Update app.py

app.py

@@ -10,6 +10,22 @@ def load_classifier(model_path: str):
     tokenizer = AutoTokenizer.from_pretrained(model_path)
     return pipeline("text-classification", model=model, tokenizer=tokenizer)
 
+def defang_url(url: str) -> str:
+    """
+    Defangs the URL to prevent it from being clickable.
+    This function replaces the protocol and dots.
+    For example:
+        https://example.com  -->  hxxps://example[.]com
+    """
+    # Replace the protocol
+    if url.startswith("https://"):
+        url = url.replace("https://", "hxxps://")
+    elif url.startswith("http://"):
+        url = url.replace("http://", "hxxp://")
+    
+    # Replace periods in the rest of the URL
+    return url.replace(".", "[.]")
+
 st.title("URL Typosquatting Detection with URLGuardian")
 st.markdown(
     "This app uses the **URLGuardian** classifier developed by Anvilogic to detect potential suspicious URL. "
@@ -26,13 +42,14 @@ if st.button("Check Safety of the url"):
         result = classifier(url)[0]
         label = result["label"]
         score = result["score"]
+        defanged_url = defang_url(url)
         if  label=='Safe':
             st.success(
-                f"The URL '{url}' is considered safe with a confidence of {score * 100:.2f}%."
+                f"The URL '{defanged_url}' is considered safe with a confidence of {score * 100:.2f}%."
             )
         else:
             st.error(
-                f"The URL '{url}' is considered suspicious with a confidence of {score * 100:.2f}%."
+                f"The URL '{defanged_url}' is considered suspicious with a confidence of {score * 100:.2f}%."
             )
         # Optionally, you can display the full result for debugging purposes:
         st.write("Full classification output:", result)