The PDF file was created by placing an existing PDF file into an InDesign document and re-exporting a PDF. In the process, the Unicode lookup table for the table legend (using the Calibri typeface) has been overwritten, so your PDF does *not* contain the text that you can see on screen. The apparently-random characters you get when copying the text are what's actually stored in the file, but the lookup table is matching them to the letter shapes you can see on screen, so it appears to be "correct".
Unfortunately there's no way back - Acrobat has no idea what the words look like, only what the characters are in the text stream. It's one of the perils of "refrying" a PDF by embedding it into another document. Some people actually use it as an intentional security measure.