prometheus/service/storage.js

/**
 * Sia-backed storage for scan-clean uploads.
 *
 * When the service is configured with S3 credentials pointing at the Silent
 * Mode hosted s3d (s3.silentmode.st), every clean upload also goes to Sia and
 * the public download URL is handed back to the submit UI. The publisher no
 * longer has to self-host — their binary gets a stable https URL right out
 * of step 1.
 *
 * Design notes
 * ------------
 * - Writes go to `s3://${bucket}/${prefix}<sha256>.bin` by content hash.
 *   Collisions on hash are a no-op: the same bytes always land at the same
 *   key, so re-uploads do not duplicate storage.
 * - Downloads are served through the s3 gateway's public URL
 *   (`${publicBase}/${bucket}/${prefix}<sha256>.bin`). The bns bucket on
 *   s3.silentmode.st is already public-read so site mirrors and Pithos
 *   share-links work from any client — nothing extra to configure here.
 * - If storage is not configured, isConfigured() returns false and the
 *   service falls back to its old behavior (publisher provides their own
 *   download_url). That keeps dev loops working without creds.
 * - Content-Type is passed through so the gateway serves apk/exe with the
 *   right MIME, which matters for mobile download prompts.
 *
 * Env
 * ---
 *   PROMETHEUS_S3_ENDPOINT     e.g. https://s3.silentmode.st
 *   PROMETHEUS_S3_ACCESS_KEY   access key for a sia-storage user
 *   PROMETHEUS_S3_SECRET       matching secret
 *   PROMETHEUS_S3_BUCKET       default: bns
 *   PROMETHEUS_S3_PREFIX       default: prometheus-storage/
 *   PROMETHEUS_PUBLIC_BASE     default: same as PROMETHEUS_S3_ENDPOINT
 */

import { createReadStream } from 'node:fs';
import { stat } from 'node:fs/promises';
import { S3Client } from './s3.js';

const CFG = {
  endpoint: process.env.PROMETHEUS_S3_ENDPOINT || '',
  accessKey: process.env.PROMETHEUS_S3_ACCESS_KEY || '',
  secret: process.env.PROMETHEUS_S3_SECRET || '',
  bucket: process.env.PROMETHEUS_S3_BUCKET || 'bns',
  prefix: normalizePrefix(process.env.PROMETHEUS_S3_PREFIX || 'prometheus-storage/'),
  publicBase: (process.env.PROMETHEUS_PUBLIC_BASE || process.env.PROMETHEUS_S3_ENDPOINT || '').replace(/\/+$/, ''),
};

let client = null;

export function isConfigured() {
  return Boolean(CFG.endpoint && CFG.accessKey && CFG.secret);
}

export function describe() {
  if (!isConfigured()) return { configured: false };
  return {
    configured: true,
    endpoint: CFG.endpoint,
    bucket: CFG.bucket,
    prefix: CFG.prefix,
    publicBase: CFG.publicBase,
  };
}

function getClient() {
  if (!isConfigured()) return null;
  if (!client) {
    client = new S3Client({
      endpoint: CFG.endpoint,
      accessKeyId: CFG.accessKey,
      secretKey: CFG.secret,
    });
  }
  return client;
}

/**
 * Upload a scan-clean file to Sia and return the public URL readers can fetch.
 *
 * Idempotent: if an object with the same key already exists (same sha256 →
 * same bytes), the PUT overwrites with identical content and the URL is the
 * same. The returned URL is stable across reruns, so a listing that already
 * references it stays valid.
 *
 * @param {object} args
 * @param {string} args.localPath — absolute path to the file on disk
 * @param {string} args.sha256    — content hash, used as the object key
 * @param {string} args.name      — original filename (for the display suffix)
 * @param {string} args.mime      — content-type for the stored object
 * @returns {Promise<{ downloadUrl: string, key: string, bytes: number }>}
 */
export async function uploadScannedFile({ localPath, sha256, name, mime }) {
  const c = getClient();
  if (!c) throw new Error('storage not configured — set PROMETHEUS_S3_*');

  const { size } = await stat(localPath);
  const key = `${CFG.prefix}${sha256}.bin`;
  const body = createReadStream(localPath);
  const contentType = mime || 'application/octet-stream';

  // The s3d upstream wants Content-Length for streaming PUTs; without it
  // s3d falls back to transfer-encoding: chunked which Sia's gateway does
  // not pass through cleanly on large bodies. Pithos has the same gotcha.
  await c.putObject(CFG.bucket, key, body, {
    contentType,
    contentLength: size,
  });

  // Content-Disposition is a client decision, not something we persist on
  // the object; the UI can add it to the download link if needed. We just
  // hand back the raw public URL — stable, cacheable, hash-addressed.
  const downloadUrl = publicUrlFor(key);
  return { downloadUrl, key, bytes: size };
}

export function publicUrlFor(key) {
  if (!CFG.publicBase) return null;
  // Path-style URL: the bucket is a path segment, matching how s3d exposes
  // the hosted gateway. (Virtual-hosted URLs would need per-bucket DNS,
  // which the public gateway does not set up.)
  return `${CFG.publicBase}/${encodeURIComponent(CFG.bucket)}/${encodeURI(key)}`;
}

function normalizePrefix(p) {
  const t = String(p || '').replace(/^\/+/, '');
  return t.endsWith('/') || t === '' ? t : t + '/';
}