To me it looks like the original document is just doing something I've heard called font substitution.
Here is Serbia back in the 80s and 90s a quick way to get our unique characters was to just use a font which had our symbols for codes of Latin letters which were on keyboard position where our own letter would be. For example, let's take word
ужички. On such systems, you'd type it as
u\i;ki and use a font which displayed ж instead of \ and ч instead of ;. This is very easy to implement, because you don't actually use a different character set and when correctly displayed, here would be no difference. Unfortunately, this doesn't work well with internationalization and is a problem if you need several different character sets on a single computer.
As far as I know, the only solution for problems which came out of this is to use a program which will translate such text to proper Unicode text. The problem here is that you have a mix of Greek and Latin characters, so program can't just to dumb substitution. As far as I can see, there is no easy way to solve the problem.
You could try getting the document and running it through a proper OCR program which will detect letters and output in Unicode, but I haven't used OCR in quite some time, so I can't recommend one which will help. Also note that some OCR programs will only accept input from a scanner, so you need to be careful when picking OCR program, unless you want to print the whole document and them scan it.