
1250 
                // wants an empty file committed apart from its lines.
1251 
                for (j, e) in wc.edits.iter().enumerate() {
1252 
                    if e.node() == *node && !edits_sel[j] {
1253 
                        edits_sel[j] = true;
1254 
                        changed = true;
1255 
                    }
1256 
                }
957fee30 Andy McCall ↶ 1257 
            }
1258 
        }
1259 
        for j in 0..wc.edits.len() {
1260 
            if !edits_sel[j] {
1261 
                continue;
1262 
            }
1263 
            let e = &wc.edits[j];
1264 
            if let Some(&a) = add_index.get(&e.node()) {
1265 
                if !ops_sel[a] {
1266 
                    ops_sel[a] = true;
1267 
                    changed = true;
1268 
                }
1269 
            }
75167672 Andy McCall 1270 
            if let Edit::Insert {
1271 
                node,
1272 
                after,
1273 
                before,
1274 
                ..
1275 
            } = e
1276 
            {
957fee30 Andy McCall ↶ 1277 
                for r in [after, before] {
1278 
                    if let AtomRef::Atom { change, index } = r {
1279 
                        if change.is_zero() {
1280 
                            if let Some(k) = insert_for(node, *index) {
1281 
                                if !edits_sel[k] {
1282 
                                    edits_sel[k] = true;
1283 
                                    changed = true;
1284 
                                }
1285 
                            }
1286 
                        }
1287 
                    }
1288 
                }
1289 
            }
1290 
        }
1291 
    }
1292 
    // Renumber self-references: old atom index → new index among selected inserts.
1293 
    let mut remap: HashMap<(NodeId, u32), u32> = HashMap::new();
1294 
    let mut new_counts: HashMap<NodeId, u32> = HashMap::new();
1295 
    for (j, r) in insert_ranges.iter().enumerate() {
1296 
        if let (true, Some((node, a, b))) = (edits_sel[j], r) {
1297 
            let next = new_counts.entry(*node).or_default();
1298 
            for old in *a..*b {
1299 
                remap.insert((*node, old), *next);
1300 
                *next += 1;
1301 
            }
1302 
        }
1303 
    }
1304 
    let map_ref = |r: AtomRef, node: &NodeId| -> AtomRef {
1305 
        match r {
75167672 Andy McCall 1306 
            AtomRef::Atom { change, index } if change.is_zero() => AtomRef::Atom {
1307 
                change,
1308 
                index: remap.get(&(*node, index)).copied().unwrap_or(index),
1309 
            },
957fee30 Andy McCall ↶ 1310 
            other => other,
1311 
        }
1312 
    };
75167672 Andy McCall 1313 
    let ops = wc
1314 
        .tree_ops
1315 
        .iter()
1316 
        .enumerate()
1317 
        .filter(|(i, _)| ops_sel[*i])
1318 
        .map(|(_, o)| o.clone())
1319 
        .collect();
957fee30 Andy McCall ↶ 1320 
    let edits = wc
1321 
        .edits
1322 
        .iter()
1323 
        .enumerate()
1324 
        .filter(|(j, _)| edits_sel[*j])
1325 
        .map(|(_, e)| match e {
75167672 Andy McCall 1326 
            Edit::Insert {
1327 
                node,
1328 
                after,
1329 
                before,
1330 
                lines,
1331 
            } => Edit::Insert {
1332 
                node: *node,
1333 
                after: map_ref(*after, node),
1334 
                before: map_ref(*before, node),
1335 
                lines: lines.clone(),
1336 
            },
957fee30 Andy McCall ↶ 1337 
            other => other.clone(),
1338 
        })
1339 
        .collect();
1340 
    (ops, edits)
1341 
}
1342
1343 
/// Render each tree operation and edit of `change` on its own, numbered.
1344 
pub fn render_hunks(change: &Change, base: &State, base_snapshot: &Snapshot) -> Vec<Hunk> {
1345 
    let by_node = base_snapshot.tree.by_node();
1346 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
1347 
    for op in &change.tree_ops {
75167672 Andy McCall 1348 
        if let TreeOp::AddNode {
1349 
            node,
1350 
            parent,
1351 
            name,
1352 
            kind,
1353 
            ..
1354 
        } = op
1355 
        {
957fee30 Andy McCall ↶ 1356 
            new_names.insert(*node, (*parent, name.clone(), *kind));
1357 
        }
1358 
    }
1359 
    let mut out = Vec::new();
1360 
    for (i, op) in change.tree_ops.iter().enumerate() {
75167672 Andy McCall 1361 
        let one = Change {
1362 
            tree_ops: vec![op.clone()],
1363 
            edits: vec![],
1364 
            ..change.clone()
1365 
        };
957fee30 Andy McCall ↶ 1366 
        let path = path_of_node(&op.node(), &by_node, &new_names).unwrap_or_default();
75167672 Andy McCall 1367 
        out.push(Hunk {
1368 
            index: i,
1369 
            path,
1370 
            text: render_change(&one, base, base_snapshot),
1371 
        });
957fee30 Andy McCall ↶ 1372 
    }
1373 
    for (j, e) in change.edits.iter().enumerate() {
75167672 Andy McCall 1374 
        let one = Change {
1375 
            tree_ops: vec![],
1376 
            edits: vec![e.clone()],
1377 
            ..change.clone()
1378 
        };
957fee30 Andy McCall ↶ 1379 
        let path = path_of_node(&e.node(), &by_node, &new_names).unwrap_or_default();
75167672 Andy McCall 1380 
        out.push(Hunk {
1381 
            index: change.tree_ops.len() + j,
1382 
            path,
1383 
            text: render_change(&one, base, base_snapshot),
1384 
        });
957fee30 Andy McCall ↶ 1385 
    }
1386 
    out
1387 
}
1388
c6aa5bd9 Andy McCall ↶ 1389 
/// One file's part of a diff, for `oxy diff --json`.
1390 
#[derive(Clone, Debug)]
1391 
pub struct FileDiff {
1392 
    pub path: String,
1393 
    /// `added`, `removed`, `modified`, `renamed`, `mode`, `replaced`.
1394 
    pub kind: String,
1395 
    pub hunks: Vec<DiffHunk>,
1396 
    /// For renames, the previous path.
1397 
    pub from: Option<String>,
1398 
}
1399
1400 
/// Lines inserted after `base_line` (0 = before the first line) and lines
1401 
/// removed, with their base line numbers.
1402 
#[derive(Clone, Debug, Default)]
1403 
pub struct DiffHunk {
1404 
    pub base_line: u32,
1405 
    pub added: Vec<Vec<u8>>,
1406 
    pub removed: Vec<(u32, Vec<u8>)>,
1407 
}
1408
1409 
impl Repo {
1410 
    /// The content of `path` in a state: `heads`, or the working copy's base.
1411 
    pub fn file_at(&self, path: &str, heads: Option<&[ChangeHash]>) -> Result<Vec<u8>> {
19303b4a Andy McCall ↶ 1412 
        // The state's history and tree, then the one file's graph: nothing
1413 
        // else is built or rendered (oxygen#92).
1414 
        // The working copy's base, and a state named by its heads (`-r
1415 
        // main` on main), come from the workspace's state index.
1416 
        let base = if self.is_bare() {
1417 
            None
1418 
        } else {
1419 
            Some(self.working_copy()?.1.deps)
1420 
        };
1421 
        let same = |a: &[ChangeHash], b: &[ChangeHash]| {
1422 
            let (mut a, mut b) = (a.to_vec(), b.to_vec());
1423 
            a.sort();
1424 
            b.sort();
1425 
            a == b
1426 
        };
1427 
        let mut state = match (heads, base) {
1428 
            (Some(h), Some(b)) if same(h, &b) => self.load_state_indexed(&b)?,
1429 
            (Some(h), _) => self.load_state_history(h)?,
1430 
            (None, Some(b)) => self.load_state_indexed(&b)?,
1431 
            (None, None) => return fail("this is a bare repository: it has no working copy"),
1432 
        };
1433 
        let view = state.tree().view();
1434 
        let Some(node) = view.entries.get(path).map(|e| e.node) else {
1435 
            return fail(format!("{path} is not in that state"));
c6aa5bd9 Andy McCall ↶ 1436 
        };
19303b4a Andy McCall ↶ 1437 
        self.load_graphs_of(&mut state, &[node])?;
1438 
        let snapshot = state.materialize_paths(view, [path]);
75167672 Andy McCall 1439 
        let Some(f) = snapshot.files.get(path) else {
1440 
            return fail(format!("{path} is not in that state"));
1441 
        };
c6aa5bd9 Andy McCall ↶ 1442 
        match &f.content {
1443 
            FileContent::Dir => fail(format!("{path} is a directory")),
1444 
            FileContent::Text { bytes, .. } => Ok(bytes.clone()),
1445 
            FileContent::Opaque { blobs } => match blobs.first() {
1446 
                Some(b) => self.read_blob(b),
1447 
                None => Ok(vec![]),
1448 
            },
1449 
        }
1450 
    }
1451
1452 
    /// Restore `paths` (files or directories) to the working copy's base:
1453 
    /// edits are undone, additions removed, removals put back. Returns the
1454 
    /// number of paths touched.
1455 
    pub fn discard(&self, paths: &[String]) -> Result<usize> {
6caacfff Andy McCall ↶ 1456 
        let _wc = self.lock_working_copy()?;
8ab265dc Andy McCall ↶ 1457 
        let snap = self.snapshot_status("oxy discard")?;
75167672 Andy McCall 1458 
        let under = |p: &str| {
1459 
            paths
1460 
                .iter()
1461 
                .any(|x| p == x || p.starts_with(&format!("{}/", x.trim_end_matches('/'))))
1462 
        };
c6aa5bd9 Andy McCall ↶ 1463 
        let attrs = Repo::state_attributes(&snap.base_snapshot);
1464 
        let mut n = 0;
1465 
        for s in &snap.statuses {
1466 
            if !under(&s.path) {
1467 
                continue;
1468 
            }
1469 
            match snap.base_snapshot.files.get(&s.path) {
1470 
                None => {
1471 
                    // Added (or the new name of a rename): remove it.
1472 
                    let full = self.root.join(&s.path);
1473 
                    if full.is_dir() {
1474 
                        let _ = fs::remove_dir_all(&full);
1475 
                    } else {
1476 
                        let _ = fs::remove_file(&full);
1477 
                    }
1478 
                    n += 1;
1479 
                }
1480 
                Some(f) => {
1481 
                    match &f.content {
1482 
                        FileContent::Dir => fs::create_dir_all(self.root.join(&s.path))?,
1483 
                        FileContent::Text { bytes, .. } => {
75167672 Andy McCall 1484 
                            let on_disk: std::borrow::Cow<[u8]> = if attrs.lookup(&s.path).eol
1485 
                                == Some(crate::attributes::Eol::Crlf)
1486 
                            {
1487 
                                crate::attributes::to_crlf(bytes).into()
1488 
                            } else {
1489 
                                bytes.as_slice().into()
1490 
                            };
c6aa5bd9 Andy McCall ↶ 1491 
                            self.write_file(&s.path, &on_disk, f.mode.exec)?;
1492 
                        }
1493 
                        FileContent::Opaque { blobs } => {
1494 
                            let bytes = match blobs.first() {
1495 
                                Some(b) => self.read_blob(b)?,
1496 
                                None => vec![],
1497 
                            };
1498 
                            if f.kind == NodeKind::Symlink {
1499 
                                self.write_symlink(&s.path, &bytes)?;
1500 
                            } else if f.kind == NodeKind::File {
1501 
                                self.write_file(&s.path, &bytes, f.mode.exec)?;
1502 
                            }
1503 
                        }
1504 
                    }
1505 
                    n += 1;
1506 
                }
1507 
            }
1508 
        }
1509 
        // A rename's old name is not in the statuses; put it back too.
1510 
        for (p, f) in &snap.base_snapshot.files {
75167672 Andy McCall 1511 
            if under(p)
1512 
                && !self.root.join(p).exists()
1513 
                && f.kind != NodeKind::Dir
1514 
                && f.kind != NodeKind::Subrepo
1515 
                && !snap.statuses.iter().any(|s| s.path == *p)
1516 
            {
c6aa5bd9 Andy McCall ↶ 1517 
                let bytes = match &f.content {
1518 
                    FileContent::Text { bytes, .. } => bytes.clone(),
75167672 Andy McCall 1519 
                    FileContent::Opaque { blobs } => blobs
1520 
                        .first()
1521 
                        .map(|b| self.read_blob(b))
1522 
                        .transpose()?
1523 
                        .unwrap_or_default(),
c6aa5bd9 Andy McCall ↶ 1524 
                    FileContent::Dir => continue,
1525 
                };
75167672 Andy McCall 1526 
                if f.kind == NodeKind::Symlink {
1527 
                    self.write_symlink(p, &bytes)?
1528 
                } else {
1529 
                    self.write_file(p, &bytes, f.mode.exec)?
1530 
                }
c6aa5bd9 Andy McCall ↶ 1531 
                n += 1;
1532 
            }
1533 
        }
1534 
        self.drop_manifest();
1535 
        Ok(n)
1536 
    }
1537
1538 
    /// Undo the working-copy hunks `indices` (as `hunks` numbers them) on
1539 
    /// disk, keeping the others: the files are rewritten from the base plus
1540 
    /// the remaining hunks.
1541 
    pub fn discard_hunks(&self, indices: &[usize]) -> Result<usize> {
6caacfff Andy McCall ↶ 1542 
        let _wc = self.lock_working_copy()?;
c6aa5bd9 Andy McCall ↶ 1543 
        let (snap, hunks) = self.hunks()?;
75167672 Andy McCall 1544 
        let keep: Vec<usize> = hunks
1545 
            .iter()
1546 
            .map(|h| h.index)
1547 
            .filter(|i| !indices.contains(i))
1548 
            .collect();
c6aa5bd9 Andy McCall ↶ 1549 
        let (ops, edits) = select_hunks(&snap.wc, &keep);
1550 
        let dropped_nodes: BTreeSet<NodeId> = hunks
1551 
            .iter()
1552 
            .filter(|h| indices.contains(&h.index))
75167672 Andy McCall 1553 
            .map(|h| {
1554 
                if h.index < snap.wc.tree_ops.len() {
1555 
                    snap.wc.tree_ops[h.index].node()
1556 
                } else {
1557 
                    snap.wc.edits[h.index - snap.wc.tree_ops.len()].node()
1558 
                }
1559 
            })
c6aa5bd9 Andy McCall ↶ 1560 
            .collect();
1561 
        let mut partial = snap.wc.clone();
1562 
        partial.tree_ops = ops;
1563 
        partial.edits = edits;
1564 
        partial.normalize();
1565 
        let mut st = snap.base.clone();
1566 
        st.add_unchecked(partial);
8ab265dc Andy McCall ↶ 1567 
        // Every file of the target is rendered: build the graphs the
1568 
        // partial snapshot left out.
1569 
        self.load_all_graphs(&mut st)?;
c6aa5bd9 Andy McCall ↶ 1570 
        let target = st.materialize();
1571 
        let by_node_target = target.tree.by_node();
1572 
        let by_node_base = snap.base_snapshot.tree.by_node();
1573 
        let attrs = Repo::state_attributes(&target);
1574 
        let mut n = 0;
1575 
        for node in dropped_nodes {
1576 
            match by_node_target.get(&node) {
1577 
                Some(e) => {
1578 
                    let f = &target.files[&e.path];
1579 
                    match &f.content {
1580 
                        FileContent::Dir => {}
1581 
                        FileContent::Text { bytes, .. } => {
75167672 Andy McCall 1582 
                            let on_disk: std::borrow::Cow<[u8]> = if attrs.lookup(&e.path).eol
1583 
                                == Some(crate::attributes::Eol::Crlf)
1584 
                            {
1585 
                                crate::attributes::to_crlf(bytes).into()
1586 
                            } else {
1587 
                                bytes.as_slice().into()
1588 
                            };
c6aa5bd9 Andy McCall ↶ 1589 
                            self.write_file(&e.path, &on_disk, f.mode.exec)?;
1590 
                        }
1591 
                        FileContent::Opaque { blobs } => {
75167672 Andy McCall 1592 
                            let bytes = blobs
1593 
                                .first()
1594 
                                .map(|b| self.read_blob(b))
1595 
                                .transpose()?
1596 
                                .unwrap_or_default();
1597 
                            if f.kind == NodeKind::Symlink {
1598 
                                self.write_symlink(&e.path, &bytes)?
1599 
                            } else {
1600 
                                self.write_file(&e.path, &bytes, f.mode.exec)?
1601 
                            }
c6aa5bd9 Andy McCall ↶ 1602 
                        }
1603 
                    }
1604 
                    // A discarded rename: the new name goes.
75167672 Andy McCall 1605 
                    if let Some(old_path) = snap
1606 
                        .statuses
1607 
                        .iter()
1608 
                        .find(|s| {
1609 
                            s.status == ChangeStatus::Renamed && !target.files.contains_key(&s.path)
1610 
                        })
1611 
                        .map(|s| s.path.clone())
1612 
                    {
c6aa5bd9 Andy McCall ↶ 1613 
                        let _ = fs::remove_file(self.root.join(old_path));
1614 
                    }
1615 
                    n += 1;
1616 
                }
1617 
                None => {
1618 
                    // The node exists in neither target nor base with a live path: an addition to remove.
75167672 Andy McCall 1619 
                    if let Some(added) = snap.statuses.iter().find(|s| {
1620 
                        s.status == ChangeStatus::Added
1621 
                            && !target.files.contains_key(&s.path)
1622 
                            && !by_node_base.contains_key(&node)
1623 
                    }) {
c6aa5bd9 Andy McCall ↶ 1624 
                        let full = self.root.join(&added.path);
75167672 Andy McCall 1625 
                        if full.is_dir() {
1626 
                            let _ = fs::remove_dir_all(&full);
1627 
                        } else {
1628 
                            let _ = fs::remove_file(&full);
1629 
                        }
c6aa5bd9 Andy McCall ↶ 1630 
                        n += 1;
1631 
                    }
1632 
                }
1633 
            }
1634 
        }
1635 
        self.drop_manifest();
1636 
        Ok(n)
1637 
    }
1638
1639 
    /// The diff of the working copy (`None`) or of a change, per file with
1640 
    /// line-numbered hunks (spec: `oxy diff --json`).
1641 
    pub fn diff_files(&self, change: Option<&ChangeHash>) -> Result<Vec<FileDiff>> {
8ab265dc Andy McCall ↶ 1642 
        let (c, base, base_snapshot) = match change {
c6aa5bd9 Andy McCall ↶ 1643 
            Some(h) => {
1644 
                let c = self.store.get_change(h)?;
8ab265dc Andy McCall ↶ 1645 
                let (base, snapshot) = self.change_base(&c)?;
1646 
                (c, base, snapshot)
c6aa5bd9 Andy McCall ↶ 1647 
            }
77445673 Andy McCall ↶ 1648 
            None => self.working_copy_with_base()?,
c6aa5bd9 Andy McCall ↶ 1649 
        };
1650 
        Ok(diff_files_of(&c, &base, &base_snapshot))
e226a6cc Andy McCall ↶ 1651 
    }
1652
1653 
    /// The snapshot a command would take, without taking it, when the
1654 
    /// status cache holds: every file is as the last snapshot left it, so
1655 
    /// that snapshot's working-copy change is the answer. Its base comes
1656 
    /// from the state index, the `.oxyattributes` files from disk. `None`
1657 
    /// when a snapshot must be taken, and when there is nothing to commit
1658 
    /// (the snapshot then says so). A conflicted base never gets here: the
1659 
    /// cache does not hold for one.
1660 
    pub(crate) fn snapshot_from_cache(&self) -> Result<Option<SnapshotResult>> {
1661 
        let Some(statuses) = self.status_by_manifest()? else {
1662 
            return Ok(None);
1663 
        };
1664 
        if statuses.is_empty() {
1665 
            return Ok(None);
1666 
        }
1667 
        let _t = crate::timing::phase("snapshot: from the status cache");
1668 
        let (wc_hash, wc) = self.working_copy()?;
1669 
        let base = self.load_state_indexed(&wc.deps)?;
1670 
        let (base, base_snapshot) = self.render_base_for(&wc, base)?;
1671 
        let attribute_files: Vec<(String, Vec<u8>)> = match self.read_manifest()? {
1672 
            Some((_, _, m)) => m
1673 
                .entries
1674 
                .keys()
1675 
                .filter(|p| is_attributes(p))
1676 
                .filter_map(|p| fs::read(self.root.join(p)).ok().map(|b| (p.clone(), b)))
1677 
                .collect(),
1678 
            None => Vec::new(),
1679 
        };
1680 
        let attributes = crate::attributes::Attributes::from_files(
1681 
            attribute_files
1682 
                .iter()
1683 
                .map(|(p, b)| (p.as_str(), b.as_slice())),
1684 
        );
1685 
        Ok(Some(SnapshotResult {
1686 
            wc_hash,
1687 
            wc,
1688 
            base,
1689 
            base_snapshot,
1690 
            statuses,
1691 
            changed: false,
1692 
            resolves: false,
1693 
            contributors: Vec::new(),
1694 
            attributes,
1695 
            untracked_nested: self.untracked_nested.borrow().clone(),
236f35c0 Andy McCall ↶ 1696 
            manifest: None,
e226a6cc Andy McCall ↶ 1697 
        }))
77445673 Andy McCall ↶ 1698 
    }
1699
1700 
    /// The working-copy change with its base state and the base's
1701 
    /// rendering, for showing what the working copy changes. While the
1702 
    /// status cache holds (every file as the last snapshot left it), the
1703 
    /// stored working-copy change is the answer and no snapshot is taken;
1704 
    /// otherwise a snapshot makes it.
1705 
    pub fn working_copy_with_base(&self) -> Result<(Change, State, Snapshot)> {
1706 
        if self.status_by_manifest()?.is_some() {
1707 
            let (_, wc) = self.working_copy()?;
1708 
            let base = self.load_state_indexed(&wc.deps)?;
1709 
            let (base, snapshot) = self.render_base_for(&wc, base)?;
1710 
            return Ok((wc, base, snapshot));
1711 
        }
1712 
        let snap = self.snapshot_status("oxy diff")?;
1713 
        Ok((snap.wc, snap.base, snap.base_snapshot))
c6aa5bd9 Andy McCall ↶ 1714 
    }
1715
1716 
    /// The log restricted to changes touching `path` (its node, so renames
1717 
    /// are followed), on the current bookmark.
1718 
    pub fn log_for_path(&self, path: &str) -> Result<Vec<LogEntry>> {
e02bf0cd Andy McCall ↶ 1719 
        let heads = self.bookmark_heads()?;
1720 
        if self.is_bare() {
1721 
            return self.log_for_path_in(path, &heads);
1722 
        }
c6aa5bd9 Andy McCall ↶ 1723 
        let entries = self.log()?;
1724 
        let (_, wc) = self.working_copy()?;
170bb436 Andy McCall ↶ 1725 
        // The path's node from the trees alone: the base's (from the
1726 
        // workspace's state index), else the bookmark's. Nothing is
1727 
        // rendered.
1728 
        let node_in = |state: State| state.tree().view().entries.get(path).map(|e| e.node);
1729 
        let node = match node_in(self.load_state_indexed(&wc.deps)?) {
1730 
            Some(n) => Some(n),
1731 
            None => node_in(self.load_state_history(&heads)?),
1732 
        };
75167672 Andy McCall 1733 
        let Some(node) = node else {
1734 
            return fail(format!("{path} is not tracked"));
1735 
        };
e02bf0cd Andy McCall ↶ 1736 
        Ok(self.entries_touching(entries, node))
1737 
    }
1738
1739 
    /// The log of the state with these heads restricted to changes touching
1740 
    /// `path` (its node, so renames are followed).
1741 
    pub fn log_for_path_in(&self, path: &str, heads: &[ChangeHash]) -> Result<Vec<LogEntry>> {
e8d38e47 Andy McCall ↶ 1742 
        self.log_for_path_of(&self.load_state_history(heads)?, path, None)
bc489b6c Andy McCall ↶ 1743 
    }
1744
1745 
    /// The changes of a loaded state touching `path`, newest first, at most
1746 
    /// `limit` of them. The path is found in the state's tree alone, so no
1747 
    /// file is rendered, and the walk stops once `limit` are found.
1748 
    pub fn log_for_path_of(
1749 
        &self,
1750 
        state: &State,
1751 
        path: &str,
1752 
        limit: Option<usize>,
1753 
    ) -> Result<Vec<LogEntry>> {
1754 
        let Some(node) = state.tree().view().entries.get(path).map(|e| e.node) else {
75167672 Andy McCall 1755 
            return fail(format!("{path} is not in that state"));
1756 
        };
bc489b6c Andy McCall ↶ 1757 
        Ok(self.entries_touching_up_to(self.log_of(state), node, limit))
e02bf0cd Andy McCall ↶ 1758 
    }
1759
1760 
    fn entries_touching(&self, entries: Vec<LogEntry>, node: NodeId) -> Vec<LogEntry> {
bc489b6c Andy McCall ↶ 1761 
        self.entries_touching_up_to(entries, node, None)
1762 
    }
1763
1764 
    fn entries_touching_up_to(
1765 
        &self,
1766 
        entries: Vec<LogEntry>,
1767 
        node: NodeId,
1768 
        limit: Option<usize>,
1769 
    ) -> Vec<LogEntry> {
170bb436 Andy McCall ↶ 1770 
        let mut touched = self.touched();
1771 
        if limit.is_none() {
1772 
            let hashes: Vec<ChangeHash> = entries.iter().map(|e| e.hash).collect();
1773 
            touched.fill(self, &hashes);
1774 
        }
1775 
        let wanted = HashSet::from([node]);
c6aa5bd9 Andy McCall ↶ 1776 
        let mut out = Vec::new();
1777 
        for e in entries {
bc489b6c Andy McCall ↶ 1778 
            if limit.is_some_and(|n| out.len() >= n) {
1779 
                break;
1780 
            }
170bb436 Andy McCall ↶ 1781 
            if touched.touches(self, &e.hash, &wanted) {
c6aa5bd9 Andy McCall ↶ 1782 
                out.push(e);
1783 
            }
1784 
        }
170bb436 Andy McCall ↶ 1785 
        touched.save();
e02bf0cd Andy McCall ↶ 1786 
        out
c6aa5bd9 Andy McCall ↶ 1787 
    }
1788 
}
1789
1790 
/// Per-file hunks of `change` against `base`.
1791 
pub fn diff_files_of(change: &Change, base: &State, base_snapshot: &Snapshot) -> Vec<FileDiff> {
1792 
    let by_node = base_snapshot.tree.by_node();
1793 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
1794 
    for op in &change.tree_ops {
75167672 Andy McCall 1795 
        if let TreeOp::AddNode {
1796 
            node,
1797 
            parent,
1798 
            name,
1799 
            kind,
1800 
            ..
1801 
        } = op
1802 
        {
c6aa5bd9 Andy McCall ↶ 1803 
            new_names.insert(*node, (*parent, name.clone(), *kind));
1804 
        }
1805 
    }
75167672 Andy McCall 1806 
    let path_of = |n: &NodeId| -> String {
1807 
        path_of_node(n, &by_node, &new_names).unwrap_or_else(|| format!("<node {}>", n.short(8)))
1808 
    };
c6aa5bd9 Andy McCall ↶ 1809 
    let mut files: BTreeMap<NodeId, FileDiff> = BTreeMap::new();
1810 
    let mut order: Vec<NodeId> = Vec::new();
1811 
    let mut entry = |files: &mut BTreeMap<NodeId, FileDiff>, n: NodeId, kind: &str| {
1812 
        if let std::collections::btree_map::Entry::Vacant(v) = files.entry(n) {
1813 
            order.push(n);
75167672 Andy McCall 1814 
            v.insert(FileDiff {
1815 
                path: path_of(&n),
1816 
                kind: kind.to_string(),
1817 
                hunks: vec![],
1818 
                from: None,
1819 
            });
c6aa5bd9 Andy McCall ↶ 1820 
        }
1821 
    };
1822 
    for op in &change.tree_ops {
1823 
        match op {
1824 
            TreeOp::AddNode { node, kind, .. } => {
1825 
                if *kind != NodeKind::Dir {
1826 
                    entry(&mut files, *node, "added");
1827 
                }
1828 
            }
1829 
            TreeOp::RemoveNode { node, .. } => {
1830 
                entry(&mut files, *node, "removed");
1831 
                if let Some(e) = by_node.get(node) {
1832 
                    files.get_mut(node).unwrap().path = e.path.clone();
1833 
                }
1834 
            }
75167672 Andy McCall 1835 
            TreeOp::Move {
1836 
                node, parent, name, ..
1837 
            } => {
c6aa5bd9 Andy McCall ↶ 1838 
                entry(&mut files, *node, "renamed");
1839 
                let f = files.get_mut(node).unwrap();
1840 
                f.from = by_node.get(node).map(|e| e.path.clone());
75167672 Andy McCall 1841 
                let dir = if parent.is_root() {
1842 
                    String::new()
1843 
                } else {
1844 
                    path_of_node(parent, &by_node, &new_names).unwrap_or_default()
1845 
                };
1846 
                f.path = if dir.is_empty() {
1847 
                    name.clone()
1848 
                } else {
1849 
                    format!("{dir}/{name}")
1850 
                };
c6aa5bd9 Andy McCall ↶ 1851 
            }
1852 
            TreeOp::SetMode { node, .. } => entry(&mut files, *node, "mode"),
1853 
        }
1854 
    }
1855 
    // Base line numbers per node, 1-based.
1856 
    let mut line_of: HashMap<NodeId, HashMap<(ChangeHash, u32), u32>> = HashMap::new();
1857 
    let mut lines_for = |n: &NodeId| -> HashMap<(ChangeHash, u32), u32> {
75167672 Andy McCall 1858 
        line_of
1859 
            .entry(*n)
1860 
            .or_insert_with(|| {
1861 
                base.live_lines(n)
1862 
                    .into_iter()
1863 
                    .enumerate()
1864 
                    .map(|(i, (k, _))| (k, i as u32 + 1))
1865 
                    .collect()
1866 
            })
1867 
            .clone()
c6aa5bd9 Andy McCall ↶ 1868 
    };
1869 
    for e in &change.edits {
1870 
        let n = e.node();
1871 
        let is_new = new_names.contains_key(&n);
1872 
        entry(&mut files, n, if is_new { "added" } else { "modified" });
1873 
        let f = files.get_mut(&n).unwrap();
1874 
        match e {
1875 
            Edit::Insert { after, lines, .. } => {
1876 
                let map = lines_for(&n);
1877 
                let base_line = match after {
1878 
                    AtomRef::Start => 0,
1879 
                    AtomRef::End => map.len() as u32,
75167672 Andy McCall 1880 
                    AtomRef::Atom { change, index } if change.is_zero() => {
1881 
                        f.hunks.last().map(|h| h.base_line).unwrap_or(0)
1882 
                    }
1883 
                    AtomRef::Atom { change, index } => {
1884 
                        map.get(&(*change, *index)).copied().unwrap_or(0)
1885 
                    }
c6aa5bd9 Andy McCall ↶ 1886 
                };
75167672 Andy McCall 1887 
                f.hunks.push(DiffHunk {
1888 
                    base_line,
01a47cd7 Andy McCall ↶ 1889 
                    added: lines.to_vec(),
75167672 Andy McCall 1890 
                    removed: vec![],
1891 
                });
c6aa5bd9 Andy McCall ↶ 1892 
            }
1893 
            Edit::Delete { atoms, .. } => {
1894 
                let map = lines_for(&n);
1895 
                let mut removed = Vec::new();
1896 
                for r in atoms {
1897 
                    for i in r.from..r.to {
1898 
                        let line = map.get(&(r.change, i)).copied().unwrap_or(0);
75167672 Andy McCall 1899 
                        let text = base
1900 
                            .file(&n)
1901 
                            .and_then(|g| g.content(&(r.change, i)))
1902 
                            .unwrap_or(b"")
1903 
                            .to_vec();
c6aa5bd9 Andy McCall ↶ 1904 
                        removed.push((line, text));
1905 
                    }
1906 
                }
75167672 Andy McCall 1907 
                let base_line = removed
1908 
                    .first()
1909 
                    .map(|(l, _)| l.saturating_sub(1))
1910 
                    .unwrap_or(0);
1911 
                f.hunks.push(DiffHunk {
1912 
                    base_line,
1913 
                    added: vec![],
1914 
                    removed,
1915 
                });
c6aa5bd9 Andy McCall ↶ 1916 
            }
1917 
            Edit::Replace { .. } => {
1918 
                if !is_new {
1919 
                    f.kind = "replaced".into();
1920 
                }
1921 
            }
1922 
        }
1923 
    }
1924 
    order.into_iter().filter_map(|n| files.remove(&n)).collect()
1925 
}
1926
98286406 Andy McCall ↶ 1927 
/// Builds the tree ops and edits of a working-copy change against a base.
dfb23a3a Andy McCall ↶ 1928 
/// `BLAKE3("oxy node 1" 0 seed kind path)`, truncated: a node identifier
1929 
/// that the same seed and path always give (never the root's).
1930 
fn seeded_node(seed: &[u8; 32], path: &str, kind: NodeKind) -> NodeId {
1931 
    let mut h = blake3::Hasher::new();
1932 
    h.update(b"oxy node 1\0");
1933 
    h.update(seed);
1934 
    h.update(&[match kind {
1935 
        NodeKind::File => 1,
1936 
        NodeKind::Dir => 2,
1937 
        NodeKind::Symlink => 3,
1938 
        NodeKind::Subrepo => 4,
1939 
    }]);
1940 
    h.update(path.as_bytes());
1941 
    let mut id = [0u8; 16];
1942 
    id.copy_from_slice(&h.finalize().as_bytes()[..16]);
1943 
    if id == [0; 16] {
1944 
        id[15] = 1;
1945 
    }
1946 
    NodeId(id)
1947 
}
1948
98286406 Andy McCall ↶ 1949 
struct ChangeBuilder<'a> {
1950 
    base: &'a State,
1951 
    base_snapshot: &'a Snapshot,
1952 
    tree_ops: Vec<TreeOp>,
1953 
    edits: Vec<Edit>,
1954 
    statuses: Vec<PathStatus>,
1955 
    dir_nodes: HashMap<String, NodeId>,
1956 
    reuse: HashMap<String, (NodeId, NodeKind)>,
1957 
    /// Changes that contributed to conflicts this working copy resolves.
1958 
    contributors: BTreeSet<ChangeHash>,
1959 
    resolves: bool,
cdb5beba Andy McCall ↶ 1960 
    attrs: crate::attributes::Attributes,
b7f102fc Andy McCall ↶ 1961 
    blobs: HashMap<String, ObjectRef>,
f7510b63 Andy McCall ↶ 1962 
    /// Opaque files whose replace conflict the content settles.
1963 
    resolve: HashSet<String>,
dfb23a3a Andy McCall ↶ 1964 
    /// When set, new nodes' identifiers derive from it and their paths.
1965 
    node_seed: Option<[u8; 32]>,
a87c7b9d Andy McCall ↶ 1966 
    /// For a Git import, the base's dead directories by where they were
1967 
    /// (see `ensure_dir`), worked out when first needed.
1968 
    dead_dirs: Option<HashMap<(NodeId, String), (NodeId, crate::tree::OpKey)>>,
360aa74a Andy McCall ↶ 1969 
    /// Text edits of edited files, worked out in parallel beforehand
1970 
    /// (`diff_in_parallel`); a file not here is diffed when it is updated.
1971 
    diffs: HashMap<String, Vec<Edit>>,
98286406 Andy McCall ↶ 1972 
}
1973
1974 
impl<'a> ChangeBuilder<'a> {
75167672 Andy McCall 1975 
    fn new(
1976 
        base: &'a State,
1977 
        base_snapshot: &'a Snapshot,
1978 
        reuse: HashMap<String, (NodeId, NodeKind)>,
1979 
    ) -> Self {
98286406 Andy McCall ↶ 1980 
        let mut dir_nodes: HashMap<String, NodeId> = HashMap::new();
1981 
        dir_nodes.insert(String::new(), NodeId::ROOT);
e83354cd Andy McCall ↶ 1982 
        for (p, e) in base_snapshot.tree.entries.iter() {
2306a4d2 Andy McCall ↶ 1983 
            if e.kind == NodeKind::Dir {
1984 
                dir_nodes.insert(p.clone(), e.node);
98286406 Andy McCall ↶ 1985 
            }
1986 
        }
75167672 Andy McCall 1987 
        ChangeBuilder {
1988 
            base,
1989 
            base_snapshot,
1990 
            tree_ops: vec![],
1991 
            edits: vec![],
1992 
            statuses: vec![],
1993 
            dir_nodes,
1994 
            reuse,
1995 
            contributors: BTreeSet::new(),
1996 
            resolves: false,
1997 
            attrs: crate::attributes::Attributes::default(),
1998 
            blobs: HashMap::new(),
f7510b63 Andy McCall ↶ 1999 
            resolve: HashSet::new(),
dfb23a3a Andy McCall ↶ 2000 
            node_seed: None,
a87c7b9d Andy McCall ↶ 2001 
            dead_dirs: None,
360aa74a Andy McCall ↶ 2002 
            diffs: HashMap::new(),
75167672 Andy McCall 2003 
        }
98286406 Andy McCall ↶ 2004 
    }
2005
360aa74a Andy McCall ↶ 2006 
    /// Diff the edited text files among `entries` against their base
2007 
    /// renderings at once, on every core, for `update_path` to take: the
2008 
    /// same diff it would make, only not one file after another.
2009 
    fn diff_in_parallel(&mut self, entries: &BTreeMap<String, FsEntry>) {
2010 
        use rayon::prelude::*;
2011 
        let files = &self.base_snapshot.files;
2012 
        let edited: Vec<(&String, &FsEntry, &crate::state::FileState)> = entries
2013 
            .iter()
2014 
            .filter_map(|(p, e)| {
2015 
                let x = files.get(p)?;
2016 
                (x.kind == e.kind && !e.unchanged && !self.opaque_for(p, e)).then_some((p, e, x))
2017 
            })
2018 
            .collect();
2019 
        // Below this, starting the threads costs more than it saves.
2020 
        if edited.len() < 16 {
2021 
            return;
2022 
        }
2023 
        let this = &*self;
2024 
        let attrs = &this.attrs;
2025 
        let diffs: Vec<(String, Vec<Edit>)> = edited
2026 
            .par_iter()
2027 
            .filter_map(|(p, e, x)| {
2028 
                let FileContent::Text {
2029 
                    bytes,
2030 
                    materialized,
2031 
                } = &x.content
2032 
                else {
2033 
                    return None;
2034 
                };
2035 
                if materialized.is_conflicted() {
2036 
                    return None;
2037 
                }
2038 
                let disk_text = if attrs.lookup(p).eol.is_some() {
2039 
                    std::sync::Arc::new(crate::attributes::normalize_eol(&e.content))
2040 
                } else {
2041 
                    e.content.clone()
2042 
                };
2043 
                if *bytes == *disk_text {
2044 
                    return None;
2045 
                }
2046 
                let keys = materialized.live_atoms();
2047 
                let lines = crate::diff::split_lines(bytes);
2048 
                (keys.len() == lines.len()).then(|| {
2049 
                    let edits = crate::diff::text_edits_of(x.node, &keys, &lines, &disk_text);
2050 
                    ((*p).clone(), edits)
2051 
                })
2052 
            })
2053 
            .collect();
2054 
        self.diffs = diffs.into_iter().collect();
2055 
    }
2056
cdb5beba Andy McCall ↶ 2057 
    fn with_attributes(mut self, attrs: crate::attributes::Attributes) -> Self {
2058 
        self.attrs = attrs;
2059 
        self
2060 
    }
2061
2062 
    /// Whether a file is opaque: forced by `binary`, forced text by `text`
2063 
    /// only when it has no NUL bytes, else sniffed.
b7f102fc Andy McCall ↶ 2064 
    /// The blob for an opaque entry: the cached one, else stored now.
75167672 Andy McCall 2065 
    fn blob_of(
2066 
        &mut self,
2067 
        repo: &Repo,
2068 
        path: &str,
2069 
        e: &FsEntry,
2070 
        nocompress: bool,
2071 
    ) -> Result<ObjectRef> {
b7f102fc Andy McCall ↶ 2072 
        let b = match e.blob {
2073 
            Some(b) => b,
2074 
            None => repo.store_blob_for(&e.content, nocompress)?,
2075 
        };
2076 
        self.blobs.insert(path.to_string(), b);
2077 
        Ok(b)
2078 
    }
2079
cdb5beba Andy McCall ↶ 2080 
    fn opaque_for(&self, path: &str, e: &FsEntry) -> bool {
b7f102fc Andy McCall ↶ 2081 
        if e.blob.is_some() {
2082 
            return true;
2083 
        }
4dee277c Andy McCall ↶ 2084 
        if e.kind == NodeKind::Symlink || e.kind == NodeKind::Subrepo {
cdb5beba Andy McCall ↶ 2085 
            return true;
2086 
        }
2087 
        match self.attrs.lookup(path).binary {
2088 
            Some(true) => true,
2089 
            Some(false) => is_binary(&e.content),
2090 
            None => is_binary(&e.content),
2091 
        }
2092 
    }
2093
2094 
    /// Text content as stored: `\r\n` normalized for `eol=` files.
40a40bb7 Andy McCall ↶ 2095 
    fn stored_text(
2096 
        &self,
2097 
        path: &str,
2098 
        content: &std::sync::Arc<Vec<u8>>,
2099 
    ) -> std::sync::Arc<Vec<u8>> {
75167672 Andy McCall 2100 
        if self.attrs.lookup(path).eol.is_some() {
40a40bb7 Andy McCall ↶ 2101 
            std::sync::Arc::new(crate::attributes::normalize_eol(content))
75167672 Andy McCall 2102 
        } else {
40a40bb7 Andy McCall ↶ 2103 
            content.clone()
dfb23a3a Andy McCall ↶ 2104 
        }
2105 
    }
2106
2107 
    /// The identifier of a node added at `path`: reused when asked, derived
2108 
    /// from the seed when there is one, random otherwise.
2109 
    fn new_node(&self, path: &str, kind: NodeKind) -> NodeId {
2110 
        if let Some(n) = self.reused_node(path, kind) {
2111 
            return n;
75167672 Andy McCall 2112 
        }
dfb23a3a Andy McCall ↶ 2113 
        match &self.node_seed {
2114 
            Some(seed) => seeded_node(seed, path, kind),
2115 
            None => NodeId::random(),
2116 
        }
cdb5beba Andy McCall ↶ 2117 
    }
2118
98286406 Andy McCall ↶ 2119 
    fn reused_node(&self, path: &str, kind: NodeKind) -> Option<NodeId> {
75167672 Andy McCall 2120 
        self.reuse
2121 
            .get(path)
2122 
            .filter(|(_, k)| *k == kind)
2123 
            .map(|(n, _)| *n)
98286406 Andy McCall ↶ 2124 
    }
2125
a87c7b9d Andy McCall ↶ 2126 
    /// A dead directory last placed at (`parent`, `name`) in the base.
2127 
    fn revivable_dir(
2128 
        &mut self,
2129 
        parent: NodeId,
2130 
        name: &str,
2131 
    ) -> Option<(NodeId, crate::tree::OpKey)> {
2132 
        let base = self.base;
2133 
        self.dead_dirs
2134 
            .get_or_insert_with(|| base.tree().dead_dirs())
2135 
            .get(&(parent, name.to_string()))
2136 
            .copied()
2137 
    }
2138
98286406 Andy McCall ↶ 2139 
    fn ensure_dir(&mut self, path: &str) -> NodeId {
2140 
        if let Some(n) = self.dir_nodes.get(path) {
2141 
            return *n;
2142 
        }
2143 
        let mut parent = NodeId::ROOT;
2144 
        let mut prefix = String::new();
2145 
        for comp in path.split('/') {
2146 
            if !prefix.is_empty() {
2147 
                prefix.push('/');
2148 
            }
2149 
            prefix.push_str(comp);
a87c7b9d Andy McCall ↶ 2150 
            let known = self.dir_nodes.get(&prefix).copied();
2151 
            let revive = match known {
2152 
                Some(_) => None,
2153 
                None if self.node_seed.is_some() => self.revivable_dir(parent, comp),
2154 
                None => None,
2155 
            };
2156 
            parent = match (known, revive) {
2157 
                (Some(n), _) => n,
2158 
                (None, Some((node, old))) => {
2159 
                    // Git has no identity for a directory: one that comes
2160 
                    // back where it was is the same directory, brought back
2161 
                    // with a move, so lines of history that placed files in
2162 
                    // it before and after agree (oxygen#30).
2163 
                    self.tree_ops.push(TreeOp::Move {
2164 
                        node,
2165 
                        old: OpRef {
2166 
                            change: old.0,
2167 
                            index: old.1,
2168 
                        },
2169 
                        parent,
2170 
                        name: comp.to_string(),
2171 
                    });
2172 
                    self.dir_nodes.insert(prefix.clone(), node);
2173 
                    node
2174 
                }
2175 
                (None, None) => {
dfb23a3a Andy McCall ↶ 2176 
                    let node = self.new_node(&prefix, NodeKind::Dir);
75167672 Andy McCall 2177 
                    self.tree_ops.push(TreeOp::AddNode {
2178 
                        node,
2179 
                        kind: NodeKind::Dir,
2180 
                        mode: Mode::default(),
2181 
                        parent,
2182 
                        name: comp.to_string(),
2183 
                    });
98286406 Andy McCall ↶ 2184 
                    self.dir_nodes.insert(prefix.clone(), node);
2185 
                    node
2186 
                }
2187 
            };
2188 
        }
2189 
        parent
2190 
    }
2191
2192 
    fn add_path(&mut self, repo: &Repo, path: &str, e: &FsEntry) -> Result<()> {
2193 
        let (dir, name) = split_path(path);
2194 
        let parent = self.ensure_dir(dir);
dfb23a3a Andy McCall ↶ 2195 
        let node = self.new_node(path, e.kind);
cdb5beba Andy McCall ↶ 2196 
        let opaque = self.opaque_for(path, e);
2197 
        let nocompress = self.attrs.lookup(path).nocompress;
75167672 Andy McCall 2198 
        self.tree_ops.push(TreeOp::AddNode {
2199 
            node,
2200 
            kind: e.kind,
2201 
            mode: Mode {
2202 
                exec: e.exec,
2203 
                opaque,
2204 
            },
2205 
            parent,
2206 
            name: name.to_string(),
2207 
        });
98286406 Andy McCall ↶ 2208 
        if opaque {
b7f102fc Andy McCall ↶ 2209 
            let blob = self.blob_of(repo, path, e, nocompress)?;
75167672 Andy McCall 2210 
            self.edits.push(Edit::Replace {
2211 
                node,
2212 
                old: None,
2213 
                new: Some(blob),
2214 
            });
98286406 Andy McCall ↶ 2215 
        } else if !e.content.is_empty() {
cdb5beba Andy McCall ↶ 2216 
            let content = self.stored_text(path, &e.content);
98286406 Andy McCall ↶ 2217 
            self.edits.push(Edit::Insert {
2218 
                node,
2219 
                after: crate::objects::AtomRef::Start,
2220 
                before: crate::objects::AtomRef::End,
40a40bb7 Andy McCall ↶ 2221 
                lines: crate::lines::Lines::split_shared(content),
98286406 Andy McCall ↶ 2222 
            });
2223 
        }
75167672 Andy McCall 2224 
        self.statuses.push(PathStatus {
2225 
            path: path.to_string(),
2226 
            status: ChangeStatus::Added,
2227 
        });
98286406 Andy McCall ↶ 2228 
        Ok(())
2229 
    }
2230
6e723b81 Andy McCall ↶ 2231 
    /// Move the node at `old` to `new` and apply `e`'s content to it.
75167672 Andy McCall 2232 
    fn move_path(
2233 
        &mut self,
2234 
        repo: &Repo,
2235 
        old: &str,
2236 
        new: &str,
2237 
        e: &FsEntry,
2238 
        expected: &crate::state::FileState,
2239 
    ) -> Result<()> {
6e723b81 Andy McCall ↶ 2240 
        let entry = &self.base_snapshot.tree.entries[old];
2241 
        let (dir, name) = split_path(new);
2242 
        let parent = self.ensure_dir(dir);
75167672 Andy McCall 2243 
        self.tree_ops.push(TreeOp::Move {
2244 
            node: expected.node,
2245 
            old: op_ref(entry.placement),
2246 
            parent,
2247 
            name: name.to_string(),
2248 
        });
6e723b81 Andy McCall ↶ 2249 
        self.update_path(repo, old, e, expected)?;
2250 
        // update_path reports under the old path; the rename is what the user sees.
2251 
        self.statuses.retain(|s| s.path != old);
75167672 Andy McCall 2252 
        self.statuses.push(PathStatus {
2253 
            path: new.to_string(),
2254 
            status: ChangeStatus::Renamed,
2255 
        });
6e723b81 Andy McCall ↶ 2256 
        Ok(())
a87c7b9d Andy McCall ↶ 2257 
    }
2258
2259 
    /// For a Git import: a node the commit keeps is in one place, the one
2260 
    /// its tree names, so any other live placement it has in the base goes;
2261 
    /// and a node the tree cannot show at all (an orphan) goes too. Such a
2262 
    /// placement is out of sight, under a directory another line of history
2263 
    /// removed, where no diff of the tree would ever reach it (oxygen#30).
2264 
    fn retire_stray_placements(&mut self) {
2265 
        let mut retired: HashSet<(ChangeHash, u32)> = HashSet::new();
2266 
        let mut placed: BTreeSet<NodeId> = BTreeSet::new();
2267 
        for op in &self.tree_ops {
2268 
            match op {
2269 
                TreeOp::Move { node, old, .. } => {
2270 
                    retired.insert((old.change, old.index));
2271 
                    placed.insert(*node);
2272 
                }
2273 
                TreeOp::RemoveNode { placement, .. } => {
2274 
                    retired.insert((placement.change, placement.index));
2275 
                }
2276 
                TreeOp::AddNode { .. } | TreeOp::SetMode { .. } => {}
2277 
            }
2278 
        }
2279 
        // Nodes kept where they were: what the base's tree view shows and
2280 
        // this change neither moves nor removes.
2281 
        let mut kept: HashMap<NodeId, (ChangeHash, u32)> = HashMap::new();
2282 
        for e in self.base_snapshot.tree.entries.values() {
2283 
            if !retired.contains(&e.placement) {
2284 
                kept.insert(e.node, e.placement);
2285 
            }
2286 
        }
2287 
        // Only a node with several live placements can have a stray one,
2288 
        // and the tree view names each as a conflict.
2289 
        let multi: BTreeSet<NodeId> = self
2290 
            .base_snapshot
2291 
            .tree
2292 
            .conflicts
2293 
            .iter()
2294 
            .filter_map(|c| match c {
2295 
                crate::tree::TreeConflict::MultiPlacement { node, .. } => Some(*node),
2296 
                _ => None,
2297 
            })
2298 
            .collect();
2299 
        let mut extra = Vec::new();
2300 
        // A node whose only placements are under a directory another line
2301 
        // removed (an orphan) is nowhere the tree can show, and Git's tree
2302 
        // has no such path: it goes.
2303 
        for c in &self.base_snapshot.tree.conflicts {
2304 
            let crate::tree::TreeConflict::Orphan { node, .. } = c else {
2305 
                continue;
2306 
            };
2307 
            if kept.contains_key(node) || placed.contains(node) {
2308 
                continue;
2309 
            }
2310 
            for (k, _) in self.base.tree().live_placements(node) {
2311 
                if retired.insert(k) {
2312 
                    extra.push(TreeOp::RemoveNode {
2313 
                        node: *node,
2314 
                        placement: op_ref(k),
2315 
                    });
2316 
                }
2317 
            }
2318 
        }
2319 
        for node in placed
2320 
            .iter()
2321 
            .chain(multi.iter().filter(|n| kept.contains_key(*n)))
2322 
        {
2323 
            for (k, _) in self.base.tree().live_placements(node) {
2324 
                if retired.contains(&k) || kept.get(node) == Some(&k) {
2325 
                    continue;
2326 
                }
2327 
                retired.insert(k);
2328 
                extra.push(TreeOp::RemoveNode {
2329 
                    node: *node,
2330 
                    placement: op_ref(k),
2331 
                });
2332 
            }
2333 
        }
2334 
        if !extra.is_empty() {
2335 
            self.resolves = true;
2336 
            self.tree_ops.extend(extra);
2337 
        }
6e723b81 Andy McCall ↶ 2338 
    }
2339
98286406 Andy McCall ↶ 2340 
    fn remove_path(&mut self, path: &str, expected: &crate::state::FileState) {
2341 
        let entry = &self.base_snapshot.tree.entries[path];
75167672 Andy McCall 2342 
        self.tree_ops.push(TreeOp::RemoveNode {
2343 
            node: expected.node,
2344 
            placement: op_ref(entry.placement),
2345 
        });
98286406 Andy McCall ↶ 2346 
        if path.contains(".oxy-conflict-") || expected.is_conflicted() {
2347 
            self.resolves = true;
2348 
        }
75167672 Andy McCall 2349 
        self.statuses.push(PathStatus {
2350 
            path: path.to_string(),
2351 
            status: ChangeStatus::Removed,
2352 
        });
98286406 Andy McCall ↶ 2353 
    }
2354
75167672 Andy McCall 2355 
    fn update_path(
2356 
        &mut self,
2357 
        repo: &Repo,
2358 
        path: &str,
2359 
        e: &FsEntry,
2360 
        expected: &crate::state::FileState,
2361 
    ) -> Result<()> {
98286406 Andy McCall ↶ 2362 
        if expected.kind != e.kind {
2363 
            self.remove_path(path, expected);
2364 
            self.statuses.pop();
2365 
            self.add_path(repo, path, e)?;
2366 
            self.statuses.pop();
75167672 Andy McCall 2367 
            self.statuses.push(PathStatus {
2368 
                path: path.to_string(),
2369 
                status: ChangeStatus::Modified,
2370 
            });
98286406 Andy McCall ↶ 2371 
            return Ok(());
2372 
        }
2373 
        let entry = &self.base_snapshot.tree.entries[path];
2374 
        let node = expected.node;
2375 
        let mode_atom = entry.mode_atom.map(op_ref);
cc7a0056 Andy McCall ↶ 2376 
        if e.unchanged {
2377 
            // Known equal to the base's unconflicted rendering (spec
2378 
            // §12.3); only the executable bit can differ.
2379 
            if e.exec != expected.mode.exec {
2380 
                if let Some(old) = mode_atom {
2381 
                    self.tree_ops.push(TreeOp::SetMode {
2382 
                        node,
2383 
                        old,
2384 
                        mode: Mode {
2385 
                            exec: e.exec,
2386 
                            opaque: false,
2387 
                        },
2388 
                    });
2389 
                    self.statuses.push(PathStatus {
2390 
                        path: path.to_string(),
2391 
                        status: ChangeStatus::ModeChanged,
2392 
                    });
2393 
                }
2394 
            }
2395 
            return Ok(());
2396 
        }
98286406 Andy McCall ↶ 2397 
        let mut status: Option<ChangeStatus> = None;
cdb5beba Andy McCall ↶ 2398 
        let wants_opaque = self.opaque_for(path, e);
2399 
        let nocompress = self.attrs.lookup(path).nocompress;
2400 
        let disk_text = self.stored_text(path, &e.content);
98286406 Andy McCall ↶ 2401 
        match &expected.content {
2402 
            FileContent::Dir => {}
75167672 Andy McCall 2403 
            FileContent::Text {
2404 
                bytes,
2405 
                materialized,
2406 
            } => {
98286406 Andy McCall ↶ 2407 
                if materialized.is_conflicted() {
40a40bb7 Andy McCall ↶ 2408 
                    if bytes == &*disk_text {
98286406 Andy McCall ↶ 2409 
                        return Ok(());
2410 
                    }
2411 
                    if has_conflict_markers(&e.content) {
75167672 Andy McCall 2412 
                        self.statuses.push(PathStatus {
2413 
                            path: path.to_string(),
2414 
                            status: ChangeStatus::ConflictEdited,
2415 
                        });
98286406 Andy McCall ↶ 2416 
                        return Ok(());
2417 
                    }
2418 
                    // Resolution (spec §9.6): diff against the unconflicted lines
2419 
                    // with each region as a gap, then retire every region atom.
2420 
                    // Whatever the author kept from a region is re-inserted as
2421 
                    // new atoms, which leaves the graph totally ordered.
2422 
                    let regions = self.base.conflict_regions(&node);
cdb5beba Andy McCall ↶ 2423 
                    let mut edits = text_edits(node, &regions.plain, &disk_text);
98286406 Andy McCall ↶ 2424 
                    if !regions.region_atoms.is_empty() {
75167672 Andy McCall 2425 
                        edits.insert(
2426 
                            0,
2427 
                            Edit::Delete {
2428 
                                node,
2429 
                                atoms: ranges(&regions.region_atoms),
2430 
                            },
2431 
                        );
98286406 Andy McCall ↶ 2432 
                    }
2433 
                    self.edits.extend(edits);
2434 
                    self.contributors.extend(regions.contributors);
2435 
                    self.resolves = true;
2436 
                    if e.exec != expected.mode.exec {
2437 
                        if let Some(old) = mode_atom {
75167672 Andy McCall 2438 
                            self.tree_ops.push(TreeOp::SetMode {
2439 
                                node,
2440 
                                old,
2441 
                                mode: Mode {
2442 
                                    exec: e.exec,
2443 
                                    opaque: false,
2444 
                                },
2445 
                            });
98286406 Andy McCall ↶ 2446 
                        }
2447 
                    }
75167672 Andy McCall 2448 
                    self.statuses.push(PathStatus {
2449 
                        path: path.to_string(),
2450 
                        status: ChangeStatus::Resolved,
2451 
                    });
98286406 Andy McCall ↶ 2452 
                    return Ok(());
2453 
                }
cdb5beba Andy McCall ↶ 2454 
                if wants_opaque {
2455 
                    // Text → opaque (spec §7.5), by content or by attribute.
2456 
                    if let Some(old) = mode_atom {
75167672 Andy McCall 2457 
                        self.tree_ops.push(TreeOp::SetMode {
2458 
                            node,
2459 
                            old,
2460 
                            mode: Mode {
2461 
                                exec: e.exec,
2462 
                                opaque: true,
2463 
                            },
2464 
                        });
cdb5beba Andy McCall ↶ 2465 
                    }
2466 
                    let live: Vec<_> = repo_live_keys(self.base, &node);
2467 
                    if !live.is_empty() {
75167672 Andy McCall 2468 
                        self.edits.push(Edit::Delete {
2469 
                            node,
2470 
                            atoms: ranges(&live),
2471 
                        });
cdb5beba Andy McCall ↶ 2472 
                    }
b7f102fc Andy McCall ↶ 2473 
                    let blob = self.blob_of(repo, path, e, nocompress)?;
75167672 Andy McCall 2474 
                    self.edits.push(Edit::Replace {
2475 
                        node,
2476 
                        old: None,
2477 
                        new: Some(blob),
2478 
                    });
cdb5beba Andy McCall ↶ 2479 
                    status = Some(ChangeStatus::Modified);
40a40bb7 Andy McCall ↶ 2480 
                } else if bytes != &*disk_text {
2306a4d2 Andy McCall ↶ 2481 
                    let mut laps = crate::timing::Laps::start();
2482 
                    // The base's rendering holds its live atoms in order,
2483 
                    // one per line (it is not conflicted here).
360aa74a Andy McCall ↶ 2484 
                    let edits = match self.diffs.remove(path) {
2485 
                        Some(edits) => edits,
2486 
                        None => {
2487 
                            let keys = materialized.live_atoms();
2488 
                            let lines = crate::diff::split_lines(bytes);
2489 
                            if keys.len() == lines.len() {
2490 
                                crate::diff::text_edits_of(node, &keys, &lines, &disk_text)
2491 
                            } else {
2492 
                                text_edits(node, &self.base.live_lines(&node), &disk_text)
2493 
                            }
2494 
                        }
2306a4d2 Andy McCall ↶ 2495 
                    };
2496 
                    laps.lap("build: text diff");
cdb5beba Andy McCall ↶ 2497 
                    if !edits.is_empty() {
2498 
                        self.edits.extend(edits);
98286406 Andy McCall ↶ 2499 
                        status = Some(ChangeStatus::Modified);
2500 
                    }
2501 
                }
cdb5beba Andy McCall ↶ 2502 
                let switched_to_opaque = wants_opaque;
98286406 Andy McCall ↶ 2503 
                if !switched_to_opaque && e.exec != expected.mode.exec {
2504 
                    if let Some(old) = mode_atom {
75167672 Andy McCall 2505 
                        self.tree_ops.push(TreeOp::SetMode {
2506 
                            node,
2507 
                            old,
2508 
                            mode: Mode {
2509 
                                exec: e.exec,
2510 
                                opaque: false,
2511 
                            },
2512 
                        });
98286406 Andy McCall ↶ 2513 
                        status.get_or_insert(ChangeStatus::ModeChanged);
2514 
                    }
2515 
                }
2516 
            }
2517 
            FileContent::Opaque { blobs } => {
b7f102fc Andy McCall ↶ 2518 
                let new_ref = e.blob.unwrap_or_else(|| Repo::blob_ref(&e.content));
2519 
                if e.kind == NodeKind::File && wants_opaque {
2520 
                    // Unchanged or replaced, the file stays opaque: remember its blob.
2521 
                    self.blobs.insert(path.to_string(), new_ref);
f7510b63 Andy McCall ↶ 2522 
                }
2523 
                if blobs.len() > 1 && self.resolve.contains(path) && e.kind == expected.kind {
2524 
                    // Settle the replace conflict (spec §9.3): replace every
2525 
                    // live candidate with the chosen blob, depending on each
2526 
                    // change that made one live, so all are superseded.
2527 
                    let chosen = self.blob_of(repo, path, e, nocompress)?;
2528 
                    for b in blobs {
2529 
                        self.edits.push(Edit::Replace {
2530 
                            node,
2531 
                            old: Some(*b),
2532 
                            new: Some(chosen),
2533 
                        });
2534 
                    }
2535 
                    self.contributors
2536 
                        .extend(self.base.blob_contributors(&node, blobs));
2537 
                    self.resolves = true;
2538 
                    self.statuses.push(PathStatus {
2539 
                        path: path.to_string(),
2540 
                        status: ChangeStatus::Resolved,
2541 
                    });
2542 
                    return Ok(());
b7f102fc Andy McCall ↶ 2543 
                }
98286406 Andy McCall ↶ 2544 
                if blobs.len() > 1 {
2545 
                    if !blobs.contains(&new_ref) {
75167672 Andy McCall 2546 
                        self.statuses.push(PathStatus {
2547 
                            path: path.to_string(),
2548 
                            status: ChangeStatus::ConflictEdited,
2549 
                        });
98286406 Andy McCall ↶ 2550 
                    }
2551 
                    return Ok(());
2552 
                }
2553 
                let current = blobs.first().copied();
2554 
                if current != Some(new_ref) {
cdb5beba Andy McCall ↶ 2555 
                    if e.kind == NodeKind::File && !wants_opaque {
98286406 Andy McCall ↶ 2556 
                        // Opaque → text (spec §7.5).
2557 
                        if let Some(old) = mode_atom {
75167672 Andy McCall 2558 
                            self.tree_ops.push(TreeOp::SetMode {
2559 
                                node,
2560 
                                old,
2561 
                                mode: Mode {
2562 
                                    exec: e.exec,
2563 
                                    opaque: false,
2564 
                                },
2565 
                            });
98286406 Andy McCall ↶ 2566 
                        }
75167672 Andy McCall 2567 
                        self.edits.push(Edit::Replace {
2568 
                            node,
2569 
                            old: current,
2570 
                            new: None,
2571 
                        });
98286406 Andy McCall ↶ 2572 
                        if !e.content.is_empty() {
2573 
                            self.edits.push(Edit::Insert {
2574 
                                node,
2575 
                                after: crate::objects::AtomRef::Start,
2576 
                                before: crate::objects::AtomRef::End,
01a47cd7 Andy McCall ↶ 2577 
                                lines: crate::lines::Lines::split(&e.content),
98286406 Andy McCall ↶ 2578 
                            });
2579 
                        }
2580 
                    } else {
b7f102fc Andy McCall ↶ 2581 
                        let blob = self.blob_of(repo, path, e, nocompress)?;
75167672 Andy McCall 2582 
                        self.edits.push(Edit::Replace {
2583 
                            node,
2584 
                            old: current,
2585 
                            new: Some(blob),
2586 
                        });
98286406 Andy McCall ↶ 2587 
                    }
2588 
                    status = Some(ChangeStatus::Modified);
2589 
                } else if e.kind == NodeKind::File && e.exec != expected.mode.exec {
2590 
                    if let Some(old) = mode_atom {
75167672 Andy McCall 2591 
                        self.tree_ops.push(TreeOp::SetMode {
2592 
                            node,
2593 
                            old,
2594 
                            mode: Mode {
2595 
                                exec: e.exec,
2596 
                                opaque: true,
2597 
                            },
2598 
                        });
98286406 Andy McCall ↶ 2599 
                        status = Some(ChangeStatus::ModeChanged);
2600 
                    }
2601 
                }
2602 
            }
2603 
        }
2604 
        if let Some(s) = status {
75167672 Andy McCall 2605 
            self.statuses.push(PathStatus {
2606 
                path: path.to_string(),
2607 
                status: s,
2608 
            });
98286406 Andy McCall ↶ 2609 
        }
2610 
        Ok(())
2611 
    }
2612 
}
2613
2614 
fn repo_live_keys(base: &State, node: &NodeId) -> Vec<crate::graph::AtomKey> {
2615 
    base.live_lines(node).into_iter().map(|(k, _)| k).collect()
2616 
}
2617
2618 
/// The changes a change's scope references (spec §8.1): introducers of every
2619 
/// atom, placement, and mode it touches, and of every node it operates on.
2620 
pub fn referenced_changes(change: &Change, base: &State) -> BTreeSet<ChangeHash> {
2621 
    let mut out = BTreeSet::new();
2622 
    let opref = |r: &OpRef, out: &mut BTreeSet<ChangeHash>| {
2623 
        if !r.change.is_zero() {
2624 
            out.insert(r.change);
2625 
        }
2626 
    };
2627 
    let mut nodes: Vec<NodeId> = Vec::new();
2628 
    for op in &change.tree_ops {
2629 
        match op {
2630 
            TreeOp::AddNode { parent, .. } => nodes.push(*parent),
2631 
            TreeOp::RemoveNode { node, placement } => {
2632 
                nodes.push(*node);
2633 
                opref(placement, &mut out);
2634 
            }
75167672 Andy McCall 2635 
            TreeOp::Move {
2636 
                node, old, parent, ..
2637 
            } => {
98286406 Andy McCall ↶ 2638 
                nodes.push(*node);
2639 
                nodes.push(*parent);
2640 
                opref(old, &mut out);
2641 
            }
2642 
            TreeOp::SetMode { node, old, .. } => {
2643 
                nodes.push(*node);
2644 
                opref(old, &mut out);
2645 
            }
2646 
        }
2647 
    }
2648 
    for e in &change.edits {
2649 
        nodes.push(e.node());
2650 
        match e {
2651 
            Edit::Insert { after, before, .. } => {
2652 
                for r in [after, before] {
2653 
                    if let crate::objects::AtomRef::Atom { change, .. } = r {
2654 
                        if !change.is_zero() {
2655 
                            out.insert(*change);
2656 
                        }
2657 
                    }
2658 
                }
2659 
            }
2660 
            Edit::Delete { atoms, .. } => {
62794ecc Andy McCall ↶ 2661 
                // Deleting an atom also depends on whoever inserted the live
2662 
                // atoms right after it: the author saw those insertions, and
2663 
                // without the dependency spec §9.1 would call the deletion
2664 
                // concurrent with them (a delete/insert conflict).
2665 
                let graph = base.file(&e.node());
98286406 Andy McCall ↶ 2666 
                for r in atoms {
2667 
                    if !r.change.is_zero() {
2668 
                        out.insert(r.change);
2669 
                    }
62794ecc Andy McCall ↶ 2670 
                    if let Some(g) = graph {
2671 
                        for i in r.from..r.to {
2672 
                            out.extend(
2673 
                                g.live_successor_changes(&(r.change, i))
2674 
                                    .into_iter()
2675 
                                    .filter(|c| !c.is_zero()),
2676 
                            );
2677 
                        }
2678 
                    }
98286406 Andy McCall ↶ 2679 
                }
2680 
            }
62794ecc Andy McCall ↶ 2681 
            // Replacing a blob depends on the change that made it current,
2682 
            // as deleting a line depends on the line's introducer.
2683 
            Edit::Replace {
2684 
                node,
2685 
                old: Some(old),
2686 
                ..
2687 
            } => {
2688 
                out.extend(
2689 
                    base.live_replaces(node)
2690 
                        .into_iter()
2691 
                        .filter(|(c, b)| b == old && !c.is_zero())
2692 
                        .map(|(c, _)| c),
2693 
                );
2694 
            }
98286406 Andy McCall ↶ 2695 
            Edit::Replace { .. } => {}
2696 
        }
2697 
    }
2698 
    for n in nodes {
2699 
        if let Some(h) = base.tree().introducer(&n) {
2700 
            out.insert(h);
2701 
        }
2702 
    }
2703 
    out
2704 
}
2705
2706 
/// Paths of the nodes a change adds on top of `base_snapshot`.
75167672 Andy McCall 2707 
pub fn added_node_paths(
2708 
    change: &Change,
2709 
    base_snapshot: &Snapshot,
2710 
) -> HashMap<String, (NodeId, NodeKind)> {
98286406 Andy McCall ↶ 2711 
    let by_node = base_snapshot.tree.by_node();
2712 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
2713 
    for op in &change.tree_ops {
75167672 Andy McCall 2714 
        if let TreeOp::AddNode {
2715 
            node,
2716 
            parent,
2717 
            name,
2718 
            kind,
2719 
            ..
2720 
        } = op
2721 
        {
98286406 Andy McCall ↶ 2722 
            new_names.insert(*node, (*parent, name.clone(), *kind));
2723 
        }
2724 
    }
2725 
    let mut out = HashMap::new();
2726 
    for (node, (_, _, kind)) in &new_names {
2727 
        if let Some(path) = path_of_node(node, &by_node, &new_names) {
2728 
            out.insert(path, (*node, *kind));
2729 
        }
2730 
    }
2731 
    out
2732 
}
2733
2734 
fn path_of_node(
2735 
    n: &NodeId,
2736 
    by_node: &HashMap<NodeId, &crate::tree::TreeEntry>,
2737 
    new_names: &HashMap<NodeId, (NodeId, String, NodeKind)>,
2738 
) -> Option<String> {
2739 
    if let Some(e) = by_node.get(n) {
2740 
        return Some(e.path.clone());
2741 
    }
2742 
    let mut parts = Vec::new();
2743 
    let mut cur = *n;
2744 
    loop {
2745 
        if cur.is_root() {
2746 
            break;
2747 
        }
2748 
        if let Some(e) = by_node.get(&cur) {
2749 
            if !e.path.is_empty() {
2750 
                parts.push(e.path.clone());
2751 
            }
2752 
            break;
2753 
        }
2754 
        let (parent, name, _) = new_names.get(&cur)?;
2755 
        parts.push(name.clone());
2756 
        cur = *parent;
2757 
    }
2758 
    parts.reverse();
2759 
    Some(parts.join("/"))
2760 
}
2761
2762 
/// Render a change's operations with line content, using `base` for deleted lines.
2763 
pub fn render_change(change: &Change, base: &State, base_snapshot: &Snapshot) -> String {
2764 
    let by_node = base_snapshot.tree.by_node();
2765 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
2766 
    for op in &change.tree_ops {
75167672 Andy McCall 2767 
        if let TreeOp::AddNode {
2768 
            node,
2769 
            parent,
2770 
            name,
2771 
            kind,
2772 
            ..
2773 
        } = op
2774 
        {
98286406 Andy McCall ↶ 2775 
            new_names.insert(*node, (*parent, name.clone(), *kind));
2776 
        }
2777 
    }
75167672 Andy McCall 2778 
    let path_of = |n: &NodeId| -> String {
2779 
        path_of_node(n, &by_node, &new_names).unwrap_or_else(|| format!("<node {}>", n.short(8)))
2780 
    };
98286406 Andy McCall ↶ 2781
2782 
    let mut out = String::new();
2783 
    for op in &change.tree_ops {
2784 
        match op {
75167672 Andy McCall 2785 
            TreeOp::AddNode {
2786 
                node, kind, mode, ..
2787 
            } => {
2788 
                out.push_str(&format!(
2789 
                    "A {} ({:?}{}{})\n",
2790 
                    path_of(node),
2791 
                    kind,
2792 
                    if mode.exec { ", exec" } else { "" },
2793 
                    if mode.opaque { ", opaque" } else { "" }
2794 
                ));
98286406 Andy McCall ↶ 2795 
            }
2796 
            TreeOp::RemoveNode { node, .. } => out.push_str(&format!("D {}\n", path_of(node))),
75167672 Andy McCall 2797 
            TreeOp::Move { node, name, .. } => {
2798 
                out.push_str(&format!("R {} -> {}\n", path_of(node), name))
2799 
            }
98286406 Andy McCall ↶ 2800 
            TreeOp::SetMode { node, mode, .. } => {
75167672 Andy McCall 2801 
                out.push_str(&format!(
2802 
                    "M {} mode: exec={} opaque={}\n",
2803 
                    path_of(node),
2804 
                    mode.exec,
2805 
                    mode.opaque
2806 
                ));
98286406 Andy McCall ↶ 2807 
            }
2808 
        }
2809 
    }
2810 
    let mut last_node: Option<NodeId> = None;
2811 
    for e in &change.edits {
2812 
        let n = e.node();
2813 
        if last_node != Some(n) {
2814 
            out.push_str(&format!("--- {}\n", path_of(&n)));
2815 
            last_node = Some(n);
2816 
        }
2817 
        match e {
2818 
            Edit::Insert { lines, .. } => {
01a47cd7 Andy McCall ↶ 2819 
                for l in lines.iter() {
98286406 Andy McCall ↶ 2820 
                    out.push('+');
2821 
                    out.push_str(&String::from_utf8_lossy(l));
2822 
                    if !l.ends_with(b"\n") {
2823 
                        out.push('\n');
2824 
                    }
2825 
                }
2826 
            }
2827 
            Edit::Delete { atoms, .. } => {
2828 
                for r in atoms {
2829 
                    for i in r.from..r.to {
75167672 Andy McCall 2830 
                        let content = base
2831 
                            .file(&n)
2832 
                            .and_then(|g| g.content(&(r.change, i)))
2833 
                            .unwrap_or(b"<unknown>\n");
98286406 Andy McCall ↶ 2834 
                        out.push('-');
2835 
                        out.push_str(&String::from_utf8_lossy(content));
2836 
                        if !content.ends_with(b"\n") {
2837 
                            out.push('\n');
2838 
                        }
2839 
                    }
2840 
                }
2841 
            }
2842 
            Edit::Replace { old, new, .. } => {
2843 
                out.push_str(&format!(
2844 
                    "  opaque content {} -> {}\n",
2845 
                    old.map(|o| o.short(12)).unwrap_or_else(|| "none".into()),
2846 
                    new.map(|o| o.short(12)).unwrap_or_else(|| "none".into())
2847 
                ));
2848 
            }
2849 
        }
2850 
    }
2851 
    out
2852 
}
6e723b81 Andy McCall ↶ 2853
2854 
/// Pair removed files with added files of the same kind whose contents are
2855 
/// at least `RENAME_THRESHOLD` percent alike, best matches first. Beyond
2856 
/// `RENAME_LIMIT` candidate pairs only identical contents are paired.
2857 
const RENAME_THRESHOLD: u32 = 50;
2858 
const RENAME_LIMIT: usize = 10_000;
2306a4d2 Andy McCall ↶ 2859
23ff1288 Andy McCall ↶ 2860 
/// How a change is built beyond its entries: for a Git import, the seed
2861 
/// its new nodes' identifiers derive from; and, when it is built from part
2862 
/// of the tree, whether the new tree has anything under a directory.
2863 
#[derive(Default)]
2864 
struct Scope<'a> {
2865 
    seed: Option<[u8; 32]>,
2306a4d2 Andy McCall ↶ 2866 
    dir_needed: Option<&'a dyn Fn(&str) -> bool>,
2867 
}
6e723b81 Andy McCall ↶ 2868
75167672 Andy McCall 2869 
fn detect_renames(
2870 
    base_snapshot: &Snapshot,
2871 
    fs_entries: &BTreeMap<String, FsEntry>,
2872 
    in_scope: impl Fn(&str) -> bool,
2873 
) -> Vec<(String, String)> {
6e723b81 Andy McCall ↶ 2874 
    let removed: Vec<(&String, &crate::state::FileState)> = base_snapshot
2875 
        .files
2876 
        .iter()
75167672 Andy McCall 2877 
        .filter(|(p, f)| {
2878 
            f.kind != NodeKind::Dir
2879 
                && !fs_entries.contains_key(*p)
2880 
                && in_scope(p)
2881 
                && !f.is_conflicted()
2882 
        })
2883 
        .collect();
2884 
    let added: Vec<(&String, &FsEntry)> = fs_entries
2885 
        .iter()
2886 
        .filter(|(p, e)| e.kind != NodeKind::Dir && !base_snapshot.files.contains_key(*p))
6e723b81 Andy McCall ↶ 2887 
        .collect();
2888 
    if removed.is_empty() || added.is_empty() {
2889 
        return vec![];
2890 
    }
2891 
    let exact_only = removed.len() * added.len() > RENAME_LIMIT;
2892 
    let old_content = |f: &crate::state::FileState| -> Option<Vec<u8>> {
2893 
        match &f.content {
2894 
            FileContent::Text { bytes, .. } => Some(bytes.clone()),
2895 
            _ => None,
2896 
        }
2897 
    };
2898 
    let mut scored: Vec<(u32, usize, usize)> = Vec::new();
2899 
    for (ri, (_, f)) in removed.iter().enumerate() {
2900 
        let old = old_content(f);
2901 
        let old_blob = match &f.content {
2902 
            FileContent::Opaque { blobs } if blobs.len() == 1 => Some(blobs[0]),
2903 
            _ => None,
2904 
        };
2905 
        for (ai, (_, e)) in added.iter().enumerate() {
2906 
            if e.kind != f.kind {
2907 
                continue;
2908 
            }
2909 
            let score = match (&old, old_blob) {
2910 
                (Some(o), _) => {
2911 
                    if exact_only {
40a40bb7 Andy McCall ↶ 2912 
                        if *o == *e.content {
75167672 Andy McCall 2913 
                            100
2914 
                        } else {
2915 
                            0
2916 
                        }
6e723b81 Andy McCall ↶ 2917 
                    } else {
2918 
                        similarity(o, &e.content)
2919 
                    }
2920 
                }
b7f102fc Andy McCall ↶ 2921 
                (None, Some(b)) if e.blob.unwrap_or_else(|| Repo::blob_ref(&e.content)) == b => 100,
6e723b81 Andy McCall ↶ 2922 
                _ => 0,
2923 
            };
2924 
            if score >= RENAME_THRESHOLD {
2925 
                scored.push((score, ri, ai));
2926 
            }
2927 
        }
2928 
    }
75167672 Andy McCall 2929 
    scored.sort_by(|a, b| {
2930 
        b.0.cmp(&a.0)
2931 
            .then(removed[a.1].0.cmp(removed[b.1].0))
2932 
            .then(added[a.2].0.cmp(added[b.2].0))
2933 
    });
6e723b81 Andy McCall ↶ 2934 
    let mut used_r = vec![false; removed.len()];
2935 
    let mut used_a = vec![false; added.len()];
2936 
    let mut out = Vec::new();
2937 
    for (_, ri, ai) in scored {
2938 
        if used_r[ri] || used_a[ai] {
2939 
            continue;
2940 
        }
2941 
        used_r[ri] = true;
2942 
        used_a[ai] = true;
2943 
        out.push((removed[ri].0.clone(), added[ai].0.clone()));
2944 
    }
2945 
    out
2946 
}
2947
2948 
/// Percentage of the larger file's bytes found in lines the two share
2949 
/// (multiset intersection of lines), like Git's similarity index.
2950 
pub fn similarity(a: &[u8], b: &[u8]) -> u32 {
2951 
    if a == b {
2952 
        return 100;
2953 
    }
2954 
    if a.is_empty() || b.is_empty() {
2955 
        return 0;
2956 
    }
2957 
    let mut counts: HashMap<&[u8], usize> = HashMap::new();
2958 
    for l in split_lines(a) {
2959 
        *counts.entry(l).or_default() += 1;
2960 
    }
2961 
    let mut shared = 0usize;
2962 
    for l in split_lines(b) {
2963 
        if let Some(n) = counts.get_mut(l) {
2964 
            if *n > 0 {
2965 
                *n -= 1;
2966 
                shared += l.len();
2967 
            }
2968 
        }
2969 
    }
2970 
    (shared * 100 / a.len().max(b.len())) as u32
d616faf5 Andy McCall ↶ 2971 
}
2972
2973 
/// Two maps side by side in key order: each key of either, with its value
2974 
/// in each. Linear, with no lookups, for maps of a whole tree.
2975 
fn join<'a, A, B>(
2976 
    a: &'a BTreeMap<String, A>,
2977 
    b: &'a BTreeMap<String, B>,
2978 
) -> impl Iterator<Item = (&'a String, Option<&'a A>, Option<&'a B>)> {
2979 
    let (mut a, mut b) = (a.iter().peekable(), b.iter().peekable());
2980 
    std::iter::from_fn(move || match (a.peek(), b.peek()) {
2981 
        (None, None) => None,
2982 
        (Some(_), None) => a.next().map(|(k, v)| (k, Some(v), None)),
2983 
        (None, Some(_)) => b.next().map(|(k, v)| (k, None, Some(v))),
2984 
        (Some((ka, _)), Some((kb, _))) => match ka.cmp(kb) {
2985 
            std::cmp::Ordering::Less => a.next().map(|(k, v)| (k, Some(v), None)),
2986 
            std::cmp::Ordering::Greater => b.next().map(|(k, v)| (k, None, Some(v))),
2987 
            std::cmp::Ordering::Equal => {
2988 
                let (k, va) = a.next()?;
2989 
                let (_, vb) = b.next()?;
2990 
                Some((k, Some(va), Some(vb)))
2991 
            }
2992 
        },
2993 
    })
2994 
}
2995
2996 
#[cfg(test)]
2997 
mod join_tests {
2998 
    use super::*;
2999
3000 
    #[test]
3001 
    fn join_pairs_every_key_in_order() {
3002 
        let a: BTreeMap<String, u8> = [("a", 1), ("b/c", 2), ("d", 3)]
3003 
            .into_iter()
3004 
            .map(|(k, v)| (k.to_string(), v))
3005 
            .collect();
3006 
        let b: BTreeMap<String, char> = [("b", 'x'), ("b/c", 'y'), ("e", 'z')]
3007 
            .into_iter()
3008 
            .map(|(k, v)| (k.to_string(), v))
3009 
            .collect();
3010 
        let got: Vec<(String, Option<u8>, Option<char>)> = join(&a, &b)
3011 
            .map(|(k, x, y)| (k.clone(), x.copied(), y.copied()))
3012 
            .collect();
3013 
        let want = vec![
3014 
            ("a".to_string(), Some(1), None),
3015 
            ("b".to_string(), None, Some('x')),
3016 
            ("b/c".to_string(), Some(2), Some('y')),
3017 
            ("d".to_string(), Some(3), None),
3018 
            ("e".to_string(), None, Some('z')),
3019 
        ];
3020 
        assert_eq!(got, want);
3021 
    }
6e723b81 Andy McCall ↶ 3022 
}