def processRequest(request):
output = ""
print("request: {}".format(request))
bucketName = request["bucketName"]
objectName = request["objectName"]
documentsTable = request["documentsTable"]
outputTable = request["outputTable"]
print("Input Object: {}/{}".format(bucketName, objectName))
ext = FileHelper.getFileExtenstion(objectName.lower())
print("Extension: {}".format(ext))
if ext and ext in ["jpg", "jpeg", "png", "pdf"]:
documentId = str(uuid.uuid1())
ds = datastore.DocumentStore(documentsTable, outputTable)
ds.createDocument(documentId, bucketName, objectName)
output = "Saved document {} for {}/{}".format(documentId, bucketName,
objectName)
print(output)
return {"statusCode": 200, "body": json.dumps(output)}
def processRequest(request):
output = ""
print("request: {}".format(request))
itemId = request["itemId"]
bucketName = request["bucketName"]
objectName = request["objectName"]
print("Input Object: {}/{}".format(bucketName, objectName))
ext = FileHelper.getFileExtenstion(objectName.lower())
print("Extension: {}".format(ext))
if(ext and ext in ["jpg", "jpeg", "png"]):
qUrl = request['syncQueueUrl']
elif (ext in ["mov", "mp4"]):
qUrl = request['asyncQueueUrl']
if(qUrl):
jsonMessage = { 'itemId' : itemId,
'bucketName': bucketName,
'objectName' : objectName }
client = AwsHelper().getClient('sqs')
response = postMessage(client, qUrl, jsonMessage)
output = "Completed routing for itemId: {}, object: {}/{}".format(itemId, bucketName, objectName)
print(output)
return response
def processRequest(request):
output = ""
print("request: {}".format(request))
bucketName = request["bucketName"]
objectName = request["objectName"]
itemsTable = request["itemsTable"]
outputBucket = request["outputBucket"]
print("Input Object: {}/{}".format(bucketName, objectName))
ext = FileHelper.getFileExtenstion(objectName.lower())
print("Extension: {}".format(ext))
if (ext and ext in ["jpg", "jpeg", "png", "mov", "mp4"]):
itemId = str(uuid.uuid1())
ds = datastore.ItemStore(itemsTable)
ds.createItem(itemId, bucketName, objectName)
output = "Saved item {} for {}/{}".format(itemId, bucketName,
objectName)
print(output)
return {'statusCode': 200, 'body': json.dumps(output)}
def processRequest(request):
output = ""
print(f"request: {request}")
bucketName = request["bucketName"]
objectName = request["objectName"]
documentsTable = request["documentsTable"]
outputTable = request["outputTable"]
print(f"Input Object: {bucketName}/{objectName}")
ext = FileHelper.getFileExtenstion(objectName.lower())
print(f"Extension: {ext}")
if(ext and ext in ["jpg", "jpeg", "png", "pdf"]):
documentId = str(uuid.uuid1())
ds = datastore.DocumentStore(documentsTable, outputTable)
ds.createDocument(documentId, bucketName, objectName)
output = f"Saved document {documentId} for {bucketName}/{objectName}"
print(output)
return {
'statusCode': 200,
'body': json.dumps(output)
}
def processRequest(request):
output = ""
print("request: {}".format(request))
documentId = request["documentId"]
bucketName = request["bucketName"]
objectName = request["objectName"]
jobErrorHandlerQueueUrl = request['errorHandlerQueueUrl']
print("Input Object: {}/{}".format(bucketName, objectName))
ext = FileHelper.getFileExtenstion(objectName.lower())
print("Extension: {}".format(ext))
client = AwsHelper().getClient('sqs')
# If not expected extension, change status to FAILED and exit
if(ext and ext not in ["jpg", "jpeg", "png", "pdf"]):
jsonErrorHandlerMessage = {
'documentId': documentId
}
postMessage(client, jobErrorHandlerQueueUrl, jsonErrorHandlerMessage)
return
if(ext and ext in ["jpg", "jpeg", "png"]):
qUrl = request['syncQueueUrl']
errorHandlerTimeoutSeconds = SYNC_JOB_TIMEOUT_SECONDS
elif (ext in ["pdf"]):
qUrl = request['asyncQueueUrl']
errorHandlerTimeoutSeconds = ASYNC_JOB_TIMEOUT_SECONDS
if(qUrl):
features = ["Text", "Forms", "Tables"]
jsonMessage = {'documentId': documentId,
"features": features,
'bucketName': bucketName,
'objectName': objectName}
postMessage(client, qUrl, jsonMessage)
jsonErrorHandlerMessage = {
'documentId': documentId
}
postMessage(client, jobErrorHandlerQueueUrl,
jsonErrorHandlerMessage, errorHandlerTimeoutSeconds)
output = "Completed routing for documentId: {}, object: {}/{}".format(
documentId, bucketName, objectName)
print(output)
def processRequest(request):
output = ""
print("request: {}".format(request))
bucketName = request["bucketName"]
objectName = request["objectName"]
itemsTable = request["itemsTable"]
outputBucket = request["outputBucket"]
jobId = request["jobId"]
invocationId = request['invocationId']
invocationSchemaVersion = request['invocationSchemaVersion']
taskId = request['taskId']
print("Input Object: {}/{}".format(bucketName, objectName))
ext = FileHelper.getFileExtenstion(objectName.lower())
print("Extension: {}".format(ext))
if (ext and ext in ["jpg", "jpeg", "png", "mov", "mp4"]):
itemId = str(uuid.uuid1())
ds = datastore.ItemStore(itemsTable)
ds.createItem(itemId, bucketName, objectName)
output = "Saved item {} for {}/{}".format(itemId, bucketName,
objectName)
print(output)
results = [{
'taskId':
taskId,
'resultCode':
'Succeeded',
'resultString':
"Item submitted for processing with Id: {}".format(itemId)
}]
return {
'invocationSchemaVersion': invocationSchemaVersion,
'treatMissingKeysAs': 'PermanentFailure',
'invocationId': invocationId,
'results': results
}
def processRequest(request):
output = ""
print("request: {}".format(request))
bucketName = request["bucketName"]
objectName = request["objectName"]
documentsTable = request["documentsTable"]
outputTable = request["outputTable"]
jobId = request["jobId"]
invocationId = request["invocationId"]
invocationSchemaVersion = request["invocationSchemaVersion"]
taskId = request["taskId"]
print("Input Object: {}/{}".format(bucketName, objectName))
ext = FileHelper.getFileExtenstion(objectName.lower())
print("Extension: {}".format(ext))
if ext and ext in ["jpg", "jpeg", "png", "pdf"]:
documentId = str(uuid.uuid1())
ds = datastore.DocumentStore(documentsTable, outputTable)
ds.createDocument(documentId, bucketName, objectName)
output = "Saved document {} for {}/{}".format(documentId, bucketName,
objectName)
print(output)
results = [{
"taskId":
taskId,
"resultCode":
"Succeeded",
"resultString":
"Document submitted for processing with Id: {}".format(documentId),
}]
return {
"invocationSchemaVersion": invocationSchemaVersion,
"treatMissingKeysAs": "PermanentFailure",
"invocationId": invocationId,
"results": results,
}
def __init__(self, bucketName, documentPath, awsRegion, detectText,
detectForms, detectTables):
ip = Input()
if (bucketName):
ip.bucketName = bucketName
if (documentPath):
ip.documentPath = documentPath
if (awsRegion):
ip.awsRegion = awsRegion
if (detectText):
ip.detectText = detectText
if (detectForms):
ip.detectForms = detectForms
if (detectTables):
ip.detectTables = detectTables
if (not ip.bucketName and not ip.documentPath):
raise Exception("Document is required.")
if (ip.bucketName):
ip.isLocalDocument = False
else:
ip.isLocalDocument = True
ext = FileHelper.getFileExtenstion(ip.documentPath).lower()
print("Extension", ext)
if (ext == "pdf"):
ip.documentType = "PDF"
elif (ext == "jpg" or ext == "jpeg" or ext == "png"):
ip.documentType = "IMAGE"
else:
raise Exception("Document should be jpg/jpeg, png or pdf.")
if (ip.documentType == "PDF" and ip.isLocalDocument):
raise Exception("PDF must be in S3 bucket.")
if (ip.detectText == False and ip.detectForms == False
and ip.detectTables == False):
raise Exception(
"Select at least one option to extract text, form or table")
self.inputParameters = ip
def processRequest(request):
output = ""
print("request: {}".format(request))
documentId = request["documentId"]
bucketName = request["bucketName"]
objectName = request["objectName"]
print("Input Object: {}/{}".format(bucketName, objectName))
ext = FileHelper.getFileExtenstion(objectName.lower())
print("Extension: {}".format(ext))
if(ext and ext in ["jpg", "jpeg", "png"]):
qUrl = request['syncQueueUrl']
elif (ext in ["pdf"]):
qUrl = request['asyncQueueUrl']
elif (ext in ["htm", "html"]):
qUrl = os.environ['HTMLTOBOUNDINGBOX_QUEUE_URL']
if(qUrl):
features = ["Text"] # , "Forms", "Tables"]
jsonMessage = { 'documentId' : documentId,
"features" : features,
'bucketName': bucketName,
'objectName' : objectName }
client = AwsHelper().getClient('sqs')
postMessage(client, qUrl, jsonMessage)
if (ext in ["pdf"]):
postMessage(client, request["pdftoimgQueueUrl"], jsonMessage)
output = "Completed routing for documentId: {}, object: {}/{}".format(documentId, bucketName, objectName)
print(output)
def processRequest(request):
output = ""
print("request: {}".format(request))
documentId = request["documentId"]
bucketName = request["bucketName"]
objectName = request["objectName"]
print("Input Object: {}/{}".format(bucketName, objectName))
ext = FileHelper.getFileExtenstion(objectName.lower())
print("Extension: {}".format(ext))
if (ext and ext in ["jpg", "jpeg", "png"]):
qUrl = request['syncQueueUrl']
elif (ext in ["pdf"]):
qUrl = request['asyncQueueUrl']
if (qUrl):
# To avoid form and tables outputs, this will be added to a configuration file
#features = ["Text", "Forms", "Tables"]
features = ["Text"]
jsonMessage = {
'documentId': documentId,
"features": features,
'bucketName': bucketName,
'objectName': objectName
}
client = AwsHelper().getClient('sqs')
postMessage(client, qUrl, jsonMessage)
output = "Completed routing for documentId: {}, object: {}/{}".format(
documentId, bucketName, objectName)
print(output)