All Classes and Interfaces
Class
Description
Captures the
x-api-key gRPC metadata header into the call Context.Utilities for API key generation and hashing.
Handles the gRPC blob upload path: persist the doc to repository
(durable, blobs are large so we don't route them through Redis),
then hand off a reference-only request to the engine.
Provides a default ChunkSizeCalculator bean when the pipestream-server
dependency does not expose it as a CDI bean.
Service for resolving Tier 1 configuration for a datasource.
Resolved configuration containing Tier 1 config and base IngestionConfig.
Thin gRPC adapter for connector intake endpoints.
Utility class for parsing and serializing connector metadata JSON.
Service for validating connector API keys and fetching connector configuration.
Document tracking entity for orphan detection.
Composite primary key for CrawlDocument.
Repository for crawl document tracking operations.
Crawl session entity tracking document ingestion sessions.
Repository for crawl session operations.
RPC service for managing crawl session lifecycle and heartbeats.
Publishes the intake layer's broadcast
DocumentDeleted signal to
Kafka.Builds engine intake handoff requests and sends them over the configured
engine transport.
Handles the unary
UploadBlob RPC.A failure attributable to the HTTP request itself — a malformed shape, a
missing header, an undeterminable document identity, or an oversized body.
Stable, client-facing error taxonomy for the intake HTTP surface.
Uniform, client-safe error body.
UUID key extractors for the intake event types, auto-discovered by the
quarkus-apicurio-registry-protobuf extension and used to key each
record for Kafka partitioning + dedup.Publishes intake lifecycle events to Kafka via the platform's
quarkus-apicurio-registry-protobuf extension: inject a typed
ProtobufEmitter per channel and call send(event) — the
extension handles protobuf serialization, the deterministic UUID key (via
IntakeEventKeyExtractors), Apicurio schema registration, and topic
wiring.Builds the
intake_events protos from upload context and maps the
service's IntakeErrorCode taxonomy onto the proto's
RejectionReason.Base type for every failure the intake HTTP surface deliberately surfaces.
Renders every
IntakeException as a uniform, client-safe JSON envelope
and guarantees the real cause is logged server-side, never serialized.Normalizes any failure into a client-safe
IntakeException.intake-service → engine streaming intake-handoff lane.
Redis Stream payload shared by connector-intake and the engine sidecar.
Single place where an intake outcome is published.
Projects the intake lifecycle event streams — accepted, rejected, error —
into the upload-receipt read-model that backs the status lookup.
Fire-and-forget durable copy to the repository service for any document
whose datasource is configured with
persist_pipedoc=true.Consumes IntakeRepoEvent messages and initiates intake handoff.
Pre-acceptance security screening for uploads.
Outcome of a security scan: allowed, or a client-safe rejection code + message.
A failure that originates beyond the request shape — resolving datasource
config, authenticating the API key, or proxying the body to
repository-service.
HTTP endpoint for raw binary uploads from connectors (
POST /uploads/raw).Client for repository-service integration.
Synchronous HTTP client for proxying raw uploads to repository-service.
Declarative REST client for proxying raw uploads to repository-service.
What the intake security scan examines for one upload.
Service for managing crawl session lifecycle.
Serializes one bidi
UploadPipeDocStream call onto a small mailbox.Handles the unary
UploadPipeDoc RPC.Accepted-upload response body.
Per-document intake status, keyed by
doc_id.Repository for
UploadReceipt, keyed by doc_id.Intake outcome for a document, as recorded on an
UploadReceipt and
reported by the status lookup.gRPC handler for
GetUploadStatus — the parity of
GET /uploads/status/{token}.Status lookup for an upload, addressed by the opaque tracking token returned
at upload time (
GET /uploads/status/{token}).Client-safe status projection — only curated fields, never internal detail.
Projects the intake lifecycle event stream —
DocumentAccepted,
DocumentRejected, DocumentError — into the UploadReceipt
read-model (upsert by doc_id) and resolves it for the status lookup.Client-safe status projection shared by the HTTP and gRPC status surfaces.
The tenant-scoped identity carried inside an
UploadTrackingToken.Mints and verifies the opaque, HMAC-SHA256-signed tracking token returned from
every upload.
Utility for canonicalizing URIs to create stable document IDs.