Save Aperio annotations to XML files

.env

USERNAME="user"
PASSWORD="pass"
IMAGE_ENDPOINT="https://images.pathology.umn.edu/dataserver/Aperio.Images/Image"
SECURITY_ENDPOINT="https://images.pathology.umn.edu/dataserver/Aperio.Security/Security2"
XML_PATH='G:\My Drive\Projects\Metzger\AnnotationBackup'
PROJECT_ID=288

get-aperio-annotations.py

import requests
import time
import os
import xml.etree.ElementTree as ET
from decouple import config

IMAGE_ENDPOINT = config(
    'IMAGE_ENDPOINT', default='https://images.pathology.umn.edu/dataserver/Aperio.Images/Image')
SECURITY_ENDPOINT = config(
    'SECURITY_ENDPOINT', default='https://images.pathology.umn.edu/dataserver/Aperio.Security/Security2')
USERNAME = config('USERNAME')
PASSWORD = config('PASSWORD')
XML_PATH = config('XML_PATH')
projectId = config('PROJECT_ID')
ns = {'aperio': 'http://www.aperio.com/webservices/'}


def getAnnotations(imageId, accessionNumber):
    headers['SOAPAction'] = 'GetAnnotations'
    data = '<soapenv:Envelope xmlns:soapenv="http://schemas.xmlsoap.org/soap/envelope/" xmlns:ns0="http://www.aperio.com/webservices/">\
    <soapenv:Body>\
    <ns0:GetAnnotations>\
    <ns0:Token>%s</ns0:Token>\
    <ns0:ImageId>%s</ns0:ImageId>\
    </ns0:GetAnnotations>\
    </soapenv:Body>\
    </soapenv:Envelope>' % (token, imageId)
    # print(data)
    r = requests.post(IMAGE_ENDPOINT, data=data, headers=headers)
    # print(r.text)
    annotations_xml = r.text.split('<AnnotationsNode>')[
        1].split('</AnnotationsNode>')[0]
    # print(annotations_xml)
    t = time.strftime("%Y%m%d-%H%M", time.localtime())
    # print(t)
    filename = '%s-annotations-%s.xml' % (imageId, t)
    # print(filename)
    filepath = os.path.join(XML_PATH, filename)
    print(filepath)
    with open(filepath, 'w') as f:
        f.write(annotations_xml)


headers = {'Content-Type': 'text/xml',
           'SOAPAction': 'Logon'}
data = '<soapenv:Envelope xmlns:soapenv="http://schemas.xmlsoap.org/soap/envelope/" xmlns:ns0="http://www.aperio.com/webservices/">\
<soapenv:Body>\
<ns0:Logon>\
<ns0:UserName>%s</ns0:UserName>\
<ns0:PassWord>%s</ns0:PassWord>\
</ns0:Logon>\
</soapenv:Body>\
</soapenv:Envelope>' % (USERNAME, PASSWORD)
r = requests.post(SECURITY_ENDPOINT, data=data, headers=headers)
resp_xml = ET.fromstring(r.text)
token = resp_xml.find('.//aperio:LogonResponse/aperio:Token', ns).text
if token is not None:
    print(token)
else:
    print(r.text)


headers['SOAPAction'] = 'GetChildList'
data = '<soapenv:Envelope xmlns:soapenv="http://schemas.xmlsoap.org/soap/envelope/" xmlns:ns0="http://www.aperio.com/webservices/">\
<soapenv:Body>\
<ns0:GetChildList>\
<ns0:Token>%s</ns0:Token>\
<ns0:ParentTableName>Project</ns0:ParentTableName>\
<ns0:ParentId>%i</ns0:ParentId>\
<ns0:ChildTableName>Specimen</ns0:ChildTableName>\
</ns0:GetChildList>\
</soapenv:Body>\
</soapenv:Envelope>' % (token, projectId)
print(data)
r = requests.post(IMAGE_ENDPOINT, data=data, headers=headers)
# print(r.text)
specimens_xml = ET.fromstring(r.text)
specimens = specimens_xml.findall(
    './/aperio:GetChildListResponse/aperio:GenericDataSet/aperio:DataRow', ns)
for specimen in specimens:
    # print(s)
    # print(dir(s))
    accessionNumber = specimen.find('./aperio:AccessionNumber', ns).text
    print(accessionNumber)
    specimenId = int(specimen.find('./aperio:Id', ns).text)
    headers['SOAPAction'] = 'GetChildList'
    data = '<soapenv:Envelope xmlns:soapenv="http://schemas.xmlsoap.org/soap/envelope/" xmlns:ns0="http://www.aperio.com/webservices/">\
    <soapenv:Body>\
    <ns0:GetChildList>\
    <ns0:Token>%s</ns0:Token>\
    <ns0:ParentTableName>Specimen</ns0:ParentTableName>\
    <ns0:ParentId>%i</ns0:ParentId>\
    <ns0:ChildTableName>Slide</ns0:ChildTableName>\
    </ns0:GetChildList>\
    </soapenv:Body>\
    </soapenv:Envelope>' % (token, specimenId)
    print(data)
    r = requests.post(IMAGE_ENDPOINT, data=data, headers=headers)
    print(r.text)
    slides_xml = ET.fromstring(r.text)
    slides = slides_xml.findall(
        './/aperio:GetChildListResponse/aperio:GenericDataSet/aperio:DataRow', ns)
    print(len(slides))
    for slide in slides:
        imageId = int(slide.find('./aperio:ImageId', ns).text)
        print(imageId)
        getAnnotations(imageId, accessionNumber)