Hello.
The way to write Perl scripts that handles unicode well with the do shell script command is to not use any utf8 features. 
This is for Perl scripts, that operates like a filter, i.e. they don’t open any files from disk, but then, with what I know now, then I would only encode the files I read into utf8, should I know they weren’t endoced as utf8 initially, and then, hope for the best! 
This can be handy to know, should you be inclined, and find yourself trying to write bom’s to the output.
I used the -CSA commandline option, and the use utf8 clause, thereby thinking it was bulletproof, by the strict parsing and interpretation, but that just don’t work, even worse, it works partially.
I have used the use 5.010 ; clause in Perl when getting those results. My version of Perl is 5.12 
Regards
McUsr