argus/src/gateway/public-gateway.mjs

// BNS public gateway (the "remote relay", convenience tier).
// Runs on the VPS. Resolves .bch from the chain (zero-dep resolver) and serves
// the content over plain HTTPS so ANY device's browser can view .bch sites with
// no install. Two URL forms:
//   path form:      https://navigate.st/bns/<name>.bch/...     (<base> injected)
//   subdomain form: https://<name-flattened>.bch.navigate.st/  (own origin, no <base>)
// where dots in a multi-level name become dashes: coinspectrum.deviant.bch
//   -> coinspectrum-deviant.bch.navigate.st
//
// Honest disclosure (convenience tier, not the trustless one): the operator can
// see which .bch sites you request. No logs — but that's trust. Install the
// resolver or Theseus for trustless resolution; use Tor for IP privacy.
import http from "http";
import https from "https";
import tls from "tls";
import { AwsClient } from "aws4fetch";
import WebSocket from "ws";
// resolver-web.js lives at ../lib/, not alongside this file — the previous
// `./resolver-web.js` import was broken (file was moved into lib/ but this
// import wasn't updated). Fix while touching the file.
import { resolveName, buildIndex, normalizeName } from "../lib/resolver-web.js";
import { pickRecord, isSubdomainOf } from "../lib/record-picker.js";

// Full-index cache (rebuilt every 30s) for fast availability lookups.
let _idx = { at: 0, index: null };
async function getIndex() {
  if (!_idx.index || Date.now() - _idx.at > 30000) {
    _idx = { at: Date.now(), index: await buildIndex({ WebSocket }) };
  }
  return _idx.index;
}

const MOUNT = process.env.BNS_MOUNT || "/bns";
const PORT = Number(process.env.BNS_PORT || 8700);
const S3_ENDPOINT = process.env.BNS_S3 || "http://127.0.0.1:8500";
const SUB_BASE = (process.env.BNS_SUB_BASE || "bch.navigate.st").toLowerCase(); // subdomain suffix
const aws = new AwsClient({
  accessKeyId: process.env.AK, secretAccessKey: process.env.SK,
  service: "s3", region: "us-east-1",
});

const cache = new Map();
async function resolve(host) {
  const c = cache.get(host);
  if (c && Date.now() - c.at < 30000) return c.entry;
  const entry = await resolveName(host, { WebSocket }); // resolveName normalizes any <sub>.<label>.<tld>
  cache.set(host, { entry, at: Date.now() });
  return entry;
}
// subfolder = labels before the registered "<label>.<tld>" (drop tld, drop label)
const subFolder = (host) => {
  const l = host.split("."); l.pop(); l.pop(); return l.reverse().join("/");
};
const MIME = { html: "text/html; charset=utf-8", css: "text/css", js: "text/javascript",
  json: "application/json", png: "image/png", jpg: "image/jpeg", jpeg: "image/jpeg",
  gif: "image/gif", svg: "image/svg+xml", ico: "image/x-icon", webp: "image/webp",
  woff2: "font/woff2", woff: "font/woff", wasm: "application/wasm", txt: "text/plain" };
const guess = (p) => MIME[p.split(".").pop()?.toLowerCase()] || "application/octet-stream";
function injectBase(html, href) {
  if (!href || /<base\s/i.test(html)) return html;
  if (/<head[^>]*>/i.test(html)) return html.replace(/<head[^>]*>/i, (m) => m + `<base href="${href}">`);
  return `<base href="${href}">` + html;
}
const nx = (h) => `<div style="font:16px system-ui;padding:2rem;color:#333"><h2>NXDOMAIN</h2><p><code>${h}</code> is not registered on the Bitcoin Cash chain.</p></div>`;

// BNS `ip`-record fetch. The default `fetch` fails here for two reasons:
//   1. It follows the site's HTTP→HTTPS 301 into `https://<name>.<tld>/`, which
//      isn't in ICANN DNS → "fetch failed".
//   2. It validates TLS against the public CA store, but BNS certs are signed
//      by per-machine BNS root CAs — the trust anchor is the on-chain `tls`
//      record's SHA-256 fingerprint, which we pin against here (see ca.js for
//      the trust model). This matches what Theseus does in-process.

// Compute a cert's SHA-256 fingerprint in the same format the on-chain `tls`
// record uses: lowercase hex, no separators.
function certFingerprint(cert) {
  const fp = cert && cert.fingerprint256;
  return fp ? fp.toLowerCase().replace(/:/g, "") : "";
}

// HTTPS GET pinned to an expected on-chain fingerprint. Standard CA validation
// is not the trust anchor for BNS names — the on-chain `tls` record is — so
// this connects over TLS and gates the request on fingerprint equality before
// any application bytes flow. A mismatch destroys the socket and rejects.
function pinnedHttpsGet(ip, port, servername, reqPath, expectedFp) {
  return new Promise((resolve, reject) => {
    const insecureOpts = { host: ip, port, servername };
    insecureOpts["rejectUnauthorized"] = false; // fingerprint pin below is the gate.
    const socket = tls.connect(insecureOpts, () => {
      const gotFp = certFingerprint(socket.getPeerCertificate(false));
      if (gotFp !== expectedFp) {
        socket.destroy();
        return reject(new Error(`tls fingerprint mismatch for ${servername}: got ${gotFp}, expected ${expectedFp}`));
      }
      const req = https.request({
        createConnection: () => socket,
        method: "GET",
        path: reqPath,
        headers: { host: servername, "user-agent": "bns-gateway/1" },
      }, (res) => {
        const chunks = [];
        res.on("data", (c) => chunks.push(c));
        res.on("end", () => resolve({ status: res.statusCode, headers: res.headers, buffer: Buffer.concat(chunks) }));
      });
      req.on("error", reject);
      req.end();
    });
    socket.setTimeout(15000, () => { socket.destroy(new Error("tls request timeout")); });
    socket.on("error", reject);
  });
}

// Fallback path — plain HTTP by IP with Host header. Used when the site has
// no on-chain `tls` record (fingerprint pinning impossible).
function httpGet(ip, reqPath, hostHeader) {
  return new Promise((resolve, reject) => {
    const req = http.request({
      host: ip, port: 80, method: "GET", path: reqPath,
      headers: { host: hostHeader, "user-agent": "bns-gateway/1" },
    }, (res) => {
      const chunks = [];
      res.on("data", (c) => chunks.push(c));
      res.on("end", () => resolve({ status: res.statusCode, headers: res.headers, buffer: Buffer.concat(chunks) }));
    });
    req.on("error", reject);
    req.end();
  });
}

// Compose: try pinned HTTPS if the on-chain `tls` fingerprint is available,
// otherwise plain HTTP by IP. If pinned HTTPS fails for any reason we do NOT
// silently fall back to HTTP — a fingerprint mismatch means "not the site the
// chain says it is" and returning HTTP anyway would defeat the pin.
async function ipRequest(ip, reqPath, hostHeader, tlsFingerprint) {
  if (tlsFingerprint) {
    return await pinnedHttpsGet(ip, 443, hostHeader, reqPath, tlsFingerprint.toLowerCase());
  }
  return await httpGet(ip, reqPath, hostHeader);
}

// Serve a resolved name. baseHref = null for subdomain form (own origin).
//
// SUBDOMAIN INHERITANCE — read before touching the record-priority ordering.
// A subdomain like `checkers.game.x` shares its parent's on-chain entry
// (`normalizeName` collapses it to `game.x`), so we serve it from the parent's
// records. But `s3` and `ip` mean *different things* about that inheritance:
//   - `ip` = "this server handles the whole namespace with Host routing" —
//     applies to the apex AND every subdomain the same way. Sending
//     `Host: checkers.game.x` lets the parent's nginx serve the subdomain.
//   - `s3` = "this Sia key is this exact name" — inheritance requires a
//     name-mangling convention (subFolder + name) that only works when the
//     parent chose to organise their bucket that way.
// So for subdomains, `ip` is the unambiguous intent and takes priority. For the
// apex, current priority stands (h → s3 → ip → u). This does not break
// Sia-per-subdomain owners: if they omit `ip` from the parent, the fallback
// still tries `s3` first as before.
async function serve(res, name, reqPath, baseHref) {
  let entry;
  try { entry = await resolve(name); } catch (e) { res.writeHead(502); return res.end("resolve error: " + e.message); }
  if (!entry) { res.writeHead(404, { "content-type": "text/html" }); return res.end(nx(name)); }
  const r = entry.records;
  // `pickRecord` is the single source of truth for record priority; keep this
  // call and the inline copy in TheseusNavigator/main.js:serveBns in step.
  const pick = pickRecord(r, { isSubdomain: isSubdomainOf(name, entry.name) });
  try {
    switch (pick) {
      case "h":
        if (reqPath !== "/") { res.writeHead(404); return res.end("not found"); }
        res.writeHead(200, { "content-type": "text/html; charset=utf-8" });
        return res.end(injectBase(r.h, baseHref));
      case "s3": {
        const sub = subFolder(name);
        let base = r.s3; if (base.endsWith("/") && sub) base += sub + "/";
        // Auto-index directory paths. S3 has no dir listing, so `/tools/`
        // would 404 (NoSuchKey) without this — matches how nginx handles
        // `try_files … /index.html` and how Ariadne's bnsd.js rewrites the
        // same URLs client-side.
        const rel = reqPath === "/" ? "index.html"
                  : reqPath.endsWith("/") ? reqPath.slice(1) + "index.html"
                  : reqPath.slice(1);
        const key = base.endsWith("/") ? base + rel : base;
        const up = await aws.fetch(`${S3_ENDPOINT}/${key}`);
        const buf = Buffer.from(await up.arrayBuffer());
        let ct = up.headers.get("content-type"); if (!ct || ct === "application/octet-stream") ct = guess(key);
        if (ct.includes("text/html")) { res.writeHead(up.status, { "content-type": ct }); return res.end(injectBase(buf.toString("utf8"), baseHref)); }
        res.writeHead(up.status, { "content-type": ct }); return res.end(buf);
      }
      case "ip": {
        const up = await ipRequest(r.ip, reqPath, name, r.tls);
        const buf = up.buffer;
        let ct = up.headers["content-type"] || guess(reqPath);
        if (ct.includes("text/html")) { res.writeHead(up.status, { "content-type": ct }); return res.end(injectBase(buf.toString("utf8"), baseHref)); }
        res.writeHead(up.status, { "content-type": ct }); return res.end(buf);
      }
      case "u":
        res.writeHead(302, { location: r.u }); return res.end();
      default:
        res.writeHead(200, { "content-type": "application/json" }); return res.end(JSON.stringify(entry, null, 2));
    }
  } catch (e) { res.writeHead(502); res.end("gateway error: " + e.message); }
}

// Native BNS TLDs — never overlap ICANN, so hijacking them here is safe.
// Dual TLDs (.com, .de, .dev) always fall through to search, so a bare
// `github.com` in the URL bar keeps going to the real GitHub, not to a
// BNS registration that happens to share the name.
const NATIVE_TLDS = new Set(["bch", "p2p", "bit", "nav"]);

// Universal "set-as-default-search-engine" shim. Users add
//   https://navigate.st/search?q=%s
// as their browser's default search engine; typing `silentmode.bch` in the URL
// bar (which Chromium/Firefox otherwise send to Google as a search query
// because .bch isn't in ICANN's known-TLD list) arrives here. If q looks like
// a native BNS name we 302 to the relay (works in every browser, no local
// daemon needed); otherwise forward to DuckDuckGo so ordinary search still
// works. Even an unregistered native-TLD query goes to the relay so the user
// gets a clear NXDOMAIN page instead of a confusing search detour.
async function handleSearch(url, res) {
  const q = (url.searchParams.get("q") || "").trim().toLowerCase();
  const bare = q.replace(/^[a-z]+:\/\//, "").split("/")[0];
  const m = bare.match(/^([a-z0-9-]+)\.([a-z0-9-]+)$/);
  let target;
  if (m && NATIVE_TLDS.has(m[2])) {
    target = `https://navigate.st/bns/${m[1]}.${m[2]}/`;
  } else {
    target = `https://duckduckgo.com/?q=${encodeURIComponent(q)}`;
  }
  res.writeHead(302, { location: target });
  res.end();
}

const server = http.createServer(async (req, res) => {
  try {
    const hostHeader = (req.headers.host || "").split(":")[0].toLowerCase();

    // --- smart search endpoint (default-search-engine shim) ---
    if (req.url.startsWith("/search")) {
      const u = new URL(req.url, "http://x");
      return handleSearch(u, res);
    }

    // --- registrar API: availability of a name ---
    // GET /api/name/<name>  ->  { name, tld, registered, records? }
    if (req.url.startsWith("/api/name/")) {
      res.setHeader("access-control-allow-origin", "*");
      res.setHeader("content-type", "application/json");
      const raw = decodeURIComponent(req.url.slice("/api/name/".length).split("?")[0]);
      try {
        const key = normalizeName(raw); // "<label>.<tld>"
        const tld = key.split(".").pop();
        const entry = (await getIndex()).get(key);
        res.writeHead(200);
        res.end(JSON.stringify({
          name: key, tld, registered: !!entry,
          records: entry?.records ?? null, category: entry?.category ?? null,
        }));
      } catch (e) {
        res.writeHead(400); res.end(JSON.stringify({ error: e.message }));
      }
      return;
    }

    // --- registrar API: all names sharing a label across every TLD ---
    // GET /api/labels/<label>  ->  { label, matches: [{ name, tld, records, category }] }
    // Powers navigate.st bare-label search: type "poutakidis" and get both
    // poutakidis.bch (if registered) and poutakidis.de rather than pinning to .bch.
    if (req.url.startsWith("/api/labels/")) {
      res.setHeader("access-control-allow-origin", "*");
      res.setHeader("content-type", "application/json");
      const raw = decodeURIComponent(req.url.slice("/api/labels/".length).split("?")[0]).toLowerCase();
      if (!/^[a-z0-9-]{1,32}$/.test(raw)) {
        res.writeHead(400);
        res.end(JSON.stringify({ error: "invalid label" }));
        return;
      }
      try {
        const idx = await getIndex();
        const matches = [];
        for (const [key, entry] of idx) {
          // Index keys: bare "<label>" for .bch, "<label>.<tld>" otherwise
          // (see normalizeName in resolver-web.js).
          const dot = key.indexOf(".");
          if (dot < 0) {
            if (key === raw) matches.push({
              name: key + ".bch", tld: "bch",
              records: entry?.records ?? null, category: entry?.category ?? null,
            });
          } else if (key.slice(0, dot) === raw) {
            matches.push({
              name: key, tld: key.slice(dot + 1),
              records: entry?.records ?? null, category: entry?.category ?? null,
            });
          }
        }
        matches.sort((a, b) => a.tld.localeCompare(b.tld));
        res.writeHead(200);
        res.end(JSON.stringify({ label: raw, matches }));
      } catch (e) {
        res.writeHead(500); res.end(JSON.stringify({ error: e.message }));
      }
      return;
    }

    // --- unified search: label + content ---
    // GET /api/search/<q>  ->  { query, count, results: [{ name, tld, matchType, score, snippet, records }] }
    // Scores: label-exact (100) > label-substring (70) > h-record content (50).
    // Content search covers on-chain inline HTML only; Sia/ip sites await Helios v1.
    // Same cached index as /api/labels/ and /api/name/ — no extra fetch cost on the hot path.
    if (req.url.startsWith("/api/search/")) {
      res.setHeader("access-control-allow-origin", "*");
      res.setHeader("content-type", "application/json");
      const q = decodeURIComponent(req.url.slice("/api/search/".length).split("?")[0]).toLowerCase().trim();
      if (!q || q.length < 2 || q.length > 64) {
        res.writeHead(400);
        res.end(JSON.stringify({ error: "query must be 2-64 chars" }));
        return;
      }
      try {
        const idx = await getIndex();
        const results = [];
        for (const [key, entry] of idx) {
          const dot = key.indexOf(".");
          const label = dot < 0 ? key : key.slice(0, dot);
          const tld = dot < 0 ? "bch" : key.slice(dot + 1);
          const displayName = dot < 0 ? key + ".bch" : key;
          const records = entry?.records ?? {};

          let matchType = null, score = 0, snippet = null;
          if (label === q) { matchType = "label-exact"; score = 100; }
          else if (label.includes(q)) { matchType = "label-substring"; score = 70; }

          // Content search over inline HTML (h record). Strip tags first so a
          // query like "poutakidis" doesn't accidentally match "class=poutakidis".
          if (typeof records.h === "string" && records.h) {
            const text = records.h.replace(/<[^>]+>/g, " ").replace(/\s+/g, " ").trim();
            const i = text.toLowerCase().indexOf(q);
            if (i >= 0) {
              if (!matchType) { matchType = "content"; score = 50; }
              const start = Math.max(0, i - 40);
              const end = Math.min(text.length, i + q.length + 60);
              snippet = (start > 0 ? "…" : "") + text.slice(start, end) + (end < text.length ? "…" : "");
            }
          }

          if (matchType) {
            results.push({ name: displayName, tld, matchType, score, snippet,
              records: entry?.records ?? null, category: entry?.category ?? null });
          }
        }
        results.sort((a, b) => b.score - a.score || a.name.localeCompare(b.name));
        const capped = results.slice(0, 50);
        res.writeHead(200);
        res.end(JSON.stringify({ query: q, count: capped.length, truncated: results.length > 50, results: capped }));
      } catch (e) {
        res.writeHead(500); res.end(JSON.stringify({ error: e.message }));
      }
      return;
    }

    // --- subdomain form: <flattened>.bch.<SUB_BASE> ---
    if (hostHeader.endsWith("." + SUB_BASE)) {
      const inner = hostHeader.slice(0, -("." + SUB_BASE).length); // "<flattened>.bch" or "<flattened>"
      const flat = inner.endsWith(".bch") ? inner.slice(0, -4) : inner;
      if (flat) {
        const name = flat.replace(/-/g, ".") + ".bch"; // dashes back to dots
        return serve(res, name, req.url || "/", null);  // own origin: no <base>
      }
    }
    // bare apex of SUB_BASE (e.g. bch.navigate.st) -> home
    if (hostHeader === SUB_BASE) { res.writeHead(200, { "content-type": "text/html" }); return res.end(home()); }

    // --- path form: /bns/<name>.bch/... ---
    let p = req.url;
    if (p === MOUNT || p === MOUNT + "/") { res.writeHead(200, { "content-type": "text/html" }); return res.end(home()); }
    if (!p.startsWith(MOUNT + "/")) { res.writeHead(404); return res.end("bns gateway"); }
    p = p.slice(MOUNT.length + 1);
    const m = p.match(/^([a-z0-9-]+(?:\.[a-z0-9-]+)+)(\/.*)?$/i); // <name>.<tld>, any TLD
    if (!m) { res.writeHead(400, { "content-type": "text/html" }); return res.end(`usage: ${MOUNT}/<name>.<tld>/`); }
    const name = m[1].toLowerCase();
    const rest = m[2] || "/";
    return serve(res, name, rest, `${MOUNT}/${name}/`);
  } catch (e) { res.writeHead(500); res.end("gateway error: " + e.message); }
});
server.listen(PORT, "127.0.0.1", () => console.log(`bns public gateway on 127.0.0.1:${PORT} (path ${MOUNT}, subdomain *.${SUB_BASE})`));

function home() {
  return `<!doctype html><meta charset=utf-8><title>BNS Gateway</title>
  <body style="font:16px system-ui;max-width:640px;margin:3rem auto;padding:0 1rem;color:#e7eaf1;background:#0f1420">
  <h1 style="color:#d6ff3d">⛓ BNS Gateway</h1>
  <p>View Bitcoin Cash <b>.bch</b> names in any browser — no install.</p>
  <ul>
    <li><a href="/bns/hello.bch">/bns/hello.bch</a> · or <a href="https://hello.${SUB_BASE}/">hello.${SUB_BASE}</a></li>
    <li><a href="/bns/coinspectrum.deviant.bch">/bns/coinspectrum.deviant.bch</a></li>
  </ul>
  <p style="color:#8b98a9;font-size:13px">This gateway resolves names for you, so it can see which .bch
  sites you request (it keeps no logs — you're trusting that). For trustless resolution install the local
  resolver/extension; for IP privacy use Tor. By Silent Mode (a Deviant project).</p></body>`;
}