spb/datacenterindex

1/**2 * Boundary between the crawl runtime (context, documents, scheduling) and the ingest layer3 * (reconciliation, persistence, provenance, events). Both sides implement against this file.4 */5import type { NormalizedEntity, DetectedChange } from "@dci/core";67export interface IngestRun {8  connectorId: string;9  sourceId: string;10  runId: string;11  sourceKind: string;12  sourcePriority: number;13  dryRun: boolean;14}1516export interface IngestDocRef {17  documentId: string;18  url: string;19  pageType: string;20  fetchedAt: string;21}2223export interface IngestStats {24  received: number;25  created: number;26  updated: number;27  unchanged: number;28  merged: number;29  pendingMatches: number;30  rejected: number;31  events: number;32  provenanceRows: number;33  byType: Record<string, number>;34  changes: DetectedChange[];35  /** entity ids touched: [{type,id}] for documents.entity_refs */36  refs: Array<{ type: string; id: string }>;37  /** data-quality counters (optional, additive): operators inferred from a canonical name prefix */38  inferredOperators?: number;39  /** coordinates dropped because the point lies in another country than the one the source states */40  geoCountryMismatch?: number;41  /** field values kept in provenance but not written to a column (e.g. pre-1980 opening dates from OSM) */42  implausibleDropped?: number;43  /** projects folded into an already-known announcement (same operator, city, MW ±10 %, within 30 days) */44  projectDedup?: number;45  /** claims rows written */46  claims?: number;47  /** quality flags written */48  qualityFlags?: number;49  /** figures kept as claims but not assigned to a column (portfolio / company / unknown scope, market statistics) */50  unscopedClaims?: number;51  /** project candidates vetoed by the announcement classifier (appointments, financing, PPAs, market research…) */52  projectsVetoed?: number;53  /** building records linked to their campus record instead of being flagged as duplicates */54  campusLinks?: number;55  /** hidden projects restored because a re-extraction qualified them again */56  projectsRequalified?: number;57}5859export type IngestFn = (run: IngestRun, entities: NormalizedEntity[], doc: IngestDocRef | null) => Promise<IngestStats>;60