Support for inverted tables within a Textract PDF

Has anyone found a good way to parse tables that come back from Textract that are inverted? Specifically, getting the table name or column headers as they are not recognized via the API itself. Here are two examples of such a tables:

• https://www.aps.anl.gov/files/APS-Uploads/SECTOR1/1-ID/mts_assy_load_capacities.pdf
• https://wiki.wholelattelove.com/images/2/27/E7_AUTO_Parts_Diagram.PDF