修复spans为空list导致的IndexError: list index out of range

magic_pdf/pre_proc/ocr_dict_merge.py

@@ -24,6 +24,9 @@ def line_sort_spans_by_left_to_right(lines):
     return line_objects
 
 def merge_spans_to_line(spans):
+    if len(spans) == 0:
+        return []
+    else:
         # 按照y0坐标排序
         spans.sort(key=lambda span: span['bbox'][1])
 

magic_pdf/pre_proc/ocr_span_list_modify.py

@@ -77,7 +77,10 @@ def adjust_bbox_for_standalone_block(spans):
 
 def modify_y_axis(spans: list, displayed_list: list, text_inline_lines: list):
     # displayed_list = []
-
+    # 如果spans为空,则不处理
+    if len(spans) == 0:
+        pass
+    else:
         spans.sort(key=lambda span: span['bbox'][1])
 
         lines = []