flagon-io/g1t

public

1	//! The repos service: repository metadata, contents, forks, landing, and
2	//! git over HTTPS.
3	//!
4	//! Other services reach it over `POST /rpc/<method>`; see
5	//! `g1t_contracts::repos` for the methods and their arguments. Any other
6	//! request is treated as git's smart HTTP protocol.
7
8	mod about;
9	mod archive_links;
10	mod backups;
11	mod blame;
12	mod catch_up;
13	mod coalesce;
14	mod commit_file;
15	mod contributors;
16	mod copy;
17	mod diff;
18	mod drift;
19	mod fallback;
20	mod forks;
21	mod git_http;
22	mod git_ops;
23	mod hidden_refs;
24	mod import;
25	mod land;
26	mod languages;
27	mod last_commits;
28	mod license;
29	mod lifecycle;
30	mod limits;
31	mod listing;
32	mod meters;
33	mod moves;
34	mod namespaces;
35	mod pack_cache;
36	mod pack_limits;
37	mod push_checks;
38	mod push_commits;
39	mod refs;
40	mod refs_cache;
41	mod registry;
42	mod resilience;
43	mod rule_facts;
44	mod rules;
45	mod run_access;
46	mod secret_scan;
47	mod shards;
48	mod shared;
49	mod signatures;
50	mod stats;
51	mod store;
52	mod test_merge;
53	mod transfer;
54	mod workflow_gate;
55
56	use g1t_contracts::events::{
57	Event, GitDeleted, GitPush, NewEvent, Publish, RepoCreated, RepoForked, RepoUpdated, WorkspaceDeleted,
58	WorkspaceDeleting, WorkspaceRenamed, WorkspaceRestored,
59	};
60	use g1t_contracts::access::{self, Capability};
61	use g1t_contracts::repos::*;
62	use g1t_contracts::time::rfc3339;
63	use g1t_contracts::{FailureCode, Outcome, PrincipalKind, Role, User, Viewer, is_valid_repo_name, new_id};
64	use g1t_kit::{args, now_ms, reply, rpc_method};
65	use std::collections::{HashMap, HashSet, VecDeque};
66	use std::rc::Rc;
67
68	use serde::Serialize;
69	use worker::{
70	Context, Env, Fetcher, MessageBatch, MessageExt, Method, Request, Response, Result, ScheduleContext, ScheduledEvent,
71	event,
72	};
73
74	use registry::{Registry, can_read, can_write, store_key};
75	use store::{ArtifactsStore, GitRepo, GitStore, Scope};
76
77	/// Namespace that holds every pull request's fork: `pulls/<pull id>`.
78	pub(crate) const PULLS_NAMESPACE: &str = "pulls";
79	const MAX_TEXT_BYTES: usize = 512 * 1024;
80	/// How far back a pull request may have forked and still be landed.
81	const MAX_ANCESTRY: u32 = 1000;
82	/// The most files one `compare` asked for by path reads.
83	const MAX_COMPARED_PATHS: usize = 50;
84	/// The most tags a repository's Tags page reads and lists.
85	const MAX_TAGS_READ: usize = 100;
86	/// Branch heads measured in one `branch_drift` call.
87	const MAX_DRIFT_HEADS: usize = 100;
88	/// How long a `last_commits` walk asked without a budget runs before it
89	/// stops and keeps its progress for the next call. The site asks from a
90	/// waitUntil, which may run 30 s past its response.
91	const LAST_COMMITS_WALK_MS: u64 = 20_000;
92	/// How long a path's history may walk when the caller gives no budget.
93	const PATH_LOG_WALK_MS: u64 = 10_000;
94
95	/// One path segment, percent-encoded for a cache key.
96	fn urlencoding_segment(segment: &str) -> String {
97	segment
98	.bytes()
99	.map(\|b\| if b.is_ascii_alphanumeric() \|\| b"-._~".contains(&b) { (b as char).to_string() } else { format!("%{b:02X}") })
100	.collect()
101	}
102	const MAX_DESCRIPTION_CHARS: usize = 200;
103	pub(crate) const SOURCE: &str = "repos";
104	pub(crate) const UNVERIFIED: &str = "Confirm your email address first. Check your inbox, or resend the link from the banner on g1t.sh.";
105
106	pub(crate) fn not_found<T>() -> Outcome<T> {
107	Outcome::fail(FailureCode::NotFound, "Repository not found.")
108	}
109
110	/// Decoded text, or `None` when the file is too large or looks binary.
111	/// Whether a ref is a full commit hash rather than a branch name.
112	fn is_commit_hash(git_ref: &str) -> bool {
113	git_ref.len() == 40 && git_ref.bytes().all(\|b\| b.is_ascii_hexdigit())
114	}
115
116	fn text_of(bytes: Vec<u8>) -> Option<String> {
117	if bytes.len() > MAX_TEXT_BYTES \|\| bytes.contains(&0) {
118	return None;
119	}
120	Some(String::from_utf8_lossy(&bytes).into_owned())
121	}
122
123	fn is_readme(name: &str) -> bool {
124	matches!(
125	name.to_lowercase().as_str(),
126	"readme" \| "readme.md" \| "readme.markdown" \| "readme.txt"
127	)
128	}
129
130	/// Whether `ancestor` is reachable from the newest commit in `history`.
131	///
132	/// `history` is the first-parent chain, which is all the store lists; a fork
133	/// that merged the target branch in has the target's head on a second
134	/// parent, so the walk follows every parent.
135	async fn descends_from<R: GitRepo>(repo: &R, history: &[Commit], ancestor: &str) -> Result<bool> {
136	let known: HashMap<&str, &[String]> = history
137	.iter()
138	.map(\|commit\| (commit.hash.as_str(), commit.parents.as_slice()))
139	.collect();
140	let mut seen = HashSet::new();
141	let mut queue: Vec<String> = history
142	.first()
143	.map(\|c\| c.hash.clone())
144	.into_iter()
145	.collect();
146	while let Some(hash) = queue.pop() {
147	if hash == ancestor {
148	return Ok(true);
149	}
150	if !seen.insert(hash.clone()) \|\| seen.len() > MAX_ANCESTRY as usize {
151	continue;
152	}
153	match known.get(hash.as_str()) {
154	Some(parents) => queue.extend(parents.iter().cloned()),
155	None => queue.extend(repo.parents(&hash).await?.unwrap_or_default()),
156	}
157	}
158	Ok(false)
159	}
160
161	/// The commit closest to the newest in `history` that is also in `shared`:
162	/// where a fork and the repository it came from last agreed.
163	async fn nearest_ancestor_in<R: GitRepo>(
164	repo: &R,
165	history: &[Commit],
166	shared: &HashSet<String>,
167	) -> Result<Option<String>> {
168	let known: HashMap<&str, &[String]> = history
169	.iter()
170	.map(\|commit\| (commit.hash.as_str(), commit.parents.as_slice()))
171	.collect();
172	let mut seen = HashSet::new();
173	let mut queue: VecDeque<String> = history
174	.first()
175	.map(\|c\| c.hash.clone())
176	.into_iter()
177	.collect();
178	while let Some(hash) = queue.pop_front() {
179	if shared.contains(&hash) {
180	return Ok(Some(hash));
181	}
182	if !seen.insert(hash.clone()) \|\| seen.len() > MAX_ANCESTRY as usize {
183	continue;
184	}
185	match known.get(hash.as_str()) {
186	Some(parents) => queue.extend(parents.iter().cloned()),
187	None => queue.extend(repo.parents(&hash).await?.unwrap_or_default()),
188	}
189	}
190	Ok(None)
191	}
192
193	thread_local! {
194	/// Targets' sides of mergeability, by head (coalesce.rs).
195	static TARGETS: std::cell::RefCell<coalesce::Memo<coalesce::TargetKey, Rc<coalesce::TargetSide>>> =
196	std::cell::RefCell::new(coalesce::Memo::new(coalesce::TARGET_TTL_MS, 32));
197	/// What targets changed between two trees.
198	static THEIRS: std::cell::RefCell<coalesce::Memo<coalesce::TheirsKey, (Vec<String>, bool)>> =
199	std::cell::RefCell::new(coalesce::Memo::new(coalesce::THEIRS_TTL_MS, 256));
200	/// What repositories hold, as read for a push's first request, for the
201	/// same push's second: a push's POST does not wait on the database.
202	static HELD: std::cell::RefCell<coalesce::Memo<String, u64>> =
203	std::cell::RefCell::new(coalesce::Memo::new(60_000, 512));
204	}
205
206	pub(crate) struct Repos<S: GitStore> {
207	registry: Registry,
208	store: S,
209	events: Fetcher,
210	/// Asked during a push which secrets have been allowed.
211	security: Option<Fetcher>,
212	/// Asked whether a workspace is on a plan, for its private storage.
213	billing: Option<Fetcher>,
214	/// Says which rulesets hold for a change to a branch or tag, and keeps
215	/// how they judged it (rules.rs). `None` where it is not deployed: the
216	/// old protection flag then holds on push.
217	work: Option<Fetcher>,
218	/// Told when a repository moves, for the tokens of agents at work on it.
219	identity: Option<Fetcher>,
220	/// What a free workspace's private repositories may hold.
221	free_private_bytes: i64,
222	/// Days a pull request's working copy is kept after it settles (forks.rs).
223	pub(crate) fork_days: u64,
224	/// The most a repository may hold (pack_limits.rs), and what happens
225	/// to a push too large to scan.
226	repo_limit: u64,
227	large_pushes: git_http::LargePushes,
228	/// Which git store namespace new repositories go in (shards.rs), and
229	/// the most each should hold (`GITSTORE_NAMESPACE_LIMITS`).
230	placement: shards::Placement,
231	limits: HashMap<String, u64>,
232	/// What isolates share: answers that list refs (refs_cache.rs).
233	shared: Option<Rc<shared::Shared>>,
234	/// Packs for fresh clones (pack_cache.rs); `None` without the bucket.
235	packs: Option<Rc<pack_cache::Packs>>,
236	/// What this request started that need not hold up its answer
237	/// (store.rs), handed to its `waitUntil` once it has answered.
238	deferred: Rc<store::Deferred>,
239	}
240
241	impl<S: GitStore> Repos<S> {
242	/// Records that the refs of the repository with this id changed, once
243	/// they have, so that the answers kept that list them go stale (see
244	/// refs_cache.rs). Everything that changes a repository's refs calls
245	/// this after it (`every_ref_writer_records_the_change` checks). A
246	/// failure is logged: the change itself happened, and what was kept
247	/// expires within `refs_cache::TTL_SECONDS` regardless.
248	pub(crate) async fn refs_moved(&self, repo_id: &str) {
249	if let Err(error) = self.registry.refs_moved(repo_id).await {
250	worker::console_error!("refs of {repo_id} changed but not recorded: {error}");
251	}
252	}
253
254	pub(crate) async fn publish<T: Serialize>(&self, event: NewEvent<T>) -> Result<()> {
255	g1t_kit::call(
256	&self.events,
257	"publish",
258	&Publish {
259	events: vec![event],
260	},
261	)
262	.await
263	}
264
265	/// Whether the viewer may read `repo`. A pull request's fork of a
266	/// private repository can be read by everyone who can read that
267	/// repository, so its members can review and check out the change, as
268	/// well as by whoever opened the pull request.
269	async fn may_read(&self, repo: &Repo, viewer: &Viewer) -> Result<bool> {
270	if can_read(repo, viewer) {
271	return Ok(true);
272	}
273	let Some(source_id) = &repo.fork_of else {
274	return Ok(false);
275	};
276	Ok(self
277	.registry
278	.by_id(source_id)
279	.await?
280	.is_some_and(\|source\| can_read(&source, viewer)))
281	}
282
283	/// `repo`, if there is one and the viewer may read it.
284	async fn visible(&self, repo: Option<Repo>, viewer: &Viewer) -> Result<Option<Repo>> {
285	Ok(match repo {
286	Some(repo) if self.may_read(&repo, viewer).await? => Some(repo),
287	_ => None,
288	})
289	}
290
291	/// Resolves a repo the viewer may read; private repos look missing.
292	pub(crate) async fn readable(&self, path: &RepoPath, viewer: &Viewer) -> Result<Option<Repo>> {
293	self.visible(self.registry.by_path(path).await?, viewer)
294	.await
295	}
296
297	async fn get(&self, a: GetArgs) -> Result<Outcome<Repo>> {
298	Ok(self
299	.readable(&a.path, &a.viewer)
300	.await?
301	.map_or_else(not_found, Outcome::Ok))
302	}
303
304	async fn get_by_id(&self, a: GetByIdArgs) -> Result<Outcome<Repo>> {
305	Ok(self
306	.visible(self.registry.by_id(&a.id).await?, &a.viewer)
307	.await?
308	.map_or_else(not_found, Outcome::Ok))
309	}
310
311	async fn update(&self, a: UpdateArgs) -> Result<Outcome<Repo>> {
312	let viewer = Some(a.actor.clone());
313	let Some(repo) = self.readable(&a.path, &viewer).await? else {
314	return Ok(not_found());
315	};
316	// Its details take Maintain; its protection, Admin; who can see it,
317	// Admin and the member privileges (below). See g1t_contracts::access.
318	let protection_changes = a.protected.is_some_and(\|protected\| protected != repo.protected);
319	let details_change = a.description.is_some() \|\| a.website.is_some() \|\| a.topics.is_some();
320	let mut needed = Vec::new();
321	if details_change \|\| !protection_changes {
322	needed.push(Capability::ManageSettings);
323	}
324	if protection_changes {
325	needed.push(Capability::ManageProtection);
326	}
327	let full_name = format!("{}/{}", repo.namespace, repo.name);
328	if repo.fork_of.is_some() {
329	return Ok(Outcome::fail(FailureCode::Forbidden, access::needs(Capability::ManageSettings, &full_name)));
330	}
331	if let Some(missing) = needed.into_iter().find(\|capability\| !registry::can(&repo, &viewer, *capability)) {
332	return Ok(Outcome::fail(FailureCode::Forbidden, access::needs(missing, &full_name)));
333	}
334	if !a.actor.verified {
335	return Ok(Outcome::fail(FailureCode::Forbidden, UNVERIFIED));
336	}
337	if let Some((code, message)) = lifecycle::archived_refusal(&repo) {
338	return Ok(Outcome::fail(code, message));
339	}
340	let description = match a.description {
341	Some(text) => Some(
342	text.trim()
343	.chars()
344	.take(MAX_DESCRIPTION_CHARS)
345	.collect::<String>(),
346	)
347	.filter(\|text\| !text.is_empty()),
348	None => repo.description.clone(),
349	};
350	let website = match a.website.as_deref() {
351	Some(text) => match clean_website(text) {
352	Ok(website) => website,
353	Err(reason) => return Ok(Outcome::fail(FailureCode::Invalid, reason)),
354	},
355	None => repo.website.clone(),
356	};
357	// Who can see it is an owner's to change, and a free workspace's
358	// storage may not take it private: see lifecycle.rs.
359	let wants_private = a.is_private.filter(\|private\| *private != repo.is_private);
360	if wants_private.is_some()
361	&& let Err((code, message)) = lifecycle::admin_only(
362	lifecycle::Asker::on(&a.actor, &repo),
363	&repo.namespace,
364	"change the visibility of",
365	Capability::ChangeVisibility,
366	)
367	{
368	return Ok(Outcome::fail(code, message));
369	}
370	if let Some(private) = wants_private
371	&& let Some(why) = lifecycle::visibility_refusal(&a.actor, &repo, private)
372	{
373	return Ok(Outcome::fail(FailureCode::Forbidden, why));
374	}
375	let is_private = repo.is_private;
376	let protected = a.protected.unwrap_or(repo.protected);
377	let topics = match &a.topics {
378	Some(topics) => match clean_topics(topics) {
379	Ok(topics) => topics,
380	Err(reason) => return Ok(Outcome::fail(FailureCode::Invalid, reason)),
381	},
382	None => repo.topics.clone(),
383	};
384	self.registry
385	.update(&repo.id, description.as_deref(), protected, &topics, website.as_deref())
386	.await?;
387	let updated = Repo {
388	description,
389	is_private,
390	protected,
391	topics,
392	website,
393	..repo
394	};
395	// Whether the default branch takes only pull requests is now its
396	// branch protection ruleset's to say (work's rulesets.rs).
397	if let (Some(protected), Some(work)) = (a.protected, &self.work) {
398	#[derive(Serialize)]
399	struct RequirePullRequest<'a> {
400	repo: &'a Repo,
401	protected: bool,
402	actor: &'a User,
403	}
404	let set: Result<Outcome<bool>> =
405	g1t_kit::call(work, "set_requires_pull_request", &RequirePullRequest { repo: &updated, protected, actor: &a.actor }).await;
406	match set {
407	Ok(Outcome::Ok(_)) => {}
408	Ok(Outcome::Fail(failure)) => return Ok(Outcome::Fail(failure)),
409	Err(error) => return Err(error),
410	}
411	}
412	if let Some(private) = wants_private {
413	return self.change_visibility(updated, private, &a.actor, a.surface).await;
414	}
415	let visibility_changed = false;
416	// Search and anything else that shows the repository hears of it;
417	// a change of visibility is announced on its own as well, so that
418	// what was public stops being shown at once.
419	self.publish(NewEvent {
420	kind: "repo.updated",
421	source: SOURCE,
422	repo_id: Some(updated.id.clone()),
423	actor: Some(a.actor.id.clone()),
424	data: RepoUpdated {
425	repo_id: updated.id.clone(),
426	namespace: updated.namespace.clone(),
427	name: updated.name.clone(),
428	is_private,
429	visibility_changed,
430	},
431	})
432	.await?;
433	Ok(Outcome::Ok(updated))
434	}
435
436	/// The repository with this id, if it is not a fork, and its store.
437	async fn stored(&self, repo_id: &str) -> Result<Option<S::Repo>> {
438	match self.registry.by_id(repo_id).await? {
439	Some(repo) if repo.fork_of.is_none() => Ok(Some(self.store.open(&store_key(&repo)).await?)),
440	_ => Ok(None),
441	}
442	}
443
444	async fn list_files(&self, a: ListFilesArgs) -> Result<FileList> {
445	let Some(repo) = self.registry.by_id(&a.repo_id).await?.filter(\|repo\| repo.fork_of.is_none()) else {
446	return Ok(FileList::default());
447	};
448	let git = self.store.open(&store_key(&repo)).await?;
449	let head = a.git_ref.unwrap_or_else(\|\| repo.default_branch.clone());
450	listing::list(&git, None, &head, &a.skip_dirs, a.limit, a.links).await
451	}
452
453	/// A link to download a branch, tag or commit as an archive, good for
454	/// a few minutes with no session or token (archive_links.rs).
455	async fn archive_link(&self, a: ArchiveLinkArgs, sealer: Option<&g1t_secrets::Sealer>) -> Result<Outcome<ArchiveLink>> {
456	let Some(repo) = self.readable(&a.path, &a.viewer).await? else {
457	return Ok(not_found());
458	};
459	let Some(sealer) = sealer else {
460	return Ok(Outcome::fail(FailureCode::Invalid, "Archive downloads are not set up on this installation."));
461	};
462	let git = self.read_git(&repo).await?;
463	let git_ref = a.git_ref.filter(\|git_ref\| !git_ref.is_empty()).unwrap_or_else(\|\| repo.default_branch.clone());
464	let Some(head) = git.log(&git_ref, 1).await?.into_iter().next() else {
465	return Ok(Outcome::fail(FailureCode::NotFound, "No such branch, tag or commit."));
466	};
467	let (token, expires) = archive_links::seal(sealer, &repo.id, &head.hash, now_ms());
468	Ok(Outcome::Ok(ArchiveLink { repo, commit: head.hash, token, expires_at: rfc3339(expires) }))
469	}
470
471	/// What an archive link's token lets its holder download.
472	async fn open_archive_token(&self, a: OpenArchiveTokenArgs, sealer: Option<&g1t_secrets::Sealer>) -> Result<Option<ArchiveGrant>> {
473	let Some(sealer) = sealer else {
474	return Ok(None);
475	};
476	let Some(repo) = self.registry.by_path(&a.path).await? else {
477	return Ok(None);
478	};
479	let Some(commit) = archive_links::open(sealer, &repo.id, &a.token, now_ms()) else {
480	return Ok(None);
481	};
482	Ok(Some(ArchiveGrant { repo_id: repo.id, namespace: repo.namespace, name: repo.name, commit }))
483	}
484
485	async fn changed_files(&self, a: ChangedFilesArgs) -> Result<FileList> {
486	let Some(git) = self.stored(&a.repo_id).await? else {
487	return Ok(FileList::default());
488	};
489	listing::list(&git, a.base.as_deref(), &a.head, &a.skip_dirs, a.limit, false).await
490	}
491
492	/// Branches and tags with their commits, for g1t's own services.
493	async fn refs_of(&self, a: RefsArgs) -> Result<Option<RepoRefs>> {
494	let Some(repo) = self.registry.by_id(&a.repo_id).await?.filter(\|repo\| repo.fork_of.is_none()) else {
495	return Ok(None);
496	};
497	let git = self.store.open(&store_key(&repo)).await?;
498	let access = git.access(Scope::Read).await?;
499	let refs = refs::heads_and_tags(refs::all(&access).await?)
500	.into_iter()
501	.map(\|(name, commit)\| GitRefEntry { name, commit })
502	.collect();
503	Ok(Some(RepoRefs { repo, refs }))
504	}
505
506	async fn raw_file(&self, a: RawFileArgs) -> Result<Option<RawFile>> {
507	use base64::Engine;
508	let Some(git) = self.stored(&a.repo_id).await? else {
509	return Ok(None);
510	};
511	Ok(git
512	.read_file(&a.git_ref, &a.path)
513	.await?
514	.filter(\|bytes\| bytes.len() <= a.max_bytes as usize)
515	.map(\|bytes\| RawFile { size: bytes.len() as u64, data: base64::engine::general_purpose::STANDARD.encode(bytes) }))
516	}
517
518	async fn raw_blobs(&self, a: RawBlobsArgs) -> Result<Vec<RawBlob>> {
519	use base64::Engine;
520	let Some(git) = self.stored(&a.repo_id).await? else {
521	return Ok(Vec::new());
522	};
523	let hashes: Vec<&String> = a.hashes.iter().take(MAX_READ_BLOBS).collect();
524	let mut out = Vec::with_capacity(hashes.len());
525	// A few at a time, as listing::read does: each is a round trip.
526	for group in hashes.chunks(8) {
527	let read = futures_util::future::try_join_all(group.iter().map(\|hash\| git.read_blob(hash))).await?;
528	for (hash, bytes) in group.iter().zip(read) {
529	let size = bytes.as_ref().map_or(0, \|bytes\| bytes.len() as u64);
530	let data = bytes
531	.filter(\|bytes\| bytes.len() <= a.max_bytes as usize)
532	.map(\|bytes\| base64::engine::general_purpose::STANDARD.encode(bytes));
533	out.push(RawBlob { hash: (*hash).clone(), size, data });
534	}
535	}
536	Ok(out)
537	}
538
539	async fn read_blobs(&self, a: ReadBlobsArgs) -> Result<Vec<BlobText>> {
540	let Some(git) = self.stored(&a.repo_id).await? else {
541	return Ok(Vec::new());
542	};
543	listing::read(&git, &a.hashes, a.max_bytes.min(MAX_TEXT_BYTES as u32)).await
544	}
545
546	async fn create(&self, a: CreateArgs) -> Result<Outcome<Repo>> {
547	if !a.owner.verified {
548	return Ok(Outcome::fail(FailureCode::Forbidden, UNVERIFIED));
549	}
550	let name = a.name.trim().to_lowercase();
551	if !is_valid_repo_name(&name) {
552	return Ok(Outcome::fail(
553	FailureCode::Invalid,
554	"Use letters, digits, dots, hyphens and underscores only.",
555	));
556	}
557	let namespace = a.namespace.trim().to_lowercase();
558	if namespace.is_empty() {
559	return Ok(Outcome::fail(
560	FailureCode::Invalid,
561	"Say which workspace to create the repository in.",
562	));
563	}
564	let Some(role) = a.owner.role_in(&namespace) else {
565	return Ok(Outcome::fail(
566	FailureCode::Forbidden,
567	"You are not a member of that workspace.",
568	));
569	};
570	// Who may create which: the workspace's member privileges. A
571	// workspace's own token acts as an owner would.
572	let role = if a.owner.kind == PrincipalKind::Workspace { Role::Owner } else { role };
573	if let Some(why) = a.owner.privileges_in(&namespace).creation_refusal(role, a.is_private, &namespace) {
574	return Ok(Outcome::fail(FailureCode::Forbidden, why));
575	}
576	let path = RepoPath { namespace, name };
577	match self.registry.by_path_any(&path).await? {
578	Some((_, None)) => {
579	return Ok(Outcome::fail(
580	FailureCode::Conflict,
581	"That workspace already has a repository with that name.",
582	));
583	}
584	Some((_, Some(_))) => {
585	return Ok(Outcome::fail(
586	FailureCode::Conflict,
587	format!(
588	"{}/{} was deleted recently and can still be restored, so its name is taken. Restore it, or delete it permanently from the workspace's Recently deleted list.",
589	path.namespace, path.name
590	),
591	));
592	}
593	None => {}
594	}
595	// With a credential (a GitHub App installation's token), everything
596	// is copied: every branch and tag. See copy.rs.
597	let mut credentialed = None;
598	if let (Some(url), Some(token)) = (a.import_url.as_deref(), a.import_token.as_deref()) {
599	let Some(url) = import::clean_url(url) else {
600	return Ok(Outcome::fail(FailureCode::Invalid, "That is not an https repository address."));
601	};
602	let source = copy::Endpoint::github(&url, token);
603	match copy::probe(&source).await? {
604	Ok(advertised) => credentialed = Some((source, advertised)),
605	Err(reason) => return Ok(Outcome::fail(FailureCode::Invalid, reason)),
606	}
607	}
608	// An import is fetched before anything is created, so that an
609	// address that does not work leaves nothing behind.
610	let mut imported = None;
611	if let Some(url) = a
612	.import_url
613	.as_deref()
614	.map(str::trim)
615	.filter(\|url\| !url.is_empty() && credentialed.is_none())
616	{
617	let Some(url) = import::clean_url(url) else {
618	return Ok(Outcome::fail(
619	FailureCode::Invalid,
620	"Give the https address of a public repository, such as https://github.com/owner/repo.",
621	));
622	};
623	let remote = match import::discover(&url).await? {
624	Ok(remote) => remote,
625	Err(reason) => return Ok(Outcome::fail(FailureCode::Invalid, reason)),
626	};
627	imported = Some((remote, url));
628	}
629	let now = now_ms();
630	let repo = Repo {
631	id: new_id("rep", now),
632	namespace: path.namespace,
633	name: path.name,
634	description: a
635	.description
636	.map(\|text\| text.trim().to_owned())
637	.filter(\|text\| !text.is_empty()),
638	is_private: a.is_private,
639	owner_id: a.owner.id.clone(),
640	default_branch: imported
641	.as_ref()
642	.map(\|(remote, _)\| remote.branch.clone())
643	.or_else(\|\| credentialed.as_ref().and_then(\|(_, advertised)\| advertised.default_branch()))
644	.unwrap_or_else(\|\| "main".to_owned()),
645	fork_of: None,
646	protected: false,
647	created_at: rfc3339(now),
648	topics: Vec::new(),
649	website: None,
650	archived_at: None,
651	};
652	let namespace = match self.place(&repo).await? {
653	Ok(namespace) => namespace,
654	Err(unplaced) => return Ok(Outcome::fail(FailureCode::Conflict, unplaced.message())),
655	};
656	self.registry
657	.claim_store_key(&repo, namespace.as_deref(), &self.store.default_namespace())
658	.await?;
659	self.store
660	.create(
661	&store_key(&repo),
662	repo.description.as_deref(),
663	&repo.default_branch,
664	)
665	.await?;
666	self.registry.insert(&repo).await?;
667	// A repository that was transferred away from this path stops
668	// redirecting here.
669	self.registry
670	.drop_redirect(&RepoPath {
671	namespace: repo.namespace.clone(),
672	name: repo.name.clone(),
673	})
674	.await?;
675	// Every branch and tag the import made, announced as pushes.
676	let mut pushed: Vec<(String, String)> = Vec::new();
677	// A public repository, read with no credential: every branch and
678	// tag is copied too, the default branch the one its HEAD names.
679	if let Some((_, url)) = imported {
680	let access = self
681	.store
682	.open(&store_key(&repo))
683	.await?
684	.access(Scope::Write)
685	.await?;
686	let target = copy::Endpoint::bearer(&access.remote, &access.token);
687	let copied = copy::copy(&copy::Endpoint::anonymous(&url), &target).await?;
688	self.refs_moved(&repo.id).await;
689	match copied {
690	Ok(copied) => pushed = copy::import_pushes(&copied.updated, &repo.default_branch),
691	Err(reason) => {
692	self.registry.remove(&repo.id).await?;
693	return Ok(Outcome::fail(
694	FailureCode::Invalid,
695	format!("The repository could not be stored: {reason}"),
696	));
697	}
698	}
699	}
700	if let Some((source, _)) = credentialed {
701	let access = self
702	.store
703	.open(&store_key(&repo))
704	.await?
705	.access(Scope::Write)
706	.await?;
707	let target = copy::Endpoint::bearer(&access.remote, &access.token);
708	let copied = copy::copy(&source, &target).await?;
709	self.refs_moved(&repo.id).await;
710	match copied {
711	Ok(copied) => pushed = copy::import_pushes(&copied.updated, &repo.default_branch),
712	Err(reason) => {
713	self.registry.remove(&repo.id).await?;
714	return Ok(Outcome::fail(
715	FailureCode::Invalid,
716	format!("The repository could not be copied: {reason}"),
717	));
718	}
719	}
720	}
721	// Whoever creates a repository is an Admin of it, as a role given
722	// to them on it, whatever the workspace's base permission.
723	if a.owner.kind == PrincipalKind::User
724	&& let Some(identity) = &self.identity
725	{
726	let granted: Result<bool> = g1t_kit::call(
727	identity,
728	"grant_creator",
729	&g1t_contracts::members::GrantCreatorArgs {
730	repo_id: repo.id.clone(),
731	namespace: repo.namespace.clone(),
732	name: repo.name.clone(),
733	user_id: a.owner.id.clone(),
734	},
735	)
736	.await;
737	if let Err(error) = granted {
738	worker::console_error!("creator of {} not given Admin: {error}", repo.id);
739	}
740	}
741	self.publish(NewEvent {
742	kind: "repo.created",
743	source: SOURCE,
744	repo_id: Some(repo.id.clone()),
745	actor: Some(a.owner.id),
746	data: RepoCreated {
747	repo_id: repo.id.clone(),
748	namespace: repo.namespace.clone(),
749	name: repo.name.clone(),
750	is_private: repo.is_private,
751	},
752	})
753	.await?;
754	for (git_ref, head) in &pushed {
755	self.publish_push(&repo, git_ref, None, head, None, None).await?;
756	}
757	Ok(Outcome::Ok(repo))
758	}
759
760	/// Where a workspace keeps its data, asked of identity only when an EU
761	/// namespace is configured: without one, every workspace's
762	/// repositories go anywhere and identity is never asked.
763	async fn residency_of(&self, workspace: &str) -> Result<shards::Residency> {
764	if self.placement.eu.is_none() {
765	return Ok(shards::Residency::Anywhere);
766	}
767	let Some(identity) = &self.identity else {
768	return Ok(shards::Residency::Anywhere);
769	};
770	let residency: Option<g1t_contracts::identity::DataResidency> = g1t_kit::call(
771	identity,
772	"workspace_residency",
773	&g1t_contracts::identity::SlugArgs { slug: workspace.to_owned() },
774	)
775	.await?;
776	Ok(match residency {
777	Some(g1t_contracts::identity::DataResidency::Eu) => shards::Residency::Eu,
778	_ => shards::Residency::Anywhere,
779	})
780	}
781
782	/// How each bound namespace stands (namespaces.rs).
783	async fn standings(&self) -> Result<Vec<namespaces::Standing>> {
784	let bound = self.store.namespaces();
785	let default = self.store.default_namespace();
786	let now = now_ms();
787	let config = namespaces::Configured {
788	bound: &bound,
789	default: &default,
790	placement: &self.placement,
791	limits: &self.limits,
792	on_fallback: &\|namespace\| self.store.on_fallback(&shards::compose(Some(namespace), "x", &default)),
793	writable: &\|namespace\| self.store.writable(namespace),
794	breaker_open: &\|namespace\| resilience::open_now(namespace, now),
795	};
796	let (held, recent) = futures_util::future::join(namespaces::held(&self.registry.db), namespaces::recent(&self.registry.db, now)).await;
797	Ok(namespaces::standings(&config, &held?, &recent?))
798	}
799
800	/// The namespace a new repository goes in (shards.rs): its workspace's
801	/// residency, then how each namespace stands, read only when there is
802	/// a choice to make. `Ok(None)` for the default.
803	async fn place(&self, repo: &Repo) -> Result<std::result::Result<Option<String>, shards::Unplaced>> {
804	let residency = self.residency_of(&repo.namespace).await?;
805	let bound = self.store.namespaces();
806	let loads = if residency == shards::Residency::Anywhere && !self.placement.needs_loads(&bound) {
807	// One namespace to choose from at most: nothing to read.
808	bound
809	.iter()
810	.map(\|namespace\| shards::Load {
811	namespace: namespace.clone(),
812	bound: true,
813	writable: self.store.writable(namespace),
814	..shards::Load::default()
815	})
816	.collect()
817	} else {
818	let default = self.store.default_namespace();
819	let now = now_ms();
820	let config = namespaces::Configured {
821	bound: &bound,
822	default: &default,
823	placement: &self.placement,
824	limits: &self.limits,
825	on_fallback: &\|namespace\| self.store.on_fallback(&shards::compose(Some(namespace), "x", &default)),
826	writable: &\|namespace\| self.store.writable(namespace),
827	breaker_open: &\|namespace\| resilience::open_now(namespace, now),
828	};
829	namespaces::loads(&self.registry.db, &config, now).await?
830	};
831	Ok(self.placement.choose(&repo.id, residency, &loads))
832	}
833
834	/// `storage_options`: what a workspace may choose about where its
835	/// repositories are kept.
836	fn storage_options(&self) -> StorageOptions {
837	let bound = self.store.namespaces();
838	StorageOptions { eu_available: self.placement.eu_available(&bound, \|namespace\| self.store.writable(namespace)) }
839	}
840
841	async fn tree(&self, a: TreeArgs) -> Result<Outcome<TreeView>> {
842	let Some(repo) = self.readable(&a.path, &a.viewer).await? else {
843	return Ok(not_found());
844	};
845	let git = self.read_git(&repo).await?;
846	let git_ref = a
847	.git_ref
848	.clone()
849	.unwrap_or_else(\|\| repo.default_branch.clone());
850
851	let Some(head) = git.log(&git_ref, 1).await?.into_iter().next() else {
852	// An unknown ref is an error; a repo with no commits is just empty.
853	if a.git_ref.is_some() {
854	return Ok(Outcome::fail(
855	FailureCode::NotFound,
856	"No such branch, tag or commit.",
857	));
858	}
859	return Ok(Outcome::Ok(TreeView {
860	repo,
861	git_ref,
862	path: a.tree_path,
863	head: None,
864	entries: Vec::new(),
865	readme: None,
866	}));
867	};
868
869	let no_directory = \|\| Outcome::fail(FailureCode::NotFound, "No such directory.");
870	let mut entries = git.read_tree(&head.tree_hash).await?;
871	for segment in a.tree_path.split('/').filter(\|segment\| !segment.is_empty()) {
872	let next = entries.as_ref().and_then(\|entries\| {
873	entries
874	.iter()
875	.find(\|entry\| entry.name == segment && entry.kind == EntryKind::Tree)
876	});
877	let Some(next) = next else {
878	return Ok(no_directory());
879	};
880	entries = git.read_tree(&next.hash).await?;
881	}
882	let Some(mut entries) = entries else {
883	return Ok(no_directory());
884	};
885	// Directories first, then by name.
886	entries.sort_by(\|a, b\| {
887	(b.kind == EntryKind::Tree)
888	.cmp(&(a.kind == EntryKind::Tree))
889	.then_with(\|\| a.name.cmp(&b.name))
890	});
891
892	let readme_entry = entries
893	.iter()
894	.find(\|entry\| entry.kind == EntryKind::Blob && is_readme(&entry.name));
895	let readme = match readme_entry {
896	Some(entry) => git.read_blob(&entry.hash).await?.map(\|bytes\| Readme {
897	name: entry.name.clone(),
898	text: text_of(bytes),
899	}),
900	None => None,
901	};
902	Ok(Outcome::Ok(TreeView {
903	repo,
904	git_ref,
905	path: a.tree_path,
906	head: Some(head),
907	entries,
908	readme,
909	}))
910	}
911
912	async fn blob(&self, a: BlobArgs) -> Result<Outcome<BlobView>> {
913	let Some(repo) = self.readable(&a.path, &a.viewer).await? else {
914	return Ok(not_found());
915	};
916	let bytes = if a.file_path.is_empty() {
917	None
918	} else {
919	let git = self.read_git(&repo).await?;
920	git.read_file(&a.git_ref, &a.file_path).await?
921	};
922	let Some(bytes) = bytes else {
923	return Ok(Outcome::fail(FailureCode::NotFound, "No such file."));
924	};
925	Ok(Outcome::Ok(BlobView {
926	repo,
927	git_ref: a.git_ref,
928	path: a.file_path,
929	size: bytes.len() as u64,
930	text: text_of(bytes),
931	}))
932	}
933
934	async fn blame(&self, a: BlameArgs) -> Result<Outcome<Blame>> {
935	let Some(repo) = self.readable(&a.path, &a.viewer).await? else {
936	return Ok(not_found());
937	};
938	let git = self.read_git(&repo).await?;
939	let git_ref = a.git_ref.unwrap_or_else(\|\| repo.default_branch.clone());
940	Ok(match blame::blame(&git, &git_ref, &a.file_path).await? {
941	Some(blame) => Outcome::Ok(blame),
942	None => not_found(),
943	})
944	}
945
946	async fn log(&self, a: LogArgs) -> Result<Outcome<Vec<Commit>>> {
947	let Some(repo) = self.readable(&a.path, &a.viewer).await? else {
948	return Ok(not_found());
949	};
950	let git = self.read_git(&repo).await?;
951	let git_ref = a.git_ref.unwrap_or_else(\|\| repo.default_branch.clone());
952	Ok(Outcome::Ok(git.log(&git_ref, a.limit).await?))
953	}
954
955	/// The commits that changed a file or directory, newest first: the
956	/// first-parent chain walked within `PATH_LOG_WALK_MS` (or the caller's
957	/// budget) and `last_commits::MAX_READS`.
958	async fn path_log(&self, a: g1t_contracts::repos::PathLogArgs) -> Result<Outcome<g1t_contracts::repos::PathLog>> {
959	let Some(repo) = self.readable(&a.path, &a.viewer).await? else {
960	return Ok(not_found());
961	};
962	let git = self.read_git(&repo).await?;
963	let git_ref = a.git_ref.unwrap_or_else(\|\| repo.default_branch.clone());
964	let Some(head) = git.log(&git_ref, 1).await?.into_iter().next() else {
965	return Ok(Outcome::fail(FailureCode::NotFound, "No such branch, tag or commit."));
966	};
967	let started = worker::Date::now().as_millis();
968	let budget = a.budget_ms.unwrap_or(PATH_LOG_WALK_MS);
969	let out_of_time = move \|\| worker::Date::now().as_millis().saturating_sub(started) > budget;
970	let limit = a.limit.clamp(1, 100) as usize;
971	let (commits, complete) = last_commits::path_log(&git, &head.hash, &a.file_path, limit, &out_of_time, last_commits::MAX_READS).await?;
972	Ok(Outcome::Ok(g1t_contracts::repos::PathLog { commits, complete }))
973	}
974
975	/// Which commit last changed each entry of a directory. A finished
976	/// answer is kept in this colo's cache by repository, head commit and
977	/// path: a commit's history never changes, so it is good for as long as
978	/// it is kept. Every walk also keeps its progress by ref and path
979	/// (last_commits.rs), so the next call goes on from it: from where it
980	/// stopped, or for a new head, only back to the old one.
981	async fn last_commits(&self, a: g1t_contracts::repos::LastCommitsArgs) -> Result<Outcome<g1t_contracts::repos::LastCommits>> {
982	let Some(repo) = self.readable(&a.path, &a.viewer).await? else {
983	return Ok(not_found());
984	};
985	let git = self.read_git(&repo).await?;
986	let git_ref = a.git_ref.unwrap_or_else(\|\| repo.default_branch.clone());
987	let Some(head) = git.log(&git_ref, 1).await?.into_iter().next() else {
988	return Ok(Outcome::fail(FailureCode::NotFound, "No such branch, tag or commit."));
989	};
990	// v2: v1 kept walks cut short at 300 commits as if finished.
991	let key = format!(
992	"https://last-commits.g1t.internal/v2/{}/{}/{}",
993	repo.id,
994	head.hash,
995	a.tree_path.split('/').map(urlencoding_segment).collect::<Vec<_>>().join("/")
996	);
997	let cache = worker::Cache::default();
998	if let Ok(Some(mut kept)) = cache.get(key.as_str(), false).await {
999	if let Ok(found) = kept.json::<g1t_contracts::repos::LastCommits>().await {
1000	return Ok(Outcome::Ok(found));
1001	}
1002	}
1003	let memo = last_commits::Memo::new(&repo.id, &git_ref, &a.tree_path);
1004	let shared = self.shared.as_deref();
1005	let progress = memo.get(shared).await;
1006	// Asked with a budget, the walk stops past it; without one, past
1007	// LAST_COMMITS_WALK_MS, well before the caller's waitUntil ends.
1008	let started = worker::Date::now().as_millis();
1009	let budget = a.budget_ms.unwrap_or(LAST_COMMITS_WALK_MS);
1010	let out_of_time = move \|\| worker::Date::now().as_millis().saturating_sub(started) > budget;
1011	let walk = last_commits::last_commits(&git, &head.hash, &a.tree_path, progress, &out_of_time, last_commits::MAX_READS).await?;
1012	if walk.reads > 0 {
1013	memo.keep(shared, &walk.progress).await;
1014	}
1015	let settled = walk.progress.settled();
1016	let found = g1t_contracts::repos::LastCommits { complete: walk.progress.complete(), entries: walk.progress.found };
1017	if !settled {
1018	return Ok(Outcome::Ok(found));
1019	}
1020	if let Ok(mut response) = worker::Response::from_json(&found) {
1021	let _ = response.headers_mut().set("cache-control", "max-age=604800");
1022	let _ = cache.put(key.as_str(), response).await;
1023	}
1024	Ok(Outcome::Ok(found))
1025	}
1026
1027	/// How far each branch head has moved from the default branch's head,
1028	/// in one call (drift.rs). Each count is kept in this colo's cache by
1029	/// repository and the pair of hashes, for good: neither history can
1030	/// change. A head that moved is the only one walked. A failed read is
1031	/// not kept, and "too far to count" only for a day.
1032	async fn branch_drift(&self, a: BranchDriftArgs) -> Result<Outcome<BranchDrifts>> {
1033	let Some(repo) = self.readable(&a.path, &a.viewer).await? else {
1034	return Ok(not_found());
1035	};
1036	if !store::is_commit_hash(&a.base) {
1037	return Ok(Outcome::fail(FailureCode::Invalid, "The default branch's head is a full commit hash."));
1038	}
1039	let heads: Vec<String> = a.heads.into_iter().take(MAX_DRIFT_HEADS).collect();
1040	let git = self.read_git(&repo).await?;
1041	// v2: answers kept before 2026-10-09 said "no count" for every
1042	// branch whose default branch took a merge since it left (drift.rs).
1043	let key = \|head: &str\| format!("https://drift.g1t.internal/v2/{}/{}/{head}", repo.id, a.base);
1044	let cache = worker::Cache::default();
1045	let (base, kept) = futures_util::future::join(
1046	git.log(&a.base, 1),
1047	futures_util::future::join_all(heads.iter().map(\|head\| {
1048	let (cache, url) = (&cache, key(head));
1049	async move {
1050	if !store::is_commit_hash(head) {
1051	return None;
1052	}
1053	let mut found = cache.get(url.as_str(), false).await.ok()??;
1054	found.json::<BranchDrift>().await.ok()
1055	}
1056	})),
1057	)
1058	.await;
1059	let missing: Vec<String> = heads
1060	.iter()
1061	.zip(&kept)
1062	.filter(\|(head, kept)\| kept.is_none() && store::is_commit_hash(head))
1063	.map(\|(head, _)\| head.clone())
1064	.collect();
1065	let measured = drift::measure(&git, &a.base, &missing).await;
1066	let mut fresh: HashMap<String, BranchDrift> = HashMap::new();
1067	for (head, found) in missing.into_iter().zip(measured) {
1068	let answer = BranchDrift { head: head.clone(), commit: found.commit, drift: found.drift };
1069	if found.settled
1070	&& let Ok(mut response) = worker::Response::from_json(&answer)
1071	{
1072	// A count is kept for good; "too far to count" for a day, so
1073	// a change to how far a walk goes reaches it.
1074	let max_age = if answer.drift.is_some() { "public, max-age=31536000, immutable" } else { "public, max-age=86400" };
1075	let _ = response.headers_mut().set("cache-control", max_age);
1076	let _ = cache.put(key(&head).as_str(), response).await;
1077	}
1078	fresh.insert(head, answer);
1079	}
1080	let branches = heads
1081	.iter()
1082	.zip(kept)
1083	.map(\|(head, kept)\| {
1084	kept.or_else(\|\| fresh.get(head).cloned())
1085	.unwrap_or_else(\|\| BranchDrift { head: head.clone(), commit: None, drift: None })
1086	})
1087	.collect();
1088	Ok(Outcome::Ok(BranchDrifts { base: base.ok().and_then(\|log\| log.into_iter().next()), branches }))
1089	}
1090
1091	/// The repository's tags, newest commit first, at most 100.
1092	async fn tags(&self, a: g1t_contracts::repos::TagsArgs) -> Result<Outcome<Vec<g1t_contracts::repos::Tag>>> {
1093	let Some(repo) = self.readable(&a.path, &a.viewer).await? else {
1094	return Ok(not_found());
1095	};
1096	// Kept until the refs move, as the branch list is (store.rs): listing
1097	// the refs is a round trip to the store on every call otherwise.
1098	let version = refs_cache::usable(registry::refs_state(&repo.id), now_ms()).filter(\|_\| !self.store.on_fallback(&store_key(&repo)));
1099	let kept_at = version.map(\|version\| format!("https://tags.g1t.internal/{}/{version}", repo.id));
1100	if let Some(url) = &kept_at
1101	&& let Ok(Some(mut kept)) = worker::Cache::default().get(url.as_str(), false).await
1102	&& let Ok(tags) = kept.json::<Vec<g1t_contracts::repos::Tag>>().await
1103	{
1104	return Ok(Outcome::Ok(tags));
1105	}
1106	let (tags, complete) = self.read_tags(&repo).await?;
1107	if complete
1108	&& let Some(url) = &kept_at
1109	&& let Ok(mut response) = worker::Response::from_json(&tags)
1110	{
1111	let _ = response.headers_mut().set("cache-control", "public, max-age=300");
1112	let _ = worker::Cache::default().put(url.as_str(), response).await;
1113	}
1114	Ok(Outcome::Ok(tags))
1115	}
1116
1117	/// The tags, and whether every one's commit was read (only then kept).
1118	async fn read_tags(&self, repo: &Repo) -> Result<(Vec<g1t_contracts::repos::Tag>, bool)> {
1119	let git = self.store.open(&store_key(repo)).await?;
1120	let access = git.access(Scope::Read).await?;
1121	let named: Vec<(String, String)> = refs::heads_and_tags(refs::all(&access).await?)
1122	.into_iter()
1123	.filter_map(\|(name, hash)\| name.strip_prefix("refs/tags/").map(\|tag\| (tag.to_owned(), hash)))
1124	.collect();
1125	let read = self.read_git(repo).await?;
1126	let commits = futures_util::future::join_all(named.iter().take(MAX_TAGS_READ).map(\|(_, hash)\| read.log(hash, 1))).await;
1127	let complete = commits.iter().all(Result::is_ok);
1128	let mut tags: Vec<g1t_contracts::repos::Tag> = named
1129	.into_iter()
1130	.zip(commits.into_iter().map(\|found\| found.ok().and_then(\|list\| list.into_iter().next())).chain(std::iter::repeat(None)))
1131	.map(\|((name, _), commit)\| g1t_contracts::repos::Tag { name, commit })
1132	.collect();
1133	tags.sort_by(\|a, b\| {
1134	let at = \|tag: &g1t_contracts::repos::Tag\| tag.commit.as_ref().map(\|c\| c.authored_at.clone()).unwrap_or_default();
1135	at(b).cmp(&at(a)).then_with(\|\| b.name.cmp(&a.name))
1136	});
1137	tags.truncate(MAX_TAGS_READ);
1138	Ok((tags, complete))
1139	}
1140
1141	/// The repository's branches, default branch first.
1142	async fn branches(&self, a: BranchesArgs) -> Result<Outcome<Vec<Branch>>> {
1143	let Some(repo) = self.readable(&a.path, &a.viewer).await? else {
1144	return Ok(not_found());
1145	};
1146	let mut branches = self.read_git(&repo).await?.branches().await?;
1147	branches.sort_by_key(\|branch\| branch.name != repo.default_branch);
1148	Ok(Outcome::Ok(branches))
1149	}
1150
1151	/// Whether a pull request's source lacks commits that the branch it
1152	/// would merge into has.
1153	async fn behind(&self, a: BehindArgs) -> Result<bool> {
1154	let Some(source) = self.registry.by_id(&a.source_id).await? else {
1155	return Ok(false);
1156	};
1157	let target = match &source.fork_of {
1158	Some(id) => self.registry.by_id(id).await?,
1159	None => Some(source.clone()),
1160	};
1161	let Some(target) = target else {
1162	return Ok(false);
1163	};
1164	let branch = a.branch.unwrap_or_else(\|\| target.default_branch.clone());
1165	let target_branch = a.target_branch.unwrap_or_else(\|\| target.default_branch.clone());
1166	let target_head = self
1167	.read_git(&target)
1168	.await?
1169	.log(&target_branch, 1)
1170	.await?
1171	.into_iter()
1172	.next()
1173	.map(\|commit\| commit.hash);
1174	let Some(target_head) = target_head else {
1175	return Ok(false);
1176	};
1177	let source_git = self.read_git(&source).await?;
1178	let history = source_git.log(&branch, MAX_ANCESTRY).await?;
1179	if history.is_empty() {
1180	return Ok(false);
1181	}
1182	Ok(!descends_from(&source_git, &history, &target_head).await?)
1183	}
1184
1185	/// The files a pull request's source and the default branch it would
1186	/// merge into each changed since they last agreed. Where the two lists
1187	/// share no file, the merge cannot conflict; where they do, it may.
1188	async fn divergence(&self, a: BehindArgs) -> Result<Option<Divergence>> {
1189	let Some(source) = self.registry.by_id(&a.source_id).await? else {
1190	return Ok(None);
1191	};
1192	let target = match &source.fork_of {
1193	Some(id) => self.registry.by_id(id).await?,
1194	None => Some(source.clone()),
1195	};
1196	let Some(target) = target else {
1197	return Ok(None);
1198	};
1199	let branch = a.branch.unwrap_or_else(\|\| target.default_branch.clone());
1200	let target_branch = a.target_branch.unwrap_or_else(\|\| target.default_branch.clone());
1201	let source_git = self.read_git(&source).await?;
1202	let target_git = self.read_git(&target).await?;
1203	// The target's side is the same for every pull request into it, and
1204	// worked out once per head (coalesce.rs).
1205	let (history, side) = futures_util::future::try_join(
1206	source_git.log(&branch, MAX_ANCESTRY),
1207	self.target_side(&target, &target_git, &target_branch),
1208	)
1209	.await?;
1210	let target_history = &side.history;
1211	let (Some(head), Some(base)) = (history.first(), target_history.first()) else {
1212	return Ok(None);
1213	};
1214	let behind = !descends_from(&source_git, &history, &base.hash).await?;
1215	let merge_base = nearest_ancestor_in(&source_git, &history, &side.shared).await?;
1216	let mut divergence = Divergence {
1217	head: head.hash.clone(),
1218	base: base.hash.clone(),
1219	merge_base: merge_base.clone(),
1220	behind,
1221	..Divergence::default()
1222	};
1223	let merge_base_tree = match &merge_base {
1224	Some(hash) => target_history
1225	.iter()
1226	.find(\|commit\| commit.hash == *hash)
1227	.map(\|commit\| commit.tree_hash.clone()),
1228	None => None,
1229	};
1230	let Some(merge_base_tree) = merge_base_tree else {
1231	// No common history to compare from: say nothing is known.
1232	divergence.truncated = true;
1233	return Ok(Some(divergence));
1234	};
1235	let (ours, truncated_ours) =
1236	diff::changed_paths(&source_git, Some(&merge_base_tree), &head.tree_hash).await?;
1237	divergence.ours = ours;
1238	divergence.truncated = truncated_ours;
1239	if behind {
1240	let now = now_ms();
1241	let key = (target.id.clone(), merge_base_tree.clone(), base.tree_hash.clone());
1242	let (theirs, truncated_theirs) = match THEIRS.with(\|memo\| memo.borrow().get(&key, now)) {
1243	Some(kept) => kept,
1244	None => {
1245	let found = diff::changed_paths(&target_git, Some(&merge_base_tree), &base.tree_hash).await?;
1246	THEIRS.with(\|memo\| memo.borrow_mut().put(key, found.clone(), now));
1247	found
1248	}
1249	};
1250	divergence.theirs = theirs;
1251	divergence.truncated \|= truncated_theirs;
1252	}
1253	Ok(Some(divergence))
1254	}
1255
1256	/// A target branch's history from its head, worked out once per head
1257	/// for every pull request asking about it (coalesce.rs). The head is
1258	/// read under the refs version; the history by its hash, which the
1259	/// object cache keeps for good.
1260	async fn target_side<R: GitRepo>(&self, target: &Repo, git: &R, branch: &str) -> Result<Rc<coalesce::TargetSide>> {
1261	let now = now_ms();
1262	let key = refs_cache::usable(registry::refs_state(&target.id), now)
1263	.map(\|version\| (target.id.clone(), branch.to_owned(), version));
1264	if let Some(key) = &key
1265	&& let Some(side) = TARGETS.with(\|memo\| memo.borrow().get(key, now))
1266	{
1267	return Ok(side);
1268	}
1269	let history = match git.log(branch, 1).await?.first() {
1270	Some(head) => git.log(&head.hash, MAX_ANCESTRY).await?,
1271	None => Vec::new(),
1272	};
1273	let side = coalesce::TargetSide::new(history);
1274	if let Some(key) = key {
1275	TARGETS.with(\|memo\| memo.borrow_mut().put(key, side.clone(), now));
1276	}
1277	Ok(side)
1278	}
1279
1280	async fn head(&self, a: HeadArgs) -> Result<Option<String>> {
1281	let Some(repo) = self.registry.by_id(&a.repo_id).await? else {
1282	return Ok(None);
1283	};
1284	let branch = if a.branch.is_empty() { &repo.default_branch } else { &a.branch };
1285	let git = self.read_git(&repo).await?;
1286	Ok(git
1287	.log(branch, 1)
1288	.await?
1289	.into_iter()
1290	.next()
1291	.map(\|commit\| commit.hash))
1292	}
1293
1294	async fn delete_branch(&self, a: DeleteBranchArgs) -> Result<Outcome<bool>> {
1295	if !deletable_branch(&a.branch, a.head.as_deref()) {
1296	return Ok(Outcome::fail(
1297	FailureCode::Forbidden,
1298	"Only branches g1t made for itself can be deleted this way.",
1299	));
1300	}
1301	let Some(repo) = self.registry.by_id(&a.repo_id).await? else {
1302	return Ok(not_found());
1303	};
1304	if a.branch == repo.default_branch {
1305	return Ok(Outcome::fail(FailureCode::Forbidden, "The default branch is never deleted."));
1306	}
1307	let repo = match self.unpaused(repo).await? {
1308	Ok(repo) => repo,
1309	Err((code, message)) => return Ok(Outcome::fail(code, message)),
1310	};
1311	self.live(&repo).await?;
1312	let git = self.store.open(&store_key(&repo)).await?;
1313	let Some(old) = git
1314	.branches()
1315	.await?
1316	.into_iter()
1317	.find(\|branch\| branch.name == a.branch)
1318	.map(\|branch\| branch.hash)
1319	else {
1320	return Ok(Outcome::Ok(false));
1321	};
1322	// Moved since the caller looked: someone else's commits are on it.
1323	if a.head.as_deref().is_some_and(\|head\| head != old) {
1324	return Ok(Outcome::fail(FailureCode::Conflict, format!("{} moved, so it was left alone.", a.branch)));
1325	}
1326	let access = git.access(Scope::Write).await?;
1327	let deleted = land::delete_ref(&access, &a.branch, &old).await?;
1328	self.refs_moved(&repo.id).await;
1329	if let Err(reason) = deleted {
1330	return Ok(Outcome::fail(
1331	FailureCode::Conflict,
1332	format!("{} could not be deleted: {reason}", a.branch),
1333	));
1334	}
1335	Ok(Outcome::Ok(true))
1336	}
1337
1338	async fn fork_for_pull(&self, a: ForkArgs) -> Result<Outcome<Repo>> {
1339	let viewer = Some(a.actor.clone());
1340	let Some(source) = self
1341	.registry
1342	.by_id(&a.source_id)
1343	.await?
1344	.filter(\|repo\| can_read(repo, &viewer))
1345	else {
1346	return Ok(not_found());
1347	};
1348	if let Some((code, message)) = lifecycle::read_only_refusal(&source) {
1349	return Ok(Outcome::fail(code, message));
1350	}
1351	// Its working copy is made in its namespace: not while it moves.
1352	let source = match self.unpaused(source).await? {
1353	Ok(source) => source,
1354	Err((code, message)) => return Ok(Outcome::fail(code, message)),
1355	};
1356	let now = now_ms();
1357	let fork = Repo {
1358	id: new_id("rep", now),
1359	namespace: PULLS_NAMESPACE.to_owned(),
1360	name: a.pull_id.clone(),
1361	description: None,
1362	// A fork is exactly as visible as the repo it came from.
1363	is_private: source.is_private,
1364	owner_id: a.actor.id.clone(),
1365	default_branch: source.default_branch.clone(),
1366	fork_of: Some(source.id.clone()),
1367	protected: false,
1368	created_at: rfc3339(now),
1369	topics: Vec::new(),
1370	website: None,
1371	archived_at: None,
1372	};
1373	// Artifacts forks within a namespace: the copy goes where its
1374	// repository is.
1375	let (namespace, _) = store::locate(&store_key(&source));
1376	self.registry
1377	.claim_store_key(&fork, Some(&namespace), &self.store.default_namespace())
1378	.await?;
1379	self.store
1380	.open(&store_key(&source))
1381	.await?
1382	.fork(&store_key(&fork))
1383	.await?;
1384	self.registry.insert(&fork).await?;
1385	self.publish(NewEvent {
1386	kind: "repo.forked",
1387	source: SOURCE,
1388	repo_id: Some(source.id.clone()),
1389	actor: Some(a.actor.id),
1390	data: RepoForked {
1391	repo_id: fork.id.clone(),
1392	source_repo_id: source.id,
1393	pull_id: a.pull_id,
1394	},
1395	})
1396	.await?;
1397	Ok(Outcome::Ok(fork))
1398	}
1399
1400	async fn git_access(&self, a: GitAccessArgs) -> Result<Outcome<GitAccess>> {
1401	let found = self.registry.by_path(&a.path).await?;
1402	Ok(match self.authorize_git(&a.path, &a.viewer, a.service, found).await? {
1403	Outcome::Ok(repo) => {
1404	self.live(&repo).await?;
1405	let write = a.service == GitService::ReceivePack;
1406	if write {
1407	// A push with this credential would not pass through
1408	// here, so nothing that lists the refs is kept until it
1409	// has expired (see refs_cache.rs).
1410	let until = now_ms() + store::CREDENTIAL_LIFE_MS + 60_000;
1411	if let Err(error) = self.registry.refs_open(&repo.id, until).await {
1412	// Before the column exists nothing is kept anyway.
1413	if registry::refs_state(&repo.id).is_some() {
1414	return Err(error);
1415	}
1416	}
1417	}
1418	let scope = if write { Scope::Write } else { Scope::Read };
1419	Outcome::Ok(self.store.handout(&store_key(&repo), scope).await?)
1420	}
1421	Outcome::Fail(failure) => Outcome::Fail(failure),
1422	})
1423	}
1424
1425	/// The repository at `path` (`found`, as just read), if the viewer may
1426	/// use `service` on it: fetch from it, or push to it. A push to a path
1427	/// with nothing there makes the repository, in a workspace the pusher
1428	/// belongs to.
1429	async fn authorize_git(
1430	&self,
1431	path: &RepoPath,
1432	viewer: &Viewer,
1433	service: GitService,
1434	found: Option<Repo>,
1435	) -> Result<Outcome<Repo>> {
1436	let mut a = GitAccessArgs {
1437	path: path.clone(),
1438	viewer: viewer.clone(),
1439	service,
1440	};
1441	let write = a.service == GitService::ReceivePack;
1442	// An access token: pushing needs code:write, reading a private
1443	// repository code:read. A public repository reads as it would for
1444	// anyone. Which repositories a token reaches is its owner's, checked
1445	// below as for anyone.
1446	if let Some(access) = a.viewer.as_ref().and_then(\|user\| user.token.as_deref()).cloned() {
1447	// A workflow job's token, and a deploy key, reach their own
1448	// repository only; a job's also the working copies of that
1449	// repository's pull requests, where their heads are.
1450	let name = format!("{}/{}", path.namespace, path.name);
1451	let source = match found.as_ref().and_then(\|repo\| repo.fork_of.as_deref()) {
1452	Some(source_id) if g1t_contracts::scopes::decide_repo(&access, &name).is_some() => self
1453	.registry
1454	.by_id(source_id)
1455	.await?
1456	.map(\|source\| format!("{}/{}", source.namespace, source.name)),
1457	_ => None,
1458	};
1459	let public = found.as_ref().is_some_and(\|repo\| !repo.is_private);
1460	if let Some(why) = git_token_refusal(&access, &name, source.as_deref(), write, public, found.is_some()) {
1461	return Ok(Outcome::fail(FailureCode::Forbidden, format!("{why}\n")));
1462	}
1463	if !write && !access.allows(g1t_contracts::scopes::Scope::CodeRead) {
1464	a.viewer = None;
1465	}
1466	}
1467
1468	// Anonymous callers are asked to authenticate whether or not the repo
1469	// exists, so private repos cannot be told apart from missing ones.
1470	let denied = \|\| match &a.viewer {
1471	Some(_) => not_found(),
1472	None => Outcome::fail(FailureCode::Unauthenticated, "Authentication required."),
1473	};
1474	// A run's token works through the API only: its sandbox has its
1475	// own way to push, to its own pull request.
1476	if a.viewer.as_ref().is_some_and(\|user\| user.kind == PrincipalKind::Agent) {
1477	return Ok(Outcome::fail(
1478	FailureCode::Forbidden,
1479	"A run's token cannot be used with git this way.",
1480	));
1481	}
1482	if let (true, Some(user)) = (write, &a.viewer)
1483	&& !user.verified
1484	{
1485	return Ok(Outcome::fail(FailureCode::Forbidden, UNVERIFIED));
1486	}
1487	let repo = match found {
1488	Some(repo) => {
1489	let allowed = if write {
1490	can_write(&repo, &a.viewer)
1491	} else {
1492	self.may_read(&repo, &a.viewer).await?
1493	};
1494	if !allowed {
1495	return Ok(denied());
1496	}
1497	// An archived repository, or a pull request's copy of one,
1498	// is read-only.
1499	if write {
1500	let archived = match &repo.fork_of {
1501	Some(source) => self.registry.by_id(source).await?,
1502	None => Some(repo.clone()),
1503	};
1504	match archived {
1505	Some(source) => {
1506	if let Some((code, message)) = lifecycle::read_only_refusal(&source) {
1507	return Ok(Outcome::fail(code, format!("{message}\n")));
1508	}
1509	}
1510	// The repository it was copied from is deleted.
1511	None => return Ok(denied()),
1512	}
1513	}
1514	repo
1515	}
1516	None => {
1517	// Push to create, in a workspace the pusher belongs to.
1518	let owner = a
1519	.viewer
1520	.as_ref()
1521	.filter(\|user\| write && user.is_member(&a.path.namespace.to_lowercase()));
1522	let Some(owner) = owner else {
1523	return Ok(denied());
1524	};
1525	let created = self.create(push_to_create(owner, &a.path)).await?;
1526	match created {
1527	Outcome::Ok(repo) => repo,
1528	Outcome::Fail(failure) => return Ok(Outcome::Fail(failure)),
1529	}
1530	}
1531	};
1532	// A push, or a credential to push with, waits while the repository
1533	// moves between namespaces (moves.rs), and goes to where it is now.
1534	if write {
1535	return Ok(match self.unpaused(repo).await? {
1536	Ok(repo) => Outcome::Ok(repo),
1537	Err((code, message)) => Outcome::fail(code, format!("{message}\n")),
1538	});
1539	}
1540	Ok(Outcome::Ok(repo))
1541	}
1542
1543	async fn land(&self, a: LandArgs) -> Result<Outcome<Landed>> {
1544	let actor: Viewer = Some(a.actor.clone());
1545	let Some(source) = self.registry.by_id(&a.source_id).await? else {
1546	return Ok(not_found());
1547	};
1548	// A fork lands on the repository it came from; a branch on its own.
1549	let target = match &source.fork_of {
1550	Some(id) => self.registry.by_id(id).await?,
1551	None => Some(source.clone()),
1552	};
1553	let Some(target) = target.filter(\|repo\| can_read(repo, &actor)) else {
1554	return Ok(not_found());
1555	};
1556	if !registry::can(&target, &actor, Capability::Merge) {
1557	return Ok(Outcome::fail(
1558	FailureCode::Forbidden,
1559	access::needs(Capability::Merge, &format!("{}/{}", target.namespace, target.name)),
1560	));
1561	}
1562	if !a.actor.verified {
1563	return Ok(Outcome::fail(FailureCode::Forbidden, UNVERIFIED));
1564	}
1565	if let Some((code, message)) = lifecycle::read_only_refusal(&target) {
1566	return Ok(Outcome::fail(code, message));
1567	}
1568	// Moving between namespaces: wait for it (moves.rs). Both are read
1569	// again once it is done, for their new keys.
1570	let (source, target) = match (self.unpaused(source).await?, self.unpaused(target).await?) {
1571	(Ok(source), Ok(target)) => (source, target),
1572	(Err((code, message)), _) \| (_, Err((code, message))) => return Ok(Outcome::fail(code, message)),
1573	};
1574
1575	let branch = &a.target_branch.clone().unwrap_or_else(\|\| target.default_branch.clone());
1576	let from_fork = source.id != target.id;
1577	let source_branch = match a.branch {
1578	Some(name) if !from_fork && name == *branch => {
1579	return Ok(Outcome::fail(
1580	FailureCode::Invalid,
1581	format!("{branch} cannot be merged into itself."),
1582	));
1583	}
1584	Some(name) => name,
1585	None if from_fork => branch.clone(),
1586	None => {
1587	return Ok(Outcome::fail(
1588	FailureCode::Invalid,
1589	"Say which branch to merge.",
1590	));
1591	}
1592	};
1593
1594	self.live(&source).await?;
1595	let source_git = self.store.open(&store_key(&source)).await?;
1596	let target_git = self.store.open(&store_key(&target)).await?;
1597	let history = source_git.log(&source_branch, MAX_ANCESTRY).await?;
1598	let Some(new) = history.first().map(\|commit\| commit.hash.clone()) else {
1599	return Ok(Outcome::fail(
1600	FailureCode::Conflict,
1601	"This pull request has no commits to merge.",
1602	));
1603	};
1604	let old = target_git
1605	.log(branch, 1)
1606	.await?
1607	.into_iter()
1608	.next()
1609	.map(\|commit\| commit.hash);
1610
1611	if old.as_deref() == Some(new.as_str()) {
1612	return Ok(Outcome::Ok(Landed {
1613	commit: new,
1614	previous: None,
1615	}));
1616	}
1617	// Moving the branch to a commit that does not descend from its
1618	// current head would discard whatever landed in between.
1619	if let Some(old) = &old
1620	&& !descends_from(&source_git, &history, old).await?
1621	{
1622	let remedy = if from_fork {
1623	format!("Pull {branch} into the pull request's fork, push, and merge again.")
1624	} else {
1625	format!("Merge {branch} into {source_branch}, push, and merge again.")
1626	};
1627	return Ok(Outcome::fail(
1628	FailureCode::Conflict,
1629	format!("{branch} has moved since this pull request was opened. {remedy}"),
1630	));
1631	}
1632
1633	// For a branch the objects are already in the target; sending them
1634	// again is harmless and keeps one way of moving a ref.
1635	let source_access = source_git.access(Scope::Read).await?;
1636	let target_access = target_git.access(Scope::Write).await?;
1637	let pushed =
1638	land::fast_forward(&source_access, &target_access, branch, old.as_deref(), &new)
1639	.await?;
1640	self.refs_moved(&target.id).await;
1641	if let Err(reason) = pushed {
1642	// Most often another pull request landed between the check and the push.
1643	return Ok(Outcome::fail(
1644	FailureCode::Conflict,
1645	format!("{branch} could not be updated: {reason}"),
1646	));
1647	}
1648	self.publish_push(
1649	&target,
1650	&format!("refs/heads/{branch}"),
1651	old.as_deref(),
1652	&new,
1653	Some(&a.actor),
1654	None,
1655	)
1656	.await?;
1657	Ok(Outcome::Ok(Landed {
1658	commit: new,
1659	previous: old,
1660	}))
1661	}
1662
1663	async fn compare(&self, a: CompareArgs) -> Result<Outcome<Comparison>> {
1664	let Some(repo) = self
1665	.visible(self.registry.by_id(&a.repo_id).await?, &a.viewer)
1666	.await?
1667	else {
1668	return Ok(not_found());
1669	};
1670	let git = self.read_git(&repo).await?;
1671	let head_ref = a.head.as_deref().unwrap_or(&repo.default_branch);
1672	// A pull request into another branch is compared from where it
1673	// left that branch.
1674	let base_branch = a.base_branch.clone();
1675	// The head's history is only searched when the base is worked out
1676	// from another branch.
1677	let depth = if a.base.is_some() \|\| is_commit_hash(head_ref) { 1 } else { MAX_ANCESTRY };
1678	let history = git.log(head_ref, depth).await?;
1679	let Some(head) = history.first() else {
1680	return Ok(Outcome::fail(
1681	FailureCode::Conflict,
1682	"There are no commits to compare.",
1683	));
1684	};
1685
1686	// Where the head's history meets the default branch of `against`,
1687	// or the branch asked for.
1688	let shared_with = async \|against: &Repo\| -> Result<Option<String>> {
1689	let against_git = self.read_git(against).await?;
1690	let branch = base_branch.as_deref().unwrap_or(&against.default_branch);
1691	let shared: HashSet<String> = against_git
1692	.log(branch, MAX_ANCESTRY)
1693	.await?
1694	.into_iter()
1695	.map(\|commit\| commit.hash)
1696	.collect();
1697	nearest_ancestor_in(&git, &history, &shared).await
1698	};
1699	let base = match (a.base, &repo.fork_of) {
1700	(Some(base), _) => Some(base),
1701	// A fork is compared with the last commit it shares with the
1702	// repository it came from.
1703	(None, Some(target_id)) => match self.registry.by_id(target_id).await? {
1704	Some(target) => shared_with(&target).await?,
1705	None => None,
1706	},
1707	// A branch, with the point where it left the default branch.
1708	// A single commit, with its first parent.
1709	(None, None) if is_commit_hash(head_ref) => head.parents.first().cloned(),
1710	(None, None) if head_ref != base_branch.as_deref().unwrap_or(&repo.default_branch) => {
1711	shared_with(&repo).await?
1712	}
1713	(None, None) => head.parents.first().cloned(),
1714	};
1715	let base_tree = match &base {
1716	Some(base) => git
1717	.log(base, 1)
1718	.await?
1719	.into_iter()
1720	.next()
1721	.map(\|commit\| commit.tree_hash),
1722	None => None,
1723	};
1724	// Only a few files of it, when asked: how the files a large
1725	// comparison lists without their lines are read, as a page reaches them.
1726	let only = a.paths.as_deref().map(\|paths\| &paths[..paths.len().min(MAX_COMPARED_PATHS)]);
1727	let (files, truncated) =
1728	diff::compare_trees(&git, base_tree.as_deref(), &head.tree_hash, only).await?;
1729	Ok(Outcome::Ok(Comparison {
1730	base,
1731	head: head.hash.clone(),
1732	files,
1733	truncated,
1734	}))
1735	}
1736
1737	/// Reports that `git_ref` of `repo` (a full ref) now points to `after`,
1738	/// moved by `actor` (marked when that was a workflow job's token).
1739	/// `commits` is how many the move brought, when the caller knows: one
1740	/// for a file committed on the site or a branch brought up to date.
1741	async fn publish_push(
1742	&self,
1743	repo: &Repo,
1744	git_ref: &str,
1745	before: Option<&str>,
1746	after: &str,
1747	actor: Option<&User>,
1748	commits: Option<u32>,
1749	) -> Result<()> {
1750	let caused_by_job = actor.and_then(g1t_contracts::events::job_run_of).map(str::to_owned);
1751	self.publish_git_push(repo, git_ref, before, after, actor.map(\|user\| user.id.clone()), false, caused_by_job, commits)
1752	.await
1753	}
1754
1755	/// `publish_push`, saying whether the push reached the store without
1756	/// being scanned for secrets first.
1757	#[allow(clippy::too_many_arguments)]
1758	async fn publish_git_push(
1759	&self,
1760	repo: &Repo,
1761	git_ref: &str,
1762	before: Option<&str>,
1763	after: &str,
1764	actor: Option<String>,
1765	unscanned: bool,
1766	caused_by_job: Option<String>,
1767	commits: Option<u32>,
1768	) -> Result<()> {
1769	self.publish(NewEvent {
1770	kind: "git.push",
1771	source: SOURCE,
1772	repo_id: Some(repo.id.clone()),
1773	actor,
1774	data: GitPush {
1775	repo_id: repo.id.clone(),
1776	git_ref: git_ref.to_owned(),
1777	before: before.map(str::to_owned),
1778	after: after.to_owned(),
1779	default_branch: git_ref.strip_prefix("refs/heads/")
1780	== Some(repo.default_branch.as_str()),
1781	unscanned,
1782	caused_by_job,
1783	commits,
1784	},
1785	})
1786	.await
1787	}
1788
1789	/// Git over HTTPS. Only what decides the answer happens before it:
1790	/// the repository, who is asking and whether they may, the free
1791	/// workspace limits, push protection, and the store's own answer. The
1792	/// audit entry and what a push changed are recorded once git has its
1793	/// answer. Each answer says how long its steps took (`Server-Timing`).
1794	async fn git_http(&self, request: Request, env: &Env, ctx: &Context) -> Result<Response> {
1795	let mut timing = git_http::Timing::start();
1796	let Some(git) = git_http::parse(&request.url()?) else {
1797	return Response::error("Not found", 404);
1798	};
1799	let response = match self.answer_git(request, &git, env, ctx, &mut timing).await {
1800	Ok(response) => response,
1801	// The git store is busy: git hears when to try again.
1802	Err(error) => match resilience::busy(&error.to_string()) {
1803	Some(busy) => git_http::busy_response(busy)?,
1804	None => return Err(error),
1805	},
1806	};
1807	timing.apply(response)
1808	}
1809
1810	async fn answer_git(
1811	&self,
1812	request: Request,
1813	git: &git_http::GitRequest,
1814	env: &Env,
1815	ctx: &Context,
1816	timing: &mut git_http::Timing,
1817	) -> Result<Response> {
1818	let write = git.service == GitService::ReceivePack;
1819	let get = request.method() == Method::Get;
1820	let identity = env.service("IDENTITY")?;
1821	// The repository and the caller's credentials, at once. A fetch may
1822	// go by the row as read a moment ago, for the same clone's next
1823	// request; a push always reads it. Anonymous callers cost nothing.
1824	let lookup = async {
1825	if write {
1826	self.registry.by_path(&git.path).await
1827	} else {
1828	self.registry.by_path_recent(&git.path).await
1829	}
1830	};
1831	let (found, viewer) =
1832	futures_util::future::join(lookup, git_http::viewer(&request, &identity)).await;
1833	let mut found = found?;
1834	timing.mark("repo");
1835	// A workspace alias staff set (identity's aliases.rs: `g1t` for
1836	// `flagon-io`) is answered in place, as the repository under the
1837	// workspace's slug: pushes and some clients do not follow
1838	// redirects. Everything after this sees only the workspace's slug.
1839	let aliased = match found {
1840	Some(_) => None,
1841	None => git_http::aliased(git, &identity).await?,
1842	};
1843	if let Some(aliased) = &aliased {
1844	found = if write {
1845	self.registry.by_path(&aliased.path).await?
1846	} else {
1847	self.registry.by_path_recent(&aliased.path).await?
1848	};
1849	timing.mark("alias");
1850	}
1851	let git = aliased.as_ref().unwrap_or(git);
1852	if found.is_none() {
1853	// A workspace that was renamed: git follows a redirect when it
1854	// first asks for refs, and uses the new address from then on.
1855	// A repository transferred to another workspace: the same, to
1856	// its new path. Fetches and pushes both follow either.
1857	let url = request.url()?;
1858	let (renamed, moved) = futures_util::future::join(
1859	git_http::renamed(&url, &identity),
1860	self.registry.resolve_moved(&git.path),
1861	)
1862	.await;
1863	timing.mark("moved");
1864	if let Some(location) = renamed? {
1865	return git_http::moved(&location, get);
1866	}
1867	if let Some(now) = moved?
1868	&& let Some(location) = git_http::transferred(&url, &now)
1869	{
1870	return git_http::moved(&location, get);
1871	}
1872	}
1873	let viewer = viewer?;
1874	// A person who has not confirmed their email address can do
1875	// nothing with git until they do: told so, not asked to sign in.
1876	if viewer.as_ref().is_some_and(g1t_contracts::User::awaits_confirmation) {
1877	let site = request.url()?.origin().ascii_serialization();
1878	return git_http::refuse(Outcome::<()>::fail(
1879	FailureCode::Forbidden,
1880	g1t_contracts::accounts::confirm_email_first(&site),
1881	));
1882	}
1883	// A run credential is checked against its grants, then acts as the
1884	// person it works for. See run_access.rs.
1885	let (request, viewer, audit) = match self.admit_git(request, git, viewer, found.as_ref()).await? {
1886	run_access::Admitted::Go { request, viewer, entry } => (request, viewer, entry),
1887	run_access::Admitted::Refused(response) => return Ok(response),
1888	};
1889	let mut after = AfterGit {
1890	audit,
1891	status: 0,
1892	message: None,
1893	push: None,
1894	};
1895	let repo = match self.authorize_git(&git.path, &viewer, git.service, found).await? {
1896	Outcome::Ok(repo) => repo,
1897	refused => {
1898	let response = git_http::refuse(refused)?;
1899	after.ended(response.status_code(), None);
1900	after.spawn(env, ctx);
1901	return Ok(response);
1902	}
1903	};
1904	// A pull request's working copy removed after it closed is made
1905	// again before git uses it (forks.rs).
1906	self.live(&repo).await?;
1907	timing.mark("access");
1908	// Clones check out the default branch g1t keeps, which can have
1909	// changed since the store made the repository.
1910	let default_branch = repo.fork_of.is_none().then(\|\| repo.default_branch.clone());
1911	let key = store_key(&repo);
1912	let scope = if write { Scope::Write } else { Scope::Read };
1913	let mut request = request;
1914	let protocol = refs_cache::protocol(request.headers().get("git-protocol")?.as_deref());
1915	// A fetch's POST is read here, to tell an `ls-refs` from a fetch of
1916	// objects; the store would have it read in full anyway.
1917	let body = if !write && !get { Some(request.bytes().await?) } else { None };
1918	// What it asks the store, for the meters (meters.rs).
1919	let call = git_ops::classify(git.service, git.endpoint, get, body.as_deref());
1920	// Answers kept from the usual store may name refs the fallback
1921	// store does not have (fallback.rs): none are used, or kept.
1922	let fallback = self.store.on_fallback(&key);
1923	// An answer that lists refs may have been kept: see refs_cache.rs.
1924	let kept_key = refs_cache::kind(git, get, protocol, body.as_deref())
1925	.filter(\|_\| !fallback)
1926	.zip(refs_cache::usable(registry::refs_state(&repo.id), now_ms()))
1927	.map(\|(kind, version)\| {
1928	refs_cache::Key::new(&repo.id, version, default_branch.as_deref(), protocol, &kind)
1929	});
1930	// A fresh clone's pack may have been kept too: see pack_cache.rs.
1931	// Under the same refs version, so never across a change to them.
1932	let pack_key = self
1933	.packs
1934	.as_ref()
1935	.filter(\|_\| !fallback)
1936	.and_then(\|_\| {
1937	let encoding = request.headers().get("content-encoding").ok().flatten();
1938	pack_cache::cacheable(git, get, protocol, encoding.as_deref(), body.as_deref())
1939	})
1940	.zip(refs_cache::usable(registry::refs_state(&repo.id), now_ms()))
1941	.map(\|(normalized, version)\| pack_cache::Key::new(&repo.id, version, &normalized));
1942	// A kept answer and the free workspace limits, with a kept
1943	// credential and, for a push, what the repository holds looked up
1944	// alongside. A kept answer goes back without waiting for the
1945	// credential, which it does not need.
1946	let pushing = write && !get;
1947	let ((answer, pack, (limited, held)), kept_access) = {
1948	let shared = self.shared.as_deref();
1949	let answer_and_limits = std::pin::pin!(futures_util::future::join3(
1950	async {
1951	match &kept_key {
1952	Some(kept_key) => refs_cache::get(shared, kept_key).await,
1953	None => None,
1954	}
1955	},
1956	async {
1957	match (&pack_key, self.packs.as_deref()) {
1958	(Some(pack_key), Some(packs)) => pack_cache::get(packs, pack_key).await,
1959	_ => None,
1960	}
1961	},
1962	futures_util::future::join(self.git_limits(call, git, &repo, env), async {
1963	if pushing { Some(self.held(&repo).await) } else { None }
1964	}),
1965	));
1966	let kept_access = std::pin::pin!(self.store.kept_access(&key, scope));
1967	match futures_util::future::select(answer_and_limits, kept_access).await {
1968	futures_util::future::Either::Left((first, kept_access)) => {
1969	let answered = first.0.is_some() \|\| first.1.is_some() \|\| matches!(first.2.0, Ok(Some(_)) \| Err(_));
1970	(first, if answered { None } else { kept_access.await })
1971	}
1972	futures_util::future::Either::Right((kept_access, first)) => (first.await, kept_access),
1973	}
1974	};
1975	timing.mark("kept");
1976	// A kept pack first: it never reaches the store, so it is never an
1977	// operation, and a free workspace past its operation cap still gets
1978	// it. (The other limits are a push's, and a pack is only a fetch.)
1979	if let Some(kept) = pack {
1980	timing.note("pack", "hit");
1981	let sent = body.as_ref().map_or(0, \|body\| body.len() as u64);
1982	meters::record(pack_cache::HIT, &key, sent, kept.size);
1983	after.ended(200, None);
1984	after.spawn(env, ctx);
1985	return kept.response();
1986	}
1987	if let Some((response, status, message)) = limited? {
1988	after.ended(status, Some(message.to_owned()));
1989	after.spawn(env, ctx);
1990	return Ok(response);
1991	}
1992	if pack_key.is_some() {
1993	timing.note("pack", "miss");
1994	}
1995	if let (Some((entry, found)), Some(kept_key)) = (answer, &kept_key) {
1996	timing.note("refs", found.as_str());
1997	if found == refs_cache::Found::Shared {
1998	let (kept_key, entry) = (kept_key.clone(), entry.clone());
1999	ctx.wait_until(async move { refs_cache::keep_in_colo(&kept_key, &entry).await });
2000	}
2001	// Never reached the store: never an operation.
2002	meters::record(call.cached_meter(), &key, 0, entry.body.len() as u64);
2003	after.ended(200, None);
2004	after.spawn(env, ctx);
2005	return entry.response();
2006	}
2007	if kept_key.is_some() {
2008	timing.note("refs", "miss");
2009	}
2010	// An anonymous fetch the store is about to answer: an operation for
2011	// the repository's workspace, so limited per repository (limits.rs).
2012	if limits::counts_as_anonymous_fetch(call, viewer.is_none())
2013	&& g1t_kit::limits::check(env, limits::ANONYMOUS_FETCH, repo.id.clone()).await.limited()
2014	{
2015	let response = limits::too_many_anonymous_fetches(&format!("{}/{}", repo.namespace, repo.name))?;
2016	after.ended(429, Some("Too many anonymous fetches.".to_owned()));
2017	after.spawn(env, ctx);
2018	return Ok(response);
2019	}
2020	// The store's credential: one made a moment ago, here or in another
2021	// isolate (see store.rs), or a new one.
2022	let access = match kept_access {
2023	Some((access, from)) => {
2024	timing.note("cred", from.as_str());
2025	access
2026	}
2027	None => {
2028	let access = self.store.mint_access(&key, scope).await?;
2029	timing.mark("mint");
2030	timing.note("cred", "mint");
2031	access
2032	}
2033	};
2034	// Should the store turn a kept credential down, a fetch's first
2035	// request is tried again with a new one; the requests after it then
2036	// have that one too.
2037	let again = if get { Some(request.clone()?) } else { None };
2038	// A push's checks: workflow files, the rules of the branches and
2039	// tags it changes, saying which rule and why (rules.rs), and push
2040	// protection, which refuses a push that adds a secret
2041	// (secret_scan.rs). See push_checks.rs.
2042	let checks = async \|head: &[u8], whole: bool, scan: bool\| self.check_receive(&repo, viewer.as_ref(), head, whole, scan).await;
2043	// What a push may bring (pack_limits.rs): the repository's size is
2044	// its own and its pull requests' working copies'.
2045	let limits = if let Some(held) = held {
2046	git_http::PushLimits {
2047	held,
2048	repo_limit: self.repo_limit,
2049	large: self.large_pushes,
2050	..git_http::PushLimits::default()
2051	}
2052	} else {
2053	git_http::PushLimits::default()
2054	};
2055	let mut outcome = git_http::forward(
2056	request,
2057	body,
2058	git,
2059	&access,
2060	checks,
2061	default_branch.as_deref(),
2062	limits,
2063	timing,
2064	)
2065	.await?;
2066	let turned_down = matches!(
2067	&outcome,
2068	git_http::Push::Forwarded(forwarded) if matches!(forwarded.response.status_code(), 401 \| 403)
2069	);
2070	if turned_down {
2071	self.store.forget_access(&key).await;
2072	if let Some(again) = again {
2073	let access = self.store.mint_access(&key, scope).await?;
2074	outcome = git_http::forward(
2075	again,
2076	None,
2077	git,
2078	&access,
2079	async \|_: &[u8], _: bool, _: bool\| Ok(push_checks::Checks::clear()),
2080	default_branch.as_deref(),
2081	git_http::PushLimits::default(),
2082	timing,
2083	)
2084	.await?;
2085	}
2086	}
2087	let forwarded =
2088	match outcome {
2089	git_http::Push::Forwarded(forwarded) => forwarded,
2090	git_http::Push::Refused(response) => {
2091	after.ended(403, Some("The push was declined by rules.".to_owned()));
2092	after.spawn(env, ctx);
2093	return Ok(response);
2094	}
2095	git_http::Push::Blocked(response) => {
2096	after.ended(403, Some("The push adds a secret.".to_owned()));
2097	after.spawn(env, ctx);
2098	return Ok(response);
2099	}
2100	git_http::Push::Declined(response, reason) => {
2101	after.ended(403, Some(format!("The push was declined: {reason}.")));
2102	after.spawn(env, ctx);
2103	return Ok(response);
2104	}
2105	};
2106	if forwarded.from_store {
2107	let received = forwarded
2108	.response
2109	.headers()
2110	.get("content-length")?
2111	.and_then(\|length\| length.parse().ok())
2112	.unwrap_or(0);
2113	meters::record(call.meter(), &key, forwarded.sent, received);
2114	}
2115	timing.mark("store");
2116	let mut response = forwarded.response;
2117	let status = response.status_code();
2118	if write && !get {
2119	// A push: the store has moved its refs once it has answered in
2120	// full, so the answer is read before the change is recorded, and
2121	// only then goes back. Whoever fetches after it sees the push.
2122	let headers = response.headers().clone();
2123	headers.delete("content-length")?;
2124	let report = response.bytes().await?;
2125	self.refs_moved(&repo.id).await;
2126	timing.mark("refs");
2127	response = Response::from_bytes(report)?.with_headers(headers).with_status(status);
2128	} else if let (Some(kept_key), 200) = (&kept_key, status) {
2129	// A miss: this answer is kept for the next to ask.
2130	let headers = response.headers().clone();
2131	headers.delete("content-length")?;
2132	let body = response.bytes().await?;
2133	if let Some(content_type) = headers.get("content-type")? {
2134	let entry = refs_cache::Entry { content_type, body: body.clone() };
2135	if entry.keepable() {
2136	let shared = self.shared.clone();
2137	let kept_key = kept_key.clone();
2138	ctx.wait_until(async move { refs_cache::keep(shared.as_deref(), &kept_key, &entry).await });
2139	}
2140	}
2141	response = Response::from_bytes(body)?.with_headers(headers).with_status(status);
2142	} else if let (Some(pack_key), Some(packs), true) = (&pack_key, &self.packs, forwarded.from_store) {
2143	// A fresh clone the bucket did not have: counted, and its pack
2144	// kept as it streams to git, when it is a whole one.
2145	meters::record(pack_cache::MISS, &key, forwarded.sent, 0);
2146	// Past the repository's limit on writes to the bucket, the pack
2147	// goes to git without being kept (limits.rs).
2148	if status == 200 && !g1t_kit::limits::check(env, limits::PACK_FILL, repo.id.clone()).await.limited() {
2149	let store_key = key.clone();
2150	let measured = Box::new(move \|bytes: u64\| meters::record_bytes(pack_cache::MISS, &store_key, 0, bytes));
2151	let (teed, filling) = pack_cache::tee(response, packs.clone(), pack_key, measured)?;
2152	response = teed;
2153	if let Some(filling) = filling {
2154	let pack_key = pack_key.clone();
2155	ctx.wait_until(async move {
2156	let filled = filling.await;
2157	if !matches!(filled, pack_cache::Filled::Kept { .. } \| pack_cache::Filled::Abandoned) {
2158	worker::console_warn!("pack {} not kept: {filled:?}", pack_key.as_str());
2159	}
2160	});
2161	}
2162	}
2163	}
2164	after.ended(status, None);
2165	if status == 200 && (forwarded.pack_bytes > 0 \|\| !forwarded.pushed.is_empty() \|\| !forwarded.deleted.is_empty()) {
2166	after.push = Some(PushDone {
2167	repo,
2168	pushed: forwarded.pushed,
2169	deleted: forwarded.deleted,
2170	pack_bytes: forwarded.pack_bytes,
2171	caused_by_job: viewer.as_ref().and_then(g1t_contracts::events::job_run_of).map(str::to_owned),
2172	// The calendar credits people, never a job's, an agent's or a workspace's token.
2173	credit: viewer
2174	.as_ref()
2175	.filter(\|user\| user.kind == PrincipalKind::User && g1t_contracts::events::job_run_of(user).is_none())
2176	.map(\|user\| user.id.clone()),
2177	actor: viewer.map(\|user: User\| user.id),
2178	unscanned: forwarded.unscanned,
2179	});
2180	}
2181	after.spawn(env, ctx);
2182	Ok(response)
2183	}
2184
2185	/// The answer for a request a free workspace's limits stop, or a push
2186	/// to a full repository, with its status and reason for the audit log;
2187	/// `None` to go on.
2188	///
2189	/// A clone, fetch or push is a git operation, which the git store
2190	/// charges g1t for: counted for billing once the answer has gone back
2191	/// (meters.rs), and a free workspace far past its share is slowed down
2192	/// rather than charged (see git_ops.rs). Whether it is past it is
2193	/// decided from counts this isolate already holds: the database is not
2194	/// asked on the way. A free workspace is never charged for private
2195	/// storage: once its private repositories hold the free amount, pushes
2196	/// to them stop, checked when a push begins so that git shows the
2197	/// reason. So do pushes to a repository at the store's size limit.
2198	async fn git_limits(
2199	&self,
2200	call: git_ops::GitCall,
2201	git: &git_http::GitRequest,
2202	repo: &Repo,
2203	env: &Env,
2204	) -> Result<Option<(Response, u16, &'static str)>> {
2205	let namespace = git.path.namespace.to_lowercase();
2206	if meters::mapping_now().billable(call.meter()) > 0.0 {
2207	let now = now_ms();
2208	let hour = git_ops::hour_key(&rfc3339(now));
2209	let limits = git_ops::Limits::from_env(env);
2210	if let Some((month, hour_ops)) = git_ops::standing(&namespace, &hour, now)
2211	&& git_ops::slow_down(month + 1, hour_ops + 1, limits.free_cap, limits.hourly)
2212	&& git_ops::is_free_kept(env.service("BILLING").ok().as_ref(), &namespace).await
2213	{
2214	return Ok(Some((
2215	git_ops::too_many(&namespace, limits.free_cap, limits.hourly)?,
2216	429,
2217	"Too many git operations this hour.",
2218	)));
2219	}
2220	}
2221	if git.service == GitService::ReceivePack && git.endpoint == "info/refs" {
2222	let held = self.held(repo).await;
2223	if held >= self.repo_limit {
2224	let message = format!(
2225	"{}/{} holds about {}, the most a repository may hold on g1t, so it takes no more pushes. Delete what you no longer need, or split it: https://docs.g1t.sh/guides/git/#size-limits\n",
2226	repo.namespace,
2227	repo.name,
2228	pack_limits::megabytes(held)
2229	);
2230	return Ok(Some((Response::error(message, 403)?, 403, "The repository is full.")));
2231	}
2232	}
2233	if git.service == GitService::ReceivePack && git.endpoint == "info/refs" && repo.is_private {
2234	let free = git_ops::free_private_bytes(env);
2235	let held = self.registry.private_bytes(&namespace).await.unwrap_or(0);
2236	if git_ops::storage_full(held, free)
2237	&& git_ops::is_free(env.service("BILLING").ok().as_ref(), &namespace).await
2238	{
2239	return Ok(Some((
2240	git_ops::storage_full_response(&namespace, held, free)?,
2241	403,
2242	"Free private storage is full.",
2243	)));
2244	}
2245	}
2246	if git.service == GitService::ReceivePack && git.endpoint == "info/refs" && !repo.is_private {
2247	let free = git_ops::free_public_bytes(env);
2248	if free > 0 {
2249	let held = self.registry.public_bytes(&namespace).await.unwrap_or(0);
2250	if git_ops::public_storage_full(held, free)
2251	&& git_ops::is_free(env.service("BILLING").ok().as_ref(), &namespace).await
2252	{
2253	return Ok(Some((
2254	git_ops::public_storage_full_response(&namespace, held, free)?,
2255	403,
2256	"Free public storage is full.",
2257	)));
2258	}
2259	}
2260	}
2261	Ok(None)
2262	}
2263
2264	/// What a repository and its pull requests' working copies hold, as
2265	/// g1t counts it: read for a push's first request, kept a minute for
2266	/// the rest of it.
2267	async fn held(&self, repo: &Repo) -> u64 {
2268	let root = repo.fork_of.clone().unwrap_or_else(\|\| repo.id.clone());
2269	let now = now_ms();
2270	if let Some(held) = HELD.with(\|held\| held.borrow().get(&root, now)) {
2271	return held;
2272	}
2273	let held = self.registry.stored_bytes(&root).await.unwrap_or(0).max(0) as u64;
2274	HELD.with(\|kept\| kept.borrow_mut().put(root, held, now));
2275	held
2276	}
2277
2278	/// A `git.deleted` event for each branch or tag a push deleted that is
2279	/// gone from the store: the store can refuse one ref and accept
2280	/// another. As many as a push announces of tags and branches it moves.
2281	async fn publish_deleted(
2282	&self,
2283	repo: &Repo,
2284	deleted: &[(String, String)],
2285	actor: Option<&str>,
2286	caused_by_job: Option<&str>,
2287	) -> Result<()> {
2288	let stored = self.store.open(&store_key(repo)).await?;
2289	let is_branch = \|git_ref: &str\| git_ref.starts_with("refs/heads/");
2290	let tags = deleted.iter().filter(\|(git_ref, _)\| !is_branch(git_ref)).count();
2291	let announced = deleted
2292	.iter()
2293	.filter(\|(git_ref, _)\| is_branch(git_ref))
2294	.take(MAX_PUSH_BRANCH_EVENTS)
2295	.chain(deleted.iter().filter(\|(git_ref, _)\| !is_branch(git_ref) && tags <= MAX_PUSH_TAG_EVENTS));
2296	for (git_ref, before) in announced {
2297	let named = git_ref.strip_prefix("refs/heads/").unwrap_or(git_ref);
2298	let gone = match stored.log(named, 1).await {
2299	Ok(head) => head.is_empty(),
2300	Err(error) => {
2301	worker::console_error!("whether {git_ref} of {} was deleted is not known: {error}", repo.name);
2302	false
2303	}
2304	};
2305	if !gone {
2306	continue;
2307	}
2308	self.publish(NewEvent {
2309	kind: "git.deleted",
2310	source: SOURCE,
2311	repo_id: Some(repo.id.clone()),
2312	actor: actor.map(str::to_owned),
2313	data: GitDeleted {
2314	repo_id: repo.id.clone(),
2315	git_ref: git_ref.clone(),
2316	before: before.clone(),
2317	caused_by_job: caused_by_job.map(str::to_owned),
2318	},
2319	})
2320	.await?;
2321	}
2322	Ok(())
2323	}
2324
2325	/// What a push changed, recorded once git has its answer.
2326	async fn record_push(&self, push: PushDone) -> Result<()> {
2327	let PushDone {
2328	repo,
2329	pushed,
2330	deleted,
2331	pack_bytes,
2332	actor,
2333	unscanned,
2334	caused_by_job,
2335	credit,
2336	} = push;
2337	// What the push stored, for billing's storage meter. A failure only
2338	// leaves the count short.
2339	if pack_bytes > 0
2340	&& let Err(error) = self.registry.add_stored_bytes(&repo, pack_bytes).await
2341	{
2342	worker::console_error!("stored bytes for {} not counted: {error}", repo.name);
2343	}
2344	if !deleted.is_empty() {
2345	self.publish_deleted(&repo, &deleted, actor.as_deref(), caused_by_job.as_deref()).await?;
2346	}
2347	if pushed.is_empty() {
2348	return Ok(());
2349	}
2350	// Artifacts' own push notifications are per repository, which does
2351	// not fit a repo per pull request, so the front end reports pushes
2352	// itself: one event for each branch that moved.
2353	let stored = self.store.open(&store_key(&repo)).await?;
2354	let announced = announced_refs(&pushed, &repo.default_branch);
2355	if announced.len() < pushed.len() {
2356	worker::console_log!(
2357	"push to {}: {} of {} refs announced",
2358	repo.name,
2359	announced.len(),
2360	pushed.len()
2361	);
2362	}
2363	for pushed in announced {
2364	// The store can refuse one ref and accept another, so each
2365	// branch is checked against where it actually is. A tag the
2366	// store cannot read back is taken as pushed.
2367	let moved = match pushed.branch() {
2368	Some(branch) => stored
2369	.log(branch, 1)
2370	.await?
2371	.first()
2372	.is_some_and(\|commit\| commit.hash == pushed.after),
2373	None => stored.log(&pushed.git_ref, 1).await.map_or(true, \|head\| {
2374	head.first().is_none_or(\|commit\| commit.hash == pushed.after)
2375	}),
2376	};
2377	// How many commits a branch push brought (push_commits.rs), for
2378	// the event and the person's contribution calendar. A log the
2379	// store cannot read leaves the event without a count and the
2380	// day short.
2381	let commits = match (moved, pushed.branch()) {
2382	(true, Some(_)) => match stored.log(&pushed.after, push_commits::MOST_PER_PUSH + 1).await {
2383	Ok(log) => Some(push_commits::new_commits(&log, pushed.before.as_deref())),
2384	Err(error) => {
2385	worker::console_error!("commits pushed to {} not counted: {error}", repo.name);
2386	None
2387	}
2388	},
2389	_ => None,
2390	};
2391	if let (Some(commits), Some(user_id), Some(branch)) = (commits, credit.as_deref(), pushed.branch())
2392	&& push_commits::counts(branch, &repo.default_branch)
2393	&& let Err(error) =
2394	push_commits::record(&self.registry.db, user_id, &repo.id, &rfc3339(now_ms())[..10], commits).await
2395	{
2396	worker::console_error!("commits pushed to {} not credited: {error}", repo.name);
2397	}
2398	if moved {
2399	self.publish_git_push(
2400	&repo,
2401	&pushed.git_ref,
2402	pushed.before.as_deref(),
2403	&pushed.after,
2404	actor.clone(),
2405	unscanned,
2406	caused_by_job.clone(),
2407	commits,
2408	)
2409	.await?;
2410	}
2411	}
2412	Ok(())
2413	}
2414	}
2415
2416	/// The most tags one push announces: a push of more (`git push --tags`
2417	/// into a new repository, say) announces none of them, so it starts no
2418	/// workflows or package reads, one per tag.
2419	const MAX_PUSH_TAG_EVENTS: usize = 3;
2420	/// The most branches one push announces. A push of more announces only
2421	/// the default branch, if it moved, which the rest of g1t reads from.
2422	const MAX_PUSH_BRANCH_EVENTS: usize = 1000;
2423
2424	/// The refs of a push that are announced with a `git.push` event each.
2425	/// Every ref is stored whatever this says; only the events are capped.
2426	fn announced_refs<'a>(pushed: &'a [git_http::Pushed], default_branch: &str) -> Vec<&'a git_http::Pushed> {
2427	let (branches, tags): (Vec<&git_http::Pushed>, Vec<&git_http::Pushed>) =
2428	pushed.iter().partition(\|pushed\| pushed.branch().is_some());
2429	let mut announced = if branches.len() > MAX_PUSH_BRANCH_EVENTS {
2430	branches.into_iter().filter(\|pushed\| pushed.branch() == Some(default_branch)).collect()
2431	} else {
2432	branches
2433	};
2434	if tags.len() <= MAX_PUSH_TAG_EVENTS {
2435	announced.extend(tags);
2436	}
2437	announced
2438	}
2439
2440	/// A push the store accepted, to be recorded once git has its answer.
2441	struct PushDone {
2442	repo: Repo,
2443	pushed: Vec<git_http::Pushed>,
2444	/// The branches and tags it asked to delete, each with the commit it
2445	/// pointed to.
2446	deleted: Vec<(String, String)>,
2447	pack_bytes: u64,
2448	actor: Option<String>,
2449	/// Too large to scan for secrets before it was stored.
2450	unscanned: bool,
2451	/// The run whose job's token pushed, if one did: its push starts no
2452	/// workflows.
2453	caused_by_job: Option<String>,
2454	/// The person credited with the push's commits on their contribution
2455	/// calendar: whoever pushed, when that is a person (push_commits.rs).
2456	credit: Option<String>,
2457	}
2458
2459	/// What a git request leaves for after its answer: its audit entry, with
2460	/// how the request ended, and what a push changed.
2461	struct AfterGit {
2462	audit: Option<Box<g1t_contracts::audit::NewAuditEntry>>,
2463	status: u16,
2464	message: Option<String>,
2465	push: Option<PushDone>,
2466	}
2467
2468	impl AfterGit {
2469	fn ended(&mut self, status: u16, message: Option<String>) {
2470	self.status = status;
2471	self.message = message;
2472	}
2473
2474	/// Does the work once the response is on its way. A failure is logged:
2475	/// git has already been told how its request went.
2476	fn spawn(self, env: &Env, ctx: &Context) {
2477	if self.audit.is_none() && self.push.is_none() {
2478	return;
2479	}
2480	let env = env.clone();
2481	ctx.wait_until(async move {
2482	let repos = match service(&env) {
2483	Ok(repos) => repos,
2484	Err(error) => {
2485	worker::console_error!("git request not recorded: {error}");
2486	return;
2487	}
2488	};
2489	repos.finish_git(self.audit, self.status, self.message).await;
2490	if let Some(push) = self.push
2491	&& let Err(error) = repos.record_push(push).await
2492	{
2493	worker::console_error!("push not recorded: {error}");
2494	}
2495	repos.deferred.settle().await;
2496	});
2497	}
2498	}
2499
2500	fn service(env: &Env) -> Result<Repos<ArtifactsStore>> {
2501	let shared = shared::Shared::from_env(env).map(Rc::new);
2502	let deferred = Rc::new(store::Deferred::default());
2503	Ok(Repos {
2504	registry: Registry { db: env.d1("DB")? },
2505	store: ArtifactsStore::new(env, shared.clone(), deferred.clone())?,
2506	deferred,
2507	shared,
2508	packs: pack_cache::Packs::from_env(env).map(Rc::new),
2509	events: env.service("EVENTS")?,
2510	security: env.service("SECURITY").ok(),
2511	billing: env.service("BILLING").ok(),
2512	identity: env.service("IDENTITY").ok(),
2513	work: env.service("WORK").ok(),
2514	free_private_bytes: git_ops::free_private_bytes(env),
2515	fork_days: forks::retention_days(env),
2516	repo_limit: env
2517	.var("REPO_STORAGE_LIMIT_BYTES")
2518	.ok()
2519	.and_then(\|value\| value.to_string().parse().ok())
2520	.unwrap_or(pack_limits::DEFAULT_REPO_LIMIT_BYTES),
2521	large_pushes: git_http::LargePushes::from_var(env.var("LARGE_PUSHES").ok().map(\|value\| value.to_string()).as_deref()),
2522	placement: shards::Placement::from_vars(
2523	env.var("GITSTORE_NEW_REPOS").ok().map(\|value\| value.to_string()).as_deref(),
2524	env.var("GITSTORE_EU_NAMESPACE").ok().map(\|value\| value.to_string()).as_deref(),
2525	),
2526	limits: shards::limits(env.var("GITSTORE_NAMESPACE_LIMITS").ok().map(\|value\| value.to_string()).as_deref()),
2527	})
2528	}
2529
2530	/// Writes what this isolate metered once the answer has gone back, every
2531	/// few seconds at most: now, or once it is due, waiting in this request's
2532	/// `wait_until` so nothing counted is left for a request that may never
2533	/// come (meters.rs).
2534	fn flush_later(env: &Env, ctx: &Context) {
2535	let Some(wait) = meters::plan_flush() else {
2536	return;
2537	};
2538	if let Ok(db) = env.d1("DB") {
2539	ctx.wait_until(async move { meters::flush_after(&db, wait).await });
2540	}
2541	}
2542
2543	/// `/backups/<job id>/parts/<number>`: the job and the part's number.
2544	fn backup_part_path(path: &str) -> Option<(String, u16)> {
2545	let rest = path.strip_prefix("/backups/")?;
2546	let (job, number) = rest.split_once("/parts/")?;
2547	let number = number.parse::<u16>().ok()?;
2548	(!job.is_empty() && !job.contains('/')).then(\|\| (job.to_owned(), number))
2549	}
2550
2551	fn backups_off<T>() -> Outcome<T> {
2552	Outcome::fail(FailureCode::Conflict, "Backups are off on this installation: it has no storage for them.")
2553	}
2554
2555	/// One part of a backup's bundle, with the job's token in its header.
2556	async fn backup_part(request: &mut Request, env: &Env, repos: &Repos<ArtifactsStore>, job_id: String, number: u16) -> Result<Response> {
2557	let Some(blobs) = backups::storage(env) else {
2558	return reply(&backups_off::<()>());
2559	};
2560	let token = request.headers().get(g1t_contracts::backups::TOKEN_HEADER)?.unwrap_or_default();
2561	let bytes = request.bytes().await?;
2562	let job = g1t_contracts::backups::BackupJobArgs { job_id, token };
2563	reply(&backups::part(&repos.registry.db, &blobs, &job, number, bytes).await?)
2564	}
2565
2566	#[cfg(test)]
2567	mod backup_path_tests {
2568	use super::backup_part_path;
2569
2570	#[test]
2571	fn a_part_is_named_by_its_job_and_number() {
2572	assert_eq!(backup_part_path("/backups/bkp_1/parts/3"), Some(("bkp_1".to_owned(), 3)));
2573	assert_eq!(backup_part_path("/backups/bkp_1/parts/x"), None);
2574	assert_eq!(backup_part_path("/backups//parts/1"), None);
2575	assert_eq!(backup_part_path("/acme/rocket.git/info/refs"), None);
2576	}
2577	}
2578
2579	/// Read methods whose answer is an `Outcome`: when the git store is busy,
2580	/// the site is told so in words instead of failing the page.
2581	const OUTCOME_READS: [&str; 7] = ["tree", "blob", "log", "branches", "blame", "compare", "branch_drift"];
2582
2583	#[event(fetch)]
2584	async fn fetch(mut request: Request, env: Env, ctx: Context) -> Result<Response> {
2585	let mut repos = service(&env)?;
2586	// A part of a backup's bundle, as the API passes it on from the
2587	// sandbox: bytes, not JSON (backups.rs).
2588	if request.method() == Method::Put
2589	&& let Some((job_id, number)) = backup_part_path(&request.path())
2590	{
2591	let answered = backup_part(&mut request, &env, &repos, job_id, number).await;
2592	repos.deferred.hand_over(&ctx);
2593	flush_later(&env, &ctx);
2594	return answered;
2595	}
2596	let Some(method) = rpc_method(&request) else {
2597	let answered = repos.git_http(request, &env, &ctx).await;
2598	repos.deferred.hand_over(&ctx);
2599	flush_later(&env, &ctx);
2600	return answered;
2601	};
2602	// A replica near the caller when it asks for one (crates/kit/src/d1.rs).
2603	// Git over HTTPS above always reads the primary.
2604	let (db, served) = g1t_kit::d1::open(&env, "DB", &request)?;
2605	repos.registry.db = db;
2606	let body: serde_json::Value = request.json().await?;
2607
2608	let answered = async { match method.as_str() {
2609	"get" => reply(&repos.get(args(body)?).await?),
2610	"get_by_id" => reply(&repos.get_by_id(args(body)?).await?),
2611	"readable" => {
2612	let a: ReadableArgs = args(body)?;
2613	reply(&repos.registry.readable(&a.ids, &a.viewer).await?)
2614	}
2615	// Work's contribution calendar: the commits a person pushed each
2616	// day, in repositories the viewer may read (push_commits.rs).
2617	"commit_days" => {
2618	let a: g1t_contracts::repos::CommitDaysArgs = args(body)?;
2619	reply(&push_commits::days(&repos.registry, &a.user_id, &a.since, &a.viewer).await?)
2620	}
2621	"public_namespaces" => {
2622	let a: PublicNamespacesArgs = args(body)?;
2623	reply(&repos.registry.public_namespaces(&a.owner_id).await?)
2624	}
2625	"path_by_id" => {
2626	let a: PathByIdArgs = args(body)?;
2627	reply(
2628	&repos
2629	.registry
2630	.by_id(&a.id)
2631	.await?
2632	.filter(\|repo\| repo.fork_of.is_none())
2633	.map(\|repo\| RepoPath {
2634	namespace: repo.namespace,
2635	name: repo.name,
2636	}),
2637	)
2638	}
2639	"list" => {
2640	let a: ListArgs = args(body)?;
2641	reply(
2642	&repos
2643	.registry
2644	.list(
2645	&a.viewer,
2646	a.query.as_deref(),
2647	a.namespace.as_deref(),
2648	a.member_only,
2649	)
2650	.await?,
2651	)
2652	}
2653	"create" => reply(&repos.create(args(body)?).await?),
2654	"transfer" => reply(&repos.transfer(args(body)?).await?),
2655	// A repository's lifecycle: see lifecycle.rs.
2656	"delete" => reply(&repos.delete(args(body)?).await?),
2657	"deleted" => reply(&repos.deleted(args(body)?).await?),
2658	"restore" => reply(&repos.restore(args(body)?).await?),
2659	"purge" => reply(&repos.purge(args(body)?).await?),
2660	"purge_due" => reply(&repos.purge_due(args(body)?).await?),
2661	"rename" => reply(&repos.rename(args(body)?).await?),
2662	"archive" => reply(&repos.archive(args(body)?).await?),
2663	"set_visibility" => reply(&repos.set_visibility(args(body)?).await?),
2664	"set_default_branch" => reply(&repos.set_default_branch(args(body)?).await?),
2665	"rename_branch" => reply(&repos.rename_branch(args(body)?).await?),
2666	"create_branch" => reply(&repos.create_branch(args(body)?).await?),
2667	"remove_branch" => reply(&repos.remove_branch(args(body)?).await?),
2668	"resolve_branch" => reply(&repos.resolve_branch(args(body)?).await?),
2669	"status_by_id" => reply(&repos.status_by_id(args(body)?).await?),
2670	"resolve_path" => {
2671	let a: ResolvePathArgs = args(body)?;
2672	reply(&repos.registry.resolve_moved(&a.path).await?)
2673	}
2674	"namespace_count" => {
2675	let a: NamespaceCountArgs = args(body)?;
2676	reply(&repos.registry.count_in(&a.namespace).await?)
2677	}
2678	"update" => reply(&repos.update(args(body)?).await?),
2679	"tree" => reply(&repos.tree(args(body)?).await?),
2680	"blob" => reply(&repos.blob(args(body)?).await?),
2681	"log" => reply(&repos.log(args(body)?).await?),
2682	"blame" => reply(&repos.blame(args(body)?).await?),
2683	"fork_for_pull" => reply(&repos.fork_for_pull(args(body)?).await?),
2684	"git_access" => reply(&repos.git_access(args(body)?).await?),
2685	"branches" => reply(&repos.branches(args(body)?).await?),
2686	"last_commits" => reply(&repos.last_commits(args(body)?).await?),
2687	"path_log" => reply(&repos.path_log(args(body)?).await?),
2688	"branch_drift" => reply(&repos.branch_drift(args(body)?).await?),
2689	"tags" => reply(&repos.tags(args(body)?).await?),
2690	// The About: what the Files page shows beside the files (about.rs).
2691	// What is kept behind the head is worked out again after the answer.
2692	"about" => {
2693	let (answer, refresh) = repos.about(args(body)?).await?;
2694	about::refresh_later(&env, &ctx, refresh);
2695	reply(&answer)
2696	}
2697	"languages" => {
2698	let (answer, refresh) = repos.languages(args(body)?).await?;
2699	about::refresh_later(&env, &ctx, refresh);
2700	reply(&answer)
2701	}
2702	"contributors" => {
2703	let (answer, refresh) = repos.contributors(args(body)?).await?;
2704	about::refresh_later(&env, &ctx, refresh);
2705	reply(&answer)
2706	}
2707	"license" => {
2708	let (answer, refresh) = repos.license(args(body)?).await?;
2709	about::refresh_later(&env, &ctx, refresh);
2710	reply(&answer)
2711	}
2712	"stars" => reply(&repos.stars(args(body)?).await?),
2713	"star" => reply(&repos.star(args(body)?).await?),
2714	"stargazers" => reply(&repos.stargazers(args(body)?).await?),
2715	"starred" => reply(&repos.starred(args(body)?).await?),
2716	"releases" => reply(&repos.releases(args(body)?).await?),
2717	"release" => reply(&repos.release(args(body)?).await?),
2718	"create_release" => reply(&repos.create_release(args(body)?).await?),
2719	"update_release" => reply(&repos.update_release(args(body)?).await?),
2720	"delete_release" => reply(&repos.delete_release(args(body)?).await?),
2721	"head" => reply(&repos.head(args(body)?).await?),
2722	"behind" => reply(&repos.behind(args(body)?).await?),
2723	"divergence" => reply(&repos.divergence(args(body)?).await?),
2724	"land" => reply(&repos.land(args(body)?).await?),
2725	"update_pull_branch" => reply(&repos.update_pull_branch(args(body)?).await?),
2726	// Services only: a pull request's test merge commit (test_merge.rs).
2727	"pull_merge_ref" => reply(&repos.pull_merge_ref(args(body)?).await?),
2728	"delete_branch" => reply(&repos.delete_branch(args(body)?).await?),
2729	"commit_file" => reply(&repos.commit_file(args(body)?).await?),
2730	"compare" => reply(&repos.compare(args(body)?).await?),
2731	// Services only: a pull request's commits, as rules look at them (rules.rs).
2732	"inspect_commits" => reply(&repos.inspect_commits(args(body)?).await?),
2733	"scan_history" => reply(&repos.scan_history(args(body)?).await?),
2734	"find_lockfiles" => reply(&repos.find_lockfiles(args(body)?).await?),
2735	"match_pattern" => reply(&repos.match_pattern(args(body)?).await?),
2736	"check_secret" => reply(&repos.check_secret(args(body)?).await?),
2737	"list_files" => reply(&repos.list_files(args(body)?).await?),
2738	// The API's tarball and zipball links, and the site opening them (archive_links.rs).
2739	"archive_link" => {
2740	let sealer = env.secret("REPOS_KEY").ok().and_then(\|key\| g1t_secrets::Sealer::new(&key.to_string()));
2741	reply(&repos.archive_link(args(body)?, sealer.as_ref()).await?)
2742	}
2743	"open_archive_token" => {
2744	let sealer = env.secret("REPOS_KEY").ok().and_then(\|key\| g1t_secrets::Sealer::new(&key.to_string()));
2745	reply(&repos.open_archive_token(args(body)?, sealer.as_ref()).await?)
2746	}
2747	"changed_files" => reply(&repos.changed_files(args(body)?).await?),
2748	"read_blobs" => reply(&repos.read_blobs(args(body)?).await?),
2749	// Services only: what the Composer registry builds packages from.
2750	"refs" => reply(&repos.refs_of(args(body)?).await?),
2751	"raw_file" => reply(&repos.raw_file(args(body)?).await?),
2752	"raw_blobs" => reply(&repos.raw_blobs(args(body)?).await?),
2753	"visibility" => {
2754	let a: g1t_contracts::repos::VisibilityArgs = args(body)?;
2755	reply(&repos.registry.visibility(&a.paths).await?)
2756	}
2757	"storage" => reply(&repos.registry.storage().await?),
2758	"git_operations" => {
2759	let a: GitOperationsArgs = args(body)?;
2760	reply(&git_ops::totals(&repos.registry.db, &a.month, a.since.as_deref(), a.namespace.as_deref().map(str::to_lowercase).as_deref()).await?)
2761	}
2762	// Identity, once: who created each repository (members.rs there).
2763	"repo_creators" => {
2764	let a: AllIdsArgs = args(body)?;
2765	let limit = a.limit.clamp(1, 500);
2766	let repos = repos.registry.creators_after(a.after.as_deref(), limit).await?;
2767	let next = (repos.len() == limit as usize).then(\|\| repos.last().map(\|repo\| repo.id.clone())).flatten();
2768	reply(&CreatorPage { repos, next })
2769	}
2770	"all_ids" => {
2771	let a: AllIdsArgs = args(body)?;
2772	let limit = a.limit.clamp(1, 500);
2773	let ids = repos.registry.ids_after(a.after.as_deref(), limit).await?;
2774	let next = (ids.len() == limit as usize).then(\|\| ids.last().cloned()).flatten();
2775	reply(&IdPage { ids, next })
2776	}
2777	// The raw meters of the git store, for reconciling with Cloudflare
2778	// (meters.rs, scripts/ops/gitstore-usage.mjs).
2779	"artifacts_usage" => {
2780	let a: meters::UsageArgs = args(body)?;
2781	reply(&meters::usage(&repos.registry.db, &a).await?)
2782	}
2783	"operation_mapping" => reply(&meters::read_mapping(&repos.registry.db).await?),
2784	// Billing: the workspace each pull request's working copy is counted
2785	// for, so Cloudflare's own count of `pulls--<id>` shares out too.
2786	"pull_owners" => {
2787	#[derive(serde::Deserialize)]
2788	struct PullOwnersArgs {
2789	pulls: Vec<String>,
2790	}
2791	let a: PullOwnersArgs = args(body)?;
2792	let pulls: Vec<String> = a.pulls.into_iter().take(500).collect();
2793	reply(&serde_json::json!({ "owners": meters::pull_owners(&repos.registry.db, &pulls).await? }))
2794	}
2795	// Services only: which meters are operations, changed without a deploy.
2796	"set_operation_mapping" => {
2797	let row: meters::MappingRow = args(body)?;
2798	meters::set_mapping(&repos.registry.db, &row, &rfc3339(now_ms())).await?;
2799	reply(&meters::read_mapping(&repos.registry.db).await?)
2800	}
2801	// Backups (backups.rs): the runner's sweep claims queued ones, and
2802	// each sandbox, through the API, asks for its job and says how it went.
2803	"claim_backups" => {
2804	let a: g1t_contracts::backups::ClaimBackupsArgs = args(body)?;
2805	let blobs = backups::storage(&env);
2806	reply(&backups::claim(&repos.registry.db, blobs.as_ref(), &a, now_ms()).await?)
2807	}
2808	"backup_spec" => match backups::storage(&env) {
2809	Some(blobs) => {
2810	let a: g1t_contracts::backups::BackupJobArgs = args(body)?;
2811	let every = backups::Settings::from_env(&env).full_every;
2812	reply(&backups::spec(&repos.registry, &blobs, &repos.store, &a, every, now_ms()).await?)
2813	}
2814	None => reply(&backups_off::<bool>()),
2815	},
2816	"backup_complete" => match backups::storage(&env) {
2817	Some(blobs) => reply(&backups::complete(&repos.registry, &blobs, &args(body)?, now_ms()).await?),
2818	None => reply(&backups_off::<bool>()),
2819	},
2820	"backup_fail" => match backups::storage(&env) {
2821	Some(blobs) => reply(&backups::fail(&repos.registry.db, &blobs, &args(body)?).await?),
2822	None => reply(&backups_off::<bool>()),
2823	},
2824	// How the git store has been answering, for the status page.
2825	"store_health" => {
2826	let a: meters::HealthArgs = args(body)?;
2827	reply(&meters::health(&repos.registry.db, &a).await?)
2828	}
2829	// Where repositories may be kept, for a workspace's settings.
2830	"storage_options" => reply(&repos.storage_options()),
2831	// Services and operators only: how each namespace stands, and
2832	// moving a repository between them (namespaces.rs, moves.rs).
2833	"namespaces" => reply(&repos.standings().await?),
2834	"move_repository" => reply(&repos.move_repository(args(body)?).await?),
2835	"repository_moves" => {
2836	let a: moves::ListMovesArgs = args(body)?;
2837	reply(&repos.registry.moves(a.limit.unwrap_or(50)).await?)
2838	}
2839	_ => Response::error("Unknown method", 404),
2840	} }
2841	.await;
2842	// The git store is busy: said in words, with when to try again.
2843	let answered = match answered {
2844	Err(error) => match resilience::busy(&error.to_string()) {
2845	Some(busy) if OUTCOME_READS.contains(&method.as_str()) => {
2846	reply(&Outcome::<()>::fail(FailureCode::Conflict, busy.message().trim()))
2847	}
2848	Some(busy) => {
2849	let response = Response::error(busy.message(), 503)?;
2850	response.headers().set("retry-after", &busy.retry_after.to_string())?;
2851	Ok(response)
2852	}
2853	None => Err(error),
2854	},
2855	answered => answered,
2856	};
2857	repos.deferred.hand_over(&ctx);
2858	flush_later(&env, &ctx);
2859	served.finish(answered)
2860	}
2861
2862	/// The nightly cron in wrangler.jsonc: tonight's backups are queued.
2863	const BACKUP_CRON: &str = "53 2 * * *";
2864
2865	/// The hourly sweep: deleted repositories whose time to be restored has
2866	/// passed are purged. See lifecycle.rs. And, at [`BACKUP_CRON`], the
2867	/// repositories whose refs moved are queued for a backup (backups.rs).
2868	#[event(scheduled)]
2869	async fn scheduled(event: ScheduledEvent, env: Env, _ctx: ScheduleContext) {
2870	let repos = match service(&env) {
2871	Ok(repos) => repos,
2872	Err(error) => {
2873	worker::console_error!("repos: the sweep could not start: {error}");
2874	return;
2875	}
2876	};
2877	if event.cron() == BACKUP_CRON {
2878	let Some(blobs) = backups::storage(&env) else { return };
2879	match backups::nightly(&repos.registry.db, &blobs, backups::Settings::from_env(&env), now_ms()).await {
2880	Ok(night) => worker::console_log!("repos: queued {} backups, removed {} of purged repositories", night.queued, night.pruned),
2881	Err(error) => worker::console_error!("repos: backups could not be queued: {error}"),
2882	}
2883	return;
2884	}
2885	match repos.purge_due(PurgeDueArgs::default()).await {
2886	Ok(0) => {}
2887	Ok(count) => worker::console_log!("repos: purged {count} deleted repositories"),
2888	Err(error) => worker::console_error!("repos: the purge sweep failed: {error}"),
2889	}
2890	// Pull requests' working copies whose time has come (forks.rs).
2891	match repos.retire_due().await {
2892	Ok(0) => {}
2893	Ok(count) => worker::console_log!("repos: removed {count} pull request working copies"),
2894	Err(error) => worker::console_error!("repos: the working copy sweep failed: {error}"),
2895	}
2896	// Repositories moving between namespaces, and old copies (moves.rs).
2897	match repos.run_moves().await {
2898	Ok(0) => {}
2899	Ok(count) => worker::console_log!("repos: moved {count} repositories between namespaces"),
2900	Err(error) => worker::console_error!("repos: the move sweep failed: {error}"),
2901	}
2902	meters::flush(&repos.registry.db).await;
2903	}
2904
2905	/// Events from the bus. A workspace's rename: its repositories move to the
2906	/// workspace's current slug, asked of identity by id, so a repeated or late
2907	/// delivery lands in the same place; their git store keys stay as they
2908	/// were. A workspace's deletion: its repositories are deleted with it,
2909	/// restored with it, or purged with it.
2910	#[event(queue)]
2911	async fn queue(batch: MessageBatch<Event>, env: Env, ctx: Context) -> Result<()> {
2912	let registry = Registry { db: env.d1("DB")? };
2913	let identity = env.service("IDENTITY")?;
2914	let handled = handle_events(&batch, &env, &registry, &identity).await;
2915	flush_later(&env, &ctx);
2916	handled
2917	}
2918
2919	/// Each event is acknowledged or retried on its own, so one that fails is
2920	/// tried again without the others before and after it running twice.
2921	async fn handle_events(batch: &MessageBatch<Event>, env: &Env, registry: &Registry, identity: &Fetcher) -> Result<()> {
2922	for message in batch.messages()? {
2923	match handle_event(message.body(), env, registry, identity).await {
2924	Ok(()) => message.ack(),
2925	Err(error) => {
2926	worker::console_error!("repos: event {} failed: {error}", message.body().id);
2927	message.retry();
2928	}
2929	}
2930	}
2931	Ok(())
2932	}
2933
2934	async fn handle_event(event: &Event, env: &Env, registry: &Registry, identity: &Fetcher) -> Result<()> {
2935	// A pull request merged, closed or reopened: its working copy is
2936	// kept or let go (forks.rs).
2937	if let Some(change) = forks::pull_change(&event.kind) {
2938	let Some(pull_id) = forks::pull_id_of(&event.data) else {
2939	worker::console_error!("{} {} names no pull request", event.kind, event.id);
2940	return Ok(());
2941	};
2942	let repos = service(env)?;
2943	match change {
2944	forks::PullChange::Settled => repos.pull_settled(&pull_id).await?,
2945	forks::PullChange::Reopened => repos.pull_reopened(&pull_id).await?,
2946	}
2947	return Ok(());
2948	}
2949	// A workspace deleted, restored or purged: its repositories go with
2950	// it, come back with it, or are purged with it (lifecycle.rs).
2951	if event.kind == "workspace.deleting" {
2952	match serde_json::from_value::<WorkspaceDeleting>(event.data.clone()) {
2953	Ok(deleting) => service(env)?.delete_with_workspace(&deleting, &protected_workspaces(env)).await?,
2954	Err(_) => worker::console_error!("workspace.deleting {} could not be read", event.id),
2955	}
2956	return Ok(());
2957	}
2958	if event.kind == "workspace.restored" {
2959	match serde_json::from_value::<WorkspaceRestored>(event.data.clone()) {
2960	Ok(restored) => service(env)?.restore_with_workspace(&restored).await?,
2961	Err(_) => worker::console_error!("workspace.restored {} could not be read", event.id),
2962	}
2963	return Ok(());
2964	}
2965	if event.kind == "workspace.deleted" {
2966	match serde_json::from_value::<WorkspaceDeleted>(event.data.clone()) {
2967	Ok(deleted) => service(env)?.purge_workspace(&deleted, &protected_workspaces(env)).await?,
2968	Err(_) => worker::console_error!("workspace.deleted {} could not be read", event.id),
2969	}
2970	return Ok(());
2971	}
2972	// An account deleted or restored: kept contributors that name it
2973	// (or ghost, for a restore) are worked out again, so it shows as
2974	// ghost for its 30 days, and as itself again if restored.
2975	if let Some(needle) = stats::shown_differently(&event.kind, &event.data) {
2976	let db = env.d1("DB")?;
2977	if let Err(error) = stats::rework_naming(&db, &needle).await {
2978	worker::console_error!("{} {}: contributors not marked to be counted again: {error}", event.kind, event.id);
2979	}
2980	return Ok(());
2981	}
2982	if event.kind != "workspace.renamed" {
2983	return Ok(());
2984	}
2985	let Ok(renamed) = serde_json::from_value::<WorkspaceRenamed>(event.data.clone()) else {
2986	worker::console_error!("workspace.renamed {} could not be read", event.id);
2987	return Ok(());
2988	};
2989	let names: HashMap<String, String> = g1t_kit::call(
2990	identity,
2991	"usernames",
2992	&g1t_contracts::identity::UsernamesArgs {
2993	ids: vec![renamed.workspace_id.clone()],
2994	},
2995	)
2996	.await?;
2997	let current = names
2998	.get(&renamed.workspace_id)
2999	.cloned()
3000	.unwrap_or_else(\|\| renamed.to.clone());
3001	let left = registry
3002	.rename_namespace(&renamed.stale_slugs(&current), &current)
3003	.await?;
3004	if left > 0 {
3005	worker::console_error!(
3006	"{left} repositories stayed under {} or {}: {current} already has repositories of the same names",
3007	renamed.from,
3008	renamed.to
3009	);
3010	}
3011	Ok(())
3012	}
3013
3014	/// The workspaces whose repositories never go with a deletion, whatever is
3015	/// published: `PROTECTED_WORKSPACES` if set here, and Flagon's always.
3016	fn protected_workspaces(env: &Env) -> Vec<String> {
3017	let configured = env.var("PROTECTED_WORKSPACES").ok().map(\|v\| v.to_string());
3018	g1t_contracts::identity::protected_names(configured.as_deref())
3019	}
3020
3021	/// What a token's own limits say about git on the repository `repo`
3022	/// (`owner/name`), before anyone's role is asked: why it is refused, or
3023	/// `None`. `source` is the repository a pull request's working copy at
3024	/// `repo` belongs to, which a workflow job's token reaches too. `public`
3025	/// is whether anyone may read it, and `exists` whether there is one.
3026	///
3027	/// A job's token and a deploy key reach their own repository only. Its
3028	/// scopes decide the rest: `code:read` to read a private repository,
3029	/// `code:write` to push, which a read-only deploy key never has. A deploy
3030	/// key never makes a repository by pushing to an empty address.
3031	pub(crate) fn git_token_refusal(
3032	access: &g1t_contracts::scopes::TokenAccess,
3033	repo: &str,
3034	source: Option<&str>,
3035	write: bool,
3036	public: bool,
3037	exists: bool,
3038	) -> Option<String> {
3039	if let Some(refused) = g1t_contracts::scopes::decide_repo(access, repo)
3040	&& !source.is_some_and(\|source\| access.reaches(source))
3041	{
3042	return Some(refused.reason.unwrap_or_default());
3043	}
3044	let decision = g1t_contracts::scopes::decide_git(access, write, public);
3045	if !decision.allowed {
3046	return Some(decision.reason.unwrap_or_default());
3047	}
3048	if access.deploy_key.is_some() && !exists {
3049	return Some(format!("This deploy key is for {repo}, which is not there any more."));
3050	}
3051	None
3052	}
3053
3054	/// The repository a push to a path that does not exist yet creates: private,
3055	/// so nothing pushed by mistake is published. An owner makes it public on
3056	/// purpose (`POST /repos/{owner}/{repo}/visibility`).
3057	fn push_to_create(owner: &User, path: &RepoPath) -> CreateArgs {
3058	CreateArgs {
3059	owner: owner.clone(),
3060	namespace: path.namespace.clone(),
3061	name: path.name.clone(),
3062	description: None,
3063	is_private: true,
3064	import_url: None,
3065	import_token: None,
3066	}
3067	}
3068
3069	/// Whether `delete_branch` may remove `branch`: one of g1t's own
3070	/// (`g1t-…`), or one whose tip the caller names, such as a dependency
3071	/// update's branch after its pull request closed.
3072	fn deletable_branch(branch: &str, head: Option<&str>) -> bool {
3073	!branch.is_empty() && (branch.starts_with(G1T_BRANCH_PREFIX) \|\| head.is_some_and(\|head\| !head.is_empty()))
3074	}
3075
3076	#[cfg(test)]
3077	mod delete_branch_tests {
3078	use super::deletable_branch;
3079
3080	#[test]
3081	fn only_g1t_branches_or_a_named_tip_are_deleted() {
3082	assert!(deletable_branch("g1t-queue-12", None));
3083	assert!(!deletable_branch("g1t/security/sharp-0.35.5", None));
3084	assert!(deletable_branch("g1t/security/sharp-0.35.5", Some("abc123")));
3085	assert!(!deletable_branch("feature", Some("")));
3086	assert!(!deletable_branch("", Some("abc123")));
3087	}
3088	}
3089
3090	#[cfg(test)]
3091	mod announced_refs_tests {
3092	use super::*;
3093
3094	fn pushed(git_ref: String) -> git_http::Pushed {
3095	git_http::Pushed { git_ref, before: None, after: "abc".into() }
3096	}
3097
3098	fn refs(announced: Vec<&git_http::Pushed>) -> Vec<&str> {
3099	announced.into_iter().map(\|pushed\| pushed.git_ref.as_str()).collect()
3100	}
3101
3102	#[test]
3103	fn a_few_tags_are_announced_and_many_are_not() {
3104	let few: Vec<_> = (1..=3).map(\|n\| pushed(format!("refs/tags/v{n}"))).chain([pushed("refs/heads/main".into())]).collect();
3105	assert_eq!(refs(announced_refs(&few, "main")), ["refs/heads/main", "refs/tags/v1", "refs/tags/v2", "refs/tags/v3"]);
3106	let many: Vec<_> = (1..=10_000).map(\|n\| pushed(format!("refs/tags/v{n}"))).chain([pushed("refs/heads/main".into())]).collect();
3107	assert_eq!(refs(announced_refs(&many, "main")), ["refs/heads/main"]);
3108	}
3109
3110	#[test]
3111	fn past_the_branch_cap_only_the_default_branch_is_announced() {
3112	let at_cap: Vec<_> = (0..MAX_PUSH_BRANCH_EVENTS).map(\|n\| pushed(format!("refs/heads/b{n}"))).collect();
3113	assert_eq!(announced_refs(&at_cap, "main").len(), MAX_PUSH_BRANCH_EVENTS);
3114	let over: Vec<_> = (0..=MAX_PUSH_BRANCH_EVENTS)
3115	.map(\|n\| pushed(format!("refs/heads/b{n}")))
3116	.chain([pushed("refs/heads/main".into())])
3117	.collect();
3118	assert_eq!(refs(announced_refs(&over, "main")), ["refs/heads/main"]);
3119	assert!(announced_refs(&over, "trunk").is_empty());
3120	}
3121	}
3122
3123	#[cfg(test)]
3124	mod push_to_create_tests {
3125	use super::*;
3126
3127	#[test]
3128	fn a_pushed_repository_starts_private() {
3129	let owner: User = serde_json::from_value(serde_json::json!({ "id": "usr_1", "username": "ada" })).unwrap();
3130	let args = push_to_create(&owner, &RepoPath { namespace: "acme".into(), name: "site".into() });
3131	assert!(args.is_private);
3132	assert_eq!((args.namespace.as_str(), args.name.as_str()), ("acme", "site"));
3133	}
3134	}
3135
3136	#[cfg(test)]
3137	mod deploy_key_git_tests {
3138	use super::*;
3139	use g1t_contracts::deploy_keys;
3140
3141	fn key(read_only: bool) -> User {
3142	deploy_keys::principal("wsp_acme", "acme", deploy_keys::access("dk_1", "CI", "acme/rocket", read_only))
3143	}
3144
3145	fn rocket(private: bool) -> Repo {
3146	serde_json::from_value(serde_json::json!({
3147	"id": "rep_rocket",
3148	"namespace": "acme",
3149	"name": "rocket",
3150	"description": null,
3151	"isPrivate": private,
3152	"ownerId": "usr_owner",
3153	"defaultBranch": "main",
3154	"forkOf": null,
3155	"protected": false,
3156	"createdAt": "",
3157	}))
3158	.unwrap()
3159	}
3160
3161	fn refusal(user: &User, repo: &str, write: bool, exists: bool) -> Option<String> {
3162	git_token_refusal(user.token.as_deref().unwrap(), repo, None, write, false, exists)
3163	}
3164
3165	#[test]
3166	fn a_read_only_deploy_key_clones_its_repository_and_never_pushes() {
3167	let user = key(true);
3168	assert_eq!(refusal(&user, "acme/rocket", false, true), None);
3169	assert!(refusal(&user, "acme/rocket", true, true).unwrap().contains("read-only"));
3170	// Its role is a workspace token's: it reads a private repository.
3171	assert!(registry::can_read(&rocket(true), &Some(user)));
3172	}
3173
3174	#[test]
3175	fn a_deploy_key_with_write_access_pushes_to_its_repository() {
3176	let user = key(false);
3177	assert_eq!(refusal(&user, "acme/rocket", true, true), None);
3178	assert!(registry::can_write(&rocket(true), &Some(user)));
3179	}
3180
3181	#[test]
3182	fn a_deploy_key_reaches_no_other_repository() {
3183	let user = key(false);
3184	for other in ["acme/booster", "other/rocket"] {
3185	for write in [false, true] {
3186	let why = refusal(&user, other, write, true).expect(other);
3187	assert!(why.contains("deploy key is for acme/rocket"), "{why}");
3188	}
3189	}
3190	// Not even a pull request's working copy of another repository.
3191	let token = user.token.as_deref().unwrap();
3192	assert!(git_token_refusal(token, "pulls/pr_1", Some("acme/booster"), false, false, true).is_some());
3193	}
3194
3195	#[test]
3196	fn a_deploy_key_never_creates_a_repository() {
3197	let why = refusal(&key(false), "acme/rocket", true, false).unwrap();
3198	assert!(why.contains("not there"), "{why}");
3199	}
3200	}