Deploy from GitHub Actions: 27118b812764ca9d2a18f390dbfd0affb5b9029d

app/config.py

@@ -20,8 +20,8 @@ EMBED_MODEL = "models/gemini-embedding-001"
 CHAT_MODEL = "gemini-2.5-flash-lite"
 
 TOP_K = 10
-CHUNK_SIZE = 1500
-CHUNK_OVERLAP = 200
+CHUNK_SIZE = 800
+CHUNK_OVERLAP = 100
 UPLOAD_BATCH_SIZE = 100
 
 MAX_FILE_COUNT = 6
@@ -49,6 +49,6 @@ PROMPT = (
 )
 
 CREATORS = [
-    {"name": "Krishnendu Das", "url": "https://itskdhere.com"},
-    {"name": "Saptarshi Roy", "url": "https://hirishi.in"}
+    {"Krishnendu Das" : "https://itskdhere.com"},
+    {"Saptarshi Roy" : "https://hirishi.in"}
 ]

app/deps.py

@@ -0,0 +1,6 @@
+from fastapi import Header, HTTPException
+
+async def get_user_id(x_user_id: str = Header(None)) -> str:
+    if not x_user_id:
+        raise HTTPException(status_code=401, detail="User ID required")
+    return x_user_id

app/main.py

@@ -1,3 +1,4 @@
+from app.routes.auth import router as auth_router
 from app.routes.ask import router as ask_router
 from app.routes.delete import router as delete_router
 from app.routes.clear import router as clear_router
@@ -27,6 +28,7 @@ app.add_middleware(
     allow_headers=["*"],
 )
 
+app.include_router(auth_router)
 app.include_router(ask_router)
 app.include_router(upload_router)
 app.include_router(delete_router)

app/rag/loader.py

@@ -2,19 +2,19 @@ from langchain_community.document_loaders import (
     CSVLoader,
     Docx2txtLoader,
     JSONLoader,
-    PDFPlumberLoader,
     TextLoader,
     UnstructuredExcelLoader,
     UnstructuredMarkdownLoader,
     UnstructuredPowerPointLoader,
 )
+from langchain_pymupdf4llm import PyMuPDF4LLMLoader
 from langchain_core.documents import Document
 
 
 # PDF
-# https://python.langchain.com/docs/integrations/document_loaders/pdfplumber
+# https://docs.langchain.com/oss/python/integrations/document_loaders/pymupdf4llm
 def read_pdf(path: str) -> list[Document]:
-    loader = PDFPlumberLoader(path)
+    loader = PyMuPDF4LLMLoader(path)
     docs = loader.load()
     return docs
 

app/rag/vectorstore.py

@@ -52,22 +52,27 @@ def delete_vectorstore(session_id: str) -> bool:
         return False
 
 
-def delete_all_vectorstores() -> bool:
+def delete_user_vectorstores(user_id: str) -> bool:
     try:
         index = _get_index()
         stats = index.describe_index_stats()
         namespaces = list(stats.namespaces.keys())
         failed: list[str] = []
-        for ns in namespaces:
+        prefix = f"{user_id}_"
+        
+        target_namespaces = [ns for ns in namespaces if ns.startswith(prefix)]
+        
+        for ns in target_namespaces:
             try:
                 index.delete(delete_all=True, namespace=ns)
             except Exception as e:
                 print(f"Failed to delete namespace '{ns}': {e}")
                 failed.append(ns)
+        
         if failed:
-            print(f"delete_all_vectorstores: {len(failed)}/{len(namespaces)} namespaces failed: {failed}")
+            print(f"delete_user_vectorstores: {len(failed)}/{len(target_namespaces)} namespaces failed: {failed}")
             return False
         return True
     except Exception as e:
-        print(f"delete_all_vectorstores: unexpected error: {e}")
+        print(f"delete_user_vectorstores: unexpected error: {e}")
         return False

app/routes/ask.py

@@ -1,6 +1,7 @@
 from app.config import CHAT_MODEL, GOOGLE_API_KEY, PROMPT, TOP_K
 from app.rag.vectorstore import get_vectorstore
-from fastapi import APIRouter, HTTPException
+from fastapi import APIRouter, HTTPException, Depends
+from app.deps import get_user_id
 from langchain_core.messages import HumanMessage
 from langchain_google_genai import ChatGoogleGenerativeAI
 from pydantic import BaseModel
@@ -20,8 +21,9 @@ class AskResponse(BaseModel):
 
 
 @router.post("/ask")
-async def ask(body: AskRequest) -> AskResponse:
-    store = get_vectorstore(body.session_id)
+async def ask(body: AskRequest, user_id: str = Depends(get_user_id)) -> AskResponse:
+    prefixed_session_id = f"{user_id}_{body.session_id}"
+    store = get_vectorstore(prefixed_session_id)
     docs = store.similarity_search(body.question, k=TOP_K)
 
     if not docs:

app/routes/auth.py

@@ -0,0 +1,8 @@
+import uuid
+from fastapi import APIRouter
+
+router = APIRouter(prefix="/auth", tags=["auth"])
+
+@router.get("/guest")
+async def create_guest_id() -> dict:
+    return {"user_id": f"guest_{uuid.uuid4().hex[:12]}"}

app/routes/chats.py

@@ -1,28 +1,36 @@
 import re
-from datetime import datetime
+from datetime import datetime, timezone
 from app.rag.vectorstore import _get_index
-from fastapi import APIRouter, HTTPException
+from fastapi import APIRouter, HTTPException, Depends
+from app.deps import get_user_id
 
 router = APIRouter()
 
 
 @router.get("/chats")
-async def get_chats() -> dict:
+async def get_chats(user_id: str = Depends(get_user_id)) -> dict:
     try:
         index = _get_index()
         stats = index.describe_index_stats()
 
         namespaces = list(stats.namespaces.keys())
         chats = []
+        prefix = f"{user_id}_"
+        
         for ns in namespaces:
-            if not re.fullmatch(r'\d+', ns):
+            if not ns.startswith(prefix):
+                continue
+            
+            session_id = ns[len(prefix):]
+            
+            if not re.fullmatch(r'\d+', session_id):
                 continue
 
-            timestamp = int(ns) / 1000.0
-            dt = datetime.fromtimestamp(timestamp)
+            timestamp = int(session_id) / 1000.0
+            dt = datetime.fromtimestamp(timestamp, tz=timezone.utc)
 
             chats.append({
-                "id": ns,
+                "id": session_id,
                 "title": f"Analysis {dt.strftime('%H:%M:%S')}",
                 "date": dt.strftime('%Y-%m-%d')
             })
@@ -30,5 +38,5 @@ async def get_chats() -> dict:
         chats.sort(key=lambda x: int(x["id"]), reverse=True)
         return {"chats": chats}
     except Exception as e:
-        print(f"Error fetching chats: {e}")
+        print(f"Error fetching chats for user {user_id}: {e}")
         raise HTTPException(500, "Failed to fetch chats from Pinecone.")

app/routes/clear.py

@@ -1,11 +1,12 @@
-from app.rag.vectorstore import delete_all_vectorstores
-from fastapi import APIRouter, HTTPException
+from app.rag.vectorstore import delete_user_vectorstores
+from fastapi import APIRouter, HTTPException, Depends
+from app.deps import get_user_id
 
 router = APIRouter()
 
 
 @router.delete("/clear")
-async def clear_index() -> dict:
-    if not delete_all_vectorstores():
-        raise HTTPException(500, "Failed to clear the vector store.")
-    return {"message": "All vector stores cleared."}
+async def clear_index(user_id: str = Depends(get_user_id)) -> dict:
+    if not delete_user_vectorstores(user_id):
+        raise HTTPException(500, "Failed to clear the user's vector stores.")
+    return {"message": "User's vector stores cleared."}

app/routes/delete.py

@@ -1,11 +1,13 @@
 from app.rag.vectorstore import delete_vectorstore
-from fastapi import APIRouter, HTTPException
+from fastapi import APIRouter, HTTPException, Depends
+from app.deps import get_user_id
 
 router = APIRouter()
 
 
 @router.delete("/delete/{session_id}")
-async def delete_specific_chat(session_id: str) -> dict:
-    if not delete_vectorstore(session_id):
+async def delete_specific_chat(session_id: str, user_id: str = Depends(get_user_id)) -> dict:
+    prefixed_session_id = f"{user_id}_{session_id}"
+    if not delete_vectorstore(prefixed_session_id):
         raise HTTPException(404, f"No vector store found for session: {session_id}")
     return {"message": f"Vector store for session {session_id} deleted."}

app/routes/upload.py

@@ -1,13 +1,19 @@
 import os
 from app.config import ALLOWED_TYPES, MAX_FILE_COUNT, MAX_FILE_SIZE, UPLOAD_DIR
 from app.rag.pipeline import process_file
-from fastapi import APIRouter, File, Form, HTTPException, UploadFile
+from fastapi import APIRouter, File, Form, HTTPException, UploadFile, Depends
+from app.deps import get_user_id
 
 router = APIRouter()
 
 
 @router.post("/upload")
-async def upload_files(files: list[UploadFile] = File(...), session_id: str = Form(...)) -> dict:
+async def upload_files(
+    files: list[UploadFile] = File(...), 
+    session_id: str = Form(...),
+    user_id: str = Depends(get_user_id)
+) -> dict:
+    prefixed_session_id = f"{user_id}_{session_id}"
     results = []
     errors = []
     total_chunks = 0
@@ -39,7 +45,7 @@ async def upload_files(files: list[UploadFile] = File(...), session_id: str = Fo
         try:
             with open(path, "wb") as f:
                 f.write(content)
-            chunks = process_file(path, ext, session_id=session_id)
+            chunks = process_file(path, ext, session_id=prefixed_session_id)
             total_chunks += chunks
             results.append({"source": original_name, "chunks": chunks})
         except Exception as e:

requirements.txt

@@ -1,6 +1,7 @@
 fastapi
 uvicorn[standard]
 python-multipart
+python-dotenv
 
 langchain-community
 langchain-core
@@ -12,12 +13,10 @@ langchain-pinecone
 
 google-generativeai
 
-pdfplumber
+langchain-pymupdf4llm
 docx2txt
 openpyxl
 python-pptx
 unstructured
 markdown
 jq
-
-python-dotenv