I'm using NiFi ExecuteScript to call a Groovy script that extracts text from PDFs. When it fails to extract, an Exception is supposed to be thrown and the flowfile is redirected to REL_FAILURE. Some PDFs go through fine, and some give the error:
ExecuteScript[id=9a39e0cb-ebcc-31e4-a169-575e367046e9] Failed to process session due to javax.script.ScriptException: javax.script.ScriptException: java.lang.IllegalStateException: StandardFlowFileRecord[uuid=2d6540f7-b7a2-48c7-8978-6b90bbfb0ff5,claim=StandardContentClaim [resourceClaim=StandardResourceClaim[id=1538596326047-12, container=default, section=12], offset=2134, length=930225],offset=0,name=1 i-9 INS rev 87 05-07-87.pdf,size=930225] already in use for an active callback or an OutputStream created by ProcessSession.write(FlowFile) has not been closed: org.apache.nifi.processor.exception.ProcessException: javax.script.ScriptException: javax.script.ScriptException: java.lang.IllegalStateException: StandardFlowFileRecord[uuid=2d6540f7-b7a2-48c7-8978-6b90bbfb0ff5,claim=StandardContentClaim [resourceClaim=StandardResourceClaim[id=1538596326047-12, container=default, section=12], offset=2134, length=930225],offset=0,name=1 i-9 INS rev 87 05-07-87.pdf,size=930225] already in use for an active callback or an OutputStream created by ProcessSession.write(FlowFile) has not been closed
My (simplified) code is below:
def flowFile = session.get()
if(!flowFile) return
flowFile = session.write(flowFile, { inputStream, outputStream ->
try {
// Load PDF from inputStream and parses text into a JSON string
// If nothing can be extracted, throw an exception so the flowfile
// can be routed to REL_FAILURE and processed further down the NiFi pipeline
if(outputLength < 15) {
throw new Exception('No output, send to REL_FAILURE')
}
// Write the string to the flowFile to be transferred
outputStream.write(json.getBytes(StandardCharsets.UTF_8))
} catch (Exception e){
System.out.println(e.getMessage())
session.transfer(flowFile, REL_FAILURE)
}
} as StreamCallback)
session.transfer(flowFile, REL_SUCCESS)
It pretty closely follows the cookbook posted in the Hortonworks community forums, and the author even mentions that closing is handled automatically.
I think that the error is caused when a PDF fails to process. This throws an exception, which should be caught in the try{}catch{} and then be transferred to REL_FAILURE. Instead it appears that the catch{} is never getting called, so the outputStream object is never closed. It works as expected and gets caught just fine when I run the same Groovy code outside of NiFi.
If you want to try running it on your own server