http://qs321.pair.com?node_id=351548


in reply to Re: Need a better way to count input lines
in thread Need a better way to count input lines

Ahh! The lights come on! (Then they go dim again) I changed the
$fname =~ s/^\s+(.+?)\W*$/$1/;
to
$fname =~ s/^\s*(.+?)\W*$/$1/;
and the problem went away. (Of course it did. I should have seen that.) I can see why with the '+' it didn't work right, but why would the failure manifest as a CR at the end of $fname?
But I am perplexed about the split itself. I replaced the (",") with (",\s*") and got the error:
Unrecognized escape \s passed through at phonelist.pl line 28.
The paren/quote format worked fine with just a comma but broke with the addition of the space char class. Did I just get lucky the first way?

As to the LIMIT option, my Llama book doesn't explain it. Does the 2 relate the expectation of having two char within the match?

-Theo-
(so many nodes and so little time ... )

Replies are listed 'Best First'.
Re: Re: Re: Need a better way to count input lines
by hv (Parson) on May 07, 2004 at 20:09 UTC

    ... but why would the failure manifest as a CR at the end of $fname?

    I suspect the text, created under Windows, has CRLF as the line terminator; running the script under Solaris the chomp is removing the LF but not the CR, so the CR was getting removed instead by the $fname substitution - but only when the substitution actually happened.

    As for split, please take a look through perldoc -f split. First, note that the first parameter is shown as /PATTERN/; if you supply this parameter as a pattern (eg /,\s*/) it gets parsed as a pattern and does the right thing. If you supply a quoted string instead (eg ",\s*") perl will treat that first as a quoted string - in which case "\s" gives the warning, and gets converted to "s" - and the resulting string is then converted to a regexp further down the line.

    You'll also find there far more than you ever wanted to know about the LIMIT argument.

    Hugo