#!/usr/bin/env npx tsx
/**
 * generate-pir-input.ts
 *
 * Erzeugt input.txtpb für den iOS-NEURLFilter-PIR-Server.
 *
 * Format (gegen Apples pir_database.proto verifiziert):
 *   rows: { keyword: "bet365.com" value: "1" }
 *   rows: { keyword: "poker.de"   value: "1" }
 *   ...
 * Eine Zeile pro Domain. KEIN Top-Level-KeywordDatabase-Wrapper.
 *
 * Modi:
 *   Default (--source db):  liest BlocklistDomain aus PostgreSQL via Prisma.
 *                           DATABASE_URL muss gesetzt sein (via `infisical run`).
 *   --source hagezi:        Fallback — fetcht HaGeZi gambling-blocklist direkt
 *                           ohne DB-Zugriff.
 *
 * Optionen:
 *   --source <db|hagezi>    Default: db
 *   --output <pfad>         Default: ./input.txtpb (relativ zum CWD)
 *   --dry-run               Nur zählen, nicht schreiben
 *
 * Aufruf:
 *   infisical run -- npx tsx scripts/generate-pir-input.ts
 *   infisical run -- npx tsx scripts/generate-pir-input.ts --output /srv/pir-server/data/input.txtpb
 *   npx tsx scripts/generate-pir-input.ts --source hagezi --output /tmp/test.txtpb
 */

import { writeFileSync } from "node:fs";
import { resolve } from "node:path";
import { PrismaClient } from "../server/generated/prisma/client.js";
import { PrismaPg } from "@prisma/adapter-pg";

// ─── Normalisierung (identisch zu server/utils/domainHash.ts) ────────────────

function normalizeDomain(input: string): string {
  return input
    .trim()
    .toLowerCase()
    .replace(/^https?:\/\//, "")
    .replace(/\/.*$/, "")
    .replace(/^www\./, "");
}

// ─── CLI-Argument-Parsing ─────────────────────────────────────────────────────

const args = process.argv.slice(2);

function getFlag(name: string): string | undefined {
  const idx = args.indexOf(name);
  if (idx === -1) return undefined;
  return args[idx + 1];
}

const source = (getFlag("--source") ?? "db") as "db" | "hagezi";
const outputArg = getFlag("--output") ?? "input.txtpb";
const outputPath = resolve(process.cwd(), outputArg);
const dryRun = args.includes("--dry-run");

if (source !== "db" && source !== "hagezi") {
  console.error(`Unbekannte Source "${source}". Erlaubt: db, hagezi`);
  process.exit(1);
}

// ─── Domain-Fetch ─────────────────────────────────────────────────────────────

async function fetchFromDb(): Promise<string[]> {
  const dbUrl = process.env.DATABASE_URL;
  if (!dbUrl) {
    console.error(
      "DATABASE_URL nicht gesetzt. Script via `infisical run --` aufrufen.",
    );
    process.exit(1);
  }

  console.log("Verbinde zu PostgreSQL...");
  const adapter = new PrismaPg({ connectionString: dbUrl });
  const prisma = new PrismaClient({ adapter, log: ["error"] });

  try {
    const count = await prisma.blocklistDomain.count({
      where: { isActive: true },
    });
    console.log(`Lade ${count.toLocaleString("de-DE")} aktive Domains aus DB...`);

    const rows = await prisma.blocklistDomain.findMany({
      where: { isActive: true },
      select: { domain: true },
    });

    return rows.map((r) => r.domain);
  } finally {
    await prisma.$disconnect();
  }
}

const HAGEZI_URL =
  "https://raw.githubusercontent.com/hagezi/dns-blocklists/main/adblock/gambling.txt";

async function fetchFromHagezi(): Promise<string[]> {
  console.log(`Fetche HaGeZi gambling-blocklist von ${HAGEZI_URL}...`);
  const res = await fetch(HAGEZI_URL);
  if (!res.ok) {
    throw new Error(`HaGeZi-Fetch fehlgeschlagen: ${res.status} ${res.statusText}`);
  }
  const raw = await res.text();

  return raw
    .split("\n")
    .map((l) => l.trim())
    .filter((l) => l.startsWith("||") && l.endsWith("^"))
    .map((l) => l.slice(2, -1).toLowerCase())
    .filter((d) => d.length > 0 && !d.includes("/") && d.includes("."));
}

// ─── Normalisieren + Dedupe + Sort ────────────────────────────────────────────

function processdomains(raw: string[]): string[] {
  const seen = new Set<string>();
  for (const d of raw) {
    const n = normalizeDomain(d);
    if (n.length > 0 && n.includes(".")) {
      seen.add(n);
    }
  }
  return Array.from(seen).sort();
}

// ─── .txtpb-Formatierung ──────────────────────────────────────────────────────

function formatTxtpb(domains: string[]): string {
  // Jede Domain → eine Zeile: rows: { keyword: "domain.com" value: "1" }
  // kein Wrapper, kein BOM, UTF-8
  const lines = domains.map((d) => `rows: { keyword: "${d}" value: "1" }`);
  return lines.join("\n") + "\n";
}

// ─── Main ─────────────────────────────────────────────────────────────────────

async function main() {
  console.log(`PIR input.txtpb Generator`);
  console.log(`Source : ${source}`);
  console.log(`Output : ${outputPath}`);
  if (dryRun) console.log(`Mode   : dry-run (kein Schreibvorgang)`);
  console.log("");

  // 1. Domains laden
  const raw =
    source === "db" ? await fetchFromDb() : await fetchFromHagezi();

  console.log(`Rohe Domains geladen: ${raw.length.toLocaleString("de-DE")}`);

  // 2. Normalisieren + Dedupe + Sort
  const domains = processdomains(raw);
  console.log(
    `Nach Normalisierung + Dedupe: ${domains.length.toLocaleString("de-DE")} Domains`,
  );

  // 3. Vorschau
  console.log("\nBeispiel (erste 5 Zeilen):");
  domains.slice(0, 5).forEach((d) =>
    console.log(`  rows: { keyword: "${d}" value: "1" }`),
  );
  console.log("  ...");
  console.log("\nBeispiel (letzte 3 Zeilen):");
  domains.slice(-3).forEach((d) =>
    console.log(`  rows: { keyword: "${d}" value: "1" }`),
  );

  if (dryRun) {
    console.log("\ndry-run: Datei wird NICHT geschrieben.");
    return;
  }

  // 4. Schreiben
  const content = formatTxtpb(domains);
  const bytes = Buffer.byteLength(content, "utf8");
  writeFileSync(outputPath, content, { encoding: "utf8" });

  console.log(
    `\nGeschrieben: ${outputPath}`,
  );
  console.log(
    `Groesse    : ${(bytes / 1024 / 1024).toFixed(2)} MB (${bytes.toLocaleString("de-DE")} Bytes)`,
  );
  console.log(`Domains    : ${domains.length.toLocaleString("de-DE")}`);
}

main().catch((err) => {
  console.error("Fehler:", err);
  process.exit(1);
});