app.py

import os
import uuid
import chromadb
import gradio as gr
from pypdf import PdfReader
import docx
from sentence_transformers import SentenceTransformer
from groq import Groq

# =========================
# 🔑 GROQ API (HF SECRET)
# =========================
# Set your secret as "GROQ_API_KEY" in HF Space Settings → Variables and secrets
groq_client = Groq(api_key=os.getenv("Multi_doc"))

# =========================
# 📄 LOAD DOCUMENTS
# =========================
def load_pdf(path):
    reader = PdfReader(path)
    return "\n".join([p.extract_text() or "" for p in reader.pages])

def load_docx(path):
    doc = docx.Document(path)
    return "\n".join([p.text for p in doc.paragraphs])

def load_txt(path):
    with open(path, "r", encoding="utf-8") as f:
        return f.read()

def load_document(path):
    ext = path.split(".")[-1].lower()
    if ext == "pdf":
        return load_pdf(path)
    if ext == "docx":
        return load_docx(path)
    if ext == "txt":
        return load_txt(path)
    raise ValueError(f"Unsupported file type: .{ext}")

# =========================
# ✂️ CHUNKING
# =========================
def chunk_text(text, size=400, overlap=80):
    words = text.split()
    chunks = []
    i = 0
    cid = 0

    while i < len(words):
        chunks.append({
            "id": cid,
            "text": " ".join(words[i:i + size])
        })
        i += size - overlap
        cid += 1

    return chunks

# =========================
# 🧠 EMBEDDINGS (LOCAL)
# =========================
embed_model = SentenceTransformer("all-MiniLM-L6-v2")

def embed(texts):
    return embed_model.encode(texts, show_progress_bar=False).tolist()

# =========================
# 🗄️ CHROMA DB
# HF Spaces has a read-only root — use /tmp for writable storage
# =========================
chroma_client = chromadb.PersistentClient(path="/tmp/chroma_db")
collection = chroma_client.get_or_create_collection("rag")

# =========================
# 📁 PROCESS FILES
# =========================
def process_files(files):
    if not files:
        return "⚠️ No files uploaded."

    all_chunks = []
    errors = []

    for f in files:
        # Gradio on HF passes file path as a string or NamedString
        file_path = f if isinstance(f, str) else f.name
        if not file_path:
            continue
        try:
            text = load_document(file_path)
            if not text.strip():
                errors.append(f"⚠️ {os.path.basename(file_path)} appears empty.")
                continue
            chunks = chunk_text(text)
            for c in chunks:
                all_chunks.append({
                    "source": os.path.basename(file_path),
                    "text": c["text"]
                })
        except Exception as e:
            errors.append(f"❌ Error reading {os.path.basename(file_path)}: {e}")

    if not all_chunks:
        return "\n".join(errors) if errors else "⚠️ No content could be extracted."

    texts = [c["text"] for c in all_chunks]
    embeddings = embed(texts)

    collection.add(
        ids=[str(uuid.uuid4()) for _ in all_chunks],
        embeddings=embeddings,
        documents=texts,
        metadatas=[{"source": c["source"]} for c in all_chunks]
    )

    result = f"✅ Indexed {len(files)} file(s) — {len(all_chunks)} chunks stored."
    if errors:
        result += "\n" + "\n".join(errors)
    return result

# =========================
# 🔍 RETRIEVAL
# =========================
def retrieve(query, k=3):
    # Guard: collection might be empty
    count = collection.count()
    if count == 0:
        return []

    k = min(k, count)  # Can't retrieve more than what's stored
    q_emb = embed([query])[0]

    results = collection.query(
        query_embeddings=[q_emb],
        n_results=k
    )

    docs = []
    for i in range(len(results["documents"][0])):
        docs.append({
            "text": results["documents"][0][i],
            "source": results["metadatas"][0][i]["source"]
        })

    return docs

# =========================
# 🤖 GROQ GENERATION
# =========================
def generate(query):
    docs = retrieve(query)

    if not docs:
        return "⚠️ No documents indexed yet. Please upload and process files first."

    context = "\n\n".join(
        [f"[{d['source']}]\n{d['text']}" for d in docs]
    )

    prompt = f"""You are a strict RAG assistant.
Answer ONLY from the context below.
If the answer is not found in the context, say: "Not found in documents."

CONTEXT:
{context}

QUESTION:
{query}

ANSWER:"""

    try:
        response = groq_client.chat.completions.create(
            model="llama-3.1-8b-instant",
            messages=[{"role": "user", "content": prompt}],
            temperature=0.2,
            max_tokens=1024,
        )
        answer = response.choices[0].message.content
    except Exception as e:
        return f"❌ Groq API error: {e}"

    sources = "\n\n".join(
        [f"📄 **{d['source']}**\n{d['text'][:200]}…" for d in docs]
    )

    return f"{answer}\n\n---\n📚 **Sources:**\n{sources}"

# =========================
# 💬 CHAT FUNCTION
# Gradio 5 uses {"role": ..., "content": ...} dicts, not tuples
# =========================
def chat(message, history):
    if not message.strip():
        return "", history
    reply = generate(message)
    history.append({"role": "user", "content": message})
    history.append({"role": "assistant", "content": reply})
    return "", history

# =========================
# 🎨 GRADIO UI
# =========================
with gr.Blocks(title="Groq RAG Assistant") as app:

    gr.Markdown(
        """# 🧠 Groq RAG Assistant
        Upload your documents, then ask questions about them.
        Powered by **Groq LLaMA3** + **ChromaDB** + **sentence-transformers**.
        """
    )

    with gr.Row():

        with gr.Column(scale=1):
            gr.Markdown("### 📂 Upload Documents")
            files = gr.File(
                file_count="multiple",
                file_types=[".pdf", ".docx", ".txt"],
                label="Upload PDF / DOCX / TXT"
            )
            process_btn = gr.Button("🚀 Process Files", variant="primary")
            status = gr.Textbox(label="Status", interactive=False)

            process_btn.click(fn=process_files, inputs=files, outputs=status)

        with gr.Column(scale=2):
            gr.Markdown("### 💬 Ask Your Documents")
            # Gradio 5: type="messages" uses the new dict format
            chatbot = gr.Chatbot(height=480, type="messages")
            msg = gr.Textbox(
                placeholder="Ask a question about your documents…",
                label="Your question",
                lines=2
            )
            with gr.Row():
                submit_btn = gr.Button("Send", variant="primary")
                clear_btn = gr.Button("Clear Chat")

            submit_btn.click(fn=chat, inputs=[msg, chatbot], outputs=[msg, chatbot])
            msg.submit(fn=chat, inputs=[msg, chatbot], outputs=[msg, chatbot])
            clear_btn.click(fn=lambda: ([], ""), outputs=[chatbot, msg])

# =========================
# 🚀 LAUNCH
# =========================
if __name__ == "__main__":
    app.launch()