import { z } from 'zod';

/** CLAUDE.md §2 — every imported fact carries provenance; every derived value is reproducible. */
export const EvidenceType = z.enum([
  'registry',
  'clinical_trial',
  'meta_analysis',
  'systematic_review',
  'cohort',
  'case_control',
  'case_series',
  'case_report',
  'preclinical',
  'regulatory',
  'guideline',
  'expert_curation',
  'database',
  'computed',
]);
export type EvidenceType = z.infer<typeof EvidenceType>;

export const AccessLevel = z.enum(['open', 'registration_required', 'controlled', 'licensed']);
export type AccessLevel = z.infer<typeof AccessLevel>;

export const ProvenanceInput = z.object({
  sourceId: z.string(),
  sourceRecordId: z.string().optional(),
  sourceUrl: z.string().url().optional(),
  dataset: z.string().optional(),
  datasetVersion: z.string().optional(),
  publicationId: z.string().optional(),
  pmid: z.string().optional(),
  doi: z.string().optional(),
  retrievedAt: z.string().datetime(),
  publishedAt: z.string().optional(),
  updatedAt: z.string().optional(),
  geography: z.string().optional(),
  population: z.string().optional(),
  cohortSize: z.number().int().nonnegative().optional(),
  methodology: z.string().optional(),
  evidenceType: EvidenceType,
  accessLevel: AccessLevel,
  confidence: z.number().min(0).max(1).optional(),
  license: z.string().optional(),
  ingestRunId: z.string().optional(),
});
export type ProvenanceInput = z.infer<typeof ProvenanceInput>;

/** Data layer a value belongs to (CLAUDE.md §2 — layers stay separable). */
export const DataLayer = z.enum(['raw', 'normalized', 'canonical', 'derived', 'ranked', 'ai_synthesis']);
export type DataLayer = z.infer<typeof DataLayer>;

/** Display category for scientific safety (CLAUDE.md §3). */
export const ClaimCategory = z.enum([
  'observed_data',
  'published_evidence',
  'curated_evidence',
  'regulatory_status',
  'clinical_guideline',
  'computed_metric',
  'ai_generated_synthesis',
]);
export type ClaimCategory = z.infer<typeof ClaimCategory>;

/** Entity mapping confidence (CLAUDE.md §221). */
export const MatchType = z.enum([
  'EXACT_IDENTIFIER',
  'CURATED_EXACT',
  'ONTOLOGY_EXACT',
  'CURATED_BROADER',
  'CURATED_NARROWER',
  'ALIAS',
  'PROBABILISTIC',
  'UNRESOLVED',
]);
export type MatchType = z.infer<typeof MatchType>;

/** Derived statistic envelope (CLAUDE.md §2). */
export interface DerivedValue {
  metric: string;
  value: number;
  computed: true;
  formulaVersion: string;
  inputs: string[];
}