Create app.py

app.py

@@ -0,0 +1,93 @@
+import gradio as gr
+from datasets import load_dataset
+from qdrant_client import QdrantClient, models
+from sentence_transformers import SentenceTransformer
+
+# --- Configuration ---
+QDRANT_HOST = "localhost"  # Or your Hugging Face Space Qdrant URL
+QDRANT_PORT = 6333
+COLLECTION_NAME = "my_text_collection"
+MODEL_NAME = 'sentence-transformers/all-MiniLM-L6-v2'
+
+# --- Load Dataset and Model ---
+# Using a simple dataset from Hugging Face
+dataset = load_dataset("ag_news", split="test")
+data = [item['text'] for item in dataset]
+# Limiting the dataset for a quicker demo
+data = data[:1000]
+
+# Load a pre-trained sentence transformer model
+model = SentenceTransformer(MODEL_NAME)
+
+# --- Qdrant Client and Collection Setup ---
+# Initialize Qdrant client
+# In a Hugging Face Space, you might use a local in-memory instance or connect to a running Qdrant container.
+qdrant_client = QdrantClient(host=QDRANT_HOST, port=QDRANT_PORT)
+
+# Create a Qdrant collection if it doesn't exist
+try:
+    qdrant_client.get_collection(collection_name=COLLECTION_NAME)
+    print("Collection already exists.")
+except Exception as e:
+    print("Creating collection...")
+    qdrant_client.recreate_collection(
+        collection_name=COLLECTION_NAME,
+        vectors_config=models.VectorParams(size=model.get_sentence_embedding_dimension(), distance=models.Distance.COSINE),
+    )
+
+    # --- Generate and Index Embeddings ---
+    print("Generating and indexing embeddings...")
+    batch_size = 128
+    for i in range(0, len(data), batch_size):
+        batch_texts = data[i:i+batch_size]
+        embeddings = model.encode(batch_texts, convert_to_tensor=True)
+
+        qdrant_client.upsert(
+            collection_name=COLLECTION_NAME,
+            points=models.Batch(
+                ids=list(range(i, i + len(batch_texts))),
+                vectors=[embedding.tolist() for embedding in embeddings],
+                payloads=[{"text": text} for text in batch_texts]
+            )
+        )
+    print("Embeddings indexed successfully.")
+
+
+# --- Search Function ---
+def search_in_qdrant(query):
+    """
+    Takes a user query, generates its embedding, and searches in Qdrant.
+    """
+    if not query:
+        return "Please enter a search query."
+
+    query_embedding = model.encode(query).tolist()
+
+    search_result = qdrant_client.search(
+        collection_name=COLLECTION_NAME,
+        query_vector=query_embedding,
+        limit=5  # Return the top 5 most similar results
+    )
+
+    results_text = ""
+    for hit in search_result:
+        results_text += f"**Score:** {hit.score:.4f}\n"
+        results_text += f"**Text:** {hit.payload['text']}\n\n"
+
+    return results_text
+
+# --- Gradio Interface ---
+with gr.Blocks() as demo:
+    gr.Markdown("# Semantic Search with Qdrant and Gradio")
+    gr.Markdown("Enter a query to search for similar news articles from the AG News dataset.")
+
+    with gr.Row():
+        search_input = gr.Textbox(label="Search Query")
+
+    search_button = gr.Button("Search")
+    search_output = gr.Markdown()
+
+    search_button.click(search_in_qdrant, inputs=search_input, outputs=search_output)
+
+if __name__ == "__main__":
+    demo.launch()