Implement support for GCP Pub/Sub #354
Conversation
WalkthroughAdds Docker emulators and two new packages: a comprehensive Google Cloud Pub/Sub integration ( Changes
Sequence Diagram(s)sequenceDiagram
participant App
participant Manager as PubSubPublisherManager
participant Factory as CommonPubSubPublisherFactory
participant Publisher as AbstractPubSubPublisher
participant PubSub as Google Pub/Sub
App->>Manager: publish(topic, message)
Manager->>Manager: resolveEventTarget & resolveCreationConfig
alt publisher exists
Manager->>Publisher: publish(message, options)
else
Manager->>Factory: buildPublisher(deps, options)
Factory->>Publisher: instantiate
Publisher->>PubSub: ensure topic (init)
Manager->>Publisher: publish(message, options)
end
Publisher->>Publisher: validate schema, check deduplication
alt duplicate
Publisher-->>App: skip (logged)
else
Publisher->>Publisher: build offload attributes
Publisher->>PubSub: send message + attributes
PubSub-->>Publisher: messageId
Publisher-->>App: success
end
sequenceDiagram
participant PubSub as Google Pub/Sub
participant Consumer as AbstractPubSubConsumer
participant Resolver as PubSubConsumerErrorResolver
participant Handler as Message Handler
PubSub->>Consumer: deliver message
Consumer->>Consumer: deserialize & parse
alt format/validation error
Consumer->>Resolver: processError(error)
Resolver-->>Consumer: standardized MessageError
Consumer->>PubSub: ack or route to DLQ per policy
else normal
Consumer->>Consumer: run pre-handlers & barriers
Consumer->>Handler: invoke handler
alt success
Handler-->>Consumer: success
Consumer->>PubSub: ack
else retry
Handler-->>Consumer: retryLater
Consumer->>PubSub: nack / retry
end
end
Consumer->>Consumer: release locks / cleanup
Estimated code review effort🎯 4 (Complex) | ⏱️ ~45 minutes Areas to focus:
Suggested reviewers
Poem
Pre-merge checks and finishing touches❌ Failed checks (1 warning)
✅ Passed checks (2 passed)
✨ Finishing touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 5
🧹 Nitpick comments (19)
docker-compose.yml (1)
72-85: New emulator services look good; optionally restrict host exposure.The GCS and Pub/Sub emulator definitions are consistent with the rest of the stack. To avoid exposing them beyond the host (like you already do for localstack), you might bind them to
127.0.0.1(e.g.,'127.0.0.1:4443:4443'and'127.0.0.1:8085:8085').packages/gcp-pubsub/test/utils/cleanupPubSub.ts (1)
3-42: Cleanup helpers are safe; consider minor DRY and idempotency tweaks.The existence checks before delete make these utilities safe for tests. If you want them fully idempotent even under races, you could (a) factor out a small “delete if exists” helper used by both subscription/topic deletions and (b) optionally tolerate not-found errors from
delete()in addition to the explicitexists()checks.packages/gcs-payload-store/package.json (1)
1-66: Manifest looks solid; consider addingtypesand narrowingdocker:stop.The package.json is generally ready for publishing. Two possible improvements:
- Add a
"types": "./dist/index.d.ts"field so TS consumers don’t rely solely onexports.- Change
"docker:stop": "docker compose down"to only stop the emulator service (e.g.,docker compose stop gcs-emulator), so running it from this package doesn’t tear down unrelated services.packages/gcs-payload-store/lib/GCSPayloadStore.ts (1)
16-120: GCS payload store looks robust; only minor polish opportunities.The config resolver, key generation with optional prefix, and handling of both string and stream payloads all look correct, and treating missing objects as
null/ no-op makes operations idempotent. If you want to tighten things further, you could:
- Extract a small helper for generating the object key (UUID + optional prefix) so
storePayloadis a bit clearer.- Share a tiny “isNotFound error” helper between
retrievePayloadanddeletePayloadto avoid repeating the(error as any)?.codecheck.packages/gcp-pubsub/lib/utils/pubSubInitter.ts (1)
8-176: Pub/Sub init/delete flow is sound; clarify config precedence and flag behavior.The topic/subscription creation and DLQ wiring logic look correct, and
deletePubSub’s delete-if-exists behavior is straightforward. Two edge cases worth considering:
- If callers accidentally pass both
locatorConfigandcreationConfig, the locator path silently wins. You might want to reject that combination explicitly or document the precedence to avoid surprises.PubSubCreationConfig.updateAttributesIfExistsandDeletionConfig.waitForConfirmation/forceDeleteInProductionaren’t used here; if higher-level code expects those flags to have an effect, this could be confusing. Either wire them in or document that they’re ignored by these helpers.packages/gcp-pubsub/vitest.config.ts (1)
17-22: Consider increasing branch coverage.Branch coverage is at 62% while function coverage is at 87%, suggesting edge cases and error paths may not be fully tested. Consider adding tests for conditional branches to improve overall coverage quality.
packages/gcp-pubsub/lib/utils/pubSubMessageDeserializer.ts (1)
6-21: Consider broadening the return type or clarifying error handling.The function returns
Either<MessageInvalidFormatError, unknown>, but line 15 may return other error types (e.g.,MessageValidationError,InternalError) whenerrorResolver.processError()transforms the error. This could lead to type inconsistencies.Consider updating the return type to reflect all possible error types:
export function deserializePubSubMessage( message: PubSubMessage, errorResolver: ErrorResolver, -): Either<MessageInvalidFormatError, unknown> { +): Either<MessageInvalidFormatError | MessageValidationError | InternalError, unknown> {Or, if the intent is to only return
MessageInvalidFormatError, add explicit type validation after error resolution.packages/gcp-pubsub/README.md (2)
70-72: Add language identifier to fenced code block.-``` +```text Publisher → Topic → Subscription → ConsumerBased on static analysis hints. --- `672-678`: **Add language identifier to fenced code block.** ```diff -``` +```text Attempt 1: Message nacked, redelivered by Pub/Sub Attempt 2: Message nacked, redelivered by Pub/Sub Attempt 3: Message nacked, redelivered by Pub/Sub ... After maxDeliveryAttempts: Sent to DLQBased on static analysis hints. </blockquote></details> <details> <summary>packages/gcp-pubsub/test/consumers/PubSubPermissionConsumer.spec.ts (2)</summary><blockquote> `94-127`: **Test name suggests ordering guarantees that aren’t asserted** `it('consumes multiple messages in order', ...)` only waits for each ID to be processed, but doesn’t assert any ordering semantics (e.g., by comparing processing sequence). If strict ordering isn’t guaranteed/required, consider renaming the test to avoid implying stronger guarantees than are actually verified; otherwise, extend the assertions to validate order explicitly (e.g., via the spy’s call sequence). --- `172-186`: **Fixed 1s sleep in invalid-format test may be brittle** The invalid-format test uses a hard-coded `setTimeout(1000)` and only inspects counters. That can be slightly brittle (slow CI, emulator hiccups) and doesn’t directly assert that the error path was exercised. If the handler spy can observe error statuses for invalid-format messages, consider waiting on the spy instead of a fixed timeout so the test becomes faster and more deterministic. </blockquote></details> <details> <summary>packages/gcp-pubsub/test/publishers/PubSubPermissionPublisher.ts (1)</summary><blockquote> `15-30`: **Tighten option typing and consider dropping redundant publish override** `PubSubPermissionPublisherOptions` works fine for tests, but there are a couple of small cleanups you might consider: - Replace `payloadStoreConfig?: any` with a more specific type (or `unknown`) to avoid leaking `any` into the test surface. - The explicit `publish` override just forwards to `super.publish` without changing behavior; unless you expect to extend it later, you could omit the override and rely on the base implementation to keep the class slimmer. These are optional and don’t affect correctness. </blockquote></details> <details> <summary>packages/gcp-pubsub/test/pubsub/PubSubPublisherManager.spec.ts (1)</summary><blockquote> `196-243`: **Minor schema choice nit in injected publisher setup** In the `injectPublisher` test you build `CommonPubSubPublisher` with `messageSchemas: [injectedSchema.consumerSchema]`. If `publisherSchema` and `consumerSchema` ever diverge, this might become confusing. Using `injectedSchema.publisherSchema` here (to mirror the publishing side) could make the intent clearer, even if they’re currently identical. </blockquote></details> <details> <summary>packages/gcs-payload-store/test/utils/gcsUtils.ts (1)</summary><blockquote> `3-49`: **Consider pagination when clearing buckets (optional)** For test usage these helpers look fine. Be aware that `bucket.getFiles()` only returns the first page of objects; if a test ever writes enough objects to trigger pagination, `assertEmptyBucket` would leave some behind. If that’s a realistic scenario, consider switching to the paged API (`getFiles({ autoPaginate: true })`) or looping over pages. </blockquote></details> <details> <summary>packages/gcp-pubsub/test/utils/testContext.ts (2)</summary><blockquote> `22-43`: **Logger stub works but could avoid `@ts-expect-error`** Using `console` as a `CommonLogger` stub is fine for tests, but the `// @ts-expect-error` suggests the interfaces don’t quite match. If you want to tighten this up later, you could define a small adapter that implements `CommonLogger` and forwards to `console`, removing the need for the suppression. --- `63-120`: **Stubbed managers typed as non-optional may hide `undefined` at runtime** `transactionObservabilityManager` and `messageMetricsManager` are registered as functions returning `undefined`, while the `Dependencies` interface types them as `TransactionObservabilityManager` and `MessageMetricsManager` respectively. That’s fine as long as all consumers treat them as optional, but it’s easy for future code to accidentally call methods on them and hit `undefined`. Two possible follow-ups (no rush, test-only): - Change their types in `Dependencies` to `TransactionObservabilityManager | undefined` / `MessageMetricsManager | undefined`, or - Provide minimal no-op implementations that satisfy the interfaces instead of returning `undefined`. Similarly, you may want to consider marking `gcsStorage` and `consumerErrorResolver` as singletons for consistency with other shared test dependencies, although it’s not required for correctness. </blockquote></details> <details> <summary>packages/gcp-pubsub/lib/pubsub/AbstractPubSubPublisher.ts (1)</summary><blockquote> `50-133`: **Publish flow looks correct; consider minor polish around serialization and error wrapping** The overall publish pipeline (schema resolution, lazy init, optional logging, dedup, offloading, and publish) is coherent and matches the core abstractions. Two small, non-blocking nits: - You serialize the message twice (`offloadMessagePayloadIfNeeded` size callback and `sendMessage`). If payloads get large and throughput is high, you might later want to let the offloader return (or reuse) the encoded `Buffer` to avoid the second stringify. - `messageSchemaResult.error` is thrown before the `try` block, while other failures are wrapped into `InternalError`. If you want a single error surface for callers, consider wrapping schema-resolution errors the same way. Nothing here is blocking; just potential future cleanup. </blockquote></details> <details> <summary>packages/gcp-pubsub/lib/pubsub/PubSubPublisherManager.ts (1)</summary><blockquote> `50-77`: **Verify schema-type mapping between `PubSubMessageSchemaType` and the manager’s `publish` return type** You introduce `PubSubMessageSchemaType<T> = z.output<T['publisherSchema']>` and pass it as the 7th generic argument to `AbstractPublisherManager`, but the overridden `publish` method still uses the core alias: ```ts override publish(...): Promise<MessageSchemaType<SupportedEventDefinitions[number]>>where
MessageSchemaTypein core is based onconsumerSchema.Depending on how
AbstractPublisherManageris parameterized, this can mean:
- The manager is instantiated with one schema type (
PubSubMessageSchemaType), but- The visible
publishreturn type exposes a different schema (MessageSchemaType).Please double-check that:
- The 7th generic argument to
AbstractPublisherManagermatches the type you actually want to expose frompublish, andMessageSchemaType<SupportedEventDefinitions[number]>here is indeed assignable to that generic argument.If they’re intentionally different (e.g. output vs input types, or publisher vs consumer schemas), it might be worth adding a short comment to document that asymmetry.
Also applies to: 101-110
packages/gcp-pubsub/lib/pubsub/AbstractPubSubConsumer.ts (1)
27-75: Minor cleanup: unused constant and unused consumer override fieldsTwo small cleanups you can consider:
_ABORT_EARLY_EITHERis defined but unused; if you’re no longer using the “abort early” pattern, it can be safely removed.PubSubConsumerOptionsexposesconsumerOverrides.batching, butstart()only appliesconsumerOverrides.flowControltosubscription.setOptions. If batching overrides are intended to be supported, you might want to wire them through here; otherwise, consider dropping them from the options to avoid confusion.Neither is blocking, but tightening these up will reduce mental overhead for future readers.
Also applies to: 196-203, 495-511
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (44)
docker-compose.yml(1 hunks)packages/gcp-pubsub/README.md(1 hunks)packages/gcp-pubsub/lib/errors/PubSubConsumerErrorResolver.ts(1 hunks)packages/gcp-pubsub/lib/fakes/FakeConsumerErrorResolver.ts(1 hunks)packages/gcp-pubsub/lib/index.ts(1 hunks)packages/gcp-pubsub/lib/pubsub/AbstractPubSubConsumer.ts(1 hunks)packages/gcp-pubsub/lib/pubsub/AbstractPubSubPublisher.ts(1 hunks)packages/gcp-pubsub/lib/pubsub/AbstractPubSubService.ts(1 hunks)packages/gcp-pubsub/lib/pubsub/CommonPubSubPublisherFactory.ts(1 hunks)packages/gcp-pubsub/lib/pubsub/PubSubPublisherManager.ts(1 hunks)packages/gcp-pubsub/lib/schemas/pubSubSchemas.ts(1 hunks)packages/gcp-pubsub/lib/types/MessageTypes.ts(1 hunks)packages/gcp-pubsub/lib/utils/messageUtils.ts(1 hunks)packages/gcp-pubsub/lib/utils/pubSubInitter.ts(1 hunks)packages/gcp-pubsub/lib/utils/pubSubMessageDeserializer.ts(1 hunks)packages/gcp-pubsub/lib/utils/pubSubMessageReader.ts(1 hunks)packages/gcp-pubsub/lib/utils/pubSubUtils.ts(1 hunks)packages/gcp-pubsub/package.json(1 hunks)packages/gcp-pubsub/test/consumers/PubSubPermissionConsumer.spec.ts(1 hunks)packages/gcp-pubsub/test/consumers/PubSubPermissionConsumer.ts(1 hunks)packages/gcp-pubsub/test/consumers/userConsumerSchemas.ts(1 hunks)packages/gcp-pubsub/test/fakes/FakeLogger.ts(1 hunks)packages/gcp-pubsub/test/publishers/PubSubPermissionPublisher.spec.ts(1 hunks)packages/gcp-pubsub/test/publishers/PubSubPermissionPublisher.ts(1 hunks)packages/gcp-pubsub/test/pubsub/PubSubPublisherManager.spec.ts(1 hunks)packages/gcp-pubsub/test/utils/cleanRedis.ts(1 hunks)packages/gcp-pubsub/test/utils/cleanupPubSub.ts(1 hunks)packages/gcp-pubsub/test/utils/testContext.ts(1 hunks)packages/gcp-pubsub/test/utils/testPubSubConfig.ts(1 hunks)packages/gcp-pubsub/test/utils/testRedisConfig.ts(1 hunks)packages/gcp-pubsub/tsconfig.build.json(1 hunks)packages/gcp-pubsub/tsconfig.json(1 hunks)packages/gcp-pubsub/vitest.config.ts(1 hunks)packages/gcs-payload-store/README.md(1 hunks)packages/gcs-payload-store/lib/GCSPayloadStore.ts(1 hunks)packages/gcs-payload-store/lib/index.ts(1 hunks)packages/gcs-payload-store/package.json(1 hunks)packages/gcs-payload-store/test/store/GCSPayloadStore.spec.ts(1 hunks)packages/gcs-payload-store/test/utils/gcsUtils.ts(1 hunks)packages/gcs-payload-store/test/utils/streamUtils.ts(1 hunks)packages/gcs-payload-store/test/utils/testGCSConfig.ts(1 hunks)packages/gcs-payload-store/tsconfig.build.json(1 hunks)packages/gcs-payload-store/tsconfig.json(1 hunks)packages/gcs-payload-store/vitest.config.ts(1 hunks)
🧰 Additional context used
🧬 Code graph analysis (17)
packages/gcp-pubsub/lib/schemas/pubSubSchemas.ts (1)
packages/gcp-pubsub/lib/pubsub/PubSubPublisherManager.ts (1)
PubSubAwareEventDefinition(22-22)
packages/gcp-pubsub/lib/utils/pubSubMessageDeserializer.ts (2)
packages/gcp-pubsub/lib/types/MessageTypes.ts (1)
PubSubMessage(4-4)packages/gcp-pubsub/lib/utils/pubSubMessageReader.ts (1)
readPubSubMessage(5-23)
packages/gcp-pubsub/lib/utils/pubSubMessageReader.ts (1)
packages/gcp-pubsub/lib/types/MessageTypes.ts (1)
PubSubMessage(4-4)
packages/gcp-pubsub/lib/pubsub/CommonPubSubPublisherFactory.ts (2)
packages/gcp-pubsub/lib/pubsub/AbstractPubSubService.ts (3)
PubSubDependencies(10-12)PubSubCreationConfig(48-52)PubSubQueueLocatorType(54-57)packages/core/lib/types/queueOptionsTypes.ts (1)
QueuePublisherOptions(115-122)
packages/gcp-pubsub/test/publishers/PubSubPermissionPublisher.spec.ts (3)
packages/gcp-pubsub/test/utils/testContext.ts (1)
registerDependencies(25-130)packages/gcp-pubsub/test/publishers/PubSubPermissionPublisher.ts (1)
PubSubPermissionPublisher(32-59)packages/gcp-pubsub/test/utils/cleanupPubSub.ts (1)
deletePubSubTopic(3-9)
packages/gcp-pubsub/test/consumers/PubSubPermissionConsumer.spec.ts (4)
packages/gcp-pubsub/test/utils/testContext.ts (1)
registerDependencies(25-130)packages/gcp-pubsub/test/consumers/PubSubPermissionConsumer.ts (1)
PubSubPermissionConsumer(60-140)packages/gcp-pubsub/test/publishers/PubSubPermissionPublisher.ts (1)
PubSubPermissionPublisher(32-59)packages/gcp-pubsub/test/utils/cleanupPubSub.ts (1)
deletePubSubTopicAndSubscription(27-42)
packages/gcs-payload-store/test/store/GCSPayloadStore.spec.ts (4)
packages/gcs-payload-store/lib/GCSPayloadStore.ts (2)
GCSPayloadStore(43-120)resolvePayloadStoreConfig(16-32)packages/gcs-payload-store/test/utils/testGCSConfig.ts (1)
createTestGCSClient(8-13)packages/gcs-payload-store/test/utils/gcsUtils.ts (3)
assertEmptyBucket(3-16)getObjectContent(18-27)objectExists(29-38)packages/gcs-payload-store/test/utils/streamUtils.ts (1)
streamToString(3-10)
packages/gcp-pubsub/test/publishers/PubSubPermissionPublisher.ts (3)
packages/gcp-pubsub/test/consumers/userConsumerSchemas.ts (4)
PERMISSIONS_ADD_MESSAGE_TYPE(18-18)PERMISSIONS_REMOVE_MESSAGE_TYPE(19-19)PERMISSIONS_ADD_MESSAGE_SCHEMA(3-9)PERMISSIONS_REMOVE_MESSAGE_SCHEMA(11-16)packages/gcp-pubsub/lib/pubsub/AbstractPubSubService.ts (1)
PubSubDependencies(10-12)packages/gcp-pubsub/lib/pubsub/AbstractPubSubPublisher.ts (1)
PubSubMessageOptions(23-26)
packages/gcp-pubsub/lib/utils/pubSubInitter.ts (2)
packages/gcp-pubsub/lib/pubsub/AbstractPubSubService.ts (2)
PubSubQueueLocatorType(54-57)PubSubCreationConfig(48-52)packages/core/lib/types/queueOptionsTypes.ts (1)
DeletionConfig(95-99)
packages/gcp-pubsub/lib/pubsub/AbstractPubSubPublisher.ts (4)
packages/gcp-pubsub/lib/pubsub/AbstractPubSubService.ts (3)
PubSubDependencies(10-12)PubSubCreationConfig(48-52)PubSubQueueLocatorType(54-57)packages/core/lib/types/queueOptionsTypes.ts (1)
QueuePublisherOptions(115-122)packages/gcp-pubsub/lib/utils/messageUtils.ts (1)
buildOffloadedPayloadAttributes(7-23)packages/gcp-pubsub/lib/types/MessageTypes.ts (1)
PubSubMessage(4-4)
packages/gcp-pubsub/test/pubsub/PubSubPublisherManager.spec.ts (4)
packages/gcp-pubsub/lib/schemas/pubSubSchemas.ts (1)
PubSubAwareEventDefinition(3-5)packages/gcp-pubsub/test/utils/testContext.ts (2)
Dependencies(134-148)registerDependencies(25-130)packages/gcp-pubsub/lib/pubsub/CommonPubSubPublisherFactory.ts (1)
CommonPubSubPublisher(20-22)packages/core/lib/messages/MetadataFiller.ts (1)
CommonMetadataFiller(29-66)
packages/gcp-pubsub/test/utils/testContext.ts (5)
packages/gcp-pubsub/test/utils/testPubSubConfig.ts (1)
TEST_PUBSUB_CONFIG(1-4)packages/gcp-pubsub/lib/errors/PubSubConsumerErrorResolver.ts (1)
PubSubConsumerErrorResolver(6-32)packages/gcp-pubsub/test/utils/testRedisConfig.ts (1)
TEST_REDIS_CONFIG(1-10)packages/gcp-pubsub/test/consumers/PubSubPermissionConsumer.ts (1)
PubSubPermissionConsumer(60-140)packages/gcp-pubsub/test/publishers/PubSubPermissionPublisher.ts (1)
PubSubPermissionPublisher(32-59)
packages/gcp-pubsub/lib/fakes/FakeConsumerErrorResolver.ts (2)
packages/gcp-pubsub/lib/errors/PubSubConsumerErrorResolver.ts (1)
PubSubConsumerErrorResolver(6-32)packages/amqp/test/fakes/FakeConsumerErrorResolver.ts (1)
FakeConsumerErrorResolver(3-23)
packages/gcp-pubsub/lib/pubsub/AbstractPubSubService.ts (3)
packages/core/lib/types/queueOptionsTypes.ts (2)
QueueDependencies(12-16)QueueOptions(109-113)packages/gcp-pubsub/lib/types/MessageTypes.ts (1)
PubSubMessage(4-4)packages/gcp-pubsub/lib/utils/pubSubInitter.ts (2)
deletePubSub(152-176)initPubSub(32-150)
packages/gcp-pubsub/lib/pubsub/AbstractPubSubConsumer.ts (6)
packages/gcp-pubsub/lib/pubsub/AbstractPubSubService.ts (3)
PubSubDependencies(10-12)PubSubCreationConfig(48-52)PubSubQueueLocatorType(54-57)packages/core/lib/types/queueOptionsTypes.ts (2)
QueueConsumerDependencies(73-76)QueueConsumerOptions(134-152)packages/gcp-pubsub/lib/utils/pubSubInitter.ts (2)
deletePubSub(152-176)initPubSub(32-150)packages/gcp-pubsub/lib/types/MessageTypes.ts (1)
PubSubMessage(4-4)packages/gcp-pubsub/lib/utils/pubSubMessageDeserializer.ts (1)
deserializePubSubMessage(6-21)packages/gcp-pubsub/lib/utils/messageUtils.ts (1)
hasOffloadedPayload(3-5)
packages/gcp-pubsub/lib/pubsub/PubSubPublisherManager.ts (6)
packages/gcp-pubsub/lib/schemas/pubSubSchemas.ts (1)
PubSubAwareEventDefinition(3-5)packages/gcp-pubsub/lib/pubsub/AbstractPubSubService.ts (3)
PubSubDependencies(10-12)PubSubCreationConfig(48-52)PubSubQueueLocatorType(54-57)packages/gcp-pubsub/lib/pubsub/CommonPubSubPublisherFactory.ts (2)
PubSubPublisherFactory(10-18)CommonPubSubPublisherFactory(24-33)packages/core/lib/messages/MetadataFiller.ts (1)
MetadataFiller(18-27)packages/gcp-pubsub/lib/pubsub/AbstractPubSubPublisher.ts (2)
PubSubMessageOptions(23-26)publish(50-119)packages/core/lib/queues/AbstractPublisherManager.ts (2)
MessagePublishType(11-11)MessageSchemaType(13-13)
packages/gcp-pubsub/test/consumers/PubSubPermissionConsumer.ts (2)
packages/gcp-pubsub/test/consumers/userConsumerSchemas.ts (4)
PERMISSIONS_ADD_MESSAGE_TYPE(18-18)PERMISSIONS_REMOVE_MESSAGE_TYPE(19-19)PERMISSIONS_REMOVE_MESSAGE_SCHEMA(11-16)PERMISSIONS_ADD_MESSAGE_SCHEMA(3-9)packages/gcp-pubsub/lib/pubsub/AbstractPubSubConsumer.ts (2)
PubSubConsumerOptions(48-74)PubSubConsumerDependencies(46-46)
🪛 LanguageTool
packages/gcp-pubsub/README.md
[style] ~78-~78: This adverb was used twice in the sentence. Consider removing one of them or replacing them with a synonym.
Context: ...to subscriptions - ❌ You CANNOT consume directly from topics - ✅ One topic can have mult...
(ADVERB_REPETITION_PREMIUM)
🪛 markdownlint-cli2 (0.18.1)
packages/gcp-pubsub/README.md
70-70: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
672-672: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
🔇 Additional comments (33)
packages/gcp-pubsub/test/utils/cleanRedis.ts (1)
1-5: LGTM! Clean test utility implementation.The function is simple, well-typed, and appropriate for its purpose as a test cleanup utility. Using
flushdb()correctly clears the current Redis database, which is the expected behavior for isolating test state.packages/gcp-pubsub/tsconfig.json (1)
1-7: LGTM! Clean TypeScript configuration.The configuration appropriately extends the base config, includes all necessary paths, and adds Vitest globals for testing.
packages/gcp-pubsub/lib/types/MessageTypes.ts (1)
1-4: LGTM! Clean type aliasing.The type alias provides a consistent abstraction over the Google Cloud SDK Message type, which is a good practice for maintaining a clean public API.
packages/gcs-payload-store/tsconfig.json (1)
1-7: LGTM! Consistent configuration across packages.The configuration mirrors the gcp-pubsub package setup, which maintains consistency across the monorepo.
packages/gcs-payload-store/tsconfig.build.json (1)
1-5: LGTM! Standard build configuration.The build configuration appropriately targets library code and excludes test files from the published package.
packages/gcp-pubsub/tsconfig.build.json (1)
1-5: LGTM! Consistent build configuration.The configuration mirrors the gcs-payload-store package, maintaining consistency in how public libraries are built.
packages/gcs-payload-store/vitest.config.ts (1)
1-25: LGTM! Well-configured test setup.The Vitest configuration is appropriate with reasonable coverage thresholds. The singleThread setting ensures deterministic test execution, which is beneficial for integration tests with external services like GCS.
packages/gcs-payload-store/test/utils/testGCSConfig.ts (1)
1-13: Port configuration verified—approval confirmed.The GCS emulator port in the test configuration (4443) correctly matches the docker-compose service port mapping (4443:4443) and the startup command (-port 4443). The test client configuration is properly aligned with the emulator setup.
packages/gcp-pubsub/test/utils/testRedisConfig.ts (1)
1-10: Redis configuration correctly aligns with docker-compose.The test configuration password ('sOmE_sEcUrE_pAsS'), port (6379), and default host settings match the docker-compose Redis service configuration. Verification confirms proper alignment.
packages/gcp-pubsub/test/consumers/userConsumerSchemas.ts (1)
1-19: Permission message schemas are well-structured for add/remove flows.Good use of a discriminating
messageTypefield and stricter timestamp/userIds requirements forremovevsadd. The inferred types should keep the publisher/consumer tests nicely aligned with the wire format.packages/gcp-pubsub/test/fakes/FakeLogger.ts (1)
1-52: FakeLogger is a pragmatic test double.The in-memory message/warning/error buffers and
child()returning the same instance are a good fit for tests. This should drop into any CommonLogger-using code without surprises.packages/gcs-payload-store/lib/index.ts (1)
1-1: LGTM!The barrel export is correctly structured for ESM compatibility.
packages/gcs-payload-store/test/utils/streamUtils.ts (1)
3-10: LGTM!The stream-to-string conversion is correctly implemented with proper error handling and Buffer management.
packages/gcp-pubsub/lib/utils/pubSubMessageReader.ts (1)
5-23: LGTM!The message reader properly handles JSON parsing errors and returns a type-safe Either result with appropriate error details.
packages/gcp-pubsub/test/utils/testPubSubConfig.ts (1)
1-4: LGTM!The test configuration appropriately points to the Pub/Sub emulator endpoint.
packages/gcp-pubsub/vitest.config.ts (1)
4-12: LGTM!The single-thread pool configuration is appropriate for integration tests that may share state with the Pub/Sub emulator.
packages/gcp-pubsub/lib/schemas/pubSubSchemas.ts (1)
3-5: LGTM!The type definition correctly extends CommonEventDefinition with an optional Pub/Sub-specific topic field.
packages/gcp-pubsub/lib/utils/pubSubUtils.ts (2)
11-13: Global environment modification side effect.Setting
process.env.PUBSUB_EMULATOR_HOSTglobally can affect other code or tests in the same process. This is the standard pattern for GCP emulators, but be aware that once set, it affects all subsequent PubSub client creations in the process.
8-18: LGTM!The type-only import combined with dynamic require is an intentional pattern to avoid bundling issues with optional peer dependencies.
packages/gcs-payload-store/README.md (1)
1-253: LGTM!The documentation is well-structured, comprehensive, and provides clear examples for installation, configuration, lifecycle management, and best practices.
Also applies to: 282-345
packages/gcs-payload-store/test/store/GCSPayloadStore.spec.ts (1)
13-145: LGTM!The test suite is comprehensive and well-structured, covering all major scenarios including:
- Payload storage (string and stream inputs)
- Key prefix functionality
- Retrieval with proper null handling
- Deletion with graceful error handling
- Configuration resolution edge cases
packages/gcp-pubsub/lib/utils/messageUtils.ts (1)
1-23: LGTM!The utility functions correctly handle offloaded payload detection and attribute building. The type guards are appropriate, and the string conversion for the size attribute is handled properly.
packages/gcp-pubsub/lib/fakes/FakeConsumerErrorResolver.ts (1)
3-23: LGTM!The test fake correctly extends
PubSubConsumerErrorResolverand provides error tracking functionality. The implementation follows established patterns from other adapters in the codebase (e.g.,packages/amqp/test/fakes/FakeConsumerErrorResolver.ts).packages/gcp-pubsub/test/publishers/PubSubPermissionPublisher.spec.ts (1)
8-145: LGTM!The test suite provides good coverage of publisher functionality including:
- Topic creation and idempotent initialization
- Publishing single and multiple messages
- Publishing with ordering keys and custom attributes
- Handler spy integration for verifying published messages
packages/gcp-pubsub/lib/errors/PubSubConsumerErrorResolver.ts (1)
6-32: LGTM!The error resolver correctly maps different error types to standardized internal errors:
SyntaxError→MessageInvalidFormatErrorZodError→MessageValidationError(with validation details)- Standardized errors →
InternalError(preserving error code)- Fallback → Generic
InternalErrorThe error handling logic is comprehensive and follows established patterns.
packages/gcp-pubsub/package.json (1)
34-34: No security advisories found.The version @google-cloud/pubsub ^5.2.0 has no known security vulnerabilities according to GitHub's security database. The dependency is safe to use.
packages/gcp-pubsub/test/consumers/PubSubPermissionConsumer.spec.ts (1)
15-40: Lifecycle setup and DI usage look solidReinitializing the consumer/publisher and cleaning up topics/subscriptions in
beforeEachgives good isolation between tests and mirrors the actual lifecycle of the components. This is a sensible pattern for these integration-style tests.packages/gcp-pubsub/lib/pubsub/CommonPubSubPublisherFactory.ts (1)
10-32: Generic publisher factory and CommonPubSubPublisher wiring look correctThe
PubSubPublisherFactorysignature andCommonPubSubPublisherFactoryimplementation correctly threadPublisherBaseEventTypeandQueuePublisherOptions<PubSubCreationConfig, PubSubQueueLocatorType, M>. The minimalCommonPubSubPublishersubclass keeps the abstraction clear without introducing extra behavior.packages/gcp-pubsub/test/pubsub/PubSubPublisherManager.spec.ts (1)
17-48: Event definitions and type plumbing give good coverageThe combination of
TestEvents,PubSubAwareEventDefinition, andTestEventPublishPayloadsTypegives strong type-safety for thePubSubPublisherManagertests while still being readable. This nicely exercises the manager against realistic event metadata and payload schemas.packages/gcp-pubsub/lib/index.ts (1)
1-14: Barrel exports are coherent and comprehensiveThis index file cleanly exposes the intended public surface (errors, abstract Pub/Sub primitives, factory, manager, schemas, and utilities) without over-coupling tests. It should make the package much easier to consume.
packages/gcp-pubsub/test/utils/testContext.ts (1)
132-148: DI config and returned container shape look consistent
DiConfigasRecord<keyof Dependencies, Resolver<any>>and theDependenciesinterface align with what you register indiConfig, and returning the Awilix container fromregisterDependenciesintegrates cleanly with the tests (AwilixContainer<Dependencies>usage). The async init/dispose setup for Redis and the permission consumer/publisher looks well thought out.packages/gcp-pubsub/lib/pubsub/AbstractPubSubService.ts (1)
10-105: Base Pub/Sub service abstraction looks consistent with core queue serviceType wiring into
AbstractQueueServiceand theinit/closelifecycle look correct, and the dependency contract (PubSubDependencies) is minimal and clear. No blocking issues from my side here.packages/gcp-pubsub/test/consumers/PubSubPermissionConsumer.ts (1)
20-138: Test consumer wiring into AbstractPubSubConsumer looks goodThe options narrowing, default
creationConfig, and handler wiring viaMessageHandlerConfigBuilderare all consistent with the core consumer abstractions. Arrow-function handlers correctly capturethisfor counters and tracking, so this should be a solid functional-test harness.
| export type PubSubPublisherManagerOptions< | ||
| T extends AbstractPubSubPublisher<EventType>, | ||
| EventType extends PublisherBaseEventType, | ||
| MetadataType, | ||
| > = { | ||
| metadataField?: string | ||
| publisherFactory?: PubSubPublisherFactory<T, EventType> | ||
| metadataFiller: MetadataFiller<EventType, MetadataType> | ||
| newPublisherOptions: Omit< | ||
| import('@message-queue-toolkit/core').QueuePublisherOptions< | ||
| PubSubCreationConfig, | ||
| PubSubQueueLocatorType, | ||
| EventType | ||
| >, | ||
| 'messageSchemas' | 'creationConfig' | 'locatorConfig' | ||
| > & { | ||
| creationConfig?: Omit<PubSubCreationConfig, 'topic'> | ||
| } | ||
| } |
There was a problem hiding this comment.
🧩 Analysis chain
MetadataFiller generic parameters look mis-ordered in options
In core, MetadataFiller is defined as:
MetadataFiller<T, D, M>where:T= message type,D= event-definition type,M= metadata type.
Here the options declare:
metadataFiller: MetadataFiller<EventType, MetadataType>which effectively treats MetadataType as the event definition (D) and leaves the metadata type at its default (PublisherMessageMetadataType). That doesn’t line up with the generics on PubSubPublisherManager, where MetadataType is meant to be the actual metadata shape.
I’d expect something closer to:
metadataFiller: MetadataFiller<
EventType,
PubSubAwareEventDefinition, // or SupportedEventDefinitions[number]
MetadataType
>and corresponding adjustments in the manager / options generics so the type system reflects how produceMetadata is actually called.
🏁 Script executed:
#!/bin/bash
# Search for MetadataFiller definition in the codebase
rg -nP "type MetadataFiller|interface MetadataFiller" --type=tsLength of output: 161
🏁 Script executed:
#!/bin/bash
# Examine the PubSubPublisherManager class definition to understand generics
rg -nP "class PubSubPublisherManager" --type=ts -A 30Length of output: 3042
🏁 Script executed:
#!/bin/bash
# Find where metadataFiller is used/assigned in PubSubPublisherManager
rg -nP "metadataFiller" --type=ts -B 2 -A 2 -g "**/*PubSubPublisherManager*"Length of output: 1553
🏁 Script executed:
#!/bin/bash
# Find AbstractPublisherManager to understand the base class generics
rg -nP "class AbstractPublisherManager|export class AbstractPublisherManager" --type=ts -A 20Length of output: 1969
🏁 Script executed:
#!/bin/bash
# Check if metadataFiller is actually called/used and what signature it expects
rg -nP "produceMetadata|metadataFiller\(" --type=ts -B 3 -A 3 -g "**/*PubSubPublisherManager*"Length of output: 0
🏁 Script executed:
#!/bin/bash
# Read the MetadataFiller type definition
cat packages/core/lib/messages/MetadataFiller.tsLength of output: 2082
🏁 Script executed:
#!/bin/bash
# Check AbstractPublisherManager for metadataFiller usage
rg -nP "metadataFiller|produceMetadata" --type=ts -B 3 -A 3 -g "**/*AbstractPublisherManager*"Length of output: 4391
🏁 Script executed:
#!/bin/bash
# Check if there are TypeScript compilation errors
cd packages/gcp-pubsub && cat tsconfig.json 2>/dev/null | head -20Length of output: 173
🏁 Script executed:
#!/bin/bash
# Verify the exact lines 30-48 in PubSubPublisherManager.ts still have the issue
sed -n '30,48p' packages/gcp-pubsub/lib/pubsub/PubSubPublisherManager.tsLength of output: 601
🏁 Script executed:
#!/bin/bash
# Check the full context of PubSubPublisherManagerOptions generics
sed -n '30,52p' packages/gcp-pubsub/lib/pubsub/PubSubPublisherManager.tsLength of output: 713
MetadataFiller generics are mis-ordered; second parameter should be an event-definition type, not MetadataType
The MetadataFiller type is defined as MetadataFiller<T, D, M> where:
T= message typeD= event-definition typeM= metadata type
Line 37 declares:
metadataFiller: MetadataFiller<EventType, MetadataType>This passes MetadataType as the second parameter (event-definition position), leaving metadata type at its default. The correct signature should be:
metadataFiller: MetadataFiller<EventType, SupportedEventDefinitions[number], MetadataType>Note: AbstractPublisherManager in core has the same issue and suppresses it with @ts-expect-error on line 238. This type mismatch should be resolved to maintain proper type safety across the hierarchy.
There was a problem hiding this comment.
I think it is fixed already, could you check it
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
# Check the current state of the metadataFiller declaration in PubSubPublisherManager.ts
sed -n '30,50p' packages/gcp-pubsub/lib/pubsub/PubSubPublisherManager.ts
Length of output: 688
@CarlosGamero, I've checked the current code and the issue is still present. Line 37 shows:
metadataFiller: MetadataFiller<EventType, MetadataType>This is still missing the event-definition type parameter. MetadataFiller expects three generic parameters:
T= message typeD= event-definition typeM= metadata type
The correct signature should be:
metadataFiller: MetadataFiller<EventType, PubSubAwareEventDefinition, MetadataType>Or if you want to be more specific:
metadataFiller: MetadataFiller<EventType, SupportedEventDefinitions[number], MetadataType>This ensures MetadataType is properly positioned as the third generic parameter (metadata type) rather than the second (event-definition type).
| "peerDependencies": { | ||
| "@google-cloud/pubsub": "^5.2.0", | ||
| "@message-queue-toolkit/core": ">=22.0.0", | ||
| "zod": ">=3.25.76 <5.0.0" |
There was a problem hiding this comment.
Verify Zod peer dependency range matches actual usage.
The peer dependency allows zod >=3.25.76 <5.0.0, but the code imports from zod/v4 (see packages/gcp-pubsub/lib/errors/PubSubConsumerErrorResolver.ts:4), which is specific to Zod 4.x. This mismatch could cause runtime failures if users install Zod 3.x.
Consider tightening the peer dependency to require Zod 4:
- "zod": ">=3.25.76 <5.0.0"
+ "zod": "^4.0.0"Alternatively, if Zod 3 support is intended, update the import to use the standard zod path instead of zod/v4.
📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| "zod": ">=3.25.76 <5.0.0" | |
| "zod": "^4.0.0" |
🤖 Prompt for AI Agents
In packages/gcp-pubsub/package.json around line 36, the declared Zod peer
dependency ">=3.25.76 <5.0.0" conflicts with code that imports from "zod/v4"
(which requires Zod 4.x); to fix either tighten the peer dependency to require
Zod 4 (e.g., restrict to ^4.0.0 <5.0.0) or modify the code to import from the
generic "zod" entry so it remains compatible with Zod 3 and 4; choose one
approach and update package.json and/or the import paths accordingly, then run
tests and update package lock files if needed.
| ### With Pub/Sub | ||
|
|
||
| ```typescript | ||
| import { Storage } from '@google-cloud/storage' | ||
| import { GCSPayloadStore } from '@message-queue-toolkit/gcs-payload-store' | ||
| import { AbstractPubSubPublisher, PUBSUB_MESSAGE_MAX_SIZE } from '@message-queue-toolkit/pubsub' | ||
|
|
||
| const storage = new Storage({ projectId: 'my-project' }) | ||
|
|
||
| class PubSubLargeMessagePublisher extends AbstractPubSubPublisher<MyMessage> { | ||
| constructor(dependencies) { | ||
| super(dependencies, { | ||
| creationConfig: { | ||
| topic: { name: 'large-events' }, | ||
| }, | ||
| messageSchemas: [MY_MESSAGE_SCHEMA], | ||
| messageTypeField: 'type', | ||
| payloadStoreConfig: { | ||
| store: new GCSPayloadStore( | ||
| { gcsStorage: storage }, | ||
| { bucketName: 'pubsub-large-payloads', keyPrefix: 'events' } | ||
| ), | ||
| messageSizeThreshold: PUBSUB_MESSAGE_MAX_SIZE, | ||
| }, | ||
| }) | ||
| } | ||
| } | ||
| ``` |
There was a problem hiding this comment.
🧩 Analysis chain
Verify package name and export references.
The example references @message-queue-toolkit/pubsub, but based on the PR context, the package appears to be @message-queue-toolkit/gcp-pubsub. Additionally, verify that AbstractPubSubPublisher and PUBSUB_MESSAGE_MAX_SIZE are actually exported from the package.
Run the following script to verify the correct package name and exports:
🏁 Script executed:
#!/bin/bash
# Verify package name and exports
# Check package.json for actual package name
fd -t f "^package\.json$" packages/gcp-pubsub --exec jq -r '.name'
# Search for PUBSUB_MESSAGE_MAX_SIZE export
rg -n "export.*PUBSUB_MESSAGE_MAX_SIZE" packages/gcp-pubsub
# Search for AbstractPubSubPublisher export
rg -n "export.*AbstractPubSubPublisher" packages/gcp-pubsub
# Check the index.ts exports
cat packages/gcp-pubsub/lib/index.tsLength of output: 1209
Update package name in README example to match the actual package.
The example imports @message-queue-toolkit/pubsub, but the actual package is @message-queue-toolkit/gcp-pubsub. Both AbstractPubSubPublisher and PUBSUB_MESSAGE_MAX_SIZE are correctly exported from the package, but users will fail to import them using the incorrect package name shown in the documentation.
Change line 259 from:
import { AbstractPubSubPublisher, PUBSUB_MESSAGE_MAX_SIZE } from '@message-queue-toolkit/pubsub'To:
import { AbstractPubSubPublisher, PUBSUB_MESSAGE_MAX_SIZE } from '@message-queue-toolkit/gcp-pubsub'🤖 Prompt for AI Agents
In packages/gcs-payload-store/README.md around lines 254 to 281, the example
imports AbstractPubSubPublisher and PUBSUB_MESSAGE_MAX_SIZE from the wrong
package name; update the import statement on line 259 to import these symbols
from @message-queue-toolkit/gcp-pubsub instead of @message-queue-toolkit/pubsub
so the example matches the actual package exports and users can import
successfully.
There was a problem hiding this comment.
Actionable comments posted: 2
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
packages/gcp-pubsub/README.md(1 hunks)
🧰 Additional context used
🪛 LanguageTool
packages/gcp-pubsub/README.md
[style] ~81-~81: This adverb was used twice in the sentence. Consider removing one of them or replacing them with a synonym.
Context: ...to subscriptions - ❌ You CANNOT consume directly from topics - ✅ One topic can have mult...
(ADVERB_REPETITION_PREMIUM)
🪛 markdownlint-cli2 (0.18.1)
packages/gcp-pubsub/README.md
73-73: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
794-794: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (16)
- GitHub Check: general (22.x, @message-queue-toolkit/sns) / build
- GitHub Check: general (20.x, @message-queue-toolkit/redis-message-deduplication-store) / build
- GitHub Check: general (20.x, @message-queue-toolkit/sqs) / build
- GitHub Check: general (22.x, @message-queue-toolkit/redis-message-deduplication-store) / build
- GitHub Check: general (22.x, @message-queue-toolkit/core) / build
- GitHub Check: general (22.x, @message-queue-toolkit/amqp) / build
- GitHub Check: general (20.x, @message-queue-toolkit/metrics) / build
- GitHub Check: general (20.x, @message-queue-toolkit/sns) / build
- GitHub Check: general (20.x, @message-queue-toolkit/s3-payload-store) / build
- GitHub Check: general (22.x, @message-queue-toolkit/sqs) / build
- GitHub Check: general (24.x, @message-queue-toolkit/sns) / build
- GitHub Check: general (22.x, @message-queue-toolkit/s3-payload-store) / build
- GitHub Check: general (20.x, @message-queue-toolkit/amqp) / build
- GitHub Check: general (24.x, @message-queue-toolkit/sqs) / build
- GitHub Check: kafka (24.x) / build
- GitHub Check: kafka (22.x) / build
🔇 Additional comments (1)
packages/gcp-pubsub/README.md (1)
1-1600: Excellent documentation coverage.This README provides comprehensive guidance on GCP Pub/Sub integration. The structure is well-organized with clear sections on configuration, advanced features, testing, and troubleshooting. Code examples are practical and demonstrate both basic and advanced patterns. The API reference and best practices sections will be valuable for users adopting the library.
| ``` | ||
| Publisher → Topic → Subscription → Consumer | ||
| ``` |
There was a problem hiding this comment.
Add language specification to fenced code block.
The diagram code block at line 73 is missing a language identifier. For Pub/Sub diagrams or plaintext, use ```text or ```plaintext.
-```
+```text
Publisher → Topic → Subscription → Consumer
-```
+```🧰 Tools
🪛 markdownlint-cli2 (0.18.1)
73-73: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
🤖 Prompt for AI Agents
In packages/gcp-pubsub/README.md around lines 73 to 75, the fenced code block
lacks a language identifier; update the opening fence to include a plaintext
language tag (e.g., ```text or ```plaintext) so the diagram is rendered as plain
text, leaving the content "Publisher → Topic → Subscription → Consumer" and the
closing fence unchanged.
| ``` | ||
| Attempt 1: Message nacked, redelivered by Pub/Sub | ||
| Attempt 2: Message nacked, redelivered by Pub/Sub | ||
| Attempt 3: Message nacked, redelivered by Pub/Sub | ||
| ... | ||
| After maxDeliveryAttempts: Sent to DLQ | ||
| ``` |
There was a problem hiding this comment.
Add language specification to fenced code block.
The exponential backoff example at line 794 is missing a language identifier. Use ```text for plaintext output examples.
- ```
+ ```text
Attempt 1: Message nacked, redelivered by Pub/Sub
Attempt 2: Message nacked, redelivered by Pub/Sub
Attempt 3: Message nacked, redelivered by Pub/Sub
...
After maxDeliveryAttempts: Sent to DLQ
- ```
+ ```🧰 Tools
🪛 markdownlint-cli2 (0.18.1)
794-794: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
🤖 Prompt for AI Agents
In packages/gcp-pubsub/README.md around lines 794 to 800, the fenced code block
showing the exponential backoff example lacks a language identifier; update the
opening fence from ``` to ```text so the block becomes a plaintext fenced code
block, preserving the contents and closing fence unchanged.
| if (isStandardizedError(error)) { | ||
| return new InternalError({ | ||
| message: error.message, | ||
| errorCode: error.code, | ||
| }) | ||
| } | ||
| return new InternalError({ | ||
| message: 'Error processing message', | ||
| errorCode: 'INTERNAL_ERROR', | ||
| }) |
There was a problem hiding this comment.
I think that adding cause here might be useful
| @@ -0,0 +1,14 @@ | |||
| export * from './errors/PubSubConsumerErrorResolver.ts' | |||
There was a problem hiding this comment.
Mmm, do we need to export everything? thinks like AbstractPubSubService, pubSubMessageDeserializer or pubSubMessageReader might not be needed, correct?
| this.errorResolver, | ||
| ) | ||
|
|
||
| if ('error' in parseResult) { |
There was a problem hiding this comment.
I see some 'error' in object conditions, isn't it better to use object.error? I guess they are Either
| "peerDependencies": { | ||
| "@google-cloud/pubsub": "^5.2.0", | ||
| "@message-queue-toolkit/core": ">=22.0.0", | ||
| "zod": ">=3.25.76 <5.0.0" |
There was a problem hiding this comment.
As it is a new lib, I think it is ok to use Zod 4 directly
Summary by CodeRabbit
New Features
Documentation
Tests