In an arch-typical file-processing loop (such as the ones described in, ummm, the multitudinous perlfaqs...), the usual procedure is to use chomp to remove any newlines, then use a regular-expression to eliminate unwanted lines. Then output the ones that pass, remembering to include the record-separator.

For such a trivial task as this, the grep command-line tool might already be exactly what you need.

Otherwise, your regular expression is probably going to be something like: /^\s*$/ which breaks-down as follows:

• The "^" character marks the beginning of the line, and the "$" marks the end. These are used to bind the enclosing terms to “the entire record,” so that you are searching for records which consist exclusively of what's between.
• The \s is a character-class reference:   anything that qualifies as “white space.”
• Finally, the "*" means “zero-or-more of the foregoing.”
• So this regular-expression will match a record that consists entirely of zero-or-more whitespace.

As I previously mentioned, you can probably do the entire job with the grep command, using the "-v" modifier to invert the meaning of the pattern:   all lines which do not match the pattern will be output. This omits the blank lines (which are the ones that match) and outputs the non-blanks. Q.E.D.

(An equivalent Perl program could of-course be constructed, but since this is trivial ...)