How do you empirically determine a file or stream's alphabet? Here's one (inelegant) way to do it:

perl -nle 'foreach $c (split //){$w{$c}++;} }{ print(sort keys %w);'

Here are some examples:

user@host$ seq 1 10 | perl -nle 'foreach $c (split //){$w{$c}++;} }{ print(sort keys %w);'

  • 0123456789

user@host$ echo 'abaaaaaac123;2ddddd,./' | perl -nle 'foreach $c (split //){$w{$c}++;} }{ print(sort keys %w);'

,./123;abcd

user@host$ perl -nle 'foreach $c (split //){$w{$c}++;} }{ print(sort keys %w);' ./test_input_file

+,-/0123456789:=ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz

That last example is an RSA private key, hence the almost-base64 set of characters it contains.