change data folder

@@ -29,8 +29,10 @@ sftp-config.json
 *.sqlite3
 *.sqlite3
 conf/*
 conf/*
 data/*
 data/*
+ocr/*
 
 
 # 脚本
 # 脚本
 src/*.sh
 src/*.sh
 
 
-test*
\ No newline at end of file
+test*
+folder_ocr_process.py
\ No newline at end of file

@@ -16,6 +16,9 @@ SPLIT_STR = '_'
 BUSINESS_TYPE_LIST = [HIL_PREFIX, AFC_PREFIX]
 BUSINESS_TYPE_LIST = [HIL_PREFIX, AFC_PREFIX]
 HIL_SET = {'HIL', 'HIl', 'HiL', 'Hil', 'hIL', 'hIl', 'hiL', 'hil', 'CO00002'}
 HIL_SET = {'HIL', 'HIl', 'HiL', 'Hil', 'hIL', 'hIl', 'hiL', 'hil', 'CO00002'}
 
 
+PRIORITY_WORDS = {'muw', 'MUW'}
+TMP_DIR_NAME = 'tmp'
+
 # -------EDMS相关---------------------------------------------------------------------------------------------------
 # -------EDMS相关---------------------------------------------------------------------------------------------------
 
 
 SESSION_PREFIX = 'FHLSID'
 SESSION_PREFIX = 'FHLSID'

@@ -373,7 +373,7 @@ class Command(BaseCommand, LoggerMixin):
 
 
             try:
             try:
                 # 2. 从EDMS获取PDF文件
                 # 2. 从EDMS获取PDF文件
-                doc_data_path = os.path.join(self.data_dir, business_type, str(doc.id))
+                doc_data_path = os.path.join(self.data_dir, business_type, consts.TMP_DIR_NAME, str(doc.id))
                 os.makedirs(doc_data_path, exist_ok=True)
                 os.makedirs(doc_data_path, exist_ok=True)
                 pdf_path = os.path.join(doc_data_path, '{0}.pdf'.format(doc.id))
                 pdf_path = os.path.join(doc_data_path, '{0}.pdf'.format(doc.id))
                 self.pdf_download(doc, pdf_path)
                 self.pdf_download(doc, pdf_path)
@@ -579,7 +579,7 @@ class Command(BaseCommand, LoggerMixin):
 
 
                     # 4.2 重构Excel文件
                     # 4.2 重构Excel文件
                     doc, business_type = self.get_doc_object(task_str)
                     doc, business_type = self.get_doc_object(task_str)
-                    doc_data_path = os.path.join(self.data_dir, business_type, str(doc.id))
+                    doc_data_path = os.path.join(self.data_dir, business_type, consts.TMP_DIR_NAME, str(doc.id))
                     excel_path = os.path.join(doc_data_path, '{0}.xlsx'.format(doc.id))
                     excel_path = os.path.join(doc_data_path, '{0}.xlsx'.format(doc.id))
                     img_save_path = os.path.join(doc_data_path, 'img')
                     img_save_path = os.path.join(doc_data_path, 'img')
                     # wb.save(src_excel_path)
                     # wb.save(src_excel_path)

@@ -303,8 +303,8 @@ class BSWorkbook(Workbook):
             period_idx = len(res_str) - 3
             period_idx = len(res_str) - 3
             if res_str[period_idx] == '.' and res_str[period_idx - 1] in {',', '.'}:   # 364,.92 364..92
             if res_str[period_idx] == '.' and res_str[period_idx - 1] in {',', '.'}:   # 364,.92 364..92
                 res_str = '{0}{1}'.format(res_str[:period_idx - 1], res_str[period_idx:])
                 res_str = '{0}{1}'.format(res_str[:period_idx - 1], res_str[period_idx:])
-            elif res_str[period_idx] == ',':
-                if res_str[period_idx - 1] in {',', '.'}:  # 364.,92  364,,92
+            elif res_str[period_idx] in {',', ':', '：'}:
+                if res_str[period_idx - 1] in {',', '.', ':', '：'}:  # 364.,92  364,,92
                     pre_idx = period_idx - 1
                     pre_idx = period_idx - 1
                 else:  # 364,92
                 else:  # 364,92
                     pre_idx = period_idx
                     pre_idx = period_idx

@@ -178,6 +178,7 @@ class PriorityDocView(GenericView, DocHandler):
         application_info = args.get('APPLICATION_INFORMATION')
         application_info = args.get('APPLICATION_INFORMATION')
         application_id = application_info.get('APPLICATION_ID')
         application_id = application_info.get('APPLICATION_ID')
         submit_datetime = application_info.get('SUBMIT_DATETIME')
         submit_datetime = application_info.get('SUBMIT_DATETIME')
+        intermediate_decision = application_info.get('INTERMEDIATE_DECISION')
         entity = application_info.get('ENTITY')
         entity = application_info.get('ENTITY')
         if submit_datetime.utcoffset() is not None:
         if submit_datetime.utcoffset() is not None:
             submit_datetime = timezone.make_naive(submit_datetime, timezone.get_current_timezone())
             submit_datetime = timezone.make_naive(submit_datetime, timezone.get_current_timezone())
@@ -187,9 +188,14 @@ class PriorityDocView(GenericView, DocHandler):
             rating=application_info.get('RATING'),
             rating=application_info.get('RATING'),
             application_id=application_id,
             application_id=application_id,
             application_version=application_info.get('APPLICATION_VERSION'),
             application_version=application_info.get('APPLICATION_VERSION'),
-            intermediate_decision=application_info.get('INTERMEDIATE_DECISION'),
+            intermediate_decision=intermediate_decision,
             submit_datetime=submit_datetime,
             submit_datetime=submit_datetime,
         )
         )
+
+        if intermediate_decision not in consts.PRIORITY_WORDS:
+            self.running_log.info('[priority doc skip] [args={0}]'.format(args))
+            return response.ok()
+
         _, created = PriorityApplication.objects.update_or_create(application_id=application_id,
         _, created = PriorityApplication.objects.update_or_create(application_id=application_id,
                                                                   defaults={'on_off': True})
                                                                   defaults={'on_off': True})
         if created:
         if created:
@@ -282,7 +288,8 @@ class DocView(GenericView, DocHandler):
         if not pdf_file.name.endswith('pdf'):
         if not pdf_file.name.endswith('pdf'):
             self.invalid_params(msg='invalid params: not a PDF file')
             self.invalid_params(msg='invalid params: not a PDF file')
 
 
-        tmp_save_path = os.path.join(conf.DATA_DIR, '{0}.pdf'.format(metadata_version_id))
+        business_type = random.choice(consts.BUSINESS_TYPE_LIST)
+        tmp_save_path = os.path.join(conf.DATA_DIR, business_type, '{0}.pdf'.format(metadata_version_id))
         file_write(pdf_file, tmp_save_path)
         file_write(pdf_file, tmp_save_path)
 
 
         try:
         try:
@@ -302,7 +309,6 @@ class DocView(GenericView, DocHandler):
         upload_finish_time = timezone.now()
         upload_finish_time = timezone.now()
         document_scheme = random.choice(consts.DOC_SCHEME_LIST)
         document_scheme = random.choice(consts.DOC_SCHEME_LIST)
         data_source = random.choice(consts.DATA_SOURCE_LIST)
         data_source = random.choice(consts.DATA_SOURCE_LIST)
-        business_type = random.choice(consts.BUSINESS_TYPE_LIST)
         UploadDocRecords.objects.create(
         UploadDocRecords.objects.create(
             metadata_version_id=metadata_version_id,
             metadata_version_id=metadata_version_id,
             application_id=application_id,
             application_id=application_id,
@@ -333,7 +339,7 @@ class DocView(GenericView, DocHandler):
         )
         )
 
 
         # 3.pdf文件移动
         # 3.pdf文件移动
-        save_dir_path = os.path.join(conf.DATA_DIR, business_type, str(doc.id))
+        save_dir_path = os.path.join(conf.DATA_DIR, business_type, consts.TMP_DIR_NAME, str(doc.id))
         save_file_path = os.path.join(save_dir_path, '{0}.pdf'.format(doc.id))
         save_file_path = os.path.join(save_dir_path, '{0}.pdf'.format(doc.id))
         os.makedirs(save_dir_path, exist_ok=True)
         os.makedirs(save_dir_path, exist_ok=True)
         # file_write(pdf_file, save_file_path)
         # file_write(pdf_file, save_file_path)

@@ -455,6 +455,30 @@ afc_sql = """
         on afc_doc (start_time, end_time);
         on afc_doc (start_time, end_time);
 """
 """
 
 
+keywords_sql = """
+    INSERT INTO afc.dbo.keywords (keyword, type, update_time, create_time) VALUES
+    (N'利息', 0, N'2020-11-09 16:14:58.000', N'2020-11-09 16:14:59.000'),
+    (N'结息', 0, N'2020-11-09 16:14:58.000', N'2020-11-09 16:14:59.000'),
+    (N'工资', 1, N'2020-11-09 16:14:58.000', N'2020-11-09 16:14:59.000'),
+    (N'代发', 1, N'2020-11-09 16:14:58.000', N'2020-11-09 16:14:59.000'),
+    (N'养老保险', 1, N'2020-11-09 16:14:58.000', N'2020-11-09 16:14:59.000'),
+    (N'奖金', 1, N'2020-11-09 16:14:58.000', N'2020-11-09 16:14:59.000'),
+    (N'理财', 2, N'2020-11-09 16:14:58.000', N'2020-11-09 16:14:59.000'),
+    (N'赎回', 2, N'2020-11-09 16:14:58.000', N'2020-11-09 16:14:59.000'),
+    (N'微信', 2, N'2020-11-09 16:14:58.000', N'2020-11-09 16:14:59.000'),
+    (N'支付宝', 2, N'2020-11-09 16:14:58.000', N'2020-11-09 16:14:59.000'),
+    (N'财付通', 2, N'2020-11-09 16:14:58.000', N'2020-11-09 16:14:59.000'),
+    (N'放款', 2, N'2020-11-09 16:14:58.000', N'2020-11-09 16:14:59.000'),
+    (N'还款', 2, N'2020-11-09 16:14:58.000', N'2020-11-09 16:14:59.000'),
+    (N'贷款', 2, N'2020-11-09 16:14:58.000', N'2020-11-09 16:14:59.000'),
+    (N'银证转账', 2, N'2020-11-09 16:14:58.000', N'2020-11-09 16:14:59.000'),
+    (N'银行卡户名（姓名）', 2, N'2020-11-09 16:14:58.000', N'2020-11-09 16:14:59.000'),
+    (N'转账/转账', 3, N'2020-11-09 16:14:58.000', N'2020-11-09 16:14:59.000'),
+    (N'商品/线下', 3, N'2020-11-09 16:14:58.000', N'2020-11-09 16:14:59.000'),
+    (N'转账', 3, N'2020-11-09 16:14:58.000', N'2020-11-09 16:14:59.000'),
+    (N'二维码收款', 3, N'2020-11-09 16:14:58.000', N'2020-11-09 16:14:59.000');
+"""
+
 hil_cnxn = pyodbc.connect('DRIVER={ODBC Driver 17 for SQL Server};', autocommit=True)
 hil_cnxn = pyodbc.connect('DRIVER={ODBC Driver 17 for SQL Server};', autocommit=True)
 
 
 hil_cursor = hil_cnxn.cursor()
 hil_cursor = hil_cnxn.cursor()
@@ -468,6 +492,7 @@ afc_cnxn = pyodbc.connect('DRIVER={ODBC Driver 17 for SQL Server};', autocommit=
 
 
 afc_cursor = afc_cnxn.cursor()
 afc_cursor = afc_cnxn.cursor()
 afc_cursor.execute(afc_sql)
 afc_cursor.execute(afc_sql)
+afc_cursor.execute(keywords_sql)
 
 
 afc_cursor.close()
 afc_cursor.close()
 afc_cnxn.close()
 afc_cnxn.close()