Textract: is it possible analyze document demo version and regular version are not aligned?

Question

Hello all,

cannot post a sample document in here, but lets say I'm working with invoices, pdf. All of a sudden, I found a couple odd balls today where my scripts that consume textract will consistently failt return 1 structured table (out of 3, the other 2 semi-strcutured) present in the document. I've tried everything, using boto3 analyze_document stripped of everything but 'TABLES' feature and I only get 2 tables for these invoices. If I put the same invoices through the Analyze Document demo (https://us-east-1.console.aws.amazon.com/textract/home?region=us-east-1#/demo) I'm always getting the 3 tables present in the invoices as expected. So I was thinking maybe demo is not actually consuming the same service that I am through scripting. Is that even a possibility? I can't think of any other explanation.

Thanks!

Answer

Seems like you are using a 1 pager PDF, and you are using sync api analyze_document with boto3? Console demo uses async api even for 1 pager pdf and sync vs async uses a different path for rendering PDF. That might explain the difference. I would recommend you to try out async [start_document_analysis](https://boto3.amazonaws.com/v1/documentation/api/latest/reference/services/textract/client/start_document_analysis.html)

Answer

Thank you! I actually found a related question right after this and the replies indeed point to the same explanation. I'll definitely try.

Best!

Textract: is it possible analyze document demo version and regular version are not aligned?

Contenido relevante