diora-web/static/js/app/48-book-cache.js

'use strict';

// ---------------------------------------------------------------------------
// Book cache — IndexedDB, evict after 4 weeks of inactivity
// ---------------------------------------------------------------------------
const _BOOK_CACHE_STORE       = 'books';
const _BOOK_META_STORE        = 'book_meta';
const _BOOK_ANNOTATIONS_STORE = 'book_annotations';
const _BOOK_PARSED_STORE      = 'parsed';   // prepared chapters, see _getParsedBook
const _BOOK_CACHE_EVICT_MS    = 4 * 7 * 24 * 60 * 60 * 1000;

function _openBookCacheDb() {
  return new Promise((resolve, reject) => {
    const req = indexedDB.open('diora_book_cache', 3);
    req.onupgradeneeded = e => {
      const db = e.target.result;
      if (!db.objectStoreNames.contains(_BOOK_CACHE_STORE))
        db.createObjectStore(_BOOK_CACHE_STORE, {keyPath: 'id'});
      if (!db.objectStoreNames.contains(_BOOK_META_STORE))
        db.createObjectStore(_BOOK_META_STORE, {keyPath: 'id'});
      if (!db.objectStoreNames.contains(_BOOK_ANNOTATIONS_STORE))
        db.createObjectStore(_BOOK_ANNOTATIONS_STORE, {keyPath: 'id'});
      if (!db.objectStoreNames.contains(_BOOK_PARSED_STORE))
        db.createObjectStore(_BOOK_PARSED_STORE, {keyPath: 'id'});
    };
    req.onsuccess  = e => resolve(e.target.result);
    req.onerror    = ()  => reject();
  });
}

// ---------------------------------------------------------------------------
// Prepared books: an EPUB's chapters after unpacking and sanitizing, so opening
// it again skips both (for a book like Das Kapital that's most of the time an
// e-reader needs to open it). Encrypted with the book key like everything else
// on the device. Valid for one version of the book file (its IV) and of the
// code that prepares chapters (a hash of its source: changing the sanitizer
// makes every book be prepared anew). Images stay in the book file: the
// chapters keep "diora-img:<path>" in place of the session's blob URLs.
// ---------------------------------------------------------------------------
let _parseCodeSig = null;

function _parsePreparationSig() {
  if (_parseCodeSig) return _parseCodeSig;
  // window.__dioraPrepareSig: what an app layer adds to the preparation — a
  // change there too.
  const src = [parseEpub, sanitizeEpubHtml, _looksLikeFootnoteLink, _isNoteBacklink, _injectImageBlobs, _resolveImageBlob, _parseEpubToc]
    .map(f => String(f)).join('\n') + (window.__dioraPrepareSig || '');
  let h = 5381;
  for (let i = 0; i < src.length; i++) h = ((h << 5) + h + src.charCodeAt(i)) | 0;
  return (_parseCodeSig = (h >>> 0).toString(36) + '.' + src.length);
}

async function _getParsedBook(bookId, dataIv, key) {
  try {
    const db = await _openBookCacheDb();
    const row = await new Promise(resolve => {
      const req = db.transaction(_BOOK_PARSED_STORE).objectStore(_BOOK_PARSED_STORE).get(bookId);
      req.onsuccess = e => resolve(e.target.result || null);
      req.onerror = () => resolve(null);
    });
    if (!row || row.sig !== dataIv + ':' + _parsePreparationSig()) return null;
    const plain = await crypto.subtle.decrypt({name: 'AES-GCM', iv: row.iv}, key, row.ct);
    return JSON.parse(new TextDecoder().decode(plain));
  } catch (e) { return null; }
}

async function _setParsedBook(bookId, dataIv, key, result) {
  try {
    // Blob URLs live as long as this page; the stored chapters name the image instead.
    const pathOf = {};
    for (const [path, url] of Object.entries(result.imageMap || {})) pathOf[url] = path;
    const blobRe = /blob:[^"'\s)]+/g;
    const chapters = result.chapters.map(c => c.replace(blobRe, u => pathOf[u] ? 'diora-img:' + pathOf[u] : u));
    const doc = {title: result.title, author: result.author, toc: result.toc, chapters,
                 images: Object.keys(result.imageMap || {})};
    // Raw bytes, not base64: IndexedDB keeps an ArrayBuffer as it is (megabytes for a big book).
    const iv = crypto.getRandomValues(new Uint8Array(12));
    const ct = await crypto.subtle.encrypt({name: 'AES-GCM', iv}, key, new TextEncoder().encode(JSON.stringify(doc)));
    const db = await _openBookCacheDb();
    await new Promise(resolve => {
      const tx = db.transaction(_BOOK_PARSED_STORE, 'readwrite');
      const store = tx.objectStore(_BOOK_PARSED_STORE);
      store.put({id: bookId, sig: dataIv + ':' + _parsePreparationSig(), iv, ct, cached_at: Date.now()});
      // Prepared books not opened for a while go (they come back on the next open).
      const all = store.getAll();
      all.onsuccess = () => all.result.forEach(r => {
        if (r.id !== bookId && Date.now() - (r.cached_at || 0) > _BOOK_CACHE_EVICT_MS) store.delete(r.id);
      });
      tx.oncomplete = resolve;
      tx.onerror = resolve;
    });
  } catch (e) {}
}

/** A prepared book back to what parseEpub returns: fresh blob URLs for its images. */
async function _restoreParsedBook(doc, epubBytes) {
  const imageMap = {};
  if (doc.images && doc.images.length) {
    const zip = await JSZip.loadAsync(epubBytes);
    await Promise.all(doc.images.map(async path => {
      const f = zip.file(path);
      if (!f) return;
      try {
        const buf = await f.async('arraybuffer');
        const ext = path.split('.').pop().toLowerCase();
        const type = {jpg: 'image/jpeg', jpeg: 'image/jpeg', png: 'image/png', gif: 'image/gif',
                      svg: 'image/svg+xml', webp: 'image/webp'}[ext] || '';
        imageMap[path] = URL.createObjectURL(new Blob([buf], type ? {type} : {}));
      } catch (e) {}
    }));
  }
  const placeholder = /diora-img:([^"'\s)]+)/g;
  const chapters = doc.chapters.map(c => c.replace(placeholder, (m, path) => imageMap[path] || ''));
  return {title: doc.title, author: doc.author, toc: doc.toc, chapters, imageMap};
}

async function _getCachedBook(bookId) {
  try {
    const db = await _openBookCacheDb();
    return await new Promise((resolve) => {
      const req = db.transaction(_BOOK_CACHE_STORE).objectStore(_BOOK_CACHE_STORE).get(bookId);
      req.onsuccess = e => resolve(e.target.result || null);
      req.onerror   = ()  => resolve(null);
    });
  } catch { return null; }
}

async function _setCachedBook(bookId, data_ct, data_iv) {
  try {
    const db = await _openBookCacheDb();
    await new Promise((resolve) => {
      const tx = db.transaction(_BOOK_CACHE_STORE, 'readwrite');
      tx.objectStore(_BOOK_CACHE_STORE).put({id: bookId, data_ct, data_iv, cached_at: Date.now()});
      tx.oncomplete = resolve;
      tx.onerror    = resolve;
    });
    // Request persistent storage so the browser doesn't evict cached books
    // under storage pressure. Chrome grants this silently for installed PWAs;
    // Firefox may show a permission prompt.
    if (navigator.storage?.persist) navigator.storage.persist();
  } catch {}
}

async function _evictCachedBook(bookId) {
  try {
    const db = await _openBookCacheDb();
    await new Promise((resolve) => {
      const tx = db.transaction(_BOOK_CACHE_STORE, 'readwrite');
      tx.objectStore(_BOOK_CACHE_STORE).delete(bookId);
      tx.oncomplete = resolve;
      tx.onerror    = resolve;
    });
  } catch {}
}

async function _evictBookCache(bookList) {
  try {
    const db = await _openBookCacheDb();
    const now = Date.now();
    const metaById = Object.fromEntries(bookList.map(b => [b.id, b]));
    await new Promise((resolve) => {
      const tx = db.transaction(_BOOK_CACHE_STORE, 'readwrite');
      const store = tx.objectStore(_BOOK_CACHE_STORE);
      store.openCursor().onsuccess = e => {
        const cursor = e.target.result;
        if (!cursor) { resolve(); return; }
        const meta = metaById[cursor.value.id];
        const lastRead = meta?.last_read ? new Date(meta.last_read).getTime() : 0;
        if (!meta || (now - lastRead) > _BOOK_CACHE_EVICT_MS) cursor.delete();
        cursor.continue();
      };
      tx.onerror = resolve;
    });
  } catch {}
}

async function _saveBookMeta(books) {
  try {
    const db = await _openBookCacheDb();
    const ids = new Set(books.map(b => String(b.id)));
    await new Promise(resolve => {
      const tx = db.transaction(_BOOK_META_STORE, 'readwrite');
      const store = tx.objectStore(_BOOK_META_STORE);
      for (const b of books) store.put(b);
      store.openCursor().onsuccess = e => {
        const cursor = e.target.result;
        if (!cursor) return;
        if (!ids.has(String(cursor.value.id))) cursor.delete();
        cursor.continue();
      };
      tx.oncomplete = resolve;
      tx.onerror    = resolve;
    });
  } catch {}
}

async function _loadCachedBookMeta() {
  try {
    const db = await _openBookCacheDb();
    return await new Promise(resolve => {
      const req = db.transaction(_BOOK_META_STORE).objectStore(_BOOK_META_STORE).getAll();
      req.onsuccess = e => resolve(e.target.result || []);
      req.onerror   = ()  => resolve([]);
    });
  } catch { return []; }
}

async function _updateCachedBookProgress(bookId, fraction, anchor, force = false, readAt = null) {
  const at = (readAt == null ? Date.now() : readAt) + _serverOffset;
  try {
    const db = await _openBookCacheDb();
    await new Promise(resolve => {
      const tx = db.transaction(_BOOK_META_STORE, 'readwrite');
      const store = tx.objectStore(_BOOK_META_STORE);
      const req = store.get(bookId);
      req.onsuccess = e => {
        const entry = e.target.result;
        // The place read last, as on the server (docs/progress-sync.md).
        if (entry && (force || entry.read_at == null || at >= entry.read_at)) {
          entry.scroll_fraction  = fraction;
          entry.position_anchor  = anchor;
          entry.read_at          = at;
          entry.device           = _progressDevice();
          entry.last_read        = new Date(at).toISOString();
          if (_cmpProgress(anchor, fraction, entry.furthest_anchor || '', entry.furthest_fraction || 0) > 0) {
            entry.furthest_anchor = anchor;
            entry.furthest_fraction = fraction;
          }
          store.put(entry);
        }
      };
      tx.oncomplete = resolve;
      tx.onerror    = resolve;
    });
  } catch {}
}

async function _saveAnnotationsCache(bookId, type, ct, iv) {
  try {
    const db = await _openBookCacheDb();
    await new Promise(resolve => {
      const tx = db.transaction(_BOOK_ANNOTATIONS_STORE, 'readwrite');
      tx.objectStore(_BOOK_ANNOTATIONS_STORE).put({id: `${bookId}_${type}`, ct, iv});
      tx.oncomplete = resolve;
      tx.onerror    = resolve;
    });
  } catch {}
}

async function _loadAnnotationsCache(bookId, type) {
  try {
    const db = await _openBookCacheDb();
    return await new Promise(resolve => {
      const req = db.transaction(_BOOK_ANNOTATIONS_STORE).objectStore(_BOOK_ANNOTATIONS_STORE).get(`${bookId}_${type}`);
      req.onsuccess = e => resolve(e.target.result || null);
      req.onerror   = ()  => resolve(null);
    });
  } catch { return null; }
}

const _PENDING_SYNC_KEY = 'diora_pending_sync';

function _markPendingSync(bookId, type) {
  try {
    const pending = JSON.parse(localStorage.getItem(_PENDING_SYNC_KEY) || '[]');
    const key = `${bookId}:${type}`;
    if (!pending.includes(key)) pending.push(key);
    localStorage.setItem(_PENDING_SYNC_KEY, JSON.stringify(pending));
  } catch {}
}

async function _syncPendingAnnotations() {
  try {
    const pending = JSON.parse(localStorage.getItem(_PENDING_SYNC_KEY) || '[]');
    if (!pending.length) return;
    const remaining = [];
    for (const key of pending) {
      const sep = key.indexOf(':');
      const bookId = key.slice(0, sep);
      const type   = key.slice(sep + 1);
      const cached = await _loadAnnotationsCache(bookId, type);
      if (!cached?.ct) continue;
      const url = type === 'bookmarks'
        ? `/books/${bookId}/bookmarks/`
        : `/books/${bookId}/highlights/`;
      try {
        await fetch(url, {
          method: 'POST',
          headers: {'Content-Type': 'application/json', 'X-CSRFToken': getCsrfToken()},
          body: JSON.stringify({ct: cached.ct, iv: cached.iv}),
        });
      } catch { remaining.push(key); }
    }
    localStorage.setItem(_PENDING_SYNC_KEY, JSON.stringify(remaining));
  } catch {}
}