优化案子处理逻辑

photo_review/auto_photo_review.py

@@ -1,7 +1,4 @@
-import json
 import logging
-import os
-import tempfile
 import time
 from collections import defaultdict
 from time import sleep
@@ -10,72 +7,24 @@ import cv2
 import fitz
 import jieba
 import numpy as np
-import requests
 import zxingcpp
 from rapidfuzz import process, fuzz
 from sqlalchemy import update
 
 from db import MysqlSession
-from db.mysql import BdYljg, BdYlks, ZxIeResult, ZxIeCost, ZxIeDischarge, ZxIeSettlement, ZxPhhd, ZxPhrec, ZxIeReview
+from db.mysql import BdYljg, BdYlks, ZxIeCost, ZxIeDischarge, ZxIeSettlement, ZxPhhd, ZxPhrec, ZxIeReview
 from log import HOSTNAME
 from photo_review import PHHD_BATCH_SIZE, SLEEP_MINUTES, HOSPITAL_ALIAS, HOSPITAL_FILTER, DEPARTMENT_ALIAS, \
     DEPARTMENT_FILTER
-from services.paddle_services import PATIENT_NAME, ADMISSION_DATE, DISCHARGE_DATE, MEDICAL_EXPENSES, \
-    PERSONAL_CASH_PAYMENT, PERSONAL_ACCOUNT_PAYMENT, PERSONAL_FUNDED_AMOUNT, MEDICAL_INSURANCE_TYPE, HOSPITAL, \
-    DEPARTMENT, DOCTOR, ADMISSION_ID, SETTLEMENT_ID, AGE, UPPERCASE_MEDICAL_EXPENSES
+from services.paddle_services import IE_KEY
 from ucloud import ufile
 from util import image_util, common_util, html_util, model_util
 from util.data_util import handle_date, handle_decimal, parse_department, handle_name, handle_insurance_type, \
     handle_original_data, handle_hospital, handle_department, handle_id, handle_age, parse_money, parse_hospital
 
 
-# 合并信息抽取结果
-def merge_result(result1, result2):
-    for key in result2:
-        result1[key] = result1.get(key, []) + result2[key]
-    return result1
-
-
-def ie_temp_image(ie, ocr, image):
-    with tempfile.NamedTemporaryFile(delete=False, suffix=".jpg") as temp_file:
-        cv2.imwrite(temp_file.name, image)
-
-    ie_result = []
-    try:
-        layout = common_util.get_ocr_layout(ocr, temp_file.name)
-        if not layout:
-            # 无识别结果
-            ie_result = []
-        else:
-            ie_result = ie({"doc": temp_file.name, "layout": layout})[0]
-    except Exception as e:
-        logging.error("信息抽取时出错", exc_info=e)
-    finally:
-        try:
-            os.remove(temp_file.name)
-        except Exception as e:
-            logging.info(f"删除临时文件 {temp_file.name} 时出错", exc_info=e)
-    return ie_result
-
-
-# 关键信息提取
-def request_ie_result(task_enum, phrecs):
-    url = task_enum.request_url()
-    identity = int(time.time())
-    images = []
-    for phrec in phrecs:
-        images.append({"name": phrec.cfjaddress, "pk": phrec.pk_phrec})
-    payload = {"images": images, "schema": task_enum.schema(), "pk_phhd": phrecs[0].pk_phhd, "identity": identity}
-    response = requests.post(url, json=payload)
-
-    if response.status_code == 200:
-        return response.json()["data"]
-    else:
-        raise Exception(f"请求信息抽取结果失败，状态码：{response.status_code}")
-
-
 # 尝试从二维码中获取高清图片
-def get_better_image_from_qrcode(image, image_id, dpi=150):
+def get_better_image_from_qrcode(img_path, image_id, dpi=150):
     def _parse_pdf_url(pdf_url_to_parse):
         pdf_file = None
         local_pdf_path = None
@@ -95,7 +44,10 @@ def get_better_image_from_qrcode(image, image_id, dpi=150):
             # 将渲染结果转换为OpenCV兼容的格式
             img = np.frombuffer(pix.samples, dtype=np.uint8).reshape((pix.height, pix.width, -1))
             img = cv2.cvtColor(img, cv2.COLOR_RGB2BGR)
-            return img, page.get_text()
+            img_name, img_ext = image_util.parse_save_path(img_path)
+            better_img_path = image_util.get_save_path(f'{img_name}.better.{img_ext}')
+            cv2.imwrite(better_img_path, img)
+            return better_img_path, page.get_text()
         except Exception as ex:
             logging.getLogger('error').error('解析pdf失败！', exc_info=ex)
             return None, None
@@ -107,7 +59,8 @@ def get_better_image_from_qrcode(image, image_id, dpi=150):
 
     jsczt_base_url = 'http://einvoice.jsczt.cn'
     try:
-        results = zxingcpp.read_barcodes(image)
+        img = cv2.imread(img_path)
+        results = zxingcpp.read_barcodes(img, text_mode=zxingcpp.TextMode.HRI)
     except Exception as e:
         logging.getLogger('error').info('二维码识别失败', exc_info=e)
         results = []
@@ -145,106 +98,52 @@ def get_better_image_from_qrcode(image, image_id, dpi=150):
 
 
 # 关键信息提取
-def information_extraction(ie, phrecs, identity):
-    result = {}
-    for phrec in phrecs:
+def information_extraction(phrec, identity):
+    """
+    处理单张图片
+    :param phrec:
+    :return:
+    """
+    img_path = image_util.get_img_path(phrec.cfjaddress)
+    if not img_path:
         img_url = ufile.get_private_url(phrec.cfjaddress)
-        if not img_url:
-            continue
-
         img_path = image_util.save_to_local(img_url)
-        image = cv2.imread(img_path)
-        # 尝试从二维码中获取高清图片
-        better_image, text = get_better_image_from_qrcode(image, phrec.cfjaddress)
-        if phrec.cRectype != '1':
-            better_image = None  # 非结算单暂时不进行替换
-        zx_ie_results = []
-        if better_image is not None:
-            img_angle = '0'
-            image = better_image
-            if text:
-                info_extract = ie(text)[0]
-            else:
-                info_extract = ie_temp_image(ie, OCR, image)
-            if not info_extract:
-                continue
 
-            ie_result = {'result': info_extract, 'angle': img_angle}
-            now = common_util.get_default_datetime()
-            result_json = json.dumps(ie_result['result'], ensure_ascii=False)
-            if len(result_json) > 5000:
-                result_json = result_json[:5000]
-            zx_ie_results.append(ZxIeResult(pk_phhd=phrec.pk_phhd, pk_phrec=phrec.pk_phrec, id=identity,
-                                            cfjaddress=phrec.cfjaddress, content=result_json,
-                                            rotation_angle=int(ie_result['angle']),
-                                            x_offset=0, y_offset=0, create_time=now,
-                                            creator=HOSTNAME, update_time=now, updater=HOSTNAME))
-
-            result = merge_result(result, ie_result['result'])
+    # 尝试从二维码中获取高清图片
+    better_img_path, text = get_better_image_from_qrcode(img_path, phrec.cfjaddress)
+    if phrec.cRectype != '1':
+        better_img_path = None  # 非结算单暂时不进行替换
+    if better_img_path is not None:
+        if text:
+            info_extract = model_util.ie_settlement_text(text)[0]
         else:
-            target_images = model_util.request_book_areas(img_path)  # 识别文档区域并裁剪
-            angle_count = defaultdict(int, {'0': 0})  # 分割后图片的最优角度统计
-            for target_image in target_images:
-                dewarped_image = model_util.dewarp(target_image)  # 去扭曲
-                angles = model_util.clas_orientation(dewarped_image)
+            info_extract = model_util.ie_settlement(better_img_path,
+                                                    common_util.ocr_result_to_layout(model_util.ocr(better_img_path)))
 
-                split_results = image_util.split(dewarped_image)
-                for split_result in split_results:
-                    if split_result['img'] is None or split_result['img'].size == 0:
-                        continue
-                    rotated_img = image_util.rotate(split_result['img'], int(angles[0]))
-                    ie_results = [{'result': ie_temp_image(ie, OCR, rotated_img), 'angle': angles[0]}]
-                    if not ie_results[0]['result'] or len(ie_results[0]['result']) < len(ie.kwargs.get('schema')):
-                        rotated_img = image_util.rotate(split_result['img'], int(angles[1]))
-                        ie_results.append({'result': ie_temp_image(ie, OCR, rotated_img), 'angle': angles[1]})
-                    now = common_util.get_default_datetime()
-                    best_angle = ['0', 0]
-                    for ie_result in ie_results:
-                        if not ie_result['result']:
-                            continue
+        return '基本医保结算单', info_extract
+    else:
+        target_image = model_util.det_book(img_path)  # 识别文档区域并裁剪
+        dewarped_image = model_util.dewarp(target_image)  # 去扭曲
+        angles = model_util.clas_orientation(dewarped_image)
+        rotated_img = image_util.rotate(dewarped_image, int(angles[0]))
+        split_results = image_util.split(rotated_img)
+        ocr_result = []
+        for split_result in split_results:
+            if split_result['img'] is None:
+                continue
+            ocr_result += model_util.ocr(rotated_img)
+        ocr_text = common_util.ocr_result_to_text(ocr_result)
+        rec_type = model_util.clas_text(ocr_text) if ocr_text else None
+        if rec_type == '基本医保结算单':
+            info_extract = model_util.ie_settlement(rotated_img, common_util.ocr_result_to_layout(ocr_result))
+        elif rec_type == '出院记录':
+            info_extract = model_util.ie_discharge(rotated_img, common_util.ocr_result_to_layout(ocr_result))
+        elif rec_type == '费用清单':
+            info_extract = model_util.ie_cost(rotated_img, common_util.ocr_result_to_layout(ocr_result))
+        else:
+            info_extract = None
 
-                        result_json = json.dumps(ie_result['result'], ensure_ascii=False)
-                        if len(result_json) > 5000:
-                            result_json = result_json[:5000]
-                        zx_ie_results.append(ZxIeResult(pk_phhd=phrec.pk_phhd, pk_phrec=phrec.pk_phrec, id=identity,
-                                                        cfjaddress=phrec.cfjaddress, content=result_json,
-                                                        rotation_angle=int(ie_result['angle']),
-                                                        x_offset=split_result['x_offset'],
-                                                        y_offset=split_result['y_offset'], create_time=now,
-                                                        creator=HOSTNAME, update_time=now, updater=HOSTNAME))
-
-                        result = merge_result(result, ie_result['result'])
-
-                        if len(ie_result['result']) > best_angle[1]:
-                            best_angle = [ie_result['angle'], len(ie_result['result'])]
-
-                    angle_count[best_angle[0]] += 1
-            img_angle = max(angle_count, key=angle_count.get)
-
-        if img_angle != '0' or better_image is not None:
-            image = image_util.rotate(image, int(img_angle))
-            with tempfile.NamedTemporaryFile(delete=False, suffix='.jpg') as temp_file:
-                cv2.imwrite(temp_file.name, image)
-            try:
-                ufile.upload_file(phrec.cfjaddress, temp_file.name)
-                if img_angle != '0':
-                    logging.info(f'旋转图片[{phrec.cfjaddress}]替换成功，已旋转{img_angle}度。')
-                    # 修正旋转角度
-                    for zx_ie_result in zx_ie_results:
-                        zx_ie_result.rotation_angle -= int(img_angle)
-                else:
-                    logging.info(f'高清图片[{phrec.cfjaddress}]替换成功！')
-            except Exception as e:
-                logging.error(f'上传图片({phrec.cfjaddress})失败', exc_info=e)
-            finally:
-                common_util.delete_temp_file(temp_file.name)
-
-        session = MysqlSession()
-        session.add_all(zx_ie_results)
-        session.commit()
-        session.close()
-
-    return result
+        return rec_type, info_extract
 
 
 # 从keys中获取准确率最高的value
@@ -359,23 +258,24 @@ def search_department(department):
     return best_match
 
 
-def settlement_task(pk_phhd, settlement_list, identity):
-    settlement_list_ie_result = information_extraction(SETTLEMENT_IE, settlement_list, identity)
+def settlement_task(pk_phhd, settlement_list_ie_result):
     settlement_data = {
         "pk_phhd": pk_phhd,
-        "name": handle_name(get_best_value_in_keys(settlement_list_ie_result, PATIENT_NAME)),
-        "admission_date_str": handle_original_data(get_best_value_in_keys(settlement_list_ie_result, ADMISSION_DATE)),
-        "discharge_date_str": handle_original_data(get_best_value_in_keys(settlement_list_ie_result, DISCHARGE_DATE)),
+        "name": handle_name(get_best_value_in_keys(settlement_list_ie_result, IE_KEY['name'])),
+        "admission_date_str": handle_original_data(
+            get_best_value_in_keys(settlement_list_ie_result, IE_KEY['admission_date'])),
+        "discharge_date_str": handle_original_data(
+            get_best_value_in_keys(settlement_list_ie_result, IE_KEY['discharge_date'])),
         "personal_cash_payment_str": handle_original_data(
-            get_best_value_in_keys(settlement_list_ie_result, PERSONAL_CASH_PAYMENT)),
+            get_best_value_in_keys(settlement_list_ie_result, IE_KEY['personal_cash_payment'])),
         "personal_account_payment_str": handle_original_data(
-            get_best_value_in_keys(settlement_list_ie_result, PERSONAL_ACCOUNT_PAYMENT)),
+            get_best_value_in_keys(settlement_list_ie_result, IE_KEY['personal_account_payment'])),
         "personal_funded_amount_str": handle_original_data(
-            get_best_value_in_keys(settlement_list_ie_result, PERSONAL_FUNDED_AMOUNT)),
+            get_best_value_in_keys(settlement_list_ie_result, IE_KEY['personal_funded_amount'])),
         "medical_insurance_type_str": handle_original_data(
-            get_best_value_in_keys(settlement_list_ie_result, MEDICAL_INSURANCE_TYPE)),
-        "admission_id": handle_id(get_best_value_in_keys(settlement_list_ie_result, ADMISSION_ID)),
-        "settlement_id": handle_id(get_best_value_in_keys(settlement_list_ie_result, SETTLEMENT_ID)),
+            get_best_value_in_keys(settlement_list_ie_result, IE_KEY['medical_insurance_type'])),
+        "admission_id": handle_id(get_best_value_in_keys(settlement_list_ie_result, IE_KEY['admission_id'])),
+        "settlement_id": handle_id(get_best_value_in_keys(settlement_list_ie_result, IE_KEY['settlement_id'])),
     }
     settlement_data["admission_date"] = handle_date(settlement_data["admission_date_str"])
     settlement_data["admission_date"] = handle_date(settlement_data["admission_date_str"])
@@ -385,28 +285,30 @@ def settlement_task(pk_phhd, settlement_list, identity):
     settlement_data["personal_funded_amount"] = handle_decimal(settlement_data["personal_funded_amount_str"])
     settlement_data["medical_insurance_type"] = handle_insurance_type(settlement_data["medical_insurance_type_str"])
 
-    parse_money_result = parse_money(get_best_value_in_keys(settlement_list_ie_result, UPPERCASE_MEDICAL_EXPENSES),
-                                     get_best_value_in_keys(settlement_list_ie_result, MEDICAL_EXPENSES))
+    parse_money_result = parse_money(
+        get_best_value_in_keys(settlement_list_ie_result, IE_KEY['upper_case_medical_expenses']),
+        get_best_value_in_keys(settlement_list_ie_result, IE_KEY['medical_expenses']))
     settlement_data["medical_expenses_str"] = handle_original_data(parse_money_result[0])
     settlement_data["medical_expenses"] = parse_money_result[1]
     save_or_update_ie(ZxIeSettlement, pk_phhd, settlement_data)
     return settlement_data
 
 
-def discharge_task(pk_phhd, discharge_record, identity):
-    discharge_record_ie_result = information_extraction(DISCHARGE_IE, discharge_record, identity)
-    hospitals = get_values_of_keys(discharge_record_ie_result, HOSPITAL)
-    departments = get_values_of_keys(discharge_record_ie_result, DEPARTMENT)
+def discharge_task(pk_phhd, discharge_record_ie_result):
+    hospitals = get_values_of_keys(discharge_record_ie_result, IE_KEY['hospital'])
+    departments = get_values_of_keys(discharge_record_ie_result, IE_KEY['department'])
     discharge_data = {
         "pk_phhd": pk_phhd,
         "hospital": handle_hospital(",".join(hospitals)),
         "department": handle_department(",".join(departments)),
-        "name": handle_name(get_best_value_in_keys(discharge_record_ie_result, PATIENT_NAME)),
-        "admission_date_str": handle_original_data(get_best_value_in_keys(discharge_record_ie_result, ADMISSION_DATE)),
-        "discharge_date_str": handle_original_data(get_best_value_in_keys(discharge_record_ie_result, DISCHARGE_DATE)),
-        "doctor": handle_name(get_best_value_in_keys(discharge_record_ie_result, DOCTOR)),
-        "admission_id": handle_id(get_best_value_in_keys(discharge_record_ie_result, ADMISSION_ID)),
-        "age": handle_age(get_best_value_in_keys(discharge_record_ie_result, AGE)),
+        "name": handle_name(get_best_value_in_keys(discharge_record_ie_result, IE_KEY['name'])),
+        "admission_date_str": handle_original_data(
+            get_best_value_in_keys(discharge_record_ie_result, IE_KEY['admission_date'])),
+        "discharge_date_str": handle_original_data(
+            get_best_value_in_keys(discharge_record_ie_result, IE_KEY['discharge_date'])),
+        "doctor": handle_name(get_best_value_in_keys(discharge_record_ie_result, IE_KEY['doctor'])),
+        "admission_id": handle_id(get_best_value_in_keys(discharge_record_ie_result, IE_KEY['admission_id'])),
+        "age": handle_age(get_best_value_in_keys(discharge_record_ie_result, IE_KEY['age'])),
     }
     discharge_data["admission_date"] = handle_date(discharge_data["admission_date_str"])
     discharge_data["discharge_date"] = handle_date(discharge_data["discharge_date_str"])
@@ -466,14 +368,16 @@ def discharge_task(pk_phhd, discharge_record, identity):
     return discharge_data
 
 
-def cost_task(pk_phhd, cost_list, identity):
-    cost_list_ie_result = information_extraction(COST_IE, cost_list, identity)
+def cost_task(pk_phhd, cost_list_ie_result):
     cost_data = {
         "pk_phhd": pk_phhd,
-        "name": handle_name(get_best_value_in_keys(cost_list_ie_result, PATIENT_NAME)),
-        "admission_date_str": handle_original_data(get_best_value_in_keys(cost_list_ie_result, ADMISSION_DATE)),
-        "discharge_date_str": handle_original_data(get_best_value_in_keys(cost_list_ie_result, DISCHARGE_DATE)),
-        "medical_expenses_str": handle_original_data(get_best_value_in_keys(cost_list_ie_result, MEDICAL_EXPENSES))
+        "name": handle_name(get_best_value_in_keys(cost_list_ie_result, IE_KEY['name'])),
+        "admission_date_str": handle_original_data(
+            get_best_value_in_keys(cost_list_ie_result, IE_KEY['admission_date'])),
+        "discharge_date_str": handle_original_data(
+            get_best_value_in_keys(cost_list_ie_result, IE_KEY['discharge_date'])),
+        "medical_expenses_str": handle_original_data(
+            get_best_value_in_keys(cost_list_ie_result, IE_KEY['medical_expenses']))
     }
     cost_data["admission_date"] = handle_date(cost_data["admission_date_str"])
     cost_data["discharge_date"] = handle_date(cost_data["discharge_date_str"])
@@ -483,28 +387,39 @@ def cost_task(pk_phhd, cost_list, identity):
 
 
 def photo_review(pk_phhd, name):
-    settlement_list = []
-    discharge_record = []
-    cost_list = []
+    """
+    处理单个报销案子
+    :param pk_phhd: 报销单主键
+    :param name: 报销人姓名
+    """
+    settlement_result = defaultdict(list)
+    discharge_result = defaultdict(list)
+    cost_result = defaultdict(list)
 
     session = MysqlSession()
-    phrecs = session.query(ZxPhrec.pk_phrec, ZxPhrec.pk_phhd, ZxPhrec.cRectype, ZxPhrec.cfjaddress).filter(
+    phrecs = session.query(ZxPhrec.pk_phrec, 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)
-
     # 同一批图的标识
     identity = int(time.time())
-    settlement_data = settlement_task(pk_phhd, settlement_list, identity)
-    discharge_data = discharge_task(pk_phhd, discharge_record, identity)
-    cost_data = cost_task(pk_phhd, cost_list, identity)
+    for phrec in phrecs:
+        rec_type, ie_result = information_extraction(phrec, identity)
+        if rec_type == '基本医保结算单':
+            rec_result = settlement_result
+        elif rec_type == '出院记录':
+            rec_result = discharge_result
+        elif rec_type == '费用清单':
+            rec_result = cost_result
+        else:
+            rec_result = None
+        if rec_result:
+            for key, value in ie_result.items():
+                rec_result[key].append(value)
+
+    settlement_data = settlement_task(pk_phhd, settlement_result)
+    discharge_data = discharge_task(pk_phhd, discharge_result)
+    cost_data = cost_task(pk_phhd, cost_result)
 
     review_result = {
         'pk_phhd': pk_phhd,
@@ -573,6 +488,9 @@ def photo_review(pk_phhd, name):
 
 
 def main():
+    """
+    照片审核批量控制
+    """
     while 1:
         session = MysqlSession()
         phhds = (session.query(ZxPhhd.pk_phhd, ZxPhhd.cXm)