优化与调整

This commit is contained in:
2024-05-29 11:31:22 +08:00
parent 1eeb2b1544
commit a94bb613a9
5 changed files with 61 additions and 19 deletions

View File

@@ -3,6 +3,7 @@ import logging
from time import sleep
from paddlenlp import Taskflow
from sqlalchemy import update
from config.keys import PATIENT_NAME, ADMISSION_DATE, DISCHARGE_DATE, MEDICAL_EXPENSES, PERSONAL_CASH_PAYMENT, \
PERSONAL_ACCOUNT_PAYMENT, PERSONAL_FUNDED_AMOUNT, MEDICAL_INSURANCE_TYPE, HOSPITAL, DEPARTMENT, DOCTOR
@@ -16,7 +17,7 @@ from photo_review.entity.zx_ie_settlement import ZxIeSettlement
from photo_review.entity.zx_ocr import ZxOcr
from photo_review.entity.zx_phhd import ZxPhhd
from photo_review.entity.zx_phrec import ZxPhrec
from photo_review.util.data_util import handle_date, handle_decimal
from photo_review.util.data_util import handle_date, handle_decimal, handle_department
from photo_review.util.ucloud import get_private_url
@@ -27,6 +28,7 @@ def information_extraction(schema, phrecs, task_path):
pic_path = get_private_url(phrec.cfjaddress)
if pic_path:
ie = Taskflow("information_extraction", schema=schema, model="uie-x-base", task_path=task_path)
# 批量抽取写法:(ie([{"doc": "./data/6.jpg"}, {"doc": "./data/7.jpg"}])
result = ie({"doc": pic_path})
result_json = json.dumps(result, ensure_ascii=False)
@@ -173,13 +175,18 @@ def photo_review(pk_phhd):
discharge_data["hospital"] = yljg.name
department_value = get_values_of_keys(discharge_record_ie_result, department_key)
if department_value:
session = MysqlSession()
ylks = session.query(BdYlks.pk_ylks, BdYlks.name) \
.filter(BdYlks.name.in_(department_value)).limit(1).one_or_none()
session.close()
if ylks:
discharge_data["pk_ylks"] = ylks.pk_ylks
discharge_data["department"] = ylks.name
department_values = []
for dept in department_value:
department_values += handle_department(dept)
department_values = list(set(department_values))
if department_values:
session = MysqlSession()
ylks = session.query(BdYlks.pk_ylks, BdYlks.name) \
.filter(BdYlks.name.in_(department_values)).limit(1).one_or_none()
session.close()
if ylks:
discharge_data["pk_ylks"] = ylks.pk_ylks
discharge_data["department"] = ylks.name
save_or_update_ie(ZxIeDischarge, pk_phhd, discharge_data)
cost_list_ie_result = information_extraction(cost_list_schema, cost_list, "config/model/cost_list_model")
@@ -192,19 +199,16 @@ def photo_review(pk_phhd):
}
cost_data["admission_date"] = handle_date(cost_data["admission_date_str"])
cost_data["discharge_date"] = handle_date(cost_data["discharge_date_str"])
cost_data["medical_expenses"] = handle_date(cost_data["medical_expenses_str"])
cost_data["medical_expenses"] = handle_decimal(cost_data["medical_expenses_str"])
save_or_update_ie(ZxIeCost, pk_phhd, cost_data)
def main():
# 最后处理的报销案子pk
last_pk_phhd = 0
# 持续检测新案子
while 1:
session = MysqlSession()
phhds = session.query(ZxPhhd.pk_phhd) \
.filter(ZxPhhd.pk_phhd > last_pk_phhd) \
.filter(ZxPhhd.cStatus == '2') \
.filter(ZxPhhd.exsuccess_flag == '1') \
.limit(PHHD_BATCH_SIZE) \
.all()
session.close()
@@ -212,7 +216,13 @@ def main():
for phhd in phhds:
pk_phhd = phhd.pk_phhd
photo_review(pk_phhd)
last_pk_phhd = pk_phhd
# 识别完成更新标识
session = MysqlSession()
stmt = (update(ZxPhhd).where(ZxPhhd.pk_phhd == pk_phhd).values(exsuccess_flag=8))
session.execute(stmt)
session.commit()
session.close()
else:
# 没有查询到新案子,等待一段时间后再查
sleep_minutes = SLEEP_MINUTES