ai.pipeline.connector.intake.pipedoc.PipeDocAcceptanceService

@ApplicationScoped public class PipeDocAcceptanceService extends Object

Per-doc acceptance for the gRPC upload paths (bidi and unary). Two responsibilities:

Route inline documents according to pipestream.intake.inline-handoff-mode. The default redis mode enqueues to pipestream:intake:ingress; compatibility modes can call the engine directly.
If resolved.shouldPersist() is true: schedule a fire-and-forget durable copy via IntakeReplayPersister. Runs on its own VT executor; failure logs but does not block, fail, or delay the engine handoff. Used for replay; not currently on any read path.

Constructor Details
- PipeDocAcceptanceService
  
  public PipeDocAcceptanceService()
Method Details
- enqueueAndMaybePersist
  
  public ai.pipestream.connector.intake.v1.UploadPipeDocResponse enqueueAndMaybePersist(ai.pipestream.data.v1.PipeDoc pipeDoc, ConfigResolutionService.ResolvedConfig resolved, String crawlId, long startTime)
  
  Accepts a fully-hydrated inline PipeDoc into the configured handoff path.
  Redis mode writes an IntakeHandoffRequest to the intake ingress stream for the sidecar to drain. Compatibility modes call engine directly. In all modes, replay persistence is optional, asynchronous, and not part of the admission decision.
  
  Parameters:
  
  pipeDoc - document after doc-id derivation and ownership stamping
  
  resolved - datasource/account configuration for this request
  
  crawlId - optional crawl session identifier
  
  startTime - start time from the transport handler, used for timing logs
  
  Returns:
  
  upload response for the connector-facing API
- acceptStreamingItem
  
  public ai.pipestream.connector.intake.v1.UploadPipeDocStreamResponse acceptStreamingItem(ai.pipestream.connector.intake.v1.PipeDocItem item, ai.pipestream.connector.intake.v1.StreamContext ctx, ConfigResolutionService.ResolvedConfig resolved)

Class PipeDocAcceptanceService

Constructor Summary