Spaces:
Runtime error
Runtime error
Commit History
-
Fix index
f36d1f9
-
Add clickable link for vector search results
c79a43d -
Update system flow and add default research papers for demo
✨ System Flow Updates: - Updated flow to include document upload and processing pipeline - Added vector index building and hybrid search steps - Integrated Modal distributed computing and Nebius AI - Enhanced with knowledge graph visualization - Updated technology stack references 🌱 Default Documents Seeding: - Added 7 important AI research papers for demo purposes - Includes: Attention Is All You Need, GPT-4, Constitutional AI, RAG, LangChain, RLHF, Emergent Abilities - Automatic seeding on first startup for populated knowledge graph - Papers include full abstracts, metadata, and author information - Enables immediate demo without requiring uploads 🎯 Demo Improvements: - Knowledge graph now populated by default in all environments - System flow reflects actual implemented functionality - Ready for hackathon demonstration with rich content 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>e57738d -
Fix Hugging Face Spaces deployment: Handle file permissions and production environment
🔧 Production File Permissions: - Use /tmp directory for uploads and database in production (NODE_ENV=production) - Add error handling for mkdir operations that fail due to permissions - Graceful fallback when directory creation fails 🚀 Hugging Face Spaces Compatibility: - Detect production environment and use writable /tmp paths - Added ENABLE_UPLOADS environment variable to control file upload functionality - Fallback routes for disabled uploads with helpful error messages 💾 Database & Storage Fixes: - SQLite database path: /tmp/knowledgebridge.db in production - Upload directory: /tmp/uploads in production vs ./uploads in development - Error handling for permission denied scenarios 🎯 Graceful Degradation: - App continues to work even when file operations fail - Clear error messages explaining disabled functionality - Traditional search and external sources still available 🔒 Environment Detection: - Automatic detection of production vs development environment - Optional ENABL…ccf1a85 -
Implement complete document upload and processing pipeline with Modal integration
✅ Real Document Upload: - Created SQLite storage replacing in-memory storage with full document metadata - Added file upload API with multer supporting PDFs, images, text files (50MB limit) - Implemented document processing pipeline with Modal integration for heavy workloads ✅ PDF/Image Processing Pipeline: - Modal integration for OCR text extraction using PyPDF2 and Tesseract - Batch processing capabilities for multiple documents concurrently - Document status tracking (pending -> processing -> completed/failed) ✅ Vector Embeddings Storage & Search: - Real embeddings generation using Nebius AI and storage in SQLite - Vector index building using Modal's FAISS implementation - High-performance vector search with relevance scoring ✅ Database Storage: - SQLite storage with proper schema for file metadata, processing status - Document lifecycle management with file path tracking - Search history and results persistence ✅ Batch Processing: - Modal distributed computing for processing la…10ac46e -
Improve result card: Remove empty additional context, enhance metadata
1. Remove 'Additional Context' section when empty (no more 'No additional context available') 2. Enhanced metadata display for academic papers: - Renamed to 'Publication Details' for academic papers - Authors prominently displayed with highlight background - Added emojis for better visual hierarchy (👥 Authors, 📅 Year, 🏛️ Venue, etc.) - Source-specific metadata (⭐ Stars for GitHub repos) - Better spacing and typography Much cleaner and more informative interface\!159f7bb -
Fix JSX syntax error in result card component
Fixed incomplete conditional expression after removing duplicate content section. Added fallback message for missing additional context.62ff453 -
Fix content truncation: Show full content instead of snippet
- Changed snippet preview to show full content with search highlighting - Added expandable preview with gradient fade for long content - Removed duplicate 'Full Content' section in expanded view - Users now see complete content immediately, not truncated snippets - Maintains clean UI with max-height and scroll for very long contentcdc4bc7 -
Make full content scrollable instead of truncated
- Replaced truncated content with full scrollable view - Added max-height (384px) with overflow-y-auto for long content - Better UX for knowledge base - users can read complete documents - Maintains clean layout while providing full access to content0e28fdd -
Remove hardcoded Replit banner script from HTML
The Replit dev banner script was hardcoded in index.html causing CSP violations in production. Removed it completely as it's not needed for Hugging Face Spaces deployment.602fcb0 -
DEFINITIVE FIX: Resolve Docker build issues for Hugging Face Spaces
Root cause: .gitignore was excluding client/src/lib/ directory with overly broad "lib/" pattern meant for Python packages. Solution: 1. Fixed .gitignore to specifically exclude Python lib dirs only 2. Added .dockerignore with explicit inclusion of client/src/lib 3. Reverted import to clean @/lib/queryClient format 4. Ensured robust file copying for containerized environments This addresses all Docker/container path resolution issues. 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>b65e9ca -
Add .ts extension to queryClient import for Docker compatibility
The Docker build environment requires explicit .ts extension for the path alias resolution to work correctly. 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>f61cf0b -
Fix Docker build by using path alias for queryClient import
Changed import from "./lib/queryClient" to "@/lib/queryClient" to use the configured path alias which resolves correctly in the Docker build environment. 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>3df58dc -
initial commit
7c012de