Better handling of spaces

app.py

@@ -226,7 +226,7 @@ class BytelatentProcessor:
                     # Boundary check, though loop structure should prevent out-of-bounds
                     if current_byte_abs_idx < len(all_token_ids):
                         label = token_to_patch_label[current_byte_abs_idx] if current_byte_abs_idx < len(token_to_patch_label) else "Error: Label Missing"
-                        display_text = f"{processed_char_text}-{j+1}".replace(" ", "_")
+                        display_text = f"{processed_char_text}-{j+1}"
                         highlighted_data.append((display_text, label))
                     else: # Should ideally not be reached
                         logging.error(f"Critical: Token index {current_byte_abs_idx} out of bounds for labeling.")
@@ -248,7 +248,7 @@ class BytelatentProcessor:
                 except Exception:
                     pass # Stick with the err_byte display_text
 
-                highlighted_data.append((display_text.replace(" ", "_"), label))
+                highlighted_data.append((display_text, label))
                 logging.warning(
                     f"Token ID {problem_byte_id} at index {current_byte_abs_idx} "
                     f"could not be part of a validly decoded character using iterative decode. Fallback: '{display_text}'."
@@ -321,7 +321,7 @@ class BytelatentProcessor:
 
             # Create highlighted text data
             _highlighted_data, patch_count = self._create_highlight_data(patch_lengths, tokens)
-            ind_highlighted_data = [(text.replace("-1", ""), label) for text, label in _highlighted_data]
+            ind_highlighted_data = [(text.replace(" -1", "_-1").replace("-1", ""), label) for text, label in _highlighted_data]
             grouped_data = defaultdict(str)
             for text, label in ind_highlighted_data:
                 grouped_data[label] += text