app.py

import os
import gradio as gr
import PyPDF2
import torch
from transformers import AutoTokenizer, AutoModel
from qdrant_client import QdrantClient
from qdrant_client.models import VectorParams, Distance, PointStruct
import cohere
from uuid import uuid4

# -------------------- CONFIG --------------------

# Qdrant Cloud config
QDRANT_URL = "https://9ecb3b08-e6fa-482c-a03f-e8cb3313a6c0.sa-east-1-0.aws.cloud.qdrant.io"   # e.g. "https://xxxxxx-xxxxx-xxxxx-xxxx-xxxxxxxxx.us-east.aws.cloud.qdrant.io:6333"
QDRANT_API_KEY = "<your-qdrant-api-key>"
COLLECTION_NAME = "Document"

# Cohere config
COHERE_API_KEY = "<your-cohere-api-key>"

# -------------------- INITIALIZE CLIENTS --------------------

qdrant = QdrantClient(
    url=QDRANT_URL,
    api_key=QDRANT_API_KEY,
)

cohere_client = cohere.Client(COHERE_API_KEY)

# -------------------- LOAD EMBEDDING MODEL --------------------

# Using MiniLM (384-dim)
tokenizer = AutoTokenizer.from_pretrained("sentence-transformers/all-MiniLM-L6-v2")
model = AutoModel.from_pretrained("sentence-transformers/all-MiniLM-L6-v2")

# -------------------- VECTOR COLLECTION SETUP --------------------

def setup_collection():
    """Create collection in Qdrant if it doesn't exist."""
    collections = qdrant.get_collections().collections
    existing_names = [c.name for c in collections]

    if COLLECTION_NAME not in existing_names:
        qdrant.create_collection(
            collection_name=COLLECTION_NAME,
            vectors_config=VectorParams(
                size=384,             # MiniLM-L6-v2 output size
                distance=Distance.COSINE
            )
        )

setup_collection()

# -------------------- UTILITY FUNCTIONS --------------------

def load_pdf(file):
    """Extract raw text from a PDF file."""
    # Gradio's File component passes a tempfile path or file object
    reader = PyPDF2.PdfReader(file)
    text = ""
    for page in reader.pages:
        page_text = page.extract_text()
        if page_text:
            text += page_text
    return text


def get_embeddings(text):
    """Compute mean-pooled embeddings using MiniLM."""
    inputs = tokenizer(
        text,
        return_tensors="pt",
        padding=True,
        truncation=True,
        max_length=512
    )
    with torch.no_grad():
        outputs = model(**inputs)
        # Mean pooling over sequence length
        embeddings = outputs.last_hidden_state.mean(dim=1).squeeze().cpu().numpy()
    return embeddings


def upload_document_chunks(chunks):
    """
    Insert document chunks into Qdrant as points.
    Each chunk becomes one point with a vector and payload.
    """
    points = []

    for chunk in chunks:
        try:
            embedding = get_embeddings(chunk)

            points.append(
                PointStruct(
                    id=str(uuid4()),          # unique ID per chunk
                    vector=embedding.tolist(),
                    payload={"content": chunk}
                )
            )
        except Exception as e:
            print(f"⚠️ Skipped chunk due to error: {e}")

    if points:
        qdrant.upsert(
            collection_name=COLLECTION_NAME,
            points=points
        )


def query_answer(query):
    """Search Qdrant for the most relevant chunks to the query."""
    query_embedding = get_embeddings(query)

    hits = qdrant.search(
        collection_name=COLLECTION_NAME,
        query_vector=query_embedding.tolist(),
        limit=3
    )
    return hits


def generate_response(context, query):
    """Use Cohere to generate a natural language answer based on context."""
    prompt = f"""You are a helpful assistant answering questions based only on the given context.

Context:
{context}

Question: {query}
Answer:"""

    response = cohere_client.generate(
        model="command",
        prompt=prompt,
        max_tokens=200,
        temperature=0.3
    )
    return response.generations[0].text.strip()


def qa_pipeline(pdf_file, query):
    """
    Full QA pipeline:
    1. Read PDF
    2. Chunk text
    3. Store chunks in Qdrant
    4. Search relevant chunks for query
    5. Generate answer via Cohere
    """
    if pdf_file is None:
        return "⚠️ Please upload a PDF first.", ""

    if not query or query.strip() == "":
        return "⚠️ Please enter a question.", ""

    # 1. Extract text
    document_text = load_pdf(pdf_file)

    if not document_text.strip():
        return "⚠️ No extractable text found in the PDF.", ""

    # 2. Simple character-based chunking
    chunk_size = 500
    document_chunks = [
        document_text[i:i + chunk_size]
        for i in range(0, len(document_text), chunk_size)
    ]

    # 3. Upload chunks to Qdrant
    upload_document_chunks(document_chunks)

    # 4. Search relevant chunks
    hits = query_answer(query)

    if not hits:
        return "⚠️ No relevant document segments found.", "I couldn't find an answer based on the document."

    context = " ".join([hit.payload.get("content", "") for hit in hits])

    # 5. Generate answer
    answer = generate_response(context, query)

    return context, answer


# -------------------- GRADIO UI --------------------

with gr.Blocks(theme="compact") as demo:
    gr.Markdown("""
        <div style="text-align: center; font-size: 28px; font-weight: bold; margin-bottom: 20px; color: #2D3748;">
            📄 Interactive PDF QA Bot (Qdrant + Cohere) 🔍
        </div>
        <p style="text-align: center; font-size: 16px; color: #4A5568;">
            Upload a PDF document, ask a question, and get answers grounded in the document content.
        </p>
        <hr style="border: 1px solid #CBD5E0; margin: 20px 0;">
    """)

    with gr.Row():
        with gr.Column(scale=1):
            pdf_input = gr.File(label="📁 Upload PDF", file_types=[".pdf"])
            query_input = gr.Textbox(
                label="❓ Ask a Question",
                placeholder="Enter your question here..."
            )
            submit_button = gr.Button("🔍 Submit")

        with gr.Column(scale=2):
            doc_segments_output = gr.Textbox(
                label="📜 Retrieved Document Segments",
                lines=10
            )
            answer_output = gr.Textbox(
                label="💬 Answer",
                lines=3
            )

    submit_button.click(
        fn=qa_pipeline,
        inputs=[pdf_input, query_input],
        outputs=[doc_segments_output, answer_output]
    )

    gr.Markdown("""
        <style>
            body {
                background-color: #EDF2F7;
            }
            input[type="file"] {
                background-color: #3182CE;
                color: white;
                padding: 8px;
                border-radius: 5px;
            }
            button {
                background-color: #3182CE;
                color: white;
                padding: 10px;
                font-size: 16px;
                border-radius: 5px;
                cursor: pointer;
                border: none;
            }
            button:hover {
                background-color: #2B6CB0;
            }
            textarea {
                border: 2px solid #CBD5E0;
                border-radius: 8px;
                padding: 10px;
                background-color: #FAFAFA;
            }
        </style>
    """)

demo.launch(share=True)