Update ocr_mkcontent.py

magic_pdf/dict2md/ocr_mkcontent.py

@@ -17,9 +17,8 @@ from magic_pdf.libs.ocr_content_type import BlockType, ContentType
 # 普通 非vllm
 from magic_pdf.dict2md.ocr_client import PredictClient,compress_image
 
-
-
-
+client = None
+status = None
 
 def __is_hyphen_at_line_end(line):
     """
@@ -127,21 +126,12 @@ def ocr_mk_markdown_with_para_core(paras_of_layout, mode, img_buket_path=''):
     return page_markdown
 
 
-def ocr_mk_markdown_with_para_core_v2(config_path,paras_of_layout,
+def ocr_mk_markdown_with_para_core_v2(paras_of_layout,
                                       mode,
                                       img_buket_path=''):
     page_markdown = []
-    config = configparser.ConfigParser()
-    config.read(config_path)
-    url = config.get('server', 'ocr_server')
-    # logger.info(f'ocr_server：{url}')
-    client = PredictClient(url)
-    status = client.check_health()
-    if not status:
-        pdf_ocr = None
-        logger.warning(f'Health check failed. The server at "{url}" is not responding as expected.')
-        logger.info(f'Qwen ocr解析服务无法正常运行,暂不使用qwen解析表格服务')
-        return None
+
+
     for para_block in paras_of_layout:
         para_text = ''
         para_type = para_block['type']
@@ -434,12 +424,24 @@ def ocr_mk_mm_standard_format(pdf_info_dict: list):
     return content_list
 
 
-def union_make(config_path: str,
+def union_make(
+               config_path: str,
                pdf_info_dict: list,
                make_mode: str,
                drop_mode: str,
                img_buket_path: str = ''):
     output_content = []
+    global client
+    global status
+    config = configparser.ConfigParser()
+    config.read(config_path)
+    url = config.get('server', 'ocr_server')
+    # logger.info(f'ocr_server：{url}')
+    client = PredictClient(url)
+    status = client.check_health()
+    if not status:
+        logger.warning(f'Health check failed. The server at "{url}" is not responding as expected.')
+        logger.info(f'Qwen ocr解析服务无法正常运行,暂不使用qwen解析表格服务')
     for page_info in pdf_info_dict:
         if page_info.get('need_drop', False):
             drop_reason = page_info.get('drop_reason')
@@ -462,11 +464,11 @@ def union_make(config_path: str,
             continue
         if make_mode == MakeMode.MM_MD:
             page_markdown = ocr_mk_markdown_with_para_core_v2(
-                config_path,paras_of_layout, 'mm', img_buket_path)
+                paras_of_layout, 'mm', img_buket_path)
             output_content.extend(page_markdown)
         elif make_mode == MakeMode.NLP_MD:
             page_markdown = ocr_mk_markdown_with_para_core_v2(
-                config_path,paras_of_layout, 'nlp')
+                paras_of_layout, 'nlp')
             output_content.extend(page_markdown)
         elif make_mode == MakeMode.STANDARD_FORMAT:
             for para_block in paras_of_layout: