ecluse
Safe HaskellNone
LanguageGHC2021

Ecluse

Description

Écluse — a supply-chain resilience proxy for package registries.

Écluse (package ecluse) is a lightweight proxy that sits between consumers (developers, CI) and a package registry, applying a configurable resilience policy before any dependency reaches a build — without taking on the cost of hosting packages itself. The name is French for a canal lock: a chamber whose gates never open at once. That is the posture — not a wall that blocks, but a controlled passage every dependency is held in and cleared through before it is admitted to a build.

The goal is resilience, not malware detection: shrink the blast radius of a bad publish — a hijacked maintainer account, a race-to-publish, a typosquat — rather than promise to recognise malice. And Écluse is not a registry: storage is delegated to whatever backend the operator runs (e.g. AWS CodeArtifact, GCP Artifact Registry), and Écluse only governs what may be fetched from, and mirrored to, those backends. npm is the first ecosystem; the domain model is deliberately ecosystem-agnostic so that PyPI and RubyGems can follow.

How a request is cleared

Écluse speaks a registry's native protocol across three read-path registries — the client's, a private upstream of already-vetted packages, and the public registry — and the two request shapes use them differently:

  • A tarball request is gated for that one version: a private-upstream hit is streamed unfiltered (already vetted); on a miss, the proxy fetches the version's public metadata, evaluates the rules, and either streams it from public and enqueues an asynchronous mirror job or returns a denial.
  • A packument (metadata) request is a merge: the private and public upstreams are fetched in parallel, public versions are filtered by the rules while private versions are trusted, and the two are combined into one document (private wins a version collision, an integrity divergence is flagged as a supply-chain signal, and latest is repointed to the newest survivor).

Two properties run through both shapes: the rules engine is deny by default — a version is admitted only if some rule allows it and none denies it — and mirroring is demand-driven, so only versions actually pulled are mirrored, and never on the request's critical path.

How the code is organized

Écluse is a functional core with effects at the edges: the policy and protocol logic is pure and trivially testable, and IO is confined to a thin shell. Swappable backends sit behind handles — records of functions chosen at a single composition root — so a new cloud or a new ecosystem is an added implementation behind an existing handle, not a structural change.

The library's vocabulary, roughly from the pure core outward:

run is the entry point the ecluse executable invokes (see Main). It lives in the library, not in app/Main.hs, so the composition root is a single importable unit and app/Main.hs stays a thin shell that only calls it.

Further reading

docs/architecture.md is the systems-design index: the vision, the end-to-end request lifecycle, and a map to the per-concern design documents. CONTRIBUTING.md covers the codebase layout and testing strategy, and STYLE.md the coding and documentation conventions.

Synopsis

Entry point

run :: IO () Source #

Start Écluse: the entry point the ecluse executable runs (see Main).

It assembles the composition root from configuration: it parses the environment layer and the optional config document, __validates everything and fails fast at boot__ on any problem (a malformed env, an unresolved rule policy, a configured mount with no adapter, a credential reference that does not resolve, or a mirror-queue backend that is not built in this binary), aggregating the failures so a single run reports them all. On success it builds the handles — the shared HTTP Manager, the config-selected mirror queue, the metadata cache, the logger, the process-global credential provider, and the telemetry substrate (off unless PROXY_TELEMETRY enables it) — into an Env, derives the served mount bindings, then runs the server and the mirror worker concurrently over that single Env (runServer and runWorker). Bracketing the Env (and the telemetry providers) for the lifetime of both means their shared resources are torn down along every exit path.

Split-ready services

runServer :: ServerConfig -> Env -> IO () Source #

Run the proxy's HTTP front door over the composition-root Env with the config-derived ServerConfig.

This is the npm-aware composition site: mountBindingFor mounts npm — its path grammar (Ecluse.Registry.Npm.Route) and its denial renderer (Ecluse.Registry.Npm.Serve) — into the otherwise ecosystem-neutral web layer (runServer), so the agnostic server stays closed over the shared Route set and only this one place names an ecosystem. Splitting the server into its own binary later reuses this same entry.

runWorker :: Env -> IO () Source #

Run the supervised mirror worker over the composition-root Env: the consume → fetch → verify → publish → ack loop against the queue, the publish-side registry client, and the credential handle, in the App orchestration monad. The loop logic lives in Ecluse.Worker; this is the composition-root entry the single-process program runs alongside runServer.

npm front door

npmServerConfig :: ServerConfig Source #

The fallback server settings: a single npm mount with no packument-serve dependencies, so the packument route is the recognised-but-unserved 501 stub. Exposed so the composed front door can be driven directly without binding a socket (e.g. embedded in another wai application, or exercised in tests through application) to assert the routing and the unwired-mount surface; a real launch derives its bindings from configuration in run.

mountBindingFor :: Ecosystem -> Maybe PackumentDeps -> Maybe MountBinding Source #

Resolve an Ecosystem to its complete MountBinding, or Nothing when that ecosystem has no adapter wired. The ecosystem selects its path grammar (the Classifier) and its denial renderer (the MountRenderer), and its path prefix is derived from it (prefixFor) rather than configured — so the ecosystem is the single thing that drives the binding (see docs/architecture/hosting.mdMounts). The packument-serve dependencies are passed in (the composition root supplies them once the per-mount registry set is resolved); Nothing for them leaves the packument route the recognised-but-unserved 501 stub.

npm is the only ecosystem with an adapter; the others have no registry client or renderer, so they resolve to Nothing — a loud miss at the call site rather than a silently half-wired mount.

Composition glue (exposed for direct testing)

orExit :: (e -> Text) -> Either e a -> IO a Source #

data BootAborted Source #

Raised to abort start-up after a boot phase has reported its aggregated failure to stderr. A distinct type — rather than a bare exitFailure — so the abort is observable in a test without the process actually exiting; uncaught, it propagates to main and the runtime exits non-zero, the operator-facing fail-fast.

Constructors

BootAborted 

Default handles

unconfiguredRegistry :: RegistryClient Source #

A registry handle with no backend behind it: every effectful field __refuses loudly__ (a typed RegistryUnconfigured) and every pure parse* field returns Left, so an unconfigured fetch/publish or parse fails explicitly rather than silently returning a fabricated success. It holds the handle slot in the composition root where a configured backend is selected elsewhere.

unconfiguredCredentials :: CredentialProvider Source #

A credential handle with no backend behind it: a static, non-expiring empty secret. It holds the CredentialProvider slot in the composition root until a live backend is selected — for the mirror-target write, and for the private-upstream read under the service / delegated-cache strategies. The default passthrough strategy needs no read credential at all (reads forward the caller's own token), so this empty placeholder is harmless on the serve path there. See docs/architecture/access-model.md.