flagon-io/g1t

public

Where people and agents ship software together. The open-source git platform for the whole job: issues, agents, checks and deploys to the edge.

git agents cloudflare rust forge mcp

3,344 linesCode BlameRaw

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 merge_methods;
33	mod meters;
34	mod moves;
35	mod namespaces;
36	mod network;
37	mod pack_cache;
38	mod pack_limits;
39	mod push_checks;
40	mod push_commits;
41	mod refs;
42	mod refs_cache;
43	mod registry;
44	mod resilience;
45	mod rule_facts;
46	mod rules;
47	mod run_access;
48	mod secret_scan;
49	mod shards;
50	mod shared;
51	mod signatures;
52	mod stats;
53	mod store;
54	mod test_merge;
55	mod transfer;
56	mod workflow_gate;
57
58	use g1t_contracts::events::{
59	Event, GitDeleted, GitPush, NewEvent, Publish, RepoCreated, RepoForked, RepoUpdated, WorkspaceDeleted,
60	WorkspaceDeleting, WorkspaceRenamed, WorkspaceRestored,
61	};
62	use g1t_contracts::access::{self, Capability};
63	use g1t_contracts::repos::*;
64	use g1t_contracts::time::rfc3339;
65	use g1t_contracts::{FailureCode, Outcome, PrincipalKind, Role, User, Viewer, is_valid_repo_name, new_id};
66	use g1t_kit::{args, now_ms, reply, rpc_method};
67	use std::collections::{HashMap, HashSet, VecDeque};
68	use std::rc::Rc;
69
70	use serde::Serialize;
71	use worker::{
72	Context, Env, Fetcher, MessageBatch, MessageExt, Method, Request, Response, Result, ScheduleContext, ScheduledEvent,
73	event,
74	};
75
76	use registry::{Registry, can_read, can_write, store_key};
77	use store::{ArtifactsStore, GitRepo, GitStore, Scope};
78
79	/// Namespace that holds every pull request's fork: `pulls/<pull id>`.
80	pub(crate) const PULLS_NAMESPACE: &str = "pulls";
81	const MAX_TEXT_BYTES: usize = 512 * 1024;
82	/// How far back a pull request may have forked and still be landed.
83	const MAX_ANCESTRY: u32 = 1000;
84	/// The highest limit on how many branches and tags one push may change
85	/// (`UpdateArgs::max_ref_updates`).
86	const MAX_REF_UPDATES_LIMIT: u32 = 1000;
87	/// How many forks up a pull request's source is looked for in the fork
88	/// network of the repository it merges into.
89	const MAX_FORK_DEPTH: usize = 10;
90	/// The most files one `compare` asked for by path reads.
91	const MAX_COMPARED_PATHS: usize = 50;
92	/// The most tags a repository's Tags page reads and lists.
93	const MAX_TAGS_READ: usize = 100;
94	/// Branch heads measured in one `branch_drift` call.
95	const MAX_DRIFT_HEADS: usize = 100;
96	/// How long a `last_commits` walk asked without a budget runs before it
97	/// stops and keeps its progress for the next call. The site asks from a
98	/// waitUntil, which may run 30 s past its response.
99	const LAST_COMMITS_WALK_MS: u64 = 20_000;
100	/// How long a path's history may walk when the caller gives no budget.
101	const PATH_LOG_WALK_MS: u64 = 10_000;
102
103	/// One path segment, percent-encoded for a cache key.
104	fn urlencoding_segment(segment: &str) -> String {
105	segment
106	.bytes()
107	.map(\|b\| if b.is_ascii_alphanumeric() \|\| b"-._~".contains(&b) { (b as char).to_string() } else { format!("%{b:02X}") })
108	.collect()
109	}
110	const MAX_DESCRIPTION_CHARS: usize = 200;
111	pub(crate) const SOURCE: &str = "repos";
112	pub(crate) const UNVERIFIED: &str = "Confirm your email address first. Check your inbox, or resend the link from the banner on g1t.sh.";
113
114	pub(crate) fn not_found<T>() -> Outcome<T> {
115	Outcome::fail(FailureCode::NotFound, "Repository not found.")
116	}
117
118	/// Decoded text, or `None` when the file is too large or looks binary.
119	/// Whether a ref is a full commit hash rather than a branch name.
120	fn is_commit_hash(git_ref: &str) -> bool {
121	git_ref.len() == 40 && git_ref.bytes().all(\|b\| b.is_ascii_hexdigit())
122	}
123
124	fn text_of(bytes: Vec<u8>) -> Option<String> {
125	if bytes.len() > MAX_TEXT_BYTES \|\| bytes.contains(&0) {
126	return None;
127	}
128	Some(String::from_utf8_lossy(&bytes).into_owned())
129	}
130
131	fn is_readme(name: &str) -> bool {
132	matches!(
133	name.to_lowercase().as_str(),
134	"readme" \| "readme.md" \| "readme.markdown" \| "readme.txt"
135	)
136	}
137
138	/// Whether `ancestor` is reachable from the newest commit in `history`.
139	///
140	/// `history` is the first-parent chain, which is all the store lists; a fork
141	/// that merged the target branch in has the target's head on a second
142	/// parent, so the walk follows every parent.
143	async fn descends_from<R: GitRepo>(repo: &R, history: &[Commit], ancestor: &str) -> Result<bool> {
144	let known: HashMap<&str, &[String]> = history
145	.iter()
146	.map(\|commit\| (commit.hash.as_str(), commit.parents.as_slice()))
147	.collect();
148	let mut seen = HashSet::new();
149	let mut queue: Vec<String> = history
150	.first()
151	.map(\|c\| c.hash.clone())
152	.into_iter()
153	.collect();
154	while let Some(hash) = queue.pop() {
155	if hash == ancestor {
156	return Ok(true);
157	}
158	if !seen.insert(hash.clone()) \|\| seen.len() > MAX_ANCESTRY as usize {
159	continue;
160	}
161	match known.get(hash.as_str()) {
162	Some(parents) => queue.extend(parents.iter().cloned()),
163	None => queue.extend(repo.parents(&hash).await?.unwrap_or_default()),
164	}
165	}
166	Ok(false)
167	}
168
169	/// The commit closest to the newest in `history` that is also in `shared`:
170	/// where a fork and the repository it came from last agreed.
171	async fn nearest_ancestor_in<R: GitRepo>(
172	repo: &R,
173	history: &[Commit],
174	shared: &HashSet<String>,
175	) -> Result<Option<String>> {
176	let known: HashMap<&str, &[String]> = history
177	.iter()
178	.map(\|commit\| (commit.hash.as_str(), commit.parents.as_slice()))
179	.collect();
180	let mut seen = HashSet::new();
181	let mut queue: VecDeque<String> = history
182	.first()
183	.map(\|c\| c.hash.clone())
184	.into_iter()
185	.collect();
186	while let Some(hash) = queue.pop_front() {
187	if shared.contains(&hash) {
188	return Ok(Some(hash));
189	}
190	if !seen.insert(hash.clone()) \|\| seen.len() > MAX_ANCESTRY as usize {
191	continue;
192	}
193	match known.get(hash.as_str()) {
194	Some(parents) => queue.extend(parents.iter().cloned()),
195	None => queue.extend(repo.parents(&hash).await?.unwrap_or_default()),
196	}
197	}
198	Ok(None)
199	}
200
201	thread_local! {
202	/// Targets' sides of mergeability, by head (coalesce.rs).
203	static TARGETS: std::cell::RefCell<coalesce::Memo<coalesce::TargetKey, Rc<coalesce::TargetSide>>> =
204	std::cell::RefCell::new(coalesce::Memo::new(coalesce::TARGET_TTL_MS, 32));
205	/// What targets changed between two trees.
206	static THEIRS: std::cell::RefCell<coalesce::Memo<coalesce::TheirsKey, (Vec<String>, bool)>> =
207	std::cell::RefCell::new(coalesce::Memo::new(coalesce::THEIRS_TTL_MS, 256));
208	/// What repositories hold, as read for a push's first request, for the
209	/// same push's second: a push's POST does not wait on the database.
210	static HELD: std::cell::RefCell<coalesce::Memo<String, u64>> =
211	std::cell::RefCell::new(coalesce::Memo::new(60_000, 512));
212	}
213
214	pub(crate) struct Repos<S: GitStore> {
215	registry: Registry,
216	store: S,
217	events: Fetcher,
218	/// Asked during a push which secrets have been allowed.
219	security: Option<Fetcher>,
220	/// Asked whether a workspace is on a plan, for its private storage.
221	billing: Option<Fetcher>,
222	/// Says which rulesets hold for a change to a branch or tag, and keeps
223	/// how they judged it (rules.rs). `None` where it is not deployed: the
224	/// old protection flag then holds on push.
225	work: Option<Fetcher>,
226	/// Told when a repository moves, for the tokens of agents at work on it.
227	identity: Option<Fetcher>,
228	/// What a free workspace's private repositories may hold.
229	free_private_bytes: i64,
230	/// Days a pull request's working copy is kept after it settles (forks.rs).
231	pub(crate) fork_days: u64,
232	/// The most a repository may hold (pack_limits.rs), and what happens
233	/// to a push too large to scan.
234	repo_limit: u64,
235	large_pushes: git_http::LargePushes,
236	/// Which git store namespace new repositories go in (shards.rs), and
237	/// the most each should hold (`GITSTORE_NAMESPACE_LIMITS`).
238	placement: shards::Placement,
239	limits: HashMap<String, u64>,
240	/// What isolates share: answers that list refs (refs_cache.rs).
241	shared: Option<Rc<shared::Shared>>,
242	/// Packs for fresh clones (pack_cache.rs); `None` without the bucket.
243	packs: Option<Rc<pack_cache::Packs>>,
244	/// What this request started that need not hold up its answer
245	/// (store.rs), handed to its `waitUntil` once it has answered.
246	deferred: Rc<store::Deferred>,
247	}
248
249	impl<S: GitStore> Repos<S> {
250	/// Records that the refs of the repository with this id changed, once
251	/// they have, so that the answers kept that list them go stale (see
252	/// refs_cache.rs). Everything that changes a repository's refs calls
253	/// this after it (`every_ref_writer_records_the_change` checks). A
254	/// failure is logged: the change itself happened, and what was kept
255	/// expires within `refs_cache::TTL_SECONDS` regardless.
256	pub(crate) async fn refs_moved(&self, repo_id: &str) {
257	if let Err(error) = self.registry.refs_moved(repo_id).await {
258	worker::console_error!("refs of {repo_id} changed but not recorded: {error}");
259	}
260	}
261
262	pub(crate) async fn publish<T: Serialize>(&self, event: NewEvent<T>) -> Result<()> {
263	g1t_kit::call(
264	&self.events,
265	"publish",
266	&Publish {
267	events: vec![event],
268	},
269	)
270	.await
271	}
272
273	/// Whether the viewer may read `repo`. A pull request's fork of a
274	/// private repository can be read by everyone who can read that
275	/// repository, so its members can review and check out the change, as
276	/// well as by whoever opened the pull request.
277	async fn may_read(&self, repo: &Repo, viewer: &Viewer) -> Result<bool> {
278	if can_read(repo, viewer) {
279	return Ok(true);
280	}
281	let Some(source_id) = &repo.fork_of else {
282	return Ok(false);
283	};
284	Ok(self
285	.registry
286	.by_id(source_id)
287	.await?
288	.is_some_and(\|source\| can_read(&source, viewer)))
289	}
290
291	/// `repo`, if there is one and the viewer may read it.
292	async fn visible(&self, repo: Option<Repo>, viewer: &Viewer) -> Result<Option<Repo>> {
293	Ok(match repo {
294	Some(repo) if self.may_read(&repo, viewer).await? => Some(repo),
295	_ => None,
296	})
297	}
298
299	/// Resolves a repo the viewer may read; private repos look missing.
300	pub(crate) async fn readable(&self, path: &RepoPath, viewer: &Viewer) -> Result<Option<Repo>> {
301	self.visible(self.registry.by_path(path).await?, viewer)
302	.await
303	}
304
305	async fn get(&self, a: GetArgs) -> Result<Outcome<Repo>> {
306	Ok(match self.readable(&a.path, &a.viewer).await? {
307	Some(repo) => Outcome::Ok(self.with_parent(repo, &a.viewer).await?),
308	None => not_found(),
309	})
310	}
311
312	async fn get_by_id(&self, a: GetByIdArgs) -> Result<Outcome<Repo>> {
313	Ok(match self.visible(self.registry.by_id(&a.id).await?, &a.viewer).await? {
314	Some(repo) => Outcome::Ok(self.with_parent(repo, &a.viewer).await?),
315	None => not_found(),
316	})
317	}
318
319	async fn update(&self, a: UpdateArgs) -> Result<Outcome<Repo>> {
320	let viewer = Some(a.actor.clone());
321	let Some(repo) = self.readable(&a.path, &viewer).await? else {
322	return Ok(not_found());
323	};
324	// Its details take Maintain; its protection, Admin; who can see it,
325	// Admin and the member privileges (below). See g1t_contracts::access.
326	let protection_changes = a.protected.is_some_and(\|protected\| protected != repo.protected);
327	let forking_changes = a.allow_forking.is_some_and(\|allow\| allow != repo.allow_forking);
328	// 0 clears the limit.
329	let max_ref_updates = match a.max_ref_updates {
330	Some(0) => None,
331	Some(limit) if limit > MAX_REF_UPDATES_LIMIT => {
332	return Ok(Outcome::fail(
333	FailureCode::Invalid,
334	format!("A push can be limited to at most {MAX_REF_UPDATES_LIMIT} branches and tags; use 0 for no limit."),
335	));
336	}
337	Some(limit) => Some(limit),
338	None => repo.max_ref_updates,
339	};
340	let limit_changes = max_ref_updates != repo.max_ref_updates;
341	let details_change = a.description.is_some() \|\| a.website.is_some() \|\| a.topics.is_some();
342	let mut needed = Vec::new();
343	if details_change \|\| !(protection_changes \|\| forking_changes \|\| limit_changes) {
344	needed.push(Capability::ManageSettings);
345	}
346	if protection_changes {
347	needed.push(Capability::ManageProtection);
348	}
349	// Who may copy a private repository elsewhere is an admin's to say.
350	// So is how much one push may change.
351	if forking_changes \|\| limit_changes {
352	needed.push(Capability::Administer);
353	}
354	let full_name = format!("{}/{}", repo.namespace, repo.name);
355	if repo.fork_of.is_some() {
356	return Ok(Outcome::fail(FailureCode::Forbidden, access::needs(Capability::ManageSettings, &full_name)));
357	}
358	if let Some(missing) = needed.into_iter().find(\|capability\| !registry::can(&repo, &viewer, *capability)) {
359	return Ok(Outcome::fail(FailureCode::Forbidden, access::needs(missing, &full_name)));
360	}
361	if !a.actor.verified {
362	return Ok(Outcome::fail(FailureCode::Forbidden, UNVERIFIED));
363	}
364	if let Some((code, message)) = lifecycle::archived_refusal(&repo) {
365	return Ok(Outcome::fail(code, message));
366	}
367	let description = match a.description {
368	Some(text) => Some(
369	text.trim()
370	.chars()
371	.take(MAX_DESCRIPTION_CHARS)
372	.collect::<String>(),
373	)
374	.filter(\|text\| !text.is_empty()),
375	None => repo.description.clone(),
376	};
377	let website = match a.website.as_deref() {
378	Some(text) => match clean_website(text) {
379	Ok(website) => website,
380	Err(reason) => return Ok(Outcome::fail(FailureCode::Invalid, reason)),
381	},
382	None => repo.website.clone(),
383	};
384	// Who can see it is an owner's to change, and a free workspace's
385	// storage may not take it private: see lifecycle.rs.
386	let wants_private = a.is_private.filter(\|private\| *private != repo.is_private);
387	if wants_private.is_some()
388	&& let Err((code, message)) = lifecycle::admin_only(
389	lifecycle::Asker::on(&a.actor, &repo),
390	&repo.namespace,
391	"change the visibility of",
392	Capability::ChangeVisibility,
393	)
394	{
395	return Ok(Outcome::fail(code, message));
396	}
397	if let Some(private) = wants_private
398	&& let Some(why) = lifecycle::visibility_refusal(&a.actor, &repo, private)
399	{
400	return Ok(Outcome::fail(FailureCode::Forbidden, why));
401	}
402	let is_private = repo.is_private;
403	let protected = a.protected.unwrap_or(repo.protected);
404	let topics = match &a.topics {
405	Some(topics) => match clean_topics(topics) {
406	Ok(topics) => topics,
407	Err(reason) => return Ok(Outcome::fail(FailureCode::Invalid, reason)),
408	},
409	None => repo.topics.clone(),
410	};
411	let allow_forking = a.allow_forking.unwrap_or(repo.allow_forking);
412	self.registry
413	.update(&repo.id, description.as_deref(), protected, &topics, website.as_deref(), allow_forking, max_ref_updates)
414	.await?;
415	let updated = Repo {
416	description,
417	is_private,
418	protected,
419	topics,
420	website,
421	allow_forking,
422	max_ref_updates,
423	..repo
424	};
425	// Whether the default branch takes only pull requests is now its
426	// branch protection ruleset's to say (work's rulesets.rs).
427	if let (Some(protected), Some(work)) = (a.protected, &self.work) {
428	#[derive(Serialize)]
429	struct RequirePullRequest<'a> {
430	repo: &'a Repo,
431	protected: bool,
432	actor: &'a User,
433	}
434	let set: Result<Outcome<bool>> =
435	g1t_kit::call(work, "set_requires_pull_request", &RequirePullRequest { repo: &updated, protected, actor: &a.actor }).await;
436	match set {
437	Ok(Outcome::Ok(_)) => {}
438	Ok(Outcome::Fail(failure)) => return Ok(Outcome::Fail(failure)),
439	Err(error) => return Err(error),
440	}
441	}
442	if let Some(private) = wants_private {
443	return self.change_visibility(updated, private, &a.actor, a.surface).await;
444	}
445	let visibility_changed = false;
446	// Search and anything else that shows the repository hears of it;
447	// a change of visibility is announced on its own as well, so that
448	// what was public stops being shown at once.
449	self.publish(NewEvent {
450	kind: "repo.updated",
451	source: SOURCE,
452	repo_id: Some(updated.id.clone()),
453	actor: Some(a.actor.id.clone()),
454	data: RepoUpdated {
455	repo_id: updated.id.clone(),
456	namespace: updated.namespace.clone(),
457	name: updated.name.clone(),
458	is_private,
459	visibility_changed,
460	},
461	})
462	.await?;
463	Ok(Outcome::Ok(updated))
464	}
465
466	/// The repository with this id, if it is not a fork, and its store.
467	async fn stored(&self, repo_id: &str) -> Result<Option<S::Repo>> {
468	match self.registry.by_id(repo_id).await? {
469	Some(repo) if repo.fork_of.is_none() => Ok(Some(self.store.open(&store_key(&repo)).await?)),
470	_ => Ok(None),
471	}
472	}
473
474	async fn list_files(&self, a: ListFilesArgs) -> Result<FileList> {
475	let Some(repo) = self.registry.by_id(&a.repo_id).await?.filter(\|repo\| repo.fork_of.is_none()) else {
476	return Ok(FileList::default());
477	};
478	let git = self.store.open(&store_key(&repo)).await?;
479	let head = a.git_ref.unwrap_or_else(\|\| repo.default_branch.clone());
480	listing::list(&git, None, &head, &a.skip_dirs, a.limit, a.links).await
481	}
482
483	/// A link to download a branch, tag or commit as an archive, good for
484	/// a few minutes with no session or token (archive_links.rs).
485	async fn archive_link(&self, a: ArchiveLinkArgs, sealer: Option<&g1t_secrets::Sealer>) -> Result<Outcome<ArchiveLink>> {
486	let Some(repo) = self.readable(&a.path, &a.viewer).await? else {
487	return Ok(not_found());
488	};
489	let Some(sealer) = sealer else {
490	return Ok(Outcome::fail(FailureCode::Invalid, "Archive downloads are not set up on this installation."));
491	};
492	let git = self.read_git(&repo).await?;
493	let git_ref = a.git_ref.filter(\|git_ref\| !git_ref.is_empty()).unwrap_or_else(\|\| repo.default_branch.clone());
494	let Some(head) = git.log(&git_ref, 1).await?.into_iter().next() else {
495	return Ok(Outcome::fail(FailureCode::NotFound, "No such branch, tag or commit."));
496	};
497	let (token, expires) = archive_links::seal(sealer, &repo.id, &head.hash, now_ms());
498	Ok(Outcome::Ok(ArchiveLink { repo, commit: head.hash, token, expires_at: rfc3339(expires) }))
499	}
500
501	/// What an archive link's token lets its holder download.
502	async fn open_archive_token(&self, a: OpenArchiveTokenArgs, sealer: Option<&g1t_secrets::Sealer>) -> Result<Option<ArchiveGrant>> {
503	let Some(sealer) = sealer else {
504	return Ok(None);
505	};
506	let Some(repo) = self.registry.by_path(&a.path).await? else {
507	return Ok(None);
508	};
509	let Some(commit) = archive_links::open(sealer, &repo.id, &a.token, now_ms()) else {
510	return Ok(None);
511	};
512	Ok(Some(ArchiveGrant { repo_id: repo.id, namespace: repo.namespace, name: repo.name, commit }))
513	}
514
515	async fn changed_files(&self, a: ChangedFilesArgs) -> Result<FileList> {
516	let Some(git) = self.stored(&a.repo_id).await? else {
517	return Ok(FileList::default());
518	};
519	listing::list(&git, a.base.as_deref(), &a.head, &a.skip_dirs, a.limit, false).await
520	}
521
522	/// Branches and tags with their commits, for g1t's own services.
523	async fn refs_of(&self, a: RefsArgs) -> Result<Option<RepoRefs>> {
524	let Some(repo) = self.registry.by_id(&a.repo_id).await?.filter(\|repo\| repo.fork_of.is_none()) else {
525	return Ok(None);
526	};
527	let git = self.store.open(&store_key(&repo)).await?;
528	let access = git.access(Scope::Read).await?;
529	let refs = refs::heads_and_tags(refs::all(&access).await?)
530	.into_iter()
531	.map(\|(name, commit)\| GitRefEntry { name, commit })
532	.collect();
533	Ok(Some(RepoRefs { repo, refs }))
534	}
535
536	async fn raw_file(&self, a: RawFileArgs) -> Result<Option<RawFile>> {
537	use base64::Engine;
538	let Some(git) = self.stored(&a.repo_id).await? else {
539	return Ok(None);
540	};
541	Ok(git
542	.read_file(&a.git_ref, &a.path)
543	.await?
544	.filter(\|bytes\| bytes.len() <= a.max_bytes as usize)
545	.map(\|bytes\| RawFile { size: bytes.len() as u64, data: base64::engine::general_purpose::STANDARD.encode(bytes) }))
546	}
547
548	async fn raw_blobs(&self, a: RawBlobsArgs) -> Result<Vec<RawBlob>> {
549	use base64::Engine;
550	let Some(git) = self.stored(&a.repo_id).await? else {
551	return Ok(Vec::new());
552	};
553	let hashes: Vec<&String> = a.hashes.iter().take(MAX_READ_BLOBS).collect();
554	let mut out = Vec::with_capacity(hashes.len());
555	// A few at a time, as listing::read does: each is a round trip.
556	for group in hashes.chunks(8) {
557	let read = futures_util::future::try_join_all(group.iter().map(\|hash\| git.read_blob(hash))).await?;
558	for (hash, bytes) in group.iter().zip(read) {
559	let size = bytes.as_ref().map_or(0, \|bytes\| bytes.len() as u64);
560	let data = bytes
561	.filter(\|bytes\| bytes.len() <= a.max_bytes as usize)
562	.map(\|bytes\| base64::engine::general_purpose::STANDARD.encode(bytes));
563	out.push(RawBlob { hash: (*hash).clone(), size, data });
564	}
565	}
566	Ok(out)
567	}
568
569	async fn read_blobs(&self, a: ReadBlobsArgs) -> Result<Vec<BlobText>> {
570	let Some(git) = self.stored(&a.repo_id).await? else {
571	return Ok(Vec::new());
572	};
573	listing::read(&git, &a.hashes, a.max_bytes.min(MAX_TEXT_BYTES as u32)).await
574	}
575
576	async fn create(&self, a: CreateArgs) -> Result<Outcome<Repo>> {
577	if !a.owner.verified {
578	return Ok(Outcome::fail(FailureCode::Forbidden, UNVERIFIED));
579	}
580	let name = a.name.trim().to_lowercase();
581	if !is_valid_repo_name(&name) {
582	return Ok(Outcome::fail(
583	FailureCode::Invalid,
584	"Use letters, digits, dots, hyphens and underscores only.",
585	));
586	}
587	let namespace = a.namespace.trim().to_lowercase();
588	if namespace.is_empty() {
589	return Ok(Outcome::fail(
590	FailureCode::Invalid,
591	"Say which workspace to create the repository in.",
592	));
593	}
594	let Some(role) = a.owner.role_in(&namespace) else {
595	return Ok(Outcome::fail(
596	FailureCode::Forbidden,
597	"You are not a member of that workspace.",
598	));
599	};
600	// Who may create which: the workspace's member privileges. A
601	// workspace's own token acts as an owner would.
602	let role = if a.owner.kind == PrincipalKind::Workspace { Role::Owner } else { role };
603	if let Some(why) = a.owner.privileges_in(&namespace).creation_refusal(role, a.is_private, &namespace) {
604	return Ok(Outcome::fail(FailureCode::Forbidden, why));
605	}
606	let path = RepoPath { namespace, name };
607	match self.registry.by_path_any(&path).await? {
608	Some((_, None)) => {
609	return Ok(Outcome::fail(
610	FailureCode::Conflict,
611	"That workspace already has a repository with that name.",
612	));
613	}
614	Some((_, Some(_))) => {
615	return Ok(Outcome::fail(
616	FailureCode::Conflict,
617	format!(
618	"{}/{} 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.",
619	path.namespace, path.name
620	),
621	));
622	}
623	None => {}
624	}
625	// With a credential (a GitHub App installation's token), everything
626	// is copied: every branch and tag. See copy.rs.
627	let mut credentialed = None;
628	if let (Some(url), Some(token)) = (a.import_url.as_deref(), a.import_token.as_deref()) {
629	let Some(url) = import::clean_url(url) else {
630	return Ok(Outcome::fail(FailureCode::Invalid, "That is not an https repository address."));
631	};
632	let source = copy::Endpoint::github(&url, token);
633	match copy::probe(&source).await? {
634	Ok(advertised) => credentialed = Some((source, advertised)),
635	Err(reason) => return Ok(Outcome::fail(FailureCode::Invalid, reason)),
636	}
637	}
638	// An import is fetched before anything is created, so that an
639	// address that does not work leaves nothing behind.
640	let mut imported = None;
641	if let Some(url) = a
642	.import_url
643	.as_deref()
644	.map(str::trim)
645	.filter(\|url\| !url.is_empty() && credentialed.is_none())
646	{
647	let Some(url) = import::clean_url(url) else {
648	return Ok(Outcome::fail(
649	FailureCode::Invalid,
650	"Give the https address of a public repository, such as https://github.com/owner/repo.",
651	));
652	};
653	let remote = match import::discover(&url).await? {
654	Ok(remote) => remote,
655	Err(reason) => return Ok(Outcome::fail(FailureCode::Invalid, reason)),
656	};
657	imported = Some((remote, url));
658	}
659	let now = now_ms();
660	let repo = Repo {
661	id: new_id("rep", now),
662	namespace: path.namespace,
663	name: path.name,
664	description: a
665	.description
666	.map(\|text\| text.trim().to_owned())
667	.filter(\|text\| !text.is_empty()),
668	is_private: a.is_private,
669	owner_id: a.owner.id.clone(),
670	default_branch: imported
671	.as_ref()
672	.map(\|(remote, _)\| remote.branch.clone())
673	.or_else(\|\| credentialed.as_ref().and_then(\|(_, advertised)\| advertised.default_branch()))
674	.unwrap_or_else(\|\| "main".to_owned()),
675	fork_of: None,
676	parent_id: None,
677	forked_from: None,
678	parent: None,
679	allow_forking: false,
680	max_ref_updates: None,
681	protected: false,
682	created_at: rfc3339(now),
683	topics: Vec::new(),
684	website: None,
685	archived_at: None,
686	};
687	let namespace = match self.place(&repo).await? {
688	Ok(namespace) => namespace,
689	Err(unplaced) => return Ok(Outcome::fail(FailureCode::Conflict, unplaced.message())),
690	};
691	self.registry
692	.claim_store_key(&repo, namespace.as_deref(), &self.store.default_namespace())
693	.await?;
694	self.store
695	.create(
696	&store_key(&repo),
697	repo.description.as_deref(),
698	&repo.default_branch,
699	)
700	.await?;
701	self.registry.insert(&repo).await?;
702	// A repository that was transferred away from this path stops
703	// redirecting here.
704	self.registry
705	.drop_redirect(&RepoPath {
706	namespace: repo.namespace.clone(),
707	name: repo.name.clone(),
708	})
709	.await?;
710	// Every branch and tag the import made, announced as pushes.
711	let mut pushed: Vec<(String, String)> = Vec::new();
712	// A public repository, read with no credential: every branch and
713	// tag is copied too, the default branch the one its HEAD names.
714	if let Some((_, url)) = imported {
715	let access = self
716	.store
717	.open(&store_key(&repo))
718	.await?
719	.access(Scope::Write)
720	.await?;
721	let target = copy::Endpoint::bearer(&access.remote, &access.token);
722	let copied = copy::copy(&copy::Endpoint::anonymous(&url), &target).await?;
723	self.refs_moved(&repo.id).await;
724	match copied {
725	Ok(copied) => pushed = copy::import_pushes(&copied.updated, &repo.default_branch),
726	Err(reason) => {
727	self.registry.remove(&repo.id).await?;
728	return Ok(Outcome::fail(
729	FailureCode::Invalid,
730	format!("The repository could not be stored: {reason}"),
731	));
732	}
733	}
734	}
735	if let Some((source, _)) = credentialed {
736	let access = self
737	.store
738	.open(&store_key(&repo))
739	.await?
740	.access(Scope::Write)
741	.await?;
742	let target = copy::Endpoint::bearer(&access.remote, &access.token);
743	let copied = copy::copy(&source, &target).await?;
744	self.refs_moved(&repo.id).await;
745	match copied {
746	Ok(copied) => pushed = copy::import_pushes(&copied.updated, &repo.default_branch),
747	Err(reason) => {
748	self.registry.remove(&repo.id).await?;
749	return Ok(Outcome::fail(
750	FailureCode::Invalid,
751	format!("The repository could not be copied: {reason}"),
752	));
753	}
754	}
755	}
756	// Whoever creates a repository is an Admin of it, as a role given
757	// to them on it, whatever the workspace's base permission.
758	if a.owner.kind == PrincipalKind::User {
759	self.grant_creator(&repo, &a.owner.id).await;
760	}
761	self.publish(NewEvent {
762	kind: "repo.created",
763	source: SOURCE,
764	repo_id: Some(repo.id.clone()),
765	actor: Some(a.owner.id),
766	data: RepoCreated {
767	repo_id: repo.id.clone(),
768	namespace: repo.namespace.clone(),
769	name: repo.name.clone(),
770	is_private: repo.is_private,
771	parent_id: None,
772	},
773	})
774	.await?;
775	for (git_ref, head) in &pushed {
776	self.publish_push(&repo, git_ref, None, head, None, None).await?;
777	}
778	Ok(Outcome::Ok(repo))
779	}
780
781	/// Makes `user_id` an Admin of `repo`, the repository they created.
782	pub(crate) async fn grant_creator(&self, repo: &Repo, user_id: &str) {
783	let Some(identity) = &self.identity else { return };
784	let granted: Result<bool> = g1t_kit::call(
785	identity,
786	"grant_creator",
787	&g1t_contracts::members::GrantCreatorArgs {
788	repo_id: repo.id.clone(),
789	namespace: repo.namespace.clone(),
790	name: repo.name.clone(),
791	user_id: user_id.to_owned(),
792	},
793	)
794	.await;
795	if let Err(error) = granted {
796	worker::console_error!("creator of {} not given Admin: {error}", repo.id);
797	}
798	}
799
800	/// Where a workspace keeps its data, asked of identity only when an EU
801	/// namespace is configured: without one, every workspace's
802	/// repositories go anywhere and identity is never asked.
803	async fn residency_of(&self, workspace: &str) -> Result<shards::Residency> {
804	if self.placement.eu.is_none() {
805	return Ok(shards::Residency::Anywhere);
806	}
807	let Some(identity) = &self.identity else {
808	return Ok(shards::Residency::Anywhere);
809	};
810	let residency: Option<g1t_contracts::identity::DataResidency> = g1t_kit::call(
811	identity,
812	"workspace_residency",
813	&g1t_contracts::identity::SlugArgs { slug: workspace.to_owned() },
814	)
815	.await?;
816	Ok(match residency {
817	Some(g1t_contracts::identity::DataResidency::Eu) => shards::Residency::Eu,
818	_ => shards::Residency::Anywhere,
819	})
820	}
821
822	/// How each bound namespace stands (namespaces.rs).
823	async fn standings(&self) -> Result<Vec<namespaces::Standing>> {
824	let bound = self.store.namespaces();
825	let default = self.store.default_namespace();
826	let now = now_ms();
827	let config = namespaces::Configured {
828	bound: &bound,
829	default: &default,
830	placement: &self.placement,
831	limits: &self.limits,
832	on_fallback: &\|namespace\| self.store.on_fallback(&shards::compose(Some(namespace), "x", &default)),
833	writable: &\|namespace\| self.store.writable(namespace),
834	breaker_open: &\|namespace\| resilience::open_now(namespace, now),
835	};
836	let (held, recent) = futures_util::future::join(namespaces::held(&self.registry.db), namespaces::recent(&self.registry.db, now)).await;
837	Ok(namespaces::standings(&config, &held?, &recent?))
838	}
839
840	/// The namespace a new repository goes in (shards.rs): its workspace's
841	/// residency, then how each namespace stands, read only when there is
842	/// a choice to make. `Ok(None)` for the default.
843	async fn place(&self, repo: &Repo) -> Result<std::result::Result<Option<String>, shards::Unplaced>> {
844	let residency = self.residency_of(&repo.namespace).await?;
845	let bound = self.store.namespaces();
846	let loads = if residency == shards::Residency::Anywhere && !self.placement.needs_loads(&bound) {
847	// One namespace to choose from at most: nothing to read.
848	bound
849	.iter()
850	.map(\|namespace\| shards::Load {
851	namespace: namespace.clone(),
852	bound: true,
853	writable: self.store.writable(namespace),
854	..shards::Load::default()
855	})
856	.collect()
857	} else {
858	let default = self.store.default_namespace();
859	let now = now_ms();
860	let config = namespaces::Configured {
861	bound: &bound,
862	default: &default,
863	placement: &self.placement,
864	limits: &self.limits,
865	on_fallback: &\|namespace\| self.store.on_fallback(&shards::compose(Some(namespace), "x", &default)),
866	writable: &\|namespace\| self.store.writable(namespace),
867	breaker_open: &\|namespace\| resilience::open_now(namespace, now),
868	};
869	namespaces::loads(&self.registry.db, &config, now).await?
870	};
871	Ok(self.placement.choose(&repo.id, residency, &loads))
872	}
873
874	/// `storage_options`: what a workspace may choose about where its
875	/// repositories are kept.
876	fn storage_options(&self) -> StorageOptions {
877	let bound = self.store.namespaces();
878	StorageOptions { eu_available: self.placement.eu_available(&bound, \|namespace\| self.store.writable(namespace)) }
879	}
880
881	async fn tree(&self, a: TreeArgs) -> Result<Outcome<TreeView>> {
882	let Some(repo) = self.readable(&a.path, &a.viewer).await? else {
883	return Ok(not_found());
884	};
885	let git = self.read_git(&repo).await?;
886	let git_ref = a
887	.git_ref
888	.clone()
889	.unwrap_or_else(\|\| repo.default_branch.clone());
890
891	let Some(head) = git.log(&git_ref, 1).await?.into_iter().next() else {
892	// An unknown ref is an error; a repo with no commits is just empty.
893	if a.git_ref.is_some() {
894	return Ok(Outcome::fail(
895	FailureCode::NotFound,
896	"No such branch, tag or commit.",
897	));
898	}
899	return Ok(Outcome::Ok(TreeView {
900	repo,
901	git_ref,
902	path: a.tree_path,
903	head: None,
904	entries: Vec::new(),
905	readme: None,
906	}));
907	};
908
909	let no_directory = \|\| Outcome::fail(FailureCode::NotFound, "No such directory.");
910	let mut entries = git.read_tree(&head.tree_hash).await?;
911	for segment in a.tree_path.split('/').filter(\|segment\| !segment.is_empty()) {
912	let next = entries.as_ref().and_then(\|entries\| {
913	entries
914	.iter()
915	.find(\|entry\| entry.name == segment && entry.kind == EntryKind::Tree)
916	});
917	let Some(next) = next else {
918	return Ok(no_directory());
919	};
920	entries = git.read_tree(&next.hash).await?;
921	}
922	let Some(mut entries) = entries else {
923	return Ok(no_directory());
924	};
925	// Directories first, then by name.
926	entries.sort_by(\|a, b\| {
927	(b.kind == EntryKind::Tree)
928	.cmp(&(a.kind == EntryKind::Tree))
929	.then_with(\|\| a.name.cmp(&b.name))
930	});
931
932	let readme_entry = entries
933	.iter()
934	.find(\|entry\| entry.kind == EntryKind::Blob && is_readme(&entry.name));
935	let readme = match readme_entry {
936	Some(entry) => git.read_blob(&entry.hash).await?.map(\|bytes\| Readme {
937	name: entry.name.clone(),
938	text: text_of(bytes),
939	}),
940	None => None,
941	};
942	Ok(Outcome::Ok(TreeView {
943	repo,
944	git_ref,
945	path: a.tree_path,
946	head: Some(head),
947	entries,
948	readme,
949	}))
950	}
951
952	async fn blob(&self, a: BlobArgs) -> Result<Outcome<BlobView>> {
953	let Some(repo) = self.readable(&a.path, &a.viewer).await? else {
954	return Ok(not_found());
955	};
956	let bytes = if a.file_path.is_empty() {
957	None
958	} else {
959	let git = self.read_git(&repo).await?;
960	git.read_file(&a.git_ref, &a.file_path).await?
961	};
962	let Some(bytes) = bytes else {
963	return Ok(Outcome::fail(FailureCode::NotFound, "No such file."));
964	};
965	Ok(Outcome::Ok(BlobView {
966	repo,
967	git_ref: a.git_ref,
968	path: a.file_path,
969	size: bytes.len() as u64,
970	text: text_of(bytes),
971	}))
972	}
973
974	async fn blame(&self, a: BlameArgs) -> Result<Outcome<Blame>> {
975	let Some(repo) = self.readable(&a.path, &a.viewer).await? else {
976	return Ok(not_found());
977	};
978	let git = self.read_git(&repo).await?;
979	let git_ref = a.git_ref.unwrap_or_else(\|\| repo.default_branch.clone());
980	Ok(match blame::blame(&git, &git_ref, &a.file_path).await? {
981	Some(blame) => Outcome::Ok(blame),
982	None => not_found(),
983	})
984	}
985
986	async fn log(&self, a: LogArgs) -> Result<Outcome<Vec<Commit>>> {
987	let Some(repo) = self.readable(&a.path, &a.viewer).await? else {
988	return Ok(not_found());
989	};
990	let git = self.read_git(&repo).await?;
991	let git_ref = a.git_ref.unwrap_or_else(\|\| repo.default_branch.clone());
992	Ok(Outcome::Ok(git.log(&git_ref, a.limit).await?))
993	}
994
995	/// The commits that changed a file or directory, newest first: the
996	/// first-parent chain walked within `PATH_LOG_WALK_MS` (or the caller's
997	/// budget) and `last_commits::MAX_READS`.
998	async fn path_log(&self, a: g1t_contracts::repos::PathLogArgs) -> Result<Outcome<g1t_contracts::repos::PathLog>> {
999	let Some(repo) = self.readable(&a.path, &a.viewer).await? else {
1000	return Ok(not_found());
1001	};
1002	let git = self.read_git(&repo).await?;
1003	let git_ref = a.git_ref.unwrap_or_else(\|\| repo.default_branch.clone());
1004	let Some(head) = git.log(&git_ref, 1).await?.into_iter().next() else {
1005	return Ok(Outcome::fail(FailureCode::NotFound, "No such branch, tag or commit."));
1006	};
1007	let started = worker::Date::now().as_millis();
1008	let budget = a.budget_ms.unwrap_or(PATH_LOG_WALK_MS);
1009	let out_of_time = move \|\| worker::Date::now().as_millis().saturating_sub(started) > budget;
1010	let limit = a.limit.clamp(1, 100) as usize;
1011	let (commits, complete) = last_commits::path_log(&git, &head.hash, &a.file_path, limit, &out_of_time, last_commits::MAX_READS).await?;
1012	Ok(Outcome::Ok(g1t_contracts::repos::PathLog { commits, complete }))
1013	}
1014
1015	/// Which commit last changed each entry of a directory. A finished
1016	/// answer is kept in this colo's cache by repository, head commit and
1017	/// path: a commit's history never changes, so it is good for as long as
1018	/// it is kept. Every walk also keeps its progress by ref and path
1019	/// (last_commits.rs), so the next call goes on from it: from where it
1020	/// stopped, or for a new head, only back to the old one.
1021	async fn last_commits(&self, a: g1t_contracts::repos::LastCommitsArgs) -> Result<Outcome<g1t_contracts::repos::LastCommits>> {
1022	let Some(repo) = self.readable(&a.path, &a.viewer).await? else {
1023	return Ok(not_found());
1024	};
1025	let git = self.read_git(&repo).await?;
1026	let git_ref = a.git_ref.unwrap_or_else(\|\| repo.default_branch.clone());
1027	let Some(head) = git.log(&git_ref, 1).await?.into_iter().next() else {
1028	return Ok(Outcome::fail(FailureCode::NotFound, "No such branch, tag or commit."));
1029	};
1030	// v2: v1 kept walks cut short at 300 commits as if finished.
1031	let key = format!(
1032	"https://last-commits.g1t.internal/v2/{}/{}/{}",
1033	repo.id,
1034	head.hash,
1035	a.tree_path.split('/').map(urlencoding_segment).collect::<Vec<_>>().join("/")
1036	);
1037	let cache = worker::Cache::default();
1038	if let Ok(Some(mut kept)) = cache.get(key.as_str(), false).await {
1039	if let Ok(found) = kept.json::<g1t_contracts::repos::LastCommits>().await {
1040	return Ok(Outcome::Ok(found));
1041	}
1042	}
1043	let memo = last_commits::Memo::new(&repo.id, &git_ref, &a.tree_path);
1044	let shared = self.shared.as_deref();
1045	let progress = memo.get(shared).await;
1046	// Asked with a budget, the walk stops past it; without one, past
1047	// LAST_COMMITS_WALK_MS, well before the caller's waitUntil ends.
1048	let started = worker::Date::now().as_millis();
1049	let budget = a.budget_ms.unwrap_or(LAST_COMMITS_WALK_MS);
1050	let out_of_time = move \|\| worker::Date::now().as_millis().saturating_sub(started) > budget;
1051	let walk = last_commits::last_commits(&git, &head.hash, &a.tree_path, progress, &out_of_time, last_commits::MAX_READS).await?;
1052	if walk.reads > 0 {
1053	memo.keep(shared, &walk.progress).await;
1054	}
1055	let settled = walk.progress.settled();
1056	let found = g1t_contracts::repos::LastCommits { complete: walk.progress.complete(), entries: walk.progress.found };
1057	if !settled {
1058	return Ok(Outcome::Ok(found));
1059	}
1060	if let Ok(mut response) = worker::Response::from_json(&found) {
1061	let _ = response.headers_mut().set("cache-control", "max-age=604800");
1062	let _ = cache.put(key.as_str(), response).await;
1063	}
1064	Ok(Outcome::Ok(found))
1065	}
1066
1067	/// How far each branch head has moved from the default branch's head,
1068	/// in one call (drift.rs). Each count is kept in this colo's cache by
1069	/// repository and the pair of hashes, for good: neither history can
1070	/// change. A head that moved is the only one walked. A failed read is
1071	/// not kept, and "too far to count" only for a day.
1072	async fn branch_drift(&self, a: BranchDriftArgs) -> Result<Outcome<BranchDrifts>> {
1073	let Some(repo) = self.readable(&a.path, &a.viewer).await? else {
1074	return Ok(not_found());
1075	};
1076	if !store::is_commit_hash(&a.base) {
1077	return Ok(Outcome::fail(FailureCode::Invalid, "The default branch's head is a full commit hash."));
1078	}
1079	let heads: Vec<String> = a.heads.into_iter().take(MAX_DRIFT_HEADS).collect();
1080	let git = self.read_git(&repo).await?;
1081	// v2: answers kept before 2026-10-09 said "no count" for every
1082	// branch whose default branch took a merge since it left (drift.rs).
1083	let key = \|head: &str\| format!("https://drift.g1t.internal/v2/{}/{}/{head}", repo.id, a.base);
1084	let cache = worker::Cache::default();
1085	let (base, kept) = futures_util::future::join(
1086	git.log(&a.base, 1),
1087	futures_util::future::join_all(heads.iter().map(\|head\| {
1088	let (cache, url) = (&cache, key(head));
1089	async move {
1090	if !store::is_commit_hash(head) {
1091	return None;
1092	}
1093	let mut found = cache.get(url.as_str(), false).await.ok()??;
1094	found.json::<BranchDrift>().await.ok()
1095	}
1096	})),
1097	)
1098	.await;
1099	let missing: Vec<String> = heads
1100	.iter()
1101	.zip(&kept)
1102	.filter(\|(head, kept)\| kept.is_none() && store::is_commit_hash(head))
1103	.map(\|(head, _)\| head.clone())
1104	.collect();
1105	let measured = drift::measure(&git, &a.base, &missing).await;
1106	let mut fresh: HashMap<String, BranchDrift> = HashMap::new();
1107	for (head, found) in missing.into_iter().zip(measured) {
1108	let answer = BranchDrift { head: head.clone(), commit: found.commit, drift: found.drift };
1109	if found.settled
1110	&& let Ok(mut response) = worker::Response::from_json(&answer)
1111	{
1112	// A count is kept for good; "too far to count" for a day, so
1113	// a change to how far a walk goes reaches it.
1114	let max_age = if answer.drift.is_some() { "public, max-age=31536000, immutable" } else { "public, max-age=86400" };
1115	let _ = response.headers_mut().set("cache-control", max_age);
1116	let _ = cache.put(key(&head).as_str(), response).await;
1117	}
1118	fresh.insert(head, answer);
1119	}
1120	let branches = heads
1121	.iter()
1122	.zip(kept)
1123	.map(\|(head, kept)\| {
1124	kept.or_else(\|\| fresh.get(head).cloned())
1125	.unwrap_or_else(\|\| BranchDrift { head: head.clone(), commit: None, drift: None })
1126	})
1127	.collect();
1128	Ok(Outcome::Ok(BranchDrifts { base: base.ok().and_then(\|log\| log.into_iter().next()), branches }))
1129	}
1130
1131	/// The repository's tags, newest commit first, at most 100.
1132	async fn tags(&self, a: g1t_contracts::repos::TagsArgs) -> Result<Outcome<Vec<g1t_contracts::repos::Tag>>> {
1133	let Some(repo) = self.readable(&a.path, &a.viewer).await? else {
1134	return Ok(not_found());
1135	};
1136	// Kept until the refs move, as the branch list is (store.rs): listing
1137	// the refs is a round trip to the store on every call otherwise.
1138	let version = refs_cache::usable(registry::refs_state(&repo.id), now_ms()).filter(\|_\| !self.store.on_fallback(&store_key(&repo)));
1139	let kept_at = version.map(\|version\| format!("https://tags.g1t.internal/{}/{version}", repo.id));
1140	if let Some(url) = &kept_at
1141	&& let Ok(Some(mut kept)) = worker::Cache::default().get(url.as_str(), false).await
1142	&& let Ok(tags) = kept.json::<Vec<g1t_contracts::repos::Tag>>().await
1143	{
1144	return Ok(Outcome::Ok(tags));
1145	}
1146	let (tags, complete) = self.read_tags(&repo).await?;
1147	if complete
1148	&& let Some(url) = &kept_at
1149	&& let Ok(mut response) = worker::Response::from_json(&tags)
1150	{
1151	let _ = response.headers_mut().set("cache-control", "public, max-age=300");
1152	let _ = worker::Cache::default().put(url.as_str(), response).await;
1153	}
1154	Ok(Outcome::Ok(tags))
1155	}
1156
1157	/// The tags, and whether every one's commit was read (only then kept).
1158	async fn read_tags(&self, repo: &Repo) -> Result<(Vec<g1t_contracts::repos::Tag>, bool)> {
1159	let git = self.store.open(&store_key(repo)).await?;
1160	let access = git.access(Scope::Read).await?;
1161	let named: Vec<(String, String)> = refs::heads_and_tags(refs::all(&access).await?)
1162	.into_iter()
1163	.filter_map(\|(name, hash)\| name.strip_prefix("refs/tags/").map(\|tag\| (tag.to_owned(), hash)))
1164	.collect();
1165	let read = self.read_git(repo).await?;
1166	let commits = futures_util::future::join_all(named.iter().take(MAX_TAGS_READ).map(\|(_, hash)\| read.log(hash, 1))).await;
1167	let complete = commits.iter().all(Result::is_ok);
1168	let mut tags: Vec<g1t_contracts::repos::Tag> = named
1169	.into_iter()
1170	.zip(commits.into_iter().map(\|found\| found.ok().and_then(\|list\| list.into_iter().next())).chain(std::iter::repeat(None)))
1171	.map(\|((name, _), commit)\| g1t_contracts::repos::Tag { name, commit })
1172	.collect();
1173	tags.sort_by(\|a, b\| {
1174	let at = \|tag: &g1t_contracts::repos::Tag\| tag.commit.as_ref().map(\|c\| c.authored_at.clone()).unwrap_or_default();
1175	at(b).cmp(&at(a)).then_with(\|\| b.name.cmp(&a.name))
1176	});
1177	tags.truncate(MAX_TAGS_READ);
1178	Ok((tags, complete))
1179	}
1180
1181	/// The repository's branches, default branch first.
1182	async fn branches(&self, a: BranchesArgs) -> Result<Outcome<Vec<Branch>>> {
1183	let Some(repo) = self.readable(&a.path, &a.viewer).await? else {
1184	return Ok(not_found());
1185	};
1186	let mut branches = self.read_git(&repo).await?.branches().await?;
1187	branches.sort_by_key(\|branch\| branch.name != repo.default_branch);
1188	Ok(Outcome::Ok(branches))
1189	}
1190
1191	/// Whether `source` is a fork of `ancestor_id`: its parent, or its
1192	/// parent's, and so on, at most [`MAX_FORK_DEPTH`] repositories up. A
1193	/// detached fork still counts as one of what it was made from
1194	/// (`forked_from`), so pull requests opened from it still merge.
1195	pub(crate) async fn forked_from(&self, source: &Repo, ancestor_id: &str) -> Result<bool> {
1196	let up = \|repo: &Repo\| repo.parent_id.clone().or_else(\|\| repo.forked_from.clone());
1197	let mut parent = up(source);
1198	for _ in 0..MAX_FORK_DEPTH {
1199	let Some(id) = parent else {
1200	return Ok(false);
1201	};
1202	if id == ancestor_id {
1203	return Ok(true);
1204	}
1205	parent = self.registry.by_id(&id).await?.and_then(\|repo\| up(&repo));
1206	}
1207	Ok(false)
1208	}
1209
1210	/// The repository a pull request whose change is in `source` merges
1211	/// into. `target_id`, when given and not `source` itself, must be the
1212	/// repository a working copy was made from or one `source` is a fork of
1213	/// ([`forked_from`](Self::forked_from)); otherwise none. Without it, a
1214	/// working copy merges into the repository it was made from, and a
1215	/// branch into its own.
1216	pub(crate) async fn pull_target(&self, source: &Repo, target_id: Option<&str>) -> Result<Option<Repo>> {
1217	match target_id.filter(\|id\| *id != source.id) {
1218	Some(id) if source.fork_of.as_deref() == Some(id) \|\| self.forked_from(source, id).await? => {
1219	self.registry.by_id(id).await
1220	}
1221	Some(_) => Ok(None),
1222	None => match &source.fork_of {
1223	Some(id) => self.registry.by_id(id).await,
1224	None => Ok(Some(source.clone())),
1225	},
1226	}
1227	}
1228
1229	/// Whether a pull request's source lacks commits that the branch it
1230	/// would merge into has.
1231	async fn behind(&self, a: BehindArgs) -> Result<bool> {
1232	let Some(source) = self.registry.by_id(&a.source_id).await? else {
1233	return Ok(false);
1234	};
1235	let target = self.pull_target(&source, a.target_id.as_deref()).await?;
1236	let Some(target) = target else {
1237	return Ok(false);
1238	};
1239	let branch = a.branch.unwrap_or_else(\|\| target.default_branch.clone());
1240	let target_branch = a.target_branch.unwrap_or_else(\|\| target.default_branch.clone());
1241	let target_head = self
1242	.read_git(&target)
1243	.await?
1244	.log(&target_branch, 1)
1245	.await?
1246	.into_iter()
1247	.next()
1248	.map(\|commit\| commit.hash);
1249	let Some(target_head) = target_head else {
1250	return Ok(false);
1251	};
1252	let source_git = self.read_git(&source).await?;
1253	let history = source_git.log(&branch, MAX_ANCESTRY).await?;
1254	if history.is_empty() {
1255	return Ok(false);
1256	}
1257	Ok(!descends_from(&source_git, &history, &target_head).await?)
1258	}
1259
1260	/// The files a pull request's source and the default branch it would
1261	/// merge into each changed since they last agreed. Where the two lists
1262	/// share no file, the merge cannot conflict; where they do, it may.
1263	async fn divergence(&self, a: BehindArgs) -> Result<Option<Divergence>> {
1264	let Some(source) = self.registry.by_id(&a.source_id).await? else {
1265	return Ok(None);
1266	};
1267	let target = self.pull_target(&source, a.target_id.as_deref()).await?;
1268	let Some(target) = target else {
1269	return Ok(None);
1270	};
1271	let branch = a.branch.unwrap_or_else(\|\| target.default_branch.clone());
1272	let target_branch = a.target_branch.unwrap_or_else(\|\| target.default_branch.clone());
1273	let source_git = self.read_git(&source).await?;
1274	let target_git = self.read_git(&target).await?;
1275	// The target's side is the same for every pull request into it, and
1276	// worked out once per head (coalesce.rs).
1277	let (history, side) = futures_util::future::try_join(
1278	source_git.log(&branch, MAX_ANCESTRY),
1279	self.target_side(&target, &target_git, &target_branch),
1280	)
1281	.await?;
1282	let target_history = &side.history;
1283	let (Some(head), Some(base)) = (history.first(), target_history.first()) else {
1284	return Ok(None);
1285	};
1286	let behind = !descends_from(&source_git, &history, &base.hash).await?;
1287	let merge_base = nearest_ancestor_in(&source_git, &history, &side.shared).await?;
1288	let mut divergence = Divergence {
1289	head: head.hash.clone(),
1290	base: base.hash.clone(),
1291	merge_base: merge_base.clone(),
1292	behind,
1293	..Divergence::default()
1294	};
1295	let merge_base_tree = match &merge_base {
1296	Some(hash) => target_history
1297	.iter()
1298	.find(\|commit\| commit.hash == *hash)
1299	.map(\|commit\| commit.tree_hash.clone()),
1300	None => None,
1301	};
1302	let Some(merge_base_tree) = merge_base_tree else {
1303	// No common history to compare from: say nothing is known.
1304	divergence.truncated = true;
1305	return Ok(Some(divergence));
1306	};
1307	let (ours, truncated_ours) =
1308	diff::changed_paths(&source_git, Some(&merge_base_tree), &head.tree_hash).await?;
1309	divergence.ours = ours;
1310	divergence.truncated = truncated_ours;
1311	if behind {
1312	let now = now_ms();
1313	let key = (target.id.clone(), merge_base_tree.clone(), base.tree_hash.clone());
1314	let (theirs, truncated_theirs) = match THEIRS.with(\|memo\| memo.borrow().get(&key, now)) {
1315	Some(kept) => kept,
1316	None => {
1317	let found = diff::changed_paths(&target_git, Some(&merge_base_tree), &base.tree_hash).await?;
1318	THEIRS.with(\|memo\| memo.borrow_mut().put(key, found.clone(), now));
1319	found
1320	}
1321	};
1322	divergence.theirs = theirs;
1323	divergence.truncated \|= truncated_theirs;
1324	}
1325	Ok(Some(divergence))
1326	}
1327
1328	/// A target branch's history from its head, worked out once per head
1329	/// for every pull request asking about it (coalesce.rs). The head is
1330	/// read under the refs version; the history by its hash, which the
1331	/// object cache keeps for good.
1332	async fn target_side<R: GitRepo>(&self, target: &Repo, git: &R, branch: &str) -> Result<Rc<coalesce::TargetSide>> {
1333	let now = now_ms();
1334	let key = refs_cache::usable(registry::refs_state(&target.id), now)
1335	.map(\|version\| (target.id.clone(), branch.to_owned(), version));
1336	if let Some(key) = &key
1337	&& let Some(side) = TARGETS.with(\|memo\| memo.borrow().get(key, now))
1338	{
1339	return Ok(side);
1340	}
1341	let history = match git.log(branch, 1).await?.first() {
1342	Some(head) => git.log(&head.hash, MAX_ANCESTRY).await?,
1343	None => Vec::new(),
1344	};
1345	let side = coalesce::TargetSide::new(history);
1346	if let Some(key) = key {
1347	TARGETS.with(\|memo\| memo.borrow_mut().put(key, side.clone(), now));
1348	}
1349	Ok(side)
1350	}
1351
1352	async fn head(&self, a: HeadArgs) -> Result<Option<String>> {
1353	let Some(repo) = self.registry.by_id(&a.repo_id).await? else {
1354	return Ok(None);
1355	};
1356	let branch = if a.branch.is_empty() { &repo.default_branch } else { &a.branch };
1357	let git = self.read_git(&repo).await?;
1358	Ok(git
1359	.log(branch, 1)
1360	.await?
1361	.into_iter()
1362	.next()
1363	.map(\|commit\| commit.hash))
1364	}
1365
1366	async fn delete_branch(&self, a: DeleteBranchArgs) -> Result<Outcome<bool>> {
1367	if !deletable_branch(&a.branch, a.head.as_deref()) {
1368	return Ok(Outcome::fail(
1369	FailureCode::Forbidden,
1370	"Only branches g1t made for itself can be deleted this way.",
1371	));
1372	}
1373	let Some(repo) = self.registry.by_id(&a.repo_id).await? else {
1374	return Ok(not_found());
1375	};
1376	if a.branch == repo.default_branch {
1377	return Ok(Outcome::fail(FailureCode::Forbidden, "The default branch is never deleted."));
1378	}
1379	let repo = match self.unpaused(repo).await? {
1380	Ok(repo) => repo,
1381	Err((code, message)) => return Ok(Outcome::fail(code, message)),
1382	};
1383	self.live(&repo).await?;
1384	let git = self.store.open(&store_key(&repo)).await?;
1385	let Some(old) = git
1386	.branches()
1387	.await?
1388	.into_iter()
1389	.find(\|branch\| branch.name == a.branch)
1390	.map(\|branch\| branch.hash)
1391	else {
1392	return Ok(Outcome::Ok(false));
1393	};
1394	// Moved since the caller looked: someone else's commits are on it.
1395	if a.head.as_deref().is_some_and(\|head\| head != old) {
1396	return Ok(Outcome::fail(FailureCode::Conflict, format!("{} moved, so it was left alone.", a.branch)));
1397	}
1398	let access = git.access(Scope::Write).await?;
1399	let deleted = land::delete_ref(&access, &a.branch, &old).await?;
1400	self.refs_moved(&repo.id).await;
1401	if let Err(reason) = deleted {
1402	return Ok(Outcome::fail(
1403	FailureCode::Conflict,
1404	format!("{} could not be deleted: {reason}", a.branch),
1405	));
1406	}
1407	Ok(Outcome::Ok(true))
1408	}
1409
1410	async fn fork_for_pull(&self, a: ForkArgs) -> Result<Outcome<Repo>> {
1411	let viewer = Some(a.actor.clone());
1412	let Some(source) = self
1413	.registry
1414	.by_id(&a.source_id)
1415	.await?
1416	.filter(\|repo\| can_read(repo, &viewer))
1417	else {
1418	return Ok(not_found());
1419	};
1420	if let Some((code, message)) = lifecycle::read_only_refusal(&source) {
1421	return Ok(Outcome::fail(code, message));
1422	}
1423	// Its working copy is made in its namespace: not while it moves.
1424	let source = match self.unpaused(source).await? {
1425	Ok(source) => source,
1426	Err((code, message)) => return Ok(Outcome::fail(code, message)),
1427	};
1428	let now = now_ms();
1429	let fork = Repo {
1430	id: new_id("rep", now),
1431	namespace: PULLS_NAMESPACE.to_owned(),
1432	name: a.pull_id.clone(),
1433	description: None,
1434	// A fork is exactly as visible as the repo it came from.
1435	is_private: source.is_private,
1436	owner_id: a.actor.id.clone(),
1437	default_branch: source.default_branch.clone(),
1438	fork_of: Some(source.id.clone()),
1439	parent_id: None,
1440	forked_from: None,
1441	parent: None,
1442	allow_forking: false,
1443	max_ref_updates: None,
1444	protected: false,
1445	created_at: rfc3339(now),
1446	topics: Vec::new(),
1447	website: None,
1448	archived_at: None,
1449	};
1450	// Artifacts forks within a namespace: the copy goes where its
1451	// repository is.
1452	let (namespace, _) = store::locate(&store_key(&source));
1453	self.registry
1454	.claim_store_key(&fork, Some(&namespace), &self.store.default_namespace())
1455	.await?;
1456	self.store
1457	.open(&store_key(&source))
1458	.await?
1459	.fork(&store_key(&fork), true)
1460	.await?;
1461	self.registry.insert(&fork).await?;
1462	self.publish(NewEvent {
1463	kind: "repo.forked",
1464	source: SOURCE,
1465	repo_id: Some(source.id.clone()),
1466	actor: Some(a.actor.id),
1467	data: RepoForked {
1468	repo_id: fork.id.clone(),
1469	source_repo_id: source.id,
1470	pull_id: Some(a.pull_id),
1471	},
1472	})
1473	.await?;
1474	Ok(Outcome::Ok(fork))
1475	}
1476
1477	async fn git_access(&self, a: GitAccessArgs) -> Result<Outcome<GitAccess>> {
1478	let found = self.registry.by_path(&a.path).await?;
1479	Ok(match self.authorize_git(&a.path, &a.viewer, a.service, found).await? {
1480	Outcome::Ok(repo) => {
1481	self.live(&repo).await?;
1482	let write = a.service == GitService::ReceivePack;
1483	if write {
1484	// A push with this credential would not pass through
1485	// here, so nothing that lists the refs is kept until it
1486	// has expired (see refs_cache.rs).
1487	let until = now_ms() + store::CREDENTIAL_LIFE_MS + 60_000;
1488	if let Err(error) = self.registry.refs_open(&repo.id, until).await {
1489	// Before the column exists nothing is kept anyway.
1490	if registry::refs_state(&repo.id).is_some() {
1491	return Err(error);
1492	}
1493	}
1494	}
1495	let scope = if write { Scope::Write } else { Scope::Read };
1496	Outcome::Ok(self.store.handout(&store_key(&repo), scope).await?)
1497	}
1498	Outcome::Fail(failure) => Outcome::Fail(failure),
1499	})
1500	}
1501
1502	/// The repository at `path` (`found`, as just read), if the viewer may
1503	/// use `service` on it: fetch from it, or push to it. A push to a path
1504	/// with nothing there makes the repository, in a workspace the pusher
1505	/// belongs to.
1506	async fn authorize_git(
1507	&self,
1508	path: &RepoPath,
1509	viewer: &Viewer,
1510	service: GitService,
1511	found: Option<Repo>,
1512	) -> Result<Outcome<Repo>> {
1513	let mut a = GitAccessArgs {
1514	path: path.clone(),
1515	viewer: viewer.clone(),
1516	service,
1517	};
1518	let write = a.service == GitService::ReceivePack;
1519	// An access token: pushing needs code:write, reading a private
1520	// repository code:read. A public repository reads as it would for
1521	// anyone. Which repositories a token reaches is its owner's, checked
1522	// below as for anyone.
1523	if let Some(access) = a.viewer.as_ref().and_then(\|user\| user.token.as_deref()).cloned() {
1524	// A workflow job's token, and a deploy key, reach their own
1525	// repository only; a job's also the working copies of that
1526	// repository's pull requests, where their heads are.
1527	let name = format!("{}/{}", path.namespace, path.name);
1528	let source = match found.as_ref().and_then(\|repo\| repo.fork_of.as_deref()) {
1529	Some(source_id) if g1t_contracts::scopes::decide_repo(&access, &name).is_some() => self
1530	.registry
1531	.by_id(source_id)
1532	.await?
1533	.map(\|source\| format!("{}/{}", source.namespace, source.name)),
1534	_ => None,
1535	};
1536	let public = found.as_ref().is_some_and(\|repo\| !repo.is_private);
1537	if let Some(why) = git_token_refusal(&access, &name, source.as_deref(), write, public, found.is_some()) {
1538	return Ok(Outcome::fail(FailureCode::Forbidden, format!("{why}\n")));
1539	}
1540	if !write && !access.allows(g1t_contracts::scopes::Scope::CodeRead) {
1541	a.viewer = None;
1542	}
1543	}
1544
1545	// Anonymous callers are asked to authenticate whether or not the repo
1546	// exists, so private repos cannot be told apart from missing ones.
1547	let denied = \|\| match &a.viewer {
1548	Some(_) => not_found(),
1549	None => Outcome::fail(FailureCode::Unauthenticated, "Authentication required."),
1550	};
1551	// A run's token works through the API only: its sandbox has its
1552	// own way to push, to its own pull request.
1553	if a.viewer.as_ref().is_some_and(\|user\| user.kind == PrincipalKind::Agent) {
1554	return Ok(Outcome::fail(
1555	FailureCode::Forbidden,
1556	"A run's token cannot be used with git this way.",
1557	));
1558	}
1559	if let (true, Some(user)) = (write, &a.viewer)
1560	&& !user.verified
1561	{
1562	return Ok(Outcome::fail(FailureCode::Forbidden, UNVERIFIED));
1563	}
1564	let repo = match found {
1565	Some(repo) => {
1566	let allowed = if write {
1567	can_write(&repo, &a.viewer)
1568	} else {
1569	self.may_read(&repo, &a.viewer).await?
1570	};
1571	if !allowed {
1572	return Ok(denied());
1573	}
1574	// An archived repository, or a pull request's copy of one,
1575	// is read-only.
1576	if write {
1577	let archived = match &repo.fork_of {
1578	Some(source) => self.registry.by_id(source).await?,
1579	None => Some(repo.clone()),
1580	};
1581	match archived {
1582	Some(source) => {
1583	if let Some((code, message)) = lifecycle::read_only_refusal(&source) {
1584	return Ok(Outcome::fail(code, format!("{message}\n")));
1585	}
1586	}
1587	// The repository it was copied from is deleted.
1588	None => return Ok(denied()),
1589	}
1590	}
1591	repo
1592	}
1593	None => {
1594	// Push to create, in a workspace the pusher belongs to.
1595	let owner = a
1596	.viewer
1597	.as_ref()
1598	.filter(\|user\| write && user.is_member(&a.path.namespace.to_lowercase()));
1599	let Some(owner) = owner else {
1600	return Ok(denied());
1601	};
1602	let created = self.create(push_to_create(owner, &a.path)).await?;
1603	match created {
1604	Outcome::Ok(repo) => repo,
1605	Outcome::Fail(failure) => return Ok(Outcome::Fail(failure)),
1606	}
1607	}
1608	};
1609	// A push, or a credential to push with, waits while the repository
1610	// moves between namespaces (moves.rs), and goes to where it is now.
1611	if write {
1612	return Ok(match self.unpaused(repo).await? {
1613	Ok(repo) => Outcome::Ok(repo),
1614	Err((code, message)) => Outcome::fail(code, format!("{message}\n")),
1615	});
1616	}
1617	Ok(Outcome::Ok(repo))
1618	}
1619
1620	async fn land(&self, a: LandArgs) -> Result<Outcome<Landed>> {
1621	let actor: Viewer = Some(a.actor.clone());
1622	let Some(source) = self.registry.by_id(&a.source_id).await? else {
1623	return Ok(not_found());
1624	};
1625	// A working copy lands on the repository it came from, a fork's
1626	// branch on the repository named, and a branch on its own.
1627	let target = self.pull_target(&source, a.target_id.as_deref()).await?;
1628	let Some(target) = target.filter(\|repo\| can_read(repo, &actor)) else {
1629	return Ok(not_found());
1630	};
1631	if !registry::can(&target, &actor, Capability::Merge) {
1632	return Ok(Outcome::fail(
1633	FailureCode::Forbidden,
1634	access::needs(Capability::Merge, &format!("{}/{}", target.namespace, target.name)),
1635	));
1636	}
1637	if !a.actor.verified {
1638	return Ok(Outcome::fail(FailureCode::Forbidden, UNVERIFIED));
1639	}
1640	if let Some((code, message)) = lifecycle::read_only_refusal(&target) {
1641	return Ok(Outcome::fail(code, message));
1642	}
1643	// Moving between namespaces: wait for it (moves.rs). Both are read
1644	// again once it is done, for their new keys.
1645	let (source, target) = match (self.unpaused(source).await?, self.unpaused(target).await?) {
1646	(Ok(source), Ok(target)) => (source, target),
1647	(Err((code, message)), _) \| (_, Err((code, message))) => return Ok(Outcome::fail(code, message)),
1648	};
1649
1650	let branch = &a.target_branch.clone().unwrap_or_else(\|\| target.default_branch.clone());
1651	let from_fork = source.id != target.id;
1652	let source_branch = match a.branch.clone() {
1653	Some(name) if !from_fork && name == *branch => {
1654	return Ok(Outcome::fail(
1655	FailureCode::Invalid,
1656	format!("{branch} cannot be merged into itself."),
1657	));
1658	}
1659	Some(name) => name,
1660	// A working copy holds the change on the branch it merges into.
1661	None if source.fork_of.is_some() => branch.clone(),
1662	None => {
1663	return Ok(Outcome::fail(
1664	FailureCode::Invalid,
1665	"Say which branch to merge.",
1666	));
1667	}
1668	};
1669
1670	self.live(&source).await?;
1671	let source_git = self.store.open(&store_key(&source)).await?;
1672	let target_git = self.store.open(&store_key(&target)).await?;
1673	let history = source_git.log(&source_branch, MAX_ANCESTRY).await?;
1674	let Some(new) = history.first().map(\|commit\| commit.hash.clone()) else {
1675	return Ok(Outcome::fail(
1676	FailureCode::Conflict,
1677	"This pull request has no commits to merge.",
1678	));
1679	};
1680	let Some(new) = land::checked_commit(&new, a.expected_commit.as_deref()) else {
1681	return Ok(Outcome::fail(FailureCode::Conflict, "The source branch changed after it was checked. Test it again before landing."));
1682	};
1683	// Every method but fast-forward makes commits (merge_methods.rs).
1684	let method = a.method.unwrap_or(g1t_contracts::rules::MergeMethod::FastForward);
1685	if method != g1t_contracts::rules::MergeMethod::FastForward {
1686	return self.land_by(method, &a, &target, &source_git, &target_git, branch, &history, new, from_fork).await;
1687	}
1688	let old = target_git
1689	.log(branch, 1)
1690	.await?
1691	.into_iter()
1692	.next()
1693	.map(\|commit\| commit.hash);
1694
1695	if old.as_deref() == Some(new.as_str()) {
1696	return Ok(Outcome::Ok(Landed {
1697	commit: new,
1698	previous: None,
1699	method,
1700	rebased: Vec::new(),
1701	}));
1702	}
1703	if a.expected_commit.is_some() && old.is_some() {
1704	// A prior attempt may have published this exact commit and stopped
1705	// before recording the merge; another change can already follow it.
1706	let target_history = target_git.log(branch, MAX_ANCESTRY).await?;
1707	if descends_from(&target_git, &target_history, &new).await? {
1708	return Ok(Outcome::Ok(Landed { commit: new, previous: None, method, rebased: Vec::new() }));
1709	}
1710	}
1711	// Moving the branch to a commit that does not descend from its
1712	// current head would discard whatever landed in between.
1713	if let Some(old) = &old
1714	&& !descends_from(&source_git, &history, old).await?
1715	{
1716	let remedy = if from_fork {
1717	format!("Pull {branch} into the pull request's fork, push, and merge again.")
1718	} else {
1719	format!("Merge {branch} into {source_branch}, push, and merge again.")
1720	};
1721	return Ok(Outcome::fail(
1722	FailureCode::Conflict,
1723	format!("{branch} has moved since this pull request was opened. {remedy}"),
1724	));
1725	}
1726
1727	// For a branch the objects are already in the target; sending them
1728	// again is harmless and keeps one way of moving a ref.
1729	let source_access = source_git.access(Scope::Read).await?;
1730	let target_access = target_git.access(Scope::Write).await?;
1731	let pushed =
1732	land::fast_forward(&source_access, &target_access, branch, old.as_deref(), &new)
1733	.await?;
1734	self.refs_moved(&target.id).await;
1735	if let Err(reason) = pushed {
1736	// Most often another pull request landed between the check and the push.
1737	return Ok(Outcome::fail(
1738	FailureCode::Conflict,
1739	format!("{branch} could not be updated: {reason}"),
1740	));
1741	}
1742	self.publish_push(
1743	&target,
1744	&format!("refs/heads/{branch}"),
1745	old.as_deref(),
1746	&new,
1747	Some(&a.actor),
1748	None,
1749	)
1750	.await?;
1751	Ok(Outcome::Ok(Landed {
1752	commit: new,
1753	previous: old,
1754	method,
1755	rebased: Vec::new(),
1756	}))
1757	}
1758
1759	async fn compare(&self, a: CompareArgs) -> Result<Outcome<Comparison>> {
1760	let found = self.registry.by_id(&a.repo_id).await?;
1761	let repo = match self.visible(found.clone(), &a.viewer).await? {
1762	Some(repo) => Some(repo),
1763	// A pull request from a private fork is reviewed by those who
1764	// read the repository it merges into, whether or not they can
1765	// read the fork: its head, a commit that repository holds as
1766	// well (`refs/pull/<n>/head`), compared with it, from a base it
1767	// holds. Nothing else of the fork is read this way.
1768	None => match (found, a.target_id.as_deref(), a.head.as_deref()) {
1769	(Some(fork), Some(target_id), Some(head))
1770	if is_commit_hash(head) && a.base.as_deref().is_none_or(is_commit_hash) =>
1771	{
1772	match self.visible(self.registry.by_id(target_id).await?, &a.viewer).await? {
1773	Some(target) if self.forked_from(&fork, target_id).await? => {
1774	let target_git = self.read_git(&target).await?;
1775	let mut held = !target_git.log(head, 1).await?.is_empty();
1776	if let Some(base) = a.base.as_deref() {
1777	held = held && !target_git.log(base, 1).await?.is_empty();
1778	}
1779	held.then_some(fork)
1780	}
1781	_ => None,
1782	}
1783	}
1784	_ => None,
1785	},
1786	};
1787	let Some(repo) = repo else {
1788	return Ok(not_found());
1789	};
1790	let git = self.read_git(&repo).await?;
1791	let head_ref = a.head.as_deref().unwrap_or(&repo.default_branch);
1792	// A pull request into another branch is compared from where it
1793	// left that branch.
1794	let base_branch = a.base_branch.clone();
1795	// The head's history is only searched when the base is worked out
1796	// from another branch.
1797	let depth = if a.base.is_some() \|\| is_commit_hash(head_ref) { 1 } else { MAX_ANCESTRY };
1798	let history = git.log(head_ref, depth).await?;
1799	let Some(head) = history.first() else {
1800	return Ok(Outcome::fail(
1801	FailureCode::Conflict,
1802	"There are no commits to compare.",
1803	));
1804	};
1805
1806	// Where the head's history meets the default branch of `against`,
1807	// or the branch asked for.
1808	let shared_with = async \|against: &Repo\| -> Result<Option<String>> {
1809	let against_git = self.read_git(against).await?;
1810	let branch = base_branch.as_deref().unwrap_or(&against.default_branch);
1811	let shared: HashSet<String> = against_git
1812	.log(branch, MAX_ANCESTRY)
1813	.await?
1814	.into_iter()
1815	.map(\|commit\| commit.hash)
1816	.collect();
1817	nearest_ancestor_in(&git, &history, &shared).await
1818	};
1819	// A fork's branch, compared from the repository it would merge into.
1820	let target = match a.target_id.as_deref().filter(\|id\| a.base.is_none() && *id != repo.id) {
1821	Some(id) if self.forked_from(&repo, id).await? => {
1822	match self.visible(self.registry.by_id(id).await?, &a.viewer).await? {
1823	Some(target) => Some(target),
1824	None => return Ok(not_found()),
1825	}
1826	}
1827	Some(_) => return Ok(not_found()),
1828	None => None,
1829	};
1830	let base = match (a.base, &repo.fork_of) {
1831	(Some(base), _) => Some(base),
1832	(None, _) if let Some(target) = &target => shared_with(target).await?,
1833	// A fork is compared with the last commit it shares with the
1834	// repository it came from.
1835	(None, Some(target_id)) => match self.registry.by_id(target_id).await? {
1836	Some(target) => shared_with(&target).await?,
1837	None => None,
1838	},
1839	// A branch, with the point where it left the default branch.
1840	// A single commit, with its first parent.
1841	(None, None) if is_commit_hash(head_ref) => head.parents.first().cloned(),
1842	(None, None) if head_ref != base_branch.as_deref().unwrap_or(&repo.default_branch) => {
1843	shared_with(&repo).await?
1844	}
1845	(None, None) => head.parents.first().cloned(),
1846	};
1847	let base_tree = match &base {
1848	Some(base) => git
1849	.log(base, 1)
1850	.await?
1851	.into_iter()
1852	.next()
1853	.map(\|commit\| commit.tree_hash),
1854	None => None,
1855	};
1856	// Only a few files of it, when asked: how the files a large
1857	// comparison lists without their lines are read, as a page reaches them.
1858	let only = a.paths.as_deref().map(\|paths\| &paths[..paths.len().min(MAX_COMPARED_PATHS)]);
1859	let (files, truncated) =
1860	diff::compare_trees(&git, base_tree.as_deref(), &head.tree_hash, only).await?;
1861	Ok(Outcome::Ok(Comparison {
1862	base,
1863	head: head.hash.clone(),
1864	files,
1865	truncated,
1866	}))
1867	}
1868
1869	/// Reports that `git_ref` of `repo` (a full ref) now points to `after`,
1870	/// moved by `actor` (marked when that was a workflow job's token).
1871	/// `commits` is how many the move brought, when the caller knows: one
1872	/// for a file committed on the site or a branch brought up to date.
1873	async fn publish_push(
1874	&self,
1875	repo: &Repo,
1876	git_ref: &str,
1877	before: Option<&str>,
1878	after: &str,
1879	actor: Option<&User>,
1880	commits: Option<u32>,
1881	) -> Result<()> {
1882	let caused_by_job = actor.and_then(g1t_contracts::events::job_run_of).map(str::to_owned);
1883	self.publish_git_push(repo, git_ref, before, after, actor.map(\|user\| user.id.clone()), false, caused_by_job, commits)
1884	.await
1885	}
1886
1887	/// `publish_push`, saying whether the push reached the store without
1888	/// being scanned for secrets first.
1889	#[allow(clippy::too_many_arguments)]
1890	async fn publish_git_push(
1891	&self,
1892	repo: &Repo,
1893	git_ref: &str,
1894	before: Option<&str>,
1895	after: &str,
1896	actor: Option<String>,
1897	unscanned: bool,
1898	caused_by_job: Option<String>,
1899	commits: Option<u32>,
1900	) -> Result<()> {
1901	self.publish(NewEvent {
1902	kind: "git.push",
1903	source: SOURCE,
1904	repo_id: Some(repo.id.clone()),
1905	actor,
1906	data: GitPush {
1907	repo_id: repo.id.clone(),
1908	git_ref: git_ref.to_owned(),
1909	before: before.map(str::to_owned),
1910	after: after.to_owned(),
1911	default_branch: git_ref.strip_prefix("refs/heads/")
1912	== Some(repo.default_branch.as_str()),
1913	unscanned,
1914	caused_by_job,
1915	commits,
1916	},
1917	})
1918	.await
1919	}
1920
1921	/// Git over HTTPS. Only what decides the answer happens before it:
1922	/// the repository, who is asking and whether they may, the free
1923	/// workspace limits, push protection, and the store's own answer. The
1924	/// audit entry and what a push changed are recorded once git has its
1925	/// answer. Each answer says how long its steps took (`Server-Timing`).
1926	async fn git_http(&self, request: Request, env: &Env, ctx: &Context) -> Result<Response> {
1927	let mut timing = git_http::Timing::start();
1928	let Some(git) = git_http::parse(&request.url()?) else {
1929	return Response::error("Not found", 404);
1930	};
1931	let response = match self.answer_git(request, &git, env, ctx, &mut timing).await {
1932	Ok(response) => response,
1933	// The git store is busy: git hears when to try again.
1934	Err(error) => match resilience::busy(&error.to_string()) {
1935	Some(busy) => git_http::busy_response(busy)?,
1936	None => return Err(error),
1937	},
1938	};
1939	timing.apply(response)
1940	}
1941
1942	async fn answer_git(
1943	&self,
1944	request: Request,
1945	git: &git_http::GitRequest,
1946	env: &Env,
1947	ctx: &Context,
1948	timing: &mut git_http::Timing,
1949	) -> Result<Response> {
1950	let write = git.service == GitService::ReceivePack;
1951	let get = request.method() == Method::Get;
1952	let identity = env.service("IDENTITY")?;
1953	// The repository and the caller's credentials, at once. A fetch may
1954	// go by the row as read a moment ago, for the same clone's next
1955	// request; a push always reads it. Anonymous callers cost nothing.
1956	let lookup = async {
1957	if write {
1958	self.registry.by_path(&git.path).await
1959	} else {
1960	self.registry.by_path_recent(&git.path).await
1961	}
1962	};
1963	let (found, viewer) =
1964	futures_util::future::join(lookup, git_http::viewer(&request, &identity)).await;
1965	let mut found = found?;
1966	timing.mark("repo");
1967	// A workspace alias staff set (identity's aliases.rs: `g1t` for
1968	// `flagon-io`) is answered in place, as the repository under the
1969	// workspace's slug: pushes and some clients do not follow
1970	// redirects. Everything after this sees only the workspace's slug.
1971	let aliased = match found {
1972	Some(_) => None,
1973	None => git_http::aliased(git, &identity).await?,
1974	};
1975	if let Some(aliased) = &aliased {
1976	found = if write {
1977	self.registry.by_path(&aliased.path).await?
1978	} else {
1979	self.registry.by_path_recent(&aliased.path).await?
1980	};
1981	timing.mark("alias");
1982	}
1983	let git = aliased.as_ref().unwrap_or(git);
1984	if found.is_none() {
1985	// A workspace that was renamed: git follows a redirect when it
1986	// first asks for refs, and uses the new address from then on.
1987	// A repository transferred to another workspace: the same, to
1988	// its new path. Fetches and pushes both follow either.
1989	let url = request.url()?;
1990	let (renamed, moved) = futures_util::future::join(
1991	git_http::renamed(&url, &identity),
1992	self.registry.resolve_moved(&git.path),
1993	)
1994	.await;
1995	timing.mark("moved");
1996	if let Some(location) = renamed? {
1997	return git_http::moved(&location, get);
1998	}
1999	if let Some(now) = moved?
2000	&& let Some(location) = git_http::transferred(&url, &now)
2001	{
2002	return git_http::moved(&location, get);
2003	}
2004	}
2005	let viewer = viewer?;
2006	// A person who has not confirmed their email address can do
2007	// nothing with git until they do: told so, not asked to sign in.
2008	if viewer.as_ref().is_some_and(g1t_contracts::User::awaits_confirmation) {
2009	let site = request.url()?.origin().ascii_serialization();
2010	return git_http::refuse(Outcome::<()>::fail(
2011	FailureCode::Forbidden,
2012	g1t_contracts::accounts::confirm_email_first(&site),
2013	));
2014	}
2015	// A run credential is checked against its grants, then acts as the
2016	// person it works for. See run_access.rs.
2017	let (request, viewer, audit) = match self.admit_git(request, git, viewer, found.as_ref()).await? {
2018	run_access::Admitted::Go { request, viewer, entry } => (request, viewer, entry),
2019	run_access::Admitted::Refused(response) => return Ok(response),
2020	};
2021	let mut after = AfterGit {
2022	audit,
2023	status: 0,
2024	message: None,
2025	push: None,
2026	};
2027	let repo = match self.authorize_git(&git.path, &viewer, git.service, found).await? {
2028	Outcome::Ok(repo) => repo,
2029	refused => {
2030	let response = git_http::refuse(refused)?;
2031	after.ended(response.status_code(), None);
2032	after.spawn(env, ctx);
2033	return Ok(response);
2034	}
2035	};
2036	// A pull request's working copy removed after it closed is made
2037	// again before git uses it (forks.rs).
2038	self.live(&repo).await?;
2039	timing.mark("access");
2040	// Clones check out the default branch g1t keeps, which can have
2041	// changed since the store made the repository.
2042	let default_branch = repo.fork_of.is_none().then(\|\| repo.default_branch.clone());
2043	let key = store_key(&repo);
2044	let scope = if write { Scope::Write } else { Scope::Read };
2045	let mut request = request;
2046	let protocol = refs_cache::protocol(request.headers().get("git-protocol")?.as_deref());
2047	// A fetch's POST is read here, to tell an `ls-refs` from a fetch of
2048	// objects; the store would have it read in full anyway.
2049	let body = if !write && !get { Some(request.bytes().await?) } else { None };
2050	// What it asks the store, for the meters (meters.rs).
2051	let call = git_ops::classify(git.service, git.endpoint, get, body.as_deref());
2052	// Answers kept from the usual store may name refs the fallback
2053	// store does not have (fallback.rs): none are used, or kept.
2054	let fallback = self.store.on_fallback(&key);
2055	// An answer that lists refs may have been kept: see refs_cache.rs.
2056	let kept_key = refs_cache::kind(git, get, protocol, body.as_deref())
2057	.filter(\|_\| !fallback)
2058	.zip(refs_cache::usable(registry::refs_state(&repo.id), now_ms()))
2059	.map(\|(kind, version)\| {
2060	refs_cache::Key::new(&repo.id, version, default_branch.as_deref(), protocol, &kind)
2061	});
2062	// A fresh clone's pack may have been kept too: see pack_cache.rs.
2063	// Under the same refs version, so never across a change to them.
2064	let pack_key = self
2065	.packs
2066	.as_ref()
2067	.filter(\|_\| !fallback)
2068	.and_then(\|_\| {
2069	let encoding = request.headers().get("content-encoding").ok().flatten();
2070	pack_cache::cacheable(git, get, protocol, encoding.as_deref(), body.as_deref())
2071	})
2072	.zip(refs_cache::usable(registry::refs_state(&repo.id), now_ms()))
2073	.map(\|(normalized, version)\| pack_cache::Key::new(&repo.id, version, &normalized));
2074	// A kept answer and the free workspace limits, with a kept
2075	// credential and, for a push, what the repository holds looked up
2076	// alongside. A kept answer goes back without waiting for the
2077	// credential, which it does not need.
2078	let pushing = write && !get;
2079	let ((answer, pack, (limited, held)), kept_access) = {
2080	let shared = self.shared.as_deref();
2081	let answer_and_limits = std::pin::pin!(futures_util::future::join3(
2082	async {
2083	match &kept_key {
2084	Some(kept_key) => refs_cache::get(shared, kept_key).await,
2085	None => None,
2086	}
2087	},
2088	async {
2089	match (&pack_key, self.packs.as_deref()) {
2090	(Some(pack_key), Some(packs)) => pack_cache::get(packs, pack_key).await,
2091	_ => None,
2092	}
2093	},
2094	futures_util::future::join(self.git_limits(call, git, &repo, env), async {
2095	if pushing { Some(self.held(&repo).await) } else { None }
2096	}),
2097	));
2098	let kept_access = std::pin::pin!(self.store.kept_access(&key, scope));
2099	match futures_util::future::select(answer_and_limits, kept_access).await {
2100	futures_util::future::Either::Left((first, kept_access)) => {
2101	let answered = first.0.is_some() \|\| first.1.is_some() \|\| matches!(first.2.0, Ok(Some(_)) \| Err(_));
2102	(first, if answered { None } else { kept_access.await })
2103	}
2104	futures_util::future::Either::Right((kept_access, first)) => (first.await, kept_access),
2105	}
2106	};
2107	timing.mark("kept");
2108	// A kept pack first: it never reaches the store, so it is never an
2109	// operation, and a free workspace past its operation cap still gets
2110	// it. (The other limits are a push's, and a pack is only a fetch.)
2111	if let Some(kept) = pack {
2112	timing.note("pack", "hit");
2113	let sent = body.as_ref().map_or(0, \|body\| body.len() as u64);
2114	meters::record(pack_cache::HIT, &key, sent, kept.size);
2115	after.ended(200, None);
2116	after.spawn(env, ctx);
2117	return kept.response();
2118	}
2119	if let Some((response, status, message)) = limited? {
2120	after.ended(status, Some(message.to_owned()));
2121	after.spawn(env, ctx);
2122	return Ok(response);
2123	}
2124	if pack_key.is_some() {
2125	timing.note("pack", "miss");
2126	}
2127	if let (Some((entry, found)), Some(kept_key)) = (answer, &kept_key) {
2128	timing.note("refs", found.as_str());
2129	if found == refs_cache::Found::Shared {
2130	let (kept_key, entry) = (kept_key.clone(), entry.clone());
2131	ctx.wait_until(async move { refs_cache::keep_in_colo(&kept_key, &entry).await });
2132	}
2133	// Never reached the store: never an operation.
2134	meters::record(call.cached_meter(), &key, 0, entry.body.len() as u64);
2135	after.ended(200, None);
2136	after.spawn(env, ctx);
2137	return entry.response();
2138	}
2139	if kept_key.is_some() {
2140	timing.note("refs", "miss");
2141	}
2142	// An anonymous fetch the store is about to answer: an operation for
2143	// the repository's workspace, so limited per repository (limits.rs).
2144	if limits::counts_as_anonymous_fetch(call, viewer.is_none())
2145	&& g1t_kit::limits::check(env, limits::ANONYMOUS_FETCH, repo.id.clone()).await.limited()
2146	{
2147	let response = limits::too_many_anonymous_fetches(&format!("{}/{}", repo.namespace, repo.name))?;
2148	after.ended(429, Some("Too many anonymous fetches.".to_owned()));
2149	after.spawn(env, ctx);
2150	return Ok(response);
2151	}
2152	// The store's credential: one made a moment ago, here or in another
2153	// isolate (see store.rs), or a new one.
2154	let access = match kept_access {
2155	Some((access, from)) => {
2156	timing.note("cred", from.as_str());
2157	access
2158	}
2159	None => {
2160	let access = self.store.mint_access(&key, scope).await?;
2161	timing.mark("mint");
2162	timing.note("cred", "mint");
2163	access
2164	}
2165	};
2166	// Should the store turn a kept credential down, a fetch's first
2167	// request is tried again with a new one; the requests after it then
2168	// have that one too.
2169	let again = if get { Some(request.clone()?) } else { None };
2170	// A push's checks: workflow files, the rules of the branches and
2171	// tags it changes, saying which rule and why (rules.rs), and push
2172	// protection, which refuses a push that adds a secret
2173	// (secret_scan.rs). See push_checks.rs.
2174	let checks = async \|head: &[u8], whole: bool, scan: bool\| self.check_receive(&repo, viewer.as_ref(), head, whole, scan).await;
2175	// What a push may bring (pack_limits.rs): the repository's size is
2176	// its own and its pull requests' working copies'.
2177	let limits = if let Some(held) = held {
2178	git_http::PushLimits {
2179	held,
2180	repo_limit: self.repo_limit,
2181	large: self.large_pushes,
2182	..git_http::PushLimits::default()
2183	}
2184	} else {
2185	git_http::PushLimits::default()
2186	};
2187	let mut outcome = git_http::forward(
2188	request,
2189	body,
2190	git,
2191	&access,
2192	checks,
2193	default_branch.as_deref(),
2194	limits,
2195	timing,
2196	)
2197	.await?;
2198	let turned_down = matches!(
2199	&outcome,
2200	git_http::Push::Forwarded(forwarded) if matches!(forwarded.response.status_code(), 401 \| 403)
2201	);
2202	if turned_down {
2203	self.store.forget_access(&key).await;
2204	if let Some(again) = again {
2205	let access = self.store.mint_access(&key, scope).await?;
2206	outcome = git_http::forward(
2207	again,
2208	None,
2209	git,
2210	&access,
2211	async \|_: &[u8], _: bool, _: bool\| Ok(push_checks::Checks::clear()),
2212	default_branch.as_deref(),
2213	git_http::PushLimits::default(),
2214	timing,
2215	)
2216	.await?;
2217	}
2218	}
2219	let forwarded =
2220	match outcome {
2221	git_http::Push::Forwarded(forwarded) => forwarded,
2222	git_http::Push::Refused(response) => {
2223	after.ended(403, Some("The push was declined by rules.".to_owned()));
2224	after.spawn(env, ctx);
2225	return Ok(response);
2226	}
2227	git_http::Push::Blocked(response) => {
2228	after.ended(403, Some("The push adds a secret.".to_owned()));
2229	after.spawn(env, ctx);
2230	return Ok(response);
2231	}
2232	git_http::Push::Declined(response, reason) => {
2233	after.ended(403, Some(format!("The push was declined: {reason}.")));
2234	after.spawn(env, ctx);
2235	return Ok(response);
2236	}
2237	};
2238	if forwarded.from_store {
2239	let received = forwarded
2240	.response
2241	.headers()
2242	.get("content-length")?
2243	.and_then(\|length\| length.parse().ok())
2244	.unwrap_or(0);
2245	meters::record(call.meter(), &key, forwarded.sent, received);
2246	}
2247	timing.mark("store");
2248	let mut response = forwarded.response;
2249	let status = response.status_code();
2250	if write && !get {
2251	// A push: the store has moved its refs once it has answered in
2252	// full, so the answer is read before the change is recorded, and
2253	// only then goes back. Whoever fetches after it sees the push.
2254	let headers = response.headers().clone();
2255	headers.delete("content-length")?;
2256	let report = response.bytes().await?;
2257	self.refs_moved(&repo.id).await;
2258	timing.mark("refs");
2259	response = Response::from_bytes(report)?.with_headers(headers).with_status(status);
2260	} else if let (Some(kept_key), 200) = (&kept_key, status) {
2261	// A miss: this answer is kept for the next to ask.
2262	let headers = response.headers().clone();
2263	headers.delete("content-length")?;
2264	let body = response.bytes().await?;
2265	if let Some(content_type) = headers.get("content-type")? {
2266	let entry = refs_cache::Entry { content_type, body: body.clone() };
2267	if entry.keepable() {
2268	let shared = self.shared.clone();
2269	let kept_key = kept_key.clone();
2270	ctx.wait_until(async move { refs_cache::keep(shared.as_deref(), &kept_key, &entry).await });
2271	}
2272	}
2273	response = Response::from_bytes(body)?.with_headers(headers).with_status(status);
2274	} else if let (Some(pack_key), Some(packs), true) = (&pack_key, &self.packs, forwarded.from_store) {
2275	// A fresh clone the bucket did not have: counted, and its pack
2276	// kept as it streams to git, when it is a whole one.
2277	meters::record(pack_cache::MISS, &key, forwarded.sent, 0);
2278	// Past the repository's limit on writes to the bucket, the pack
2279	// goes to git without being kept (limits.rs).
2280	if status == 200 && !g1t_kit::limits::check(env, limits::PACK_FILL, repo.id.clone()).await.limited() {
2281	let store_key = key.clone();
2282	let measured = Box::new(move \|bytes: u64\| meters::record_bytes(pack_cache::MISS, &store_key, 0, bytes));
2283	let (teed, filling) = pack_cache::tee(response, packs.clone(), pack_key, measured)?;
2284	response = teed;
2285	if let Some(filling) = filling {
2286	let pack_key = pack_key.clone();
2287	ctx.wait_until(async move {
2288	let filled = filling.await;
2289	if !matches!(filled, pack_cache::Filled::Kept { .. } \| pack_cache::Filled::Abandoned) {
2290	worker::console_warn!("pack {} not kept: {filled:?}", pack_key.as_str());
2291	}
2292	});
2293	}
2294	}
2295	}
2296	after.ended(status, None);
2297	if status == 200 && (forwarded.pack_bytes > 0 \|\| !forwarded.pushed.is_empty() \|\| !forwarded.deleted.is_empty()) {
2298	after.push = Some(PushDone {
2299	repo,
2300	pushed: forwarded.pushed,
2301	deleted: forwarded.deleted,
2302	pack_bytes: forwarded.pack_bytes,
2303	caused_by_job: viewer.as_ref().and_then(g1t_contracts::events::job_run_of).map(str::to_owned),
2304	// The calendar credits people, never a job's, an agent's or a workspace's token.
2305	credit: viewer
2306	.as_ref()
2307	.filter(\|user\| user.kind == PrincipalKind::User && g1t_contracts::events::job_run_of(user).is_none())
2308	.map(\|user\| user.id.clone()),
2309	actor: viewer.map(\|user: User\| user.id),
2310	unscanned: forwarded.unscanned,
2311	});
2312	}
2313	after.spawn(env, ctx);
2314	Ok(response)
2315	}
2316
2317	/// The answer for a request a free workspace's limits stop, or a push
2318	/// to a full repository, with its status and reason for the audit log;
2319	/// `None` to go on.
2320	///
2321	/// A clone, fetch or push is a git operation, which the git store
2322	/// charges g1t for: counted for billing once the answer has gone back
2323	/// (meters.rs), and a free workspace far past its share is slowed down
2324	/// rather than charged (see git_ops.rs). Whether it is past it is
2325	/// decided from counts this isolate already holds: the database is not
2326	/// asked on the way. A free workspace is never charged for private
2327	/// storage: once its private repositories hold the free amount, pushes
2328	/// to them stop, checked when a push begins so that git shows the
2329	/// reason. So do pushes to a repository at the store's size limit.
2330	async fn git_limits(
2331	&self,
2332	call: git_ops::GitCall,
2333	git: &git_http::GitRequest,
2334	repo: &Repo,
2335	env: &Env,
2336	) -> Result<Option<(Response, u16, &'static str)>> {
2337	let namespace = git.path.namespace.to_lowercase();
2338	if meters::mapping_now().billable(call.meter()) > 0.0 {
2339	let now = now_ms();
2340	let hour = git_ops::hour_key(&rfc3339(now));
2341	let limits = git_ops::Limits::from_env(env);
2342	if let Some((month, hour_ops)) = git_ops::standing(&namespace, &hour, now)
2343	&& git_ops::slow_down(month + 1, hour_ops + 1, limits.free_cap, limits.hourly)
2344	&& git_ops::is_free_kept(env.service("BILLING").ok().as_ref(), &namespace).await
2345	{
2346	return Ok(Some((
2347	git_ops::too_many(&namespace, limits.free_cap, limits.hourly)?,
2348	429,
2349	"Too many git operations this hour.",
2350	)));
2351	}
2352	}
2353	if git.service == GitService::ReceivePack && git.endpoint == "info/refs" {
2354	let held = self.held(repo).await;
2355	if held >= self.repo_limit {
2356	let message = format!(
2357	"{}/{} 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",
2358	repo.namespace,
2359	repo.name,
2360	pack_limits::megabytes(held)
2361	);
2362	return Ok(Some((Response::error(message, 403)?, 403, "The repository is full.")));
2363	}
2364	}
2365	if git.service == GitService::ReceivePack && git.endpoint == "info/refs" && repo.is_private {
2366	let free = git_ops::free_private_bytes(env);
2367	let held = self.registry.private_bytes(&namespace).await.unwrap_or(0);
2368	if git_ops::storage_full(held, free)
2369	&& git_ops::is_free(env.service("BILLING").ok().as_ref(), &namespace).await
2370	{
2371	return Ok(Some((
2372	git_ops::storage_full_response(&namespace, held, free)?,
2373	403,
2374	"Free private storage is full.",
2375	)));
2376	}
2377	}
2378	if git.service == GitService::ReceivePack && git.endpoint == "info/refs" && !repo.is_private {
2379	let free = git_ops::free_public_bytes(env);
2380	if free > 0 {
2381	let held = self.registry.public_bytes(&namespace).await.unwrap_or(0);
2382	if git_ops::public_storage_full(held, free)
2383	&& git_ops::is_free(env.service("BILLING").ok().as_ref(), &namespace).await
2384	{
2385	return Ok(Some((
2386	git_ops::public_storage_full_response(&namespace, held, free)?,
2387	403,
2388	"Free public storage is full.",
2389	)));
2390	}
2391	}
2392	}
2393	Ok(None)
2394	}
2395
2396	/// What a repository and its pull requests' working copies hold, as
2397	/// g1t counts it: read for a push's first request, kept a minute for
2398	/// the rest of it.
2399	async fn held(&self, repo: &Repo) -> u64 {
2400	let root = repo.fork_of.clone().unwrap_or_else(\|\| repo.id.clone());
2401	let now = now_ms();
2402	if let Some(held) = HELD.with(\|held\| held.borrow().get(&root, now)) {
2403	return held;
2404	}
2405	let held = self.registry.stored_bytes(&root).await.unwrap_or(0).max(0) as u64;
2406	HELD.with(\|kept\| kept.borrow_mut().put(root, held, now));
2407	held
2408	}
2409
2410	/// A `git.deleted` event for each branch or tag a push deleted that is
2411	/// gone from the store: the store can refuse one ref and accept
2412	/// another. As many as a push announces of tags and branches it moves.
2413	async fn publish_deleted(
2414	&self,
2415	repo: &Repo,
2416	deleted: &[(String, String)],
2417	actor: Option<&str>,
2418	caused_by_job: Option<&str>,
2419	) -> Result<()> {
2420	let stored = self.store.open(&store_key(repo)).await?;
2421	let is_branch = \|git_ref: &str\| git_ref.starts_with("refs/heads/");
2422	let tags = deleted.iter().filter(\|(git_ref, _)\| !is_branch(git_ref)).count();
2423	let announced = deleted
2424	.iter()
2425	.filter(\|(git_ref, _)\| is_branch(git_ref))
2426	.take(MAX_PUSH_BRANCH_EVENTS)
2427	.chain(deleted.iter().filter(\|(git_ref, _)\| !is_branch(git_ref) && tags <= MAX_PUSH_TAG_EVENTS));
2428	for (git_ref, before) in announced {
2429	let named = git_ref.strip_prefix("refs/heads/").unwrap_or(git_ref);
2430	let gone = match stored.log(named, 1).await {
2431	Ok(head) => head.is_empty(),
2432	Err(error) => {
2433	worker::console_error!("whether {git_ref} of {} was deleted is not known: {error}", repo.name);
2434	false
2435	}
2436	};
2437	if !gone {
2438	continue;
2439	}
2440	self.publish(NewEvent {
2441	kind: "git.deleted",
2442	source: SOURCE,
2443	repo_id: Some(repo.id.clone()),
2444	actor: actor.map(str::to_owned),
2445	data: GitDeleted {
2446	repo_id: repo.id.clone(),
2447	git_ref: git_ref.clone(),
2448	before: before.clone(),
2449	caused_by_job: caused_by_job.map(str::to_owned),
2450	},
2451	})
2452	.await?;
2453	}
2454	Ok(())
2455	}
2456
2457	/// What a push changed, recorded once git has its answer.
2458	async fn record_push(&self, push: PushDone) -> Result<()> {
2459	let PushDone {
2460	repo,
2461	pushed,
2462	deleted,
2463	pack_bytes,
2464	actor,
2465	unscanned,
2466	caused_by_job,
2467	credit,
2468	} = push;
2469	// What the push stored, for billing's storage meter. A failure only
2470	// leaves the count short.
2471	if pack_bytes > 0
2472	&& let Err(error) = self.registry.add_stored_bytes(&repo, pack_bytes).await
2473	{
2474	worker::console_error!("stored bytes for {} not counted: {error}", repo.name);
2475	}
2476	if !deleted.is_empty() {
2477	self.publish_deleted(&repo, &deleted, actor.as_deref(), caused_by_job.as_deref()).await?;
2478	}
2479	if pushed.is_empty() {
2480	return Ok(());
2481	}
2482	// Artifacts' own push notifications are per repository, which does
2483	// not fit a repo per pull request, so the front end reports pushes
2484	// itself: one event for each branch that moved.
2485	let stored = self.store.open(&store_key(&repo)).await?;
2486	let announced = announced_refs(&pushed, &repo.default_branch);
2487	if announced.len() < pushed.len() {
2488	worker::console_log!(
2489	"push to {}: {} of {} refs announced",
2490	repo.name,
2491	announced.len(),
2492	pushed.len()
2493	);
2494	}
2495	for pushed in announced {
2496	// The store can refuse one ref and accept another, so each
2497	// branch is checked against where it actually is. A tag the
2498	// store cannot read back is taken as pushed.
2499	let moved = match pushed.branch() {
2500	Some(branch) => stored
2501	.log(branch, 1)
2502	.await?
2503	.first()
2504	.is_some_and(\|commit\| commit.hash == pushed.after),
2505	None => stored.log(&pushed.git_ref, 1).await.map_or(true, \|head\| {
2506	head.first().is_none_or(\|commit\| commit.hash == pushed.after)
2507	}),
2508	};
2509	// How many commits a branch push brought (push_commits.rs), for
2510	// the event and the person's contribution calendar. A log the
2511	// store cannot read leaves the event without a count and the
2512	// day short.
2513	let commits = match (moved, pushed.branch()) {
2514	(true, Some(_)) => match stored.log(&pushed.after, push_commits::MOST_PER_PUSH + 1).await {
2515	Ok(log) => Some(push_commits::new_commits(&log, pushed.before.as_deref())),
2516	Err(error) => {
2517	worker::console_error!("commits pushed to {} not counted: {error}", repo.name);
2518	None
2519	}
2520	},
2521	_ => None,
2522	};
2523	if let (Some(commits), Some(user_id), Some(branch)) = (commits, credit.as_deref(), pushed.branch())
2524	&& push_commits::counts(branch, &repo.default_branch)
2525	&& let Err(error) =
2526	push_commits::record(&self.registry.db, user_id, &repo.id, &rfc3339(now_ms())[..10], commits).await
2527	{
2528	worker::console_error!("commits pushed to {} not credited: {error}", repo.name);
2529	}
2530	if moved {
2531	self.publish_git_push(
2532	&repo,
2533	&pushed.git_ref,
2534	pushed.before.as_deref(),
2535	&pushed.after,
2536	actor.clone(),
2537	unscanned,
2538	caused_by_job.clone(),
2539	commits,
2540	)
2541	.await?;
2542	}
2543	}
2544	Ok(())
2545	}
2546	}
2547
2548	/// The most tags one push announces: a push of more (`git push --tags`
2549	/// into a new repository, say) announces none of them, so it starts no
2550	/// workflows or package reads, one per tag.
2551	const MAX_PUSH_TAG_EVENTS: usize = 3;
2552	/// The most branches one push announces. A push of more announces only
2553	/// the default branch, if it moved, which the rest of g1t reads from.
2554	const MAX_PUSH_BRANCH_EVENTS: usize = 1000;
2555
2556	/// The refs of a push that are announced with a `git.push` event each.
2557	/// Every ref is stored whatever this says; only the events are capped.
2558	fn announced_refs<'a>(pushed: &'a [git_http::Pushed], default_branch: &str) -> Vec<&'a git_http::Pushed> {
2559	let (branches, tags): (Vec<&git_http::Pushed>, Vec<&git_http::Pushed>) =
2560	pushed.iter().partition(\|pushed\| pushed.branch().is_some());
2561	let mut announced = if branches.len() > MAX_PUSH_BRANCH_EVENTS {
2562	branches.into_iter().filter(\|pushed\| pushed.branch() == Some(default_branch)).collect()
2563	} else {
2564	branches
2565	};
2566	if tags.len() <= MAX_PUSH_TAG_EVENTS {
2567	announced.extend(tags);
2568	}
2569	announced
2570	}
2571
2572	/// A push the store accepted, to be recorded once git has its answer.
2573	struct PushDone {
2574	repo: Repo,
2575	pushed: Vec<git_http::Pushed>,
2576	/// The branches and tags it asked to delete, each with the commit it
2577	/// pointed to.
2578	deleted: Vec<(String, String)>,
2579	pack_bytes: u64,
2580	actor: Option<String>,
2581	/// Too large to scan for secrets before it was stored.
2582	unscanned: bool,
2583	/// The run whose job's token pushed, if one did: its push starts no
2584	/// workflows.
2585	caused_by_job: Option<String>,
2586	/// The person credited with the push's commits on their contribution
2587	/// calendar: whoever pushed, when that is a person (push_commits.rs).
2588	credit: Option<String>,
2589	}
2590
2591	/// What a git request leaves for after its answer: its audit entry, with
2592	/// how the request ended, and what a push changed.
2593	struct AfterGit {
2594	audit: Option<Box<g1t_contracts::audit::NewAuditEntry>>,
2595	status: u16,
2596	message: Option<String>,
2597	push: Option<PushDone>,
2598	}
2599
2600	impl AfterGit {
2601	fn ended(&mut self, status: u16, message: Option<String>) {
2602	self.status = status;
2603	self.message = message;
2604	}
2605
2606	/// Does the work once the response is on its way. A failure is logged:
2607	/// git has already been told how its request went.
2608	fn spawn(self, env: &Env, ctx: &Context) {
2609	if self.audit.is_none() && self.push.is_none() {
2610	return;
2611	}
2612	let env = env.clone();
2613	ctx.wait_until(async move {
2614	let repos = match service(&env) {
2615	Ok(repos) => repos,
2616	Err(error) => {
2617	worker::console_error!("git request not recorded: {error}");
2618	return;
2619	}
2620	};
2621	repos.finish_git(self.audit, self.status, self.message).await;
2622	if let Some(push) = self.push
2623	&& let Err(error) = repos.record_push(push).await
2624	{
2625	worker::console_error!("push not recorded: {error}");
2626	}
2627	repos.deferred.settle().await;
2628	});
2629	}
2630	}
2631
2632	fn service(env: &Env) -> Result<Repos<ArtifactsStore>> {
2633	let shared = shared::Shared::from_env(env).map(Rc::new);
2634	let deferred = Rc::new(store::Deferred::default());
2635	Ok(Repos {
2636	registry: Registry { db: env.d1("DB")? },
2637	store: ArtifactsStore::new(env, shared.clone(), deferred.clone())?,
2638	deferred,
2639	shared,
2640	packs: pack_cache::Packs::from_env(env).map(Rc::new),
2641	events: env.service("EVENTS")?,
2642	security: env.service("SECURITY").ok(),
2643	billing: env.service("BILLING").ok(),
2644	identity: env.service("IDENTITY").ok(),
2645	work: env.service("WORK").ok(),
2646	free_private_bytes: git_ops::free_private_bytes(env),
2647	fork_days: forks::retention_days(env),
2648	repo_limit: env
2649	.var("REPO_STORAGE_LIMIT_BYTES")
2650	.ok()
2651	.and_then(\|value\| value.to_string().parse().ok())
2652	.unwrap_or(pack_limits::DEFAULT_REPO_LIMIT_BYTES),
2653	large_pushes: git_http::LargePushes::from_var(env.var("LARGE_PUSHES").ok().map(\|value\| value.to_string()).as_deref()),
2654	placement: shards::Placement::from_vars(
2655	env.var("GITSTORE_NEW_REPOS").ok().map(\|value\| value.to_string()).as_deref(),
2656	env.var("GITSTORE_EU_NAMESPACE").ok().map(\|value\| value.to_string()).as_deref(),
2657	),
2658	limits: shards::limits(env.var("GITSTORE_NAMESPACE_LIMITS").ok().map(\|value\| value.to_string()).as_deref()),
2659	})
2660	}
2661
2662	/// Writes what this isolate metered once the answer has gone back, every
2663	/// few seconds at most: now, or once it is due, waiting in this request's
2664	/// `wait_until` so nothing counted is left for a request that may never
2665	/// come (meters.rs).
2666	fn flush_later(env: &Env, ctx: &Context) {
2667	let Some(wait) = meters::plan_flush() else {
2668	return;
2669	};
2670	if let Ok(db) = env.d1("DB") {
2671	ctx.wait_until(async move { meters::flush_after(&db, wait).await });
2672	}
2673	}
2674
2675	/// `/backups/<job id>/parts/<number>`: the job and the part's number.
2676	fn backup_part_path(path: &str) -> Option<(String, u16)> {
2677	let rest = path.strip_prefix("/backups/")?;
2678	let (job, number) = rest.split_once("/parts/")?;
2679	let number = number.parse::<u16>().ok()?;
2680	(!job.is_empty() && !job.contains('/')).then(\|\| (job.to_owned(), number))
2681	}
2682
2683	fn backups_off<T>() -> Outcome<T> {
2684	Outcome::fail(FailureCode::Conflict, "Backups are off on this installation: it has no storage for them.")
2685	}
2686
2687	/// One part of a backup's bundle, with the job's token in its header.
2688	async fn backup_part(request: &mut Request, env: &Env, repos: &Repos<ArtifactsStore>, job_id: String, number: u16) -> Result<Response> {
2689	let Some(blobs) = backups::storage(env) else {
2690	return reply(&backups_off::<()>());
2691	};
2692	let token = request.headers().get(g1t_contracts::backups::TOKEN_HEADER)?.unwrap_or_default();
2693	let bytes = request.bytes().await?;
2694	let job = g1t_contracts::backups::BackupJobArgs { job_id, token };
2695	reply(&backups::part(&repos.registry.db, &blobs, &job, number, bytes).await?)
2696	}
2697
2698	#[cfg(test)]
2699	mod backup_path_tests {
2700	use super::backup_part_path;
2701
2702	#[test]
2703	fn a_part_is_named_by_its_job_and_number() {
2704	assert_eq!(backup_part_path("/backups/bkp_1/parts/3"), Some(("bkp_1".to_owned(), 3)));
2705	assert_eq!(backup_part_path("/backups/bkp_1/parts/x"), None);
2706	assert_eq!(backup_part_path("/backups//parts/1"), None);
2707	assert_eq!(backup_part_path("/acme/rocket.git/info/refs"), None);
2708	}
2709	}
2710
2711	/// Read methods whose answer is an `Outcome`: when the git store is busy,
2712	/// the site is told so in words instead of failing the page.
2713	const OUTCOME_READS: [&str; 7] = ["tree", "blob", "log", "branches", "blame", "compare", "branch_drift"];
2714
2715	#[event(fetch)]
2716	async fn fetch(mut request: Request, env: Env, ctx: Context) -> Result<Response> {
2717	let mut repos = service(&env)?;
2718	// A part of a backup's bundle, as the API passes it on from the
2719	// sandbox: bytes, not JSON (backups.rs).
2720	if request.method() == Method::Put
2721	&& let Some((job_id, number)) = backup_part_path(&request.path())
2722	{
2723	let answered = backup_part(&mut request, &env, &repos, job_id, number).await;
2724	repos.deferred.hand_over(&ctx);
2725	flush_later(&env, &ctx);
2726	return answered;
2727	}
2728	let Some(method) = rpc_method(&request) else {
2729	let answered = repos.git_http(request, &env, &ctx).await;
2730	repos.deferred.hand_over(&ctx);
2731	flush_later(&env, &ctx);
2732	return answered;
2733	};
2734	// A replica near the caller when it asks for one (crates/kit/src/d1.rs).
2735	// Git over HTTPS above always reads the primary.
2736	let (db, served) = g1t_kit::d1::open(&env, "DB", &request)?;
2737	repos.registry.db = db;
2738	let body: serde_json::Value = request.json().await?;
2739
2740	let answered = async { match method.as_str() {
2741	"get" => reply(&repos.get(args(body)?).await?),
2742	"get_by_id" => reply(&repos.get_by_id(args(body)?).await?),
2743	"readable" => {
2744	let a: ReadableArgs = args(body)?;
2745	reply(&repos.registry.readable(&a.ids, &a.viewer).await?)
2746	}
2747	// Work's contribution calendar: the commits a person pushed each
2748	// day, in repositories the viewer may read (push_commits.rs).
2749	"commit_days" => {
2750	let a: g1t_contracts::repos::CommitDaysArgs = args(body)?;
2751	reply(&push_commits::days(&repos.registry, &a.user_id, &a.since, &a.viewer).await?)
2752	}
2753	"public_namespaces" => {
2754	let a: PublicNamespacesArgs = args(body)?;
2755	reply(&repos.registry.public_namespaces(&a.owner_id).await?)
2756	}
2757	"path_by_id" => {
2758	let a: PathByIdArgs = args(body)?;
2759	reply(
2760	&repos
2761	.registry
2762	.by_id(&a.id)
2763	.await?
2764	.filter(\|repo\| repo.fork_of.is_none())
2765	.map(\|repo\| RepoPath {
2766	namespace: repo.namespace,
2767	name: repo.name,
2768	}),
2769	)
2770	}
2771	"list" => {
2772	let a: ListArgs = args(body)?;
2773	reply(
2774	&repos
2775	.registry
2776	.list(
2777	&a.viewer,
2778	a.query.as_deref(),
2779	a.namespace.as_deref(),
2780	a.member_only,
2781	)
2782	.await?,
2783	)
2784	}
2785	"create" => reply(&repos.create(args(body)?).await?),
2786	"transfer" => reply(&repos.transfer(args(body)?).await?),
2787	// A repository's lifecycle: see lifecycle.rs.
2788	"delete" => reply(&repos.delete(args(body)?).await?),
2789	"deleted" => reply(&repos.deleted(args(body)?).await?),
2790	"restore" => reply(&repos.restore(args(body)?).await?),
2791	"purge" => reply(&repos.purge(args(body)?).await?),
2792	"purge_due" => reply(&repos.purge_due(args(body)?).await?),
2793	"rename" => reply(&repos.rename(args(body)?).await?),
2794	"archive" => reply(&repos.archive(args(body)?).await?),
2795	"set_visibility" => reply(&repos.set_visibility(args(body)?).await?),
2796	"set_default_branch" => reply(&repos.set_default_branch(args(body)?).await?),
2797	"rename_branch" => reply(&repos.rename_branch(args(body)?).await?),
2798	"create_branch" => reply(&repos.create_branch(args(body)?).await?),
2799	"remove_branch" => reply(&repos.remove_branch(args(body)?).await?),
2800	"resolve_branch" => reply(&repos.resolve_branch(args(body)?).await?),
2801	"status_by_id" => reply(&repos.status_by_id(args(body)?).await?),
2802	"resolve_path" => {
2803	let a: ResolvePathArgs = args(body)?;
2804	reply(&repos.registry.resolve_moved(&a.path).await?)
2805	}
2806	"namespace_count" => {
2807	let a: NamespaceCountArgs = args(body)?;
2808	reply(&repos.registry.count_in(&a.namespace).await?)
2809	}
2810	"update" => reply(&repos.update(args(body)?).await?),
2811	"tree" => reply(&repos.tree(args(body)?).await?),
2812	"blob" => reply(&repos.blob(args(body)?).await?),
2813	"log" => reply(&repos.log(args(body)?).await?),
2814	"blame" => reply(&repos.blame(args(body)?).await?),
2815	"fork_for_pull" => reply(&repos.fork_for_pull(args(body)?).await?),
2816	// Forks people make into their workspaces: see network.rs.
2817	"fork" => reply(&repos.fork(args(body)?).await?),
2818	"forks" => reply(&repos.forks(args(body)?).await?),
2819	"sync_fork" => reply(&repos.sync_fork(args(body)?).await?),
2820	"detach_fork" => reply(&repos.detach_fork(args(body)?).await?),
2821	"git_access" => reply(&repos.git_access(args(body)?).await?),
2822	"branches" => reply(&repos.branches(args(body)?).await?),
2823	"last_commits" => reply(&repos.last_commits(args(body)?).await?),
2824	"path_log" => reply(&repos.path_log(args(body)?).await?),
2825	"branch_drift" => reply(&repos.branch_drift(args(body)?).await?),
2826	"tags" => reply(&repos.tags(args(body)?).await?),
2827	// The About: what the Files page shows beside the files (about.rs).
2828	// What is kept behind the head is worked out again after the answer.
2829	"about" => {
2830	let (answer, refresh) = repos.about(args(body)?).await?;
2831	about::refresh_later(&env, &ctx, refresh);
2832	reply(&answer)
2833	}
2834	"languages" => {
2835	let (answer, refresh) = repos.languages(args(body)?).await?;
2836	about::refresh_later(&env, &ctx, refresh);
2837	reply(&answer)
2838	}
2839	"contributors" => {
2840	let (answer, refresh) = repos.contributors(args(body)?).await?;
2841	about::refresh_later(&env, &ctx, refresh);
2842	reply(&answer)
2843	}
2844	"license" => {
2845	let (answer, refresh) = repos.license(args(body)?).await?;
2846	about::refresh_later(&env, &ctx, refresh);
2847	reply(&answer)
2848	}
2849	"stars" => reply(&repos.stars(args(body)?).await?),
2850	"star" => reply(&repos.star(args(body)?).await?),
2851	"stargazers" => reply(&repos.stargazers(args(body)?).await?),
2852	"starred" => reply(&repos.starred(args(body)?).await?),
2853	"releases" => reply(&repos.releases(args(body)?).await?),
2854	"release" => reply(&repos.release(args(body)?).await?),
2855	"create_release" => reply(&repos.create_release(args(body)?).await?),
2856	"update_release" => reply(&repos.update_release(args(body)?).await?),
2857	"delete_release" => reply(&repos.delete_release(args(body)?).await?),
2858	"head" => reply(&repos.head(args(body)?).await?),
2859	"behind" => reply(&repos.behind(args(body)?).await?),
2860	"divergence" => reply(&repos.divergence(args(body)?).await?),
2861	"land" => reply(&repos.land(args(body)?).await?),
2862	"update_pull_branch" => reply(&repos.update_pull_branch(args(body)?).await?),
2863	// Services only: a pull request's test merge commit (test_merge.rs).
2864	"pull_merge_ref" => reply(&repos.pull_merge_ref(args(body)?).await?),
2865	"delete_branch" => reply(&repos.delete_branch(args(body)?).await?),
2866	"commit_file" => reply(&repos.commit_file(args(body)?).await?),
2867	"compare" => reply(&repos.compare(args(body)?).await?),
2868	// Services only: a pull request's commits, as rules look at them (rules.rs).
2869	"inspect_commits" => reply(&repos.inspect_commits(args(body)?).await?),
2870	"scan_history" => reply(&repos.scan_history(args(body)?).await?),
2871	"find_lockfiles" => reply(&repos.find_lockfiles(args(body)?).await?),
2872	"match_pattern" => reply(&repos.match_pattern(args(body)?).await?),
2873	"check_secret" => reply(&repos.check_secret(args(body)?).await?),
2874	"list_files" => reply(&repos.list_files(args(body)?).await?),
2875	// The API's tarball and zipball links, and the site opening them (archive_links.rs).
2876	"archive_link" => {
2877	let sealer = env.secret("REPOS_KEY").ok().and_then(\|key\| g1t_secrets::Sealer::new(&key.to_string()));
2878	reply(&repos.archive_link(args(body)?, sealer.as_ref()).await?)
2879	}
2880	"open_archive_token" => {
2881	let sealer = env.secret("REPOS_KEY").ok().and_then(\|key\| g1t_secrets::Sealer::new(&key.to_string()));
2882	reply(&repos.open_archive_token(args(body)?, sealer.as_ref()).await?)
2883	}
2884	"changed_files" => reply(&repos.changed_files(args(body)?).await?),
2885	"read_blobs" => reply(&repos.read_blobs(args(body)?).await?),
2886	// Services only: what the Composer registry builds packages from.
2887	"refs" => reply(&repos.refs_of(args(body)?).await?),
2888	"raw_file" => reply(&repos.raw_file(args(body)?).await?),
2889	"raw_blobs" => reply(&repos.raw_blobs(args(body)?).await?),
2890	"visibility" => {
2891	let a: g1t_contracts::repos::VisibilityArgs = args(body)?;
2892	reply(&repos.registry.visibility(&a.paths).await?)
2893	}
2894	"storage" => reply(&repos.registry.storage().await?),
2895	"git_operations" => {
2896	let a: GitOperationsArgs = args(body)?;
2897	reply(&git_ops::totals(&repos.registry.db, &a.month, a.since.as_deref(), a.namespace.as_deref().map(str::to_lowercase).as_deref()).await?)
2898	}
2899	// Identity, once: who created each repository (members.rs there).
2900	"repo_creators" => {
2901	let a: AllIdsArgs = args(body)?;
2902	let limit = a.limit.clamp(1, 500);
2903	let repos = repos.registry.creators_after(a.after.as_deref(), limit).await?;
2904	let next = (repos.len() == limit as usize).then(\|\| repos.last().map(\|repo\| repo.id.clone())).flatten();
2905	reply(&CreatorPage { repos, next })
2906	}
2907	"all_ids" => {
2908	let a: AllIdsArgs = args(body)?;
2909	let limit = a.limit.clamp(1, 500);
2910	let ids = repos.registry.ids_after(a.after.as_deref(), limit).await?;
2911	let next = (ids.len() == limit as usize).then(\|\| ids.last().cloned()).flatten();
2912	reply(&IdPage { ids, next })
2913	}
2914	// The raw meters of the git store, for reconciling with Cloudflare
2915	// (meters.rs, scripts/ops/gitstore-usage.mjs).
2916	"artifacts_usage" => {
2917	let a: meters::UsageArgs = args(body)?;
2918	reply(&meters::usage(&repos.registry.db, &a).await?)
2919	}
2920	"operation_mapping" => reply(&meters::read_mapping(&repos.registry.db).await?),
2921	// Billing: the workspace each pull request's working copy is counted
2922	// for, so Cloudflare's own count of `pulls--<id>` shares out too.
2923	"pull_owners" => {
2924	#[derive(serde::Deserialize)]
2925	struct PullOwnersArgs {
2926	pulls: Vec<String>,
2927	}
2928	let a: PullOwnersArgs = args(body)?;
2929	let pulls: Vec<String> = a.pulls.into_iter().take(500).collect();
2930	reply(&serde_json::json!({ "owners": meters::pull_owners(&repos.registry.db, &pulls).await? }))
2931	}
2932	// Services only: which meters are operations, changed without a deploy.
2933	"set_operation_mapping" => {
2934	let row: meters::MappingRow = args(body)?;
2935	meters::set_mapping(&repos.registry.db, &row, &rfc3339(now_ms())).await?;
2936	reply(&meters::read_mapping(&repos.registry.db).await?)
2937	}
2938	// Backups (backups.rs): the runner's sweep claims queued ones, and
2939	// each sandbox, through the API, asks for its job and says how it went.
2940	"claim_backups" => {
2941	let a: g1t_contracts::backups::ClaimBackupsArgs = args(body)?;
2942	let blobs = backups::storage(&env);
2943	reply(&backups::claim(&repos.registry.db, blobs.as_ref(), &a, now_ms()).await?)
2944	}
2945	"backup_spec" => match backups::storage(&env) {
2946	Some(blobs) => {
2947	let a: g1t_contracts::backups::BackupJobArgs = args(body)?;
2948	let every = backups::Settings::from_env(&env).full_every;
2949	reply(&backups::spec(&repos.registry, &blobs, &repos.store, &a, every, now_ms()).await?)
2950	}
2951	None => reply(&backups_off::<bool>()),
2952	},
2953	"backup_complete" => match backups::storage(&env) {
2954	Some(blobs) => reply(&backups::complete(&repos.registry, &blobs, &args(body)?, now_ms()).await?),
2955	None => reply(&backups_off::<bool>()),
2956	},
2957	"backup_fail" => match backups::storage(&env) {
2958	Some(blobs) => reply(&backups::fail(&repos.registry.db, &blobs, &args(body)?).await?),
2959	None => reply(&backups_off::<bool>()),
2960	},
2961	// How the git store has been answering, for the status page.
2962	"store_health" => {
2963	let a: meters::HealthArgs = args(body)?;
2964	reply(&meters::health(&repos.registry.db, &a).await?)
2965	}
2966	// Where repositories may be kept, for a workspace's settings.
2967	"storage_options" => reply(&repos.storage_options()),
2968	// Services and operators only: how each namespace stands, and
2969	// moving a repository between them (namespaces.rs, moves.rs).
2970	"namespaces" => reply(&repos.standings().await?),
2971	"move_repository" => reply(&repos.move_repository(args(body)?).await?),
2972	"repository_moves" => {
2973	let a: moves::ListMovesArgs = args(body)?;
2974	reply(&repos.registry.moves(a.limit.unwrap_or(50)).await?)
2975	}
2976	_ => Response::error("Unknown method", 404),
2977	} }
2978	.await;
2979	// The git store is busy: said in words, with when to try again.
2980	let answered = match answered {
2981	Err(error) => match resilience::busy(&error.to_string()) {
2982	Some(busy) if OUTCOME_READS.contains(&method.as_str()) => {
2983	reply(&Outcome::<()>::fail(FailureCode::Conflict, busy.message().trim()))
2984	}
2985	Some(busy) => {
2986	let response = Response::error(busy.message(), 503)?;
2987	response.headers().set("retry-after", &busy.retry_after.to_string())?;
2988	Ok(response)
2989	}
2990	None => Err(error),
2991	},
2992	answered => answered,
2993	};
2994	repos.deferred.hand_over(&ctx);
2995	flush_later(&env, &ctx);
2996	served.finish(answered)
2997	}
2998
2999	/// The nightly cron in wrangler.jsonc: tonight's backups are queued.
3000	const BACKUP_CRON: &str = "53 2 * * *";
3001
3002	/// The hourly sweep: deleted repositories whose time to be restored has
3003	/// passed are purged. See lifecycle.rs. And, at [`BACKUP_CRON`], the
3004	/// repositories whose refs moved are queued for a backup (backups.rs).
3005	#[event(scheduled)]
3006	async fn scheduled(event: ScheduledEvent, env: Env, _ctx: ScheduleContext) {
3007	let repos = match service(&env) {
3008	Ok(repos) => repos,
3009	Err(error) => {
3010	worker::console_error!("repos: the sweep could not start: {error}");
3011	return;
3012	}
3013	};
3014	if event.cron() == BACKUP_CRON {
3015	let Some(blobs) = backups::storage(&env) else { return };
3016	match backups::nightly(&repos.registry.db, &blobs, backups::Settings::from_env(&env), now_ms()).await {
3017	Ok(night) => worker::console_log!("repos: queued {} backups, removed {} of purged repositories", night.queued, night.pruned),
3018	Err(error) => worker::console_error!("repos: backups could not be queued: {error}"),
3019	}
3020	return;
3021	}
3022	match repos.purge_due(PurgeDueArgs::default()).await {
3023	Ok(0) => {}
3024	Ok(count) => worker::console_log!("repos: purged {count} deleted repositories"),
3025	Err(error) => worker::console_error!("repos: the purge sweep failed: {error}"),
3026	}
3027	// Pull requests' working copies whose time has come (forks.rs).
3028	match repos.retire_due().await {
3029	Ok(0) => {}
3030	Ok(count) => worker::console_log!("repos: removed {count} pull request working copies"),
3031	Err(error) => worker::console_error!("repos: the working copy sweep failed: {error}"),
3032	}
3033	// Private forks whose makers can no longer read where they came from
3034	// (network.rs).
3035	match repos.check_private_forks().await {
3036	Ok(0) => {}
3037	Ok(count) => worker::console_log!("repos: deleted {count} private forks their makers can no longer read the source of"),
3038	Err(error) => worker::console_error!("repos: the private fork sweep failed: {error}"),
3039	}
3040	// Repositories moving between namespaces, and old copies (moves.rs).
3041	match repos.run_moves().await {
3042	Ok(0) => {}
3043	Ok(count) => worker::console_log!("repos: moved {count} repositories between namespaces"),
3044	Err(error) => worker::console_error!("repos: the move sweep failed: {error}"),
3045	}
3046	meters::flush(&repos.registry.db).await;
3047	}
3048
3049	/// Events from the bus. A workspace's rename: its repositories move to the
3050	/// workspace's current slug, asked of identity by id, so a repeated or late
3051	/// delivery lands in the same place; their git store keys stay as they
3052	/// were. A workspace's deletion: its repositories are deleted with it,
3053	/// restored with it, or purged with it.
3054	#[event(queue)]
3055	async fn queue(batch: MessageBatch<Event>, env: Env, ctx: Context) -> Result<()> {
3056	let registry = Registry { db: env.d1("DB")? };
3057	let identity = env.service("IDENTITY")?;
3058	let handled = handle_events(&batch, &env, &registry, &identity).await;
3059	flush_later(&env, &ctx);
3060	handled
3061	}
3062
3063	/// Each event is acknowledged or retried on its own, so one that fails is
3064	/// tried again without the others before and after it running twice.
3065	async fn handle_events(batch: &MessageBatch<Event>, env: &Env, registry: &Registry, identity: &Fetcher) -> Result<()> {
3066	for message in batch.messages()? {
3067	match handle_event(message.body(), env, registry, identity).await {
3068	Ok(()) => message.ack(),
3069	Err(error) => {
3070	worker::console_error!("repos: event {} failed: {error}", message.body().id);
3071	message.retry();
3072	}
3073	}
3074	}
3075	Ok(())
3076	}
3077
3078	async fn handle_event(event: &Event, env: &Env, registry: &Registry, identity: &Fetcher) -> Result<()> {
3079	// A pull request merged, closed or reopened: its working copy is
3080	// kept or let go (forks.rs).
3081	if let Some(change) = forks::pull_change(&event.kind) {
3082	let Some(pull_id) = forks::pull_id_of(&event.data) else {
3083	worker::console_error!("{} {} names no pull request", event.kind, event.id);
3084	return Ok(());
3085	};
3086	let repos = service(env)?;
3087	match change {
3088	forks::PullChange::Settled => repos.pull_settled(&pull_id).await?,
3089	forks::PullChange::Reopened => repos.pull_reopened(&pull_id).await?,
3090	}
3091	return Ok(());
3092	}
3093	// A workspace deleted, restored or purged: its repositories go with
3094	// it, come back with it, or are purged with it (lifecycle.rs).
3095	if event.kind == "workspace.deleting" {
3096	match serde_json::from_value::<WorkspaceDeleting>(event.data.clone()) {
3097	Ok(deleting) => service(env)?.delete_with_workspace(&deleting, &protected_workspaces(env)).await?,
3098	Err(_) => worker::console_error!("workspace.deleting {} could not be read", event.id),
3099	}
3100	return Ok(());
3101	}
3102	if event.kind == "workspace.restored" {
3103	match serde_json::from_value::<WorkspaceRestored>(event.data.clone()) {
3104	Ok(restored) => service(env)?.restore_with_workspace(&restored).await?,
3105	Err(_) => worker::console_error!("workspace.restored {} could not be read", event.id),
3106	}
3107	return Ok(());
3108	}
3109	if event.kind == "workspace.deleted" {
3110	match serde_json::from_value::<WorkspaceDeleted>(event.data.clone()) {
3111	Ok(deleted) => service(env)?.purge_workspace(&deleted, &protected_workspaces(env)).await?,
3112	Err(_) => worker::console_error!("workspace.deleted {} could not be read", event.id),
3113	}
3114	return Ok(());
3115	}
3116	// An account deleted or restored: kept contributors that name it
3117	// (or ghost, for a restore) are worked out again, so it shows as
3118	// ghost for its 30 days, and as itself again if restored.
3119	if let Some(needle) = stats::shown_differently(&event.kind, &event.data) {
3120	let db = env.d1("DB")?;
3121	if let Err(error) = stats::rework_naming(&db, &needle).await {
3122	worker::console_error!("{} {}: contributors not marked to be counted again: {error}", event.kind, event.id);
3123	}
3124	return Ok(());
3125	}
3126	if event.kind != "workspace.renamed" {
3127	return Ok(());
3128	}
3129	let Ok(renamed) = serde_json::from_value::<WorkspaceRenamed>(event.data.clone()) else {
3130	worker::console_error!("workspace.renamed {} could not be read", event.id);
3131	return Ok(());
3132	};
3133	let names: HashMap<String, String> = g1t_kit::call(
3134	identity,
3135	"usernames",
3136	&g1t_contracts::identity::UsernamesArgs {
3137	ids: vec![renamed.workspace_id.clone()],
3138	},
3139	)
3140	.await?;
3141	let current = names
3142	.get(&renamed.workspace_id)
3143	.cloned()
3144	.unwrap_or_else(\|\| renamed.to.clone());
3145	let left = registry
3146	.rename_namespace(&renamed.stale_slugs(&current), &current)
3147	.await?;
3148	if left > 0 {
3149	worker::console_error!(
3150	"{left} repositories stayed under {} or {}: {current} already has repositories of the same names",
3151	renamed.from,
3152	renamed.to
3153	);
3154	}
3155	Ok(())
3156	}
3157
3158	/// The workspaces whose repositories never go with a deletion, whatever is
3159	/// published: `PROTECTED_WORKSPACES` if set here, and Flagon's always.
3160	fn protected_workspaces(env: &Env) -> Vec<String> {
3161	let configured = env.var("PROTECTED_WORKSPACES").ok().map(\|v\| v.to_string());
3162	g1t_contracts::identity::protected_names(configured.as_deref())
3163	}
3164
3165	/// What a token's own limits say about git on the repository `repo`
3166	/// (`owner/name`), before anyone's role is asked: why it is refused, or
3167	/// `None`. `source` is the repository a pull request's working copy at
3168	/// `repo` belongs to, which a workflow job's token reaches too. `public`
3169	/// is whether anyone may read it, and `exists` whether there is one.
3170	///
3171	/// A job's token and a deploy key reach their own repository only. Its
3172	/// scopes decide the rest: `code:read` to read a private repository,
3173	/// `code:write` to push, which a read-only deploy key never has. A deploy
3174	/// key never makes a repository by pushing to an empty address.
3175	pub(crate) fn git_token_refusal(
3176	access: &g1t_contracts::scopes::TokenAccess,
3177	repo: &str,
3178	source: Option<&str>,
3179	write: bool,
3180	public: bool,
3181	exists: bool,
3182	) -> Option<String> {
3183	if let Some(refused) = g1t_contracts::scopes::decide_repo(access, repo)
3184	&& !source.is_some_and(\|source\| access.reaches(source))
3185	{
3186	return Some(refused.reason.unwrap_or_default());
3187	}
3188	let decision = g1t_contracts::scopes::decide_git(access, write, public);
3189	if !decision.allowed {
3190	return Some(decision.reason.unwrap_or_default());
3191	}
3192	if access.deploy_key.is_some() && !exists {
3193	return Some(format!("This deploy key is for {repo}, which is not there any more."));
3194	}
3195	None
3196	}
3197
3198	/// The repository a push to a path that does not exist yet creates: private,
3199	/// so nothing pushed by mistake is published. An owner makes it public on
3200	/// purpose (`POST /repos/{owner}/{repo}/visibility`).
3201	fn push_to_create(owner: &User, path: &RepoPath) -> CreateArgs {
3202	CreateArgs {
3203	owner: owner.clone(),
3204	namespace: path.namespace.clone(),
3205	name: path.name.clone(),
3206	description: None,
3207	is_private: true,
3208	import_url: None,
3209	import_token: None,
3210	}
3211	}
3212
3213	/// Whether `delete_branch` may remove `branch`: one of g1t's own
3214	/// (`g1t-…`), or one whose tip the caller names, such as a dependency
3215	/// update's branch after its pull request closed.
3216	fn deletable_branch(branch: &str, head: Option<&str>) -> bool {
3217	!branch.is_empty() && (branch.starts_with(G1T_BRANCH_PREFIX) \|\| head.is_some_and(\|head\| !head.is_empty()))
3218	}
3219
3220	#[cfg(test)]
3221	mod delete_branch_tests {
3222	use super::deletable_branch;
3223
3224	#[test]
3225	fn only_g1t_branches_or_a_named_tip_are_deleted() {
3226	assert!(deletable_branch("g1t-queue-12", None));
3227	assert!(!deletable_branch("g1t/security/sharp-0.35.5", None));
3228	assert!(deletable_branch("g1t/security/sharp-0.35.5", Some("abc123")));
3229	assert!(!deletable_branch("feature", Some("")));
3230	assert!(!deletable_branch("", Some("abc123")));
3231	}
3232	}
3233
3234	#[cfg(test)]
3235	mod announced_refs_tests {
3236	use super::*;
3237
3238	fn pushed(git_ref: String) -> git_http::Pushed {
3239	git_http::Pushed { git_ref, before: None, after: "abc".into() }
3240	}
3241
3242	fn refs(announced: Vec<&git_http::Pushed>) -> Vec<&str> {
3243	announced.into_iter().map(\|pushed\| pushed.git_ref.as_str()).collect()
3244	}
3245
3246	#[test]
3247	fn a_few_tags_are_announced_and_many_are_not() {
3248	let few: Vec<_> = (1..=3).map(\|n\| pushed(format!("refs/tags/v{n}"))).chain([pushed("refs/heads/main".into())]).collect();
3249	assert_eq!(refs(announced_refs(&few, "main")), ["refs/heads/main", "refs/tags/v1", "refs/tags/v2", "refs/tags/v3"]);
3250	let many: Vec<_> = (1..=10_000).map(\|n\| pushed(format!("refs/tags/v{n}"))).chain([pushed("refs/heads/main".into())]).collect();
3251	assert_eq!(refs(announced_refs(&many, "main")), ["refs/heads/main"]);
3252	}
3253
3254	#[test]
3255	fn past_the_branch_cap_only_the_default_branch_is_announced() {
3256	let at_cap: Vec<_> = (0..MAX_PUSH_BRANCH_EVENTS).map(\|n\| pushed(format!("refs/heads/b{n}"))).collect();
3257	assert_eq!(announced_refs(&at_cap, "main").len(), MAX_PUSH_BRANCH_EVENTS);
3258	let over: Vec<_> = (0..=MAX_PUSH_BRANCH_EVENTS)
3259	.map(\|n\| pushed(format!("refs/heads/b{n}")))
3260	.chain([pushed("refs/heads/main".into())])
3261	.collect();
3262	assert_eq!(refs(announced_refs(&over, "main")), ["refs/heads/main"]);
3263	assert!(announced_refs(&over, "trunk").is_empty());
3264	}
3265	}
3266
3267	#[cfg(test)]
3268	mod push_to_create_tests {
3269	use super::*;
3270
3271	#[test]
3272	fn a_pushed_repository_starts_private() {
3273	let owner: User = serde_json::from_value(serde_json::json!({ "id": "usr_1", "username": "ada" })).unwrap();
3274	let args = push_to_create(&owner, &RepoPath { namespace: "acme".into(), name: "site".into() });
3275	assert!(args.is_private);
3276	assert_eq!((args.namespace.as_str(), args.name.as_str()), ("acme", "site"));
3277	}
3278	}
3279
3280	#[cfg(test)]
3281	mod deploy_key_git_tests {
3282	use super::*;
3283	use g1t_contracts::deploy_keys;
3284
3285	fn key(read_only: bool) -> User {
3286	deploy_keys::principal("wsp_acme", "acme", deploy_keys::access("dk_1", "CI", "acme/rocket", read_only))
3287	}
3288
3289	fn rocket(private: bool) -> Repo {
3290	serde_json::from_value(serde_json::json!({
3291	"id": "rep_rocket",
3292	"namespace": "acme",
3293	"name": "rocket",
3294	"description": null,
3295	"isPrivate": private,
3296	"ownerId": "usr_owner",
3297	"defaultBranch": "main",
3298	"forkOf": null,
3299	"protected": false,
3300	"createdAt": "",
3301	}))
3302	.unwrap()
3303	}
3304
3305	fn refusal(user: &User, repo: &str, write: bool, exists: bool) -> Option<String> {
3306	git_token_refusal(user.token.as_deref().unwrap(), repo, None, write, false, exists)
3307	}
3308
3309	#[test]
3310	fn a_read_only_deploy_key_clones_its_repository_and_never_pushes() {
3311	let user = key(true);
3312	assert_eq!(refusal(&user, "acme/rocket", false, true), None);
3313	assert!(refusal(&user, "acme/rocket", true, true).unwrap().contains("read-only"));
3314	// Its role is a workspace token's: it reads a private repository.
3315	assert!(registry::can_read(&rocket(true), &Some(user)));
3316	}
3317
3318	#[test]
3319	fn a_deploy_key_with_write_access_pushes_to_its_repository() {
3320	let user = key(false);
3321	assert_eq!(refusal(&user, "acme/rocket", true, true), None);
3322	assert!(registry::can_write(&rocket(true), &Some(user)));
3323	}
3324
3325	#[test]
3326	fn a_deploy_key_reaches_no_other_repository() {
3327	let user = key(false);
3328	for other in ["acme/booster", "other/rocket"] {
3329	for write in [false, true] {
3330	let why = refusal(&user, other, write, true).expect(other);
3331	assert!(why.contains("deploy key is for acme/rocket"), "{why}");
3332	}
3333	}
3334	// Not even a pull request's working copy of another repository.
3335	let token = user.token.as_deref().unwrap();
3336	assert!(git_token_refusal(token, "pulls/pr_1", Some("acme/booster"), false, false, true).is_some());
3337	}
3338
3339	#[test]
3340	fn a_deploy_key_never_creates_a_repository() {
3341	let why = refusal(&key(false), "acme/rocket", true, false).unwrap();
3342	assert!(why.contains("not there"), "{why}");
3343	}
3344	}