Fetching metadata from the HF Docker repository...
Jun 25 13:09 Fix: Apply buttons now REPLACE text (not append), add translation reference/copy/apply buttons 12:12 Fix: remove extra closing div that broke summarize-area nesting 12:03 UI: Footer only in write tab, remove Quran toolbar btn, summary always custom with import 11:49 UI: Make window dots decorative only, rename floating bar Quran label 11:32 UI/UX: Add Quran standalone tab, summarize textarea, floating selection bar, fix clear editor, remove duplicate button Jun 21 16:34 feat: Add popover dismiss button (تجاهل) + dismiss CSS styling 16:28 fix: Correct hero badge error count from 3 to 8 (4+1+3) 16:26 fix: Remove Quran + Dialect links from footer 16:19 feat: Complete ALL plan items — gradient tokens, nav scroll glow, focus rings, dropdown keyboard nav, color reset, network delay indicator 16:09 feat: Add Quran + Dialect product links to footer 15:57 fix: Correct toast types (warning/error) + analyze API error toast + cleanup 15:48 fix: tool count ٨→٧ + Bayyinah arrow ↗ 15:42 fix: Bayyinah external link arrow direction (← → ↗) 15:41 feat: UI/UX Overhaul Phase 1-9 — CSS modernization, RTL arrows, pricing features, confirm dialog, meta SEO, footer links, hero copy, shimmer/glow/glassmorphism 15:02 ux: Escape closes Quran modal + char counter on dialect textarea (red when >5000) 15:00 fix: Quran XSS escape + 30s timeout + hide buttons during loading (same fixes as dialect) 14:58 fix: XSS escape error messages + 30s fetch timeout for dialect 14:56 remove: جديد badge from dialect panel 14:39 polish: remove 'قيد التطوير' badge, add 'جديد' badge + loading state on convert button 14:23 feat: dialect-to-MSA conversion — backend API + frontend + Docker model cache 13:14 fix: unified undo/redo - toolbar+keyboard use same stack, works with typing+formatting+Quran apply 13:07 fix: Ctrl+Q works with Arabic keyboard (e.code) + copy button for translations 13:00 feat: Amiri Quran font + copy button + Ctrl+Q shortcut 12:00 fix: use fullwidth brackets for ref delimiter to avoid [O Muhammad] collision 11:23 fix: quran text excluded from analysis models + styled ref + score preserved 11:16 feat: apply includes surah ref + quran-applied span protects from analysis models 10:54 fix: use correct editor ID (editor-container) - root cause of apply failures 10:43 fix: simplify apply to direct textContent replace (guaranteed to work) 10:39 fix: apply uses window.find() + add translation apply button 10:34 fix: robust apply using TreeWalker text search instead of fragile Range 10:23 fix: apply replaces text properly + translation shows ref + bigger font 10:16 feat: add 'Apply' button to replace editor text with verified Quran text 10:04 feat: Quran verification & translation endpoint + modal UI (13 languages) 09:42 UI: add Quran verse translation to 14 languages feature in home + features page 09:38 feat: add Quran verse translation to 14 languages in features page 09:27 UI: add Quran verse translation to 14 languages in feature cards 01:37 UI: swap nav order - pricing before editor (الأسعار then المحرر) 01:35 UI: remove pricing from nav, fix dialect colors to match theme, remove placeholder example 01:01 fix: grammar/quran icons different, tooltip no strikethrough, punctuation no duplicate dots 00:54 UI: nav logo, bayyinah hover, quran icon, hero autocomplete shadow, clean tooltip, fix punctuation dup + hamza كان 00:23 Full UI/UX modernization: fix hero examples, pricing beta, Apply All cursor/format fix, design polish Jun 20 23:34 fix: move Bayyinah section inside max-w-7xl container to prevent stretching 23:30 fix: replace Bayyinah icons with actual book-open-text logo SVG 23:25 fix: features page layout + update Bayyinah color to #14919b 23:19 feat: add Bayyinah RAG feature to home page + features page 23:13 fix: remove extra closing div causing gap above editor 23:08 update: replace Quran example with Yusuf:2 verse 23:05 feat: add Quran verification + Dialect conversion features (UI only, no backend) 22:52 feat: add Autocomplete feature to home page + features page 14:36 feat(NLP-4): AutoComplete integration — hybrid bigram + GPT-2 COMPLETELY INDEPENDENT from the correction pipeline. Zero interaction
with /api/analyze, StageLocker, OffsetMapper, PatchSet, or highlights.
Backend:
- src/nlp/autocomplete/ package (service, rules, init)
- HybridAutoComplete: lazy-loaded singleton, LRU cache, OOM fallback
- Bigram model from bayan10/AutoComplete (bigram_model_v4.pkl)
- GPT-2 from aubmindlab/aragpt2-base (optional, falls back to bigram)
- POST /api/autocomplete with {context, n} → {status, suggestions}
- /api/health now reports autocomplete status
Frontend:
- src/js/autocomplete.js — self-contained module
- Ghost text overlay (gray, semi-transparent, pointer-events:none)
- Dropdown with keyboard nav (↑/↓), TAB accept, ESC dismiss
- 400ms debounce, min 3 chars, cursor-aware context extraction
- CSS in components.css with dark/light theme support
59/59 existing pipeline tests passing