OXYGENHUB
◆ push-notes oxygen / crates / oxy-core / src / repo / snapshot.rs
137.1 KiB · Rust · last changed Repo::union_resolution: a host settles a state whose conflicts are all union (refs #145) 9 days ago source blame history raw
98286406 Andy McCall ↶ 1 
//! Working-copy snapshots, commits, and the change builder (spec §12.3).
2
3 
use super::*;
957fee30 Andy McCall ↶ 4 
use crate::objects::AtomRef;
98286406 Andy McCall ↶ 5
6 
#[derive(Clone, Copy, Debug, PartialEq, Eq)]
7 
pub enum ChangeStatus {
8 
    Added,
9 
    Modified,
10 
    Removed,
6e723b81 Andy McCall ↶ 11 
    /// Moved from another path (detected by similarity, tracked as a placement change).
12 
    Renamed,
98286406 Andy McCall ↶ 13 
    ModeChanged,
14 
    /// The file is conflicted in the base state and still contains conflict markers.
15 
    ConflictEdited,
16 
    /// The file was conflicted in the base state and the working copy resolves it.
17 
    Resolved,
18 
}
19
20 
/// Lines that mark a conflict region in a materialized file (spec §9.5).
21 
pub fn has_conflict_markers(content: &[u8]) -> bool {
22 
    split_lines(content).iter().any(|l| {
75167672 Andy McCall 23 
        l.starts_with(b"<<<<<<< oxy conflict")
24 
            || l.starts_with(b"======= oxy conflict")
25 
            || l.starts_with(b">>>>>>> oxy conflict")
98286406 Andy McCall ↶ 26 
    })
27 
}
28
3652104a Andy McCall ↶ 29 
/// Whether `content` still holds a conflict marker oxy wrote for this file
30 
/// (oxygen#126): a marker-like line of the materialized conflicted file
31 
/// that is not one of its unconflicted lines (`plain`). Text that merely
32 
/// looks like a marker, as oxy's own documentation or a test fixture has,
33 
/// is in the unconflicted lines, so it is not taken for one.
34 
pub fn has_own_conflict_markers(content: &[u8], materialized: &[u8], plain: &[&[u8]]) -> bool {
35 
    let line = |l: &[u8]| -> Vec<u8> {
36 
        let l = l.strip_suffix(b"\n").unwrap_or(l);
37 
        l.strip_suffix(b"\r").unwrap_or(l).to_vec()
38 
    };
39 
    let marker = |l: &[u8]| {
40 
        l.starts_with(b"<<<<<<< oxy conflict")
41 
            || l.starts_with(b"======= oxy conflict")
42 
            || l.starts_with(b">>>>>>> oxy conflict")
43 
    };
44 
    let unconflicted: HashSet<Vec<u8>> = plain.iter().map(|l| line(l)).collect();
45 
    let own: HashSet<Vec<u8>> = split_lines(materialized)
46 
        .into_iter()
47 
        .map(line)
48 
        .filter(|l| marker(l) && !unconflicted.contains(l))
49 
        .collect();
50 
    !own.is_empty()
51 
        && split_lines(content)
52 
            .into_iter()
53 
            .any(|l| own.contains(&line(l)))
54 
}
55
98286406 Andy McCall ↶ 56 
#[derive(Clone, Debug, PartialEq, Eq)]
57 
pub struct PathStatus {
58 
    pub path: String,
59 
    pub status: ChangeStatus,
60 
}
61
614541c3 Andy McCall ↶ 62 
/// Operations of a change built from the working directory against a base.
63 
pub(crate) struct BuiltChange {
64 
    pub tree_ops: Vec<TreeOp>,
65 
    pub edits: Vec<Edit>,
66 
    pub contributors: Vec<ChangeHash>,
67 
    pub resolves: bool,
68 
    pub statuses: Vec<PathStatus>,
b7f102fc Andy McCall ↶ 69 
    /// The stored blob of every opaque path, for the manifest's stat cache.
70 
    pub blobs: HashMap<String, ObjectRef>,
614541c3 Andy McCall ↶ 71 
}
72
98286406 Andy McCall ↶ 73 
/// Result of snapshotting the working copy.
74 
pub struct SnapshotResult {
5b4c5130 Andy McCall ↶ 75 
    /// The working-copy change's hash; for a commit's own snapshot, which
76 
    /// does not encode it, the hash it had before.
98286406 Andy McCall ↶ 77 
    pub wc_hash: ChangeHash,
78 
    pub wc: Change,
79 
    /// The base state (closure of the working-copy change's dependencies).
80 
    pub base: State,
81 
    pub base_snapshot: Snapshot,
82 
    pub statuses: Vec<PathStatus>,
83 
    /// Whether the snapshot rewrote the working-copy change.
84 
    pub changed: bool,
85 
    /// Whether the working copy resolves conflicts of its base (spec §9.6).
86 
    pub resolves: bool,
87 
    /// Changes that contributed to the conflicts being resolved.
88 
    pub contributors: Vec<ChangeHash>,
f9466021 Andy McCall ↶ 89 
    /// The `.oxyattributes` in force on disk, from the files the walk read.
90 
    pub(crate) attributes: crate::attributes::Attributes,
d4dbab97 Andy McCall ↶ 91 
    /// Folders holding a repository of their own (`.git` or `.oxy`) that
92 
    /// are not subrepos, so nothing in them is tracked (oxygen#43).
93 
    pub untracked_nested: Vec<String>,
236f35c0 Andy McCall ↶ 94 
    /// A commit's snapshot leaves the manifest to the commit, which writes
95 
    /// it once with the new base's graph keys.
96 
    pub(crate) manifest: Option<PendingManifest>,
98286406 Andy McCall ↶ 97 
}
98
236f35c0 Andy McCall ↶ 99 
/// The manifest a commit's snapshot would have written, as of the moment
100 
/// it would have written it (that moment is the racy-window stamp).
101 
pub(crate) struct PendingManifest {
102 
    entries: BTreeMap<String, FsEntry>,
103 
    blobs: HashMap<String, ObjectRef>,
104 
    dirs: Option<super::WalkedDirs>,
105 
    written: u128,
106 
}
107
157d02e9 Andy McCall ↶ 108 
/// Whether `base` has two or more live versions of the opaque file or
109 
/// subrepo pin at `path` (a replace conflict, spec §9.3).
110 
fn has_replace_conflict(base: &Snapshot, path: &str) -> bool {
111 
    matches!(base.files.get(path).map(|f| &f.content), Some(FileContent::Opaque { blobs }) if blobs.len() > 1)
112 
}
113
9bd98cfb Andy McCall ↶ 114 
/// What `oxy resolve --both` did to one file (oxygen#121).
115 
#[derive(Debug, Default, Clone, PartialEq, Eq)]
116 
pub struct KeptBoth {
117 
    pub path: String,
118 
    /// The conflicts settled by keeping every side, by number.
119 
    pub settled: Vec<usize>,
120 
    /// The conflicts that keep their markers, by number, and why.
121 
    pub left: Vec<(usize, &'static str)>,
122 
}
123
124 
/// `text` (a working-copy file) with the marker block of each conflict that
125 
/// `sides` (from [`State::both_sides`]) says keeping both settles replaced
126 
/// by its sides' lines; the conflicts settled, and those left and why.
127 
pub(crate) fn keep_both_text(
128 
    text: &[u8],
129 
    sides: &[Option<Vec<Vec<u8>>>],
130 
    reverse: bool,
131 
) -> (Vec<u8>, Vec<usize>, Vec<(usize, &'static str)>) {
132 
    // A side as the markers hold it: the renderer ends each with a newline.
133 
    let with_newline = |s: &[u8]| {
134 
        let mut s = s.to_vec();
135 
        if !s.is_empty() && !s.ends_with(b"\n") {
136 
            s.push(b'\n');
137 
        }
138 
        s
139 
    };
140 
    let lines: Vec<&[u8]> = text.split_inclusive(|&b| b == b'\n').collect();
141 
    let (mut out, mut settled, mut left) = (Vec::new(), Vec::new(), Vec::new());
142 
    let mut i = 0;
143 
    while i < lines.len() {
144 
        let Some((n, total)) = conflict_header(lines[i]) else {
145 
            out.extend_from_slice(lines[i]);
146 
            i += 1;
147 
            continue;
148 
        };
149 
        let end = format!(">>>>>>> oxy conflict {n} of {total} end");
150 
        let next_side = format!("======= oxy conflict {n} of {total}, side ");
151 
        let Some(stop) =
152 
            (i + 1..lines.len()).find(|&j| lines[j].trim_ascii_end() == end.as_bytes())
153 
        else {
154 
            left.push((n, "its end marker is missing; edit it by hand"));
155 
            out.extend_from_slice(lines[i]);
156 
            i += 1;
157 
            continue;
158 
        };
159 
        let mut found: Vec<Vec<u8>> = vec![Vec::new()];
160 
        for line in &lines[i + 1..stop] {
161 
            if line.starts_with(next_side.as_bytes()) {
162 
                found.push(Vec::new());
163 
            } else {
164 
                found.last_mut().unwrap().extend_from_slice(line);
165 
            }
166 
        }
167 
        let wanted = match sides.get(n.wrapping_sub(1)) {
168 
            _ if total != sides.len() => {
169 
                Err("the markers are not the file's conflicts now; see `oxy status`")
170 
            }
171 
            Some(Some(w)) => Ok(w),
172 
            _ => Err("both sides changed these lines; edit the markers"),
173 
        };
174 
        let keep = wanted.and_then(|w| {
175 
            let mut a: Vec<Vec<u8>> = w.iter().map(|s| with_newline(s)).collect();
176 
            let mut b = found.clone();
177 
            a.sort();
178 
            b.sort();
179 
            if a == b {
180 
                Ok(w)
181 
            } else {
182 
                Err("the lines between its markers were edited; finish it by hand")
183 
            }
184 
        });
185 
        match keep {
186 
            Ok(w) => {
187 
                let order: Vec<&Vec<u8>> = if reverse {
188 
                    w.iter().rev().collect()
189 
                } else {
190 
                    w.iter().collect()
191 
                };
192 
                for s in order {
193 
                    out.extend_from_slice(&with_newline(s));
194 
                }
195 
                settled.push(n);
196 
            }
197 
            Err(why) => {
198 
                left.push((n, why));
199 
                for line in &lines[i..=stop] {
200 
                    out.extend_from_slice(line);
201 
                }
202 
            }
203 
        }
204 
        i = stop + 1;
205 
    }
206 
    (out, settled, left)
207 
}
208
209 
/// `(n, total)` from a conflict's opening marker (spec §9.5).
210 
fn conflict_header(line: &[u8]) -> Option<(usize, usize)> {
211 
    let rest = line.strip_prefix(b"<<<<<<< oxy conflict ")?;
212 
    let rest = std::str::from_utf8(rest).ok()?;
213 
    let (n, rest) = rest.split_once(" of ")?;
214 
    let total = rest.split(|c: char| !c.is_ascii_digit()).next()?;
215 
    Some((n.parse().ok()?, total.parse().ok()?))
216 
}
217
98286406 Andy McCall ↶ 218 
pub fn is_binary(content: &[u8]) -> bool {
219 
    content[..content.len().min(BINARY_SNIFF)].contains(&0)
220 
}
221
222 
impl Repo {
614541c3 Andy McCall ↶ 223 
    /// Diff the files on disk against `base` and build the operations of a
224 
    /// change from `base` to the working directory. `reuse` maps paths to the
225 
    /// NodeIds an earlier version of the same change assigned to new paths.
75167672 Andy McCall 226 
    pub(crate) fn build_against(
227 
        &self,
228 
        base: &State,
229 
        base_snapshot: &Snapshot,
230 
        reuse: HashMap<String, (NodeId, NodeKind)>,
231 
    ) -> Result<BuiltChange> {
c863c7a7 Andy McCall ↶ 232 
        let mut fs_entries =
696bbf29 Andy McCall ↶ 233 
            self.walk_working_copy_with(None, &Self::tracked_kinds(base_snapshot), None)?;
c863c7a7 Andy McCall ↶ 234 
        self.platform_normalize(&mut fs_entries, base_snapshot);
c3069537 Andy McCall ↶ 235 
        self.build_from_entries(base, base_snapshot, reuse, &fs_entries)
236 
    }
237
c863c7a7 Andy McCall ↶ 238 
    /// Read the working copy as the platform left it (spec §7.7): without
239 
    /// executable bits a file keeps the bit it has in the base, and without
240 
    /// symbolic links a plain file where the base has a symlink is that
241 
    /// symlink, written out as its target, not a new file.
242 
    fn platform_normalize(&self, entries: &mut BTreeMap<String, FsEntry>, base: &Snapshot) {
243 
        let p = self.platform;
244 
        if p.exec_bits && p.symlinks {
245 
            return;
246 
        }
247 
        for (path, e) in entries.iter_mut() {
248 
            let Some(f) = base.files.get(path) else {
249 
                continue;
250 
            };
251 
            if !p.symlinks && f.kind == NodeKind::Symlink && e.kind == NodeKind::File {
252 
                e.kind = NodeKind::Symlink;
253 
                e.exec = false;
254 
            }
255 
            if !p.exec_bits && f.kind == NodeKind::File && e.kind == NodeKind::File {
256 
                e.exec = f.mode.exec;
257 
            }
258 
        }
259 
    }
260
c3069537 Andy McCall ↶ 261 
    /// Like `build_against`, for files given in memory (imports).
75167672 Andy McCall 262 
    pub(crate) fn build_from_entries(
263 
        &self,
264 
        base: &State,
265 
        base_snapshot: &Snapshot,
266 
        reuse: HashMap<String, (NodeId, NodeKind)>,
267 
        fs_entries: &BTreeMap<String, FsEntry>,
268 
    ) -> Result<BuiltChange> {
f7510b63 Andy McCall ↶ 269 
        self.build_from_entries_resolving(base, base_snapshot, reuse, fs_entries, &HashSet::new())
270 
    }
271
272 
    /// Like `build_from_entries`, settling the replace conflicts of the
273 
    /// opaque files in `resolve` to their content (spec §9.3, §9.6): a
274 
    /// caller that knows the user chose it, such as the hub's resolver. A
275 
    /// snapshot never settles one on its own, since a checkout writes one
276 
    /// candidate at the path.
277 
    pub(crate) fn build_from_entries_resolving(
278 
        &self,
279 
        base: &State,
280 
        base_snapshot: &Snapshot,
281 
        reuse: HashMap<String, (NodeId, NodeKind)>,
282 
        fs_entries: &BTreeMap<String, FsEntry>,
283 
        resolve: &HashSet<String>,
284 
    ) -> Result<BuiltChange> {
23ff1288 Andy McCall ↶ 285 
        self.build_from_entries_with(
286 
            base,
287 
            base_snapshot,
288 
            reuse,
289 
            fs_entries,
290 
            resolve,
291 
            Scope::default(),
292 
        )
dfb23a3a Andy McCall ↶ 293 
    }
294
295 
    /// `build_from_entries` whose new nodes take identifiers derived from
296 
    /// `seed` and their paths instead of random ones, so the same input
297 
    /// always builds the same change (a Git import, spec §23).
2306a4d2 Andy McCall ↶ 298 
    ///
299 
    /// With `dir_needed`, the snapshot and entries hold only the paths the
300 
    /// import touches (and the directories a removal may leave empty); a
301 
    /// directory stays while `dir_needed` says the new tree has something
302 
    /// under it.
dfb23a3a Andy McCall ↶ 303 
    pub(crate) fn build_from_entries_seeded(
304 
        &self,
305 
        base: &State,
306 
        base_snapshot: &Snapshot,
307 
        fs_entries: &BTreeMap<String, FsEntry>,
308 
        seed: [u8; 32],
2306a4d2 Andy McCall ↶ 309 
        dir_needed: Option<&dyn Fn(&str) -> bool>,
dfb23a3a Andy McCall ↶ 310 
    ) -> Result<BuiltChange> {
311 
        self.build_from_entries_with(
312 
            base,
313 
            base_snapshot,
314 
            HashMap::new(),
315 
            fs_entries,
316 
            &HashSet::new(),
23ff1288 Andy McCall ↶ 317 
            Scope {
318 
                seed: Some(seed),
319 
                dir_needed,
320 
            },
dfb23a3a Andy McCall ↶ 321 
        )
322 
    }
323
23ff1288 Andy McCall ↶ 324 
    /// `build_from_entries` for a snapshot most of whose files are known
325 
    /// unchanged (oxygen#41): only the paths that may differ are built
326 
    /// (files read rather than known unchanged, base files gone from disk,
327 
    /// and the directories those may leave empty), against a snapshot of
328 
    /// just those, so a large tree with a few edits costs what they touch.
329 
    fn build_changed_only(
330 
        &self,
331 
        base: &State,
332 
        base_snapshot: &Snapshot,
333 
        reuse: HashMap<String, (NodeId, NodeKind)>,
334 
        fs_entries: &BTreeMap<String, FsEntry>,
335 
    ) -> Result<BuiltChange> {
d616faf5 Andy McCall ↶ 336 
        let pick = crate::timing::phase("snapshot:   of which picking the changed");
23ff1288 Andy McCall ↶ 337 
        // A file known unchanged may still have changed mode: that alters
338 
        // neither its size nor its time.
d616faf5 Andy McCall ↶ 339 
        let changed: BTreeMap<String, FsEntry> = join(fs_entries, &base_snapshot.files)
340 
            .filter_map(|(p, e, f)| {
341 
                let e = e?;
342 
                (!e.unchanged || f.is_none_or(|f| f.kind != e.kind || f.mode.exec != e.exec))
343 
                    .then(|| (p.clone(), e.clone()))
23ff1288 Andy McCall ↶ 344 
            })
345 
            .collect();
346 
        let mut files: BTreeMap<String, crate::state::FileState> = BTreeMap::new();
347 
        for p in changed.keys() {
348 
            if let Some(f) = base_snapshot.files.get(p) {
349 
                files.insert(p.clone(), f.clone());
350 
            }
351 
        }
d616faf5 Andy McCall ↶ 352 
        // Paths the base has and the disk does not.
353 
        let gone = join(&base_snapshot.files, fs_entries)
354 
            .filter_map(|(p, f, e)| (e.is_none()).then_some((p, f?)));
355 
        for (p, f) in gone {
356 
            if f.kind == NodeKind::Dir {
23ff1288 Andy McCall ↶ 357 
                continue;
358 
            }
359 
            files.insert(p.clone(), f.clone());
360 
            let mut end = p.len();
361 
            while let Some(i) = p[..end].rfind('/') {
362 
                let dir = &p[..i];
363 
                if files.contains_key(dir) {
364 
                    break;
365 
                }
366 
                if let Some(d) = base_snapshot.files.get(dir) {
367 
                    files.insert(dir.to_string(), d.clone());
368 
                }
369 
                end = i;
370 
            }
371 
        }
372 
        let part = Snapshot {
373 
            tree: base_snapshot.tree.clone(),
374 
            files,
375 
            unrendered: BTreeSet::new(),
376 
        };
d616faf5 Andy McCall ↶ 377 
        drop(pick);
378 
        let _build = crate::timing::phase("snapshot:   of which building the change");
23ff1288 Andy McCall ↶ 379 
        // Whether anything on disk is under `dir`.
380 
        let dir_needed = |dir: &str| {
381 
            let prefix = format!("{dir}/");
382 
            fs_entries
383 
                .range(prefix.clone()..)
384 
                .next()
385 
                .is_some_and(|(k, _)| k.starts_with(&prefix))
386 
        };
387 
        self.build_from_entries_with(
388 
            base,
389 
            &part,
390 
            reuse,
391 
            &changed,
392 
            &HashSet::new(),
393 
            Scope {
394 
                seed: None,
395 
                dir_needed: Some(&dir_needed),
396 
            },
397 
        )
398 
    }
399
dfb23a3a Andy McCall ↶ 400 
    fn build_from_entries_with(
401 
        &self,
402 
        base: &State,
403 
        base_snapshot: &Snapshot,
404 
        reuse: HashMap<String, (NodeId, NodeKind)>,
405 
        fs_entries: &BTreeMap<String, FsEntry>,
406 
        resolve: &HashSet<String>,
23ff1288 Andy McCall ↶ 407 
        scope: Scope,
dfb23a3a Andy McCall ↶ 408 
    ) -> Result<BuiltChange> {
23ff1288 Andy McCall ↶ 409 
        let Scope { seed, dir_needed } = scope;
2306a4d2 Andy McCall ↶ 410 
        let mut laps = crate::timing::Laps::start();
cdb5beba Andy McCall ↶ 411 
        // Attributes come from the files being committed, so a new
412 
        // `.oxyattributes` applies to the same change that adds it.
413 
        let attrs = crate::attributes::Attributes::from_files(
75167672 Andy McCall 414 
            fs_entries
415 
                .iter()
416 
                .filter(|(p, _)| p.as_str() == ".oxyattributes" || p.ends_with("/.oxyattributes"))
417 
                .map(|(p, e)| (p.as_str(), e.content.as_slice())),
cdb5beba Andy McCall ↶ 418 
        );
419 
        let mut b = ChangeBuilder::new(base, base_snapshot, reuse).with_attributes(attrs);
f7510b63 Andy McCall ↶ 420 
        b.resolve = resolve.clone();
dfb23a3a Andy McCall ↶ 421 
        b.node_seed = seed;
2306a4d2 Andy McCall ↶ 422 
        laps.lap("build: builder");
6e723b81 Andy McCall ↶ 423 
        // Rename detection (spec §7.1, §23): a removed file and an added file
424 
        // of the same kind whose contents are at least half alike become one
425 
        // move of the same node, so identity survives across the rename.
426 
        let renames = detect_renames(base_snapshot, fs_entries, |p| self.in_scope(p));
75167672 Andy McCall 427 
        let renamed_from: HashMap<&str, &str> = renames
428 
            .iter()
429 
            .map(|(old, new)| (new.as_str(), old.as_str()))
430 
            .collect();
431 
        let renamed_to: HashMap<&str, &str> = renames
432 
            .iter()
433 
            .map(|(old, new)| (old.as_str(), new.as_str()))
434 
            .collect();
2306a4d2 Andy McCall ↶ 435 
        laps.lap("build: renames");
360aa74a Andy McCall ↶ 436 
        b.diff_in_parallel(fs_entries);
437 
        laps.lap("build: diffs");
c3069537 Andy McCall ↶ 438 
        for (path, fs_entry) in fs_entries {
98286406 Andy McCall ↶ 439 
            match base_snapshot.files.get(path) {
6e723b81 Andy McCall ↶ 440 
                None => match renamed_from.get(path.as_str()) {
75167672 Andy McCall 441 
                    Some(old) => {
442 
                        b.move_path(self, old, path, fs_entry, &base_snapshot.files[*old])?
443 
                    }
6e723b81 Andy McCall ↶ 444 
                    None => b.add_path(self, path, fs_entry)?,
445 
                },
98286406 Andy McCall ↶ 446 
                Some(expected) => b.update_path(self, path, fs_entry, expected)?,
447 
            }
448 
        }
2306a4d2 Andy McCall ↶ 449 
        laps.lap("build: paths");
958fb27d Andy McCall ↶ 450 
        // A directory stays while anything under it exists or is out of
451 
        // scope: the directories above each such path, gathered once.
452 
        let mut needed_dirs: HashSet<&str> = HashSet::new();
453 
        fn parents_of<'a>(p: &'a str, set: &mut HashSet<&'a str>) {
454 
            let mut end = p.len();
455 
            while let Some(i) = p[..end].rfind('/') {
456 
                if !set.insert(&p[..i]) {
457 
                    break;
458 
                }
459 
                end = i;
460 
            }
461 
        }
462 
        for p in fs_entries.keys() {
463 
            parents_of(p, &mut needed_dirs);
464 
        }
465 
        for p in base_snapshot.files.keys() {
466 
            if !self.in_scope(p) {
467 
                parents_of(p, &mut needed_dirs);
468 
            }
469 
        }
98286406 Andy McCall ↶ 470 
        for (path, expected) in &base_snapshot.files {
75167672 Andy McCall 471 
            if fs_entries.contains_key(path)
472 
                || !self.in_scope(path)
473 
                || renamed_to.contains_key(path.as_str())
474 
            {
98286406 Andy McCall ↶ 475 
                continue;
476 
            }
2306a4d2 Andy McCall ↶ 477 
            let still_needed = expected.kind == NodeKind::Dir
478 
                && match dir_needed {
479 
                    Some(f) => f(path),
480 
                    None => needed_dirs.contains(path.as_str()),
481 
                };
98286406 Andy McCall ↶ 482 
            if !still_needed {
483 
                b.remove_path(path, expected);
484 
            }
485 
        }
2306a4d2 Andy McCall ↶ 486 
        laps.lap("build: removals");
a87c7b9d Andy McCall ↶ 487 
        if seed.is_some() {
488 
            b.retire_stray_placements();
489 
        }
c2e1d090 Andy McCall 490 
        b.retire_unified();
2306a4d2 Andy McCall ↶ 491 
        laps.lap("build: stray placements");
614541c3 Andy McCall ↶ 492 
        let mut statuses = b.statuses;
493 
        statuses.sort_by(|a, b| a.path.cmp(&b.path));
75167672 Andy McCall 494 
        Ok(BuiltChange {
495 
            tree_ops: b.tree_ops,
496 
            edits: b.edits,
497 
            contributors: b.contributors.into_iter().collect(),
498 
            resolves: b.resolves,
499 
            statuses,
500 
            blobs: b.blobs,
501 
        })
614541c3 Andy McCall ↶ 502 
    }
98286406 Andy McCall ↶ 503
c07e0e7d Andy McCall ↶ 504 
    /// Store a change and, when `core.sign = change` and we are its author,
505 
    /// sign it. A rewritten change by another author is stored unsigned: only
506 
    /// its author (or a delegate) can vouch for it.
614541c3 Andy McCall ↶ 507 
    pub(crate) fn store_signed(&self, change: &Change) -> Result<ChangeHash> {
c07e0e7d Andy McCall ↶ 508 
        let ours = change.author == self.author();
509 
        if self.config.sign_changes && ours && self.secret_key()?.is_none() {
614541c3 Andy McCall ↶ 510 
            return fail("core.sign is \"change\" but no secret key is available for author.key");
511 
        }
5b4c5130 Andy McCall ↶ 512 
        let storing = crate::timing::phase("store: the change");
614541c3 Andy McCall ↶ 513 
        let hash = self.store.put_change(change)?;
5b4c5130 Andy McCall ↶ 514 
        drop(storing);
c07e0e7d Andy McCall ↶ 515 
        if self.config.sign_changes && ours {
5b4c5130 Andy McCall ↶ 516 
            let _t = crate::timing::phase("store: sign it");
614541c3 Andy McCall ↶ 517 
            self.sign_change(&hash)?;
518 
        }
519 
        Ok(hash)
157d02e9 Andy McCall ↶ 520 
    }
521
522 
    fn resolutions_path(&self) -> PathBuf {
523 
        self.oxy
524 
            .join("workspaces")
525 
            .join(self.workspace.to_hex())
526 
            .join("resolve")
527 
    }
528
529 
    /// The paths whose replace conflicts the user chose to settle with what
530 
    /// the working copy holds there (`oxy resolve PATH`, spec §9.3).
531 
    pub fn resolutions(&self) -> Result<BTreeSet<String>> {
532 
        match fs::read_to_string(self.resolutions_path()) {
533 
            Ok(text) => Ok(text
534 
                .lines()
535 
                .filter(|l| !l.is_empty())
536 
                .map(String::from)
537 
                .collect()),
538 
            Err(e) if e.kind() == std::io::ErrorKind::NotFound => Ok(BTreeSet::new()),
539 
            Err(e) => Err(e.into()),
540 
        }
541 
    }
542
543 
    /// Settle the replace conflicts at `paths` (a file's or a subrepo pin's,
544 
    /// spec §9.3) with what the working copy holds there: the next snapshot
545 
    /// replaces every live candidate with it. Refused for a path whose base
546 
    /// has no replace conflict.
547 
    pub fn choose_resolutions(&self, base: &Snapshot, paths: &[String]) -> Result<()> {
548 
        for p in paths {
549 
            if !has_replace_conflict(base, p) {
550 
                return Err(RepoError::Msg(format!(
9bd98cfb Andy McCall ↶ 551 
                    "{p} has no replace conflict to settle (a text file's conflict is settled by editing out its markers, or `oxy resolve --both {p}`)"
157d02e9 Andy McCall ↶ 552 
                )));
553 
            }
554 
        }
555 
        let mut all = self.resolutions()?;
556 
        all.extend(paths.iter().cloned());
557 
        let text: String = all.iter().map(|p| format!("{p}\n")).collect();
558 
        crate::durable::write_atomic(&self.resolutions_path(), text.as_bytes())?;
559 
        Ok(())
9bd98cfb Andy McCall ↶ 560 
    }
561
562 
    /// Settle the adjacent-insertion conflicts of the text files at `paths`
563 
    /// (oxygen#121) by keeping every side, earliest change first (latest
564 
    /// first with `reverse`): each one's markers in the working copy give
565 
    /// way to its sides' lines, and the next snapshot records the result.
566 
    /// A conflict whose sides changed the same lines, or whose text between
567 
    /// the markers was edited, keeps its markers. With no paths, every
568 
    /// conflicted text file.
569 
    pub fn keep_both(
570 
        &self,
571 
        snap: &SnapshotResult,
572 
        paths: &[String],
573 
        reverse: bool,
574 
    ) -> Result<Vec<KeptBoth>> {
575 
        let paths: Vec<String> = if paths.is_empty() {
576 
            snap.base_snapshot
577 
                .files
578 
                .iter()
579 
                .filter(|(_, f)| matches!(&f.content, FileContent::Text { materialized, .. } if materialized.is_conflicted()))
580 
                .map(|(p, _)| p.clone())
581 
                .collect()
582 
        } else {
583 
            paths.to_vec()
584 
        };
585 
        let mut out = Vec::new();
586 
        for p in paths {
587 
            let conflicted = match snap.base_snapshot.files.get(&p).map(|f| (f, &f.content)) {
588 
                Some((f, FileContent::Text { materialized, .. }))
589 
                    if materialized.is_conflicted() =>
590 
                {
591 
                    f
592 
                }
593 
                Some((_, FileContent::Opaque { blobs })) if blobs.len() > 1 => {
594 
                    return Err(RepoError::Msg(format!(
595 
                        "{p} has a replace conflict: keep one version with `oxy resolve {p}`"
596 
                    )))
597 
                }
598 
                _ => {
599 
                    return Err(RepoError::Msg(format!(
600 
                        "{p} has no text conflict to settle"
601 
                    )))
602 
                }
603 
            };
604 
            let sides = snap.base.both_sides(&conflicted.node);
605 
            let disk = self.root().join(&p);
606 
            let text = fs::read(&disk)?;
607 
            let (bytes, settled, left) = keep_both_text(&text, &sides, reverse);
608 
            if !settled.is_empty() {
609 
                fs::write(&disk, bytes)?;
610 
            }
611 
            out.push(KeptBoth {
612 
                path: p,
613 
                settled,
614 
                left,
615 
            });
616 
        }
617 
        Ok(out)
157d02e9 Andy McCall ↶ 618 
    }
619
4afe288b Andy McCall ↶ 620 
    /// Settle the adjacent insertions of each conflicted text file whose
621 
    /// `.oxyattributes` say `merge=union` (oxygen#145) by keeping both
622 
    /// sides, earliest change first, as [`Repo::keep_both`] does: a
623 
    /// checkout calls it, and the next commit records the resolution. The
624 
    /// files it settled something in.
625 
    pub(crate) fn settle_unions(&self, command: &str) -> Result<Vec<KeptBoth>> {
626 
        let snap = self.snapshot(command)?;
627 
        let paths: Vec<String> = snap
628 
            .base_snapshot
629 
            .files
630 
            .iter()
631 
            .filter(|(p, f)| {
632 
                matches!(&f.content, FileContent::Text { materialized, .. } if materialized.is_conflicted())
633 
                    && snap.attributes.lookup(p).union
634 
            })
635 
            .map(|(p, _)| p.clone())
636 
            .collect();
637 
        if paths.is_empty() {
638 
            return Ok(Vec::new());
639 
        }
640 
        let kept = self.keep_both(&snap, &paths, false)?;
641 
        Ok(kept.into_iter().filter(|k| !k.settled.is_empty()).collect())
642 
    }
643
ac4ec124 Andy McCall ↶ 644 
    /// Every conflict of `snap`, `state`'s materialization, settled by
645 
    /// keeping both sides (oxygen#145): the new text of each conflicted
646 
    /// file, when each is a text file its `.oxyattributes` mark
647 
    /// `merge=union` and each of its conflicts is an adjacent insertion;
648 
    /// `None` when any conflict needs a person, or there is none. A host
649 
    /// lands a proposal with these as a resolution, as a checkout settles
650 
    /// them in a working copy.
651 
    pub fn union_resolution(state: &State, snap: &Snapshot) -> Option<Vec<(String, Vec<u8>)>> {
652 
        if snap.tree.is_conflicted() {
653 
            return None;
654 
        }
655 
        let attrs = Repo::state_attributes(snap);
656 
        let mut out = Vec::new();
657 
        for (p, f) in &snap.files {
658 
            if !f.is_conflicted() {
659 
                continue;
660 
            }
661 
            let FileContent::Text { bytes, .. } = &f.content else {
662 
                return None;
663 
            };
664 
            if !attrs.lookup(p).union {
665 
                return None;
666 
            }
667 
            let sides = state.both_sides(&f.node);
668 
            let (text, _, left) = keep_both_text(bytes, &sides, false);
669 
            if !left.is_empty() || sides.iter().any(Option::is_none) {
670 
                return None;
671 
            }
672 
            out.push((p.clone(), text));
673 
        }
674 
        (!out.is_empty()).then_some(out)
675 
    }
676
157d02e9 Andy McCall ↶ 677 
    /// The chosen resolutions `base` still needs; the rest are forgotten.
678 
    fn resolutions_for(&self, base: &Snapshot) -> Result<HashSet<String>> {
679 
        let all = self.resolutions()?;
680 
        if all.is_empty() {
681 
            return Ok(HashSet::new());
682 
        }
683 
        let live: HashSet<String> = all
684 
            .iter()
685 
            .filter(|p| has_replace_conflict(base, p))
686 
            .cloned()
687 
            .collect();
688 
        if live.is_empty() {
689 
            match fs::remove_file(self.resolutions_path()) {
690 
                Ok(()) => {}
691 
                Err(e) if e.kind() == std::io::ErrorKind::NotFound => {}
692 
                Err(e) => return Err(e.into()),
693 
            }
694 
        }
695 
        Ok(live)
614541c3 Andy McCall ↶ 696 
    }
697
698 
    /// Snapshot the working copy into the working-copy change (spec §12.3).
cc7a0056 Andy McCall ↶ 699 
    /// The result carries the full base state and its whole rendering.
614541c3 Andy McCall ↶ 700 
    pub fn snapshot(&self, command: &str) -> Result<SnapshotResult> {
f9466021 Andy McCall ↶ 701 
        self.snapshot_as(command, false, false)
cc7a0056 Andy McCall ↶ 702 
    }
703
704 
    /// Snapshot the working copy as [`Repo::snapshot`] does, for commands
8ab265dc Andy McCall ↶ 705 
    /// that need its statuses, conflicts and base history, and the graphs
706 
    /// of the files it changes: status, log, show, commit, diff, hunks and
707 
    /// discard. Text files known unchanged since the last manifest are
cc7a0056 Andy McCall ↶ 708 
    /// neither read nor rendered: the result's `base` is a history-only
709 
    /// state without their graphs, and its `base_snapshot` lists them in
8ab265dc Andy McCall ↶ 710 
    /// `unrendered` with empty bytes. They are never conflicted, and no
711 
    /// change the working copy makes touches them.
cc7a0056 Andy McCall ↶ 712 
    pub fn snapshot_status(&self, command: &str) -> Result<SnapshotResult> {
f9466021 Andy McCall ↶ 713 
        self.snapshot_as(command, true, false)
cc7a0056 Andy McCall ↶ 714 
    }
715
cd4eb55e Andy McCall ↶ 716 
    /// `snapshot_as`; `for_commit` writes the manifest even where a status
717 
    /// would not bother (an edited file is read next time whatever it
718 
    /// says), since after the commit that file is unchanged, and records no
719 
    /// operation: the commit's own, recorded straight after, replaces the
720 
    /// working copy, so the snapshot's would cost a durable write for
721 
    /// nothing. Should the commit fail, the files are still on disk and the
722 
    /// next snapshot finds them.
f9466021 Andy McCall ↶ 723 
    fn snapshot_as(
724 
        &self,
725 
        command: &str,
726 
        partial: bool,
cd4eb55e Andy McCall ↶ 727 
        for_commit: bool,
f9466021 Andy McCall ↶ 728 
    ) -> Result<SnapshotResult> {
4e88e774 Andy McCall ↶ 729 
        let _lock = self.lock_snapshot()?;
8f10a07a Andy McCall ↶ 730 
        let batch = self.store.begin_batch()?;
cd4eb55e Andy McCall ↶ 731 
        let result = self.snapshot_locked(command, partial, for_commit);
8f10a07a Andy McCall ↶ 732 
        batch.finish()?;
733 
        result
734 
    }
735
f9466021 Andy McCall ↶ 736 
    fn snapshot_locked(
737 
        &self,
738 
        command: &str,
739 
        partial: bool,
cd4eb55e Andy McCall ↶ 740 
        for_commit: bool,
f9466021 Andy McCall ↶ 741 
    ) -> Result<SnapshotResult> {
0bdc3f91 Andy McCall ↶ 742 
        let _all = crate::timing::phase("snapshot");
614541c3 Andy McCall ↶ 743 
        let (op_ref, op) = self.head_op()?;
744 
        let (old_hash, old_wc) = self.working_copy()?;
cc7a0056 Andy McCall ↶ 745 
        // The base's history first: its graph keys say which files the
746 
        // manifest already knows equal to it.
747 
        let mut base = {
748 
            let _t = crate::timing::phase("snapshot: load base history");
f9575918 Andy McCall ↶ 749 
            self.load_state_indexed(&old_wc.deps)?
0bdc3f91 Andy McCall ↶ 750 
        };
d0d52107 Andy McCall ↶ 751 
        let view = {
752 
            let _t = crate::timing::phase("snapshot: tree view");
c2e1d090 Andy McCall 753 
            self.tree_view(&mut base)?
d0d52107 Andy McCall ↶ 754 
        };
8b276e01 Andy McCall ↶ 755 
        let manifest = {
756 
            let _t = crate::timing::phase("snapshot: read manifest");
757 
            self.read_manifest()?.map(|(_, _, m)| m)
758 
        };
cc7a0056 Andy McCall ↶ 759 
        let tracked: BTreeMap<String, (NodeKind, Option<ObjectRef>)> = {
760 
            let _t = crate::timing::phase("snapshot: tracked paths");
36e51aa8 Andy McCall ↶ 761 
            // In the view's path order, so the map is built without searching.
762 
            view.entries
763 
                .iter()
764 
                .filter(|(_, e)| e.kind != NodeKind::Dir)
765 
                .map(|(p, e)| {
766 
                    let blob = if e.kind != NodeKind::File || e.mode.opaque {
767 
                        base.live_blobs(&e.node).first().copied()
768 
                    } else {
769 
                        None
770 
                    };
771 
                    (p.clone(), (e.kind, blob))
772 
                })
773 
                .collect()
cc7a0056 Andy McCall ↶ 774 
        };
992467d3 Andy McCall ↶ 775 
        // A text file's graph key, worked out when the walk asks for it.
1966f5db Andy McCall ↶ 776 
        let key_of = |p: &str| -> Option<[u8; 32]> {
992467d3 Andy McCall ↶ 777 
            let e = view.entries.get(p)?;
778 
            if e.kind != NodeKind::File || e.mode.opaque {
779 
                return None;
780 
            }
1966f5db Andy McCall ↶ 781 
            base.graph_key(&e.node)
992467d3 Andy McCall ↶ 782 
        };
cc7a0056 Andy McCall ↶ 783 
        let mut fs_entries = {
784 
            let _t = crate::timing::phase("snapshot: read working copy");
992467d3 Andy McCall ↶ 785 
            self.walk_working_copy_with(manifest.as_deref(), &tracked, Some(&key_of))?
cc7a0056 Andy McCall ↶ 786 
        };
992467d3 Andy McCall ↶ 787 
        // The keys of the files read, before building graphs drops them: a
788 
        // file found equal to its base rendering keeps its key.
1966f5db Andy McCall ↶ 789 
        let keys: HashMap<String, [u8; 32]> = fs_entries
992467d3 Andy McCall ↶ 790 
            .iter()
791 
            .filter(|(_, e)| !e.unchanged && e.kind == NodeKind::File)
792 
            .filter_map(|(p, _)| Some((p.clone(), key_of(p)?)))
793 
            .collect();
0bdc3f91 Andy McCall ↶ 794 
        let base_snapshot = {
cc7a0056 Andy McCall ↶ 795 
            let _t = crate::timing::phase("snapshot: load and render base");
796 
            if partial {
d616faf5 Andy McCall ↶ 797 
                let sets = crate::timing::phase("snapshot:   of which choosing graphs");
798 
                // Every file but those on disk known unchanged.
799 
                let need: Vec<(&str, NodeId)> = join(&view.entries, &fs_entries)
800 
                    .filter_map(|(p, v, f)| {
801 
                        let v = v?;
802 
                        (v.kind == NodeKind::File && !f.is_some_and(|f| f.unchanged))
803 
                            .then_some((p.as_str(), v.node))
804 
                    })
cc7a0056 Andy McCall ↶ 805 
                    .collect();
d616faf5 Andy McCall ↶ 806 
                let render: std::collections::HashSet<&str> =
807 
                    need.iter().map(|(p, _)| *p).collect();
808 
                let need: Vec<NodeId> = need.iter().map(|(_, n)| *n).collect();
809 
                drop(sets);
d0990c2b Andy McCall ↶ 810 
                {
811 
                    let _t = crate::timing::phase("snapshot:   of which building graphs");
812 
                    self.load_graphs_of(&mut base, &need)?;
813 
                }
814 
                let _t = crate::timing::phase("snapshot:   of which rendering");
d616faf5 Andy McCall ↶ 815 
                base.materialize_only(&render)
cc7a0056 Andy McCall ↶ 816 
            } else {
817 
                self.load_all_graphs(&mut base)?;
818 
                base.materialize()
819 
            }
0bdc3f91 Andy McCall ↶ 820 
        };
614541c3 Andy McCall ↶ 821
822 
        // Reuse the NodeIds the previous working-copy version assigned to new
823 
        // paths, so an unchanged working copy hashes identically (spec §12.3).
824 
        let reuse = added_node_paths(&old_wc, &base_snapshot);
c863c7a7 Andy McCall ↶ 825 
        self.platform_normalize(&mut fs_entries, &base_snapshot);
157d02e9 Andy McCall ↶ 826 
        // Replace conflicts the user settled with `oxy resolve PATH`
827 
        // (oxygen#140): those still in the base; a choice the base no longer
828 
        // needs (once committed) is forgotten.
829 
        let resolve = self.resolutions_for(&base_snapshot)?;
0bdc3f91 Andy McCall ↶ 830 
        let built = {
831 
            let _t = crate::timing::phase("snapshot: diff against base");
157d02e9 Andy McCall ↶ 832 
            if !resolve.is_empty() {
833 
                self.build_from_entries_resolving(
834 
                    &base,
835 
                    &base_snapshot,
836 
                    reuse,
837 
                    &fs_entries,
838 
                    &resolve,
839 
                )?
840 
            } else if partial && self.config.sparse_paths.is_empty() {
23ff1288 Andy McCall ↶ 841 
                #[cfg(test)]
842 
                let full =
843 
                    self.build_from_entries(&base, &base_snapshot, reuse.clone(), &fs_entries)?;
844 
                let built = self.build_changed_only(&base, &base_snapshot, reuse, &fs_entries)?;
845 
                #[cfg(test)]
846 
                {
847 
                    // New nodes take random identifiers: compare them by
848 
                    // order of appearance.
849 
                    let norm = |b: &BuiltChange| {
850 
                        let text = format!("{:?} {:?} {:?}", b.tree_ops, b.edits, b.statuses);
851 
                        let mut seen: Vec<String> = Vec::new();
852 
                        let mut out = String::new();
853 
                        let mut rest = text.as_str();
854 
                        while let Some(i) = rest.find("NodeId(") {
855 
                            out.push_str(&rest[..i]);
856 
                            let end = i + rest[i..].find(')').unwrap() + 1;
857 
                            let id = rest[i..end].to_string();
858 
                            let n = seen.iter().position(|x| *x == id).unwrap_or_else(|| {
859 
                                seen.push(id);
860 
                                seen.len() - 1
861 
                            });
862 
                            out.push_str(&format!("N{n}"));
863 
                            rest = &rest[end..];
864 
                        }
865 
                        out.push_str(rest);
866 
                        out
867 
                    };
868 
                    assert_eq!(
869 
                        norm(&built),
870 
                        norm(&full),
871 
                        "a changed-only build is the full one"
872 
                    );
873 
                }
874 
                built
875 
            } else {
876 
                self.build_from_entries(&base, &base_snapshot, reuse, &fs_entries)?
877 
            }
0bdc3f91 Andy McCall ↶ 878 
        };
cc7a0056 Andy McCall ↶ 879 
        // Text files found equal to an unconflicted base rendering keep its
880 
        // graph key, so the next snapshot need not read or render them.
881 
        {
882 
            let changed: HashSet<&str> = built.statuses.iter().map(|s| s.path.as_str()).collect();
883 
            for (p, e) in fs_entries.iter_mut() {
884 
                if e.unchanged || e.kind != NodeKind::File || changed.contains(p.as_str()) {
885 
                    continue;
886 
                }
887 
                e.graph_key = match base_snapshot.files.get(p).map(|f| &f.content) {
888 
                    Some(FileContent::Text {
889 
                        bytes,
890 
                        materialized,
40a40bb7 Andy McCall ↶ 891 
                    }) if !materialized.is_conflicted() && *bytes == *e.content => {
1966f5db Andy McCall ↶ 892 
                        keys.get(p).copied()
cc7a0056 Andy McCall ↶ 893 
                    }
894 
                    _ => None,
895 
                };
896 
            }
897 
        }
614541c3 Andy McCall ↶ 898
98286406 Andy McCall ↶ 899 
        let mut wc = Change {
6f8aeb99 Andy McCall ↶ 900 
            version: crate::objects::CHANGE_VERSION,
98286406 Andy McCall ↶ 901 
            change_id: old_wc.change_id,
902 
            deps: vec![],
903 
            author: self.author(),
904 
            hlc: old_wc.hlc,
905 
            message: old_wc.message.clone(),
614541c3 Andy McCall ↶ 906 
            tree_ops: built.tree_ops,
907 
            edits: built.edits,
98286406 Andy McCall ↶ 908 
            kind: ChangeKind::WorkingCopy,
909 
        };
910 
        let base_heads = base.maximal(&old_wc.deps);
911 
        let mut deps: BTreeSet<ChangeHash> = base_heads.into_iter().collect();
912 
        deps.extend(referenced_changes(&wc, &base));
614541c3 Andy McCall ↶ 913 
        deps.extend(built.contributors.iter().copied());
98286406 Andy McCall ↶ 914 
        wc.deps = deps.into_iter().collect();
915 
        wc.normalize();
614541c3 Andy McCall ↶ 916 
        let resolves = built.resolves;
917 
        let contributors = built.contributors;
918 
        let statuses = built.statuses;
98286406 Andy McCall ↶ 919 
        // Empty snapshot: hash the candidate before storing; only record an op on change.
8b276e01 Andy McCall ↶ 920 
        let record = crate::timing::phase("snapshot: hash and record");
17f94b13 Andy McCall ↶ 921 
        // Encoded once, its texts stored a batch at a time as they are
922 
        // made (a stored blob is not stored again); the change itself only
923 
        // when it changed.
5b4c5130 Andy McCall ↶ 924 
        // For a commit the change is neither encoded, stored nor recorded:
925 
        // the commit stores its own, built from this one's operations, and
926 
        // its texts with it. The hash it would have is not needed.
927 
        let (new_hash, changed, encoded) = if for_commit {
928 
            (old_hash, true, Vec::new())
929 
        } else {
930 
            let encoded = wc.encode_streaming(|batch| self.store.put_texts(&batch))?;
931 
            let new_hash = crate::hash::hash_object(crate::hash::ObjectType::Change, &encoded);
932 
            (new_hash, new_hash != old_hash, encoded)
933 
        };
cd4eb55e Andy McCall ↶ 934 
        if changed && !for_commit {
17f94b13 Andy McCall ↶ 935 
            self.store.put(crate::hash::ObjectType::Change, &encoded)?;
98286406 Andy McCall ↶ 936 
            let mut view = op.view.clone();
937 
            view.workspaces.insert(self.workspace, new_hash);
938 
            self.record_op(&[op_ref], &format!("{command}{SNAPSHOT_SUFFIX}"), view)?;
939 
        }
8b276e01 Andy McCall ↶ 940 
        drop(record);
135212de Andy McCall ↶ 941 
        // The prompt compares against this without recording anything.
0bdc3f91 Andy McCall ↶ 942 
        let _t = crate::timing::phase("snapshot: write manifest");
992467d3 Andy McCall ↶ 943 
        let conflicted = base_snapshot.is_conflicted();
944 
        let uncommitted = !wc.tree_ops.is_empty() || !wc.edits.is_empty();
945 
        // Rewriting the manifest pays only when it would teach the next
946 
        // snapshot something: other flags, a file found equal to the base
947 
        // (it need not be read again), a new blob, a path added or gone.
948 
        // An edited file is read next time whatever it says.
949 
        let worth = match self.read_manifest()? {
cd4eb55e Andy McCall ↶ 950 
            _ if for_commit => true,
992467d3 Andy McCall ↶ 951 
            None => true,
952 
            Some((c, u, m)) => {
953 
                c != conflicted
954 
                    || u != uncommitted
955 
                    || m.entries.len() != fs_entries.len()
956 
                    || fs_entries.iter().any(|(p, e)| {
957 
                        !e.unchanged && (e.graph_key.is_some() || !m.entries.contains_key(p))
958 
                    })
959 
                    || built
960 
                        .blobs
961 
                        .iter()
962 
                        .any(|(p, b)| m.entries.get(p).and_then(|e| e.blob) != Some(*b))
36041796 Andy McCall ↶ 963 
                    // With uncommitted work, the status cache below holds
964 
                    // only while the manifest describes the edited files as
965 
                    // they are now.
966 
                    || (uncommitted && statuses.iter().any(|st| {
967 
                        let Some(e) = fs_entries.get(&st.path) else {
968 
                            return m.entries.contains_key(&st.path);
969 
                        };
970 
                        let Some(me) = m.entries.get(&st.path) else {
971 
                            return true;
972 
                        };
973 
                        let hash = e
974 
                            .content_hash
975 
                            .unwrap_or_else(|| *blake3::hash(&e.content).as_bytes());
976 
                        let stat = e.stat.or_else(|| {
977 
                            fs::symlink_metadata(self.root.join(&st.path))
978 
                                .ok()
979 
                                .map(|md| (md.len(), super::prompt::mtime_ns(&md)))
980 
                        });
e226a6cc Andy McCall ↶ 981 
                        // The executable bit too: a mode change moves
982 
                        // neither size nor time.
983 
                        me.hash != hash || me.exec != e.exec || stat != Some((me.size, me.mtime_ns))
36041796 Andy McCall ↶ 984 
                    }))
992467d3 Andy McCall ↶ 985 
            }
986 
        };
f9466021 Andy McCall ↶ 987 
        let attributes = crate::attributes::Attributes::from_files(
988 
            fs_entries
989 
                .iter()
990 
                .filter(|(p, _)| is_attributes(p))
991 
                .map(|(p, e)| (p.as_str(), e.content.as_slice())),
992 
        );
236f35c0 Andy McCall ↶ 993 
        let mut manifest = None;
994 
        if for_commit {
995 
            manifest = Some(PendingManifest {
996 
                entries: fs_entries,
997 
                blobs: built.blobs.clone(),
998 
                dirs: self.walked_dirs.take(),
999 
                written: super::prompt::now_ns(),
1000 
            });
1001 
        } else {
1002 
            if worth {
1003 
                self.write_manifest_entries(conflicted, uncommitted, &fs_entries, &built.blobs)?;
1004 
            }
1005 
            self.write_status_cache((uncommitted && !conflicted).then_some(&new_hash), &statuses);
1006 
        }
75167672 Andy McCall 1007 
        Ok(SnapshotResult {
1008 
            wc_hash: new_hash,
1009 
            wc,
1010 
            base,
1011 
            base_snapshot,
1012 
            statuses,
1013 
            changed,
1014 
            resolves,
1015 
            contributors,
f9466021 Andy McCall ↶ 1016 
            attributes,
d4dbab97 Andy McCall ↶ 1017 
            untracked_nested: self.untracked_nested.borrow().clone(),
236f35c0 Andy McCall ↶ 1018 
            manifest,
75167672 Andy McCall 1019 
        })
98286406 Andy McCall ↶ 1020 
    }
1021
50a046e4 Andy McCall ↶ 1022 
    /// `oxy status --json` (spec §9.5): the changed paths and every conflict
1023 
    /// of the working copy's base with its kind, path, and contributing
1024 
    /// ChangeIds.
1025 
    pub fn status_json(&self, snap: &SnapshotResult) -> Result<String> {
1026 
        let heads = self.bookmark_heads()?;
1027 
        let base_snapshot = &snap.base_snapshot;
1028 
        let by_node = base_snapshot.tree.by_node();
1029 
        let path_of = |n: &NodeId| by_node.get(n).map(|e| e.path.clone());
75167672 Andy McCall 1030 
        let id_of = |h: &ChangeHash| {
1031 
            snap.base
1032 
                .get(h)
1033 
                .map(|m| m.change_id.to_hex())
1034 
                .unwrap_or_else(|| h.to_hex())
1035 
        };
1036 
        let paths = json_array(snap.statuses.iter().map(|s| {
1037 
            format!(
1038 
                "{{\"path\":{},\"status\":{}}}",
1039 
                json_str(&s.path),
1040 
                json_str(&format!("{:?}", s.status))
1041 
            )
1042 
        }));
50a046e4 Andy McCall ↶ 1043 
        let mut conflicts: Vec<String> = Vec::new();
1044 
        for (p, f) in &base_snapshot.files {
1045 
            if !f.is_conflicted() {
1046 
                continue;
1047 
            }
1048 
            let (kind, changes) = match &f.content {
75167672 Andy McCall 1049 
                FileContent::Text { .. } => {
1050 
                    ("text", snap.base.conflict_regions(&f.node).contributors)
1051 
                }
1052 
                FileContent::Opaque { blobs } => {
1053 
                    ("replace", snap.base.blob_contributors(&f.node, blobs))
1054 
                }
50a046e4 Andy McCall ↶ 1055 
                FileContent::Dir => continue,
1056 
            };
75167672 Andy McCall 1057 
            let resolved = match snap
1058 
                .statuses
1059 
                .iter()
1060 
                .find(|s| s.path == *p)
1061 
                .map(|s| s.status)
1062 
            {
50a046e4 Andy McCall ↶ 1063 
                Some(ChangeStatus::Resolved) => "true",
1064 
                Some(ChangeStatus::ConflictEdited) => "\"markers\"",
1065 
                _ => "false",
1066 
            };
1067 
            conflicts.push(format!(
1068 
                "{{\"kind\":{},\"path\":{},\"node\":{},\"changes\":{},\"resolved\":{resolved}}}",
1069 
                json_str(kind),
1070 
                json_str(p),
1071 
                json_str(&f.node.to_hex()),
1072 
                json_array(changes.iter().map(|h| json_str(&id_of(h))))
1073 
            ));
1074 
        }
1075 
        for c in &base_snapshot.tree.conflicts {
1076 
            let (kind, nodes): (&str, Vec<NodeId>) = match c {
1077 
                crate::tree::TreeConflict::Name { nodes, .. } => ("name", nodes.clone()),
75167672 Andy McCall 1078 
                crate::tree::TreeConflict::MultiPlacement { node, .. } => {
1079 
                    ("multi-placement", vec![*node])
1080 
                }
50a046e4 Andy McCall ↶ 1081 
                crate::tree::TreeConflict::Orphan { node, .. } => ("orphan", vec![*node]),
1082 
                crate::tree::TreeConflict::Cycle { node } => ("cycle", vec![*node]),
1083 
                crate::tree::TreeConflict::Mode { node, .. } => ("mode", vec![*node]),
1084 
            };
1085 
            let detail = match c {
1086 
                crate::tree::TreeConflict::Name { name, parent, .. } => {
75167672 Andy McCall 1087 
                    let dir = if *parent == NodeId::ROOT {
1088 
                        String::new()
1089 
                    } else {
1090 
                        path_of(parent).unwrap_or_default()
1091 
                    };
1092 
                    format!(
1093 
                        ",\"path\":{}",
1094 
                        json_str(&if dir.is_empty() {
1095 
                            name.clone()
1096 
                        } else {
1097 
                            format!("{dir}/{name}")
1098 
                        })
1099 
                    )
50a046e4 Andy McCall ↶ 1100 
                }
75167672 Andy McCall 1101 
                _ => nodes
1102 
                    .first()
1103 
                    .and_then(path_of)
1104 
                    .map(|p| format!(",\"path\":{}", json_str(&p)))
1105 
                    .unwrap_or_default(),
50a046e4 Andy McCall ↶ 1106 
            };
1107 
            conflicts.push(format!(
1108 
                "{{\"kind\":{},\"nodes\":{}{detail}}}",
1109 
                json_str(kind),
1110 
                json_array(nodes.iter().map(|n| json_str(&n.to_hex())))
1111 
            ));
1112 
        }
1113 
        Ok(format!(
b689779e Andy McCall ↶ 1114 
            "{{\"bookmark\":{},\"working_copy\":{},\"heads\":{},\"clean\":{},\"conflicted\":{},\"paths\":{paths},\"conflicts\":{},\"untracked_nested\":{},\"skipped_large\":{}}}",
50a046e4 Andy McCall ↶ 1115 
            json_str(&self.config.bookmark),
1116 
            json_str(&snap.wc.change_id.to_hex()),
1117 
            heads.len(),
1118 
            snap.statuses.is_empty(),
1119 
            base_snapshot.is_conflicted(),
d4dbab97 Andy McCall ↶ 1120 
            json_array(conflicts),
b689779e Andy McCall ↶ 1121 
            json_array(snap.untracked_nested.iter().map(|p| json_str(p))),
1122 
            json_array(
1123 
                self.skipped_large
1124 
                    .borrow()
1125 
                    .iter()
1126 
                    .map(|(p, n)| format!("{{\"path\":{},\"size\":{n}}}", json_str(p)))
1127 
            )
50a046e4 Andy McCall ↶ 1128 
        ))
1129 
    }
1130
98286406 Andy McCall ↶ 1131 
    /// Turn the working-copy change into a normal change on the bookmark (spec §12.3).
1132 
    pub fn commit(&self, message: &str, extra_deps: &[ChangeHash]) -> Result<ChangeHash> {
957fee30 Andy McCall ↶ 1133 
        self.commit_selection(message, extra_deps, &Selection::All)
1134 
    }
1135
1136 
    /// The hunks of the working copy, for `oxy commit --interactive`: each
1137 
    /// tree operation and each edit, in the order `select_hunks` numbers them.
1138 
    pub fn hunks(&self) -> Result<(SnapshotResult, Vec<Hunk>)> {
8ab265dc Andy McCall ↶ 1139 
        let snap = self.snapshot_status("oxy commit")?;
957fee30 Andy McCall ↶ 1140 
        let hunks = render_hunks(&snap.wc, &snap.base, &snap.base_snapshot);
1141 
        Ok((snap, hunks))
1142 
    }
1143
1144 
    /// Commit part of the working copy (spec §12.3). What is not selected
1145 
    /// stays on disk as uncommitted work: the next snapshot rebuilds it
1146 
    /// against the new base, so the working copy always describes the
1147 
    /// directory.
f9466021 Andy McCall ↶ 1148 
    /// The graph key in `st` of each text file at `paths`, the paths a
1149 
    /// commit changed: only their graphs changed, so the keys the manifest
1150 
    /// holds for the others still stand (and any that does not is simply
1151 
    /// not trusted by the next snapshot, which compares keys). Files with
1152 
    /// an `eol` attribute are left out: their bytes on disk are not the
1153 
    /// rendering.
1154 
    fn committed_graph_keys(
1155 
        &self,
1156 
        st: &State,
1157 
        paths: &[String],
1158 
        attrs: &crate::attributes::Attributes,
1159 
    ) -> HashMap<String, [u8; 32]> {
86e1824f Andy McCall ↶ 1160 
        let _t = crate::timing::phase("commit: graph keys");
c2e1d090 Andy McCall 1161 
        let view = st.view();
f9466021 Andy McCall ↶ 1162 
        paths
86e1824f Andy McCall ↶ 1163 
            .iter()
f9466021 Andy McCall ↶ 1164 
            .filter_map(|p| {
1165 
                let e = view.entries.get(p)?;
1166 
                (e.kind == NodeKind::File && !e.mode.opaque && attrs.lookup(p).eol.is_none())
1167 
                    .then_some(())?;
1168 
                Some((p.clone(), st.graph_key(&e.node)?))
86e1824f Andy McCall ↶ 1169 
            })
8457a6cc Andy McCall ↶ 1170 
            .collect()
86e1824f Andy McCall ↶ 1171 
    }
1172
75167672 Andy McCall 1173 
    pub fn commit_selection(
1174 
        &self,
1175 
        message: &str,
1176 
        extra_deps: &[ChangeHash],
1177 
        selection: &Selection,
1178 
    ) -> Result<ChangeHash> {
75051941 Andy McCall ↶ 1179 
        Ok(self
1180 
            .commit_selection_partial(message, extra_deps, selection)?
1181 
            .0)
1182 
    }
1183
1184 
    /// As [`Repo::commit_selection`], and whether the commit left part of
1185 
    /// the working copy uncommitted: when it did not, nothing is left, and
1186 
    /// no snapshot is needed to say so.
1187 
    pub fn commit_selection_partial(
1188 
        &self,
1189 
        message: &str,
1190 
        extra_deps: &[ChangeHash],
1191 
        selection: &Selection,
1192 
    ) -> Result<(ChangeHash, bool)> {
b30a880f Andy McCall ↶ 1193 
        // The commit's objects are written together, made durable at once
1194 
        // before the operation that names them (spec §13.6).
1195 
        let batch = self.store.begin_batch()?;
1196 
        let out = self.commit_selection_batched(message, extra_deps, selection);
1197 
        batch.finish()?;
1198 
        out
1199 
    }
1200
236f35c0 Andy McCall ↶ 1201 
    /// Write the manifest a commit's snapshot left to the commit, with the
1202 
    /// graph keys `keys` sets.
1203 
    fn write_pending_manifest(
1204 
        &self,
1205 
        mut m: PendingManifest,
1206 
        conflicted: bool,
1207 
        keys: HashMap<String, [u8; 32]>,
1208 
    ) -> Result<()> {
1209 
        for (path, k) in keys {
1210 
            if let Some(e) = m.entries.get_mut(&path) {
1211 
                e.graph_key = Some(k);
1212 
            }
1213 
        }
1214 
        self.write_manifest_entries_as_of(
1215 
            conflicted,
1216 
            false,
1217 
            &m.entries,
1218 
            &m.blobs,
1219 
            m.written,
1220 
            m.dirs.unwrap_or_default(),
1221 
        )
1222 
    }
1223
b30a880f Andy McCall ↶ 1224 
    fn commit_selection_batched(
1225 
        &self,
1226 
        message: &str,
1227 
        extra_deps: &[ChangeHash],
1228 
        selection: &Selection,
75051941 Andy McCall ↶ 1229 
    ) -> Result<(ChangeHash, bool)> {
6caacfff Andy McCall ↶ 1230 
        let _wc = self.lock_working_copy()?;
e226a6cc Andy McCall ↶ 1231 
        let mut snap = match self.snapshot_from_cache()? {
1232 
            Some(snap) => snap,
1233 
            None => self.snapshot_as("oxy commit", true, true)?,
1234 
        };
236f35c0 Andy McCall ↶ 1235 
        let pending = snap.manifest.take();
98286406 Andy McCall ↶ 1236 
        if snap.wc.tree_ops.is_empty() && snap.wc.edits.is_empty() {
236f35c0 Andy McCall ↶ 1237 
            if let Some(m) = pending {
1238 
                let conflicted = snap.base_snapshot.is_conflicted();
1239 
                self.write_pending_manifest(m, conflicted, HashMap::new())?;
1240 
            }
98286406 Andy McCall ↶ 1241 
            return fail("nothing to commit: the working copy matches its base");
1242 
        }
75167672 Andy McCall 1243 
        if let Some(s) = snap
1244 
            .statuses
1245 
            .iter()
1246 
            .find(|s| s.status == ChangeStatus::ConflictEdited)
1247 
        {
98286406 Andy McCall ↶ 1248 
            return fail(format!("{} still contains conflict markers", s.path));
1249 
        }
1250 
        if message.trim().is_empty() {
1251 
            return fail("a commit message is required");
1252 
        }
192fad1e Andy McCall ↶ 1253 
        // What the whole working copy references, before a whole commit
1254 
        // takes its operations over (they are moved, not copied: for a
1255 
        // large tree they are most of the memory).
1256 
        let wc_refs = referenced_changes(&snap.wc, &snap.base);
f9466021 Andy McCall ↶ 1257 
        let changed_paths: Vec<String> = snap.statuses.iter().map(|s| s.path.clone()).collect();
957fee30 Andy McCall ↶ 1258 
        let (tree_ops, edits, partial) = match selection {
192fad1e Andy McCall ↶ 1259 
            Selection::All => (
1260 
                std::mem::take(&mut snap.wc.tree_ops),
1261 
                std::mem::take(&mut snap.wc.edits),
1262 
                false,
1263 
            ),
957fee30 Andy McCall ↶ 1264 
            Selection::Paths(paths) => {
1265 
                let (o, e) = select_paths(&snap.wc, &snap.base_snapshot, paths);
1266 
                let partial = o.len() != snap.wc.tree_ops.len() || e.len() != snap.wc.edits.len();
1267 
                (o, e, partial)
1268 
            }
1269 
            Selection::Hunks(indices) => {
c2e1d090 Andy McCall 1270 
                let (o, e) = select_hunks(&snap.wc, indices, &snap.base_snapshot.tree.unified);
957fee30 Andy McCall ↶ 1271 
                let partial = o.len() != snap.wc.tree_ops.len() || e.len() != snap.wc.edits.len();
1272 
                (o, e, partial)
1273 
            }
1274 
        };
1275 
        if tree_ops.is_empty() && edits.is_empty() {
1276 
            return fail("nothing selected to commit");
1277 
        }
69953b50 Andy McCall ↶ 1278 
        // A resolution may leave other work uncommitted (oxygen#118), but
1279 
        // takes every file it resolves, whole.
957fee30 Andy McCall ↶ 1280 
        if partial && snap.resolves {
69953b50 Andy McCall ↶ 1281 
            let resolved = snap
1282 
                .statuses
1283 
                .iter()
1284 
                .filter(|s| s.status == ChangeStatus::Resolved)
1285 
                .map(|s| s.path.as_str());
1286 
            let missing: Vec<&str> = match selection {
1287 
                Selection::Paths(paths) => resolved
1288 
                    .filter(|p| {
1289 
                        !paths.iter().any(|x| {
1290 
                            p == x || p.starts_with(&format!("{}/", x.trim_end_matches('/')))
1291 
                        })
1292 
                    })
1293 
                    .collect(),
1294 
                _ => resolved.collect(),
1295 
            };
1296 
            if !missing.is_empty() {
1297 
                return fail(format!(
1298 
                    "a conflict resolution takes every file it resolves, whole: commit by paths, including {}",
1299 
                    missing.join(", ")
1300 
                ));
1301 
            }
957fee30 Andy McCall ↶ 1302 
        }
6d3eefbc Andy McCall ↶ 1303 
        // pre-commit hook: may veto.
1304 
        self.run_hook(
1305 
            "pre-commit",
1306 
            &format!(
1307 
                "{{\"event\":\"pre-commit\",\"message\":{},\"paths\":{}}}",
1308 
                json_str(message),
75167672 Andy McCall 1309 
                json_array(snap.statuses.iter().map(|s| format!(
1310 
                    "{{\"path\":{},\"status\":{}}}",
1311 
                    json_str(&s.path),
1312 
                    json_str(&format!("{:?}", s.status))
1313 
                )))
6d3eefbc Andy McCall ↶ 1314 
            ),
1315 
        )?;
98286406 Andy McCall ↶ 1316 
        let (op_ref, op) = self.head_op()?;
192fad1e Andy McCall ↶ 1317 
        let mut deps: BTreeSet<ChangeHash> = wc_refs;
6d3eefbc Andy McCall ↶ 1318 
        // semantic-deps hook: may add explicit dependencies (spec §8.1).
1319 
        if self.config.semantic_deps {
1320 
            let touched: Vec<String> = snap.statuses.iter().map(|s| s.path.clone()).collect();
1321 
            let by_node = snap.base_snapshot.tree.by_node();
1322 
            let mut candidates: Vec<(ChangeId, ChangeHash, Vec<String>)> = Vec::new();
1323 
            let mut seen: BTreeSet<ChangeHash> = BTreeSet::new();
1324 
            for h in deps.iter().chain(snap.base.maximal(&snap.wc.deps).iter()) {
1325 
                if !seen.insert(*h) {
1326 
                    continue;
1327 
                }
1328 
                let c = self.store.get_change(h)?;
1329 
                let mut paths: BTreeSet<String> = BTreeSet::new();
75167672 Andy McCall 1330 
                for n in c
1331 
                    .tree_ops
1332 
                    .iter()
1333 
                    .map(|o| o.node())
1334 
                    .chain(c.edits.iter().map(|e| e.node()))
1335 
                {
6d3eefbc Andy McCall ↶ 1336 
                    if let Some(e) = by_node.get(&n) {
1337 
                        paths.insert(e.path.clone());
1338 
                    }
1339 
                }
1340 
                candidates.push((c.change_id, *h, paths.into_iter().collect()));
1341 
            }
1342 
            for d in self.semantic_deps(&snap.wc.change_id, &touched, &candidates)? {
1343 
                if snap.base.contains(&d) {
1344 
                    deps.insert(d);
1345 
                }
1346 
            }
1347 
        }
98286406 Andy McCall ↶ 1348 
        // A resolution depends on every change that contributed to the
1349 
        // conflict (spec §8.1). Tree and opaque conflicts are resolved through
1350 
        // ordinary operations whose references already cover the contributors.
1351 
        deps.extend(snap.contributors.iter().copied());
1352 
        for d in extra_deps {
1353 
            if !snap.base.contains(d) {
75167672 Andy McCall 1354 
                return fail(format!(
1355 
                    "--depends-on {d}: not in the working copy's base state"
1356 
                ));
98286406 Andy McCall ↶ 1357 
            }
1358 
            deps.insert(*d);
1359 
        }
1360 
        let mut change = Change {
6f8aeb99 Andy McCall ↶ 1361 
            version: crate::objects::CHANGE_VERSION,
75167672 Andy McCall 1362 
            change_id: if partial {
1363 
                ChangeId::random()
1364 
            } else {
1365 
                snap.wc.change_id
1366 
            },
957fee30 Andy McCall ↶ 1367 
            deps: vec![],
98286406 Andy McCall ↶ 1368 
            author: self.author(),
1369 
            hlc: Hlc::now(),
1370 
            message: message.trim_end().to_string(),
957fee30 Andy McCall ↶ 1371 
            tree_ops,
1372 
            edits,
75167672 Andy McCall 1373 
            kind: if snap.resolves {
1374 
                ChangeKind::Resolution
1375 
            } else {
1376 
                ChangeKind::Normal
1377 
            },
98286406 Andy McCall ↶ 1378 
        };
957fee30 Andy McCall ↶ 1379 
        if partial {
1380 
            // Dependencies come from what the selected part references, not
1381 
            // from the whole working copy.
1382 
            deps = referenced_changes(&change, &snap.base);
1383 
            for d in extra_deps {
1384 
                deps.insert(*d);
1385 
            }
1386 
        }
1387 
        change.deps = deps.into_iter().collect();
98286406 Andy McCall ↶ 1388 
        change.normalize();
f9466021 Andy McCall ↶ 1389 
        let signing = crate::timing::phase("commit: sign and store");
614541c3 Andy McCall ↶ 1390 
        let hash = self.store_signed(&change)?;
f9466021 Andy McCall ↶ 1391 
        drop(signing);
1392 
        let extending = crate::timing::phase("commit: extend the state");
98286406 Andy McCall ↶ 1393
1394 
        // New bookmark heads: maximal elements of old heads plus the new change.
1395 
        // The working copy continues on its own base plus the new change, which
1396 
        // is exactly what the files on disk contain; the bookmark may hold
1397 
        // other heads (and conflicts) the working copy has not checked out.
1398 
        let old_heads = self.heads_of(&op.view, &self.config.bookmark)?;
1399 
        let old_base = snap.base.maximal(&snap.wc.deps);
1400 
        let mut all: Vec<ChangeHash> = old_heads.clone();
1401 
        all.extend(old_base.iter().copied());
1402 
        all.push(hash);
07eb99be Andy McCall ↶ 1403 
        // The snapshot already loaded the working copy's base; when that also
1404 
        // holds every bookmark head, it is the state to extend.
1405 
        let mut st = if old_heads.iter().all(|h| snap.base.contains(h)) {
1406 
            snap.base
1407 
        } else {
1408 
            self.load_state(&all)?
1409 
        };
192fad1e Andy McCall ↶ 1410 
        st.add_hashed(hash, change)?;
98286406 Andy McCall ↶ 1411 
        let mut with_new = old_heads.clone();
1412 
        with_new.push(hash);
1413 
        let new_heads = st.maximal(&with_new);
252e8cb2 Andy McCall ↶ 1414 
        // Automatic checkpoint once enough changes accumulated (spec §11).
75167672 Andy McCall 1415 
        let since = self
1416 
            .latest_checkpoint(&self.config.bookmark)?
1417 
            .map(|(_, cp)| cp.count)
1418 
            .unwrap_or(0);
1419 
        if self.config.checkpoint_interval > 0
1420 
            && (st.len() as u64).saturating_sub(since) >= self.config.checkpoint_interval
1421 
        {
252e8cb2 Andy McCall ↶ 1422 
            let cp_state = self.load_state_full(&new_heads)?;
1423 
            self.write_checkpoint(&self.config.bookmark, &cp_state, &new_heads)?;
1424 
        }
98286406 Andy McCall ↶ 1425 
        let mut wc_base = old_base.clone();
1426 
        wc_base.push(hash);
1427 
        let wc_base = st.maximal(&wc_base);
f9466021 Andy McCall ↶ 1428 
        drop(extending);
98286406 Andy McCall ↶ 1429
8457a6cc Andy McCall ↶ 1430 
        // The graph keys of the files as they now are, from the state the
1431 
        // commit just extended: a file whose graph it built has none, and
1432 
        // is read by the next snapshot; a key the next base does not share
1433 
        // (the bookmark held other heads) is simply not trusted then.
236f35c0 Andy McCall ↶ 1434 
        let keying = crate::timing::phase("commit: graph keys of the files");
f9466021 Andy McCall ↶ 1435 
        let committed_keys =
1436 
            (!partial).then(|| self.committed_graph_keys(&st, &changed_paths, &snap.attributes));
236f35c0 Andy McCall ↶ 1437 
        drop(keying);
f9466021 Andy McCall ↶ 1438 
        let recording = crate::timing::phase("commit: record");
98286406 Andy McCall ↶ 1439 
        let wc = self.new_working_copy(wc_base, "");
1440 
        let wc_hash = self.store.put_change(&wc)?;
1441 
        let mut view = op.view.clone();
75167672 Andy McCall 1442 
        view.bookmarks.insert(
1443 
            self.config.bookmark.clone(),
1444 
            self.store.put_state(&StateObject::new(new_heads))?,
1445 
        );
98286406 Andy McCall ↶ 1446 
        view.workspaces.insert(self.workspace, wc_hash);
1447 
        self.record_op(&[op_ref], &format!("oxy commit -m {message:?}"), view)?;
f9466021 Andy McCall ↶ 1448 
        drop(recording);
236f35c0 Andy McCall ↶ 1449 
        let rewriting = crate::timing::phase("commit: manifest");
957fee30 Andy McCall ↶ 1450 
        if partial {
1451 
            // The rest is rebuilt from disk by the next snapshot.
1452 
            self.drop_manifest();
1453 
        } else {
86e1824f Andy McCall ↶ 1454 
            let conflicted = snap.base_snapshot.is_conflicted() && !snap.resolves;
1455 
            // The files on disk are now exactly the new base's renderings:
1456 
            // record their graph keys, so the next snapshot neither reads nor
1457 
            // renders them (oxygen#35).
1458 
            let keys = if conflicted {
1459 
                HashMap::new()
1460 
            } else {
8457a6cc Andy McCall ↶ 1461 
                committed_keys.unwrap_or_default()
86e1824f Andy McCall ↶ 1462 
            };
236f35c0 Andy McCall ↶ 1463 
            match pending {
1464 
                Some(m) => self.write_pending_manifest(m, conflicted, keys)?,
1465 
                None => self.rewrite_manifest(conflicted, false, &keys)?,
1466 
            }
957fee30 Andy McCall ↶ 1467 
        }
236f35c0 Andy McCall ↶ 1468 
        drop(rewriting);
1469 
        let packing = crate::timing::phase("commit: autopack");
d02e545d Andy McCall ↶ 1470 
        self.maybe_autopack()?;
236f35c0 Andy McCall ↶ 1471 
        drop(packing);
1472 
        let _hook = crate::timing::phase("commit: hook");
75167672 Andy McCall 1473 
        let _ = self.run_hook(
1474 
            "post-commit",
1475 
            &format!(
1476 
                "{{\"event\":\"post-commit\",\"change_id\":{},\"hash\":{},\"message\":{}}}",
1477 
                json_str(&snap.wc.change_id.to_hex()),
1478 
                json_str(&hash.to_hex()),
1479 
                json_str(message)
1480 
            ),
1481 
        );
75051941 Andy McCall ↶ 1482 
        Ok((hash, partial))
98286406 Andy McCall ↶ 1483 
    }
1484 
}
1485
957fee30 Andy McCall ↶ 1486 
/// What part of the working copy a commit takes (spec §12.3).
1487 
#[derive(Clone, Debug)]
1488 
pub enum Selection {
1489 
    All,
1490 
    /// Every operation and edit on nodes under these paths.
1491 
    Paths(Vec<String>),
1492 
    /// Hunk indices as `Repo::hunks` numbers them; dependent hunks are added.
1493 
    Hunks(Vec<usize>),
1494 
}
1495
1496 
/// One selectable piece of the working copy: a tree operation or an edit.
1497 
#[derive(Clone, Debug)]
1498 
pub struct Hunk {
1499 
    pub index: usize,
1500 
    pub path: String,
1501 
    /// Rendered like `oxy diff`.
1502 
    pub text: String,
1503 
}
1504
1505 
/// The tree operations and edits on nodes under `paths`, plus the
1506 
/// additions of any new parent directories they need.
75167672 Andy McCall 1507 
fn select_paths(
1508 
    wc: &Change,
1509 
    base_snapshot: &Snapshot,
1510 
    paths: &[String],
1511 
) -> (Vec<TreeOp>, Vec<Edit>) {
1512 
    let selected = |p: &str| {
1513 
        paths
1514 
            .iter()
1515 
            .any(|x| p == x || p.starts_with(&format!("{}/", x.trim_end_matches('/'))))
1516 
    };
957fee30 Andy McCall ↶ 1517 
    let by_node = base_snapshot.tree.by_node();
1518 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
1519 
    for op in &wc.tree_ops {
75167672 Andy McCall 1520 
        if let TreeOp::AddNode {
1521 
            node,
1522 
            parent,
1523 
            name,
1524 
            kind,
1525 
            ..
1526 
        } = op
1527 
        {
957fee30 Andy McCall ↶ 1528 
            new_names.insert(*node, (*parent, name.clone(), *kind));
1529 
        }
1530 
    }
1531 
    let mut nodes: BTreeSet<NodeId> = BTreeSet::new();
75167672 Andy McCall 1532 
    for n in wc
1533 
        .tree_ops
1534 
        .iter()
1535 
        .map(|o| o.node())
1536 
        .chain(wc.edits.iter().map(|e| e.node()))
1537 
    {
957fee30 Andy McCall ↶ 1538 
        if path_of_node(&n, &by_node, &new_names).is_some_and(|p| selected(&p)) {
1539 
            nodes.insert(n);
1540 
        }
1541 
    }
c2e1d090 Andy McCall 1542 
    // The nodes shown as one with a selected node are retired with it.
1543 
    for (dup, shown) in &base_snapshot.tree.unified {
1544 
        if nodes.contains(shown) {
1545 
            nodes.insert(*dup);
1546 
        }
1547 
    }
957fee30 Andy McCall ↶ 1548 
    // New parents of selected new nodes come along.
1549 
    let mut changed = true;
1550 
    while changed {
1551 
        changed = false;
1552 
        for (n, (parent, _, _)) in &new_names {
1553 
            if nodes.contains(n) && new_names.contains_key(parent) && nodes.insert(*parent) {
1554 
                changed = true;
1555 
            }
1556 
        }
1557 
    }
75167672 Andy McCall 1558 
    let ops = wc
1559 
        .tree_ops
1560 
        .iter()
1561 
        .filter(|o| nodes.contains(&o.node()))
1562 
        .cloned()
1563 
        .collect();
1564 
    let edits = wc
1565 
        .edits
1566 
        .iter()
1567 
        .filter(|e| nodes.contains(&e.node()))
1568 
        .cloned()
1569 
        .collect();
957fee30 Andy McCall ↶ 1570 
    (ops, edits)
1571 
}
1572
1573 
/// The hunks `indices` name plus whatever they refer to within the working
1574 
/// copy (the parent directory of an added node, the insert that introduced
1575 
/// an anchor line), renumbered so the result is a well-formed change.
c2e1d090 Andy McCall 1576 
/// The nodes `unified` shows as one with another (a base's
1577 
/// [`TreeView::unified`](crate::tree::TreeView::unified)) are retired with
1578 
/// any hunk of the node shown.
1579 
fn select_hunks(
1580 
    wc: &Change,
1581 
    indices: &[usize],
1582 
    unified: &BTreeMap<NodeId, NodeId>,
1583 
) -> (Vec<TreeOp>, Vec<Edit>) {
957fee30 Andy McCall ↶ 1584 
    let n_ops = wc.tree_ops.len();
1585 
    let mut ops_sel = vec![false; n_ops];
1586 
    let mut edits_sel = vec![false; wc.edits.len()];
1587 
    for &i in indices {
1588 
        if i < n_ops {
1589 
            ops_sel[i] = true;
1590 
        } else if i - n_ops < wc.edits.len() {
1591 
            edits_sel[i - n_ops] = true;
1592 
        }
1593 
    }
1594 
    // Atom ranges each insert introduces, per node, in the working copy's numbering.
1595 
    let mut per_node: HashMap<NodeId, u32> = HashMap::new();
1596 
    let mut insert_ranges: Vec<Option<(NodeId, u32, u32)>> = Vec::new();
1597 
    for e in &wc.edits {
1598 
        insert_ranges.push(match e {
1599 
            Edit::Insert { node, lines, .. } => {
1600 
                let start = per_node.entry(*node).or_default();
1601 
                let r = (*node, *start, *start + lines.len() as u32);
1602 
                *start += lines.len() as u32;
1603 
                Some(r)
1604 
            }
1605 
            _ => None,
1606 
        });
1607 
    }
75167672 Andy McCall 1608 
    let add_index: HashMap<NodeId, usize> = wc
1609 
        .tree_ops
1610 
        .iter()
1611 
        .enumerate()
1612 
        .filter_map(|(i, o)| {
1613 
            if let TreeOp::AddNode { node, .. } = o {
1614 
                Some((*node, i))
1615 
            } else {
1616 
                None
1617 
            }
1618 
        })
1619 
        .collect();
1620 
    let insert_for = |node: &NodeId, index: u32| -> Option<usize> {
1621 
        insert_ranges
1622 
            .iter()
1623 
            .position(|r| matches!(r, Some((n, a, b)) if n == node && *a <= index && index < *b))
1624 
    };
957fee30 Andy McCall ↶ 1625 
    // Closure.
1626 
    let mut changed = true;
1627 
    while changed {
1628 
        changed = false;
1629 
        for i in 0..n_ops {
1630 
            if !ops_sel[i] {
1631 
                continue;
1632 
            }
d8c086ca Andy McCall ↶ 1633 
            if let TreeOp::AddNode { node, parent, .. } = &wc.tree_ops[i] {
957fee30 Andy McCall ↶ 1634 
                if let Some(&p) = add_index.get(parent) {
1635 
                    if !ops_sel[p] {
1636 
                        ops_sel[p] = true;
1637 
                        changed = true;
1638 
                    }
1639 
                }
d8c086ca Andy McCall ↶ 1640 
                // A new node's content belongs with its addition: nobody
1641 
                // wants an empty file committed apart from its lines.
1642 
                for (j, e) in wc.edits.iter().enumerate() {
1643 
                    if e.node() == *node && !edits_sel[j] {
1644 
                        edits_sel[j] = true;
1645 
                        changed = true;
1646 
                    }
1647 
                }
957fee30 Andy McCall ↶ 1648 
            }
1649 
        }
1650 
        for j in 0..wc.edits.len() {
1651 
            if !edits_sel[j] {
1652 
                continue;
1653 
            }
1654 
            let e = &wc.edits[j];
1655 
            if let Some(&a) = add_index.get(&e.node()) {
1656 
                if !ops_sel[a] {
1657 
                    ops_sel[a] = true;
1658 
                    changed = true;
1659 
                }
1660 
            }
75167672 Andy McCall 1661 
            if let Edit::Insert {
1662 
                node,
1663 
                after,
1664 
                before,
1665 
                ..
1666 
            } = e
1667 
            {
957fee30 Andy McCall ↶ 1668 
                for r in [after, before] {
1669 
                    if let AtomRef::Atom { change, index } = r {
1670 
                        if change.is_zero() {
1671 
                            if let Some(k) = insert_for(node, *index) {
1672 
                                if !edits_sel[k] {
1673 
                                    edits_sel[k] = true;
1674 
                                    changed = true;
1675 
                                }
1676 
                            }
1677 
                        }
1678 
                    }
c2e1d090 Andy McCall 1679 
                }
1680 
            }
1681 
        }
1682 
    }
1683 
    // The retirement of a node shown as one with a selected node.
1684 
    if !unified.is_empty() {
1685 
        let chosen: HashSet<NodeId> = (0..n_ops)
1686 
            .filter(|i| ops_sel[*i])
1687 
            .map(|i| wc.tree_ops[i].node())
1688 
            .chain(
1689 
                (0..wc.edits.len())
1690 
                    .filter(|j| edits_sel[*j])
1691 
                    .map(|j| wc.edits[j].node()),
1692 
            )
1693 
            .collect();
1694 
        for (i, o) in wc.tree_ops.iter().enumerate() {
1695 
            if let TreeOp::RemoveNode { node, .. } = o {
1696 
                if unified.get(node).is_some_and(|s| chosen.contains(s)) {
1697 
                    ops_sel[i] = true;
957fee30 Andy McCall ↶ 1698 
                }
1699 
            }
1700 
        }
1701 
    }
1702 
    // Renumber self-references: old atom index → new index among selected inserts.
1703 
    let mut remap: HashMap<(NodeId, u32), u32> = HashMap::new();
1704 
    let mut new_counts: HashMap<NodeId, u32> = HashMap::new();
1705 
    for (j, r) in insert_ranges.iter().enumerate() {
1706 
        if let (true, Some((node, a, b))) = (edits_sel[j], r) {
1707 
            let next = new_counts.entry(*node).or_default();
1708 
            for old in *a..*b {
1709 
                remap.insert((*node, old), *next);
1710 
                *next += 1;
1711 
            }
1712 
        }
1713 
    }
1714 
    let map_ref = |r: AtomRef, node: &NodeId| -> AtomRef {
1715 
        match r {
75167672 Andy McCall 1716 
            AtomRef::Atom { change, index } if change.is_zero() => AtomRef::Atom {
1717 
                change,
1718 
                index: remap.get(&(*node, index)).copied().unwrap_or(index),
1719 
            },
957fee30 Andy McCall ↶ 1720 
            other => other,
1721 
        }
1722 
    };
75167672 Andy McCall 1723 
    let ops = wc
1724 
        .tree_ops
1725 
        .iter()
1726 
        .enumerate()
1727 
        .filter(|(i, _)| ops_sel[*i])
1728 
        .map(|(_, o)| o.clone())
1729 
        .collect();
957fee30 Andy McCall ↶ 1730 
    let edits = wc
1731 
        .edits
1732 
        .iter()
1733 
        .enumerate()
1734 
        .filter(|(j, _)| edits_sel[*j])
1735 
        .map(|(_, e)| match e {
75167672 Andy McCall 1736 
            Edit::Insert {
1737 
                node,
1738 
                after,
1739 
                before,
1740 
                lines,
1741 
            } => Edit::Insert {
1742 
                node: *node,
1743 
                after: map_ref(*after, node),
1744 
                before: map_ref(*before, node),
1745 
                lines: lines.clone(),
1746 
            },
957fee30 Andy McCall ↶ 1747 
            other => other.clone(),
1748 
        })
1749 
        .collect();
1750 
    (ops, edits)
1751 
}
1752
1753 
/// Render each tree operation and edit of `change` on its own, numbered.
1754 
pub fn render_hunks(change: &Change, base: &State, base_snapshot: &Snapshot) -> Vec<Hunk> {
1755 
    let by_node = base_snapshot.tree.by_node();
1756 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
1757 
    for op in &change.tree_ops {
75167672 Andy McCall 1758 
        if let TreeOp::AddNode {
1759 
            node,
1760 
            parent,
1761 
            name,
1762 
            kind,
1763 
            ..
1764 
        } = op
1765 
        {
957fee30 Andy McCall ↶ 1766 
            new_names.insert(*node, (*parent, name.clone(), *kind));
1767 
        }
1768 
    }
1769 
    let mut out = Vec::new();
1770 
    for (i, op) in change.tree_ops.iter().enumerate() {
c2e1d090 Andy McCall 1771 
        // A retired duplicate goes with the node shown: no hunk of its own.
1772 
        if is_unified_retirement(op, base_snapshot) {
1773 
            continue;
1774 
        }
75167672 Andy McCall 1775 
        let one = Change {
1776 
            tree_ops: vec![op.clone()],
1777 
            edits: vec![],
1778 
            ..change.clone()
1779 
        };
957fee30 Andy McCall ↶ 1780 
        let path = path_of_node(&op.node(), &by_node, &new_names).unwrap_or_default();
75167672 Andy McCall 1781 
        out.push(Hunk {
1782 
            index: i,
1783 
            path,
1784 
            text: render_change(&one, base, base_snapshot),
1785 
        });
957fee30 Andy McCall ↶ 1786 
    }
1787 
    for (j, e) in change.edits.iter().enumerate() {
75167672 Andy McCall 1788 
        let one = Change {
1789 
            tree_ops: vec![],
1790 
            edits: vec![e.clone()],
1791 
            ..change.clone()
1792 
        };
957fee30 Andy McCall ↶ 1793 
        let path = path_of_node(&e.node(), &by_node, &new_names).unwrap_or_default();
75167672 Andy McCall 1794 
        out.push(Hunk {
1795 
            index: change.tree_ops.len() + j,
1796 
            path,
1797 
            text: render_change(&one, base, base_snapshot),
1798 
        });
957fee30 Andy McCall ↶ 1799 
    }
1800 
    out
1801 
}
1802
c6aa5bd9 Andy McCall ↶ 1803 
/// One file's part of a diff, for `oxy diff --json`.
1804 
#[derive(Clone, Debug)]
1805 
pub struct FileDiff {
1806 
    pub path: String,
1807 
    /// `added`, `removed`, `modified`, `renamed`, `mode`, `replaced`.
1808 
    pub kind: String,
1809 
    pub hunks: Vec<DiffHunk>,
1810 
    /// For renames, the previous path.
1811 
    pub from: Option<String>,
1812 
}
1813
1814 
/// Lines inserted after `base_line` (0 = before the first line) and lines
1815 
/// removed, with their base line numbers.
1816 
#[derive(Clone, Debug, Default)]
1817 
pub struct DiffHunk {
1818 
    pub base_line: u32,
1819 
    pub added: Vec<Vec<u8>>,
1820 
    pub removed: Vec<(u32, Vec<u8>)>,
1821 
}
1822
1823 
impl Repo {
1824 
    /// The content of `path` in a state: `heads`, or the working copy's base.
1825 
    pub fn file_at(&self, path: &str, heads: Option<&[ChangeHash]>) -> Result<Vec<u8>> {
19303b4a Andy McCall ↶ 1826 
        // The state's history and tree, then the one file's graph: nothing
1827 
        // else is built or rendered (oxygen#92).
1828 
        // The working copy's base, and a state named by its heads (`-r
1829 
        // main` on main), come from the workspace's state index.
1830 
        let base = if self.is_bare() {
1831 
            None
1832 
        } else {
1833 
            Some(self.working_copy()?.1.deps)
1834 
        };
1835 
        let same = |a: &[ChangeHash], b: &[ChangeHash]| {
1836 
            let (mut a, mut b) = (a.to_vec(), b.to_vec());
1837 
            a.sort();
1838 
            b.sort();
1839 
            a == b
1840 
        };
1841 
        let mut state = match (heads, base) {
1842 
            (Some(h), Some(b)) if same(h, &b) => self.load_state_indexed(&b)?,
1843 
            (Some(h), _) => self.load_state_history(h)?,
1844 
            (None, Some(b)) => self.load_state_indexed(&b)?,
1845 
            (None, None) => return fail("this is a bare repository: it has no working copy"),
1846 
        };
c2e1d090 Andy McCall 1847 
        let view = self.tree_view(&mut state)?;
19303b4a Andy McCall ↶ 1848 
        let Some(node) = view.entries.get(path).map(|e| e.node) else {
1849 
            return fail(format!("{path} is not in that state"));
c6aa5bd9 Andy McCall ↶ 1850 
        };
19303b4a Andy McCall ↶ 1851 
        self.load_graphs_of(&mut state, &[node])?;
1852 
        let snapshot = state.materialize_paths(view, [path]);
75167672 Andy McCall 1853 
        let Some(f) = snapshot.files.get(path) else {
1854 
            return fail(format!("{path} is not in that state"));
1855 
        };
c6aa5bd9 Andy McCall ↶ 1856 
        match &f.content {
1857 
            FileContent::Dir => fail(format!("{path} is a directory")),
1858 
            FileContent::Text { bytes, .. } => Ok(bytes.clone()),
1859 
            FileContent::Opaque { blobs } => match blobs.first() {
1860 
                Some(b) => self.read_blob(b),
1861 
                None => Ok(vec![]),
1862 
            },
1863 
        }
1864 
    }
1865
1866 
    /// Restore `paths` (files or directories) to the working copy's base:
1867 
    /// edits are undone, additions removed, removals put back. Returns the
1868 
    /// number of paths touched.
1869 
    pub fn discard(&self, paths: &[String]) -> Result<usize> {
6caacfff Andy McCall ↶ 1870 
        let _wc = self.lock_working_copy()?;
8ab265dc Andy McCall ↶ 1871 
        let snap = self.snapshot_status("oxy discard")?;
75167672 Andy McCall 1872 
        let under = |p: &str| {
1873 
            paths
1874 
                .iter()
1875 
                .any(|x| p == x || p.starts_with(&format!("{}/", x.trim_end_matches('/'))))
1876 
        };
c6aa5bd9 Andy McCall ↶ 1877 
        let attrs = Repo::state_attributes(&snap.base_snapshot);
1878 
        let mut n = 0;
1879 
        for s in &snap.statuses {
1880 
            if !under(&s.path) {
1881 
                continue;
1882 
            }
1883 
            match snap.base_snapshot.files.get(&s.path) {
1884 
                None => {
1885 
                    // Added (or the new name of a rename): remove it.
1886 
                    let full = self.root.join(&s.path);
1887 
                    if full.is_dir() {
1888 
                        let _ = fs::remove_dir_all(&full);
1889 
                    } else {
1890 
                        let _ = fs::remove_file(&full);
1891 
                    }
1892 
                    n += 1;
1893 
                }
1894 
                Some(f) => {
1895 
                    match &f.content {
1896 
                        FileContent::Dir => fs::create_dir_all(self.root.join(&s.path))?,
1897 
                        FileContent::Text { bytes, .. } => {
75167672 Andy McCall 1898 
                            let on_disk: std::borrow::Cow<[u8]> = if attrs.lookup(&s.path).eol
1899 
                                == Some(crate::attributes::Eol::Crlf)
1900 
                            {
1901 
                                crate::attributes::to_crlf(bytes).into()
1902 
                            } else {
1903 
                                bytes.as_slice().into()
1904 
                            };
c6aa5bd9 Andy McCall ↶ 1905 
                            self.write_file(&s.path, &on_disk, f.mode.exec)?;
1906 
                        }
1907 
                        FileContent::Opaque { blobs } => {
1908 
                            let bytes = match blobs.first() {
1909 
                                Some(b) => self.read_blob(b)?,
1910 
                                None => vec![],
1911 
                            };
1912 
                            if f.kind == NodeKind::Symlink {
1913 
                                self.write_symlink(&s.path, &bytes)?;
1914 
                            } else if f.kind == NodeKind::File {
1915 
                                self.write_file(&s.path, &bytes, f.mode.exec)?;
1916 
                            }
1917 
                        }
1918 
                    }
1919 
                    n += 1;
1920 
                }
1921 
            }
1922 
        }
1923 
        // A rename's old name is not in the statuses; put it back too.
1924 
        for (p, f) in &snap.base_snapshot.files {
75167672 Andy McCall 1925 
            if under(p)
1926 
                && !self.root.join(p).exists()
1927 
                && f.kind != NodeKind::Dir
1928 
                && f.kind != NodeKind::Subrepo
1929 
                && !snap.statuses.iter().any(|s| s.path == *p)
1930 
            {
c6aa5bd9 Andy McCall ↶ 1931 
                let bytes = match &f.content {
1932 
                    FileContent::Text { bytes, .. } => bytes.clone(),
75167672 Andy McCall 1933 
                    FileContent::Opaque { blobs } => blobs
1934 
                        .first()
1935 
                        .map(|b| self.read_blob(b))
1936 
                        .transpose()?
1937 
                        .unwrap_or_default(),
c6aa5bd9 Andy McCall ↶ 1938 
                    FileContent::Dir => continue,
1939 
                };
75167672 Andy McCall 1940 
                if f.kind == NodeKind::Symlink {
1941 
                    self.write_symlink(p, &bytes)?
1942 
                } else {
1943 
                    self.write_file(p, &bytes, f.mode.exec)?
1944 
                }
c6aa5bd9 Andy McCall ↶ 1945 
                n += 1;
1946 
            }
1947 
        }
1948 
        self.drop_manifest();
1949 
        Ok(n)
1950 
    }
1951
1952 
    /// Undo the working-copy hunks `indices` (as `hunks` numbers them) on
1953 
    /// disk, keeping the others: the files are rewritten from the base plus
1954 
    /// the remaining hunks.
1955 
    pub fn discard_hunks(&self, indices: &[usize]) -> Result<usize> {
6caacfff Andy McCall ↶ 1956 
        let _wc = self.lock_working_copy()?;
c6aa5bd9 Andy McCall ↶ 1957 
        let (snap, hunks) = self.hunks()?;
75167672 Andy McCall 1958 
        let keep: Vec<usize> = hunks
1959 
            .iter()
1960 
            .map(|h| h.index)
1961 
            .filter(|i| !indices.contains(i))
1962 
            .collect();
c2e1d090 Andy McCall 1963 
        let (ops, edits) = select_hunks(&snap.wc, &keep, &snap.base_snapshot.tree.unified);
c6aa5bd9 Andy McCall ↶ 1964 
        let dropped_nodes: BTreeSet<NodeId> = hunks
1965 
            .iter()
1966 
            .filter(|h| indices.contains(&h.index))
75167672 Andy McCall 1967 
            .map(|h| {
1968 
                if h.index < snap.wc.tree_ops.len() {
1969 
                    snap.wc.tree_ops[h.index].node()
1970 
                } else {
1971 
                    snap.wc.edits[h.index - snap.wc.tree_ops.len()].node()
1972 
                }
1973 
            })
c6aa5bd9 Andy McCall ↶ 1974 
            .collect();
1975 
        let mut partial = snap.wc.clone();
1976 
        partial.tree_ops = ops;
1977 
        partial.edits = edits;
1978 
        partial.normalize();
1979 
        let mut st = snap.base.clone();
1980 
        st.add_unchecked(partial);
8ab265dc Andy McCall ↶ 1981 
        // Every file of the target is rendered: build the graphs the
1982 
        // partial snapshot left out.
1983 
        self.load_all_graphs(&mut st)?;
c6aa5bd9 Andy McCall ↶ 1984 
        let target = st.materialize();
1985 
        let by_node_target = target.tree.by_node();
1986 
        let by_node_base = snap.base_snapshot.tree.by_node();
1987 
        let attrs = Repo::state_attributes(&target);
1988 
        let mut n = 0;
1989 
        for node in dropped_nodes {
1990 
            match by_node_target.get(&node) {
1991 
                Some(e) => {
1992 
                    let f = &target.files[&e.path];
1993 
                    match &f.content {
1994 
                        FileContent::Dir => {}
1995 
                        FileContent::Text { bytes, .. } => {
75167672 Andy McCall 1996 
                            let on_disk: std::borrow::Cow<[u8]> = if attrs.lookup(&e.path).eol
1997 
                                == Some(crate::attributes::Eol::Crlf)
1998 
                            {
1999 
                                crate::attributes::to_crlf(bytes).into()
2000 
                            } else {
2001 
                                bytes.as_slice().into()
2002 
                            };
c6aa5bd9 Andy McCall ↶ 2003 
                            self.write_file(&e.path, &on_disk, f.mode.exec)?;
2004 
                        }
2005 
                        FileContent::Opaque { blobs } => {
75167672 Andy McCall 2006 
                            let bytes = blobs
2007 
                                .first()
2008 
                                .map(|b| self.read_blob(b))
2009 
                                .transpose()?
2010 
                                .unwrap_or_default();
2011 
                            if f.kind == NodeKind::Symlink {
2012 
                                self.write_symlink(&e.path, &bytes)?
2013 
                            } else {
2014 
                                self.write_file(&e.path, &bytes, f.mode.exec)?
2015 
                            }
c6aa5bd9 Andy McCall ↶ 2016 
                        }
2017 
                    }
2018 
                    // A discarded rename: the new name goes.
75167672 Andy McCall 2019 
                    if let Some(old_path) = snap
2020 
                        .statuses
2021 
                        .iter()
2022 
                        .find(|s| {
2023 
                            s.status == ChangeStatus::Renamed && !target.files.contains_key(&s.path)
2024 
                        })
2025 
                        .map(|s| s.path.clone())
2026 
                    {
c6aa5bd9 Andy McCall ↶ 2027 
                        let _ = fs::remove_file(self.root.join(old_path));
2028 
                    }
2029 
                    n += 1;
2030 
                }
2031 
                None => {
2032 
                    // The node exists in neither target nor base with a live path: an addition to remove.
75167672 Andy McCall 2033 
                    if let Some(added) = snap.statuses.iter().find(|s| {
2034 
                        s.status == ChangeStatus::Added
2035 
                            && !target.files.contains_key(&s.path)
2036 
                            && !by_node_base.contains_key(&node)
2037 
                    }) {
c6aa5bd9 Andy McCall ↶ 2038 
                        let full = self.root.join(&added.path);
75167672 Andy McCall 2039 
                        if full.is_dir() {
2040 
                            let _ = fs::remove_dir_all(&full);
2041 
                        } else {
2042 
                            let _ = fs::remove_file(&full);
2043 
                        }
c6aa5bd9 Andy McCall ↶ 2044 
                        n += 1;
2045 
                    }
2046 
                }
2047 
            }
2048 
        }
2049 
        self.drop_manifest();
2050 
        Ok(n)
2051 
    }
2052
2053 
    /// The diff of the working copy (`None`) or of a change, per file with
2054 
    /// line-numbered hunks (spec: `oxy diff --json`).
2055 
    pub fn diff_files(&self, change: Option<&ChangeHash>) -> Result<Vec<FileDiff>> {
8ab265dc Andy McCall ↶ 2056 
        let (c, base, base_snapshot) = match change {
c6aa5bd9 Andy McCall ↶ 2057 
            Some(h) => {
2058 
                let c = self.store.get_change(h)?;
8ab265dc Andy McCall ↶ 2059 
                let (base, snapshot) = self.change_base(&c)?;
2060 
                (c, base, snapshot)
c6aa5bd9 Andy McCall ↶ 2061 
            }
74b4a3c6 Andy McCall ↶ 2062 
            None => match self.working_copy_with_base_unless_unchanged()? {
2063 
                Some(all) => all,
2064 
                None => return Ok(Vec::new()),
2065 
            },
c6aa5bd9 Andy McCall ↶ 2066 
        };
2067 
        Ok(diff_files_of(&c, &base, &base_snapshot))
e226a6cc Andy McCall ↶ 2068 
    }
2069
2070 
    /// The snapshot a command would take, without taking it, when the
2071 
    /// status cache holds: every file is as the last snapshot left it, so
2072 
    /// that snapshot's working-copy change is the answer. Its base comes
2073 
    /// from the state index, the `.oxyattributes` files from disk. `None`
2074 
    /// when a snapshot must be taken, and when there is nothing to commit
2075 
    /// (the snapshot then says so). A conflicted base never gets here: the
2076 
    /// cache does not hold for one.
2077 
    pub(crate) fn snapshot_from_cache(&self) -> Result<Option<SnapshotResult>> {
2078 
        let Some(statuses) = self.status_by_manifest()? else {
2079 
            return Ok(None);
2080 
        };
2081 
        if statuses.is_empty() {
2082 
            return Ok(None);
2083 
        }
2084 
        let _t = crate::timing::phase("snapshot: from the status cache");
2085 
        let (wc_hash, wc) = self.working_copy()?;
2086 
        let base = self.load_state_indexed(&wc.deps)?;
2087 
        let (base, base_snapshot) = self.render_base_for(&wc, base)?;
2088 
        let attribute_files: Vec<(String, Vec<u8>)> = match self.read_manifest()? {
2089 
            Some((_, _, m)) => m
2090 
                .entries
2091 
                .keys()
2092 
                .filter(|p| is_attributes(p))
2093 
                .filter_map(|p| fs::read(self.root.join(p)).ok().map(|b| (p.clone(), b)))
2094 
                .collect(),
2095 
            None => Vec::new(),
2096 
        };
2097 
        let attributes = crate::attributes::Attributes::from_files(
2098 
            attribute_files
2099 
                .iter()
2100 
                .map(|(p, b)| (p.as_str(), b.as_slice())),
2101 
        );
2102 
        Ok(Some(SnapshotResult {
2103 
            wc_hash,
2104 
            wc,
2105 
            base,
2106 
            base_snapshot,
2107 
            statuses,
2108 
            changed: false,
2109 
            resolves: false,
2110 
            contributors: Vec::new(),
2111 
            attributes,
2112 
            untracked_nested: self.untracked_nested.borrow().clone(),
236f35c0 Andy McCall ↶ 2113 
            manifest: None,
e226a6cc Andy McCall ↶ 2114 
        }))
77445673 Andy McCall ↶ 2115 
    }
2116
dd6147dc Andy McCall ↶ 2117 
    /// The working-copy change with its base state and a snapshot of the
2118 
    /// base, for showing what the working copy changes ([`render_change`],
2119 
    /// [`diff_files_of`]). While the status cache holds (every file as the
2120 
    /// last snapshot left it), the stored working-copy change is the
2121 
    /// answer, no snapshot is taken, and the base's snapshot holds its tree
2122 
    /// but no file; otherwise a snapshot makes all three.
77445673 Andy McCall ↶ 2123 
    pub fn working_copy_with_base(&self) -> Result<(Change, State, Snapshot)> {
74b4a3c6 Andy McCall ↶ 2124 
        match self.working_copy_with_base_unless_unchanged()? {
2125 
            Some(all) => Ok(all),
2126 
            None => {
2127 
                let (_, wc) = self.working_copy()?;
2128 
                let base = self.load_state_indexed(&wc.deps)?;
2129 
                let (base, snapshot) = self.tree_base_for(&wc, base)?;
2130 
                Ok((wc, base, snapshot))
2131 
            }
2132 
        }
2133 
    }
2134
2135 
    /// [`Repo::working_copy_with_base`], or `None` when the status cache
2136 
    /// holds and the working-copy change changes nothing: its diff is
2137 
    /// empty, and the base is not loaded for it (oxygen#94).
2138 
    pub fn working_copy_with_base_unless_unchanged(
2139 
        &self,
2140 
    ) -> Result<Option<(Change, State, Snapshot)>> {
77445673 Andy McCall ↶ 2141 
        if self.status_by_manifest()?.is_some() {
2142 
            let (_, wc) = self.working_copy()?;
74b4a3c6 Andy McCall ↶ 2143 
            if wc.tree_ops.is_empty() && wc.edits.is_empty() {
2144 
                return Ok(None);
2145 
            }
77445673 Andy McCall ↶ 2146 
            let base = self.load_state_indexed(&wc.deps)?;
dd6147dc Andy McCall ↶ 2147 
            let (base, snapshot) = self.tree_base_for(&wc, base)?;
74b4a3c6 Andy McCall ↶ 2148 
            return Ok(Some((wc, base, snapshot)));
77445673 Andy McCall ↶ 2149 
        }
2150 
        let snap = self.snapshot_status("oxy diff")?;
74b4a3c6 Andy McCall ↶ 2151 
        Ok(Some((snap.wc, snap.base, snap.base_snapshot)))
c6aa5bd9 Andy McCall ↶ 2152 
    }
2153
2154 
    /// The log restricted to changes touching `path` (its node, so renames
2155 
    /// are followed), on the current bookmark.
2156 
    pub fn log_for_path(&self, path: &str) -> Result<Vec<LogEntry>> {
e02bf0cd Andy McCall ↶ 2157 
        let heads = self.bookmark_heads()?;
2158 
        if self.is_bare() {
2159 
            return self.log_for_path_in(path, &heads);
2160 
        }
c6aa5bd9 Andy McCall ↶ 2161 
        let entries = self.log()?;
2162 
        let (_, wc) = self.working_copy()?;
170bb436 Andy McCall ↶ 2163 
        // The path's node from the trees alone: the base's (from the
2164 
        // workspace's state index), else the bookmark's. Nothing is
2165 
        // rendered.
c2e1d090 Andy McCall 2166 
        let node_in = |mut state: State| -> Result<Option<HashSet<NodeId>>> {
2167 
            Ok(nodes_at(&self.tree_view(&mut state)?, path))
2168 
        };
2169 
        let nodes = match node_in(self.load_state_indexed(&wc.deps)?)? {
170bb436 Andy McCall ↶ 2170 
            Some(n) => Some(n),
c2e1d090 Andy McCall 2171 
            None => node_in(self.load_state_history(&heads)?)?,
170bb436 Andy McCall ↶ 2172 
        };
c2e1d090 Andy McCall 2173 
        let Some(nodes) = nodes else {
75167672 Andy McCall 2174 
            return fail(format!("{path} is not tracked"));
2175 
        };
c2e1d090 Andy McCall 2176 
        Ok(self.entries_touching_up_to(entries, nodes, None))
e02bf0cd Andy McCall ↶ 2177 
    }
2178
2179 
    /// The log of the state with these heads restricted to changes touching
2180 
    /// `path` (its node, so renames are followed).
2181 
    pub fn log_for_path_in(&self, path: &str, heads: &[ChangeHash]) -> Result<Vec<LogEntry>> {
e8d38e47 Andy McCall ↶ 2182 
        self.log_for_path_of(&self.load_state_history(heads)?, path, None)
bc489b6c Andy McCall ↶ 2183 
    }
2184
2185 
    /// The changes of a loaded state touching `path`, newest first, at most
2186 
    /// `limit` of them. The path is found in the state's tree alone, so no
2187 
    /// file is rendered, and the walk stops once `limit` are found.
2188 
    pub fn log_for_path_of(
2189 
        &self,
2190 
        state: &State,
2191 
        path: &str,
2192 
        limit: Option<usize>,
2193 
    ) -> Result<Vec<LogEntry>> {
c2e1d090 Andy McCall 2194 
        let Some(nodes) = nodes_at(&state.view(), path) else {
75167672 Andy McCall 2195 
            return fail(format!("{path} is not in that state"));
2196 
        };
c2e1d090 Andy McCall 2197 
        Ok(self.entries_touching_up_to(self.log_of(state), nodes, limit))
bc489b6c Andy McCall ↶ 2198 
    }
2199
2200 
    fn entries_touching_up_to(
2201 
        &self,
2202 
        entries: Vec<LogEntry>,
c2e1d090 Andy McCall 2203 
        wanted: HashSet<NodeId>,
bc489b6c Andy McCall ↶ 2204 
        limit: Option<usize>,
2205 
    ) -> Vec<LogEntry> {
170bb436 Andy McCall ↶ 2206 
        let mut touched = self.touched();
2207 
        if limit.is_none() {
2208 
            let hashes: Vec<ChangeHash> = entries.iter().map(|e| e.hash).collect();
2209 
            touched.fill(self, &hashes);
2210 
        }
c6aa5bd9 Andy McCall ↶ 2211 
        let mut out = Vec::new();
2212 
        for e in entries {
bc489b6c Andy McCall ↶ 2213 
            if limit.is_some_and(|n| out.len() >= n) {
2214 
                break;
2215 
            }
170bb436 Andy McCall ↶ 2216 
            if touched.touches(self, &e.hash, &wanted) {
c6aa5bd9 Andy McCall ↶ 2217 
                out.push(e);
2218 
            }
2219 
        }
170bb436 Andy McCall ↶ 2220 
        touched.save();
e02bf0cd Andy McCall ↶ 2221 
        out
c6aa5bd9 Andy McCall ↶ 2222 
    }
c2e1d090 Andy McCall 2223 
}
2224
2225 
/// The node at `path` in `view`, with the nodes it shows as one with it
2226 
/// (identical additions, spec §9.2): the file's history is all of theirs.
2227 
fn nodes_at(view: &crate::tree::TreeView, path: &str) -> Option<HashSet<NodeId>> {
2228 
    let node = view.entries.get(path)?.node;
2229 
    let mut out = HashSet::from([node]);
2230 
    out.extend(
2231 
        view.unified
2232 
            .iter()
2233 
            .filter(|(_, shown)| **shown == node)
2234 
            .map(|(dup, _)| *dup),
2235 
    );
2236 
    Some(out)
c6aa5bd9 Andy McCall ↶ 2237 
}
2238
2239 
/// Per-file hunks of `change` against `base`.
2240 
pub fn diff_files_of(change: &Change, base: &State, base_snapshot: &Snapshot) -> Vec<FileDiff> {
2241 
    let by_node = base_snapshot.tree.by_node();
2242 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
2243 
    for op in &change.tree_ops {
75167672 Andy McCall 2244 
        if let TreeOp::AddNode {
2245 
            node,
2246 
            parent,
2247 
            name,
2248 
            kind,
2249 
            ..
2250 
        } = op
2251 
        {
c6aa5bd9 Andy McCall ↶ 2252 
            new_names.insert(*node, (*parent, name.clone(), *kind));
2253 
        }
2254 
    }
75167672 Andy McCall 2255 
    let path_of = |n: &NodeId| -> String {
2256 
        path_of_node(n, &by_node, &new_names).unwrap_or_else(|| format!("<node {}>", n.short(8)))
2257 
    };
c6aa5bd9 Andy McCall ↶ 2258 
    let mut files: BTreeMap<NodeId, FileDiff> = BTreeMap::new();
2259 
    let mut order: Vec<NodeId> = Vec::new();
2260 
    let mut entry = |files: &mut BTreeMap<NodeId, FileDiff>, n: NodeId, kind: &str| {
2261 
        if let std::collections::btree_map::Entry::Vacant(v) = files.entry(n) {
2262 
            order.push(n);
75167672 Andy McCall 2263 
            v.insert(FileDiff {
2264 
                path: path_of(&n),
2265 
                kind: kind.to_string(),
2266 
                hunks: vec![],
2267 
                from: None,
2268 
            });
c6aa5bd9 Andy McCall ↶ 2269 
        }
2270 
    };
c2e1d090 Andy McCall 2271 
    for op in change
2272 
        .tree_ops
2273 
        .iter()
2274 
        .filter(|o| !is_unified_retirement(o, base_snapshot))
2275 
    {
c6aa5bd9 Andy McCall ↶ 2276 
        match op {
2277 
            TreeOp::AddNode { node, kind, .. } => {
2278 
                if *kind != NodeKind::Dir {
2279 
                    entry(&mut files, *node, "added");
2280 
                }
2281 
            }
2282 
            TreeOp::RemoveNode { node, .. } => {
2283 
                entry(&mut files, *node, "removed");
2284 
                if let Some(e) = by_node.get(node) {
2285 
                    files.get_mut(node).unwrap().path = e.path.clone();
2286 
                }
2287 
            }
75167672 Andy McCall 2288 
            TreeOp::Move {
2289 
                node, parent, name, ..
2290 
            } => {
c6aa5bd9 Andy McCall ↶ 2291 
                entry(&mut files, *node, "renamed");
2292 
                let f = files.get_mut(node).unwrap();
2293 
                f.from = by_node.get(node).map(|e| e.path.clone());
75167672 Andy McCall 2294 
                let dir = if parent.is_root() {
2295 
                    String::new()
2296 
                } else {
2297 
                    path_of_node(parent, &by_node, &new_names).unwrap_or_default()
2298 
                };
2299 
                f.path = if dir.is_empty() {
2300 
                    name.clone()
2301 
                } else {
2302 
                    format!("{dir}/{name}")
2303 
                };
c6aa5bd9 Andy McCall ↶ 2304 
            }
2305 
            TreeOp::SetMode { node, .. } => entry(&mut files, *node, "mode"),
2306 
        }
2307 
    }
2308 
    // Base line numbers per node, 1-based.
2309 
    let mut line_of: HashMap<NodeId, HashMap<(ChangeHash, u32), u32>> = HashMap::new();
2310 
    let mut lines_for = |n: &NodeId| -> HashMap<(ChangeHash, u32), u32> {
75167672 Andy McCall 2311 
        line_of
2312 
            .entry(*n)
2313 
            .or_insert_with(|| {
2314 
                base.live_lines(n)
2315 
                    .into_iter()
2316 
                    .enumerate()
2317 
                    .map(|(i, (k, _))| (k, i as u32 + 1))
2318 
                    .collect()
2319 
            })
2320 
            .clone()
c6aa5bd9 Andy McCall ↶ 2321 
    };
2322 
    for e in &change.edits {
2323 
        let n = e.node();
2324 
        let is_new = new_names.contains_key(&n);
2325 
        entry(&mut files, n, if is_new { "added" } else { "modified" });
2326 
        let f = files.get_mut(&n).unwrap();
2327 
        match e {
2328 
            Edit::Insert { after, lines, .. } => {
2329 
                let map = lines_for(&n);
2330 
                let base_line = match after {
2331 
                    AtomRef::Start => 0,
2332 
                    AtomRef::End => map.len() as u32,
75167672 Andy McCall 2333 
                    AtomRef::Atom { change, index } if change.is_zero() => {
2334 
                        f.hunks.last().map(|h| h.base_line).unwrap_or(0)
2335 
                    }
2336 
                    AtomRef::Atom { change, index } => {
2337 
                        map.get(&(*change, *index)).copied().unwrap_or(0)
2338 
                    }
c6aa5bd9 Andy McCall ↶ 2339 
                };
75167672 Andy McCall 2340 
                f.hunks.push(DiffHunk {
2341 
                    base_line,
01a47cd7 Andy McCall ↶ 2342 
                    added: lines.to_vec(),
75167672 Andy McCall 2343 
                    removed: vec![],
2344 
                });
c6aa5bd9 Andy McCall ↶ 2345 
            }
2346 
            Edit::Delete { atoms, .. } => {
2347 
                let map = lines_for(&n);
2348 
                let mut removed = Vec::new();
2349 
                for r in atoms {
2350 
                    for i in r.from..r.to {
2351 
                        let line = map.get(&(r.change, i)).copied().unwrap_or(0);
75167672 Andy McCall 2352 
                        let text = base
2353 
                            .file(&n)
2354 
                            .and_then(|g| g.content(&(r.change, i)))
2355 
                            .unwrap_or(b"")
2356 
                            .to_vec();
c6aa5bd9 Andy McCall ↶ 2357 
                        removed.push((line, text));
2358 
                    }
2359 
                }
75167672 Andy McCall 2360 
                let base_line = removed
2361 
                    .first()
2362 
                    .map(|(l, _)| l.saturating_sub(1))
2363 
                    .unwrap_or(0);
2364 
                f.hunks.push(DiffHunk {
2365 
                    base_line,
2366 
                    added: vec![],
2367 
                    removed,
2368 
                });
c6aa5bd9 Andy McCall ↶ 2369 
            }
2370 
            Edit::Replace { .. } => {
2371 
                if !is_new {
2372 
                    f.kind = "replaced".into();
2373 
                }
2374 
            }
2375 
        }
2376 
    }
2377 
    order.into_iter().filter_map(|n| files.remove(&n)).collect()
2378 
}
2379
98286406 Andy McCall ↶ 2380 
/// Builds the tree ops and edits of a working-copy change against a base.
dfb23a3a Andy McCall ↶ 2381 
/// `BLAKE3("oxy node 1" 0 seed kind path)`, truncated: a node identifier
2382 
/// that the same seed and path always give (never the root's).
2383 
fn seeded_node(seed: &[u8; 32], path: &str, kind: NodeKind) -> NodeId {
2384 
    let mut h = blake3::Hasher::new();
2385 
    h.update(b"oxy node 1\0");
2386 
    h.update(seed);
2387 
    h.update(&[match kind {
2388 
        NodeKind::File => 1,
2389 
        NodeKind::Dir => 2,
2390 
        NodeKind::Symlink => 3,
2391 
        NodeKind::Subrepo => 4,
2392 
    }]);
2393 
    h.update(path.as_bytes());
2394 
    let mut id = [0u8; 16];
2395 
    id.copy_from_slice(&h.finalize().as_bytes()[..16]);
2396 
    if id == [0; 16] {
2397 
        id[15] = 1;
2398 
    }
2399 
    NodeId(id)
2400 
}
2401
98286406 Andy McCall ↶ 2402 
struct ChangeBuilder<'a> {
2403 
    base: &'a State,
2404 
    base_snapshot: &'a Snapshot,
2405 
    tree_ops: Vec<TreeOp>,
2406 
    edits: Vec<Edit>,
2407 
    statuses: Vec<PathStatus>,
2408 
    dir_nodes: HashMap<String, NodeId>,
2409 
    reuse: HashMap<String, (NodeId, NodeKind)>,
2410 
    /// Changes that contributed to conflicts this working copy resolves.
2411 
    contributors: BTreeSet<ChangeHash>,
2412 
    resolves: bool,
cdb5beba Andy McCall ↶ 2413 
    attrs: crate::attributes::Attributes,
b7f102fc Andy McCall ↶ 2414 
    blobs: HashMap<String, ObjectRef>,
f7510b63 Andy McCall ↶ 2415 
    /// Opaque files whose replace conflict the content settles.
2416 
    resolve: HashSet<String>,
dfb23a3a Andy McCall ↶ 2417 
    /// When set, new nodes' identifiers derive from it and their paths.
2418 
    node_seed: Option<[u8; 32]>,
a87c7b9d Andy McCall ↶ 2419 
    /// For a Git import, the base's dead directories by where they were
2420 
    /// (see `ensure_dir`), worked out when first needed.
2421 
    dead_dirs: Option<HashMap<(NodeId, String), (NodeId, crate::tree::OpKey)>>,
360aa74a Andy McCall ↶ 2422 
    /// Text edits of edited files, worked out in parallel beforehand
2423 
    /// (`diff_in_parallel`); a file not here is diffed when it is updated.
2424 
    diffs: HashMap<String, Vec<Edit>>,
98286406 Andy McCall ↶ 2425 
}
2426
2427 
impl<'a> ChangeBuilder<'a> {
75167672 Andy McCall 2428 
    fn new(
2429 
        base: &'a State,
2430 
        base_snapshot: &'a Snapshot,
2431 
        reuse: HashMap<String, (NodeId, NodeKind)>,
2432 
    ) -> Self {
98286406 Andy McCall ↶ 2433 
        let mut dir_nodes: HashMap<String, NodeId> = HashMap::new();
2434 
        dir_nodes.insert(String::new(), NodeId::ROOT);
e83354cd Andy McCall ↶ 2435 
        for (p, e) in base_snapshot.tree.entries.iter() {
2306a4d2 Andy McCall ↶ 2436 
            if e.kind == NodeKind::Dir {
2437 
                dir_nodes.insert(p.clone(), e.node);
98286406 Andy McCall ↶ 2438 
            }
2439 
        }
75167672 Andy McCall 2440 
        ChangeBuilder {
2441 
            base,
2442 
            base_snapshot,
2443 
            tree_ops: vec![],
2444 
            edits: vec![],
2445 
            statuses: vec![],
2446 
            dir_nodes,
2447 
            reuse,
2448 
            contributors: BTreeSet::new(),
2449 
            resolves: false,
2450 
            attrs: crate::attributes::Attributes::default(),
2451 
            blobs: HashMap::new(),
f7510b63 Andy McCall ↶ 2452 
            resolve: HashSet::new(),
dfb23a3a Andy McCall ↶ 2453 
            node_seed: None,
a87c7b9d Andy McCall ↶ 2454 
            dead_dirs: None,
360aa74a Andy McCall ↶ 2455 
            diffs: HashMap::new(),
75167672 Andy McCall 2456 
        }
98286406 Andy McCall ↶ 2457 
    }
2458
360aa74a Andy McCall ↶ 2459 
    /// Diff the edited text files among `entries` against their base
2460 
    /// renderings at once, on every core, for `update_path` to take: the
2461 
    /// same diff it would make, only not one file after another.
2462 
    fn diff_in_parallel(&mut self, entries: &BTreeMap<String, FsEntry>) {
2463 
        use rayon::prelude::*;
2464 
        let files = &self.base_snapshot.files;
2465 
        let edited: Vec<(&String, &FsEntry, &crate::state::FileState)> = entries
2466 
            .iter()
2467 
            .filter_map(|(p, e)| {
2468 
                let x = files.get(p)?;
2469 
                (x.kind == e.kind && !e.unchanged && !self.opaque_for(p, e)).then_some((p, e, x))
2470 
            })
2471 
            .collect();
2472 
        // Below this, starting the threads costs more than it saves.
2473 
        if edited.len() < 16 {
2474 
            return;
2475 
        }
2476 
        let this = &*self;
2477 
        let attrs = &this.attrs;
2478 
        let diffs: Vec<(String, Vec<Edit>)> = edited
2479 
            .par_iter()
2480 
            .filter_map(|(p, e, x)| {
2481 
                let FileContent::Text {
2482 
                    bytes,
2483 
                    materialized,
2484 
                } = &x.content
2485 
                else {
2486 
                    return None;
2487 
                };
2488 
                if materialized.is_conflicted() {
2489 
                    return None;
2490 
                }
2491 
                let disk_text = if attrs.lookup(p).eol.is_some() {
2492 
                    std::sync::Arc::new(crate::attributes::normalize_eol(&e.content))
2493 
                } else {
2494 
                    e.content.clone()
2495 
                };
2496 
                if *bytes == *disk_text {
2497 
                    return None;
2498 
                }
2499 
                let keys = materialized.live_atoms();
2500 
                let lines = crate::diff::split_lines(bytes);
2501 
                (keys.len() == lines.len()).then(|| {
2502 
                    let edits = crate::diff::text_edits_of(x.node, &keys, &lines, &disk_text);
2503 
                    ((*p).clone(), edits)
2504 
                })
2505 
            })
2506 
            .collect();
2507 
        self.diffs = diffs.into_iter().collect();
2508 
    }
2509
cdb5beba Andy McCall ↶ 2510 
    fn with_attributes(mut self, attrs: crate::attributes::Attributes) -> Self {
2511 
        self.attrs = attrs;
2512 
        self
2513 
    }
2514
2515 
    /// Whether a file is opaque: forced by `binary`, forced text by `text`
2516 
    /// only when it has no NUL bytes, else sniffed.
b7f102fc Andy McCall ↶ 2517 
    /// The blob for an opaque entry: the cached one, else stored now.
75167672 Andy McCall 2518 
    fn blob_of(
2519 
        &mut self,
2520 
        repo: &Repo,
2521 
        path: &str,
2522 
        e: &FsEntry,
2523 
        nocompress: bool,
2524 
    ) -> Result<ObjectRef> {
b7f102fc Andy McCall ↶ 2525 
        let b = match e.blob {
2526 
            Some(b) => b,
2527 
            None => repo.store_blob_for(&e.content, nocompress)?,
2528 
        };
2529 
        self.blobs.insert(path.to_string(), b);
2530 
        Ok(b)
2531 
    }
2532
cdb5beba Andy McCall ↶ 2533 
    fn opaque_for(&self, path: &str, e: &FsEntry) -> bool {
b7f102fc Andy McCall ↶ 2534 
        if e.blob.is_some() {
2535 
            return true;
2536 
        }
4dee277c Andy McCall ↶ 2537 
        if e.kind == NodeKind::Symlink || e.kind == NodeKind::Subrepo {
cdb5beba Andy McCall ↶ 2538 
            return true;
2539 
        }
2540 
        match self.attrs.lookup(path).binary {
2541 
            Some(true) => true,
2542 
            Some(false) => is_binary(&e.content),
2543 
            None => is_binary(&e.content),
2544 
        }
2545 
    }
2546
2547 
    /// Text content as stored: `\r\n` normalized for `eol=` files.
40a40bb7 Andy McCall ↶ 2548 
    fn stored_text(
2549 
        &self,
2550 
        path: &str,
2551 
        content: &std::sync::Arc<Vec<u8>>,
2552 
    ) -> std::sync::Arc<Vec<u8>> {
75167672 Andy McCall 2553 
        if self.attrs.lookup(path).eol.is_some() {
40a40bb7 Andy McCall ↶ 2554 
            std::sync::Arc::new(crate::attributes::normalize_eol(content))
75167672 Andy McCall 2555 
        } else {
40a40bb7 Andy McCall ↶ 2556 
            content.clone()
dfb23a3a Andy McCall ↶ 2557 
        }
2558 
    }
2559
2560 
    /// The identifier of a node added at `path`: reused when asked, derived
2561 
    /// from the seed when there is one, random otherwise.
2562 
    fn new_node(&self, path: &str, kind: NodeKind) -> NodeId {
2563 
        if let Some(n) = self.reused_node(path, kind) {
2564 
            return n;
75167672 Andy McCall 2565 
        }
dfb23a3a Andy McCall ↶ 2566 
        match &self.node_seed {
2567 
            Some(seed) => seeded_node(seed, path, kind),
2568 
            None => NodeId::random(),
2569 
        }
cdb5beba Andy McCall ↶ 2570 
    }
2571
98286406 Andy McCall ↶ 2572 
    fn reused_node(&self, path: &str, kind: NodeKind) -> Option<NodeId> {
75167672 Andy McCall 2573 
        self.reuse
2574 
            .get(path)
2575 
            .filter(|(_, k)| *k == kind)
2576 
            .map(|(n, _)| *n)
98286406 Andy McCall ↶ 2577 
    }
2578
a87c7b9d Andy McCall ↶ 2579 
    /// A dead directory last placed at (`parent`, `name`) in the base.
2580 
    fn revivable_dir(
2581 
        &mut self,
2582 
        parent: NodeId,
2583 
        name: &str,
2584 
    ) -> Option<(NodeId, crate::tree::OpKey)> {
2585 
        let base = self.base;
2586 
        self.dead_dirs
2587 
            .get_or_insert_with(|| base.tree().dead_dirs())
2588 
            .get(&(parent, name.to_string()))
2589 
            .copied()
2590 
    }
2591
98286406 Andy McCall ↶ 2592 
    fn ensure_dir(&mut self, path: &str) -> NodeId {
2593 
        if let Some(n) = self.dir_nodes.get(path) {
2594 
            return *n;
2595 
        }
2596 
        let mut parent = NodeId::ROOT;
2597 
        let mut prefix = String::new();
2598 
        for comp in path.split('/') {
2599 
            if !prefix.is_empty() {
2600 
                prefix.push('/');
2601 
            }
2602 
            prefix.push_str(comp);
a87c7b9d Andy McCall ↶ 2603 
            let known = self.dir_nodes.get(&prefix).copied();
2604 
            let revive = match known {
2605 
                Some(_) => None,
2606 
                None if self.node_seed.is_some() => self.revivable_dir(parent, comp),
2607 
                None => None,
2608 
            };
2609 
            parent = match (known, revive) {
2610 
                (Some(n), _) => n,
2611 
                (None, Some((node, old))) => {
2612 
                    // Git has no identity for a directory: one that comes
2613 
                    // back where it was is the same directory, brought back
2614 
                    // with a move, so lines of history that placed files in
2615 
                    // it before and after agree (oxygen#30).
2616 
                    self.tree_ops.push(TreeOp::Move {
2617 
                        node,
2618 
                        old: OpRef {
2619 
                            change: old.0,
2620 
                            index: old.1,
2621 
                        },
2622 
                        parent,
2623 
                        name: comp.to_string(),
2624 
                    });
2625 
                    self.dir_nodes.insert(prefix.clone(), node);
2626 
                    node
2627 
                }
2628 
                (None, None) => {
dfb23a3a Andy McCall ↶ 2629 
                    let node = self.new_node(&prefix, NodeKind::Dir);
75167672 Andy McCall 2630 
                    self.tree_ops.push(TreeOp::AddNode {
2631 
                        node,
2632 
                        kind: NodeKind::Dir,
2633 
                        mode: Mode::default(),
2634 
                        parent,
2635 
                        name: comp.to_string(),
2636 
                    });
98286406 Andy McCall ↶ 2637 
                    self.dir_nodes.insert(prefix.clone(), node);
2638 
                    node
2639 
                }
2640 
            };
2641 
        }
2642 
        parent
2643 
    }
2644
2645 
    fn add_path(&mut self, repo: &Repo, path: &str, e: &FsEntry) -> Result<()> {
2646 
        let (dir, name) = split_path(path);
2647 
        let parent = self.ensure_dir(dir);
dfb23a3a Andy McCall ↶ 2648 
        let node = self.new_node(path, e.kind);
cdb5beba Andy McCall ↶ 2649 
        let opaque = self.opaque_for(path, e);
2650 
        let nocompress = self.attrs.lookup(path).nocompress;
75167672 Andy McCall 2651 
        self.tree_ops.push(TreeOp::AddNode {
2652 
            node,
2653 
            kind: e.kind,
2654 
            mode: Mode {
2655 
                exec: e.exec,
2656 
                opaque,
2657 
            },
2658 
            parent,
2659 
            name: name.to_string(),
2660 
        });
98286406 Andy McCall ↶ 2661 
        if opaque {
b7f102fc Andy McCall ↶ 2662 
            let blob = self.blob_of(repo, path, e, nocompress)?;
75167672 Andy McCall 2663 
            self.edits.push(Edit::Replace {
2664 
                node,
2665 
                old: None,
2666 
                new: Some(blob),
2667 
            });
98286406 Andy McCall ↶ 2668 
        } else if !e.content.is_empty() {
cdb5beba Andy McCall ↶ 2669 
            let content = self.stored_text(path, &e.content);
98286406 Andy McCall ↶ 2670 
            self.edits.push(Edit::Insert {
2671 
                node,
2672 
                after: crate::objects::AtomRef::Start,
2673 
                before: crate::objects::AtomRef::End,
40a40bb7 Andy McCall ↶ 2674 
                lines: crate::lines::Lines::split_shared(content),
98286406 Andy McCall ↶ 2675 
            });
2676 
        }
75167672 Andy McCall 2677 
        self.statuses.push(PathStatus {
2678 
            path: path.to_string(),
2679 
            status: ChangeStatus::Added,
2680 
        });
98286406 Andy McCall ↶ 2681 
        Ok(())
2682 
    }
2683
6e723b81 Andy McCall ↶ 2684 
    /// Move the node at `old` to `new` and apply `e`'s content to it.
75167672 Andy McCall 2685 
    fn move_path(
2686 
        &mut self,
2687 
        repo: &Repo,
2688 
        old: &str,
2689 
        new: &str,
2690 
        e: &FsEntry,
2691 
        expected: &crate::state::FileState,
2692 
    ) -> Result<()> {
e8e6e0ca Andy McCall ↶ 2693 
        let placement = self.base_snapshot.tree.entries[old].placement;
6e723b81 Andy McCall ↶ 2694 
        let (dir, name) = split_path(new);
2695 
        let parent = self.ensure_dir(dir);
75167672 Andy McCall 2696 
        self.tree_ops.push(TreeOp::Move {
2697 
            node: expected.node,
e8e6e0ca Andy McCall ↶ 2698 
            old: op_ref(placement),
75167672 Andy McCall 2699 
            parent,
2700 
            name: name.to_string(),
2701 
        });
e8e6e0ca Andy McCall ↶ 2702 
        for k in self.same_place(&expected.node, placement) {
2703 
            self.tree_ops.push(TreeOp::RemoveNode {
2704 
                node: expected.node,
2705 
                placement: op_ref(k),
2706 
            });
2707 
        }
6e723b81 Andy McCall ↶ 2708 
        self.update_path(repo, old, e, expected)?;
2709 
        // update_path reports under the old path; the rename is what the user sees.
2710 
        self.statuses.retain(|s| s.path != old);
75167672 Andy McCall 2711 
        self.statuses.push(PathStatus {
2712 
            path: new.to_string(),
2713 
            status: ChangeStatus::Renamed,
2714 
        });
6e723b81 Andy McCall ↶ 2715 
        Ok(())
a87c7b9d Andy McCall ↶ 2716 
    }
2717
2718 
    /// For a Git import: a node the commit keeps is in one place, the one
2719 
    /// its tree names, so any other live placement it has in the base goes;
2720 
    /// and a node the tree cannot show at all (an orphan) goes too. Such a
2721 
    /// placement is out of sight, under a directory another line of history
2722 
    /// removed, where no diff of the tree would ever reach it (oxygen#30).
2723 
    fn retire_stray_placements(&mut self) {
2724 
        let mut retired: HashSet<(ChangeHash, u32)> = HashSet::new();
2725 
        let mut placed: BTreeSet<NodeId> = BTreeSet::new();
2726 
        for op in &self.tree_ops {
2727 
            match op {
2728 
                TreeOp::Move { node, old, .. } => {
2729 
                    retired.insert((old.change, old.index));
2730 
                    placed.insert(*node);
2731 
                }
2732 
                TreeOp::RemoveNode { placement, .. } => {
2733 
                    retired.insert((placement.change, placement.index));
2734 
                }
2735 
                TreeOp::AddNode { .. } | TreeOp::SetMode { .. } => {}
2736 
            }
2737 
        }
2738 
        // Nodes kept where they were: what the base's tree view shows and
2739 
        // this change neither moves nor removes.
2740 
        let mut kept: HashMap<NodeId, (ChangeHash, u32)> = HashMap::new();
2741 
        for e in self.base_snapshot.tree.entries.values() {
2742 
            if !retired.contains(&e.placement) {
2743 
                kept.insert(e.node, e.placement);
2744 
            }
2745 
        }
2746 
        // Only a node with several live placements can have a stray one,
2747 
        // and the tree view names each as a conflict.
2748 
        let multi: BTreeSet<NodeId> = self
2749 
            .base_snapshot
2750 
            .tree
2751 
            .conflicts
2752 
            .iter()
2753 
            .filter_map(|c| match c {
2754 
                crate::tree::TreeConflict::MultiPlacement { node, .. } => Some(*node),
2755 
                _ => None,
2756 
            })
2757 
            .collect();
2758 
        let mut extra = Vec::new();
2759 
        // A node whose only placements are under a directory another line
2760 
        // removed (an orphan) is nowhere the tree can show, and Git's tree
2761 
        // has no such path: it goes.
2762 
        for c in &self.base_snapshot.tree.conflicts {
2763 
            let crate::tree::TreeConflict::Orphan { node, .. } = c else {
2764 
                continue;
2765 
            };
2766 
            if kept.contains_key(node) || placed.contains(node) {
2767 
                continue;
2768 
            }
2769 
            for (k, _) in self.base.tree().live_placements(node) {
2770 
                if retired.insert(k) {
2771 
                    extra.push(TreeOp::RemoveNode {
2772 
                        node: *node,
2773 
                        placement: op_ref(k),
2774 
                    });
2775 
                }
2776 
            }
2777 
        }
2778 
        for node in placed
2779 
            .iter()
2780 
            .chain(multi.iter().filter(|n| kept.contains_key(*n)))
2781 
        {
2782 
            for (k, _) in self.base.tree().live_placements(node) {
2783 
                if retired.contains(&k) || kept.get(node) == Some(&k) {
2784 
                    continue;
2785 
                }
2786 
                retired.insert(k);
2787 
                extra.push(TreeOp::RemoveNode {
2788 
                    node: *node,
2789 
                    placement: op_ref(k),
2790 
                });
2791 
            }
2792 
        }
2793 
        if !extra.is_empty() {
2794 
            self.resolves = true;
2795 
            self.tree_ops.extend(extra);
2796 
        }
6e723b81 Andy McCall ↶ 2797 
    }
2798
c2e1d090 Andy McCall 2799 
    /// Nodes the base shows as one with another (identical additions at one
2800 
    /// path, spec §9.2) are retired by the change that first touches the
2801 
    /// node shown: edited, moved, removed or given a mode alone, it would
2802 
    /// differ from them and the name conflict would come back. Nothing
2803 
    /// else changes for the user, so no status reports it.
2804 
    fn retire_unified(&mut self) {
2805 
        let unified = &self.base_snapshot.tree.unified;
2806 
        if unified.is_empty() {
2807 
            return;
2808 
        }
2809 
        let touched: HashSet<NodeId> = self
2810 
            .tree_ops
2811 
            .iter()
2812 
            .map(|o| o.node())
2813 
            .chain(self.edits.iter().map(|e| e.node()))
2814 
            .collect();
2815 
        let mut retired: HashSet<(ChangeHash, u32)> = HashSet::new();
2816 
        for op in &self.tree_ops {
2817 
            match op {
2818 
                TreeOp::Move { old, .. } => {
2819 
                    retired.insert((old.change, old.index));
2820 
                }
2821 
                TreeOp::RemoveNode { placement, .. } => {
2822 
                    retired.insert((placement.change, placement.index));
2823 
                }
2824 
                TreeOp::AddNode { .. } | TreeOp::SetMode { .. } => {}
2825 
            }
2826 
        }
2827 
        let mut extra = Vec::new();
2828 
        for (dup, shown) in unified {
2829 
            if !touched.contains(shown) {
2830 
                continue;
2831 
            }
2832 
            for (k, _) in self.base.tree().live_placements(dup) {
2833 
                if retired.insert(k) {
2834 
                    extra.push(TreeOp::RemoveNode {
2835 
                        node: *dup,
2836 
                        placement: op_ref(k),
2837 
                    });
2838 
                }
2839 
            }
2840 
        }
2841 
        self.tree_ops.extend(extra);
2842 
    }
2843
e8e6e0ca Andy McCall ↶ 2844 
    /// The node's other live placements at the same place as `k`, the one
2845 
    /// the tree shows: two lines of history that put a node back where it
2846 
    /// was each leave a placement, and the tree shows them as one file. A
2847 
    /// change that removes or moves it retires them together, or the copy
2848 
    /// left behind outlives its directory (oxygen#88).
2849 
    fn same_place(&self, node: &NodeId, k: (ChangeHash, u32)) -> Vec<(ChangeHash, u32)> {
2850 
        let tree = self.base.tree();
2851 
        let Some(shown) = tree.placement(&k) else {
2852 
            return Vec::new();
2853 
        };
2854 
        tree.live_placements(node)
2855 
            .into_iter()
2856 
            .filter(|(o, p)| *o != k && p.parent == shown.parent && p.name == shown.name)
2857 
            .map(|(o, _)| o)
2858 
            .collect()
2859 
    }
2860
98286406 Andy McCall ↶ 2861 
    fn remove_path(&mut self, path: &str, expected: &crate::state::FileState) {
2862 
        let entry = &self.base_snapshot.tree.entries[path];
e8e6e0ca Andy McCall ↶ 2863 
        let placement = entry.placement;
75167672 Andy McCall 2864 
        self.tree_ops.push(TreeOp::RemoveNode {
2865 
            node: expected.node,
e8e6e0ca Andy McCall ↶ 2866 
            placement: op_ref(placement),
75167672 Andy McCall 2867 
        });
e8e6e0ca Andy McCall ↶ 2868 
        for k in self.same_place(&expected.node, placement) {
2869 
            self.tree_ops.push(TreeOp::RemoveNode {
2870 
                node: expected.node,
2871 
                placement: op_ref(k),
2872 
            });
2873 
        }
98286406 Andy McCall ↶ 2874 
        if path.contains(".oxy-conflict-") || expected.is_conflicted() {
2875 
            self.resolves = true;
2876 
        }
75167672 Andy McCall 2877 
        self.statuses.push(PathStatus {
2878 
            path: path.to_string(),
2879 
            status: ChangeStatus::Removed,
2880 
        });
98286406 Andy McCall ↶ 2881 
    }
2882
75167672 Andy McCall 2883 
    fn update_path(
2884 
        &mut self,
2885 
        repo: &Repo,
2886 
        path: &str,
2887 
        e: &FsEntry,
2888 
        expected: &crate::state::FileState,
2889 
    ) -> Result<()> {
98286406 Andy McCall ↶ 2890 
        if expected.kind != e.kind {
2891 
            self.remove_path(path, expected);
2892 
            self.statuses.pop();
2893 
            self.add_path(repo, path, e)?;
2894 
            self.statuses.pop();
75167672 Andy McCall 2895 
            self.statuses.push(PathStatus {
2896 
                path: path.to_string(),
2897 
                status: ChangeStatus::Modified,
2898 
            });
98286406 Andy McCall ↶ 2899 
            return Ok(());
2900 
        }
2901 
        let entry = &self.base_snapshot.tree.entries[path];
2902 
        let node = expected.node;
2903 
        let mode_atom = entry.mode_atom.map(op_ref);
cc7a0056 Andy McCall ↶ 2904 
        if e.unchanged {
2905 
            // Known equal to the base's unconflicted rendering (spec
2906 
            // §12.3); only the executable bit can differ.
2907 
            if e.exec != expected.mode.exec {
2908 
                if let Some(old) = mode_atom {
2909 
                    self.tree_ops.push(TreeOp::SetMode {
2910 
                        node,
2911 
                        old,
2912 
                        mode: Mode {
2913 
                            exec: e.exec,
2914 
                            opaque: false,
2915 
                        },
2916 
                    });
2917 
                    self.statuses.push(PathStatus {
2918 
                        path: path.to_string(),
2919 
                        status: ChangeStatus::ModeChanged,
2920 
                    });
2921 
                }
2922 
            }
2923 
            return Ok(());
2924 
        }
98286406 Andy McCall ↶ 2925 
        let mut status: Option<ChangeStatus> = None;
cdb5beba Andy McCall ↶ 2926 
        let wants_opaque = self.opaque_for(path, e);
2927 
        let nocompress = self.attrs.lookup(path).nocompress;
2928 
        let disk_text = self.stored_text(path, &e.content);
98286406 Andy McCall ↶ 2929 
        match &expected.content {
2930 
            FileContent::Dir => {}
75167672 Andy McCall 2931 
            FileContent::Text {
2932 
                bytes,
2933 
                materialized,
2934 
            } => {
98286406 Andy McCall ↶ 2935 
                if materialized.is_conflicted() {
40a40bb7 Andy McCall ↶ 2936 
                    if bytes == &*disk_text {
98286406 Andy McCall ↶ 2937 
                        return Ok(());
2938 
                    }
3652104a Andy McCall ↶ 2939 
                    let regions = self.base.conflict_regions(&node);
2940 
                    let plain: Vec<&[u8]> =
2941 
                        regions.plain.iter().map(|(_, l)| l.as_slice()).collect();
2942 
                    if has_own_conflict_markers(&e.content, bytes, &plain) {
75167672 Andy McCall 2943 
                        self.statuses.push(PathStatus {
2944 
                            path: path.to_string(),
2945 
                            status: ChangeStatus::ConflictEdited,
2946 
                        });
98286406 Andy McCall ↶ 2947 
                        return Ok(());
2948 
                    }
2949 
                    // Resolution (spec §9.6): diff against the unconflicted lines
2950 
                    // with each region as a gap, then retire every region atom.
2951 
                    // Whatever the author kept from a region is re-inserted as
2952 
                    // new atoms, which leaves the graph totally ordered.
cdb5beba Andy McCall ↶ 2953 
                    let mut edits = text_edits(node, &regions.plain, &disk_text);
98286406 Andy McCall ↶ 2954 
                    if !regions.region_atoms.is_empty() {
75167672 Andy McCall 2955 
                        edits.insert(
2956 
                            0,
2957 
                            Edit::Delete {
2958 
                                node,
2959 
                                atoms: ranges(&regions.region_atoms),
2960 
                            },
2961 
                        );
98286406 Andy McCall ↶ 2962 
                    }
2963 
                    self.edits.extend(edits);
2964 
                    self.contributors.extend(regions.contributors);
2965 
                    self.resolves = true;
2966 
                    if e.exec != expected.mode.exec {
2967 
                        if let Some(old) = mode_atom {
75167672 Andy McCall 2968 
                            self.tree_ops.push(TreeOp::SetMode {
2969 
                                node,
2970 
                                old,
2971 
                                mode: Mode {
2972 
                                    exec: e.exec,
2973 
                                    opaque: false,
2974 
                                },
2975 
                            });
98286406 Andy McCall ↶ 2976 
                        }
2977 
                    }
75167672 Andy McCall 2978 
                    self.statuses.push(PathStatus {
2979 
                        path: path.to_string(),
2980 
                        status: ChangeStatus::Resolved,
2981 
                    });
98286406 Andy McCall ↶ 2982 
                    return Ok(());
2983 
                }
cdb5beba Andy McCall ↶ 2984 
                if wants_opaque {
2985 
                    // Text → opaque (spec §7.5), by content or by attribute.
2986 
                    if let Some(old) = mode_atom {
75167672 Andy McCall 2987 
                        self.tree_ops.push(TreeOp::SetMode {
2988 
                            node,
2989 
                            old,
2990 
                            mode: Mode {
2991 
                                exec: e.exec,
2992 
                                opaque: true,
2993 
                            },
2994 
                        });
cdb5beba Andy McCall ↶ 2995 
                    }
2996 
                    let live: Vec<_> = repo_live_keys(self.base, &node);
2997 
                    if !live.is_empty() {
75167672 Andy McCall 2998 
                        self.edits.push(Edit::Delete {
2999 
                            node,
3000 
                            atoms: ranges(&live),
3001 
                        });
cdb5beba Andy McCall ↶ 3002 
                    }
b7f102fc Andy McCall ↶ 3003 
                    let blob = self.blob_of(repo, path, e, nocompress)?;
75167672 Andy McCall 3004 
                    self.edits.push(Edit::Replace {
3005 
                        node,
3006 
                        old: None,
3007 
                        new: Some(blob),
3008 
                    });
cdb5beba Andy McCall ↶ 3009 
                    status = Some(ChangeStatus::Modified);
40a40bb7 Andy McCall ↶ 3010 
                } else if bytes != &*disk_text {
2306a4d2 Andy McCall ↶ 3011 
                    let mut laps = crate::timing::Laps::start();
3012 
                    // The base's rendering holds its live atoms in order,
3013 
                    // one per line (it is not conflicted here).
360aa74a Andy McCall ↶ 3014 
                    let edits = match self.diffs.remove(path) {
3015 
                        Some(edits) => edits,
3016 
                        None => {
3017 
                            let keys = materialized.live_atoms();
3018 
                            let lines = crate::diff::split_lines(bytes);
3019 
                            if keys.len() == lines.len() {
3020 
                                crate::diff::text_edits_of(node, &keys, &lines, &disk_text)
3021 
                            } else {
3022 
                                text_edits(node, &self.base.live_lines(&node), &disk_text)
3023 
                            }
3024 
                        }
2306a4d2 Andy McCall ↶ 3025 
                    };
3026 
                    laps.lap("build: text diff");
cdb5beba Andy McCall ↶ 3027 
                    if !edits.is_empty() {
3028 
                        self.edits.extend(edits);
98286406 Andy McCall ↶ 3029 
                        status = Some(ChangeStatus::Modified);
3030 
                    }
3031 
                }
cdb5beba Andy McCall ↶ 3032 
                let switched_to_opaque = wants_opaque;
98286406 Andy McCall ↶ 3033 
                if !switched_to_opaque && e.exec != expected.mode.exec {
3034 
                    if let Some(old) = mode_atom {
75167672 Andy McCall 3035 
                        self.tree_ops.push(TreeOp::SetMode {
3036 
                            node,
3037 
                            old,
3038 
                            mode: Mode {
3039 
                                exec: e.exec,
3040 
                                opaque: false,
3041 
                            },
3042 
                        });
98286406 Andy McCall ↶ 3043 
                        status.get_or_insert(ChangeStatus::ModeChanged);
3044 
                    }
3045 
                }
3046 
            }
3047 
            FileContent::Opaque { blobs } => {
b7f102fc Andy McCall ↶ 3048 
                let new_ref = e.blob.unwrap_or_else(|| Repo::blob_ref(&e.content));
3049 
                if e.kind == NodeKind::File && wants_opaque {
3050 
                    // Unchanged or replaced, the file stays opaque: remember its blob.
3051 
                    self.blobs.insert(path.to_string(), new_ref);
f7510b63 Andy McCall ↶ 3052 
                }
3053 
                if blobs.len() > 1 && self.resolve.contains(path) && e.kind == expected.kind {
3054 
                    // Settle the replace conflict (spec §9.3): replace every
3055 
                    // live candidate with the chosen blob, depending on each
3056 
                    // change that made one live, so all are superseded.
3057 
                    let chosen = self.blob_of(repo, path, e, nocompress)?;
3058 
                    for b in blobs {
3059 
                        self.edits.push(Edit::Replace {
3060 
                            node,
3061 
                            old: Some(*b),
3062 
                            new: Some(chosen),
3063 
                        });
3064 
                    }
3065 
                    self.contributors
3066 
                        .extend(self.base.blob_contributors(&node, blobs));
3067 
                    self.resolves = true;
3068 
                    self.statuses.push(PathStatus {
3069 
                        path: path.to_string(),
3070 
                        status: ChangeStatus::Resolved,
3071 
                    });
3072 
                    return Ok(());
b7f102fc Andy McCall ↶ 3073 
                }
98286406 Andy McCall ↶ 3074 
                if blobs.len() > 1 {
3075 
                    if !blobs.contains(&new_ref) {
75167672 Andy McCall 3076 
                        self.statuses.push(PathStatus {
3077 
                            path: path.to_string(),
3078 
                            status: ChangeStatus::ConflictEdited,
3079 
                        });
98286406 Andy McCall ↶ 3080 
                    }
3081 
                    return Ok(());
3082 
                }
3083 
                let current = blobs.first().copied();
3084 
                if current != Some(new_ref) {
cdb5beba Andy McCall ↶ 3085 
                    if e.kind == NodeKind::File && !wants_opaque {
98286406 Andy McCall ↶ 3086 
                        // Opaque → text (spec §7.5).
3087 
                        if let Some(old) = mode_atom {
75167672 Andy McCall 3088 
                            self.tree_ops.push(TreeOp::SetMode {
3089 
                                node,
3090 
                                old,
3091 
                                mode: Mode {
3092 
                                    exec: e.exec,
3093 
                                    opaque: false,
3094 
                                },
3095 
                            });
98286406 Andy McCall ↶ 3096 
                        }
75167672 Andy McCall 3097 
                        self.edits.push(Edit::Replace {
3098 
                            node,
3099 
                            old: current,
3100 
                            new: None,
3101 
                        });
98286406 Andy McCall ↶ 3102 
                        if !e.content.is_empty() {
3103 
                            self.edits.push(Edit::Insert {
3104 
                                node,
3105 
                                after: crate::objects::AtomRef::Start,
3106 
                                before: crate::objects::AtomRef::End,
01a47cd7 Andy McCall ↶ 3107 
                                lines: crate::lines::Lines::split(&e.content),
98286406 Andy McCall ↶ 3108 
                            });
3109 
                        }
3110 
                    } else {
b7f102fc Andy McCall ↶ 3111 
                        let blob = self.blob_of(repo, path, e, nocompress)?;
75167672 Andy McCall 3112 
                        self.edits.push(Edit::Replace {
3113 
                            node,
3114 
                            old: current,
3115 
                            new: Some(blob),
3116 
                        });
6eae06b6 Andy McCall ↶ 3117 
                        // New contents and a new executable bit at once: the
3118 
                        // bit was dropped (VS Code's inno_updater.exe,
3119 
                        // 2018-02-28, lost +x for 49 commits; oxygen#87).
3120 
                        if e.kind == NodeKind::File && e.exec != expected.mode.exec {
3121 
                            if let Some(old) = mode_atom {
3122 
                                self.tree_ops.push(TreeOp::SetMode {
3123 
                                    node,
3124 
                                    old,
3125 
                                    mode: Mode {
3126 
                                        exec: e.exec,
3127 
                                        opaque: true,
3128 
                                    },
3129 
                                });
3130 
                            }
3131 
                        }
98286406 Andy McCall ↶ 3132 
                    }
3133 
                    status = Some(ChangeStatus::Modified);
3134 
                } else if e.kind == NodeKind::File && e.exec != expected.mode.exec {
3135 
                    if let Some(old) = mode_atom {
75167672 Andy McCall 3136 
                        self.tree_ops.push(TreeOp::SetMode {
3137 
                            node,
3138 
                            old,
3139 
                            mode: Mode {
3140 
                                exec: e.exec,
3141 
                                opaque: true,
3142 
                            },
3143 
                        });
98286406 Andy McCall ↶ 3144 
                        status = Some(ChangeStatus::ModeChanged);
3145 
                    }
3146 
                }
3147 
            }
3148 
        }
3149 
        if let Some(s) = status {
75167672 Andy McCall 3150 
            self.statuses.push(PathStatus {
3151 
                path: path.to_string(),
3152 
                status: s,
3153 
            });
98286406 Andy McCall ↶ 3154 
        }
3155 
        Ok(())
3156 
    }
3157 
}
3158
3159 
fn repo_live_keys(base: &State, node: &NodeId) -> Vec<crate::graph::AtomKey> {
3160 
    base.live_lines(node).into_iter().map(|(k, _)| k).collect()
3161 
}
3162
3163 
/// The changes a change's scope references (spec §8.1): introducers of every
3164 
/// atom, placement, and mode it touches, and of every node it operates on.
3165 
pub fn referenced_changes(change: &Change, base: &State) -> BTreeSet<ChangeHash> {
3166 
    let mut out = BTreeSet::new();
3167 
    let opref = |r: &OpRef, out: &mut BTreeSet<ChangeHash>| {
3168 
        if !r.change.is_zero() {
3169 
            out.insert(r.change);
3170 
        }
3171 
    };
3172 
    let mut nodes: Vec<NodeId> = Vec::new();
3173 
    for op in &change.tree_ops {
3174 
        match op {
3175 
            TreeOp::AddNode { parent, .. } => nodes.push(*parent),
3176 
            TreeOp::RemoveNode { node, placement } => {
3177 
                nodes.push(*node);
3178 
                opref(placement, &mut out);
3179 
            }
75167672 Andy McCall 3180 
            TreeOp::Move {
3181 
                node, old, parent, ..
3182 
            } => {
98286406 Andy McCall ↶ 3183 
                nodes.push(*node);
3184 
                nodes.push(*parent);
3185 
                opref(old, &mut out);
3186 
            }
3187 
            TreeOp::SetMode { node, old, .. } => {
3188 
                nodes.push(*node);
3189 
                opref(old, &mut out);
3190 
            }
3191 
        }
3192 
    }
3193 
    for e in &change.edits {
3194 
        nodes.push(e.node());
3195 
        match e {
3196 
            Edit::Insert { after, before, .. } => {
3197 
                for r in [after, before] {
3198 
                    if let crate::objects::AtomRef::Atom { change, .. } = r {
3199 
                        if !change.is_zero() {
3200 
                            out.insert(*change);
3201 
                        }
3202 
                    }
3203 
                }
3204 
            }
3205 
            Edit::Delete { atoms, .. } => {
62794ecc Andy McCall ↶ 3206 
                // Deleting an atom also depends on whoever inserted the live
3207 
                // atoms right after it: the author saw those insertions, and
3208 
                // without the dependency spec §9.1 would call the deletion
3209 
                // concurrent with them (a delete/insert conflict).
3210 
                let graph = base.file(&e.node());
98286406 Andy McCall ↶ 3211 
                for r in atoms {
3212 
                    if !r.change.is_zero() {
3213 
                        out.insert(r.change);
3214 
                    }
62794ecc Andy McCall ↶ 3215 
                    if let Some(g) = graph {
3216 
                        for i in r.from..r.to {
3217 
                            out.extend(
3218 
                                g.live_successor_changes(&(r.change, i))
3219 
                                    .into_iter()
3220 
                                    .filter(|c| !c.is_zero()),
3221 
                            );
3222 
                        }
3223 
                    }
98286406 Andy McCall ↶ 3224 
                }
3225 
            }
62794ecc Andy McCall ↶ 3226 
            // Replacing a blob depends on the change that made it current,
3227 
            // as deleting a line depends on the line's introducer.
3228 
            Edit::Replace {
3229 
                node,
3230 
                old: Some(old),
3231 
                ..
3232 
            } => {
3233 
                out.extend(
3234 
                    base.live_replaces(node)
3235 
                        .into_iter()
3236 
                        .filter(|(c, b)| b == old && !c.is_zero())
3237 
                        .map(|(c, _)| c),
3238 
                );
3239 
            }
98286406 Andy McCall ↶ 3240 
            Edit::Replace { .. } => {}
3241 
        }
3242 
    }
3243 
    for n in nodes {
3244 
        if let Some(h) = base.tree().introducer(&n) {
3245 
            out.insert(h);
3246 
        }
3247 
    }
3248 
    out
3249 
}
3250
3251 
/// Paths of the nodes a change adds on top of `base_snapshot`.
75167672 Andy McCall 3252 
pub fn added_node_paths(
3253 
    change: &Change,
3254 
    base_snapshot: &Snapshot,
3255 
) -> HashMap<String, (NodeId, NodeKind)> {
98286406 Andy McCall ↶ 3256 
    let by_node = base_snapshot.tree.by_node();
3257 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
3258 
    for op in &change.tree_ops {
75167672 Andy McCall 3259 
        if let TreeOp::AddNode {
3260 
            node,
3261 
            parent,
3262 
            name,
3263 
            kind,
3264 
            ..
3265 
        } = op
3266 
        {
98286406 Andy McCall ↶ 3267 
            new_names.insert(*node, (*parent, name.clone(), *kind));
3268 
        }
3269 
    }
3270 
    let mut out = HashMap::new();
3271 
    for (node, (_, _, kind)) in &new_names {
3272 
        if let Some(path) = path_of_node(node, &by_node, &new_names) {
3273 
            out.insert(path, (*node, *kind));
3274 
        }
3275 
    }
3276 
    out
3277 
}
3278
c2e1d090 Andy McCall 3279 
/// Whether `op` retires a node `base_snapshot` shows as one with another
3280 
/// (spec §9.2): part of touching that node, never a removal the user made.
3281 
fn is_unified_retirement(op: &TreeOp, base_snapshot: &Snapshot) -> bool {
3282 
    matches!(op, TreeOp::RemoveNode { node, .. } if base_snapshot.tree.unified.contains_key(node))
3283 
}
3284
98286406 Andy McCall ↶ 3285 
fn path_of_node(
3286 
    n: &NodeId,
3287 
    by_node: &HashMap<NodeId, &crate::tree::TreeEntry>,
3288 
    new_names: &HashMap<NodeId, (NodeId, String, NodeKind)>,
3289 
) -> Option<String> {
3290 
    if let Some(e) = by_node.get(n) {
3291 
        return Some(e.path.clone());
3292 
    }
3293 
    let mut parts = Vec::new();
3294 
    let mut cur = *n;
3295 
    loop {
3296 
        if cur.is_root() {
3297 
            break;
3298 
        }
3299 
        if let Some(e) = by_node.get(&cur) {
3300 
            if !e.path.is_empty() {
3301 
                parts.push(e.path.clone());
3302 
            }
3303 
            break;
3304 
        }
3305 
        let (parent, name, _) = new_names.get(&cur)?;
3306 
        parts.push(name.clone());
3307 
        cur = *parent;
3308 
    }
3309 
    parts.reverse();
3310 
    Some(parts.join("/"))
3311 
}
3312
3313 
/// Render a change's operations with line content, using `base` for deleted lines.
3314 
pub fn render_change(change: &Change, base: &State, base_snapshot: &Snapshot) -> String {
3315 
    let by_node = base_snapshot.tree.by_node();
3316 
    let mut new_names: HashMap<NodeId, (NodeId, String, NodeKind)> = HashMap::new();
3317 
    for op in &change.tree_ops {
75167672 Andy McCall 3318 
        if let TreeOp::AddNode {
3319 
            node,
3320 
            parent,
3321 
            name,
3322 
            kind,
3323 
            ..
3324 
        } = op
3325 
        {
98286406 Andy McCall ↶ 3326 
            new_names.insert(*node, (*parent, name.clone(), *kind));
3327 
        }
3328 
    }
75167672 Andy McCall 3329 
    let path_of = |n: &NodeId| -> String {
3330 
        path_of_node(n, &by_node, &new_names).unwrap_or_else(|| format!("<node {}>", n.short(8)))
3331 
    };
98286406 Andy McCall ↶ 3332
3333 
    let mut out = String::new();
c2e1d090 Andy McCall 3334 
    for op in change
3335 
        .tree_ops
3336 
        .iter()
3337 
        .filter(|o| !is_unified_retirement(o, base_snapshot))
3338 
    {
98286406 Andy McCall ↶ 3339 
        match op {
75167672 Andy McCall 3340 
            TreeOp::AddNode {
3341 
                node, kind, mode, ..
3342 
            } => {
3343 
                out.push_str(&format!(
3344 
                    "A {} ({:?}{}{})\n",
3345 
                    path_of(node),
3346 
                    kind,
3347 
                    if mode.exec { ", exec" } else { "" },
3348 
                    if mode.opaque { ", opaque" } else { "" }
3349 
                ));
98286406 Andy McCall ↶ 3350 
            }
3351 
            TreeOp::RemoveNode { node, .. } => out.push_str(&format!("D {}\n", path_of(node))),
0912a033 Andy McCall ↶ 3352 
            TreeOp::Move {
3353 
                node, parent, name, ..
3354 
            } => {
3355 
                let to = if parent.is_root() {
3356 
                    name.clone()
3357 
                } else {
3358 
                    format!("{}/{name}", path_of(parent))
3359 
                };
3360 
                out.push_str(&format!("R {} -> {to}\n", path_of(node)))
75167672 Andy McCall 3361 
            }
98286406 Andy McCall ↶ 3362 
            TreeOp::SetMode { node, mode, .. } => {
75167672 Andy McCall 3363 
                out.push_str(&format!(
3364 
                    "M {} mode: exec={} opaque={}\n",
3365 
                    path_of(node),
3366 
                    mode.exec,
3367 
                    mode.opaque
3368 
                ));
98286406 Andy McCall ↶ 3369 
            }
0912a033 Andy McCall ↶ 3370 
        }
3371 
    }
3372 
    // Every other file the change edits is modified: the summary names
3373 
    // every path it touches, not only those with a tree operation.
3374 
    let with_op: HashSet<NodeId> = change.tree_ops.iter().map(|o| o.node()).collect();
3375 
    let mut listed: HashSet<NodeId> = HashSet::new();
3376 
    for e in &change.edits {
3377 
        let n = e.node();
3378 
        if !with_op.contains(&n) && listed.insert(n) {
3379 
            out.push_str(&format!("M {}\n", path_of(&n)));
98286406 Andy McCall ↶ 3380 
        }
3381 
    }
3382 
    let mut last_node: Option<NodeId> = None;
3383 
    for e in &change.edits {
3384 
        let n = e.node();
3385 
        if last_node != Some(n) {
3386 
            out.push_str(&format!("--- {}\n", path_of(&n)));
3387 
            last_node = Some(n);
3388 
        }
3389 
        match e {
3390 
            Edit::Insert { lines, .. } => {
01a47cd7 Andy McCall ↶ 3391 
                for l in lines.iter() {
98286406 Andy McCall ↶ 3392 
                    out.push('+');
3393 
                    out.push_str(&String::from_utf8_lossy(l));
3394 
                    if !l.ends_with(b"\n") {
3395 
                        out.push('\n');
3396 
                    }
3397 
                }
3398 
            }
3399 
            Edit::Delete { atoms, .. } => {
3400 
                for r in atoms {
3401 
                    for i in r.from..r.to {
75167672 Andy McCall 3402 
                        let content = base
3403 
                            .file(&n)
3404 
                            .and_then(|g| g.content(&(r.change, i)))
3405 
                            .unwrap_or(b"<unknown>\n");
98286406 Andy McCall ↶ 3406 
                        out.push('-');
3407 
                        out.push_str(&String::from_utf8_lossy(content));
3408 
                        if !content.ends_with(b"\n") {
3409 
                            out.push('\n');
3410 
                        }
3411 
                    }
3412 
                }
3413 
            }
3414 
            Edit::Replace { old, new, .. } => {
3415 
                out.push_str(&format!(
3416 
                    "  opaque content {} -> {}\n",
3417 
                    old.map(|o| o.short(12)).unwrap_or_else(|| "none".into()),
3418 
                    new.map(|o| o.short(12)).unwrap_or_else(|| "none".into())
3419 
                ));
3420 
            }
3421 
        }
3422 
    }
3423 
    out
3424 
}
6e723b81 Andy McCall ↶ 3425
3426 
/// Pair removed files with added files of the same kind whose contents are
3427 
/// at least `RENAME_THRESHOLD` percent alike, best matches first. Beyond
3428 
/// `RENAME_LIMIT` candidate pairs only identical contents are paired.
3429 
const RENAME_THRESHOLD: u32 = 50;
3430 
const RENAME_LIMIT: usize = 10_000;
2306a4d2 Andy McCall ↶ 3431
23ff1288 Andy McCall ↶ 3432 
/// How a change is built beyond its entries: for a Git import, the seed
3433 
/// its new nodes' identifiers derive from; and, when it is built from part
3434 
/// of the tree, whether the new tree has anything under a directory.
3435 
#[derive(Default)]
3436 
struct Scope<'a> {
3437 
    seed: Option<[u8; 32]>,
2306a4d2 Andy McCall ↶ 3438 
    dir_needed: Option<&'a dyn Fn(&str) -> bool>,
3439 
}
6e723b81 Andy McCall ↶ 3440
75167672 Andy McCall 3441 
fn detect_renames(
3442 
    base_snapshot: &Snapshot,
3443 
    fs_entries: &BTreeMap<String, FsEntry>,
3444 
    in_scope: impl Fn(&str) -> bool,
3445 
) -> Vec<(String, String)> {
6e723b81 Andy McCall ↶ 3446 
    let removed: Vec<(&String, &crate::state::FileState)> = base_snapshot
3447 
        .files
3448 
        .iter()
75167672 Andy McCall 3449 
        .filter(|(p, f)| {
3450 
            f.kind != NodeKind::Dir
3451 
                && !fs_entries.contains_key(*p)
3452 
                && in_scope(p)
3453 
                && !f.is_conflicted()
3454 
        })
3455 
        .collect();
3456 
    let added: Vec<(&String, &FsEntry)> = fs_entries
3457 
        .iter()
3458 
        .filter(|(p, e)| e.kind != NodeKind::Dir && !base_snapshot.files.contains_key(*p))
6e723b81 Andy McCall ↶ 3459 
        .collect();
3460 
    if removed.is_empty() || added.is_empty() {
3461 
        return vec![];
3462 
    }
3463 
    let exact_only = removed.len() * added.len() > RENAME_LIMIT;
3464 
    let old_content = |f: &crate::state::FileState| -> Option<Vec<u8>> {
3465 
        match &f.content {
3466 
            FileContent::Text { bytes, .. } => Some(bytes.clone()),
3467 
            _ => None,
3468 
        }
3469 
    };
3470 
    let mut scored: Vec<(u32, usize, usize)> = Vec::new();
3471 
    for (ri, (_, f)) in removed.iter().enumerate() {
3472 
        let old = old_content(f);
3473 
        let old_blob = match &f.content {
3474 
            FileContent::Opaque { blobs } if blobs.len() == 1 => Some(blobs[0]),
3475 
            _ => None,
3476 
        };
3477 
        for (ai, (_, e)) in added.iter().enumerate() {
3478 
            if e.kind != f.kind {
3479 
                continue;
3480 
            }
3481 
            let score = match (&old, old_blob) {
3482 
                (Some(o), _) => {
3483 
                    if exact_only {
40a40bb7 Andy McCall ↶ 3484 
                        if *o == *e.content {
75167672 Andy McCall 3485 
                            100
3486 
                        } else {
3487 
                            0
3488 
                        }
6e723b81 Andy McCall ↶ 3489 
                    } else {
3490 
                        similarity(o, &e.content)
3491 
                    }
3492 
                }
b7f102fc Andy McCall ↶ 3493 
                (None, Some(b)) if e.blob.unwrap_or_else(|| Repo::blob_ref(&e.content)) == b => 100,
6e723b81 Andy McCall ↶ 3494 
                _ => 0,
3495 
            };
3496 
            if score >= RENAME_THRESHOLD {
3497 
                scored.push((score, ri, ai));
3498 
            }
3499 
        }
3500 
    }
75167672 Andy McCall 3501 
    scored.sort_by(|a, b| {
3502 
        b.0.cmp(&a.0)
3503 
            .then(removed[a.1].0.cmp(removed[b.1].0))
3504 
            .then(added[a.2].0.cmp(added[b.2].0))
3505 
    });
6e723b81 Andy McCall ↶ 3506 
    let mut used_r = vec![false; removed.len()];
3507 
    let mut used_a = vec![false; added.len()];
3508 
    let mut out = Vec::new();
3509 
    for (_, ri, ai) in scored {
3510 
        if used_r[ri] || used_a[ai] {
3511 
            continue;
3512 
        }
3513 
        used_r[ri] = true;
3514 
        used_a[ai] = true;
3515 
        out.push((removed[ri].0.clone(), added[ai].0.clone()));
3516 
    }
3517 
    out
3518 
}
3519
3520 
/// Percentage of the larger file's bytes found in lines the two share
3521 
/// (multiset intersection of lines), like Git's similarity index.
3522 
pub fn similarity(a: &[u8], b: &[u8]) -> u32 {
3523 
    if a == b {
3524 
        return 100;
3525 
    }
3526 
    if a.is_empty() || b.is_empty() {
3527 
        return 0;
3528 
    }
3529 
    let mut counts: HashMap<&[u8], usize> = HashMap::new();
3530 
    for l in split_lines(a) {
3531 
        *counts.entry(l).or_default() += 1;
3532 
    }
3533 
    let mut shared = 0usize;
3534 
    for l in split_lines(b) {
3535 
        if let Some(n) = counts.get_mut(l) {
3536 
            if *n > 0 {
3537 
                *n -= 1;
3538 
                shared += l.len();
3539 
            }
3540 
        }
3541 
    }
3542 
    (shared * 100 / a.len().max(b.len())) as u32
d616faf5 Andy McCall ↶ 3543 
}
3544
3545 
#[cfg(test)]
3546 
mod join_tests {
3547 
    use super::*;
f57bf670 Andy McCall ↶ 3548
3549 
    #[test]
3550 
    fn missing_from_finds_what_the_other_lacks() {
3551 
        let a: BTreeMap<String, u8> = [("a", 1), ("b", 2), ("c/d", 3), ("e", 4)]
3552 
            .into_iter()
3553 
            .map(|(k, v)| (k.to_string(), v))
3554 
            .collect();
3555 
        let b: BTreeMap<String, ()> = [("0", ()), ("b", ()), ("c", ()), ("e", ())]
3556 
            .into_iter()
3557 
            .map(|(k, v)| (k.to_string(), v))
3558 
            .collect();
3559 
        let got: Vec<&str> = missing_from(&a, &b)
3560 
            .iter()
3561 
            .map(|(k, _)| k.as_str())
3562 
            .collect();
3563 
        assert_eq!(got, vec!["a", "c/d"]);
3564 
    }
d616faf5 Andy McCall ↶ 3565
3566 
    #[test]
3567 
    fn join_pairs_every_key_in_order() {
3568 
        let a: BTreeMap<String, u8> = [("a", 1), ("b/c", 2), ("d", 3)]
3569 
            .into_iter()
3570 
            .map(|(k, v)| (k.to_string(), v))
3571 
            .collect();
3572 
        let b: BTreeMap<String, char> = [("b", 'x'), ("b/c", 'y'), ("e", 'z')]
3573 
            .into_iter()
3574 
            .map(|(k, v)| (k.to_string(), v))
3575 
            .collect();
3576 
        let got: Vec<(String, Option<u8>, Option<char>)> = join(&a, &b)
3577 
            .map(|(k, x, y)| (k.clone(), x.copied(), y.copied()))
3578 
            .collect();
3579 
        let want = vec![
3580 
            ("a".to_string(), Some(1), None),
3581 
            ("b".to_string(), None, Some('x')),
3582 
            ("b/c".to_string(), Some(2), Some('y')),
3583 
            ("d".to_string(), Some(3), None),
3584 
            ("e".to_string(), None, Some('z')),
3585 
        ];
3586 
        assert_eq!(got, want);
3587 
    }
6e723b81 Andy McCall ↶ 3588 
}