//! A repository on disk: the `.oxy` directory, one workspace, the operation //! log, and the working-copy change (spec §12, §13.1). //! //! Not yet implemented from the spec: signatures and identities (the author //! is `ObjectRef::ZERO`), packs, checkpoints, multiple workspaces, remotes, //! and resolving conflicts by editing markers. use crate::diff::{ranges, split_lines, text_edits}; use crate::hash::{hash_object, ChangeHash, ObjectRef, ObjectType}; use crate::ids::{ChangeId, NodeId, WorkspaceId}; use crate::objects::{Blob, Change, ChangeKind, Edit, Hlc, Mode, NodeKind, OpRef, TreeOp}; use crate::oplog::{Op, View, SNAPSHOT_SUFFIX}; use crate::state::{FileContent, Snapshot, State}; use crate::store::{Store, StoreError}; use crate::tree::OpKey; use std::collections::{BTreeMap, BTreeSet, HashMap}; use std::fs; use std::io::Write; use std::path::{Path, PathBuf}; pub const OXY_DIR: &str = ".oxy"; pub const FORMAT_LINE: &str = "oxy repository format 1\n"; const BINARY_SNIFF: usize = 8 * 1024; #[derive(Debug, thiserror::Error)] pub enum RepoError { #[error("io: {0}")] Io(#[from] std::io::Error), #[error(transparent)] Store(#[from] StoreError), #[error(transparent)] Object(#[from] crate::objects::ObjectError), #[error(transparent)] State(#[from] crate::state::StateError), #[error("{0}")] Msg(String), } pub type Result = std::result::Result; fn fail(s: impl Into) -> Result { Err(RepoError::Msg(s.into())) } /// Repository-local settings (`.oxy/config`, `key = value` lines). #[derive(Clone, Debug, PartialEq, Eq)] pub struct Config { pub author_name: String, pub author_email: String, pub bookmark: String, } impl Config { fn default_for_user() -> Config { let user = std::env::var("USER").unwrap_or_else(|_| "unknown".into()); Config { author_name: std::env::var("OXY_AUTHOR_NAME").unwrap_or_else(|_| user.clone()), author_email: std::env::var("OXY_AUTHOR_EMAIL").unwrap_or_else(|_| format!("{user}@localhost")), bookmark: "main".into(), } } fn load(path: &Path) -> Result { let mut c = Config::default_for_user(); for line in fs::read_to_string(path)?.lines() { let line = line.trim(); if line.is_empty() || line.starts_with('#') { continue; } let Some((k, v)) = line.split_once('=') else { continue }; let v = v.trim().trim_matches('"').to_string(); match k.trim() { "author.name" => c.author_name = v, "author.email" => c.author_email = v, "workspace.bookmark" => c.bookmark = v, _ => {} } } Ok(c) } fn save(&self, path: &Path) -> Result<()> { let text = format!( "author.name = \"{}\"\nauthor.email = \"{}\"\nworkspace.bookmark = \"{}\"\n", self.author_name, self.author_email, self.bookmark ); write_atomic(path, text.as_bytes()) } } fn write_atomic(path: &Path, data: &[u8]) -> Result<()> { let tmp = path.with_extension("tmp"); { let mut f = fs::File::create(&tmp)?; f.write_all(data)?; f.sync_all()?; } fs::rename(&tmp, path)?; Ok(()) } #[derive(Clone, Copy, Debug, PartialEq, Eq)] pub enum ChangeStatus { Added, Modified, Removed, ModeChanged, /// The file is conflicted in the base state and was edited in the working copy. ConflictEdited, } #[derive(Clone, Debug, PartialEq, Eq)] pub struct PathStatus { pub path: String, pub status: ChangeStatus, } /// Result of snapshotting the working copy. pub struct SnapshotResult { pub wc_hash: ChangeHash, pub wc: Change, /// The base state (closure of the working-copy change's dependencies). pub base: State, pub base_snapshot: Snapshot, pub statuses: Vec, /// Whether the snapshot rewrote the working-copy change. pub changed: bool, } pub struct LogEntry { pub hash: ChangeHash, pub change: Change, pub is_working_copy: bool, pub is_head: bool, } struct FsEntry { kind: NodeKind, exec: bool, content: Vec, } pub struct Repo { root: PathBuf, oxy: PathBuf, store: Store, workspace: WorkspaceId, pub config: Config, } pub fn is_binary(content: &[u8]) -> bool { content[..content.len().min(BINARY_SNIFF)].contains(&0) } impl Repo { // ------------------------------------------------------------ lifecycle pub fn init(root: &Path) -> Result { let root = root.canonicalize().unwrap_or_else(|_| root.to_path_buf()); let oxy = root.join(OXY_DIR); if oxy.exists() { return fail(format!("{} is already an oxy repository", root.display())); } fs::create_dir_all(oxy.join("ops"))?; write_atomic(&oxy.join("format"), FORMAT_LINE.as_bytes())?; let config = Config::default_for_user(); config.save(&oxy.join("config"))?; let workspace = WorkspaceId::random(); write_atomic(&oxy.join("workspace"), workspace.to_hex().as_bytes())?; let store = Store::open(oxy.join("objects"))?; let repo = Repo { root, oxy, store, workspace, config }; let wc = repo.new_working_copy(vec![], ""); let wc_hash = repo.store.put_change(&wc)?; let mut view = View::default(); view.bookmarks.insert(repo.config.bookmark.clone(), vec![]); view.workspaces.insert(workspace, wc_hash); repo.record_op(&[], "oxy init", view)?; Ok(repo) } /// Open the repository containing `start` (searching upwards). pub fn open(start: &Path) -> Result { let start = start.canonicalize()?; let mut dir: Option<&Path> = Some(&start); while let Some(d) = dir { let oxy = d.join(OXY_DIR); if oxy.is_dir() { let format = fs::read_to_string(oxy.join("format"))?; if format != FORMAT_LINE { return fail(format!("unsupported repository format: {}", format.trim())); } let config = Config::load(&oxy.join("config"))?; let workspace = WorkspaceId::from_hex(fs::read_to_string(oxy.join("workspace"))?.trim()) .ok_or_else(|| RepoError::Msg("corrupt .oxy/workspace".into()))?; let store = Store::open(oxy.join("objects"))?; return Ok(Repo { root: d.to_path_buf(), oxy, store, workspace, config }); } dir = d.parent(); } fail(format!("no oxy repository found at or above {}", start.display())) } pub fn root(&self) -> &Path { &self.root } pub fn store(&self) -> &Store { &self.store } pub fn workspace(&self) -> WorkspaceId { self.workspace } fn new_working_copy(&self, deps: Vec, message: &str) -> Change { let mut c = Change { change_id: ChangeId::random(), deps, author: ObjectRef::ZERO, hlc: Hlc::now(), message: message.into(), tree_ops: vec![], edits: vec![], kind: ChangeKind::WorkingCopy, }; c.normalize(); c } // ------------------------------------------------------------ operation log pub fn head_op(&self) -> Result<(ObjectRef, Op)> { let hex = fs::read_to_string(self.oxy.join("ops").join("head"))?; let r = ObjectRef::from_hex(hex.trim()).ok_or_else(|| RepoError::Msg("corrupt .oxy/ops/head".into()))?; let op = Op::decode(&self.store.get_typed(&r, ObjectType::Op)?)?; Ok((r, op)) } pub fn view(&self) -> Result { Ok(self.head_op()?.1.view) } fn record_op(&self, parents: &[ObjectRef], command: &str, view: View) -> Result { let op = Op { parents: parents.to_vec(), hlc: Hlc::now(), command: command.into(), view }; let r = self.store.put(ObjectType::Op, &op.encode())?; write_atomic(&self.oxy.join("ops").join("head"), r.to_hex().as_bytes())?; Ok(r) } pub fn op_log(&self) -> Result> { let mut out = Vec::new(); let (mut r, mut op) = self.head_op()?; loop { let next = op.parents.first().copied(); out.push((r, op)); match next { Some(p) => { r = p; op = Op::decode(&self.store.get_typed(&r, ObjectType::Op)?)?; } None => break, } } Ok(out) } /// Undo the last user command (spec §12.1). Operations that only recorded /// a working-copy snapshot are skipped, since the files are still on disk /// and the next snapshot recreates them. pub fn undo(&self) -> Result { let (_, mut op) = self.head_op()?; while op.is_snapshot() { let Some(parent) = op.parents.first() else { break }; op = Op::decode(&self.store.get_typed(parent, ObjectType::Op)?)?; } let Some(parent) = op.parents.first() else { return fail("nothing to undo"); }; let parent_op = Op::decode(&self.store.get_typed(parent, ObjectType::Op)?)?; write_atomic(&self.oxy.join("ops").join("head"), parent.to_hex().as_bytes())?; Ok(parent_op) } /// The last operation that was not a snapshot. pub fn last_command(&self) -> Result { let (_, mut op) = self.head_op()?; while op.is_snapshot() { let Some(parent) = op.parents.first() else { break }; op = Op::decode(&self.store.get_typed(parent, ObjectType::Op)?)?; } Ok(op) } // ------------------------------------------------------------ states /// Load the dependency closure of `heads` from the store. pub fn load_state(&self, heads: &[ChangeHash]) -> Result { let mut state = State::new(); let mut stack: Vec = heads.to_vec(); let mut seen: BTreeSet = BTreeSet::new(); while let Some(h) = stack.pop() { if !seen.insert(h) { continue; } let c = self.store.get_change(&h)?; stack.extend(c.deps.iter().copied()); state.add_unchecked(c); } state.validate()?; Ok(state) } pub fn working_copy(&self) -> Result<(ChangeHash, Change)> { let view = self.view()?; let h = *view .workspaces .get(&self.workspace) .ok_or_else(|| RepoError::Msg("this workspace is not in the current view".into()))?; Ok((h, self.store.get_change(&h)?)) } pub fn bookmark_heads(&self) -> Result> { Ok(self.view()?.bookmarks.get(&self.config.bookmark).cloned().unwrap_or_default()) } // ------------------------------------------------------------ blobs fn blob_for(content: &[u8]) -> Blob { Blob { length: content.len() as u64, chunks: crate::chunk::chunks(content).into_iter().map(|c| hash_object(ObjectType::Chunk, c)).collect(), } } pub fn blob_ref(content: &[u8]) -> ObjectRef { Self::blob_for(content).hash() } pub fn store_blob(&self, content: &[u8]) -> Result { for c in crate::chunk::chunks(content) { self.store.put_chunk(c)?; } Ok(self.store.put(ObjectType::Blob, &Self::blob_for(content).encode())?) } pub fn read_blob(&self, r: &ObjectRef) -> Result> { let blob = Blob::decode(&self.store.get_typed(r, ObjectType::Blob)?)?; let mut out = Vec::with_capacity(blob.length as usize); for c in &blob.chunks { out.extend(self.store.get_typed(c, ObjectType::Chunk)?); } Ok(out) } // ------------------------------------------------------------ working copy fn walk_working_copy(&self) -> Result> { let mut wb = ignore::WalkBuilder::new(&self.root); wb.hidden(false) .git_ignore(false) .git_global(false) .git_exclude(false) .require_git(false) .follow_links(false) .add_custom_ignore_filename(".oxyignore"); wb.filter_entry(|e| e.file_name() != OXY_DIR); let mut out = BTreeMap::new(); for entry in wb.build() { let entry = entry.map_err(|e| RepoError::Msg(e.to_string()))?; let path = entry.path(); if path == self.root { continue; } let rel = path.strip_prefix(&self.root).unwrap(); let rel: String = rel.components().map(|c| c.as_os_str().to_string_lossy().to_string()).collect::>().join("/"); let Some(ft) = entry.file_type() else { continue }; if ft.is_dir() { continue; } if ft.is_symlink() { let target = fs::read_link(path)?; out.insert(rel, FsEntry { kind: NodeKind::Symlink, exec: false, content: target.to_string_lossy().as_bytes().to_vec() }); } else if ft.is_file() { let exec = is_executable(&entry.metadata().map_err(|e| RepoError::Msg(e.to_string()))?); out.insert(rel, FsEntry { kind: NodeKind::File, exec, content: fs::read(path)? }); } } Ok(out) } /// Snapshot the working copy into the working-copy change (spec §12.3). pub fn snapshot(&self, command: &str) -> Result { let (op_ref, op) = self.head_op()?; let (old_hash, old_wc) = self.working_copy()?; let base = self.load_state(&old_wc.deps)?; let base_snapshot = base.materialize(); let fs_entries = self.walk_working_copy()?; // Reuse the NodeIds the previous working-copy version assigned to new // paths, so an unchanged working copy hashes identically (spec §12.3). let reuse = added_node_paths(&old_wc, &base_snapshot); let mut b = ChangeBuilder::new(&base, &base_snapshot, reuse); for (path, fs_entry) in &fs_entries { match base_snapshot.files.get(path) { None => b.add_path(self, path, fs_entry)?, Some(expected) => b.update_path(self, path, fs_entry, expected)?, } } for (path, expected) in &base_snapshot.files { if fs_entries.contains_key(path) { continue; } let still_needed = expected.kind == NodeKind::Dir && fs_entries.keys().any(|p| p.starts_with(&format!("{path}/"))); if !still_needed { b.remove_path(path, expected); } } let mut wc = Change { change_id: old_wc.change_id, deps: vec![], author: ObjectRef::ZERO, hlc: old_wc.hlc, message: old_wc.message.clone(), tree_ops: b.tree_ops, edits: b.edits, kind: ChangeKind::WorkingCopy, }; let base_heads = base.maximal(&old_wc.deps); let mut deps: BTreeSet = base_heads.into_iter().collect(); deps.extend(referenced_changes(&wc, &base)); wc.deps = deps.into_iter().collect(); wc.normalize(); let mut statuses = b.statuses; statuses.sort_by(|a, b| a.path.cmp(&b.path)); // Empty snapshot: hash the candidate before storing; only record an op on change. let new_hash = wc.hash(); let changed = new_hash != old_hash; if changed { self.store.put_change(&wc)?; let mut view = op.view.clone(); view.workspaces.insert(self.workspace, new_hash); self.record_op(&[op_ref], &format!("{command}{SNAPSHOT_SUFFIX}"), view)?; } Ok(SnapshotResult { wc_hash: new_hash, wc, base, base_snapshot, statuses, changed }) } /// Turn the working-copy change into a normal change on the bookmark (spec §12.3). pub fn commit(&self, message: &str, extra_deps: &[ChangeHash]) -> Result { let snap = self.snapshot("oxy commit")?; if snap.wc.tree_ops.is_empty() && snap.wc.edits.is_empty() { return fail("nothing to commit: the working copy matches its base"); } if snap.statuses.iter().any(|s| s.status == ChangeStatus::ConflictEdited) { return fail("a conflicted file was edited; resolving conflicts is not implemented yet"); } if message.trim().is_empty() { return fail("a commit message is required"); } let (op_ref, op) = self.head_op()?; let mut deps: BTreeSet = referenced_changes(&snap.wc, &snap.base); for d in extra_deps { if !snap.base.contains(d) { return fail(format!("--depends-on {d}: not in the working copy's base state")); } deps.insert(*d); } let mut change = Change { change_id: snap.wc.change_id, deps: deps.into_iter().collect(), author: ObjectRef::ZERO, hlc: Hlc::now(), message: message.trim_end().to_string(), tree_ops: snap.wc.tree_ops.clone(), edits: snap.wc.edits.clone(), kind: ChangeKind::Normal, }; change.normalize(); let hash = self.store.put_change(&change)?; // New bookmark heads: maximal elements of old heads plus the new change. let old_heads = op.view.bookmarks.get(&self.config.bookmark).cloned().unwrap_or_default(); let mut with_new: Vec = old_heads.clone(); with_new.push(hash); let mut st = self.load_state(&with_new)?; st.add(change)?; let new_heads = st.maximal(&with_new); let wc = self.new_working_copy(new_heads.clone(), ""); let wc_hash = self.store.put_change(&wc)?; let mut view = op.view.clone(); view.bookmarks.insert(self.config.bookmark.clone(), new_heads); view.workspaces.insert(self.workspace, wc_hash); self.record_op(&[op_ref], &format!("oxy commit -m {message:?}"), view)?; Ok(hash) } /// Move the working copy to the state whose heads are `heads` (spec §12.3): /// the working directory is rewritten to that state and a fresh /// working-copy change is started on it. The bookmark does not move. pub fn checkout(&self, heads: &[ChangeHash], discard: bool, command: &str) -> Result { let snap = self.snapshot(command)?; if !discard && (!snap.wc.tree_ops.is_empty() || !snap.wc.edits.is_empty()) { return fail("the working copy has changes; commit them first or pass --discard"); } let target_state = self.load_state(heads)?; let heads = target_state.maximal(heads); let target = target_state.materialize(); let current = self.walk_working_copy()?; let mut report = CheckoutReport::default(); // Remove tracked paths that the target does not have. let mut dirs_to_prune: BTreeSet = BTreeSet::new(); for path in current.keys() { if !target.files.contains_key(path) { let full = self.safe_path(path)?; fs::remove_file(&full)?; report.removed += 1; if let Some(parent) = full.parent() { dirs_to_prune.insert(parent.to_path_buf()); } } } // Write target files that differ. for (path, f) in &target.files { match &f.content { FileContent::Dir => continue, FileContent::Text { bytes, .. } => { let same = current.get(path).is_some_and(|c| c.kind == NodeKind::File && &c.content == bytes && c.exec == f.mode.exec); if !same { self.write_file(path, bytes, f.mode.exec)?; report.written += 1; } } FileContent::Opaque { blobs } => { let Some(blob) = blobs.first() else { continue }; let bytes = self.read_blob(blob)?; if f.kind == NodeKind::Symlink { let same = current.get(path).is_some_and(|c| c.kind == NodeKind::Symlink && c.content == bytes); if !same { self.write_symlink(path, &bytes)?; report.written += 1; } } else { let same = current.get(path).is_some_and(|c| c.kind == NodeKind::File && c.content == bytes && c.exec == f.mode.exec); if !same { self.write_file(path, &bytes, f.mode.exec)?; report.written += 1; } } if blobs.len() > 1 { for (i, b) in blobs.iter().enumerate().skip(1) { self.write_file(&format!("{path}.oxy-conflict-{}", i + 1), &self.read_blob(b)?, f.mode.exec)?; } } } } } // Prune directories left empty, deepest first. for d in dirs_to_prune.iter().rev() { let mut d = d.clone(); while d != self.root && fs::read_dir(&d).map(|mut i| i.next().is_none()).unwrap_or(false) { fs::remove_dir(&d)?; let Some(p) = d.parent() else { break }; d = p.to_path_buf(); } } let (op_ref, op) = self.head_op()?; let wc = self.new_working_copy(heads, ""); let wc_hash = self.store.put_change(&wc)?; let mut view = op.view.clone(); view.workspaces.insert(self.workspace, wc_hash); self.record_op(&[op_ref], command, view)?; report.conflicted = target.is_conflicted(); Ok(report) } /// The absolute path for a repository path, refusing anything that would /// escape the workspace or write through a symlink (spec §7.7). fn safe_path(&self, rel: &str) -> Result { let mut full = self.root.clone(); let comps: Vec<&str> = rel.split('/').collect(); for (i, comp) in comps.iter().enumerate() { if comp.is_empty() || *comp == "." || *comp == ".." || comp.contains('\\') { return fail(format!("refusing unsafe path {rel:?}")); } if i == 0 && *comp == OXY_DIR { return fail(format!("refusing to write inside {OXY_DIR}: {rel:?}")); } if i + 1 < comps.len() { full.push(comp); if let Ok(m) = fs::symlink_metadata(&full) { if m.file_type().is_symlink() { return fail(format!("refusing to write through symlink {}", full.display())); } } } else { full.push(comp); } } Ok(full) } fn write_file(&self, rel: &str, bytes: &[u8], exec: bool) -> Result<()> { let full = self.safe_path(rel)?; if let Some(parent) = full.parent() { fs::create_dir_all(parent)?; } if fs::symlink_metadata(&full).is_ok_and(|m| m.file_type().is_symlink() || m.is_dir()) { remove_any(&full)?; } fs::write(&full, bytes)?; set_executable(&full, exec)?; Ok(()) } fn write_symlink(&self, rel: &str, target: &[u8]) -> Result<()> { let full = self.safe_path(rel)?; if let Some(parent) = full.parent() { fs::create_dir_all(parent)?; } if fs::symlink_metadata(&full).is_ok() { remove_any(&full)?; } #[cfg(unix)] { std::os::unix::fs::symlink(String::from_utf8_lossy(target).as_ref(), &full)?; Ok(()) } #[cfg(not(unix))] { // Spec §7.7: without symlink support, write the target as a plain file. fs::write(&full, target)?; Ok(()) } } /// Changes reachable from the bookmark and the working copy, newest first. pub fn log(&self) -> Result> { let (wc_hash, _) = self.working_copy()?; let mut heads = self.bookmark_heads()?; heads.push(wc_hash); let state = self.load_state(&heads)?; let head_set: BTreeSet = self.bookmark_heads()?.into_iter().collect(); let mut entries: Vec = state .changes() .map(|(h, c)| LogEntry { hash: *h, change: c.clone(), is_working_copy: *h == wc_hash, is_head: head_set.contains(h) }) .collect(); entries.sort_by(|a, b| b.change.hlc.cmp(&a.change.hlc).then(b.hash.cmp(&a.hash))); Ok(entries) } /// Resolve a change by ChangeId prefix, `#hash` prefix, `@` (working copy), or bookmark. pub fn resolve_change(&self, spec: &str) -> Result { if spec == "@" { return Ok(self.working_copy()?.0); } if spec.is_empty() || spec == "#" { return fail("empty change specifier"); } let entries = self.log()?; let matches: Vec<&LogEntry> = if let Some(hp) = spec.strip_prefix('#') { entries.iter().filter(|e| e.hash.starts_with_hex(hp)).collect() } else { entries.iter().filter(|e| e.change.change_id.starts_with_hex(spec)).collect() }; match matches.len() { 1 => Ok(matches[0].hash), 0 => fail(format!("no change matches {spec:?}")), _ => { let ids: BTreeSet = matches.iter().map(|e| e.change.change_id).collect(); if ids.len() == 1 { fail(format!("change {spec} is divergent: {} variants", matches.len())) } else { fail(format!("{spec:?} is ambiguous")) } } } } /// Human-readable description of what a change does, with line content. pub fn describe_change(&self, hash: &ChangeHash) -> Result { let change = self.store.get_change(hash)?; let base = self.load_state(&change.deps)?; let base_snapshot = base.materialize(); Ok(render_change(&change, &base, &base_snapshot)) } } #[derive(Debug, Default, Clone, PartialEq, Eq)] pub struct CheckoutReport { pub written: usize, pub removed: usize, pub conflicted: bool, } fn remove_any(path: &Path) -> std::io::Result<()> { let m = fs::symlink_metadata(path)?; if m.is_dir() { fs::remove_dir_all(path) } else { fs::remove_file(path) } } #[cfg(unix)] fn set_executable(path: &Path, exec: bool) -> std::io::Result<()> { use std::os::unix::fs::PermissionsExt; let mut perms = fs::metadata(path)?.permissions(); let mode = perms.mode(); let want = if exec { mode | 0o111 } else { mode & !0o111 }; if want != mode { perms.set_mode(want); fs::set_permissions(path, perms)?; } Ok(()) } #[cfg(not(unix))] fn set_executable(_path: &Path, _exec: bool) -> std::io::Result<()> { Ok(()) } #[cfg(unix)] fn is_executable(m: &fs::Metadata) -> bool { use std::os::unix::fs::PermissionsExt; m.permissions().mode() & 0o111 != 0 } #[cfg(not(unix))] fn is_executable(_m: &fs::Metadata) -> bool { false } fn op_ref(k: OpKey) -> OpRef { OpRef { change: k.0, index: k.1 } } fn split_path(path: &str) -> (&str, &str) { match path.rfind('/') { Some(i) => (&path[..i], &path[i + 1..]), None => ("", path), } } /// Builds the tree ops and edits of a working-copy change against a base. struct ChangeBuilder<'a> { base: &'a State, base_snapshot: &'a Snapshot, tree_ops: Vec, edits: Vec, statuses: Vec, dir_nodes: HashMap, reuse: HashMap, } impl<'a> ChangeBuilder<'a> { fn new(base: &'a State, base_snapshot: &'a Snapshot, reuse: HashMap) -> Self { let mut dir_nodes: HashMap = HashMap::new(); dir_nodes.insert(String::new(), NodeId::ROOT); for (p, f) in &base_snapshot.files { if f.kind == NodeKind::Dir { dir_nodes.insert(p.clone(), f.node); } } ChangeBuilder { base, base_snapshot, tree_ops: vec![], edits: vec![], statuses: vec![], dir_nodes, reuse } } fn reused_node(&self, path: &str, kind: NodeKind) -> Option { self.reuse.get(path).filter(|(_, k)| *k == kind).map(|(n, _)| *n) } fn ensure_dir(&mut self, path: &str) -> NodeId { if let Some(n) = self.dir_nodes.get(path) { return *n; } let mut parent = NodeId::ROOT; let mut prefix = String::new(); for comp in path.split('/') { if !prefix.is_empty() { prefix.push('/'); } prefix.push_str(comp); parent = match self.dir_nodes.get(&prefix) { Some(n) => *n, None => { let node = self.reused_node(&prefix, NodeKind::Dir).unwrap_or_else(NodeId::random); self.tree_ops.push(TreeOp::AddNode { node, kind: NodeKind::Dir, mode: Mode::default(), parent, name: comp.to_string() }); self.dir_nodes.insert(prefix.clone(), node); node } }; } parent } fn add_path(&mut self, repo: &Repo, path: &str, e: &FsEntry) -> Result<()> { let (dir, name) = split_path(path); let parent = self.ensure_dir(dir); let node = self.reused_node(path, e.kind).unwrap_or_else(NodeId::random); let opaque = e.kind == NodeKind::Symlink || is_binary(&e.content); self.tree_ops.push(TreeOp::AddNode { node, kind: e.kind, mode: Mode { exec: e.exec, opaque }, parent, name: name.to_string() }); if opaque { let blob = repo.store_blob(&e.content)?; self.edits.push(Edit::Replace { node, old: None, new: Some(blob) }); } else if !e.content.is_empty() { self.edits.push(Edit::Insert { node, after: crate::objects::AtomRef::Start, before: crate::objects::AtomRef::End, lines: split_lines(&e.content).iter().map(|l| l.to_vec()).collect(), }); } self.statuses.push(PathStatus { path: path.to_string(), status: ChangeStatus::Added }); Ok(()) } fn remove_path(&mut self, path: &str, expected: &crate::state::FileState) { let entry = &self.base_snapshot.tree.entries[path]; self.tree_ops.push(TreeOp::RemoveNode { node: expected.node, placement: op_ref(entry.placement) }); self.statuses.push(PathStatus { path: path.to_string(), status: ChangeStatus::Removed }); } fn update_path(&mut self, repo: &Repo, path: &str, e: &FsEntry, expected: &crate::state::FileState) -> Result<()> { if expected.kind != e.kind { self.remove_path(path, expected); self.statuses.pop(); self.add_path(repo, path, e)?; self.statuses.pop(); self.statuses.push(PathStatus { path: path.to_string(), status: ChangeStatus::Modified }); return Ok(()); } let entry = &self.base_snapshot.tree.entries[path]; let node = expected.node; let mode_atom = entry.mode_atom.map(op_ref); let mut status: Option = None; match &expected.content { FileContent::Dir => {} FileContent::Text { bytes, materialized } => { if materialized.is_conflicted() { if bytes != &e.content { self.statuses.push(PathStatus { path: path.to_string(), status: ChangeStatus::ConflictEdited }); } return Ok(()); } if bytes != &e.content { if is_binary(&e.content) { // Text → opaque (spec §7.5). if let Some(old) = mode_atom { self.tree_ops.push(TreeOp::SetMode { node, old, mode: Mode { exec: e.exec, opaque: true } }); } let live: Vec<_> = repo_live_keys(self.base, &node); if !live.is_empty() { self.edits.push(Edit::Delete { node, atoms: ranges(&live) }); } let blob = repo.store_blob(&e.content)?; self.edits.push(Edit::Replace { node, old: None, new: Some(blob) }); status = Some(ChangeStatus::Modified); } else { let old = self.base.live_lines(&node); let edits = text_edits(node, &old, &e.content); if !edits.is_empty() { self.edits.extend(edits); status = Some(ChangeStatus::Modified); } } } let switched_to_opaque = status == Some(ChangeStatus::Modified) && is_binary(&e.content); if !switched_to_opaque && e.exec != expected.mode.exec { if let Some(old) = mode_atom { self.tree_ops.push(TreeOp::SetMode { node, old, mode: Mode { exec: e.exec, opaque: false } }); status.get_or_insert(ChangeStatus::ModeChanged); } } } FileContent::Opaque { blobs } => { let new_ref = Repo::blob_ref(&e.content); if blobs.len() > 1 { if !blobs.contains(&new_ref) { self.statuses.push(PathStatus { path: path.to_string(), status: ChangeStatus::ConflictEdited }); } return Ok(()); } let current = blobs.first().copied(); if current != Some(new_ref) { if e.kind == NodeKind::File && !is_binary(&e.content) { // Opaque → text (spec §7.5). if let Some(old) = mode_atom { self.tree_ops.push(TreeOp::SetMode { node, old, mode: Mode { exec: e.exec, opaque: false } }); } self.edits.push(Edit::Replace { node, old: current, new: None }); if !e.content.is_empty() { self.edits.push(Edit::Insert { node, after: crate::objects::AtomRef::Start, before: crate::objects::AtomRef::End, lines: split_lines(&e.content).iter().map(|l| l.to_vec()).collect(), }); } } else { let blob = repo.store_blob(&e.content)?; self.edits.push(Edit::Replace { node, old: current, new: Some(blob) }); } status = Some(ChangeStatus::Modified); } else if e.kind == NodeKind::File && e.exec != expected.mode.exec { if let Some(old) = mode_atom { self.tree_ops.push(TreeOp::SetMode { node, old, mode: Mode { exec: e.exec, opaque: true } }); status = Some(ChangeStatus::ModeChanged); } } } } if let Some(s) = status { self.statuses.push(PathStatus { path: path.to_string(), status: s }); } Ok(()) } } fn repo_live_keys(base: &State, node: &NodeId) -> Vec { base.live_lines(node).into_iter().map(|(k, _)| k).collect() } /// The changes a change's scope references (spec §8.1): introducers of every /// atom, placement, and mode it touches, and of every node it operates on. pub fn referenced_changes(change: &Change, base: &State) -> BTreeSet { let mut out = BTreeSet::new(); let opref = |r: &OpRef, out: &mut BTreeSet| { if !r.change.is_zero() { out.insert(r.change); } }; let mut nodes: Vec = Vec::new(); for op in &change.tree_ops { match op { TreeOp::AddNode { parent, .. } => nodes.push(*parent), TreeOp::RemoveNode { node, placement } => { nodes.push(*node); opref(placement, &mut out); } TreeOp::Move { node, old, parent, .. } => { nodes.push(*node); nodes.push(*parent); opref(old, &mut out); } TreeOp::SetMode { node, old, .. } => { nodes.push(*node); opref(old, &mut out); } } } for e in &change.edits { nodes.push(e.node()); match e { Edit::Insert { after, before, .. } => { for r in [after, before] { if let crate::objects::AtomRef::Atom { change, .. } = r { if !change.is_zero() { out.insert(*change); } } } } Edit::Delete { atoms, .. } => { for r in atoms { if !r.change.is_zero() { out.insert(r.change); } } } Edit::Replace { .. } => {} } } for n in nodes { if let Some(h) = base.tree().introducer(&n) { out.insert(h); } } out } /// Paths of the nodes a change adds on top of `base_snapshot`. pub fn added_node_paths(change: &Change, base_snapshot: &Snapshot) -> HashMap { let by_node = base_snapshot.tree.by_node(); let mut new_names: HashMap = HashMap::new(); for op in &change.tree_ops { if let TreeOp::AddNode { node, parent, name, kind, .. } = op { new_names.insert(*node, (*parent, name.clone(), *kind)); } } let mut out = HashMap::new(); for (node, (_, _, kind)) in &new_names { if let Some(path) = path_of_node(node, &by_node, &new_names) { out.insert(path, (*node, *kind)); } } out } fn path_of_node( n: &NodeId, by_node: &HashMap, new_names: &HashMap, ) -> Option { if let Some(e) = by_node.get(n) { return Some(e.path.clone()); } let mut parts = Vec::new(); let mut cur = *n; loop { if cur.is_root() { break; } if let Some(e) = by_node.get(&cur) { if !e.path.is_empty() { parts.push(e.path.clone()); } break; } let (parent, name, _) = new_names.get(&cur)?; parts.push(name.clone()); cur = *parent; } parts.reverse(); Some(parts.join("/")) } /// Render a change's operations with line content, using `base` for deleted lines. pub fn render_change(change: &Change, base: &State, base_snapshot: &Snapshot) -> String { let by_node = base_snapshot.tree.by_node(); let mut new_names: HashMap = HashMap::new(); for op in &change.tree_ops { if let TreeOp::AddNode { node, parent, name, kind, .. } = op { new_names.insert(*node, (*parent, name.clone(), *kind)); } } let path_of = |n: &NodeId| -> String { path_of_node(n, &by_node, &new_names).unwrap_or_else(|| format!("", n.short(8))) }; let mut out = String::new(); for op in &change.tree_ops { match op { TreeOp::AddNode { node, kind, mode, .. } => { out.push_str(&format!("A {} ({:?}{}{})\n", path_of(node), kind, if mode.exec { ", exec" } else { "" }, if mode.opaque { ", opaque" } else { "" })); } TreeOp::RemoveNode { node, .. } => out.push_str(&format!("D {}\n", path_of(node))), TreeOp::Move { node, name, .. } => out.push_str(&format!("R {} -> {}\n", path_of(node), name)), TreeOp::SetMode { node, mode, .. } => { out.push_str(&format!("M {} mode: exec={} opaque={}\n", path_of(node), mode.exec, mode.opaque)); } } } let mut last_node: Option = None; for e in &change.edits { let n = e.node(); if last_node != Some(n) { out.push_str(&format!("--- {}\n", path_of(&n))); last_node = Some(n); } match e { Edit::Insert { lines, .. } => { for l in lines { out.push('+'); out.push_str(&String::from_utf8_lossy(l)); if !l.ends_with(b"\n") { out.push('\n'); } } } Edit::Delete { atoms, .. } => { for r in atoms { for i in r.from..r.to { let content = base.file(&n).and_then(|g| g.content(&(r.change, i))).unwrap_or(b"\n"); out.push('-'); out.push_str(&String::from_utf8_lossy(content)); if !content.ends_with(b"\n") { out.push('\n'); } } } } Edit::Replace { old, new, .. } => { out.push_str(&format!( " opaque content {} -> {}\n", old.map(|o| o.short(12)).unwrap_or_else(|| "none".into()), new.map(|o| o.short(12)).unwrap_or_else(|| "none".into()) )); } } } out } #[cfg(test)] mod tests { use super::*; fn write(root: &Path, rel: &str, content: &[u8]) { let p = root.join(rel); fs::create_dir_all(p.parent().unwrap()).unwrap(); fs::write(p, content).unwrap(); } fn statuses(s: &SnapshotResult) -> Vec<(String, ChangeStatus)> { s.statuses.iter().map(|p| (p.path.clone(), p.status)).collect() } #[test] fn init_snapshot_commit_log_undo() { let dir = tempfile::tempdir().unwrap(); let root = dir.path(); let repo = Repo::init(root).unwrap(); assert!(Repo::init(root).is_err()); let ops = repo.op_log().unwrap(); assert_eq!(ops.len(), 1); assert_eq!(ops[0].1.command, "oxy init"); // Empty working copy: snapshot records no operation (spec §12.3). let s = repo.snapshot("oxy status").unwrap(); assert!(!s.changed); assert_eq!(repo.op_log().unwrap().len(), 1); write(root, "src/main.rs", b"fn main() {}\n"); write(root, "README.md", b"# Hello\n"); write(root, ".oxyignore", b"/target\n"); write(root, "target/junk.o", b"\0\0\0"); let s = repo.snapshot("oxy status").unwrap(); assert!(s.changed); assert_eq!( statuses(&s), vec![ (".oxyignore".into(), ChangeStatus::Added), ("README.md".into(), ChangeStatus::Added), ("src/main.rs".into(), ChangeStatus::Added), ] ); assert_eq!(repo.op_log().unwrap().len(), 2); // Snapshotting again with nothing changed records no op. let s2 = repo.snapshot("oxy status").unwrap(); assert!(!s2.changed); assert_eq!(s2.wc_hash, s.wc_hash); assert_eq!(repo.op_log().unwrap().len(), 2); let h1 = repo.commit("Initial import", &[]).unwrap(); let c1 = repo.store().get_change(&h1).unwrap(); assert_eq!(c1.kind, ChangeKind::Normal); assert!(c1.deps.is_empty(), "first change has nothing to depend on"); assert_eq!(repo.bookmark_heads().unwrap(), vec![h1]); let (wc_hash, wc) = repo.working_copy().unwrap(); assert_eq!(wc.deps, vec![h1]); assert!(wc.edits.is_empty()); assert!(repo.commit("nothing", &[]).is_err()); // Modify one file, remove another, add a binary file. write(root, "src/main.rs", b"fn main() {\n println!(\"hi\");\n}\n"); fs::remove_file(root.join("README.md")).unwrap(); write(root, "logo.png", b"\x89PNG\0\0\x01"); let s = repo.snapshot("oxy status").unwrap(); assert_eq!( statuses(&s), vec![ ("README.md".into(), ChangeStatus::Removed), ("logo.png".into(), ChangeStatus::Added), ("src/main.rs".into(), ChangeStatus::Modified), ] ); assert_ne!(s.wc_hash, wc_hash); // The rewritten working copy keeps its ChangeId and depends on h1. assert_eq!(s.wc.change_id, wc.change_id); assert!(s.wc.deps.contains(&h1)); let h2 = repo.commit("Second", &[]).unwrap(); let c2 = repo.store().get_change(&h2).unwrap(); assert_eq!(c2.deps, vec![h1], "everything it touches was introduced by h1"); assert_eq!(repo.bookmark_heads().unwrap(), vec![h2]); // The materialized head state matches the working copy. let st = repo.load_state(&[h2]).unwrap(); let snap = st.materialize(); assert!(!snap.is_conflicted()); assert_eq!(snap.files.keys().cloned().collect::>(), [".oxyignore", "logo.png", "src", "src/main.rs"]); assert!(matches!(&snap.files["src/main.rs"].content, FileContent::Text { bytes, .. } if bytes == b"fn main() {\n println!(\"hi\");\n}\n")); assert!(snap.files["logo.png"].mode.opaque); if let FileContent::Opaque { blobs } = &snap.files["logo.png"].content { assert_eq!(repo.read_blob(&blobs[0]).unwrap(), b"\x89PNG\0\0\x01"); } else { panic!("logo should be opaque"); } // Log: newest first, working copy included. let log = repo.log().unwrap(); assert_eq!(log.len(), 3); assert!(log[0].is_working_copy); assert_eq!(log[1].hash, h2); assert!(log[1].is_head); assert_eq!(log[2].hash, h1); let d = repo.describe_change(&h2).unwrap(); assert!(d.contains("D README.md"), "{d}"); assert!(d.contains("+ println!(\"hi\");"), "{d}"); assert!(d.contains("-fn main() {}"), "{d}"); // Resolution by prefix. assert_eq!(repo.resolve_change(&c2.change_id.short(6)).unwrap(), h2); assert_eq!(repo.resolve_change(&format!("#{}", h1.short(8))).unwrap(), h1); assert!(repo.resolve_change("zzzz").is_err()); // Undo the commit: bookmark and working copy go back. repo.undo().unwrap(); assert_eq!(repo.bookmark_heads().unwrap(), vec![h1]); let s = repo.snapshot("oxy status").unwrap(); assert!(!s.changed, "files on disk still match the pre-commit working copy"); assert_eq!(s.statuses.len(), 3); } #[test] fn checkout_moves_the_working_copy() { let dir = tempfile::tempdir().unwrap(); let root = dir.path(); let repo = Repo::init(root).unwrap(); write(root, "a.txt", b"one\n"); write(root, "sub/deep/b.txt", b"b\n"); let h1 = repo.commit("first", &[]).unwrap(); write(root, "a.txt", b"two\n"); fs::remove_file(root.join("sub/deep/b.txt")).unwrap(); write(root, "bin.dat", &[0, 1, 2, 3]); let h2 = repo.commit("second", &[]).unwrap(); let r = repo.checkout(&[h1], false, "oxy checkout").unwrap(); assert_eq!(r, CheckoutReport { written: 2, removed: 1, conflicted: false }); assert_eq!(fs::read(root.join("a.txt")).unwrap(), b"one\n"); assert_eq!(fs::read(root.join("sub/deep/b.txt")).unwrap(), b"b\n"); assert!(!root.join("bin.dat").exists()); let (_, wc) = repo.working_copy().unwrap(); assert_eq!(wc.deps, vec![h1]); assert!(!repo.snapshot("status").unwrap().changed, "checkout leaves a clean working copy"); // The bookmark still points at h2. assert_eq!(repo.bookmark_heads().unwrap(), vec![h2]); // Dirty working copy refuses without --discard. write(root, "a.txt", b"dirty\n"); assert!(repo.checkout(&[h2], false, "oxy checkout").is_err()); let r = repo.checkout(&[h2], true, "oxy checkout --discard").unwrap(); assert_eq!(fs::read(root.join("a.txt")).unwrap(), b"two\n"); assert_eq!(fs::read(root.join("bin.dat")).unwrap(), [0, 1, 2, 3]); assert!(!root.join("sub").exists(), "empty directories are pruned"); assert_eq!(r.removed, 1); assert!(!repo.snapshot("status").unwrap().changed); // Refuses to write through a symlinked directory. #[cfg(unix)] { fs::remove_dir_all(root.join("sub")).ok(); std::os::unix::fs::symlink("/tmp", root.join("sub")).unwrap(); assert!(repo.checkout(&[h1], true, "oxy checkout").is_err()); } } #[test] fn large_files_are_chunked_and_read_back() { let dir = tempfile::tempdir().unwrap(); let repo = Repo::init(dir.path()).unwrap(); let mut big = vec![0u8; 700 * 1024]; for (i, b) in big.iter_mut().enumerate() { *b = (i as u32).wrapping_mul(2654435761).to_le_bytes()[0]; } let r = repo.store_blob(&big).unwrap(); let blob = Blob::decode(&repo.store().get_typed(&r, ObjectType::Blob).unwrap()).unwrap(); assert!(blob.chunks.len() >= 3, "{}", blob.chunks.len()); assert_eq!(repo.read_blob(&r).unwrap(), big); assert_eq!(Repo::blob_ref(&big), r); } #[test] fn undo_skips_snapshot_operations() { let dir = tempfile::tempdir().unwrap(); let root = dir.path(); let repo = Repo::init(root).unwrap(); write(root, "a.txt", b"a\n"); let h1 = repo.commit("a", &[]).unwrap(); // A later command snapshots new edits on top of the commit. write(root, "b.txt", b"b\n"); let s = repo.snapshot("oxy status").unwrap(); assert!(s.changed); assert!(repo.head_op().unwrap().1.is_snapshot()); assert_eq!(repo.last_command().unwrap().command, "oxy commit -m \"a\""); // Undo must undo the commit, not the snapshot. repo.undo().unwrap(); assert_eq!(repo.bookmark_heads().unwrap(), vec![]); let s = repo.snapshot("oxy status").unwrap(); let paths: Vec<_> = s.statuses.iter().map(|p| p.path.as_str()).collect(); assert_eq!(paths, ["a.txt", "b.txt"], "both files are back in the working copy"); assert!(repo.store().get_change(&h1).unwrap().deps.is_empty()); // Only snapshots remain above `init`, so there is nothing left to undo. assert_eq!(repo.last_command().unwrap().command, "oxy init"); assert!(repo.undo().is_err()); } #[test] fn independent_changes_commute_and_make_multiple_heads() { let dir = tempfile::tempdir().unwrap(); let root = dir.path(); let repo = Repo::init(root).unwrap(); write(root, "a.txt", b"a\n"); let h1 = repo.commit("a", &[]).unwrap(); write(root, "b.txt", b"b\n"); let h2 = repo.commit("b", &[]).unwrap(); let c2 = repo.store().get_change(&h2).unwrap(); assert!(c2.deps.is_empty(), "b.txt touches nothing from h1"); let mut heads = repo.bookmark_heads().unwrap(); heads.sort(); let mut want = vec![h1, h2]; want.sort(); assert_eq!(heads, want); // Editing a.txt depends on h1 only. write(root, "a.txt", b"a\na2\n"); let h3 = repo.commit("a2", &[]).unwrap(); assert_eq!(repo.store().get_change(&h3).unwrap().deps, vec![h1]); let heads: BTreeSet<_> = repo.bookmark_heads().unwrap().into_iter().collect(); assert_eq!(heads, [h2, h3].into_iter().collect()); // Explicit dependency. write(root, "c.txt", b"c\n"); let h4 = repo.commit("c", &[h2]).unwrap(); assert_eq!(repo.store().get_change(&h4).unwrap().deps, vec![h2]); } #[test] fn mode_change_and_text_binary_switch() { let dir = tempfile::tempdir().unwrap(); let root = dir.path(); let repo = Repo::init(root).unwrap(); write(root, "run.sh", b"#!/bin/sh\n"); repo.commit("add script", &[]).unwrap(); #[cfg(unix)] { use std::os::unix::fs::PermissionsExt; fs::set_permissions(root.join("run.sh"), fs::Permissions::from_mode(0o755)).unwrap(); let s = repo.snapshot("status").unwrap(); assert_eq!(statuses(&s), vec![("run.sh".into(), ChangeStatus::ModeChanged)]); let h = repo.commit("exec", &[]).unwrap(); let snap = repo.load_state(&[h]).unwrap().materialize(); assert!(snap.files["run.sh"].mode.exec); } // Text → binary → text. write(root, "run.sh", b"\0binary"); let h = repo.commit("binary", &[]).unwrap(); let snap = repo.load_state(&[h]).unwrap().materialize(); assert!(snap.files["run.sh"].mode.opaque); write(root, "run.sh", b"text again\n"); let h = repo.commit("text", &[]).unwrap(); let snap = repo.load_state(&[h]).unwrap().materialize(); assert!(!snap.files["run.sh"].mode.opaque); assert!(matches!(&snap.files["run.sh"].content, FileContent::Text { bytes, .. } if bytes == b"text again\n")); assert!(!snap.is_conflicted()); } }