Merge pull request #113 from AOSSIE-Org/feat/chat

Feat/ RAG chat endpoint + Pinecone metadata fix

Author: ManavSarkar

.github/workflows/deploy-backend-to-hf.yml

@@ -3,57 +3,67 @@ name: 🚀 Deploy Backend to HF Space
 on:
   push:
     branches:
-      - main # or your primary branch
+      - main
     paths:
-      - "backend/**" # only trigger when anything under backend/ changes
+      - "backend/**"
 
 jobs:
   deploy:
     runs-on: ubuntu-latest
+    # set your HF username here (or replace with a secret if you prefer)
+    env:
+      HF_USER: Thunder1245
+      HF_REPO: perspective-backend
+
     steps:
       - name: 👉 Checkout code
         uses: actions/checkout@v4
         with:
           fetch-depth: 0
 
-      - name: 🔒 Install HF CLI
-        run: pip install huggingface_hub
-
-      - name: 🔑 HF login
-        env:
-          HF_TOKEN: ${{ secrets.HF_TOKEN }}
-        run: huggingface-cli login --token "$HF_TOKEN"
+      - name: 🔍 Ensure HF_TOKEN is set
+        run: |
+          if [ -z "${{ secrets.HF_TOKEN }}" ]; then
+            echo "ERROR: HF_TOKEN secret is not set. Add it in repository secrets: Settings → Secrets & variables → Actions."
+            exit 1
+          fi
 
-      - name: 📂 Prepare Space repo
+      - name: 📂 Prepare Space repo (clone)
         env:
           HF_TOKEN: ${{ secrets.HF_TOKEN }}
         run: |
-          rm -rf space-backend
+          rm -rf space-backend || true
+          # clone using token in URL (this authenticates the clone)
           git clone https://Thunder1245:${HF_TOKEN}@huggingface.co/spaces/Thunder1245/perspective-backend.git space-backend
 
       - name: 📦 Install rsync
         run: |
           sudo apt-get update
           sudo apt-get install -y rsync
 
-      - name: 📤 Sync backend code
+      - name: 📤 Sync backend code to Space
         env:
           HF_TOKEN: ${{ secrets.HF_TOKEN }}
         run: |
+          set -e
+
           cd space-backend
 
-          # Only remove tracked files (preserve .git and config)
+          # Remove tracked files while preserving .git and config (ignore failure)
           git rm -r . || true
           cd ..
 
-          # Copy new backend files in
+          # Copy backend files into the cloned space directory
           cp -R backend/. space-backend/
 
-          # Push new code to HF Space
+          # Commit & push
           cd space-backend
           git config user.name "github-actions[bot]"
           git config user.email "github-actions[bot]@users.noreply.github.com"
           git add --all
-          git commit -m "Auto‑deploy backend: ${{ github.sha }}" || echo "No changes to commit"
+          git commit -m "Auto-deploy backend: ${{ github.sha }}" || echo "No changes to commit"
           git push origin main
 
+      - name: ✅ Done
+        run: |
+          echo "Backend deployed to Hugging Face Space: https://huggingface.co/spaces/${HF_USER}/${HF_REPO}"

README.md

@@ -2,15 +2,32 @@
 ![Perspective banner](frontend/public/perspective_banner.jpg)
 
 ### Table of Contents
-- [System Overview](#system-overview)
-- [Architecture Components](#architecture-components)
-- [Technical Stack](#technical-stack)
-- [Core Features](#core-features)
-- [Data Flow & Security](#data-flow--security)
-- [Setup & Deployment](#setup--deployment)
-- [Detailed Architecture Diagram](#detailed-architecture-diagram)
-- [Expected Outcomes](#expected-outcomes)
-- [Required Skills](#required-skills)
+- [Perspective-AI](#perspective-ai)
+    - [Table of Contents](#table-of-contents)
+  - [System Overview](#system-overview)
+    - [High-Level Concept](#high-level-concept)
+  - [Architecture Components](#architecture-components)
+    - [1. Frontend Layer](#1-frontend-layer)
+    - [3. Core Backend](#3-core-backend)
+    - [4. AI \& NLP Integration](#4-ai--nlp-integration)
+    - [5. Data Storage](#5-data-storage)
+  - [Technical Stack](#technical-stack)
+    - [Frontend Technologies](#frontend-technologies)
+    - [Backend Technologies](#backend-technologies)
+    - [I Integration](#i-integration)
+  - [Core Features](#core-features)
+    - [1. Counter-Perspective Generation](#1-counter-perspective-generation)
+    - [2. Reasoned Thinking](#2-reasoned-thinking)
+    - [3. Updated Facts](#3-updated-facts)
+    - [4. Seamless Integration](#4-seamless-integration)
+    - [5. Real-Time Analysis](#5-real-time-analysis)
+  - [Data Flow \& Security](#data-flow--security)
+  - [Setup \& Deployment](#setup--deployment)
+    - [Frontend Setup](#frontend-setup)
+    - [Backend Setup](#backend-setup)
+  - [Architecture Diagram](#architecture-diagram)
+  - [Expected Outcomes](#expected-outcomes)
+  - [Required Skills](#required-skills)
 
 ---
 
@@ -137,20 +154,25 @@ npm run dev
   - add .env file in `/new-backend`directory.
   - add following environment variable in your .env file.
   ```
-  HF_TOKEN = <Your_hugging_face_access_token>
+  GROQ_API_KEY= <groq_api_key>
+PINECONE_API_KEY = <your_pinecone_API_KEY>
+PORT = 8000
+SEARCH_KEY = <your_Google_custom_search_engine_API_key>
   ```
 
 *Run backend:*
 ```bash
-cd new-backend
+cd backend
 uv sync # Creating virtual environment at: .venv
 uv run main.py #Runs the backend server
 ```
 
 ---
 
+
 ## Architecture Diagram
 
+
 ```mermaid
 graph TB
     %% Define Subgraphs with Colors and Text Styles
@@ -168,6 +190,7 @@ graph TB
         Analyzer[Content Analyzer]
         CNEngine[Counter-Narrative Engine]
         Context[Context Manager]
+
     end
 
     subgraph AI & NLP Layer
@@ -212,7 +235,7 @@ graph TB
 
 ## Required Skills
 
-- **Frontend Development**: Experience with Next.js and modern UI frameworks.
+- **Frontend Development**:  Experience with Next.js and modern UI frameworks.
 - **Backend Development**: Proficiency in Python and FastAPI.
 - **AI & NLP**: Familiarity with LangChain, Langgraph, and prompt engineering techniques.
 - **Database Management**: Knowledge of vector databases system.

backend/app/modules/chat/__init__.py

@@ -0,0 +1,10 @@
+from sentence_transformers import SentenceTransformer
+
+embedder = SentenceTransformer("all-MiniLM-L6-v2")
+
+
+def embed_query(query: str):
+
+    embeddings = embedder.encode(query).tolist()
+
+    return embeddings

backend/app/modules/chat/embed_query.py

@@ -0,0 +1,31 @@
+from pinecone import Pinecone
+from dotenv import load_dotenv
+from app.modules.chat.embed_query import embed_query
+import os
+
+load_dotenv()
+
+pc = Pinecone(os.getenv("PINECONE_API_KEY"))
+index = pc.Index("perspective")
+
+
+def search_pinecone(query: str, top_k: int = 5):
+
+    embeddings = embed_query(query)
+
+    results = index.query(
+        vector=embeddings,
+        top_k=top_k,
+        include_metadata=True,
+        namespace="default"
+
+    )
+
+    matches = []
+    for match in results["matches"]:
+        matches.append({
+            "id": match["id"],
+            "score": match["score"],
+            "metadata": match["metadata"]
+        })
+    return matches

backend/app/modules/chat/get_rag_data.py

@@ -0,0 +1,35 @@
+import os
+from groq import Groq
+from dotenv import load_dotenv
+
+load_dotenv()
+
+client = Groq(api_key=os.getenv("GROQ_API_KEY"))
+
+
+def build_context(docs):
+
+    return "\n".join(f"{m['metadata'].get('explanation') or m['metadata'].get('reasoning', '')}"for m in docs)
+
+
+def ask_llm(question, docs):
+    context = build_context(docs)
+    print(context)
+    prompt = f"""You are an assistant that answers based on context.
+
+Context:
+{context}
+
+Question:
+{question}
+"""
+
+    response = client.chat.completions.create(
+        model="gemma2-9b-it",
+        messages=[
+            {"role": "system", "content": "Use only the context to answer."},
+            {"role": "user", "content": prompt}
+        ]
+    )
+
+    return response.choices[0].message.content

backend/app/modules/chat/llm_processing.py

@@ -28,3 +28,4 @@ def embed_chunks(chunks: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
             "metadata": chunk["metadata"]
         })
     return vectors
+

backend/app/modules/vector_store/embed.py

@@ -3,31 +3,49 @@
 from app.modules.pipeline import run_scraper_pipeline
 from app.modules.pipeline import run_langgraph_workflow
 from app.modules.bias_detection.check_bias import check_bias
+from app.modules.chat.get_rag_data import search_pinecone
+from app.modules.chat.llm_processing import ask_llm
 import asyncio
 import json
 
 router = APIRouter()
 
+
 class URlRequest(BaseModel):
     url: str
 
 
+class ChatQuery(BaseModel):
+    message: str
+
+
 @router.get("/")
 async def home():
     return {"message": "Perspective API is live!"}
 
+
 @router.post("/bias")
 async def bias_detection(request: URlRequest):
-    content = await asyncio.to_thread(run_scraper_pipeline,(request.url))
-    bias_score = await asyncio.to_thread(check_bias,(content))
+    content = await asyncio.to_thread(run_scraper_pipeline, (request.url))
+    bias_score = await asyncio.to_thread(check_bias, (content))
     print(bias_score)
     return bias_score
-    
 
 
 @router.post("/process")
 async def run_pipelines(request: URlRequest):
-    article_text = await asyncio.to_thread(run_scraper_pipeline,(request.url))
+    article_text = await asyncio.to_thread(run_scraper_pipeline, (request.url))
     print(json.dumps(article_text, indent=2))
-    data = await asyncio.to_thread(run_langgraph_workflow,(article_text))
+    data = await asyncio.to_thread(run_langgraph_workflow, (article_text))
     return data
+
+
+@router.post("/chat")
+async def answer_query(request: ChatQuery):
+
+    query = request.message
+    results = search_pinecone(query)
+    answer = ask_llm(query, results)
+    print(answer)
+
+    return {"answer": answer}

backend/app/routes/routes.py

@@ -20,6 +20,7 @@ import {
 import { Input } from "@/components/ui/input";
 import { Badge } from "@/components/ui/badge";
 import BiasMeter from "@/components/bias-meter";
+import axios from "axios";
 
 
 /**
@@ -79,23 +80,24 @@ export default function AnalyzePage() {
 
   }, [router]);
 
-  const handleSendMessage = (e: React.FormEvent) => {
+  async function handleSendMessage(e: React.FormEvent){
     e.preventDefault();
     if (!message.trim()) return;
     const newMessages = [...messages, { role: "user", content: message }];
     setMessages(newMessages);
     setMessage("");
-    setTimeout(() => {
-      setMessages([
-        ...newMessages,
-        {
-          role: "system",
-          content:
-            "Based on the article... let me know if you want more details.",
-        },
-      ]);
-    }, 1000);
-  };
+
+  const res = await axios.post("http://Thunder1245-perspective-backend.hf.space/api/chat", {
+    message: message
+    });
+  const data = res.data;
+
+  console.log(data)
+
+  // 🔹 Step 2: Append LLM’s response
+  setMessages([...newMessages, { role: "assistant", content: data.answer }]);
+};
+
 
   if (isLoading || !analysisData || !biasScore) {
     return (
@@ -113,7 +115,7 @@ export default function AnalyzePage() {
     score,
   } = analysisData;
 
-  return (
+  return(
     <div className="flex flex-col min-h-screen">
       {/* Header omitted for brevity */}
       <main className="flex-1 pt-16 container mx-auto px-4">