增加对姓名的处理,新增和更新数据库时插入当前时间

This commit is contained in:
2024-05-29 14:06:44 +08:00
parent 5d6ab60496
commit 451b693a54
2 changed files with 18 additions and 9 deletions

View File

@@ -1,6 +1,6 @@
import json import json
import logging import logging
from time import sleep from time import sleep, time
from paddlenlp import Taskflow from paddlenlp import Taskflow
from sqlalchemy import update from sqlalchemy import update
@@ -17,7 +17,7 @@ from photo_review.entity.zx_ie_settlement import ZxIeSettlement
from photo_review.entity.zx_ocr import ZxOcr from photo_review.entity.zx_ocr import ZxOcr
from photo_review.entity.zx_phhd import ZxPhhd from photo_review.entity.zx_phhd import ZxPhhd
from photo_review.entity.zx_phrec import ZxPhrec from photo_review.entity.zx_phrec import ZxPhrec
from photo_review.util.data_util import handle_date, handle_decimal, handle_department from photo_review.util.data_util import handle_date, handle_decimal, handle_department, handle_name
from photo_review.util.ucloud import get_private_url from photo_review.util.ucloud import get_private_url
@@ -37,7 +37,7 @@ def information_extraction(schema, phrecs, task_path):
# 提取完保存每张图片的结果 # 提取完保存每张图片的结果
session = MysqlSession() session = MysqlSession()
zx_ocr = ZxOcr(pk_phhd=phrec.pk_phhd, pk_phrec=phrec.pk_phrec, cfjaddress=phrec.cfjaddress, zx_ocr = ZxOcr(pk_phhd=phrec.pk_phhd, pk_phrec=phrec.pk_phrec, cfjaddress=phrec.cfjaddress,
content=result_json) content=result_json, create_time=time(), update_time=time())
session.add(zx_ocr) session.add(zx_ocr)
session.commit() session.commit()
session.close() session.close()
@@ -81,9 +81,14 @@ def save_or_update_ie(table, pk_phhd, data):
session = MysqlSession() session = MysqlSession()
db_data = session.query(table).filter_by(pk_phhd=pk_phhd).one_or_none() db_data = session.query(table).filter_by(pk_phhd=pk_phhd).one_or_none()
if db_data: if db_data:
# 更新
db_data.update_time = time()
for k, v in data.items(): for k, v in data.items():
setattr(db_data, k, v) setattr(db_data, k, v)
else: else:
# 新增
obj.create_time = time()
obj.update_time = time()
session.add(obj) session.add(obj)
session.commit() session.commit()
session.close() session.close()
@@ -133,7 +138,7 @@ def photo_review(pk_phhd):
"config/model/settlement_list_model") "config/model/settlement_list_model")
settlement_data = { settlement_data = {
"pk_phhd": pk_phhd, "pk_phhd": pk_phhd,
"name": get_best_value_in_keys(settlement_list_ie_result, name_key), "name": handle_name(get_best_value_in_keys(settlement_list_ie_result, name_key)),
"admission_date_str": get_best_value_in_keys(settlement_list_ie_result, admission_date_key), "admission_date_str": get_best_value_in_keys(settlement_list_ie_result, admission_date_key),
"discharge_date_str": get_best_value_in_keys(settlement_list_ie_result, discharge_date_key), "discharge_date_str": get_best_value_in_keys(settlement_list_ie_result, discharge_date_key),
"medical_expenses_str": get_best_value_in_keys(settlement_list_ie_result, medical_expenses_key), "medical_expenses_str": get_best_value_in_keys(settlement_list_ie_result, medical_expenses_key),
@@ -157,10 +162,10 @@ def photo_review(pk_phhd):
"pk_phhd": pk_phhd, "pk_phhd": pk_phhd,
"hospital": get_best_value_in_keys(discharge_record_ie_result, hospital_key), "hospital": get_best_value_in_keys(discharge_record_ie_result, hospital_key),
"department": get_best_value_in_keys(discharge_record_ie_result, department_key), "department": get_best_value_in_keys(discharge_record_ie_result, department_key),
"name": get_best_value_in_keys(discharge_record_ie_result, name_key), "name": handle_name(get_best_value_in_keys(discharge_record_ie_result, name_key)),
"admission_date_str": get_best_value_in_keys(discharge_record_ie_result, admission_date_key), "admission_date_str": get_best_value_in_keys(discharge_record_ie_result, admission_date_key),
"discharge_date_str": get_best_value_in_keys(discharge_record_ie_result, discharge_date_key), "discharge_date_str": get_best_value_in_keys(discharge_record_ie_result, discharge_date_key),
"doctor": get_best_value_in_keys(discharge_record_ie_result, doctor_key) "doctor": handle_name(get_best_value_in_keys(discharge_record_ie_result, doctor_key))
} }
discharge_data["admission_date"] = handle_date(discharge_data["admission_date_str"]) discharge_data["admission_date"] = handle_date(discharge_data["admission_date_str"])
discharge_data["discharge_date"] = handle_date(discharge_data["discharge_date_str"]) discharge_data["discharge_date"] = handle_date(discharge_data["discharge_date_str"])
@@ -192,7 +197,7 @@ def photo_review(pk_phhd):
cost_list_ie_result = information_extraction(cost_list_schema, cost_list, "config/model/cost_list_model") cost_list_ie_result = information_extraction(cost_list_schema, cost_list, "config/model/cost_list_model")
cost_data = { cost_data = {
"pk_phhd": pk_phhd, "pk_phhd": pk_phhd,
"name": get_best_value_in_keys(cost_list_ie_result, name_key), "name": handle_name(get_best_value_in_keys(cost_list_ie_result, name_key)),
"admission_date_str": get_best_value_in_keys(cost_list_ie_result, admission_date_key), "admission_date_str": get_best_value_in_keys(cost_list_ie_result, admission_date_key),
"discharge_date_str": get_best_value_in_keys(cost_list_ie_result, discharge_date_key), "discharge_date_str": get_best_value_in_keys(cost_list_ie_result, discharge_date_key),
"medical_expenses_str": get_best_value_in_keys(cost_list_ie_result, medical_expenses_key) "medical_expenses_str": get_best_value_in_keys(cost_list_ie_result, medical_expenses_key)

View File

@@ -79,5 +79,9 @@ def handle_department(string):
result.append(pure_string) result.append(pure_string)
return result return result
if __name__ == '__main__':
print(handle_decimal("~202312167,214.83金额1,920.008,888.38740.00交三医专用章广东省医疗压院收票据电子政策性游其他支付0.00医保类型跨省异地预缴金额11564.15备注病历号2165996治疗费手术费护理费中成药费项目名称政电电子“东省结算方式个人账户支付0.00医保编号补缴金额0.00住院号216599612,971.56金额7,483.062,192.0211.96复核人小写68,214.38")) # 处理姓名类数据
def handle_name(string):
if not string:
return ""
return re.sub(r'[^⺀-鿿·]', '', string)