[Chicago-talk] folding multiple newlines at eof

Steven Lembark lembark at wrkhors.com
Tue Aug 16 10:19:37 PDT 2005



-- Jason Perkins <jperkins at sneer.org>

> I'm attempting to convert runs of multiple newlines at the end of  
> files into a single newline. The following one liner is converting  
> all runs of newlines throughout the file to a single newline:
> 
> find . -name '*.php' -exec perl -00 -pi -e 's/\n{2,}$/\n/s' {} \;
> 
> Can anyone suggest a means of correcting the behaviour? Explanations  
> with your fix even more greatly appreciated so that I can figure this  
> out on my own next time.

slurp the file and zap the newlines at EOL:

    #!/path/to/perl -p

    BEGIN { undef $/ };

    s/\n+\Z//;

or, if you prefer

    {
        local $/;

        my $string = <$ARGV>;

        $string =~ s{ \n+ \Z}{}x;

        print $string;
    }

(someone else'll have to remember what use English 
does to $/... :-).

Both ^ and $ are designed to apply mutlple times within
a string and locate themselves based on embedded newlines.
The \A and \Z meta char's are zero-width and only apply
at the start and end of the entire string. By anchoring
a search for newlines at the end-of-string (vs. end-of-
line) you'll strip only the trailing newlines.

Damian suggests in PBP using only \A and \Z, dropping
^ and $ entirely, becuase people make the mis-assumption
about EOL and EOS too frequently. 

-- 
Steven Lembark                                       85-09 90th Street
Workhorse Computing                                Woodhaven, NY 11421
lembark at wrkhors.com                                     1 888 359 3508


More information about the Chicago-talk mailing list