mod

冯轩
Showing 1 changed file with 45 additions and 44 deletions
src/apps/doc/management/commands/ocr_process.py
--- a/src/apps/doc/management/commands/ocr_process.py
View file @f821dc2
+++ b/src/apps/doc/management/commands/ocr_process.py
View file @f821dc2
@@ -1542,50 +1542,50 @@ class Command(BaseCommand, LoggerMixin):
                                self.log_base, traceback.format_exc()))
                        # 获取算法结果
-                        # ocr_result = wxbs_predict(pdf_handler.pdf_info)
+                        ocr_result = wxbs_predict(pdf_handler.pdf_info)
-                        # page_res = {}
+                        page_res = {}
-                        # for page_num, page_info in ocr_result.get('page_info', {}).items():
+                        for page_num, page_info in ocr_result.get('page_info', {}).items():
-                        #     if isinstance(page_num, str) and page_num.startswith('page_'):
+                            if isinstance(page_num, str) and page_num.startswith('page_'):
-                        #         page_res[page_num] = {
+                                page_res[page_num] = {
-                        #             'classify': int(classify_1_str),
+                                    'classify': int(classify_1_str),
-                        #             "is_asp": ocr_result.get('is_asp', False),
+                                    "is_asp": ocr_result.get('is_asp', False),
-                        #             'page_num': page_num,
+                                    'page_num': page_num,
-                        #             'page_info': page_info
+                                    'page_info': page_info
-                        #         }
+                                }
-                        # contract_res = {}
+                        contract_res = {}
-                        # for img_path_tmp, page_key in pdf_handler.img_path_pno_list:
+                        for img_path_tmp, page_key in pdf_handler.img_path_pno_list:
-                        #     if classify_1_str == str(consts.HMH_CLASSIFY):
+                            if classify_1_str == str(consts.HMH_CLASSIFY):
-                        #         img_contract_res = {
+                                img_contract_res = {
-                        #                 'code': 1,
+                                        'code': 1,
-                        #                 'data': [
+                                        'data': [
-                        #                     {
+                                            {
-                        #                         'classify': consts.HMH_CLASSIFY,
+                                                'classify': consts.HMH_CLASSIFY,
-                        #                         'data': hmh_predict(pdf_handler.pdf_info)
+                                                'data': hmh_predict(pdf_handler.pdf_info)
-                        #                     }
+                                            }
-                        #                 ]
+                                        ]
-                        #             }
+                                    }
-                        #     else:
+                            else:
-                        #         if page_key in page_res:
+                                if page_key in page_res:
-                        #             img_contract_res = {
+                                    img_contract_res = {
-                        #                 'code': 1,
+                                        'code': 1,
-                        #                 'data': [
+                                        'data': [
-                        #                     {
+                                            {
-                        #                         'classify': page_res[page_key].pop('classify', consts.OTHER_CLASSIFY),
+                                                'classify': page_res[page_key].pop('classify', consts.OTHER_CLASSIFY),
-                        #                         'data': page_res[page_key]
+                                                'data': page_res[page_key]
-                        #                     }
+                                            }
-                        #                 ]
+                                        ]
-                        #             }
+                                    }
-                        #         else:
+                                else:
-                        #             img_contract_res = {
+                                    img_contract_res = {
-                        #                 'code': 1,
+                                        'code': 1,
-                        #                 'data': [
+                                        'data': [
-                        #                     {
+                                            {
-                        #                         'classify': int(classify_1_str),
+                                                'classify': int(classify_1_str),
-                        #                     }
+                                            }
-                        #                 ]
+                                        ]
-                        #             }
+                                    }
-                        #     contract_res[img_path_tmp] = img_contract_res
+                            contract_res[img_path_tmp] = img_contract_res
                        with lock:
                            res_str = '''{"/data/AFC/OCR_Files/60671/img/page_0_img_0.png":{"code":1,"msg":"success","data":[{"classify":12,"confidence":0.9999922513961792,"data":[{"summary":["冯轩","110109199202260310",null,null,"2025-06-30","2025-06-25","2025-06-30"],"cells":[{"start_row":0,"start_column":0,"words":"交易单号"},{"start_row":0,"start_column":1,"words":"交易时间"},{"start_row":0,"start_column":2,"words":"交易类型"},{"start_row":0,"start_column":3,"words":"收/支/其他"},{"start_row":0,"start_column":4,"words":"交易方式"},{"start_row":0,"start_column":5,"words":"金额(元)"},{"start_row":0,"start_column":6,"words":"交易对方"},{"start_row":0,"start_column":7,"words":"商户单号"},{"start_row":1,"start_column":0,"words":"4200002687202506298013\\n175397"},{"start_row":1,"start_column":1,"words":"2025-06-29\\n13:47:15"},{"start_row":1,"start_column":2,"words":"商户消费"},{"start_row":1,"start_column":3,"words":"支出"},{"start_row":1,"start_column":4,"words":"浦发银行储\\n蓄卡(9862)"},{"start_row":1,"start_column":5,"words":"333.00"},{"start_row":1,"start_column":6,"words":"美团平台商\\n户"},{"start_row":1,"start_column":7,"words":"20250629134707U\\n8927844819024394"},{"start_row":2,"start_column":0,"words":"1000039901017506286326\\n541033022"},{"start_row":2,"start_column":1,"words":"2025-06-28\\n21:07:37"},{"start_row":2,"start_column":2,"words":"微信红包"},{"start_row":2,"start_column":3,"words":"收入"},{"start_row":2,"start_column":4,"words":"/"},{"start_row":2,"start_column":5,"words":"1.20"},{"start_row":2,"start_column":6,"words":"默"},{"start_row":2,"start_column":7,"words":"1000039901202506\\n286326541033022"},{"start_row":3,"start_column":0,"words":"1000039901004506276276\\n744394049"},{"start_row":3,"start_column":1,"words":"2025-06-27\\n23:58:18"},{"start_row":3,"start_column":2,"words":"微信红包"},{"start_row":3,"start_column":3,"words":"收入"},{"start_row":3,"start_column":4,"words":"/"},{"start_row":3,"start_column":5,"words":"1.36"},{"start_row":3,"start_column":6,"words":"九日"},{"start_row":3,"start_column":7,"words":"1000039901202506\\n276276744394049"},{"start_row":4,"start_column":0,"words":"1000039901008506277372\\n907681001"},{"start_row":4,"start_column":1,"words":"2025-06-27\\n23:57:27"},{"start_row":4,"start_column":2,"words":"微信红包"},{"start_row":4,"start_column":3,"words":"收入"},{"start_row":4,"start_column":4,"words":"/"},{"start_row":4,"start_column":5,"words":"1.68"},{"start_row":4,"start_column":6,"words":"希望"},{"start_row":4,"start_column":7,"words":"1000039901202506\\n277372907681001"},{"start_row":5,"start_column":0,"words":"4200002672202506275479\\n881478"},{"start_row":5,"start_column":1,"words":"2025-06-27\\n19:50:21"},{"start_row":5,"start_column":2,"words":"商户消费"},{"start_row":5,"start_column":3,"words":"支出"},{"start_row":5,"start_column":4,"words":"浦发银行储\\n蓄卡(9862)"},{"start_row":5,"start_column":5,"words":"124.00"},{"start_row":5,"start_column":6,"words":"呷哺呷哺(北\\n京首联世纪\\n广场）"},{"start_row":5,"start_column":7,"words":";8011433A2506271\\n95001M000105"}],"verify":{"verify_res":"real","verify_info":[]}}]}]}}'''
@@ -1808,6 +1808,7 @@ class Command(BaseCommand, LoggerMixin):
                            self.online_log.info('{0} [get lock] [img={1}] '.format(self.log_base, img_path))
                            doc_res_dict = res_dict.setdefault(task_str, {})
                            doc_res_dict[img_path] = ocr_1_res
+                            self.online_log.info('{0} [ocr_1_res={1}] '.format(self.log_base, ocr_1_res))
                            res_dict[task_str] = doc_res_dict
                            todo_count = todo_count_dict.get(task_str)
                            if todo_count == 1: