Skip to content

周伟奇 / bmw-ocr

Go to a project
Toggle navigation
Toggle navigation pinning
dca5de92 by 周伟奇
Options

se compare part 6

1 parent eccdff28
Inline Side-by-side
Showing 3 changed files with 432 additions and 10 deletions
src/apps/doc/consts.py
......@@ -1023,12 +1023,12 @@ IC_RES_MAPPING = {
1: '真实证件',
}
# ----------------- compare ---------------------
# ----------------- CA compare ---------------------
ENTITY = ['HIL', 'AFC']
CUSTOMER_TYPE = ['TCCOR', 'TCDAS', 'TCFRE', 'TCIAS', 'TCIND', 'TCSEP', 'TCURE']
VEHICLE_STATUS = ['PCUSD', 'PCNEW']
APPLICANT_TYPE = ['COAPP', 'CUSTR', 'GAUTR1', 'GAUTR2']
APPLICANT_TYPE = ['CUSTR', 'COAPP', 'GAUTR1', 'GAUTR2']
ID_TYPE = ['ITARI', 'ITHKM', 'ITPRC', 'ITPSP', 'ITRES', 'ITTID', 'ITUSC', 'ITCCU']
SECOND_ID_TYPE = ['ITARI', 'ITHKM', 'ITPRC', 'ITPSP', 'ITRES', 'ITTID']
SUB_TYPE = ['CSIBM', 'CSOTH', 'CSSME']
......@@ -1074,7 +1074,18 @@ RESULT_MAPPING = {
# VAT_CLASSIFY: 'vat_ocr',
}
COMPARE_FIELDS = (IC_OCR_FIELD, RP_OCR_FIELD, BL_OCR_FIELD, EEP_OCR_FIELD, DL_OCR_FIELD, PP_OCR_FIELD, MVC_OCR_FIELD)
COMPARE_FIELDS = (MVI_OCR_FIELD,
IC_OCR_FIELD,
RP_OCR_FIELD,
BC_OCR_FIELD,
BL_OCR_FIELD,
UCI_OCR_FIELD,
EEP_OCR_FIELD,
DL_OCR_FIELD,
PP_OCR_FIELD,
MVC_OCR_FIELD,
DDA_OCR_FIELD,
)
# 身份证
ITPRC = [
......@@ -1224,3 +1235,163 @@ DDA_MAPPING = [
(DDA_BC_ID, BC_FIELD),
]
# ----------------------- SE compare -----------------------
SE_FIRST_ID_FIELD = ['customerName', 'idNum', 'dateOfBirth', 'idExpiryDate']
SE_ID_CARD_FIELD = ['customerName', 'idNum', 'dateOfBirth', 'idExpiryDate', 'hukouProvince', 'hukouCity']
SE_SECOND_ID_FIELD = [
('customerName', 'customerName'),
('secondIdNum', 'idNum'),
]
SE_CORPORATE_ID_FIELD = [
('legalRepName', 'customerName'),
('firstIdNo', 'idNum')
]
SE_SEP_FIELD = [
('companyName', 'companyName'),
('customerName', 'legalRepName'),
('registeredCapital', 'registeredCapital'),
('selfEmployedSubType', 'organizationType')
]
SE_CORPORATE_FIELD = ['companyName', 'legalRepName', 'businessLicenseNo', 'organizationCreditCode',
'taxRegistrationCertificateNo', 'establishmentDate', 'businessLicenseDueDate', 'organizationType']
SE_CORPORATE_ALL_FIELD = ['companyName', 'legalRepName', 'firstIdNo', 'businessLicenseNo', 'organizationCreditCode',
'taxRegistrationCertificateNo', 'establishmentDate', 'businessLicenseDueDate',
'organizationType']
SE_NEW_FIELD = ['vinNo', 'dealer', 'vehicleTransactionAmount']
SE_USED_FIELD = ['vinNo', 'vehicleTransactionAmount']
SE_NEW_ADD_FIELD = ['customerName', 'idNum', 'dateOfInvoice']
SE_BANK_FIELD = ['accountNo', 'bankName']
SE_DDA_FIELD = ['applicationId(1)', 'applicationId(2)', 'bankName', 'companyName', 'customerName', 'idNum', 'accountHolderName', 'accountNo']
HIL_COMPANY_NAME = '先锋国际融资租赁有限公司'
AFC_COMPANY_NAME = '宝马汽车金融(中国)有限公司'
ID_EN = 'PRC ID'
PP_EN = 'Passport'
EEP_EN = 'Resident Permit to Mainland'
RP_EN = 'Resident ID'
BL_EN = 'Business permit'
MVI_EN = 'newCar Invoice'
UCI_EN = 'usedCar Invoice'
BC_EN = 'Bank Card'
DDA_EN = 'DDA'
ID_COMPARE_LOGIC = [
('customerName', '姓名'),
('idNum', '公民身份号码'),
('dateOfBirth', '出生年月'),
('idExpiryDate', '有效期限'),
('hukouProvince', '住址'),
('hukouCity', '住址'),
]
PP_COMPARE_LOGIC = [
('customerName', '英文姓名'),
('idNum', '护照号码'),
('dateOfBirth', '出生日期'),
('idExpiryDate', '有效期至'),
]
EEP_COMPARE_LOGIC = [
('customerName', '中文名'),
('idNum', '证件号码'),
('dateOfBirth', '出生日期'),
('idExpiryDate', '有效期限'),
]
RP_COMPARE_LOGIC = [
('customerName', '姓名'),
('idNum', '公民身份号码'),
('dateOfBirth', '出生年月'),
('idExpiryDate', '有效期限'),
('hukouProvince', '住址'),
('hukouCity', '住址'),
]
BL_COMPARE_LOGIC = [
('companyName', '企业名称', ),
('legalRepName', '经营者姓名', ),
('firstIdNo', '注册号', ),
('businessLicenseNo', '注册号', ),
('organizationCreditCode', '注册号', ),
('taxRegistrationCertificateNo', '注册号', ),
('establishmentDate', '成立日期', ), # 2017年07月11日
# 2017年07月11日至长期 1. OCR识别为长期，向GCAP发送：2099-12-31 00:00:00.0
('businessLicenseDueDate', '营业期限'),
('organizationType', '企业类型', ), # 有限责任公司
('registeredCapital', '注册资本', ), # 壹拾万元整 将OCR识别结果（人民币大写）转化为数字
]
MVI_COMPARE_LOGIC = [
('vinNo', '车辆识别代码'),
('dealer', '销方名称'),
('vehicleTransactionAmount', '价税合计小写'),
('customerName', '购方名称'),
('idNum', '购买方身份证号或组织机构代码'),
('dateOfInvoice', '开票日期'),
]
UCI_COMPARE_LOGIC = [
('vinNo', '车架号'),
('vehicleTransactionAmount', '价税合计小写'),
('customerName', '购方名称'),
('idNum', '购方纳税人识别号'),
]
SE_DDA_FIELD = ['applicationId(1)', 'applicationId(2)', 'bankName', 'companyName', 'customerName', 'idNum', 'accountHolderName', 'accountNo']
BC_COMPARE_LOGIC = [
('accountNo', 'CardNum'),
('bankName', 'BankName'),
]
DDA_COMPARE_LOGIC = [
('applicationId(1)', 'CardNum'),
('applicationId(2)', 'BankName'),
('bankName', 'BankName'),
('companyName', 'BankName'),
('customerName', 'BankName'),
('idNum', 'BankName'),
('accountHolderName', 'BankName'),
('accountNo', 'BankName'),
]
# MVC_OCR_FIELD = 'mvc_ocr'
SE_COMPARE_FIELD = {
ID_EN: (IC_OCR_FIELD, ID_COMPARE_LOGIC, True),
PP_EN: (PP_OCR_FIELD, PP_COMPARE_LOGIC, False),
EEP_EN: (EEP_OCR_FIELD, EEP_COMPARE_LOGIC, False),
RP_EN: (RP_OCR_FIELD, RP_COMPARE_LOGIC, True),
BL_EN: (BL_OCR_FIELD, BL_COMPARE_LOGIC, False),
MVI_EN: (MVI_OCR_FIELD, MVI_COMPARE_LOGIC, False),
UCI_EN: (UCI_OCR_FIELD, UCI_COMPARE_LOGIC, False),
BC_EN: (BC_OCR_FIELD, BC_COMPARE_LOGIC, False),
DDA_EN: (DDA_OCR_FIELD, DDA_COMPARE_LOGIC, False),
}
SE_FIRST_ID_FIELD_MAPPING = {
'ITPRC': (ID_EN, SE_ID_CARD_FIELD),
'ITPSP': (PP_EN, SE_FIRST_ID_FIELD),
'ITHKM': (EEP_EN, SE_FIRST_ID_FIELD),
'ITTID': (EEP_EN, SE_FIRST_ID_FIELD),
'ITRES': (RP_EN, SE_FIRST_ID_FIELD),
}
SE_SECOND_ID_FIELD_MAPPING = {
'ITPRC': (ID_EN, SE_SECOND_ID_FIELD),
'ITPSP': (PP_EN, SE_SECOND_ID_FIELD),
'ITHKM': (EEP_EN, SE_SECOND_ID_FIELD),
'ITTID': (EEP_EN, SE_SECOND_ID_FIELD),
'ITRES': (RP_EN, SE_SECOND_ID_FIELD),
}
HEAD_LIST = ['Info', 'Index', 'License', 'Field', 'Input', 'OCR', 'Result', 'Position', 'Image', 'errorType']
......
src/apps/doc/views.py
......@@ -778,12 +778,11 @@ class CompareResultView(GenericView):
if len(compare_result_list) == 0:
body_html = "<h1>没有比对结果</h1>"
else:
head_list = ['Info', 'Index', 'License', 'Field', 'Result', 'Input', 'OCR', 'Position', 'Image', 'errorType']
head_content = ''.join(['<th>{0}</th>'.format(head_name) for head_name in head_list])
head_content = ''.join(['<th>{0}</th>'.format(head_name) for head_name in consts.HEAD_LIST])
head_html = '<tr>{0}</tr>'.format(head_content)
row_html_list = []
for row_dict in compare_result_list:
row_list = [row_dict.get(head, '') for head in head_list]
row_list = [row_dict.get(head, '') for head in consts.HEAD_LIST]
row_content = ''.join(['<td>{0}</td>'.format(row_str) for row_str in row_list])
row_html = '<tr>{0}</tr>'.format(row_content)
row_html_list.append(row_html)
......
src/celery_compare/tasks.py
......@@ -435,10 +435,261 @@ def ca_compare(application_id, application_entity, ocr_res_id, last_obj, ocr_res
traceback.format_exc()))
def get_se_compare_info(last_obj, application_entity):
compare_info = {}
individual_info_dict = {}
main_role_info = {}
for individual_info in json.loads(last_obj.individual_cus_info):
main_role_info.setdefault(individual_info['applicantType'], []).append(
(individual_info['customerName'], individual_info['idNum']))
license_dict = {}
if individual_info['idType'] in consts.SE_FIRST_ID_FIELD_MAPPING:
license_en, field_list = consts.SE_FIRST_ID_FIELD_MAPPING[individual_info['idType']]
field_input = []
for field in field_list:
if field in individual_info:
field_input.append((field, individual_info[field]))
license_dict[license_en] = field_input
if individual_info['secondIdType'] in consts.SE_SECOND_ID_FIELD_MAPPING:
second_license_en, second_field_list = consts.SE_SECOND_ID_FIELD_MAPPING[individual_info['secondIdType']]
if second_license_en not in license_dict:
second_field_input = []
for second_field, write_field in second_field_list:
if second_field in individual_info:
second_field_input.append((write_field, individual_info[second_field]))
license_dict[second_license_en] = second_field_input
if individual_info['customerType'] == consts.CUSTOMER_TYPE[5]:
sep_field_input = []
for sep_field, sep_write_field in consts.SE_SEP_FIELD:
if sep_field in individual_info:
sep_field_input.append((sep_write_field, individual_info[sep_field]))
license_dict[consts.BL_EN] = sep_field_input
if len(license_dict) > 0:
individual_info_dict.setdefault(individual_info['applicantType'], []).append(license_dict)
compare_info['individualCusInfo'] = individual_info_dict
for applicant_type in consts.APPLICANT_TYPE:
if applicant_type in main_role_info:
main_name, main_num = main_role_info[applicant_type][0]
break
del main_role_info
company_info = None
if isinstance(last_obj.corporate_cus_info, str):
corporate_info_dict = json.loads(last_obj.corporate_cus_info)
corporate_info = {}
company_info = (corporate_info_dict['companyName'], corporate_info_dict['businessLicenseNo'])
if corporate_info_dict['firstIdType'] == consts.ID_TYPE[6]:
bl_field_input = []
for bl_field in consts.SE_CORPORATE_ALL_FIELD:
bl_field_input.append((bl_field, corporate_info_dict[bl_field]))
corporate_info[consts.BL_EN] = bl_field_input
else:
if corporate_info_dict['firstIdType'] in consts.SE_FIRST_ID_FIELD_MAPPING:
license_en, _ = consts.SE_FIRST_ID_FIELD_MAPPING[corporate_info_dict['firstIdType']]
first_id_field_input = []
for first_id_field, first_id_write_field in consts.SE_CORPORATE_ID_FIELD:
first_id_field_input.append((first_id_write_field, corporate_info_dict[first_id_field]))
corporate_info[license_en] = first_id_field_input
bl_field_input = []
for bl_field in consts.SE_CORPORATE_FIELD:
bl_field_input.append((bl_field, corporate_info_dict[bl_field]))
corporate_info[consts.BL_EN] = bl_field_input
compare_info['corporateCusInfo'] = corporate_info
vehicle_info_dict = json.loads(last_obj.vehicle_info) # TODO 车辆登记证
vehicle_info = {}
vehicle_field_input = []
if vehicle_info_dict['vehicleStatus'] == 'New':
for vehicle_field in consts.SE_NEW_FIELD:
vehicle_field_input.append((vehicle_field, vehicle_info_dict[vehicle_field]))
vehicle_field_input.append((consts.SE_NEW_ADD_FIELD[0], main_name))
vehicle_field_input.append((consts.SE_NEW_ADD_FIELD[1], main_num))
vehicle_field_input.append((consts.SE_NEW_ADD_FIELD[2], last_obj.first_submmison_date))
vehicle_info[consts.MVI_EN] = vehicle_field_input
else:
for vehicle_field in consts.SE_USED_FIELD:
vehicle_field_input.append((vehicle_field, vehicle_info_dict[vehicle_field]))
if isinstance(company_info, tuple):
vehicle_field_input.append((consts.SE_NEW_ADD_FIELD[0], company_info[0]))
vehicle_field_input.append((consts.SE_NEW_ADD_FIELD[1], company_info[1]))
else:
vehicle_field_input.append((consts.SE_NEW_ADD_FIELD[0], main_name))
vehicle_field_input.append((consts.SE_NEW_ADD_FIELD[1], main_num))
vehicle_info[consts.UCI_EN] = vehicle_field_input
compare_info['vehicleInfo'] = vehicle_info
bank_info_dict = json.loads(last_obj.bank_info)
bank_info = {}
bank_field_input = []
for bank_field in consts.SE_BANK_FIELD:
bank_field_input.append((bank_field, bank_info_dict[bank_field]))
bank_info[consts.BC_EN] = bank_field_input
dda_field_input = []
for dda_field in consts.SE_DDA_FIELD:
if dda_field.startswith('applicationId'):
dda_field_input.append((dda_field, last_obj.application_id))
elif dda_field == 'companyName':
company_name = consts.HIL_COMPANY_NAME if application_entity in consts.HIL_SET else consts.AFC_COMPANY_NAME
dda_field_input.append((dda_field, company_name))
elif dda_field == 'customerName':
dda_field_input.append((dda_field, main_name))
elif dda_field == 'idNum':
dda_field_input.append((dda_field, main_num))
else:
dda_field_input.append((dda_field, bank_info_dict[dda_field]))
bank_info[consts.DDA_EN] = dda_field_input
compare_info['bankInfo'] = bank_info
return compare_info
def se_compare_license(license_en, ocr_res_dict, field_list):
ocr_field, compare_logic, special_expiry_date = consts.SE_COMPARE_FIELD[license_en]
is_find = False
ocr_res_str = ocr_res_dict.get(ocr_field)
if ocr_res_str is not None:
ocr_res_list = json.loads(ocr_res_str)
length = len(ocr_res_list)
# 身份证、居住证 过期期限特殊处理
if special_expiry_date:
expiry_dates = []
key = compare_logic[2][1]
# for ocr_res in tmp_list:
for ocr_res in ocr_res_list:
if ocr_res.get(key):
expiry_dates.append(ocr_res.get(key))
else:
expiry_dates = []
# for ocr_res in tmp_list:
for ocr_res in ocr_res_list:
if is_find:
break
for idx, compare_tuple in enumerate(compare_list):
input_str = info_dict.get(compare_tuple[0])
ocr_str = ocr_res.get(compare_tuple[1])
compare_res, ocr_output = getattr(cp, compare_tuple[2])(
input_str, ocr_str, idx, **compare_tuple[3])
# print('type: {0}, idx: {1}, field: {2}, input: {3}, ocr: {4}, res: {5}, out: {6}'.format(ocr_field, idx, compare_tuple[0], input_str, ocr_str, compare_res, ocr_output))
if idx == 0 and compare_res in [consts.RESULT_N, consts.RESULT_NA] and length > 1:
break
is_find = True
# 过期期限特殊处理
if idx == 2 and has_expiry_date and compare_res == consts.RESULT_NA:
for expiry_date in expiry_dates:
expiry_date_compare_res, expiry_date_ocr_output = getattr(cp, compare_tuple[2])(
input_str, expiry_date, idx, **compare_tuple[3]
)
if expiry_date_compare_res == consts.RESULT_Y:
compare_res = consts.RESULT_Y
ocr_output = expiry_date_ocr_output
ocr_str = expiry_date
break
info_dict[compare_tuple[4]] = compare_res
if input_str is not None:
if ocr_str is None or ocr_output is None:
del info_dict[compare_tuple[0]]
else:
info_dict[compare_tuple[0]] = ocr_output
res_set.add(compare_res)
if not is_find:
res_set.add(consts.RESULT_N)
for compare_tuple in compare_list:
info_dict[compare_tuple[4]] = consts.RESULT_NA
if compare_tuple[0] in info_dict:
del info_dict[compare_tuple[0]]
def se_compare_process(compare_info, ocr_res_dict):
# individualCusInfo
# corporateCusInfo
# vehicleInfo
# bankInfo
compare_result = []
total_fields = 0
failed_count = 0
successful_at_this_level = True
failure_reason = {}
for info_key, info_value in compare_info.items():
if info_key == 'individualCusInfo':
for idx, license_dict in info_value.items():
for license_en, field_list in license_dict.items():
failure_field = []
result_field_list = se_compare_license(license_en, ocr_res_dict, field_list)
for name, value, result, ocr_str in result_field_list:
total_fields += 1
if result == consts.RESULT_N:
failed_count += 1
successful_at_this_level = False
failure_field.append(name)
compare_result.append(
{
'Info': info_key,
'Index': idx,
'License': license_en,
'Field': name,
'Input': value,
'OCR': ocr_str,
'Result': result,
'Position': '',
'Image': '',
'errorType': '',
}
)
failure_reason.setdefault(license_en, []).append('/'.join(failure_field))
else:
for license_en, field_list in info_value.items():
failure_field = []
result_field_list = se_compare_license(license_en, ocr_res_dict, field_list)
for name, value, result, ocr_str in result_field_list:
total_fields += 1
if result == consts.RESULT_N:
failed_count += 1
successful_at_this_level = False
failure_field.append(name)
compare_result.append(
{
'Info': info_key,
'Index': 0,
'License': license_en,
'Field': name,
'Input': value,
'OCR': ocr_str,
'Result': result,
'Position': '',
'Image': '',
'errorType': '',
}
)
failure_reason.setdefault(license_en, []).append('/'.join(failure_field))
failure_reason_list = []
for license_en, failed_field_list in failure_reason.items():
failure_reason_list.append('{0}:{1}'.format(license_en, ';'.join(failed_field_list)))
failure_reason_str = '、'.join(failure_reason_list)
return compare_result, total_fields, failed_count, successful_at_this_level, failure_reason_str
def se_compare(application_id, application_entity, ocr_res_id, last_obj, ocr_res_dict):
try:
# 比对逻辑
pass
start_time = datetime.now()
compare_info = get_se_compare_info(last_obj, application_entity)
compare_result, total_fields, failed_count, successful_at_this_level, failure_reason_str = se_compare_process(
compare_info, ocr_res_dict)
except Exception as e:
pass
else:
......@@ -452,7 +703,7 @@ def se_compare(application_id, application_entity, ocr_res_id, last_obj, ocr_res
res_obj.compare_count = total_fields
res_obj.failed_count = failed_count
# res_obj.reason1_count = reason1_count
res_obj.result = json.dumps(compare_result_str)
res_obj.result = json.dumps(compare_result)
res_obj.save()
compare_log.info('{0} [SE] [result save success] [entity={1}] [id={2}] [ocr_res_id={3}]'.format(
log_base, application_entity, application_id, ocr_res_id))
......@@ -462,6 +713,7 @@ def se_compare(application_id, application_entity, ocr_res_id, last_obj, ocr_res
traceback.format_exc()))
# report
end_time = datetime.now()
try:
request_trigger = RequestTrigger.SUBMITING.value if ocr_res_id is None else RequestTrigger.UPLOADING.value
report_class = HILCompareReport if application_entity == consts.HIL_PREFIX else AFCCompareReport
......@@ -472,10 +724,10 @@ def se_compare(application_id, application_entity, ocr_res_id, last_obj, ocr_res
transaction_start=start_time,
transaction_end=end_time,
successful_at_this_level=successful_at_this_level,
failure_reason=failure_reason,
failure_reason=failure_reason_str,
process_name=ProcessName.SE_CACOMPARE.value,
total_fields=total_fields,
workflow_name=customer_type,
workflow_name=last_obj.customer_type,
)
compare_log.info('{0} [SE] [report save success] [entity={1}] [id={2}] [ocr_res_id={3}]'.format(
log_base, application_entity, application_id, ocr_res_id))
......
Styling with Markdown is supported
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!