/**
 * earth-now.co
 * Author:  Simon-Pierre Boucher
 * Contact: contact@spboucher.ai
 * File:    apps/ingest/src/runners.ts
 * Purpose: Per-source run functions (fetch → archive raw → parse → summary) shared by the CLI and the queue worker
 */

import { archiveRaw } from "./archive.js";
import { type FetchLike, defaultFetch } from "./fetch-like.js";
import { NOAA_CO2_URL, parseNoaaMonthlyCo2 } from "./sources/noaa-co2.js";
import { parseUsgsCount, usgsCountUrl } from "./sources/usgs.js";

/** USGS live counter parameters: quakes M ≥ 4.5 over the trailing 24 h. */
export const USGS_WINDOW_HOURS = 24;
// Must match the registry definition of earthquakes_24h (catalog: M ≥ 2.5).
export const USGS_MIN_MAGNITUDE = 2.5;

export interface ObservationPoint {
  time: string;
  value: number;
}

export interface RunSummary {
  sourceId: string;
  observationCount: number;
  first: ObservationPoint | null;
  last: ObservationPoint | null;
  rawPath: string;
  sha256: string;
}

async function fetchText(url: string, fetchImpl: FetchLike): Promise<string> {
  const res = await fetchImpl(url);
  if (!res.ok) throw new Error(`fetch failed: HTTP ${res.status} for ${url}`);
  return res.text();
}

/** USGS FDSN: one observation — the M≥4.5 quake count over the trailing 24 h window. */
export async function runUsgsFdsn(fetchImpl: FetchLike = defaultFetch): Promise<RunSummary> {
  const url = usgsCountUrl(USGS_WINDOW_HOURS, USGS_MIN_MAGNITUDE);
  const text = await fetchText(url, fetchImpl);
  const archived = await archiveRaw("usgs_fdsn", text);
  const count = parseUsgsCount(JSON.parse(text));
  const point: ObservationPoint = { time: new Date().toISOString(), value: count };
  return {
    sourceId: "usgs_fdsn",
    observationCount: 1,
    first: point,
    last: point,
    rawPath: archived.rawPath,
    sha256: archived.sha256,
  };
}

/** NOAA GML Mauna Loa: full monthly CO₂ series (Keeling curve). */
export async function runNoaaGmlMlo(fetchImpl: FetchLike = defaultFetch): Promise<RunSummary> {
  const text = await fetchText(NOAA_CO2_URL, fetchImpl);
  const archived = await archiveRaw("noaa_gml_mlo", text);
  const observations = parseNoaaMonthlyCo2(text);
  return {
    sourceId: "noaa_gml_mlo",
    observationCount: observations.length,
    first: observations[0] ?? null,
    last: observations[observations.length - 1] ?? null,
    rawPath: archived.rawPath,
    sha256: archived.sha256,
  };
}

export interface SourceSpec {
  id: string;
  /** Scheduling cadence in ms (usgs 5 min, noaa daily). */
  cadenceMs: number;
  run: (fetchImpl?: FetchLike) => Promise<RunSummary>;
}

/** Every source the ingest service knows how to run, with its scheduling cadence. */
export const SOURCES: readonly SourceSpec[] = [
  { id: "usgs_fdsn", cadenceMs: 5 * 60_000, run: runUsgsFdsn },
  { id: "noaa_gml_mlo", cadenceMs: 24 * 3_600_000, run: runNoaaGmlMlo },
];

export const KNOWN_SOURCE_IDS: readonly string[] = SOURCES.map((s) => s.id);

/** Run one source by id. Throws on unknown id or on any fetch/parse failure. */
export async function runSource(
  sourceId: string,
  fetchImpl: FetchLike = defaultFetch,
): Promise<RunSummary> {
  const spec = SOURCES.find((s) => s.id === sourceId);
  if (!spec) throw new Error(`Unknown source '${sourceId}'. Known: ${KNOWN_SOURCE_IDS.join(", ")}`);
  return spec.run(fetchImpl);
}