Update utils.py

utils.py

@@ -334,27 +334,44 @@ def merge_overlapping_ranges(ranges):
     return merged_ranges
 
 def sort_text_boxes_in_reading_order(text_bboxes, sorted_panel_bboxes):
+    """
+    Sort text boxes in reading order (left-to-right, top-to-bottom) within panels.
+
+    Args:
+        text_bboxes: List of text bounding boxes [x1, y1, x2, y2].
+        sorted_panel_bboxes: List of panel bounding boxes, assumed to be sorted in reading order.
+
+    Returns:
+        List of indices of text boxes sorted in reading order.
+    """
     text_bboxes = convert_to_list_of_lists(text_bboxes)
     sorted_panel_bboxes = convert_to_list_of_lists(sorted_panel_bboxes)
 
     if len(text_bboxes) == 0:
         return []
 
-    def indices_of_same_elements(nums):
-        groups = groupby(range(len(nums)), key=lambda i: nums[i])
-        return [list(indices) for _, indices in groups]
-
+    # Step 1: Map each text box to its panel
     panel_id_for_text = get_text_to_panel_mapping(text_bboxes, sorted_panel_bboxes)
-    indices_of_texts = list(range(len(text_bboxes)))
-    indices_of_texts, panel_id_for_text = zip(*sorted(zip(indices_of_texts, panel_id_for_text), key=lambda x: x[1]))
-    indices_of_texts = list(indices_of_texts)
-    grouped_indices = indices_of_same_elements(panel_id_for_text)
-    for group in grouped_indices:
-        subset_of_text_indices = [indices_of_texts[i] for i in group]
-        text_bboxes_of_subset = [text_bboxes[i] for i in subset_of_text_indices]
-        sorted_subset_indices = sort_texts_within_panel(text_bboxes_of_subset)
-        indices_of_texts[group[0] : group[-1] + 1] = [subset_of_text_indices[i] for i in sorted_subset_indices]
-    return indices_of_texts
+
+    # Step 2: Sort text boxes first by their panel and then within each panel
+    text_indices = list(range(len(text_bboxes)))
+    # Sort by panel ID first
+    text_indices.sort(key=lambda i: panel_id_for_text[i])
+
+    # Step 3: Group by panel ID and sort within each panel
+    from itertools import groupby
+
+    sorted_indices = []
+    for panel_id, group in groupby(text_indices, key=lambda i: panel_id_for_text[i]):
+        group_indices = list(group)
+        text_bboxes_within_panel = [text_bboxes[i] for i in group_indices]
+
+        # Sort within the panel (left-to-right, top-to-bottom)
+        sorted_within_panel = sort_texts_within_panel(text_bboxes_within_panel)
+        sorted_indices.extend([group_indices[i] for i in sorted_within_panel])
+
+    return sorted_indices
+
 
 def get_text_to_panel_mapping(text_bboxes, sorted_panel_bboxes):
     text_to_panel_mapping = []