Update app.py

app.py

@@ -0,0 +1,78 @@
+import gradio as gr
+import pandas as pd
+import numpy as np
+from sentence_transformers import SentenceTransformer
+from sklearn.metrics.pairwise import cosine_similarity
+
+model = SentenceTransformer("sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2")
+
+embeddings_hotels = np.load("normalized_embeddings.pkl", allow_pickle=True)
+embeddings_ar = np.load("normalized_embeddings_ar.pkl", allow_pickle=True)
+df_hotels = pd.read_csv("hotel_dataset_processed.csv")
+df_ar = pd.read_csv("df_ar_1.csv")
+
+def search_in_combined(query_text, model, k=5):
+    query_embedding = model.encode(query_text, convert_to_tensor=True).cpu().numpy().reshape(1, -1)
+    similarities_hotels = cosine_similarity(query_embedding, embeddings_hotels).flatten()
+    similarities_ar = cosine_similarity(query_embedding, embeddings_ar).flatten()
+    top_indices_hotels = np.argsort(similarities_hotels)[::-1][:k]
+    top_indices_ar = np.argsort(similarities_ar)[::-1][:k]
+    top_hotels = df_hotels.iloc[top_indices_hotels].copy()
+    top_ar = df_ar.iloc[top_indices_ar].copy()
+    top_hotels["similarity"] = similarities_hotels[top_indices_hotels]
+    top_ar["similarity"] = similarities_ar[top_indices_ar]
+    combined_top_results = pd.concat([top_hotels, top_ar], ignore_index=True)
+    combined_top_results = combined_top_results.sort_values(by="similarity", ascending=False)
+    return combined_top_results.head(k)
+
+def format_results(results):
+    formatted_results = []
+    for _, row in results.iterrows():
+        if not pd.isna(row.get("hotel_name", "")):
+            google_maps_url = f"https://www.google.com/maps/search/?api=1&query={row.get('hotel_name', 'N/A').replace(' ', '+')}"
+            result = (
+                f"<b>Hotel Name</b>: {row.get('hotel_name', 'N/A')}<br>"
+                f"<b>Description</b>: {row.get('hotel_description', 'N/A')}<br>"
+                f"<b>Review Title</b>: {row.get('review_title', 'N/A')}<br>"
+                f"<b>Review Text</b>: {row.get('review_text', 'N/A')}<br>"
+                f"<b>Rating</b>: {row.get('rate', 'N/A')}<br>"
+                f"<b>Trip Date</b>: {row.get('tripdate', 'N/A')}<br>"
+                f"<b>Price Range</b>: {row.get('price_range', 'N/A')}<br>"
+                f"<b>Location</b>: {row.get('locality', 'N/A')}, {row.get('country', 'N/A')}<br>"
+                f"<b>Hotel Website URL</b>: <a href='{row.get('hotel_url', 'N/A')}' target='_blank'>Link</a><br>"
+                f"<b>Google Maps</b>: <a href='{google_maps_url}' target='_blank'>View on Maps</a><br>"
+                f"<b>Image</b>: <img src='{row.get('hotel_image', 'N/A')}' width='200' /><br>"
+            )
+        else:
+            google_maps_url = f"https://www.google.com/maps/search/?api=1&query={row.get('name', 'N/A').replace(' ', '+')}"
+            result = (
+                f"<b>Name</b>: {row.get('name', 'N/A')}<br>"
+                f"<b>Location</b>: {row.get('location', 'N/A')}<br>"
+                f"<b>Price</b>: {row.get('price', 'N/A')}<br>"
+                f"<b>Price For</b>: {row.get('price_for', 'N/A')}<br>"
+                f"<b>Room Type</b>: {row.get('room_type', 'N/A')}<br>"
+                f"<b>Beds</b>: {row.get('beds', 'N/A')}<br>"
+                f"<b>Rating</b>: {row.get('rating', 'N/A')}<br>"
+                f"<b>Rating Title</b>: {row.get('rating_title', 'N/A')}<br>"
+                f"<b>Google Maps</b>: <a href='{google_maps_url}' target='_blank'>View on Maps</a><br>"
+                f"<b>Number of Ratings</b>: {row.get('number_of_ratings', 'N/A')}<br>"
+                f"<b>Hotel Website URL</b>: <a href='{row.get('url', 'N/A')}' target='_blank'>Link</a><br>"
+                f"<b>Additional Info</b>: {row.get('cm', 'N/A')}<br>"
+            )
+        formatted_results.append(result)
+    return "<br><br>".join(formatted_results)
+
+def search_interface(query_text):
+    results = search_in_combined(query_text, model, 7)
+    return format_results(results)
+
+iface = gr.Interface(
+    fn=search_interface,
+    inputs=gr.Textbox(label="Enter your search query"),
+    outputs=gr.HTML(label="Search Results"),
+    title="Hotel and Arabic Data Search",
+    description="Enter a query to search for hotels or Arabic data. The results will show the top matches based on similarity and provide a Google Maps URL for hotel locations.",
+    examples=["Riyadh", "Deluxe Room"]
+)
+
+iface.launch()