Create app.py

app.py

@@ -0,0 +1,194 @@
+import os
+import faiss
+import numpy as np
+import gradio as gr
+from pypdf import PdfReader
+from sentence_transformers import SentenceTransformer
+from groq import Groq
+
+# -----------------------------
+# Initialize Models
+# -----------------------------
+embedder = SentenceTransformer("all-MiniLM-L6-v2")
+
+# Safely load API key
+GROQ_API_KEY = os.getenv("Rag")
+client = Groq(api_key=GROQ_API_KEY) if GROQ_API_KEY else None
+
+# -----------------------------
+# Global Storage
+# -----------------------------
+index = None
+documents = []
+
+# -----------------------------
+# PDF Processing
+# -----------------------------
+def read_pdf(file):
+    try:
+        reader = PdfReader(file.name)  # FIX for Hugging Face
+        text = ""
+        for page in reader.pages:
+            if page.extract_text():
+                text += page.extract_text()
+        return text
+    except Exception as e:
+        return f"Error reading PDF: {str(e)}"
+
+
+def chunk_text(text, chunk_size=500, overlap=100):
+    chunks = []
+    start = 0
+
+    while start < len(text):
+        end = start + chunk_size
+        chunks.append(text[start:end])
+        start += chunk_size - overlap
+
+    return chunks
+
+
+# -----------------------------
+# Create FAISS Index
+# -----------------------------
+def create_index(chunks):
+    global index, documents
+
+    documents = chunks
+    embeddings = embedder.encode(chunks)
+
+    embeddings = np.array(embeddings).astype("float32")
+
+    dimension = embeddings.shape[1]
+    index = faiss.IndexFlatL2(dimension)
+    index.add(embeddings)
+
+
+# -----------------------------
+# Retrieval
+# -----------------------------
+def retrieve(query, k=3, threshold=1.2):
+    global index
+
+    if index is None:
+        return [], None
+
+    query_embedding = embedder.encode([query])
+    query_embedding = np.array(query_embedding).astype("float32")
+
+    distances, indices = index.search(query_embedding, k)
+
+    relevant_chunks = []
+    valid_distances = []
+
+    for i, dist in zip(indices[0], distances[0]):
+        if i < len(documents) and dist < threshold:
+            relevant_chunks.append(documents[i])
+            valid_distances.append(dist)
+
+    # Confidence
+    confidence = None
+    if valid_distances:
+        avg = np.mean(valid_distances)
+        if avg < 0.5:
+            confidence = "High"
+        elif avg < 1.0:
+            confidence = "Medium"
+        else:
+            confidence = "Low"
+
+    return relevant_chunks, confidence
+
+
+# -----------------------------
+# LLM (Groq)
+# -----------------------------
+def ask_groq(context_chunks, question):
+    if client is None:
+        return "Error: GROQ_API_KEY not set in Hugging Face Secrets."
+
+    context = "\n".join(context_chunks)
+
+    prompt = f"""
+You are an intelligent assistant.
+
+Rules:
+1. If answer is clearly in context → answer normally.
+2. If related but not exact → say:
+   "This is not explicitly mentioned in the document, but based on related context..."
+3. If irrelevant → say:
+   "The document does not contain information related to this question."
+
+Context:
+{context}
+
+Question:
+{question}
+"""
+
+    try:
+        response = client.chat.completions.create(
+            messages=[{"role": "user", "content": prompt}],
+            model="llama-3.3-70b-versatile",
+        )
+        return response.choices[0].message.content
+    except Exception as e:
+        return f"Groq API Error: {str(e)}"
+
+
+# -----------------------------
+# Main Functions
+# -----------------------------
+def process_pdf(file):
+    if file is None:
+        return "Please upload a PDF."
+
+    text = read_pdf(file)
+
+    if not text or "Error" in text:
+        return text
+
+    chunks = chunk_text(text)
+    create_index(chunks)
+
+    return f"✅ PDF processed! Chunks: {len(chunks)}"
+
+
+def answer_question(question):
+    if index is None:
+        return "Please upload and process a PDF first."
+
+    context_chunks, confidence = retrieve(question)
+
+    if not context_chunks:
+        return "The document does not contain information related to this question."
+
+    answer = ask_groq(context_chunks, question)
+
+    if confidence:
+        answer = f"(Confidence: {confidence})\n\n{answer}"
+
+    return answer
+
+
+# -----------------------------
+# Gradio UI
+# -----------------------------
+with gr.Blocks() as demo:
+    gr.Markdown("## 📄 RAG PDF Q&A (Groq + FAISS)")
+
+    file_input = gr.File(label="Upload PDF")
+    upload_btn = gr.Button("Process PDF")
+    status = gr.Textbox(label="Status")
+
+    question = gr.Textbox(label="Ask a question")
+    answer = gr.Textbox(label="Answer")
+
+    upload_btn.click(process_pdf, inputs=file_input, outputs=status)
+    question.submit(answer_question, inputs=question, outputs=answer)
+
+# -----------------------------
+# Launch
+# -----------------------------
+if __name__ == "__main__":
+    demo.launch(server_name="0.0.0.0", server_port=7860)