Skip to content

周伟奇 / bmw-ocr

Go to a project
Toggle navigation
Toggle navigation pinning
b198d2cf by 周伟奇
Options

fix bug

1 parent aeefa795
Inline Side-by-side
Showing 4 changed files with 80 additions and 66 deletions
src/apps/doc/consts.py
......@@ -1044,36 +1044,36 @@ COMPARE_FIELDS = (IC_OCR_FIELD, RP_OCR_FIELD, BL_OCR_FIELD, EEP_OCR_FIELD, DL_OC
# 身份证
ITPRC = [
('customerChineseName', '姓名', 'name_compare', {}),
('idNum', '公民身份号码', 'common_compare', {}),
('customerChineseName', '姓名', 'name_compare', {}, 'customerChineseNameResult'),
('idNum', '公民身份号码', 'common_compare', {}, 'idNumResult'),
# 20200410-20250410 OCR识别为长期，向GCAP发送：2099-12-31 00:00:00.0
('idExpiryDate', '有效期限', 'date_compare', {'long': True, 'ocr_split': True, 'input_replace': ''}),
('dateOfBirth', '出生年月', 'date_compare', {'input_replace': ''}), # 20231023
('idExpiryDate', '有效期限', 'date_compare', {'long': True, 'ocr_split': True, 'input_replace': ''}, 'idExpiryDateResult'),
('dateOfBirth', '出生年月', 'date_compare', {'input_replace': ''}, 'dateOfBirthResult'), # 20231023
]
# 护照
ITPSP = [
('customerChineseName', '英文姓名', 'name_compare', {'is_passport': True}),
('idNum', '护照号码', 'common_compare', {}),
('idExpiryDate', '有效期至', 'date_compare', {'input_replace': ''}), # 20250410
('dateOfBirth', '出生日期', 'date_compare', {'input_replace': ''}), # 20250410
('customerChineseName', '英文姓名', 'name_compare', {'is_passport': True}, 'customerChineseNameResult'),
('idNum', '护照号码', 'common_compare', {}, 'idNumResult'),
('idExpiryDate', '有效期至', 'date_compare', {'input_replace': ''}, 'idExpiryDateResult'), # 20250410
('dateOfBirth', '出生日期', 'date_compare', {'input_replace': ''}, 'dateOfBirthResult'), # 20250410
]
# 港澳台通行证
ITHKM_ITTID = [
('customerChineseName', '中文名', 'common_compare', {}),
('idNum', '证件号码', 'common_compare', {}),
('idExpiryDate', '有效期限', 'date_compare', {'ocr_split': True, 'input_replace': '.'}), # 2013.10.24-2023.10.23
('dateOfBirth', '出生日期', 'date_compare', {'input_replace': '.'}), # 2023.10.23
('customerChineseName', '中文名', 'common_compare', {}, 'customerChineseNameResult'),
('idNum', '证件号码', 'common_compare', {}, 'idNumResult'),
('idExpiryDate', '有效期限', 'date_compare', {'ocr_split': True, 'input_replace': '.'}, 'idExpiryDateResult'), # 2013.10.24-2023.10.23
('dateOfBirth', '出生日期', 'date_compare', {'input_replace': '.'}, 'dateOfBirthResult'), # 2023.10.23
]
# 居住证
ITRES = [
('customerChineseName', '姓名', 'common_compare', {}),
('idNum', '公民身份号码', 'common_compare', {}),
('secondIdNum', '通行证号码', 'common_compare', {}),
('idExpiryDate', '有效期限', 'date_compare', {'ocr_split': True, 'input_replace': ''}), # 20200410-20250410
('dateOfBirth', '出生年月', 'date_compare', {'input_replace': ''}), # 20231023
('customerChineseName', '姓名', 'common_compare', {}, 'customerChineseName'),
('idNum', '公民身份号码', 'common_compare', {}, 'idNum'),
('secondIdNum', '通行证号码', 'common_compare', {}, 'secondIdNum'),
('idExpiryDate', '有效期限', 'date_compare', {'ocr_split': True, 'input_replace': ''}, 'idExpiryDate'), # 20200410-20250410
('dateOfBirth', '出生年月', 'date_compare', {'input_replace': ''}, 'dateOfBirth'), # 20231023
]
ID_TYPE_COMPARE = {
......@@ -1090,33 +1090,33 @@ ID_TYPE_COMPARE = {
# c）两中比对 均不一致：N，向GCAP发送：车辆登记证 & 行驶证识别结果
# 车辆登记证
PCUSD_MVC = [
('vinNo', '9.车辆识别代号/车架号', 'common_compare', {}),
('manufactureDate', '32.车辆出厂日期', 'common_compare', {}),
('firstRegistrationDate', '3.登记日期', 'common_compare', {}),
('vinNo', '9.车辆识别代号/车架号', 'common_compare', {}, 'vinNoResult'),
('manufactureDate', '32.车辆出厂日期', 'common_compare', {}, 'manufactureDateResult'),
('firstRegistrationDate', '3.登记日期', 'common_compare', {}, 'firstRegistrationDateResult'),
]
# 行驶证
PCUSD_DL = [
('vinNo', '车辆识别代码', 'common_compare', {}),
('vinNo', '车辆识别代码', 'common_compare', {}, 'vinNoResult'),
]
# 营业执照
TCCOR = [
('customerChineseName', '企业名称', 'common_compare', {}),
('legalRepName', '经营者姓名', 'name_compare', {}),
('idNum', '注册号', 'common_compare', {}),
('businessLicenseNo', '注册号', 'common_compare', {}),
('taxRegistrationCode', '注册号', 'common_compare', {}),
('incorporationDate', '成立日期', 'date_compare', {'ocr_replace': True}), # 2017年07月11日
('customerChineseName', '企业名称', 'common_compare', {}, 'customerChinessNameResult'),
('legalRepName', '经营者姓名', 'name_compare', {}, 'legalRepNameResult'),
('idNum', '注册号', 'common_compare', {}, 'idNumResult'),
('businessLicenseNo', '注册号', 'common_compare', {}, 'businessLicenseNoResult'),
('taxRegistrationCode', '注册号', 'common_compare', {}, 'taxRegistrationCodeResult'),
('incorporationDate', '成立日期', 'date_compare', {'ocr_replace': True}, 'incorporationDateResult'), # 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', 'date_compare', {'long': True, 'ocr_replace': True}),
('capitalRegAmount', '注册资本', 'rmb_compare', {}), # 壹拾万元整 将OCR识别结果（人民币大写）转化为数字
('businessLicenseDueDate', '营业期限', 'date_compare', {'long': True, 'ocr_replace': True}, 'businessLicenseDueDateResult'),
('capitalRegAmount', '注册资本', 'rmb_compare', {}, 'capitalRegAmountResult'), # 壹拾万元整 将OCR识别结果（人民币大写）转化为数字
]
TCSEP = [
('companyName', '企业名称', 'common_compare', {}),
('registeredCapital', '注册资本', 'rmb_compare', {}), # 壹拾万元整 将OCR识别结果（人民币大写）转化为数字
('selfEmployedSubType', '企业类型', 'type_compare', {}) # 有限责任公司
('companyName', '企业名称', 'common_compare', {}, 'companyNameResult'),
('registeredCapital', '注册资本', 'rmb_compare', {}, 'registeredCapitalResult'), # 壹拾万元整 将OCR识别结果（人民币大写）转化为数字
('selfEmployedSubType', '企业类型', 'type_compare', {}, 'selfEmployedSubTypeResult') # 有限责任公司
]
# 1. POS数据OCR识别结果对应关系如下：
......@@ -1131,7 +1131,7 @@ RESULT_NA = 'NA'
IN_ORDER = ('applicantType', 'idType', 'secondIdType', 'customerType', 'customerChineseName', 'idNum', 'secondIdNum',
"idExpiryDate", "dateOfBirth", 'companyName', "registeredCapital", 'selfEmployedSubType',)
UC_ORDER = ('vinNo', 'manufactureDate', 'firstRegistrationDate')
CO_ORDER = ('customerChineseName', 'legalRepName', 'idNum', 'businessLicenseNo', 'taxRegistrationCode',
CO_ORDER = ('customerChineseName', 'legalRepName', 'idNum', 'customerType', 'businessLicenseNo', 'taxRegistrationCode',
'incorporationDate', 'businessLicenseDueDate', 'capitalRegAmount')
......
src/apps/doc/ocr/gcap.py
......@@ -25,7 +25,7 @@ class GCAP:
response = requests.post(self.url, headers=self.headers, data=data, verify=False, auth=self.auth)
if response.status_code != 200:
raise GCAPException('GCAP response with code: {0}'.format(response.status_code))
return response.status_code
return response.text
def test_send(self):
test_res = OrderedDict({
......
src/celery_compare/tasks.py
......@@ -35,22 +35,25 @@ def field_compare(info_dict, ocr_res_dict, ocr_field, compare_list, res_set):
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 = getattr(cp, compare_tuple[2])(
info_dict.get(compare_tuple[0]), ocr_res.get(compare_tuple[1]), **compare_tuple[3])
ocr_str = ocr_res.get(compare_tuple[1])
compare_res, ocr_output = getattr(cp, compare_tuple[2])(
info_dict.get(compare_tuple[0]), ocr_res.get(compare_tuple[1]), idx, **compare_tuple[3])
if idx == 0 and compare_res == consts.RESULT_N:
break
is_find = True
info_dict[compare_tuple[0] + 'Result'] = compare_res
info_dict[compare_tuple[4]] = compare_res
if input_str is not None:
info_dict[compare_tuple[0]] = ocr_str
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[0] + 'Result'] = consts.RESULT_NA
info_dict[compare_tuple[4]] = consts.RESULT_NA
if compare_tuple[0] in info_dict:
info_dict[compare_tuple[0]] = ''
del info_dict[compare_tuple[0]]
def usedcar_info_compare(info_dict, ocr_res_dict, ocr_field, compare_list, res_set):
......@@ -66,21 +69,21 @@ def usedcar_info_compare(info_dict, ocr_res_dict, ocr_field, compare_list, res_s
input_str = info_dict.get(compare_tuple[0])
ocr_str = ocr_res.get(compare_tuple[1], '')
compare_res = getattr(cp, compare_tuple[2])(
info_dict.get(compare_tuple[0]), ocr_res.get(compare_tuple[1]), **compare_tuple[3])
info_dict.get(compare_tuple[0]), ocr_res.get(compare_tuple[1]), idx, **compare_tuple[3])
if idx == 0 and compare_res == consts.RESULT_N:
no_match_vino.append(ocr_str)
break
is_find = True
info_dict[compare_tuple[0] + 'Result'] = compare_res
info_dict[compare_tuple[4]] = compare_res
if input_str is not None:
info_dict[compare_tuple[0]] = ocr_str
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[0] + 'Result'] = consts.RESULT_N
info_dict[compare_tuple[4]] = consts.RESULT_N
if compare_tuple[0] in info_dict:
info_dict[compare_tuple[0]] = ''
del info_dict[compare_tuple[0]]
return is_find, no_match_vino
......@@ -123,7 +126,7 @@ def compare(application_id, application_entity, uniq_seq, ocr_res_id):
'vehicleStatus': last_obj.vehicle_status,
'wholeResult': 'Y',
'wholeResultMessage': '',
'applicationLink': '',
'applicationLink': r'\\china.bmw.corp\WINFS\SF-CN-data\SF3-CN-S\SF3-CN-S-1-New',
}
})
......@@ -165,7 +168,7 @@ def compare(application_id, application_entity, uniq_seq, ocr_res_id):
elif mvc_find is False and dl_find is True:
vino = mvc_vinos[0] if len(mvc_vinos) > 0 else ''
order_usedcar_info[consts.PCUSD_MVC[0][0]] = vino
order_usedcar_info[consts.PCUSD_MVC[0][0] + 'Result'] = consts.RESULT_N
order_usedcar_info[consts.PCUSD_MVC[0][4]] = consts.RESULT_N
elif mvc_find is False and dl_find is False:
vino_list = []
if len(mvc_vinos) > 0:
......@@ -213,7 +216,7 @@ def compare(application_id, application_entity, uniq_seq, ocr_res_id):
try:
for times in range(consts.RETRY_TIMES):
try:
gcap.send(data)
res_text = gcap.send(data)
except Exception as e:
gcap_exc = str(e)
else:
......@@ -225,5 +228,8 @@ def compare(application_id, application_entity, uniq_seq, ocr_res_id):
'[error={5}]'.format(log_base, application_entity, application_id, uniq_seq,
ocr_res_id, traceback.format_exc()))
else:
compare_log.info('{0} [gcap success] [entity={1}] [id={2}] [uniq_seq={3}] [ocr_res_id={4}] '
'[response={5}]'.format(log_base, application_entity, application_id, uniq_seq,
ocr_res_id, res_text))
compare_log.info('{0} [task success] [entity={1}] [id={2}] [uniq_seq={3}] [ocr_res_id={4}]'.format(
log_base, application_entity, application_id, uniq_seq, ocr_res_id))
......
src/common/tools/comparison.py
import re
from datetime import datetime
from .rmb_upper import to_rmb_upper
......@@ -33,14 +34,14 @@ class Comparison:
else:
return self.RESULT_N
def common_compare(self, input_str, ocr_str, **kwargs):
def common_compare(self, input_str, ocr_str, idx, **kwargs):
if not isinstance(input_str, str) or not isinstance(ocr_str, str):
return self.RESULT_N
return self.build_res(input_str == ocr_str)
return self.build_res(input_str == ocr_str), ocr_str
def name_compare(self, input_str, ocr_str, **kwargs):
def name_compare(self, input_str, ocr_str, idx, **kwargs):
if not isinstance(input_str, str) or not isinstance(ocr_str, str):
return self.RESULT_N
return self.RESULT_N, ocr_str
if kwargs.get('is_passport'):
input_obj = re.search(r'[a-zA-Z]]!', input_str)
if input_obj:
......@@ -48,39 +49,46 @@ class Comparison:
ocr_obj = re.search(r'[a-zA-Z]]!', ocr_str)
if ocr_obj:
ocr_s = ocr_obj.group()
return self.build_res(input_s == ocr_s)
return self.RESULT_N
return self.build_res(input_s == ocr_s), ocr_str
return self.RESULT_N, ocr_str
else:
return self.build_res((input_str == ocr_str))
return self.build_res((input_str == ocr_str)), ocr_str
else:
if re.search(r'[a-zA-Z]]', input_str):
return self.RESULT_NA
return self.RESULT_NA, ocr_str
input_s = input_str.translate(self.TRANS)
ocr_s = ocr_str.translate(self.TRANS)
return self.build_res(input_s == ocr_s)
return self.build_res(input_s == ocr_s), ocr_str
def date_compare(self, input_str, ocr_str, **kwargs):
def date_compare(self, input_str, ocr_str, idx, **kwargs):
if not isinstance(input_str, str) or not isinstance(ocr_str, str):
return self.RESULT_N
return self.RESULT_N, ocr_str
if kwargs.get('long', False) and '长期' in ocr_str:
return '2099-12-31'
return self.RESULT_Y, '2099-12-31'
if kwargs.get('ocr_split', False):
ocr_str = ocr_str.split('-')[-1]
if kwargs.get('ocr_replace', False):
ocr_str = ocr_str.replace('年', '-').replace('月', '-').replace('日', '')
if kwargs.get('input_replace') is not None:
input_str = input_str.replace('-', kwargs.get('input_replace'))
return self.build_res(input_str == ocr_str)
try:
ocr_output = datetime.strptime(ocr_str, '%Y{0}%m{0}%d'.format(
kwargs.get('input_replace'))).strftime('%Y-%m-%d')
except Exception as e:
ocr_output = None
else:
ocr_output = ocr_str
return self.build_res(input_str == ocr_str), ocr_output
def rmb_compare(self, input_str, ocr_str, **kwargs):
def rmb_compare(self, input_str, ocr_str, idx, **kwargs):
if not isinstance(input_str, str) or not isinstance(ocr_str, str):
return self.RESULT_N
input_rmb_upper = to_rmb_upper(float(input_str))
return self.build_res(input_rmb_upper == ocr_str)
return self.build_res(input_rmb_upper == ocr_str), input_str
def type_compare(self, input_str, ocr_str, **kwargs):
def type_compare(self, input_str, ocr_str, idx, **kwargs):
if not isinstance(input_str, str) or not isinstance(ocr_str, str):
return self.RESULT_N
return self.RESULT_N, ocr_str
for map_tuple in self.TYPE_MAPPING:
if re.search(map_tuple[0], ocr_str) is not None:
compare_str = map_tuple[1]
......@@ -88,7 +96,7 @@ class Comparison:
else:
compare_str = self.CSOTH
return self.build_res(input_str == compare_str)
return self.build_res(input_str == compare_str), ocr_str
cp = Comparison()
......
Styling with Markdown is supported
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!