Update app.py

app.py

@@ -35,11 +35,14 @@ category_data = df[category_column].dropna().unique().tolist()
 model_name = "sentence-transformers/all-MiniLM-L6-v2"
 model = SentenceTransformer(model_name)
 
-# Function to find synonyms dynamically
+# Function to find synonyms dynamically with fallback
 def find_synonym(word, top_n=1):
     query_embedding = model.encode(word, convert_to_tensor=True)
     results = util.semantic_search(query_embedding, model.encode(designer_data + category_data, convert_to_tensor=True), top_k=top_n)
-    return [designer_data + category_data[result['corpus_id']] for result in results[0] if result['score'] > 0.6]
+    # Check if results exist
+    if results and len(results[0]) > 0:
+        return [designer_data + category_data[result['corpus_id']] for result in results[0] if result['score'] > 0.6]
+    return []  # Return an empty list if no results
 
 # Function to correct spellings
 def correct_spelling(word):
@@ -50,14 +53,15 @@ def correct_spelling(word):
             return best_match
     return word
 
-# Autocomplete function
+# Autocomplete function with safe handling of synonyms
 def autocomplete(query):
     if not query.strip():
         return "None", "None", [], []
 
     original_query = query.strip()
     corrected_query = correct_spelling(original_query)
-    synonym_query = find_synonym(corrected_query, top_n=1)[0] if corrected_query != original_query else corrected_query
+    synonym_results = find_synonym(corrected_query, top_n=1)
+    synonym_query = synonym_results[0] if synonym_results else corrected_query
 
     # Perform fuzzy matching for designers and categories separately
     designer_matches = process.extract(synonym_query, designer_data, scorer=fuzz.partial_ratio, limit=5)
@@ -89,4 +93,4 @@ with gr.Blocks() as demo:
         outputs=[correction_output, synonym_output, designer_output, category_output]
     )
 
-demo.launch()
+demo.launch(share=True)