Skip to content

周伟奇 / bmw-ocr

Go to a project
Toggle navigation
Toggle navigation pinning
efad9db2 by chenyao
Options

更新财务报表的三个报表的处理,dict为空,则excel中sheet不创建

1 parent b859310a
Inline Side-by-side
Showing 3 changed files with 65 additions and 30 deletions
src/apps/doc/consts.py
...@@ -98,6 +98,7 @@ RES_SHEET_HEADER = ('页码', '图片序号', '检测图片序号', '结果') ...@@ -98,6 +98,7 @@ RES_SHEET_HEADER = ('页码', '图片序号', '检测图片序号', '结果')
98 RES_SUCCESS = '识别成功' 98 RES_SUCCESS = '识别成功'
99 RES_SUCCESS_OTHER = '识别成功（其他类）' 99 RES_SUCCESS_OTHER = '识别成功（其他类）'
100 RES_SUCCESS_EMPTY = '识别成功（空数据）' 100 RES_SUCCESS_EMPTY = '识别成功（空数据）'
101 RES_SUCCESS_FINANCIAL_STATEMENT = '识别成功（财务报表类）'
101 RES_FAILED = '识别失败' 102 RES_FAILED = '识别失败'
102 RES_FAILED_1 = '识别失败(阶段1)' 103 RES_FAILED_1 = '识别失败(阶段1)'
103 RES_FAILED_2 = '识别失败(阶段2)' 104 RES_FAILED_2 = '识别失败(阶段2)'
......
src/apps/doc/management/commands/ocr_process.py
...@@ -1725,7 +1725,7 @@ class Command(BaseCommand, LoggerMixin): ...@@ -1725,7 +1725,7 @@ class Command(BaseCommand, LoggerMixin):
1725 contract_result = {} 1725 contract_result = {}
1726 contract_result_compare = {} 1726 contract_result_compare = {}
1727 # 添加财报三个报表的处理 1727 # 添加财报三个报表的处理
1728 financial_statement_dict = {"code": {}, "stamp": {}} 1728 financial_statement_dict = {}
1729 res_list = [] 1729 res_list = []
1730 interest_keyword = Keywords.objects.filter( 1730 interest_keyword = Keywords.objects.filter(
1731 type=KeywordsType.INTEREST.value, on_off=True).values_list('keyword', flat=True) 1731 type=KeywordsType.INTEREST.value, on_off=True).values_list('keyword', flat=True)
...@@ -1756,6 +1756,14 @@ class Command(BaseCommand, LoggerMixin): ...@@ -1756,6 +1756,14 @@ class Command(BaseCommand, LoggerMixin):
1756 self.log_base, img_path)) 1756 self.log_base, img_path))
1757 continue 1757 continue
1758 elif classify in consts.FINANCIAL_STATEMENT_CLASSIFY_LIST: 1758 elif classify in consts.FINANCIAL_STATEMENT_CLASSIFY_LIST:
1759 # 添加到 res_list 中
1760 res_list.append((pno, ino, part_idx, consts.RES_SUCCESS_FINANCIAL_STATEMENT))
1761 # 只要分类为财报三个报表的，就在 financial_statement_dict 中添加对应的 code 和 stamp 两个dict
1762 if "code" not in financial_statement_dict:
1763 financial_statement_dict["code"] = {}
1764 if "stamp" not in financial_statement_dict:
1765 financial_statement_dict["stamp"] = {}
1766
1759 financial_statement_table_name = None 1767 financial_statement_table_name = None
1760 if classify == 97: 1768 if classify == 97:
1761 financial_statement_table_name = "balance_sheet" 1769 financial_statement_table_name = "balance_sheet"
...@@ -2057,7 +2065,51 @@ class Command(BaseCommand, LoggerMixin): ...@@ -2057,7 +2065,51 @@ class Command(BaseCommand, LoggerMixin):
2057 license_summary[consts.BS_CLASSIFY] = bs_rebuild 2065 license_summary[consts.BS_CLASSIFY] = bs_rebuild
2058 2066
2059 # 比对 2067 # 比对
2060 if len(license_summary) > 0 and doc.document_scheme != consts.DOC_SCHEME_LIST[2]: 2068 if len(license_summary) > 0:
2069 if doc.document_scheme != consts.DOC_SCHEME_LIST[2]:
2070 # if len(license_summary) > 0 and doc.document_scheme != consts.DOC_SCHEME_LIST[2]:
2071 try:
2072 is_ca = True if doc.document_scheme == consts.DOC_SCHEME_LIST[0] else False
2073 # 更新OCR累计识别结果表
2074 if business_type == consts.HIL_PREFIX:
2075 result_class = HILOCRResult if is_ca else HILSEOCRResult
2076 res_obj = atomicSaveDBHIL(self,result_class,doc,license_summary,ic_merge,rp_merge, task_str, financial_statement_dict)
2077 else:
2078 result_class = AFCOCRResult if is_ca else AFCSEOCRResult
2079 res_obj = atomicSaveDBAFC(self,result_class,doc,license_summary,ic_merge,rp_merge, task_str, financial_statement_dict)
2080
2081 except Exception as e:
2082 self.online_log.error(
2083 '{0} [process error (ocr result save)] [task={1}] [error={2}]'.format(
2084 self.log_base, task_str, traceback.format_exc()))
2085 else:
2086 self.online_log.info('{0} [ocr result save success] [task={1}] [res_id={2}]'.format(
2087 self.log_base, task_str, res_obj.id))
2088 # 触发比对
2089 try:
2090 # 是否fsm
2091 cms_status_class = HILCmsStatusInfo if business_type in consts.HIL_SET else AFCCmsStatusInfo
2092 cms_status_info = cms_status_class.objects.filter(application_id=doc.application_id).first()
2093 is_fsm = cms_status_info is not None and cms_status_info.is_fsm == 1
2094 self.online_log.info('{0} [isfsm] [task={1}] [true or false={2}]'.format(
2095 self.log_base, task_str, is_fsm))
2096 if is_fsm:
2097 fsm_compare.apply_async((doc.application_id, business_type, None, res_obj.id, is_ca, True),
2098 queue='queue_compare')
2099 else:
2100 # pass
2101 compare.apply_async((doc.application_id, business_type, None, res_obj.id,
2102 is_ca, True), queue='queue_compare')
2103 except Exception as e:
2104 self.online_log.error(
2105 '{0} [process error (comparison info send)] [task={1}] [error={2}]'.format(
2106 self.log_base, task_str, traceback.format_exc()))
2107 else:
2108 self.online_log.info('{0} [comparison info send success] [task={1}] '
2109 '[res_id={2}]'.format(self.log_base, task_str, res_obj.id))
2110 else:
2111 # license_summary 为空
2112 self.online_log.info('{0} [task={1}] [no license_summary]'.format(self.log_base, task_str))
2061 try: 2113 try:
2062 is_ca = True if doc.document_scheme == consts.DOC_SCHEME_LIST[0] else False 2114 is_ca = True if doc.document_scheme == consts.DOC_SCHEME_LIST[0] else False
2063 # 更新OCR累计识别结果表 2115 # 更新OCR累计识别结果表
...@@ -2067,36 +2119,10 @@ class Command(BaseCommand, LoggerMixin): ...@@ -2067,36 +2119,10 @@ class Command(BaseCommand, LoggerMixin):
2067 else: 2119 else:
2068 result_class = AFCOCRResult if is_ca else AFCSEOCRResult 2120 result_class = AFCOCRResult if is_ca else AFCSEOCRResult
2069 res_obj = atomicSaveDBAFC(self,result_class,doc,license_summary,ic_merge,rp_merge, task_str, financial_statement_dict) 2121 res_obj = atomicSaveDBAFC(self,result_class,doc,license_summary,ic_merge,rp_merge, task_str, financial_statement_dict)
2070
2071 except Exception as e: 2122 except Exception as e:
2072 self.online_log.error( 2123 self.online_log.error(
2073 '{0} [process error (ocr result save)] [task={1}] [error={2}]'.format( 2124 '{0} [process error (ocr result save)] [task={1}] [error={2}]'.format(
2074 self.log_base, task_str, traceback.format_exc())) 2125 self.log_base, task_str, traceback.format_exc()))
2075 else:
2076 self.online_log.info('{0} [ocr result save success] [task={1}] [res_id={2}]'.format(
2077 self.log_base, task_str, res_obj.id))
2078 # 触发比对
2079 try:
2080 # 是否fsm
2081 cms_status_class = HILCmsStatusInfo if business_type in consts.HIL_SET else AFCCmsStatusInfo
2082 cms_status_info = cms_status_class.objects.filter(application_id=doc.application_id).first()
2083 is_fsm = cms_status_info is not None and cms_status_info.is_fsm == 1
2084 self.online_log.info('{0} [isfsm] [task={1}] [true or false={2}]'.format(
2085 self.log_base, task_str, is_fsm))
2086 if is_fsm:
2087 fsm_compare.apply_async((doc.application_id, business_type, None, res_obj.id, is_ca, True),
2088 queue='queue_compare')
2089 else:
2090 # pass
2091 compare.apply_async((doc.application_id, business_type, None, res_obj.id,
2092 is_ca, True), queue='queue_compare')
2093 except Exception as e:
2094 self.online_log.error(
2095 '{0} [process error (comparison info send)] [task={1}] [error={2}]'.format(
2096 self.log_base, task_str, traceback.format_exc()))
2097 else:
2098 self.online_log.info('{0} [comparison info send success] [task={1}] '
2099 '[res_id={2}]'.format(self.log_base, task_str, res_obj.id))
2100 2126
2101 # DDA处理 2127 # DDA处理
2102 if do_dda: 2128 if do_dda:
...@@ -2451,7 +2477,9 @@ def atomicSaveDBAFC(self,result_class,doc,license_summary,ic_merge,rp_merge,task ...@@ -2451,7 +2477,9 @@ def atomicSaveDBAFC(self,result_class,doc,license_summary,ic_merge,rp_merge,task
2451 res_obj = result_class() 2477 res_obj = result_class()
2452 res_obj.application_id = doc.application_id 2478 res_obj.application_id = doc.application_id
2453 self.online_log.info('{0} [res_obj is None application_id={1} doc_id={2}]'.format(self.log_base, doc.application_id,doc.id)) 2479 self.online_log.info('{0} [res_obj is None application_id={1} doc_id={2}]'.format(self.log_base, doc.application_id,doc.id))
2454 res_obj.fs_ocr = json.dumps([financial_statement_dict]) 2480 if res_obj is not None:
2481 if financial_statement_dict:
2482 res_obj.fs_ocr = json.dumps([financial_statement_dict])
2455 for classify, field in consts.RESULT_MAPPING.items(): 2483 for classify, field in consts.RESULT_MAPPING.items():
2456 if not hasattr(res_obj, field): 2484 if not hasattr(res_obj, field):
2457 continue 2485 continue
...@@ -2485,7 +2513,9 @@ def atomicSaveDBHIL(self,result_class,doc,license_summary,ic_merge,rp_merge, tas ...@@ -2485,7 +2513,9 @@ def atomicSaveDBHIL(self,result_class,doc,license_summary,ic_merge,rp_merge, tas
2485 res_obj = result_class() 2513 res_obj = result_class()
2486 res_obj.application_id = doc.application_id 2514 res_obj.application_id = doc.application_id
2487 self.online_log.info('{0} [res_obj is None application_id={1} doc_id={2}]'.format(self.log_base, doc.application_id,doc.id)) 2515 self.online_log.info('{0} [res_obj is None application_id={1} doc_id={2}]'.format(self.log_base, doc.application_id,doc.id))
2488 res_obj.fs_ocr = json.dumps([financial_statement_dict]) 2516 if res_obj is not None:
2517 if financial_statement_dict:
2518 res_obj.fs_ocr = json.dumps([financial_statement_dict])
2489 for classify, field in consts.RESULT_MAPPING.items(): 2519 for classify, field in consts.RESULT_MAPPING.items():
2490 if not hasattr(res_obj, field): 2520 if not hasattr(res_obj, field):
2491 continue 2521 continue
......
src/apps/doc/ocr/wb.py
...@@ -829,6 +829,10 @@ class BSWorkbook(Workbook): ...@@ -829,6 +829,10 @@ class BSWorkbook(Workbook):
829 829
830 830
831 def financial_rebuild(self, financial_statement_dict): 831 def financial_rebuild(self, financial_statement_dict):
832 # 如果 financial_statement_dict 为空，则不创建表
833 if not financial_statement_dict:
834 return
835 # 如果 financial_statement_dict 不为空，则创建表
832 ws = self.create_sheet(consts.FINANCIAL_SHEET_NAME) 836 ws = self.create_sheet(consts.FINANCIAL_SHEET_NAME)
833 for fin_key, fin_value in financial_statement_dict.items(): 837 for fin_key, fin_value in financial_statement_dict.items():
834 table_str = "识别码" 838 table_str = "识别码"
......
Styling with Markdown is supported
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!