fix ocr_mk_markdown_with_para_core_v2

magic_pdf/cli/magicpdf.py

@@ -25,6 +25,8 @@ import os
 import json as json_parse
 from datetime import datetime
 import click
+from loguru import logger
+
 from magic_pdf.pipe.UNIPipe import UNIPipe
 from magic_pdf.pipe.OCRPipe import OCRPipe
 from magic_pdf.pipe.TXTPipe import TXTPipe
@@ -77,13 +79,13 @@ def _do_parse(pdf_bytes, model_list, parse_method, image_writer, md_writer, imag
         path=f"{part_file_name}.json",
         mode=AbsReaderWriter.MODE_TXT,
     )
-    try:
-        content_list = pipe.pipe_mk_uni_format()
-    except Exception as e:
-        print(e)
-    md_writer.write(
-        str(content_list), f"{part_file_name}.txt", AbsReaderWriter.MODE_TXT
-    )
+    # try:
+    #     content_list = pipe.pipe_mk_uni_format()
+    # except Exception as e:
+    #     logger.exception(e)
+    # md_writer.write(
+    #     str(content_list), f"{part_file_name}.txt", AbsReaderWriter.MODE_TXT
+    # )
 
 
 @click.group()

magic_pdf/dict2md/ocr_mkcontent.py

@@ -32,7 +32,7 @@ def ocr_mk_nlp_markdown_with_para(pdf_info_dict: list):
     markdown = []
     for page_info in pdf_info_dict:
         paras_of_layout = page_info.get("para_blocks")
-        page_markdown = ocr_mk_markdown_with_para_core(paras_of_layout, "nlp")
+        page_markdown = ocr_mk_markdown_with_para_core_v2(paras_of_layout, "nlp")
         markdown.extend(page_markdown)
     return '\n\n'.join(markdown)
 
@@ -44,7 +44,7 @@ def ocr_mk_mm_markdown_with_para_and_pagination(pdf_info_dict: list, img_buket_p
         paras_of_layout = page_info.get("para_blocks")
         if not paras_of_layout:
             continue
-        page_markdown = ocr_mk_markdown_with_para_core(paras_of_layout, "mm", img_buket_path)
+        page_markdown = ocr_mk_markdown_with_para_core_v2(paras_of_layout, "mm", img_buket_path)
         markdown_with_para_and_pagination.append({
             'page_no': page_no,
             'md_content': '\n\n'.join(page_markdown)
@@ -93,20 +93,19 @@ def ocr_mk_markdown_with_para_core(paras_of_layout, mode, img_buket_path=""):
 
 def ocr_mk_markdown_with_para_core_v2(paras_of_layout, mode, img_buket_path=""):
     page_markdown = []
-    for paras in paras_of_layout:
-        for para in paras:
-            para_type = para.get('type')
+    for para_block in paras_of_layout:
+        para_type = para_block.get('type')
         if para_type == BlockType.Text:
-                para_text = merge_para_with_text(para)
+            para_text = merge_para_with_text(para_block)
         elif para_type == BlockType.Title:
-                para_text = f"# {merge_para_with_text(para)}"
+            para_text = f"# {merge_para_with_text(para_block)}"
         elif para_type == BlockType.InterlineEquation:
-                para_text = merge_para_with_text(para)
+            para_text = merge_para_with_text(para_block)
         elif para_type == BlockType.Image:
             if mode == 'nlp':
                 continue
             elif mode == 'mm':
-                    img_blocks = para.get('blocks')
+                img_blocks = para_block.get('blocks')
                 for img_block in img_blocks:
                     if img_block.get('type') == BlockType.ImageBody:
                         for line in img_block.get('lines'):
@@ -120,7 +119,7 @@ def ocr_mk_markdown_with_para_core_v2(paras_of_layout, mode, img_buket_path=""):
             if mode == 'nlp':
                 continue
             elif mode == 'mm':
-                    table_blocks = para.get('blocks')
+                table_blocks = para_block.get('blocks')
                 for table_block in table_blocks:
                     if table_block.get('type') == BlockType.TableBody:
                         for line in table_block.get('lines'):