Skip to content

周伟奇 / bmw-ocr

Go to a project
Toggle navigation
Toggle navigation pinning
8b3fd09a by 周伟奇
Options

add bs compare

1 parent 44e363a2
Inline Side-by-side
Showing 6 changed files with 243 additions and 6 deletions
src/apps/doc/consts.py
......@@ -1179,6 +1179,8 @@ HMH_OCR_FIELD = 'hmh_ocr'
JYPZ_OCR_FIELD = 'jypz_ocr'
HT_FIELD = 'ht_ocr'
BD_FIELD = 'bd_ocr'
BS_FIELD = 'bss_ocr'
BS_CLASSIFY = 10087
RESULT_MAPPING = {
MVI_CLASSIFY: MVI_OCR_FIELD,
......@@ -1197,9 +1199,10 @@ RESULT_MAPPING = {
JYPZ_CLASSIFY: JYPZ_OCR_FIELD,
CONTRACT_CLASSIFY: HT_FIELD,
INSURANCE_CLASSIFY: BD_FIELD,
BS_CLASSIFY: BS_FIELD,
}
CA_ADD_COMPARE_FIELDS = (IC_OCR_FIELD, BL_OCR_FIELD)
CA_ADD_COMPARE_FIELDS = (IC_OCR_FIELD, BL_OCR_FIELD, BS_FIELD)
COMPARE_FIELDS = (MVI_OCR_FIELD,
IC_OCR_FIELD,
......@@ -1216,6 +1219,7 @@ COMPARE_FIELDS = (MVI_OCR_FIELD,
JYPZ_OCR_FIELD,
HT_FIELD,
BD_FIELD,
BS_FIELD,
)
# 身份证
......@@ -1396,6 +1400,7 @@ SE_STAMP_VALUE = '有'
SE_LAYOUT_VALUE = '旧版-旧打印、新版-新打印'
SE_GB_NEW_FIELD = ['vinNo']
SE_GB_USED_FIELD = ['customerName', 'idNum', 'date']
SE_BS_FIELD = ['户名', '打印日期', '流水日期', '(担保人)户名', '(担保人)打印日期', '(担保人)流水日期']
SE_HMH_FIELD = ['借款人/承租人姓名', '借款人/承租人证件号', '申请号', '渠道', '签字']
SE_BD_FIELD = ['被保险人姓名', '被保险人证件号码', '车架号', '机动车损失保险金额', '第三者责任保险金额', '绝对免赔率', '保险起始日期', '保险截止日期', '保单章', '第一受益人', '保险费合计']
JDMPV_VALUE = ['-', '--', '0%', '0.00', '/', '0']
......@@ -1508,6 +1513,7 @@ HMH_EN = 'Mortgage Waiver Letter'
JYPZ_EN = 'Used Car Document'
AFC_CONTRACT_EN = 'AFC Contract'
BD_EN = 'Insurance'
BS_EN = 'Bank Statement'
SKIP_CARD = {SME_BL_EN}
......@@ -1663,6 +1669,17 @@ BD_COMPARE_LOGIC = {
'保险费合计': ('保险费合计', 'se_amount_compare', {}, '保单保费疑似无法覆盖ASP保险融资'),
}
BS_COMPARE_LOGIC = {
'户名': ('role', 'se_one_compare', {}, '主共借人未提供银行流水'),
'(担保人)户名': ('role', 'se_one_compare', {}, '担保人未提供银行流水，需人工查看直系亲属关系'),
'打印日期': ('print_time', 'se_bs_print_date_compare', {}, '主共借人银行流水打印日期异常'),
'(担保人)打印日期': ('print_time', 'se_bs_print_date_compare', {}, '担保人银行流水打印日期异常'),
'流水日期': ('date', 'se_bs_date_compare', {}, '主共借人银行流水日期异常'),
'(担保人)流水日期': ('date', 'se_bs_date_compare', {}, '担保人银行流水日期异常'),
}
SPECIAL_REASON = '主共借人未提供银行流水，含担保人需人工查看直系亲属关系'
HT_COMPARE_LOGIC = {
'合同编号-每页': ('合同编号-每页', 'se_list_compare', {}, '合同编号与系统不一致'),
'所购车辆价格-小写-重要条款': ('所购车辆价格-小写-重要条款', 'se_amount_str_compare', {}, '合同首页中车辆价格与系统不一致'),
......@@ -1732,6 +1749,7 @@ SE_COMPARE_FIELD = {
JYPZ_EN: (JYPZ_OCR_FIELD, JYPZ_COMPARE_LOGIC, False),
AFC_CONTRACT_EN: (HT_FIELD, HT_COMPARE_LOGIC, False),
BD_EN: (BD_FIELD, BD_COMPARE_LOGIC, False),
BS_EN: (BS_FIELD, BS_COMPARE_LOGIC, False),
}
......
src/apps/doc/management/commands/ocr_process.py
......@@ -872,6 +872,43 @@ class Command(BaseCommand, LoggerMixin):
summary['role'] = self.get_most(summary['role'])
return bs_summary
def rebuild_bs(self, bs_summary):
# bs_summary = {
# '卡号': {
# 'classify': 0,
# 'confidence': 0.9,
# 'role': '柳雪',
# 'code': [('page', 'code'), ],
# 'verify': [(pno, ino, reason_str), ]
# 'print_time': 'datetime',
# 'start_date': 'datetime',
# 'end_date': 'datetime',
# 'sheet': ['sheet_name']
# }
# }
res = []
for bs_info in bs_summary.values():
try:
print_date = bs_info.get('print_time', '').strftime("%Y-%m-%d")
except Exception as e:
print_date = ''
try:
start_date = bs_info.get('start_date', '').strftime("%Y-%m-%d")
except Exception as e:
start_date = ''
try:
end_date = bs_info.get('end_date', '').strftime("%Y-%m-%d")
except Exception as e:
end_date = ''
res.append(
{
'role': bs_info.get('role', ''),
'print_time': print_date,
'date': [start_date, end_date],
}
)
return res
def rebuild_contract(self, contract_result_compare):
res_list = []
for classify, page_info_dict in contract_result_compare.items():
......@@ -1626,6 +1663,10 @@ class Command(BaseCommand, LoggerMixin):
if len(contract_result_compare_rebuild) > 0:
license_summary[consts.CONTRACT_CLASSIFY] = contract_result_compare_rebuild
bs_rebuild = self.rebuild_bs(merged_bs_summary)
if len(bs_rebuild) > 0:
license_summary[consts.BS_CLASSIFY] = bs_rebuild
# 比对
if len(license_summary) > 0 and doc.document_scheme != consts.DOC_SCHEME_LIST[2]:
try:
......
src/apps/doc/models.py
......@@ -315,6 +315,7 @@ class AFCOCRResult(models.Model):
jypz_ocr = models.TextField(null=True, verbose_name="二手车交易凭证")
ht_ocr = models.TextField(null=True, verbose_name="AFC合同")
bd_ocr = models.TextField(null=True, verbose_name="保单")
bss_ocr = models.TextField(null=True, verbose_name="银行流水")
update_time = models.DateTimeField(auto_now=True, verbose_name='修改时间')
create_time = models.DateTimeField(auto_now_add=True, verbose_name='创建时间')
......@@ -346,6 +347,7 @@ class HILOCRResult(models.Model):
jypz_ocr = models.TextField(null=True, verbose_name="二手车交易凭证")
ht_ocr = models.TextField(null=True, verbose_name="AFC合同")
bd_ocr = models.TextField(null=True, verbose_name="保单")
bss_ocr = models.TextField(null=True, verbose_name="银行流水")
update_time = models.DateTimeField(auto_now=True, verbose_name='修改时间')
create_time = models.DateTimeField(auto_now_add=True, verbose_name='创建时间')
......@@ -376,6 +378,7 @@ class AFCSEOCRResult(models.Model):
jypz_ocr = models.TextField(null=True, verbose_name="二手车交易凭证")
ht_ocr = models.TextField(null=True, verbose_name="AFC合同")
bd_ocr = models.TextField(null=True, verbose_name="保单")
bss_ocr = models.TextField(null=True, verbose_name="银行流水")
update_time = models.DateTimeField(auto_now=True, verbose_name='修改时间')
create_time = models.DateTimeField(auto_now_add=True, verbose_name='创建时间')
......@@ -407,6 +410,7 @@ class HILSEOCRResult(models.Model):
jypz_ocr = models.TextField(null=True, verbose_name="二手车交易凭证")
ht_ocr = models.TextField(null=True, verbose_name="AFC合同")
bd_ocr = models.TextField(null=True, verbose_name="保单")
bss_ocr = models.TextField(null=True, verbose_name="银行流水")
update_time = models.DateTimeField(auto_now=True, verbose_name='修改时间')
create_time = models.DateTimeField(auto_now_add=True, verbose_name='创建时间')
......
src/celery_compare/tasks.py
......@@ -971,7 +971,7 @@ def get_se_cms_compare_info(last_obj, application_entity, detect_list):
vehicle_info = {}
vehicle_field_input = []
vehicle_status = cms_info.get('vehicleStatus', '')
first_submmison_date = cms_info.get('submissionDate', '')
first_submission_date = cms_info.get('submissionDate', '')
vin_no = cms_info.get('vehicleInformation', {}).get('vinNo', '')
amount = str(cms_info.get('financialInformation', {}).get('vehiclePrice', '0.0'))
# 新车发票----------------------------------------------------------------------------------------------------------
......@@ -986,7 +986,7 @@ def get_se_cms_compare_info(last_obj, application_entity, detect_list):
else:
vehicle_field_input.append((consts.SE_NEW_ADD_FIELD[0], hmh_name))
vehicle_field_input.append((consts.SE_NEW_ADD_FIELD[1], hmh_id))
vehicle_field_input.append((consts.SE_NEW_ADD_FIELD[2], first_submmison_date))
vehicle_field_input.append((consts.SE_NEW_ADD_FIELD[2], first_submission_date))
vehicle_field_input.append((consts.SE_NEW_ADD_FIELD[3], consts.SE_STAMP_VALUE))
vehicle_field_input.append((consts.SE_NEW_ADD_FIELD[4], consts.SE_FPL_VALUE))
bhsj = float(amount) / 1.13
......@@ -1020,12 +1020,12 @@ def get_se_cms_compare_info(last_obj, application_entity, detect_list):
jypz_field_input.append((consts.SE_NEW_ADD_FIELD[1], main_num))
gb34_field_input.append((consts.SE_GB_USED_FIELD[0], main_name))
gb34_field_input.append((consts.SE_GB_USED_FIELD[1], main_num))
gb34_field_input.append((consts.SE_GB_USED_FIELD[2], first_submmison_date))
vehicle_field_input.append((consts.SE_NEW_ADD_FIELD[2], first_submmison_date))
gb34_field_input.append((consts.SE_GB_USED_FIELD[2], first_submission_date))
vehicle_field_input.append((consts.SE_NEW_ADD_FIELD[2], first_submission_date))
vehicle_field_input.append((consts.SE_NEW_ADD_FIELD[3], consts.SE_STAMP_VALUE))
jypz_field_input.append(('vinNo', vin_no))
jypz_field_input.append(('vehicleTransactionAmount', amount))
jypz_field_input.append((consts.SE_GB_USED_FIELD[-1], first_submmison_date))
jypz_field_input.append((consts.SE_GB_USED_FIELD[-1], first_submission_date))
vehicle_info[consts.MVC_EN] = gb_field_input
vehicle_info[consts.MVC34_EN] = gb34_field_input
......@@ -1064,6 +1064,34 @@ def get_se_cms_compare_info(last_obj, application_entity, detect_list):
bank_info[consts.DDA_EN] = dda_field_input
compare_info['bankInfo'] = bank_info
# 银行流水 --------------------------------------------------------------------
if cms_info.get('autoApprovedDetails', {}).get('aaType', '') in ['CAA1', 'CAA2']:
bs_role_list = []
for applicant_type in consts.APPLICANT_TYPE_ORDER[:2]:
if applicant_type in main_role_info:
for bs_role, _, _ in main_role_info[applicant_type]:
bs_role_list.append(bs_role)
bs_info = dict()
bs_field_input = [
(consts.SE_BS_FIELD[0], bs_role_list),
(consts.SE_BS_FIELD[1], first_submission_date),
(consts.SE_BS_FIELD[2], '90天'),
]
dbr_bs_role_list = []
for dbr_bs_role, _, _ in main_role_info.get(consts.APPLICANT_TYPE_ORDER[2], []):
dbr_bs_role_list.append(dbr_bs_role)
if len(dbr_bs_role_list) > 0:
bs_field_input.extend([
(consts.SE_BS_FIELD[3], dbr_bs_role_list),
(consts.SE_BS_FIELD[4], first_submission_date),
(consts.SE_BS_FIELD[5], '90天'),
])
bs_info[consts.BS_EN] = bs_field_input
compare_info['Bank Statement'] = bs_info
# 抵押登记豁免函----------------------------------------------------------------------------------------------------
other_info = {}
full_no = cms_info.get('settlemnetVerification', {}).get('applicationNo', '')
......@@ -1408,6 +1436,124 @@ def get_se_compare_info(last_obj, application_entity, detect_list):
return compare_info
def se_bs_compare(license_en, ocr_res_dict, strip_list):
# 主共借至少提供一个
# 有担保人，担保人必须提供。主共借没有时，修改comment：人工查看担保人亲属关系
ocr_field, compare_logic, _ = consts.SE_COMPARE_FIELD[license_en]
ocr_res_str = ocr_res_dict.get(ocr_field)
result_field_list = []
field_img_path_dict = dict()
if ocr_res_str is not None:
pre_field_list = strip_list[:3]
suf_field_list = strip_list[3:]
ocr_res_list = json.loads(ocr_res_str)
# length = len(ocr_res_list)
# 主共借人
pre_best_res = {}
max_correct_count = 0
for ocr_res in ocr_res_list:
correct_count = 0
pre_tmp_res_part = {}
for idx, (name, value) in enumerate(pre_field_list):
ocr_str_or_list = ocr_res.get(compare_logic[name][0])
if isinstance(ocr_str_or_list, str) or isinstance(ocr_str_or_list, list):
result = getattr(cp, compare_logic[name][1])(value, ocr_str_or_list, **compare_logic[name][2])
if isinstance(ocr_str_or_list, list):
ocr_str = json.dumps(ocr_str_or_list, ensure_ascii=False)
else:
ocr_str = ocr_str_or_list
reason = compare_logic[name][3]
else:
result = consts.RESULT_N
ocr_str = empty_str
reason = compare_logic[name][3]
if idx == 0 and result == consts.RESULT_N:
break
if result == consts.RESULT_Y:
correct_count += 1
pre_tmp_res_part[name] = (result, ocr_str, reason)
if correct_count > 0 and correct_count >= max_correct_count:
max_correct_count = correct_count
pre_best_res = pre_tmp_res_part
for name, value in pre_field_list:
if len(pre_best_res) > 0:
result, ocr_str, reason = pre_best_res[name]
else:
result = consts.RESULT_N
ocr_str = empty_str
if len(suf_field_list) > 0:
reason = consts.SPECIAL_REASON
else:
reason = compare_logic[pre_field_list[0][0]][3]
img_path = empty_str
error_type = empty_error_type if result == consts.RESULT_Y else ErrorType.OCR.value
if isinstance(value, list):
value = json.dumps(value, ensure_ascii=False)
result_field_list.append((name, value, result, ocr_str, img_path, error_type, reason))
# 担保人
if len(suf_field_list) > 0:
suf_best_res = {}
max_correct_count = 0
for ocr_res in ocr_res_list:
correct_count = 0
suf_tmp_res_part = {}
for idx, (name, value) in enumerate(suf_field_list):
ocr_str_or_list = ocr_res.get(compare_logic[name][0])
if isinstance(ocr_str_or_list, str) or isinstance(ocr_str_or_list, list):
result = getattr(cp, compare_logic[name][1])(value, ocr_str_or_list, **compare_logic[name][2])
if isinstance(ocr_str_or_list, list):
ocr_str = json.dumps(ocr_str_or_list, ensure_ascii=False)
else:
ocr_str = ocr_str_or_list
reason = compare_logic[name][3]
else:
result = consts.RESULT_N
ocr_str = empty_str
reason = compare_logic[name][3]
if idx == 0 and result == consts.RESULT_N:
break
if result == consts.RESULT_Y:
correct_count += 1
suf_tmp_res_part[name] = (result, ocr_str, reason)
if correct_count > 0 and correct_count >= max_correct_count:
max_correct_count = correct_count
suf_best_res = suf_tmp_res_part
for name, value in suf_field_list:
if len(suf_best_res) > 0:
result, ocr_str, reason = suf_best_res[name]
else:
result = consts.RESULT_N
ocr_str = empty_str
reason = compare_logic[suf_field_list[0][0]][3]
img_path = empty_str
error_type = empty_error_type if result == consts.RESULT_Y else ErrorType.OCR.value
if isinstance(value, list):
value = json.dumps(value, ensure_ascii=False)
result_field_list.append((name, value, result, ocr_str, img_path, error_type, reason))
else:
for name, value in strip_list:
if isinstance(value, list):
value = json.dumps(value, ensure_ascii=False)
result_field_list.append((name, value, consts.RESULT_N, empty_str, empty_str, ErrorType.NF.value,
'未提供银行流水'))
return result_field_list, field_img_path_dict
def se_compare_license(license_en, ocr_res_dict, field_list):
ocr_field, compare_logic, special_expiry_date = consts.SE_COMPARE_FIELD[license_en]
......@@ -1816,6 +1962,8 @@ def se_compare_process(compare_info, ocr_res_dict):
result_field_list, field_img_path_dict = se_mvc34_compare(license_en, ocr_res_dict, strip_list)
elif license_en == consts.AFC_CONTRACT_EN:
result_field_list, field_img_path_dict = se_afc_contract_compare(license_en, ocr_res_dict, strip_list)
elif license_en == consts.BS_EN:
result_field_list, field_img_path_dict = se_bs_compare(license_en, ocr_res_dict, strip_list)
else:
result_field_list, _, field_img_path_dict = se_compare_license(license_en, ocr_res_dict, strip_list)
for name, value, result, ocr_str, img_path, error_type, cn_reason in result_field_list:
......
src/common/tools/comparison.py
......@@ -472,6 +472,28 @@ class Comparison:
except Exception as e:
return self.RESULT_N
def se_bs_print_date_compare(self, input_str, ocr_str, **kwargs):
try:
input_date = datetime.strptime(input_str, "%Y-%m-%d")
ocr_date = datetime.strptime(ocr_str, "%Y-%m-%d")
if ocr_date >= input_date - relativedelta(days=15):
return self.RESULT_Y
return self.RESULT_N
except Exception as e:
return self.RESULT_N
def se_bs_date_compare(self, input_str, ocr_str_list, **kwargs):
if isinstance(ocr_str_list, list) and len(ocr_str_list) == 2:
try:
start_date = datetime.strptime(ocr_str_list[0], "%Y-%m-%d")
end_date = datetime.strptime(ocr_str_list[1], "%Y-%m-%d")
if end_date >= start_date + relativedelta(months=3):
return self.RESULT_Y
return self.RESULT_N
except Exception as e:
return self.RESULT_N
return self.RESULT_N
cp = Comparison()
......
src/common/tools/mssql_script11.py
......@@ -2,11 +2,15 @@ import pyodbc
hil_sql = """
ALTER TABLE hil_ocr_result ADD bd_ocr nvarchar(max);
ALTER TABLE hil_ocr_result ADD bss_ocr nvarchar(max);
ALTER TABLE hil_se_ocr_result ADD bd_ocr nvarchar(max);
ALTER TABLE hil_se_ocr_result ADD bss_ocr nvarchar(max);
"""
afc_sql = """
ALTER TABLE afc_ocr_result ADD bd_ocr nvarchar(max);
ALTER TABLE afc_ocr_result ADD bss_ocr nvarchar(max);
ALTER TABLE afc_se_ocr_result ADD bss_ocr nvarchar(max);
ALTER TABLE afc_se_ocr_result ADD bd_ocr nvarchar(max);
"""
......
Styling with Markdown is supported
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!