Ok. The regex reflects the structure of you paragraphs - lines separated by empty lines (sed does some cleaning for it). -00 switch says to perl to read a file by paragraphs. The structure is:
some starting chars (not in the regex)
colon :
spaces \s*
shars (.) - remember them in $1
before newline \n
chars .
before colon :
spaces \s*
remember digits in $2 (\d+)
chars, including newline (for that is /s in the end of the regex) .*
before colon :
spaces \s*
digits to $3 (\d+)
some ending shars (not in the regex)
Though the perl solution works perfect for me I just found out that I don't
have perl installed on all my machines. Therefore, can somebody please
supply me with an AWK or sed to solution.