import { z } from "zod";
import { SENSOR_TYPES, TIERS } from "./taxonomy";

/**
 * Source-registry seed schema (shared by the engine's seed loader, the validator CLI and the
 * API's bulk-import endpoint). Dependency-free apart from zod; no filesystem access here.
 */
export const sensorSchema = z.object({
  id: z.string().optional(),
  name: z.string(),
  url: z.string().url(),
  type: z.enum(SENSOR_TYPES),
  connector: z.string().default("http"),
  tier: z.enum(TIERS).optional(),
  interval: z.number().int().positive().optional(),
  weight: z.number().positive().optional(),
  config: z.record(z.string(), z.unknown()).default({}),
});
export type SensorSeed = z.infer<typeof sensorSchema>;

export const productSchema = z.object({ id: z.string().optional(), name: z.string(), type: z.string().default("product"), aliases: z.array(z.string()).default([]) });

export const sourceSchema = z.object({
  id: z.string().regex(/^[a-z0-9][a-z0-9-]*$/, "id must be kebab-case"),
  name: z.string(),
  domain: z.string(),
  homepage: z.string().url().optional(),
  description: z.string().optional(),
  categories: z.array(z.string()).default([]),
  tier: z.enum(TIERS).default("B"),
  weight: z.number().positive().default(1),
  entity_type: z.string().default("organization"),
  aliases: z.array(z.string()).default([]),
  products: z.array(productSchema).default([]),
  discover: z.object({ rss: z.boolean().optional(), sitemap: z.boolean().optional(), status: z.boolean().optional(), pages: z.boolean().optional() }).default({}),
  fallback: z.object({ firecrawl: z.boolean().optional(), scrapfly: z.boolean().optional() }).default({}),
  sensors: z.array(sensorSchema).default([]),
  notes: z.string().optional(),
  enabled: z.boolean().default(true),
  /** false = heuristics only (high-volume feeds such as news wires) */
  llm: z.boolean().default(true),
  /** ISO 3166-1 alpha-2 (upper-case) when the organization is national/regional; "EU" for EU bodies; "INT" for global bodies */
  country: z.string().regex(/^[A-Z]{2,3}$/).optional(),
  /** primary language of the published content (ISO 639-1); defaults to "en" downstream */
  language: z.string().regex(/^[a-z]{2}$/).optional(),
  /** false for media/aggregators reporting about others (news wires, tech press); true for the organization's own channels */
  first_party: z.boolean().optional(),
  /** fragment-only: merge into an already declared source instead of redefining it */
  extend: z.boolean().default(false),
});
export type SourceSeed = z.infer<typeof sourceSchema>;

/** Partial schema for `extend: true` entries — only `id` is required. */
export const extendSchema = z.object({
  id: z.string(),
  extend: z.literal(true),
  aliases: z.array(z.string()).default([]),
  products: z.array(productSchema).default([]),
  sensors: z.array(sensorSchema).default([]),
  categories: z.array(z.string()).default([]),
  fallback: z.object({ firecrawl: z.boolean().optional(), scrapfly: z.boolean().optional() }).optional(),
  notes: z.string().optional(),
  country: z.string().regex(/^[A-Z]{2,3}$/).optional(),
  language: z.string().regex(/^[a-z]{2}$/).optional(),
  first_party: z.boolean().optional(),
});

/** Bulk import document (JSON / YAML already parsed): `{ sources: [...] }`. */
export const importDocumentSchema = z.object({ sources: z.array(z.unknown()).min(1).max(2000) });