/**
 * ─────────────────────────────────────────────
 *  SPB Drive — Personal Cloud Drive
 * ─────────────────────────────────────────────
 *  Author  : Simon-Pierre Boucher
 *  Contact : contact@spboucher.ai
 *  File    : src/storage/blobs.mjs
 *  Purpose : Content-addressed blob store — sha256 CAS, refcounts, GC
 *  License : MIT © Simon-Pierre Boucher
 * ─────────────────────────────────────────────
 */

import { createHash, randomBytes } from 'node:crypto';
import { createReadStream, createWriteStream, existsSync } from 'node:fs';
import { mkdir, rename, rm, stat } from 'node:fs/promises';
import path from 'node:path';
import { pipeline } from 'node:stream/promises';
import { config } from '../config.mjs';
import { getDb } from '../db/db.mjs';

/** Absolute on-disk path for a blob sha. */
export function blobPath(sha) {
  return path.join(config.filesDir, sha.slice(0, 2), sha.slice(2, 4), sha);
}

/**
 * Store a readable stream in the CAS. Streams to a temp file while hashing,
 * then moves into place — identical content dedupes naturally.
 * @param {NodeJS.ReadableStream} stream
 * @returns {Promise<{sha: string, size: number, deduped: boolean}>}
 */
export async function putBlob(stream) {
  const tmp = path.join(config.cacheDir, 'tmp', `put-${randomBytes(8).toString('hex')}`);
  const hash = createHash('sha256');
  let size = 0;

  await pipeline(
    stream,
    async function* (source) {
      for await (const chunk of source) {
        hash.update(chunk);
        size += chunk.length;
        yield chunk;
      }
    },
    createWriteStream(tmp, { mode: 0o600 }),
  );

  const sha = hash.digest('hex');
  const dest = blobPath(sha);
  const deduped = existsSync(dest);
  if (deduped) {
    await rm(tmp, { force: true });
  } else {
    await mkdir(path.dirname(dest), { recursive: true, mode: 0o700 });
    await rename(tmp, dest);
  }

  getDb()
    .prepare(
      `INSERT INTO blobs (sha, size, refcount, created) VALUES (?, ?, 0, ?)
       ON CONFLICT(sha) DO NOTHING`,
    )
    .run(sha, size, Date.now());
  return { sha, size, deduped };
}

/** Increment a blob's refcount (one node now points at it). */
export function refBlob(sha) {
  getDb().prepare('UPDATE blobs SET refcount = refcount + 1 WHERE sha = ?').run(sha);
}

/** Decrement a blob's refcount. Physical deletion happens in gcBlobs(). */
export function unrefBlob(sha) {
  getDb().prepare('UPDATE blobs SET refcount = MAX(refcount - 1, 0) WHERE sha = ?').run(sha);
}

/** Readable stream over a blob (optionally a byte range). */
export function blobStream(sha, { start, end } = {}) {
  return createReadStream(blobPath(sha), start !== undefined ? { start, end } : {});
}

/** Blob size from disk (throws if missing). */
export async function blobSize(sha) {
  return (await stat(blobPath(sha))).size;
}

/**
 * Delete blobs with refcount 0 (and their cache artifacts).
 * @returns {Promise<number>} number of blobs removed
 */
export async function gcBlobs() {
  const db = getDb();
  const orphans = db.prepare('SELECT sha FROM blobs WHERE refcount <= 0').all();
  for (const { sha } of orphans) {
    await rm(blobPath(sha), { force: true });
    for (const sub of ['thumbs', 'transcode', 'office', 'peaks', 'text']) {
      const dir = path.join(config.cacheDir, sub);
      for (const suffix of ['', '.256.webp', '.512.webp', '.pdf', '.mp4', '.json', '.txt']) {
        await rm(path.join(dir, `${sha}${suffix}`), { force: true });
      }
    }
    db.prepare('DELETE FROM blobs WHERE sha = ?').run(sha);
  }
  return orphans.length;
}

/** Aggregate storage stats for /healthz and the storage meter. */
export function storageStats() {
  const db = getDb();
  const blobs = db.prepare('SELECT COUNT(*) AS n, COALESCE(SUM(size), 0) AS bytes FROM blobs WHERE refcount > 0').get();
  const nodes = db.prepare("SELECT COUNT(*) AS n FROM nodes WHERE trashed_at IS NULL AND id != 1").get();
  const byType = db
    .prepare(
      `SELECT CASE
         WHEN mime LIKE 'image/%' THEN 'images'
         WHEN mime LIKE 'video/%' THEN 'video'
         WHEN mime LIKE 'audio/%' THEN 'audio'
         WHEN mime LIKE 'application/pdf' OR mime LIKE '%word%' OR mime LIKE '%sheet%'
           OR mime LIKE '%presentation%' OR mime LIKE 'text/%' THEN 'docs'
         WHEN mime LIKE '%zip%' OR mime LIKE '%tar%' OR mime LIKE '%7z%' OR mime LIKE '%rar%' THEN 'archives'
         ELSE 'other' END AS bucket,
         COALESCE(SUM(size), 0) AS bytes, COUNT(*) AS n
       FROM nodes WHERE type = 'file' AND trashed_at IS NULL GROUP BY bucket`,
    )
    .all();
  return { blobCount: blobs.n, usedBytes: blobs.bytes, nodeCount: nodes.n, byType };
}