compare update

周伟奇
Showing 4 changed files with 104 additions and 63 deletions
src/apps/doc/consts.py
src/apps/doc/views.py
src/celery_compare/tasks.py
src/common/tools/comparison.py
--- a/src/apps/doc/consts.py
View file @8914936
+++ b/src/apps/doc/consts.py
View file @8914936
@@ -1018,61 +1018,69 @@ BASE_XML_TEXT = """<?xml version="1.0" encoding="utf-8"?>
 CDATA_TEXT = """<![CDATA[<Exec xmlns="http://tempuri.org/"><strXMLParm>&lt;Request&gt;&lt;Framework&gt;&lt;UserName&gt;SFCHINA\qqcout0&lt;/UserName&gt;&lt;GUID&gt;70d0efcb-3bc2-4018-ac4e-681c8f3131b6&lt;/GUID&gt;&lt;DetailedTracingEnabled&gt;False&lt;/DetailedTracingEnabled&gt;&lt;ServiceName&gt;AMSWebService&lt;/ServiceName&gt;&lt;SupportsRedirection&gt;true&lt;/SupportsRedirection&gt;&lt;ServiceType&gt;Service&lt;/ServiceType&gt;&lt;/Framework&gt;&lt;Parms&gt;&lt;InputXML type="string"&gt;&amp;lt;?xml version="1.0" encoding="utf-16"?&amp;gt;&amp;lt;InputXML&amp;gt; &amp;lt;Result&amp;gt; {0} &amp;lt;/Result&amp;gt;&amp;lt;AuthorizationData&amp;gt;&amp;lt;ServiceComponent&amp;gt;OCR&amp;lt;/ServiceComponent&amp;gt;&amp;lt;RoleId/&amp;gt;&amp;lt;CompanyId/&amp;gt;&amp;lt;/AuthorizationData&amp;gt;&amp;lt;/InputXML&amp;gt;&lt;/InputXML&gt;&lt;/Parms&gt;&lt;/Request&gt;</strXMLParm></Exec>]]>"""
+IC_OCR_FIELD = 'ic_ocr'
+RP_OCR_FIELD = 'rp_ocr'
+BL_OCR_FIELD = 'bl_ocr'
+EEP_OCR_FIELD = 'eep_ocr'
+DL_OCR_FIELD = 'dl_ocr'
+PP_OCR_FIELD = 'pp_ocr'
+MVC_OCR_FIELD = 'mvc_ocr'
 RESULT_MAPPING = {
    # MVI_CLASSIFY: 'mvi_ocr',
-    IC_CLASSIFY: 'ic_ocr',
+    IC_CLASSIFY: IC_OCR_FIELD,
-    RP_CLASSIFY: 'rp_ocr',
+    RP_CLASSIFY: RP_OCR_FIELD,
    # BC_CLASSIFY: 'bc_ocr',
-    BL_CLASSIFY: 'bl_ocr',
+    BL_CLASSIFY: BL_OCR_FIELD,
    # UCI_CLASSIFY: 'uci_ocr',
-    EEP_CLASSIFY: 'eep_ocr',
+    EEP_CLASSIFY: EEP_OCR_FIELD,
-    DL_CLASSIFY: 'dl_ocr',
+    DL_CLASSIFY: DL_OCR_FIELD,
-    PP_CLASSIFY: 'pp_ocr',
+    PP_CLASSIFY: PP_OCR_FIELD,
-    MVC_CLASSIFY: 'mvc_ocr',
+    MVC_CLASSIFY: MVC_OCR_FIELD,
    # VAT_CLASSIFY: 'vat_ocr',
 }
-COMPARE_FIELDS = ('ic_ocr', 'rp_ocr', 'bl_ocr', 'eep_ocr', 'dl_ocr', 'pp_ocr', 'mvc_ocr')
+COMPARE_FIELDS = (IC_OCR_FIELD, RP_OCR_FIELD, BL_OCR_FIELD, EEP_OCR_FIELD, DL_OCR_FIELD, PP_OCR_FIELD, MVC_OCR_FIELD)
 # 身份证
-ITPRC = {
+ITPRC = [
-    'customerChineseName': ('姓名', 'common_compare', {}),
+    ('customerChineseName', '姓名', 'name_compare', {}),
-    'idNum': ('公民身份号码', 'common_compare', {}),
+    ('idNum', '公民身份号码', 'common_compare', {}),
    # 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': ''}),
+    ('idExpiryDate', '有效期限', 'date_compare', {'long': True, 'ocr_split': True, 'input_replace': ''}),
-}
+]
 # 护照
-ITPSP = {
+ITPSP = [
-    'customerChineseName': ('英文姓名', 'common_compare', {}),
+    ('customerChineseName', '英文姓名', 'name_compare', {'is_passport': True}),
-    'idNum': ('护照号码', 'common_compare', {}),
+    ('idNum', '护照号码', 'common_compare', {}),
-    'idExpiryDate': ('有效期至', 'date_compare', {'input_replace': ''}),  # 20250410
+    ('idExpiryDate', '有效期至', 'date_compare', {'input_replace': ''}),  # 20250410
-    'dateOfBirth': ('出生日期', 'date_compare', {'input_replace': ''}),  # 20250410
+    ('dateOfBirth', '出生日期', 'date_compare', {'input_replace': ''}),  # 20250410
-}
+]
 # 港澳台通行证
-ITHKM_ITTID = {
+ITHKM_ITTID = [
-    'customerChineseName': ('中文名', 'common_compare', {}),
+    ('customerChineseName', '中文名', 'common_compare', {}),
-    'idNum': ('证件号码', '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
+    ('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
+    ('dateOfBirth', '出生日期', 'date_compare', {'input_replace': '.'}),  # 2023.10.23
    # 'secondIdNum': ''
-}
+]
 # 居住证
-ITRES = {
+ITRES = [
-    'customerChineseName': ('姓名', 'common_compare', {}),
+    ('customerChineseName', '姓名', 'common_compare', {}),
-    'idNum': ('公民身份号码', 'common_compare', {}),
+    ('idNum', '公民身份号码', 'common_compare', {}),
-    'idExpiryDate': ('有效期限', 'date_compare', {'ocr_split': True, 'input_replace': ''}),  # 20200410-20250410
+    ('idExpiryDate', '有效期限', 'date_compare', {'ocr_split': True, 'input_replace': ''}),  # 20200410-20250410
-    'secondIdNum': ('通行证号码', 'common_compare', {})
+    ('secondIdNum', '通行证号码', 'common_compare', {})
-}
+]
 ID_TYPE_COMPARE = {
-    'ITPRC': {'model_field': 'ic_ocr', 'compare_field': ITPRC},
+    'ITPRC': [IC_OCR_FIELD, ITPRC],
-    'ITPSP': {'model_field': 'pp_ocr', 'compare_field': ITPSP},
+    'ITPSP': [PP_OCR_FIELD, ITPSP],
-    'ITHKM': {'model_field': 'eep_ocr', 'compare_field': ITHKM_ITTID},
+    'ITHKM': [EEP_OCR_FIELD, ITHKM_ITTID],
-    'ITTID': {'model_field': 'eep_ocr', 'compare_field': ITHKM_ITTID},
+    'ITTID': [EEP_OCR_FIELD, ITHKM_ITTID],
-    'ITRES': {'model_field': 'rp_ocr', 'compare_field': ITRES},
+    'ITRES': [RP_OCR_FIELD, ITRES],
 }
 # 1. 分别对比 POS车架号 vs 车辆登记正，POS车架号 vs 行驶证
@@ -1080,37 +1088,35 @@ ID_TYPE_COMPARE = {
 #     b）其中一个 不一致：N，向GCAP发送：不一致的OCR识别结果
 #     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 = {
+PCUSD_MVC = [
-    'vinNo': ('9.车辆识别代号/车架号', 'common_compare', {}),
+    ('vinNo', '9.车辆识别代号/车架号', 'common_compare', {}),
-    'manufactureDate': ('32.车辆出厂日期', 'common_compare', {}),
+    ('manufactureDate', '32.车辆出厂日期', 'common_compare', {}),
-    'firstRegistrationDate': ('3.登记日期', 'common_compare', {}),
+    ('firstRegistrationDate', '3.登记日期', 'common_compare', {}),
-}
+]
 # 行驶证
-PCUSD_DL = {
+PCUSD_DL = [
-    'vinNo': ('车辆识别代码', 'common_compare', {}),
+    ('vinNo', '车辆识别代码', 'common_compare', {}),
-    # 'manufactureDate': '',
+]
-    # 'firstRegistrationDate': '',
-}
 # 营业执照
-TCCOR = {
+TCCOR = [
-    'customerChineseName': ('企业名称', 'common_compare', {}),
+    ('customerChineseName', '企业名称', 'common_compare', {}),
-    'legalRepName': ('经营者姓名', 'common_compare', {}),
+    ('legalRepName', '经营者姓名', 'name_compare', {}),
-    'idNum': ('注册号', 'common_compare', {}),
+    ('idNum', '注册号', 'common_compare', {}),
-    'businessLicenseNo': ('注册号', 'common_compare', {}),
+    ('businessLicenseNo', '注册号', 'common_compare', {}),
-    'taxRegistrationCode': ('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
+    ('incorporationDate', '成立日期', 'date_compare', {'ocr_replace': True}),  # 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}),
+    ('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
+    ('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
-}
+]
-TCSEP = {
+TCSEP = [
-    'companyName': ('企业名称', 'common_compare', {}),
+    ('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
+    ('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
+    ('selfEmployedSubType', '企业类型', 'type_compare', {})  # 有限责任公司
-}
+]
 # 1. POS数据OCR识别结果对应关系如下：
 #     a）Individual Businessman CSIBM => 个体工商户
--- a/src/apps/doc/views.py
View file @8914936
+++ b/src/apps/doc/views.py
View file @8914936
@@ -444,7 +444,7 @@ class DocView(GenericView, DocHandler):
        file.close()
        # 1. 上传信息记录
        # application_id = '{0}{1}'.format(consts.FIXED_APPLICATION_ID_PREFIX, metadata_version_id)
-        application_id = 'CH-S200012772'
+        application_id = 'CH-S200012727'
        upload_finish_time = timezone.now()
        document_scheme = random.choice(consts.DOC_SCHEME_LIST)
        data_source = random.choice(consts.DATA_SOURCE_LIST)
--- a/src/celery_compare/tasks.py
View file @8914936
+++ b/src/celery_compare/tasks.py
View file @8914936
--- a/src/common/tools/comparison.py
View file @8914936
+++ b/src/common/tools/comparison.py
View file @8914936
@@ -19,7 +19,13 @@ class Comparison:
        self.RESULT_Y = 'Y'
        self.RESULT_N = 'N'
-        self.RESULT_NA = 'NA'  # TODO NA情况
+        self.RESULT_NA = 'NA'
+        self.TRANS_MAP = {
+            ' ': '',
+            '·': '',
+        }
+        self.TRANS = str.maketrans(self.TRANS_MAP)
    def build_res(self, result):
        if result:
@@ -28,9 +34,34 @@ class Comparison:
            return self.RESULT_N
    def common_compare(self, input_str, ocr_str, **kwargs):
+        if not isinstance(input_str, str) or isinstance(ocr_str, str):
+            return self.RESULT_N
        return self.build_res(input_str == ocr_str)
+    def name_compare(self, input_str, ocr_str, **kwargs):
+        if not isinstance(input_str, str) or isinstance(ocr_str, str):
+            return self.RESULT_N
+        if kwargs.get('is_passport'):
+            input_obj = re.search(r'[a-zA-Z]]!', input_str)
+            if input_obj:
+                input_s = input_obj.group()
+                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
+            else:
+                return self.build_res((input_str == ocr_str))
+        else:
+            if re.search(r'[a-zA-Z]]', input_str):
+                return self.RESULT_NA
+            input_s = input_str.translate(self.TRANS)
+            ocr_s = ocr_str.translate(self.TRANS)
+            return self.build_res(input_s == ocr_s)
    def date_compare(self, input_str, ocr_str, **kwargs):
+        if not isinstance(input_str, str) or isinstance(ocr_str, str):
+            return self.RESULT_N
        if kwargs.get('long', False) and '长期' in ocr_str:
            return '2099-12-31'
        if kwargs.get('ocr_split', False):
@@ -42,10 +73,14 @@ class Comparison:
        return self.build_res(input_str == ocr_str)
    def rmb_compare(self, input_str, ocr_str, **kwargs):
+        if not isinstance(input_str, str) or 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)
    def type_compare(self, input_str, ocr_str, **kwargs):
+        if not isinstance(input_str, str) or isinstance(ocr_str, str):
+            return self.RESULT_N
        for map_tuple in self.TYPE_MAPPING:
            if re.search(map_tuple[0], ocr_str) is not None:
                compare_str = map_tuple[1]