app.py

import gradio as gr
import uuid
import base64
import io
from qdrant_client import QdrantClient
from qdrant_client.models import PointStruct, VectorParams, Distance
from sentence_transformers import SentenceTransformer
from PIL import Image

# --------------------------
# Qdrant Cloud Connection
# --------------------------
QDRANT_URL = "https://ff4da494-27b1-413c-ba58-d5ea14932fe1.europe-west3-0.gcp.cloud.qdrant.io:6333"
QDRANT_API_KEY = "eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJhY2Nlc3MiOiJtIn0.jjeB1JgnUSlb1hOOKMdRpVvMrUER57-udT-X1AWXT1E"
COLLECTION_NAME = "lost_and_found"

# CLIP model (text + image embeddings)
MODEL_NAME = "sentence-transformers/clip-ViT-B-32"
embedder = SentenceTransformer(MODEL_NAME)

# CLIP ViT-B/32 always gives 512-dimensional embeddings
VECTOR_SIZE = 512

# Qdrant Client (Cloud)
qclient = QdrantClient(
    url=QDRANT_URL,
    api_key=QDRANT_API_KEY
)

# Ensure collection exists
qclient.recreate_collection(
    collection_name=COLLECTION_NAME,
    vectors_config=VectorParams(size=VECTOR_SIZE, distance=Distance.COSINE),
)

# --------------------------
# Helper Functions
# --------------------------

def image_to_base64(img: Image.Image) -> str:
    buf = io.BytesIO()
    img.save(buf, format="PNG")
    return base64.b64encode(buf.getvalue()).decode("utf-8")

def base64_to_image(b64_str: str) -> Image.Image:
    img_bytes = base64.b64decode(b64_str)
    return Image.open(io.BytesIO(img_bytes))

def embed_text(text: str):
    return embedder.encode(text).tolist()

def embed_image(img: Image.Image):
    return embedder.encode(img).tolist()

def add_item(image, description, finder_name, finder_phone):
    if image is None or description.strip() == "":
        return "Please provide both an image and a description."

    embedding = embed_image(image)
    img_b64 = image_to_base64(image)

    metadata = {
        "description": description,
        "finder_name": finder_name if finder_name.strip() else "NA",
        "finder_phone": finder_phone if finder_phone.strip() else "NA",
        "image_b64": img_b64
    }

    qclient.upsert(
        collection_name=COLLECTION_NAME,
        points=[
            PointStruct(
                id=str(uuid.uuid4()),
                vector=embedding,
                payload=metadata
            )
        ]
    )
    return "Item successfully added!"

def search_items(query_text, query_image):
    if not query_text and query_image is None:
        return "Please enter text or upload an image to search.", []

    if query_image:
        query_vector = embed_image(query_image)
    else:
        query_vector = embed_text(query_text)

    results = qclient.search(
        collection_name=COLLECTION_NAME,
        query_vector=query_vector,
        limit=5
    )

    if not results:
        return "No matches found.", []

    gallery = []
    output_text = "### Matches Found\n\n"
    for r in results:
        desc = r.payload.get("description", "No description")
        name = r.payload.get("finder_name", "NA")
        phone = r.payload.get("finder_phone", "NA")
        output_text += f"- **{desc}** — Finder: {name}, Phone: {phone}\n"

        if "image_b64" in r.payload:
            try:
                img = base64_to_image(r.payload["image_b64"])
                gallery.append(img)
            except Exception:
                pass

    return output_text, gallery

def clear_database():
    qclient.delete_collection(COLLECTION_NAME)
    qclient.recreate_collection(
        collection_name=COLLECTION_NAME,
        vectors_config=VectorParams(size=VECTOR_SIZE, distance=Distance.COSINE),
    )
    return "Database cleared successfully."

# --------------------------
# Gradio UI
# --------------------------

with gr.Blocks() as demo:
    gr.Markdown(
        """
        # Lost & Found System  
        A simple platform to add and search lost items.  
        """
    )

    with gr.Tab("Add Item"):
        gr.Markdown("### Add a Found Item")
        with gr.Row():
            image_in = gr.Image(type="pil", label="Upload Image")
            desc_in = gr.Textbox(label="Item Description")
        with gr.Row():
            finder_name = gr.Textbox(label="Finder's Name (optional)")
            finder_phone = gr.Textbox(label="Finder's Phone Number (optional)")
        add_btn = gr.Button("Submit Item", variant="primary")
        add_output = gr.Textbox(label="Status", interactive=False)

    with gr.Tab("Search"):
        gr.Markdown("### Search Lost Items")
        with gr.Row():
            search_text = gr.Textbox(label="Search by Text")
            search_image = gr.Image(type="pil", label="Or Upload Image")
        search_btn = gr.Button("Search", variant="primary")
        search_output = gr.Markdown()
        gallery = gr.Gallery(label="Matched Items", show_label=True, elem_id="gallery")

    with gr.Tab("Admin"):
        gr.Markdown("### Admin Controls")
        clear_btn = gr.Button("Clear Database", variant="stop")
        clear_output = gr.Textbox(label="Status", interactive=False)

    # Button actions
    add_btn.click(add_item, inputs=[image_in, desc_in, finder_name, finder_phone], outputs=add_output)
    search_btn.click(search_items, inputs=[search_text, search_image], outputs=[search_output, gallery])
    clear_btn.click(clear_database, outputs=clear_output)

# --------------------------
# Launch App
# --------------------------
if __name__ == "__main__":
    demo.launch(server_name="0.0.0.0", server_port=7860)