This is a fundamental problem every integration engineer needs to ask. There are many answers to this question:
If the system has /wiki/spaces/IX/pages/2004156468 then this might be a convenient way to extract the data you need.
Does one needs to build an adapter?
If the system uses a database can you query data from this and export it?
Does the system have some type of automated capability to generate output files - one can read these files and extract the data from here.
Do we have to resort to screen scraping? This is where we simulate a human user programmatically sending keyboard actions to the user interface of the program and extract the data in this way.