disclosure-bureau/web/lib/fm-types.ts

/**
 * Frontmatter type contracts derived from CLAUDE-schema-full.md.
 *
 * Used by the UI primitives in `components/fm/*` to render each field with
 * the correct semantic UI (chip / link / badge / bbox / etc.).
 *
 * NOTE: every field is optional because the schema is permissive — earlier
 * extraction passes left gaps that later phases fill. The UI must be robust
 * to missing data.
 */

export type ConfidenceBand = "high" | "medium" | "low" | "speculation";
export type EnrichmentStatus = "deep" | "shallow" | "none";
export type ClassificationLevel = "UNCLASSIFIED" | "CUI" | "CONFIDENTIAL" | "SECRET" | "TOP SECRET";
export type ContentClass =
  | "text-only" | "contains-photos" | "contains-sketches" | "contains-diagrams"
  | "contains-maps" | "contains-tables" | "contains-signatures" | "contains-stamps"
  | "redaction-heavy" | "mixed" | "blank";

export interface BBox { x: number; y: number; w: number; h: number }
export interface Coords { lat?: number | null; lon?: number | null; raw_text?: string; confidence_band?: ConfidenceBand }

export interface ClassificationMarking {
  level?: ClassificationLevel;
  caveats?: string[];
  location?: "header" | "footer" | "banner" | "stamp";
  bbox?: BBox;
}

export interface Redaction {
  code?: "(b)(1) 1.4(a)" | "(b)(3)" | "(b)(6)" | "other" | string;
  description?: string;
  bbox?: BBox;
  text_inferred?: string | null;
}

export interface SignatureObserved {
  signer_inferred?: string | null;
  confidence_band?: ConfidenceBand;
  bbox?: BBox;
  notes?: string;
}

export interface TableDetected {
  local_table_index?: number;
  bbox?: BBox;
  spans_multi_page?: boolean;
  continues_from_prev_page?: boolean;
  likely_continues_next_page?: boolean;
  row_count_estimate?: number;
  col_count_estimate?: number;
  headers_summary?: string;
  table_id?: string;          // populated by Phase 4.8 consolidate-tables
}

export interface ImageDetected {
  local_image_index?: number;
  image_type?: "photo" | "sketch" | "map" | "chart" | "stamp" | "signature" | "redaction" | "logo" | "seal" | "diagram" | "other";
  bbox?: BBox;
  caption_ocr?: string;
}

export interface ExternalSource {
  url?: string;
  title?: string;
  publisher?: string;
  accessed_at?: string;
  key_facts?: string[];
  reliability_band?: ConfidenceBand;
}

export interface EntityRef {
  name?: string;
  role_in_page?: "subject" | "witness" | "author" | "signer" | "mentioned";
  aliases?: string[];
  type?: string;
  class?: string;
  shape?: string;
  color?: string;
  size_estimate?: string;
  label?: string;
  date?: string;
}

export interface EntitiesExtracted {
  people?: EntityRef[];
  organizations?: EntityRef[];
  locations?: EntityRef[];
  events?: EntityRef[];
  uap_objects?: EntityRef[];
  vehicles?: EntityRef[];
  operations?: EntityRef[];
  concepts?: EntityRef[];
}

export interface UapObservation {
  date_time_utc?: string;
  duration_seconds?: number | null;
  shape?: string;
  color?: string;
  size_estimate?: string;
  altitude_ft?: number | null;
  speed_kts?: number | null;
  bearing_deg?: number | null;
  distance_nm?: number | null;
  coordinates?: Coords;
}

export interface MentionedIn {
  page?: string;          // wiki-link string
  page_ref?: string;      // alternative naming
  mention_count?: number;
  role_in_page?: string;
}

/** Universal frontmatter shape — covers document/page/entity/etc.
 * Use this with `(fm as AnyFrontmatter).field` for safe access. */
export interface AnyFrontmatter {
  schema_version?: string;
  type?: string;
  wiki_version?: string;
  last_ingest?: string;
  last_lint?: string;
  last_enriched?: string;

  // identity
  doc_id?: string;
  page_id?: string;
  page_number?: number;
  total_pages?: number;
  person_id?: string;
  organization_id?: string;
  location_id?: string;
  event_id?: string;
  uap_object_id?: string;
  vehicle_id?: string;
  operation_id?: string;
  concept_id?: string;
  table_id?: string;
  image_id?: string;
  entity_class?: string;
  canonical_name?: string;
  canonical_title?: string;
  aliases?: string[];

  // classification
  highest_classification?: ClassificationLevel;
  classification_markings?: ClassificationMarking[];
  language_detected?: string;
  languages_detected?: string[];
  content_classification?: ContentClass[];
  page_type?: string;
  document_class?: string;
  collection?: string;
  redactions?: Redaction[];
  signatures_observed?: SignatureObserved[];

  // page-level extraction
  tables_detected?: TableDetected[];
  images_detected?: ImageDetected[];
  entities_extracted?: EntitiesExtracted;
  uap_observation_fields?: UapObservation | null;
  vision_description?: string;
  vision_description_pt_br?: string;
  ocr_quality_score?: number;
  vision_quality_score?: number;
  flags?: string[];

  // document-level
  page_count?: number;
  total_redactions?: number;
  total_signatures?: number;
  total_tables?: number;
  total_images?: number;
  key_entities?: EntitiesExtracted;

  // entity-level relational
  mentioned_in?: MentionedIn[];
  total_mentions?: number;
  documents_count?: number;
  enrichment_status?: EnrichmentStatus;
  external_sources?: ExternalSource[];
  disambiguation_note?: string;

  // location-specific
  location_type?: string;
  country?: string | string[];
  region?: string;
  parent_location?: string;
  coordinates?: Coords;
  events_here?: string[];

  // event-specific
  event_class?: string;
  date_start?: string;
  date_end?: string;
  date_confidence?: ConfidenceBand;
  primary_location?: string | null;
  observers?: string[];
  uap_objects?: string[];
  documented_in?: string[];
  narrative_summary?: string;
  narrative_summary_pt_br?: string;

  // uap-object-specific
  observed_in_event?: string;
  secondary_events?: string[];
  shape?: string;
  color?: string;
  size_estimate_m?: { min?: number | null; max?: number | null; confidence_band?: ConfidenceBand };
  features?: string[];
  altitude_ft?: { min?: number | null; max?: number | null; confidence_band?: ConfidenceBand };
  speed_kts?: { min?: number | null; max?: number | null; confidence_band?: ConfidenceBand };
  maneuver_descriptors?: string[];
  sensor_observations?: string[];
  visual_records?: string[];
  confidence_band_overall?: ConfidenceBand;

  // person-specific
  roles?: string[];
  dates?: { born?: string | null; died?: string | null };
  primary_role?: string;
  primary_organization?: string;

  // organization-specific
  organization_type?: string;
  founded?: string;

  // vehicle-specific
  vehicle_class?: string;
  operator?: string;
  model?: string;

  // operation-specific
  operation_type?: string;
  status?: string;

  // concept-specific
  concept_class?: string;
  domain?: string;
  definition_short?: string;
  definition_short_pt_br?: string;

  // table/image specific
  source_doc?: string;
  multi_page?: boolean;
  spans_pages?: Array<{ page?: string; bbox?: BBox; role?: "start" | "middle" | "end" }>;
  headers_summary?: string;
  total_rows_estimate?: number;
  total_columns_estimate?: number;
  extraction_quality?: number | null;
  csv_path?: string;
  headers?: string[];
  row_count_extracted?: number;
  column_count_extracted?: number;
  extraction_notes?: string;
  extraction_model?: string;
  extracted_at?: string;

  // war.gov enrichment (from 02b script)
  war_gov?: Record<string, unknown>;

  // catch-all
  [k: string]: unknown;
}