in code/textract_async/textract_starter.py [0:0]
def startJob(bucketName, objectName, documentId, snsTopic, snsRole):
print("Starting job with documentId: {}, bucketName: {}, objectName: {}".format(documentId, bucketName, objectName))
response = None
client = AwsHelper().getClient('textract')
response = client.start_document_analysis(
ClientRequestToken = documentId,
DocumentLocation={
'S3Object': {
'Bucket': bucketName,
'Name': objectName
}
},
FeatureTypes=["FORMS", "TABLES"],
NotificationChannel= {
"RoleArn": snsRole,
"SNSTopicArn": snsTopic
},
OutputConfig = {
"S3Bucket": targetBucketName,
"S3Prefix": objectName + "/textract-output"
},
JobTag = documentId
)
return response["JobId"]