Cap author bios at 500 chars; use Wikipedia exsentences=4; block AbeBooks boilerplate OG descriptions

Reject boilerplate descriptions in sanitizer; prefer Wikipedia over OL name-search for author bios (fixes wrong-person disambiguation)

Parse title and author from AbeBooks listing pages, reject boilerplate description.

Add Wikipedia author-bio fallback for reliable About the Author.

Add ThriftBooks parser with Book Overview, author-page bio, and iTunes fallback.

Reject JSON-LD leaks, invalid years, and foreign-language genres in book scraper.

Fill B&N description and genre via iTunes when page scrape is blocked.

Fix B&N slug parsing and ISBN fallbacks when HTML or catalog APIs fail.

Trigger Hugging Face Space rebuild for latest book URL scraper.

Ship B&N scraper refinements, platform backtest results, and project docs.

Fix Barnes and Noble book import for Shopify storefront pages.

Harden book URL import across top platforms and swap Bookshop for WorldCat in UI.

Improve multi-platform book URL import with ranked catalog matching.

Fix: Make Amazon author URL extraction more resilient to capture follow-the-author-card links

Fix: Add Amazon data-testid pattern for modern Author Central biography extraction

Fix: Add Amazon Author profile link extraction and secondary fetch to get biography if missing from book page

Fix: Amazon description and bio extraction by stripping scripts and parsing JSON blobs

Fix: bypass Redis cache on preview (force_refresh) so improved Amazon scraper runs fresh

Fix Amazon: title/author/description/bio now fetched from page DOM instead of misleading OG meta tags

Fix URL input bottom padding

Redesign Upload Book section with modern UI

fix: correct DATA_API base URL in superadmin.html to prevent duplicate /super path

fix: correct API base URL from /api/v1 to /api in all frontend files

feat: universal genre fallback via OL Search + Google Books

fix: cover+genre now actually persist in DB

feat: full cover+genre extraction from all platforms - platform HTML parsers + DB fields

feat: rich shimmer skeleton loading animation for books table

fix: 3 bugs — stalled false positive, unknown platform crash, genre/processing UI

fix: SyntaxError in books.py — close preview return dict before import-url decorator

feat: URL import now auto-indexes scraped metadata instantly — no file upload needed

fix: indexing-stalled false positive, genre not saved, book table UI

feat: replace Upload Book UI with full URL-import-first flow in real admin SPA

feat: full admin SPA + superadmin persistent login fix

fix: resolve Pyrefly unknown-name errors in book.py

feat: scraper v3 - API-first, never-blocked architecture

feat: deep book metadata extraction - author bio, genre, ratings, pages

feat: book URL import - 10 platforms, 3-step UI, scraper service

feat: show IP address in sessions + persistent login until logout - Add ip_address field to ChatSession model (VARCHAR 45, handles IPv4+IPv6) - Store real client IP at session/init time via get_real_ip() - Expose ip_address in /sessions list and /sessions/{id}/transcript responses - DB migration: ip_address column added to chat_sessions - Bump REFRESH_TOKEN_EXPIRE_DAYS from 7 to 30 days (stay logged in 1 month) - Fix auth-client.js: refreshAccessToken only clears session on 401/403, not on network/5xx errors (HF cold-start safe) - Fix restoreSession: keeps session alive when refresh fails transiently, only clears on explicit auth rejection

fix: geo lookup now works on HF Spaces without MaxMind DB - Add ip-api.com async HTTP fallback when GeoLite2 mmdb is not present - 3-tier resolution: Redis cache (24h) -> MaxMind (offline) -> ip-api.com (free, no key) - Cache key uses SHA-256 hash of IP (raw IP never stored) - Pass redis into get_geo_info from both chat.py and visitor_tracker.py - Fixes country showing as dash on all sessions

feat: proper visitor analytics system - New Visitor model (visitor_uid + fingerprint fallback, first_seen/last_seen/page_views, geo, device) - VisitorRepository with upsert + analytics queries - visitor_tracker.py: record_visitor() upserts on every session/init - session/init accepts visitor_uid in request body (HF-Spaces iframe-safe, no cookies) - 4 new analytics endpoints: /visitors /geo /devices /sessions/stats - geo.py: region extraction + X-Real-IP support - DB migration: visitor_uid column on chat_sessions - ARCHITECTURE.md updated

fix: 5 critical storage bugs + 3 SQLite/SSE bugs -- tracker.py: db.flush() after add() so chat messages/analytics are persisted -- chat.py: flush->commit for durable ChatSession row -- settings.py: fix session mismatch in all PUT routes (widget-config, profile, personality, notifications, password) - setattr on cross-session user was discarded; now uses UPDATE stmt on route db -- aggregator.py: replace PostgreSQL-only gen_random_uuid() with uuid4() param for SQLite -- ingest.py: fix SSE channel name mismatch subscriber vs publisher -- superadmin_service.py: fix naive/aware datetime TypeError in extend_subscription

fix: Audit pass — critical bug fix + lazy import cleanup (qa/books/schemas_router)

feat: Phase 10 — Analytics endpoints (unanswered/books/heatmap)

test: Phase 9 — 55 tests all passing (23 new in Suite F)

feat: Phase 7 — Polling streaming (stream-start + poll endpoints)

fix: keep run_backup as lazy import (celery not in local dev env)

refactor: Phase 6A+6B — top-level imports and Pydantic schema upgrades

feat: Phases 2B + 3B — semantic Q&A and vector store fixes

Docs: update ARCHITECTURE.md change log for Phases 1-5

feat: Phases 1-5 production improvements + 8 bug fixes