Revert "尝试使用多进程来优化整体识别速度"

This reverts commit 66c0d5e950.
This commit is contained in:
2024-07-01 13:47:51 +08:00
parent 66c0d5e950
commit 8b4d67fd2b

View File

@@ -1,4 +1,3 @@
import concurrent.futures
import json import json
import logging import logging
import math import math
@@ -291,10 +290,25 @@ def save_or_update_ie(table, pk_phhd, data):
session.close() session.close()
def settlement_task(pk_phhd, settlement_list): def photo_review(pk_phhd, task_flows):
settlement_list_ie_result = information_extraction( settlement_list = []
Taskflow("information_extraction", schema=SETTLEMENT_LIST_SCHEMA, model="uie-x-base", discharge_record = []
task_path="config/model/settlement_list_model", layout_analysis=LAYOUT_ANALYSIS), settlement_list) cost_list = []
session = MysqlSession()
phrecs = session.query(ZxPhrec.pk_phrec, ZxPhrec.pk_phhd, ZxPhrec.cRectype, ZxPhrec.cfjaddress) \
.filter(ZxPhrec.pk_phhd == pk_phhd) \
.all()
session.close()
for phrec in phrecs:
if phrec.cRectype == "1":
settlement_list.append(phrec)
elif phrec.cRectype == "3":
discharge_record.append(phrec)
elif phrec.cRectype == "4":
cost_list.append(phrec)
settlement_list_ie_result = information_extraction(task_flows[0], settlement_list)
settlement_data = { settlement_data = {
"pk_phhd": pk_phhd, "pk_phhd": pk_phhd,
"name": handle_name(get_best_value_in_keys(settlement_list_ie_result, PATIENT_NAME)), "name": handle_name(get_best_value_in_keys(settlement_list_ie_result, PATIENT_NAME)),
@@ -320,11 +334,7 @@ def settlement_task(pk_phhd, settlement_list):
settlement_data["personal_funded_amount"] = handle_decimal(settlement_data["personal_funded_amount_str"]) settlement_data["personal_funded_amount"] = handle_decimal(settlement_data["personal_funded_amount_str"])
save_or_update_ie(ZxIeSettlement, pk_phhd, settlement_data) save_or_update_ie(ZxIeSettlement, pk_phhd, settlement_data)
discharge_record_ie_result = information_extraction(task_flows[1], discharge_record)
def discharge_task(pk_phhd, discharge_record):
discharge_record_ie_result = information_extraction(
Taskflow("information_extraction", schema=DISCHARGE_RECORD_SCHEMA, model="uie-x-base",
task_path="config/model/discharge_record_model", layout_analysis=LAYOUT_ANALYSIS), discharge_record)
discharge_data = { discharge_data = {
"pk_phhd": pk_phhd, "pk_phhd": pk_phhd,
"hospital": handle_hospital(get_best_value_in_keys(discharge_record_ie_result, HOSPITAL)), "hospital": handle_hospital(get_best_value_in_keys(discharge_record_ie_result, HOSPITAL)),
@@ -361,11 +371,7 @@ def discharge_task(pk_phhd, discharge_record):
discharge_data["department"] = ylks.name discharge_data["department"] = ylks.name
save_or_update_ie(ZxIeDischarge, pk_phhd, discharge_data) save_or_update_ie(ZxIeDischarge, pk_phhd, discharge_data)
cost_list_ie_result = information_extraction(task_flows[2], cost_list)
def cost_task(pk_phhd, cost_list):
cost_list_ie_result = information_extraction(
Taskflow("information_extraction", schema=COST_LIST_SCHEMA, model="uie-x-base",
task_path="config/model/cost_list_model", layout_analysis=LAYOUT_ANALYSIS), cost_list)
cost_data = { cost_data = {
"pk_phhd": pk_phhd, "pk_phhd": pk_phhd,
"name": handle_name(get_best_value_in_keys(cost_list_ie_result, PATIENT_NAME)), "name": handle_name(get_best_value_in_keys(cost_list_ie_result, PATIENT_NAME)),
@@ -379,30 +385,6 @@ def cost_task(pk_phhd, cost_list):
save_or_update_ie(ZxIeCost, pk_phhd, cost_data) save_or_update_ie(ZxIeCost, pk_phhd, cost_data)
def photo_review(pk_phhd):
settlement_list = []
discharge_record = []
cost_list = []
session = MysqlSession()
phrecs = session.query(ZxPhrec.pk_phrec, ZxPhrec.pk_phhd, ZxPhrec.cRectype, ZxPhrec.cfjaddress) \
.filter(ZxPhrec.pk_phhd == pk_phhd) \
.all()
session.close()
for phrec in phrecs:
if phrec.cRectype == "1":
settlement_list.append(phrec)
elif phrec.cRectype == "3":
discharge_record.append(phrec)
elif phrec.cRectype == "4":
cost_list.append(phrec)
with concurrent.futures.ProcessPoolExecutor() as executor:
executor.submit(settlement_task, pk_phhd, settlement_list)
executor.submit(discharge_task, pk_phhd, discharge_record)
executor.submit(cost_task, pk_phhd, cost_list)
def main(): def main():
# 持续检测新案子 # 持续检测新案子
while 1: while 1:
@@ -411,10 +393,18 @@ def main():
phhds = session.query(ZxPhhd.pk_phhd).filter(ZxPhhd.exsuccess_flag == '1').limit(PHHD_BATCH_SIZE).all() phhds = session.query(ZxPhhd.pk_phhd).filter(ZxPhhd.exsuccess_flag == '1').limit(PHHD_BATCH_SIZE).all()
session.close() session.close()
if phhds: if phhds:
ie_task_flows = [
Taskflow("information_extraction", schema=SETTLEMENT_LIST_SCHEMA, model="uie-x-base",
task_path="config/model/settlement_list_model", layout_analysis=LAYOUT_ANALYSIS),
Taskflow("information_extraction", schema=DISCHARGE_RECORD_SCHEMA, model="uie-x-base",
task_path="config/model/discharge_record_model", layout_analysis=LAYOUT_ANALYSIS),
Taskflow("information_extraction", schema=COST_LIST_SCHEMA, model="uie-x-base",
task_path="config/model/cost_list_model", layout_analysis=LAYOUT_ANALYSIS),
]
for phhd in phhds: for phhd in phhds:
pk_phhd = phhd.pk_phhd pk_phhd = phhd.pk_phhd
logging.info(f"开始识别:{pk_phhd}") logging.info(f"开始识别:{pk_phhd}")
photo_review(pk_phhd) photo_review(pk_phhd, ie_task_flows)
# 识别完成更新标识 # 识别完成更新标识
session = MysqlSession() session = MysqlSession()