Update para_split_v2.py

magic_pdf/para/para_split_v2.py

@@ -140,9 +140,9 @@ def __detect_list_lines(lines, new_layout_bboxes, lang):
     # 然后根据编码进行分段, 选出来 1,2,3连续出现至少2次的行，认为是列表。
 
     list_indice, list_start_idx = find_repeating_patterns2(line_fea_encode)
-    if len(list_indice) > 0:
-        if debug_able:
-            logger.info(f"发现了列表，列表行数：{list_indice}， {list_start_idx}")
+    # if len(list_indice) > 0:
+    #     if debug_able:
+    #         logger.info(f"发现了列表，列表行数：{list_indice}， {list_start_idx}")
 
     # TODO check一下这个特列表里缩进的行左侧是不是对齐的。
     segments = []
@@ -150,12 +150,12 @@ def __detect_list_lines(lines, new_layout_bboxes, lang):
         for i in range(start, end + 1):
             if i > 0:
                 if line_fea_encode[i] == 4:
-                    if debug_able:
-                        logger.info(f"列表行的第{i}行不是顶格的")
+                    # if debug_able:
+                    #     logger.info(f"列表行的第{i}行不是顶格的")
                     break
-        else:
-            if debug_able:
-                logger.info(f"列表行的第{start}到第{end}行是列表")
+        # else:
+        #     if debug_able:
+        #         logger.info(f"列表行的第{start}到第{end}行是列表")
 
     return split_indices(total_lines, list_indice), list_start_idx
 
@@ -435,8 +435,8 @@ def __connect_list_inter_layout(blocks_group, new_layout_bbox, layout_list_info,
 
         if pre_layout_list_info[1] and not next_layout_list_info[0] and next_first_para[
             "type"] == BlockType.Text:  # 前一个是列表结尾，后一个是非列表开头，此时检测是否有相同的缩进
-            if debug_able:
-                logger.info(f"连接page {page_num} 内的list")
+            # if debug_able:
+            #     logger.info(f"连接page {page_num} 内的list")
             # 向layout_paras[i] 寻找开头具有相同缩进的连续的行
             may_list_lines = []
             lines = next_first_para.get("lines", [])
@@ -467,8 +467,8 @@ def __connect_list_inter_page(pre_page_paras, next_page_paras, pre_page_layout_b
     if pre_page_paras[-1][-1]["type"] != BlockType.Text or next_page_paras[0][0]["type"] != BlockType.Text:
         return False
     if pre_page_list_info[1] and not next_page_list_info[0]:  # 前一个是列表结尾，后一个是非列表开头，此时检测是否有相同的缩进
-        if debug_able:
-            logger.info(f"连接page {page_num} 内的list")
+        # if debug_able:
+        #     logger.info(f"连接page {page_num} 内的list")
         # 向layout_paras[i] 寻找开头具有相同缩进的连续的行
         may_list_lines = []
         next_page_first_para = next_page_paras[0][0]
@@ -680,8 +680,8 @@ def __connect_middle_align_text(page_paras, new_layout_bbox, page_num, lang):
                 first_line_text = ''.join([__get_span_text(span) for span in layout_para[start]["lines"][0]['spans']])
                 if "Table" in first_line_text or "Figure" in first_line_text:
                     pass
-                if debug_able:
-                    logger.info(line_hi.std())
+                # if debug_able:
+                #     logger.info(line_hi.std())
 
                 if line_hi.std() < 2:
                     """行高度相同，那么判断是否居中"""
@@ -693,8 +693,8 @@ def __connect_middle_align_text(page_paras, new_layout_bbox, page_num, lang):
                             and not all([x1 == layout_box[2] for x1 in all_right_x1]):
                         merge_para = [block["lines"][0] for block in layout_para[start:end + 1]]
                         para_text = ''.join([__get_span_text(span) for line in merge_para for span in line['spans']])
-                        if debug_able:
-                            logger.info(para_text)
+                        # if debug_able:
+                        #     logger.info(para_text)
                         layout_para[start]["lines"] = merge_para
                         for i_para in range(start + 1, end + 1):
                             layout_para[i_para]["lines"] = []