add todo about interline_equation

magic_pdf/pdf_parse_union_core.py

@@ -111,6 +111,7 @@ def parse_page_core(pdf_docs, magic_model, page_id, pdf_bytes_md5, imageWriter,
     spans = ocr_cut_image_and_table(spans, pdf_docs[page_id], page_id, pdf_bytes_md5, imageWriter)
 
     '''将所有区块的bbox整理到一起'''
+    # @todo interline_equation_blocks参数不够准，后面切换到interline_equations上
     if len(interline_equation_blocks) > 0:
         all_bboxes, all_discarded_blocks, drop_reasons = ocr_prepare_bboxes_for_layout_split(
             img_blocks, table_blocks, discarded_blocks, text_blocks, title_blocks,

magic_pdf/pre_proc/ocr_detect_all_bboxes.py

@@ -36,6 +36,9 @@ def ocr_prepare_bboxes_for_layout_split(img_blocks, table_blocks, discarded_bloc
     all_bboxes = fix_text_overlap_title_blocks(all_bboxes)
     '''任何框体与舍弃框重叠，优先信任舍弃框'''
     all_bboxes = remove_need_drop_blocks(all_bboxes, discarded_blocks)
+    # @todo interline_equation 与title或text框冲突的情况，分两种情况处理
+    '''interline_equation框与文本类型框iou比较接近1的时候，信任行间公式框'''
+    '''interline_equation框被包含在文本类型框内，且interline_equation比文本区块小很多时信任文本框，这时需要舍弃公式框'''
 
     '''discarded_blocks中只保留宽度超过1/3页面宽度的，高度超过10的，处于页面下半50%区域的（限定footnote）'''
     for discarded in discarded_blocks: