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