#!/usr/bin/perl
$/ = undef;
print smoosh( scalar <> );

use constant DEBUG => !!$ENV{DEBUG};
use constant {
    PRE  => ( DEBUG ? '<' : '' ),
    POST => ( DEBUG ? '>' : '' ),
};

my ( $UnicodeEscapeSequence, $IdentifierStart, $WhiteSpace, $LineTermi
+nator );

BEGIN {
    $UnicodeEscapeSequence = '\u[\da-fA-F]{4}';
    $IdentifierStart
        = '$_\p{Lu}\p{Ll}\p{Lt}\p{Lm}\p{Lo}\p{Nl}\p{Mn}\p{Mc}\p{Nd}\p{
+Pc}';
    $WhiteSpace     = '\t\x0b\f \xa0\p{Zs}';
    $LineTerminator = '\r\n\x{2028}\x{2029}';
}

sub smoosh {
    local $_ = shift @_;
    my $out = '';
    my %stat if DEBUG;

    while ( not m{\G\z}gc ) {
        if (0) { }

        # whitespace
        elsif (m{\G[$WhiteSpace]+}gco) {
            DEBUG and ++$stat{WhiteSpace};

            my $pos = pos;

            # Insert whitespace only if I'm joining things that could
            # be identifiers.
            if ((      $out =~ /[$IdentifierStart]\z/o
                    or $out =~ /$UnicodeEscapeSequence\z/o
                )
                and (  m{\G[$IdentifierStart]}gco
                    or m{\G$UnicodeEscapeSequence}gco )
                )
            {
                $out .= ' ';
            }

            pos() = $pos;
        }

        # everything else
        elsif (m{\G[^'"/$LineTerminator$WhiteSpace]+}gco) {
            DEBUG and ++$stat{'*'};
            $out .= PRE . substr( $_, $-[0], $+[0] - $-[0] ) . POST;
        }

        # line endings. Must be separated from other whitespace
        # because somtimes there are implicit semicolons here.
        elsif (m{\G[$LineTerminator]+}gco) {
            DEBUG and ++$stat{LineTerminator};
            if ( $out !~ /[$LineTerminator]\z/o ) {
                $out .= "\n";
            }
        }

        # String literals
        elsif (m{\G"(?:[^\\"$LineTerminator]*|\\.)*"}sgco) {
            DEBUG and ++$stat{'"StringLiteral"'};
            $out .= PRE . substr( $_, $-[0], $+[0] - $-[0] ) . POST;
        }
        elsif (m{\G'(?:[^\\'$LineTerminator]*|\\.)*'}sgco) {
           DEBUG and ++$stat{q['StringLiteral']};
            $out .= PRE . substr( $_, $-[0], $+[0] - $-[0] ) . POST;
        }

        # Comments
        elsif (m{\G//[^$LineTerminator]*[$LineTerminator]}gco) {
            DEBUG and ++$stat{SingleLineComment};
        }
        elsif (m{\G/\*(?:[^*\\]*|\\.)*\*/}sgc) {
            DEBUG and ++$stat{MultiLineComment};
        }

        # Regexps
        elsif (m{\G/(?:[^/\\$LineTerminator]*|\\.)+/}gco) {
            DEBUG and ++$stat{RegularExpressionLiteral};
            $out .= PRE . substr( $_, $-[0], $+[0] - $-[0] ) . POST;
        }

        # Division
        elsif (m{\G/(?![/*])}gc) {
            DEBUG and ++$stat{Division};
            $out .= PRE . '/' . POST;
        }

        else {
            my $pos  = pos;
            my $line = 1 + tr/\x0a\x0d\x2028\x2029//;
            die 'Error parsing JavaScript: "'
                . substr( $_, pos(), 20 )
                . q[" at line $line.\n];
        }
    }

    if (DEBUG) {
        print "STATS:\n";
        for ( sort { $stat{$b} <=> $stat{$a} } keys %stat ) {
            print "    $_: $stat{$_}\n";
        }
    }

    return $out;
}
[download]