Boost scores when query grade matches standard title grade, penalize mismatches. Add part disambiguation to correctly route queries to specific standard parts (e.g., IS 12269 (Part 1) vs (Part 2)). Regenerate retrieval results with improved ranking.
Add Pass 3 to recover scope text incorrectly placed in previous block, and Pass 4 to truncate bleed from the following standard. Regenerate standards.json and standards_chunks.json with the improved parser.
- Add faiss-cpu, rank-bm25, sentence-transformers, numpy to requirements.txt.
(previously only pymupdf was listed; other deps were manual-install only)
- Cast score to float() before round() to avoid numpy type serialization errors.
- Pass expected_standards through _format_result for eval script compatibility.
- Update retrieval_results.json with expected_standards per query for eval.
- Add i18next + react-i18next + i18next-browser-languagedetector.
- EN/HI translation files covering all UI strings across every page and component.
- Language switcher button in Navbar; choice persisted to localStorage.
- document.documentElement.lang synced to active language in App.
- Skip-nav link and #main-content anchor for keyboard accessibility.
- aria-describedby on modal dialog; page title and meta description in index.html.
- Secure page title set to 'BIS SP-21 Standards.'
- Add helmet for secure HTTP response headers.
- Add express-rate-limit: 60 req/min general, 20 req/min on LLM endpoints.
- Restrict CORS to localhost origins in dev, CORS_ORIGIN env var in prod.
- Cap request body at 16kb.
- Add sanitizeText() to strip control chars on all string inputs.
- Add isValidStandardId() regex guard on :id param and standard_id fields.
- All route handlers use sanitized values; no raw req.body/req.query access.
Kshitij
notkshitij
atharvaombase