Chat with a PDF file

• This demo application uses OpenAI, LangChain.js, Supabase and Cloudflare R2 to allow you to chat with a PDF.
• PDF is uploaded into Cloudflare R2, then split into chunks and embedded in Supabase vector storage.
• The main frontend logic is found in the app/chat/pdf folder.
• The main configuration file can be found in app/pdf/toolConfig.ts file. You can specify the GPT model used & number messages included in the history.
• The API endpoint and logic can be found in app/api/pdf/chat/route.ts.
• The API to generate embeddings be found in app/api/pdf/vectorize/route.ts.
• The API to upload PDF can be found in app/api/pdf/upload/route.ts.
• The API to delete PDF & embeddings can be found in app/api/pdf/delete/route.ts.
• Hope you like it - give it a try!