Hi Jack, the thought is within workflow, to use Pattern Matching to Identify and Break out the text for Line # Sections on the pages into a Multi-value Token. Then using that Token, as you loop through each sections Text, use additional Pattern Matching to pull the specific information from that's tokens text to create the individuals item in a table or metadata fields. There are lots of caveats here that have to be considered such as if a Line # section cross pages, differing length of sections, etc. It would require a lot of testing and possible exception handling to ensure reliability.

Yes - OCR the document. Make sure the text displays accurately and identify the pattern. Create a workflow.

***Here's the trick - you need to know regex. 

In the workflow you can use the "Retrieve Document Text" activity to get the text. Then use the "Pattern Matching" activity to parse the text.

This workflow:

• Gets the document
• Create Tokens (for pattern matching)
• Gets the Document Text
• Regex the individual pages (in this case the pages had headers and it was easier to work with the text in-between the header and footer)
• Then for each page:: run different pattern matches to get the data, drop them in to a token, assign field values