g1t/services/packages/src/lib.rs

823 lines35,983 bytesCodeBlame
1//! The packages service: the registries a workspace publishes to and
2//! installs from, beside its code (docs/PACKAGES.md). Container images
3//! first, over OCI Distribution 1.1 on `g1t.sh/v2/` (oci.rs).
4//!
5//! Other services reach it over `POST /rpc/<method>`; see
6//! `g1t_contracts::packages` for the methods and their arguments. Any
7//! other request is a registry's own protocol. Files are kept through the
8//! `BlobStore` port (store/), metadata in D1 (db.rs).
9
10mod access;
11mod archive;
12mod cargo;
13mod cargo_http;
14mod composer;
15mod composer_http;
16mod db;
17mod digest;
18mod limits;
19mod manifest;
20mod marshal;
21mod maven;
22mod maven_http;
23mod names;
24mod npm;
25mod npm_http;
26mod nuget;
27mod nuget_http;
28mod oci;
29mod quota;
30mod range;
31mod rubygems;
32mod rubygems_http;
33mod store;
34mod token;
35mod upload;
36mod xml;
37mod yaml;
38
39use std::cell::RefCell;
40use std::collections::HashMap;
41
42use g1t_contracts::audit::{AuditActor, AuditTarget, NewAuditEntry, RecordAuditArgs, Surface};
43use g1t_contracts::credentials::Decision;
44use g1t_contracts::events::{Event, NewEvent, PackageEvent, Publish};
45use g1t_contracts::identity::GitCredentialsArgs;
46use g1t_contracts::packages::*;
47use g1t_contracts::repos::{GetArgs, Repo, RepoPath};
48use g1t_contracts::{FailureCode, Outcome, User, Viewer, new_id};
49use g1t_kit::{args, now_ms, reply, rpc_method};
50use worker::{Context, Env, Fetcher, MessageBatch, Request, Response, Result, ScheduleContext, ScheduledEvent, event};
51
52use access::{Action, LinkedTo, Target};
53use db::{Db, PackageRow};
54use store::{BlobStore, Store};
55
56pub(crate) const SOURCE: &str = "packages";
57/// How large a request body may be: Cloudflare's limit on the zone's plan.
58const DEFAULT_MAX_REQUEST_BYTES: u64 = 100_000_000;
59/// How many packages a listing shows.
60const LIST_LIMIT: u32 = 200;
61const VERSIONS_SHOWN: u32 = 200;
62/// How much one sweep lets go of.
63const SWEEP_BATCH: u32 = 200;
64
65thread_local! {
66 /// Pulls counted since the last write, by package (and by version,
67 /// where it is counted too): written at most every few seconds, so a
68 /// busy image costs one write, not one a pull. What an isolate holds
69 /// when it goes away is lost: the count is approximate.
70 static DOWNLOADS: RefCell<(HashMap<(String, Option<String>), u64>, u64)> = RefCell::new((HashMap::new(), 0));
71}
72const DOWNLOADS_FLUSH_MS: u64 = 10_000;
73
74thread_local! {
75 /// What billing allows each workspace, as asked last, and when.
76 static ALLOWANCES: RefCell<HashMap<String, (quota::Allowance, u64)>> = RefCell::new(HashMap::new());
77}
78/// How long billing's answer is kept.
79const ALLOWANCE_TTL_MS: u64 = 5 * 60 * 1000;
80
81/// Who made a registry request, as its audit entries and versions name them.
82pub struct Caller {
83 pub actor: Option<AuditActor>,
84}
85
86/// What access decisions need about a package, owned.
87pub struct TargetOf {
88 workspace: String,
89 repo: Option<(String, String, bool)>,
90 public: bool,
91}
92
93impl TargetOf {
94 pub fn package(row: &PackageRow) -> TargetOf {
95 TargetOf {
96 workspace: row.workspace.clone(),
97 repo: row
98 .repo_id
99 .as_ref()
100 .map(|id| (id.clone(), row.repo_name.clone().unwrap_or_default(), !row.public())),
101 public: row.public(),
102 }
103 }
104
105 pub fn view(&self) -> Target<'_> {
106 Target {
107 workspace: &self.workspace,
108 repo: self.repo.as_ref().map(|(id, name, private)| LinkedTo { id, name, private: *private }),
109 public: self.public,
110 }
111 }
112}
113
114pub struct Packages {
115 pub db: Db,
116 pub store: Store,
117 identity: Fetcher,
118 repos: Fetcher,
119 events: Fetcher,
120 /// Signs registry tokens (PACKAGES_TOKEN_SECRET).
121 secret: Vec<u8>,
122 max_request: u64,
123 /// The host package addresses start with: `g1t.sh`.
124 host: String,
125 /// Whether free workspaces' package storage is limited (STORAGE_LIMITS
126 /// `on`); off when self-hosted.
127 storage_limits: bool,
128 env: Env,
129}
130
131/// What a workspace's deletion or restoring does to its packages.
132#[derive(Debug, PartialEq, Eq)]
133pub(crate) enum WorkspaceMark {
134 /// Hide them from `at`, while the workspace may still be restored.
135 Hide { workspace: String, at: String },
136 /// Show them again.
137 Show { workspace: String },
138}
139
140/// The mark `event` asks for, if any. A protected workspace (`protected`,
141/// slugs or ids, as identity's PROTECTED_WORKSPACES) is never hidden: it
142/// cannot be deleted, so an event saying so is a mistake.
143pub(crate) fn workspace_mark(event: &Event, protected: &[String], now: &str) -> Option<WorkspaceMark> {
144 let text = |key: &str| event.data[key].as_str().map(|v| v.trim().to_lowercase()).filter(|v| !v.is_empty());
145 let workspace = text("slug")?;
146 match event.kind.as_str() {
147 "workspace.deleting" => {
148 let id = text("workspaceId").unwrap_or_default();
149 if protected.iter().any(|name| *name == workspace || (!id.is_empty() && *name == id)) {
150 return None;
151 }
152 Some(WorkspaceMark::Hide { workspace, at: now.to_owned() })
153 }
154 "workspace.restored" => Some(WorkspaceMark::Show { workspace }),
155 _ => None,
156 }
157}
158
159fn not_found<T>() -> Outcome<T> {
160 Outcome::fail(FailureCode::NotFound, "Package not found.")
161}
162
163impl Packages {
164 pub fn from_env(env: &Env) -> Result<Packages> {
165 let secret = store::var(env, "PACKAGES_TOKEN_SECRET");
166 if secret.len() < 32 {
167 return Err(worker::Error::RustError("PACKAGES_TOKEN_SECRET is missing or shorter than 32 characters".into()));
168 }
169 // 0 is no limit, as self-hosted (nothing in front cuts bodies short).
170 let max_request = match store::var(env, "MAX_REQUEST_BYTES").parse::<u64>() {
171 Ok(0) => u64::MAX,
172 Ok(limit) => limit,
173 Err(_) => DEFAULT_MAX_REQUEST_BYTES,
174 };
175 let host = store::var(env, "REGISTRY_HOST");
176 Ok(Packages {
177 db: Db { db: env.d1("DB")? },
178 store: store::from_env(env)?,
179 identity: env.service("IDENTITY")?,
180 repos: env.service("REPOS")?,
181 events: env.service("EVENTS")?,
182 secret: secret.into_bytes(),
183 max_request,
184 host: if host.is_empty() { "g1t.sh".to_owned() } else { host },
185 storage_limits: store::var(env, "STORAGE_LIMITS") == "on",
186 env: env.clone(),
187 })
188 }
189
190 async fn viewer_for(&self, username: &str, secret: &str) -> Result<Viewer> {
191 g1t_kit::call(
192 &self.identity,
193 "user_for_git_credentials",
194 &GitCredentialsArgs { username: username.to_owned(), secret: secret.to_owned() },
195 )
196 .await
197 }
198
199 /// A repository of the workspace by name, whoever may see it.
200 async fn repo_by_name(&self, workspace: &str, name: &str) -> Result<Option<Repo>> {
201 if !g1t_contracts::is_valid_repo_name(name) {
202 return Ok(None);
203 }
204 let found: Outcome<Repo> = g1t_kit::call(
205 &self.repos,
206 "get",
207 &GetArgs {
208 path: RepoPath { namespace: workspace.to_owned(), name: name.to_owned() },
209 viewer: Some(User::system(workspace)),
210 },
211 )
212 .await?;
213 Ok(found.into_result().ok().filter(|repo| repo.fork_of.is_none()))
214 }
215
216 /// Who may do what to an image: the package's own settings, or, before
217 /// its first push, the repository it will be linked to.
218 async fn target(&self, name: &names::ImageName, found: Option<&PackageRow>) -> Result<TargetOf> {
219 if let Some(found) = found {
220 return Ok(TargetOf::package(found));
221 }
222 let repo = self.repo_by_name(&name.workspace, name.repo_name()).await?;
223 Ok(TargetOf {
224 workspace: name.workspace.clone(),
225 repo: repo.map(|repo| (repo.id, repo.name, repo.is_private)),
226 public: false,
227 })
228 }
229
230 /// What billing allows the workspace, kept for five minutes. `None`
231 /// when billing cannot be asked: the push is then let through.
232 async fn allowance(&self, workspace: &str) -> Option<quota::Allowance> {
233 let now = now_ms();
234 let kept = ALLOWANCES.with(|kept| kept.borrow().get(workspace).copied());
235 if let Some((allowance, at)) = kept
236 && now.saturating_sub(at) < ALLOWANCE_TTL_MS
237 {
238 return Some(allowance);
239 }
240 let billing = self.env.service("BILLING").ok()?;
241 let asked: Result<quota::Allowance> = g1t_kit::call(
242 &billing,
243 "entitlements",
244 &g1t_contracts::billing::EntitlementsArgs { workspace: workspace.to_owned() },
245 )
246 .await;
247 match asked {
248 Ok(allowance) => {
249 ALLOWANCES.with(|kept| kept.borrow_mut().insert(workspace.to_owned(), (allowance, now)));
250 Some(allowance)
251 }
252 Err(error) => {
253 worker::console_error!("packages: billing could not be asked about {workspace}, letting the push through: {error}");
254 None
255 }
256 }
257 }
258
259 /// Why a push of these files (digest and size) into the package may
260 /// not be kept, if it may not: a free workspace past its free storage.
261 /// Files the workspace already holds add nothing.
262 pub(crate) async fn storage_refusal(&self, package: &PackageRow, files: &[(String, u64)]) -> Result<Option<String>> {
263 if !self.storage_limits || files.is_empty() {
264 return Ok(None);
265 }
266 let digests: Vec<String> = files.iter().map(|(digest, _)| digest.clone()).collect();
267 let held = self.db.held(&package.workspace, &digests).await?;
268 let mut seen = std::collections::HashSet::new();
269 let adding: u64 = files
270 .iter()
271 .filter(|(digest, _)| !held.contains(digest) && seen.insert(digest.clone()))
272 .map(|(_, size)| size)
273 .sum();
274 if adding == 0 {
275 return Ok(None);
276 }
277 let Some(allowance) = self.allowance(&package.workspace).await else {
278 return Ok(None);
279 };
280 let (public_bytes, private_bytes) = self.db.storage(&package.workspace).await?;
281 let public = package.public();
282 let used = if public { public_bytes } else { private_bytes };
283 Ok(quota::decide(&allowance, public, used, adding)
284 .err()
285 .map(|refusal| quota::message(&package.workspace, &refusal)))
286 }
287
288 fn count_download(&self, package_id: &str, ctx: &Context) {
289 self.count_downloads(package_id, None, ctx);
290 }
291
292 /// A download of one version, counted for it and its package.
293 fn count_version_download(&self, package_id: &str, version_id: &str, ctx: &Context) {
294 self.count_downloads(package_id, Some(version_id), ctx);
295 }
296
297 fn count_downloads(&self, package_id: &str, version_id: Option<&str>, ctx: &Context) {
298 let due = DOWNLOADS.with(|counts| {
299 let mut counts = counts.borrow_mut();
300 *counts.0.entry((package_id.to_owned(), version_id.map(str::to_owned))).or_default() += 1;
301 let now = now_ms();
302 if now.saturating_sub(counts.1) < DOWNLOADS_FLUSH_MS {
303 return None;
304 }
305 counts.1 = now;
306 Some(counts.0.drain().collect::<Vec<_>>())
307 });
308 if let Some(due) = due {
309 let env = self.env.clone();
310 ctx.wait_until(async move {
311 let written = match env.d1("DB") {
312 Ok(db) => Db { db }.add_downloads(&due).await,
313 Err(error) => Err(error),
314 };
315 if let Err(error) = written {
316 worker::console_error!("packages: downloads not counted: {error}");
317 }
318 });
319 }
320 }
321
322 async fn announce(&self, kind: &'static str, package: &PackageRow, data: PackageEvent, caller: &Caller) {
323 let event = NewEvent {
324 kind,
325 source: SOURCE,
326 repo_id: package.repo_id.clone(),
327 actor: caller.actor.as_ref().map(|actor| actor.actor_id.clone()),
328 data,
329 };
330 let published: Result<serde_json::Value> = g1t_kit::call(&self.events, "publish", &Publish { events: vec![event] }).await;
331 if let Err(error) = published {
332 worker::console_error!("packages: {kind} not published: {error}");
333 }
334 }
335
336 async fn audit(&self, caller: &Caller, action: &str, package: &PackageRow, path: Option<&str>, surface: Option<Surface>) {
337 let Some(actor) = caller.actor.clone() else {
338 return;
339 };
340 let target = AuditTarget {
341 workspace: package.workspace.clone(),
342 repo: package.repo_name.as_ref().map(|name| format!("{}/{name}", package.workspace)),
343 path: Some(path.map_or_else(|| format!("{}:{}/{}", package.ecosystem, package.workspace, package.name), str::to_owned)),
344 ..AuditTarget::default()
345 };
346 let mut entry = NewAuditEntry::new(
347 actor,
348 action,
349 surface.unwrap_or(Surface::Registry),
350 target,
351 &Decision::allow("packages"),
352 new_id("req", now_ms()),
353 );
354 entry.result = Some("ok".to_owned());
355 let recorded: Result<u32> = g1t_kit::call(&self.events, "audit_record", &RecordAuditArgs { entries: vec![entry] }).await;
356 if let Err(error) = recorded {
357 worker::console_error!("packages: audit entry not recorded: {error}");
358 }
359 }
360
361 fn summary(&self, row: &db::ListedRow) -> PackageSummary {
362 let p = &row.package;
363 PackageSummary {
364 id: p.id.clone(),
365 workspace: p.workspace.clone(),
366 ecosystem: Ecosystem::parse(&p.ecosystem).unwrap_or(Ecosystem::Container),
367 name: p.name.clone(),
368 address: match p.ecosystem.as_str() {
369 "npm" => format!("{}/-/npm/@{}/{}", self.host, p.workspace, p.name),
370 "composer" => format!("{}/-/composer/{}/{}", self.host, p.workspace, p.name),
371 "cargo" => format!("{}/-/cargo/{}/{}", self.host, p.workspace, p.name),
372 "maven" => format!("{}/-/maven/{}/{}", self.host, p.workspace, p.name),
373 "nuget" => format!("{}/-/nuget/{}/{}", self.host, p.workspace, p.name),
374 "rubygems" => format!("{}/-/rubygems/{}/{}", self.host, p.workspace, p.name),
375 _ => format!("{}/{}/{}", self.host, p.workspace, p.name),
376 },
377 visibility: Visibility::parse(&p.visibility),
378 repo: p.repo_id.as_ref().map(|id| LinkedRepo {
379 id: id.clone(),
380 namespace: p.workspace.clone(),
381 name: p.repo_name.clone().unwrap_or_default(),
382 }),
383 description: p.description.clone(),
384 versions: row.version_count,
385 latest: db::latest_shown(row),
386 size: row.bytes,
387 downloads: p.downloads,
388 created_at: p.created_at.clone(),
389 updated_at: p.updated_at.clone(),
390 }
391 }
392
393 async fn listed(&self, workspace: &str, ecosystem: Ecosystem, name: &str) -> Result<Option<db::ListedRow>> {
394 let rows = self.db.list(workspace, Some(ecosystem.as_str()), None, None, LIST_LIMIT).await?;
395 if let Some(row) = rows.into_iter().find(|row| row.package.name == name) {
396 return Ok(Some(row));
397 }
398 // Past the first page: read it alone.
399 Ok(self.db.package(workspace, ecosystem.as_str(), name).await?.filter(|package| !package.hidden()).map(|package| db::ListedRow {
400 package,
401 version_count: 0,
402 bytes: 0,
403 latest_tag: None,
404 latest_tag_version: None,
405 latest_version: None,
406 }))
407 }
408
409 async fn list_packages(&self, a: ListPackagesArgs) -> Result<Outcome<Vec<PackageSummary>>> {
410 let workspace = a.workspace.to_lowercase();
411 let rows = self
412 .db
413 .list(&workspace, a.ecosystem.map(Ecosystem::as_str), a.repo_id.as_deref(), a.query.as_deref(), LIST_LIMIT)
414 .await?;
415 let visible = rows
416 .iter()
417 .filter(|row| access::decide(a.viewer.as_ref(), &TargetOf::package(&row.package).view(), Action::Pull).allowed)
418 .map(|row| self.summary(row))
419 .collect();
420 Ok(Outcome::Ok(visible))
421 }
422
423 async fn get_package(&self, a: GetPackageArgs) -> Result<Outcome<PackageDetail>> {
424 let workspace = a.workspace.to_lowercase();
425 let Some(row) = self.listed(&workspace, a.ecosystem, &a.name).await? else {
426 return Ok(not_found());
427 };
428 let target = TargetOf::package(&row.package);
429 let permissions = access::permissions(a.viewer.as_ref(), &target.view());
430 if !permissions.pull {
431 return Ok(not_found());
432 }
433 let tags = self.db.tags(&row.package.id).await?;
434 // A yanked or unlisted version reads as deprecated: still there
435 // for lockfiles that name it, no longer picked for new ones.
436 let withdrawn = match row.package.ecosystem.as_str() {
437 "nuget" => "Unlisted: still restored by projects that name it, no longer shown in search.",
438 "rubygems" => "Yanked: Bundler no longer picks this version for new lockfiles.",
439 _ => "Yanked: Cargo no longer picks this version for new lockfiles.",
440 };
441 let versions = self
442 .db
443 .versions(&row.package.id, VERSIONS_SHOWN)
444 .await?
445 .into_iter()
446 .map(|version| {
447 let meta = version.meta();
448 let text = |key: &str| meta[key].as_str().map(str::to_owned);
449 PackageVersion {
450 tags: tags.iter().filter(|tag| tag.version_id == version.id).map(|tag| tag.tag.clone()).collect(),
451 media_type: text("media_type"),
452 artifact_type: text("artifact_type"),
453 platforms: meta["platforms"]
454 .as_array()
455 .map(|list| list.iter().filter_map(|p| p.as_str().map(str::to_owned)).collect())
456 .unwrap_or_default(),
457 id: version.id,
458 version: version.version,
459 digest: version.digest,
460 size: version.size,
461 subject: version.subject,
462 published_by: version.published_by,
463 published_at: version.published_at,
464 deprecated: if version.yanked != 0 {
465 Some(withdrawn.to_owned())
466 } else {
467 version.deprecated
468 },
469 symbols: meta["symbols"] == true,
470 downloads: (row.package.ecosystem == "nuget").then_some(version.downloads),
471 }
472 })
473 .collect();
474 // The README its page shows: npm's, from the latest version.
475 let readme = match self.db.readme_digest(&row.package.id).await?.and_then(|d| digest::Digest::parse(&d)) {
476 Some(digest) => match self.db.blob(&digest).await? {
477 Some(blob) => self.store.read(&blob.object_key).await?.map(|b| String::from_utf8_lossy(&b).into_owned()),
478 None => None,
479 },
480 None => None,
481 };
482 Ok(Outcome::Ok(PackageDetail {
483 readme,
484 package: self.summary(&row),
485 versions,
486 tags: tags
487 .into_iter()
488 .map(|tag| PackageTag { tag: tag.tag, digest: tag.digest, updated_at: tag.updated_at })
489 .collect(),
490 permissions,
491 }))
492 }
493
494 /// The package `actor` asks to change, if they may `action` it.
495 async fn for_change(&self, actor: &User, workspace: &str, ecosystem: Ecosystem, name: &str, action: Action) -> Result<Outcome<PackageRow>> {
496 let Some(package) = self.db.package(&workspace.to_lowercase(), ecosystem.as_str(), name).await?.filter(|p| !p.hidden()) else {
497 return Ok(not_found());
498 };
499 let target = TargetOf::package(&package);
500 let decision = access::decide(Some(actor), &target.view(), action);
501 if decision.allowed {
502 return Ok(Outcome::Ok(package));
503 }
504 if !access::decide(Some(actor), &target.view(), Action::Pull).allowed {
505 return Ok(not_found());
506 }
507 Ok(Outcome::fail(FailureCode::Forbidden, decision.reason.unwrap_or_else(|| "Not allowed.".to_owned())))
508 }
509
510 async fn delete_version(&self, a: DeleteVersionArgs) -> Result<Outcome<()>> {
511 let package = match self.for_change(&a.actor, &a.workspace, a.ecosystem, &a.name, Action::Delete).await? {
512 Outcome::Ok(package) => package,
513 Outcome::Fail(failure) => return Ok(Outcome::Fail(failure)),
514 };
515 let Some(version) = self.db.find_version(&package.id, &a.version).await? else {
516 return Ok(Outcome::fail(FailureCode::NotFound, "Version not found."));
517 };
518 let caller = Caller { actor: Some(AuditActor::of(&a.actor)) };
519 self.remove_version(&package, &version, &caller).await?;
520 Ok(Outcome::Ok(()))
521 }
522
523 async fn delete_package(&self, a: DeletePackageArgs) -> Result<Outcome<()>> {
524 let package = match self.for_change(&a.actor, &a.workspace, a.ecosystem, &a.name, Action::Delete).await? {
525 Outcome::Ok(package) => package,
526 Outcome::Fail(failure) => return Ok(Outcome::Fail(failure)),
527 };
528 self.db.delete_package(&package.id).await?;
529 self.db.measure(&package.workspace).await?;
530 let caller = Caller { actor: Some(AuditActor::of(&a.actor)) };
531 self.announce("package.deleted", &package, self.event_of(&package), &caller).await;
532 self.audit(&caller, "package.delete", &package, None, a.surface).await;
533 Ok(Outcome::Ok(()))
534 }
535
536 async fn set_package(&self, a: SetPackageArgs) -> Result<Outcome<PackageSummary>> {
537 let package = match self.for_change(&a.actor, &a.workspace, a.ecosystem, &a.name, Action::Delete).await? {
538 Outcome::Ok(package) => package,
539 Outcome::Fail(failure) => return Ok(Outcome::Fail(failure)),
540 };
541 let caller = Caller { actor: Some(AuditActor::of(&a.actor)) };
542 let now = now_ms();
543 let before = package.visibility.clone();
544 if a.unlink {
545 self.db.set_link(&package.id, None, "private", now).await?;
546 } else if let Some(link) = a.link.as_deref() {
547 let Some(repo) = self.repo_by_name(&package.workspace, link).await? else {
548 return Ok(Outcome::fail(FailureCode::NotFound, format!("There is no repository {}/{link}.", package.workspace)));
549 };
550 // Linking hands the package to the repository's roles: only
551 // someone who administers that repository may.
552 let role = g1t_contracts::access::permission(Some(&a.actor), &repo);
553 if role < Some(g1t_contracts::access::RepoRole::Admin) {
554 return Ok(Outcome::fail(
555 FailureCode::Forbidden,
556 format!("You need the Admin role on {}/{} to link a package to it.", repo.namespace, repo.name),
557 ));
558 }
559 let visibility = if repo.is_private { "private" } else { "public" };
560 self.db.set_link(&package.id, Some((&repo.id, &repo.name)), visibility, now).await?;
561 }
562 if let Some(visibility) = a.visibility {
563 let linked = if a.unlink { false } else { a.link.is_some() || package.repo_id.is_some() };
564 if linked {
565 return Ok(Outcome::fail(
566 FailureCode::Invalid,
567 "A linked package has its repository's visibility. Change the repository's, or unlink the package.",
568 ));
569 }
570 self.db.set_visibility(&package.id, visibility.as_str(), now).await?;
571 }
572 self.audit(&caller, "package.update", &package, None, a.surface).await;
573 let Some(row) = self.listed(&package.workspace, a.ecosystem, &package.name).await? else {
574 return Ok(not_found());
575 };
576 if row.package.visibility != before {
577 self.db.measure(&package.workspace).await?;
578 let event = PackageEvent { visibility: Some(row.package.visibility.clone()), ..self.event_of(&row.package) };
579 self.announce("package.visibility_changed", &row.package, event, &caller).await;
580 }
581 Ok(Outcome::Ok(self.summary(&row)))
582 }
583
584 /// Lets go of expired uploads, and of blobs no version has used for a day.
585 async fn sweep(&self) -> Result<(u32, u32)> {
586 let now = now_ms();
587 let mut uploads = 0;
588 for row in self.db.expired_uploads(now, SWEEP_BATCH).await? {
589 if let Some(progress) = row.progress()
590 && let Err(error) = upload::abort(&self.store, &progress).await
591 {
592 worker::console_error!("packages: upload {} not aborted: {error}", row.id);
593 }
594 self.db.delete_upload(&row.id).await?;
595 uploads += 1;
596 }
597 let mut blobs = 0;
598 for blob in self.db.unused_blobs(now, SWEEP_BATCH).await? {
599 self.store.delete(&blob.object_key).await?;
600 self.db.forget_blob(&blob.digest).await?;
601 blobs += 1;
602 }
603 Ok((uploads, blobs))
604 }
605
606 /// Follows what happens elsewhere: workspaces renamed and deleted,
607 /// repositories that change visibility, are renamed, move or go.
608 async fn on_event(&self, env: &Env, event: &Event) -> Result<()> {
609 // Composer packages are read from repositories: again when one is
610 // pushed to (its default branch may have gained a composer.json),
611 // restored, renamed or moved; gone when it is deleted. A failure is
612 // logged, not retried with the batch: the next push reads it again.
613 let repo_id = event.repo_id.clone().or_else(|| event.data["repoId"].as_str().map(str::to_owned));
614 if let Some(repo_id) = repo_id.as_deref() {
615 let synced = match event.kind.as_str() {
616 "git.push" => {
617 let known = self.db.package_for_repo(repo_id, "composer").await?.is_some();
618 if known || event.data["defaultBranch"].as_bool() == Some(true) {
619 Some(self.sync_composer(repo_id).await.map(|_| ()))
620 } else {
621 None
622 }
623 }
624 "repo.deleted" => Some(self.composer_repo_gone(repo_id).await),
625 "repo.restored" | "repo.renamed" | "repo.transferred" => Some(self.sync_composer(repo_id).await.map(|_| ())),
626 _ => None,
627 };
628 if let Some(Err(error)) = synced {
629 worker::console_error!("packages: composer {} for {repo_id}: {error}", event.kind);
630 }
631 if event.kind == "git.push" || event.kind == "repo.deleted" || event.kind == "repo.restored" {
632 return Ok(());
633 }
634 }
635 let db = &self.db.db;
636 let protected = g1t_contracts::identity::protected_names(Some(&store::var(env, "PROTECTED_WORKSPACES")));
637 match workspace_mark(event, &protected, &g1t_contracts::time::rfc3339(now_ms())) {
638 Some(WorkspaceMark::Hide { workspace, at }) => return self.db.mark_workspace(&workspace, Some(&at)).await,
639 Some(WorkspaceMark::Show { workspace }) => return self.db.mark_workspace(&workspace, None).await,
640 None => {}
641 }
642 if g1t_kit::rename::on_event(env, db, event, &[
643 "UPDATE packages SET workspace = ?1 WHERE workspace = ?2",
644 "UPDATE uploads SET workspace = ?1 WHERE workspace = ?2",
645 "UPDATE workspace_blobs SET workspace = ?1 WHERE workspace = ?2",
646 ])
647 .await?
648 {
649 return Ok(());
650 }
651 if g1t_kit::deleted::on_event(db, event, &[
652 "DELETE FROM tags WHERE package_id IN (SELECT id FROM packages WHERE workspace = ?1)",
653 "DELETE FROM version_files WHERE version_id IN (SELECT v.id FROM versions v JOIN packages p ON p.id = v.package_id WHERE p.workspace = ?1)",
654 "DELETE FROM versions WHERE package_id IN (SELECT id FROM packages WHERE workspace = ?1)",
655 "DELETE FROM package_blobs WHERE package_id IN (SELECT id FROM packages WHERE workspace = ?1)",
656 "DELETE FROM uploads WHERE workspace = ?1",
657 "DELETE FROM workspace_blobs WHERE workspace = ?1",
658 "DELETE FROM packages WHERE workspace = ?1",
659 ])
660 .await?
661 {
662 return Ok(());
663 }
664 // A repository renamed keeps its packages linked under its new
665 // name; one moved to another workspace leaves them behind, unlinked.
666 if g1t_kit::transfer::on_event(env, db, event, &[
667 "UPDATE packages SET repo_name = ?6 WHERE repo_id = ?5 AND workspace = ?3",
668 "UPDATE packages SET repo_id = NULL, repo_name = NULL WHERE repo_id = ?5 AND workspace <> ?3",
669 ])
670 .await?
671 {
672 return Ok(());
673 }
674 if g1t_kit::lifecycle::on_purged(db, event, &["UPDATE packages SET repo_id = NULL, repo_name = NULL, visibility = 'private' WHERE repo_id = ?1"])
675 .await?
676 {
677 return Ok(());
678 }
679 if event.kind == "repo.visibility_changed" {
680 #[derive(serde::Deserialize)]
681 #[serde(rename_all = "camelCase")]
682 struct Changed {
683 repo_id: String,
684 is_private: bool,
685 }
686
687 let Ok(changed) = serde_json::from_value::<Changed>(event.data.clone()) else {
688 worker::console_error!("repo.visibility_changed {} could not be read", event.id);
689 return Ok(());
690 };
691 self.db.follow_visibility(&changed.repo_id, changed.is_private).await?;
692 for workspace in self.db.workspaces_linked_to(&changed.repo_id).await? {
693 self.db.measure(&workspace).await?;
694 }
695 }
696 Ok(())
697 }
698}
699
700#[event(fetch)]
701async fn fetch(mut request: Request, env: Env, ctx: Context) -> Result<Response> {
702 let packages = Packages::from_env(&env)?;
703 let Some(method) = rpc_method(&request) else {
704 if request.path().starts_with("/-/npm/") || request.path() == "/-/npm" {
705 return packages.npm(request, &ctx).await;
706 }
707 if request.path().starts_with("/-/composer/") {
708 return packages.composer(request, &ctx).await;
709 }
710 if request.path().starts_with("/-/cargo/") {
711 return packages.cargo(request, &ctx).await;
712 }
713 if request.path().starts_with("/-/maven/") {
714 return packages.maven(request, &ctx).await;
715 }
716 if request.path().starts_with("/-/nuget/") {
717 return packages.nuget(request, &ctx).await;
718 }
719 if request.path().starts_with("/-/rubygems/") {
720 return packages.rubygems(request, &ctx).await;
721 }
722 return packages.registry(request, &ctx).await;
723 };
724 let body: serde_json::Value = request.json().await?;
725 match method.as_str() {
726 "list_packages" => reply(&packages.list_packages(args(body)?).await?),
727 "get_package" => reply(&packages.get_package(args(body)?).await?),
728 "delete_version" => reply(&packages.delete_version(args(body)?).await?),
729 "delete_package" => reply(&packages.delete_package(args(body)?).await?),
730 "set_package" => reply(&packages.set_package(args(body)?).await?),
731 // For billing: what a workspace's packages hold.
732 "storage_all" => reply(&packages.db.storage_all().await?),
733 // Read a repository's Composer package again now, as a push would.
734 "sync_composer" => {
735 let a: SyncComposerArgs = args(body)?;
736 reply(&packages.sync_composer(&a.repo_id).await?)
737 }
738 "storage" => {
739 let a: StorageArgs = args(body)?;
740 let (public_bytes, private_bytes) = packages.db.storage(&a.workspace.to_lowercase()).await?;
741 reply(&PackageStorage { public_bytes, private_bytes })
742 }
743 _ => Response::error("Unknown method", 404),
744 }
745}
746
747/// Every hour: expired uploads are let go, and blobs no version has used
748/// for a day are deleted from the store.
749#[event(scheduled)]
750async fn scheduled(_event: ScheduledEvent, env: Env, _ctx: ScheduleContext) {
751 let packages = match Packages::from_env(&env) {
752 Ok(packages) => packages,
753 Err(error) => {
754 worker::console_error!("packages: the sweep could not start: {error}");
755 return;
756 }
757 };
758 match packages.sweep().await {
759 Ok((0, 0)) => {}
760 Ok((uploads, blobs)) => worker::console_log!("packages: let go of {uploads} uploads and {blobs} blobs"),
761 Err(error) => worker::console_error!("packages: the sweep failed: {error}"),
762 }
763 // Repositories that had a composer.json before the registry did.
764 match packages.composer_backfill().await {
765 Ok(0) => {}
766 Ok(found) => worker::console_log!("packages: found {found} Composer packages"),
767 Err(error) => worker::console_error!("packages: the Composer backfill failed: {error}"),
768 }
769}
770
771#[event(queue)]
772async fn queue(batch: MessageBatch<Event>, env: Env, _ctx: Context) -> Result<()> {
773 let packages = Packages::from_env(&env)?;
774 for message in batch.messages()? {
775 packages.on_event(&env, message.body()).await?;
776 }
777 Ok(())
778}
779
780#[cfg(test)]
781mod tests {
782 use super::*;
783 use serde_json::json;
784
785 fn event(kind: &str, data: serde_json::Value) -> Event {
786 Event {
787 id: "evt_1".into(),
788 kind: kind.into(),
789 source: "identity".into(),
790 time: "2026-10-06T00:00:00.000Z".into(),
791 repo_id: None,
792 actor: None,
793 data,
794 }
795 }
796
797 #[test]
798 fn a_deleted_workspace_is_hidden_and_a_restored_one_shown() {
799 let protected = g1t_contracts::identity::protected_names(Some("wsp_keep"));
800 let now = "2026-10-06T12:00:00.000Z";
801 let deleting = event("workspace.deleting", json!({ "workspaceId": "wsp_1", "slug": "Acme", "by": "ana", "purgeAfter": "x" }));
802 assert_eq!(
803 workspace_mark(&deleting, &protected, now),
804 Some(WorkspaceMark::Hide { workspace: "acme".into(), at: now.into() })
805 );
806 let restored = event("workspace.restored", json!({ "workspaceId": "wsp_1", "slug": "acme" }));
807 assert_eq!(workspace_mark(&restored, &protected, now), Some(WorkspaceMark::Show { workspace: "acme".into() }));
808 }
809
810 #[test]
811 fn protected_workspaces_and_other_events_are_left_alone() {
812 let protected = g1t_contracts::identity::protected_names(Some("wsp_keep"));
813 let now = "2026-10-06T12:00:00.000Z";
814 let flagon = event("workspace.deleting", json!({ "workspaceId": "wsp_f", "slug": "flagon-io" }));
815 assert_eq!(workspace_mark(&flagon, &protected, now), None, "always protected");
816 let by_id = event("workspace.deleting", json!({ "workspaceId": "wsp_keep", "slug": "kept" }));
817 assert_eq!(workspace_mark(&by_id, &protected, now), None, "protected by id");
818 let purged = event("workspace.deleted", json!({ "workspaceId": "wsp_1", "slug": "acme" }));
819 assert_eq!(workspace_mark(&purged, &protected, now), None, "the purge is handled apart");
820 let nameless = event("workspace.deleting", json!({ "workspaceId": "wsp_1", "slug": "" }));
821 assert_eq!(workspace_mark(&nameless, &protected, now), None);
822 }
823}