[Chicago-talk] Performance issue

[Steven Lembark]

On Wed, 20 Apr 2011 22:42:56 -0500
Jay Strauss <me at heyjay.com> wrote:

> I was under the impression that regexs were slow.

In general replacing the contents of a string in-place
will take much less time than using a regex to break 
up the string into an array and then paste the array
back together. 

For example, stripping the closing quote with a regex
should be quite fast:

    while( <> )
    {
        s{^ \s*? " }{}x;    # shortest match to the first quote
        s{ " \s*  $}{}x;    # shortest match to the end of line

        s{ [|] }{-}gx;      # pipes to dashes
        s{ ,   }{|}gx;      # commas to pipes

        # use the line here
    }

should be about the fastest way you can perform the
replacement. 

I'd suggest avoiding a "substr $x, -2" to strip the 
trailing quote since any minor change in the file
format (e.g. the source providing ^M^J instead of just
^J for the line endings) will completely break the 
process. With the regex you always get everything after
the final quote stripped.


-- 
Steven Lembark                                             3646 Flora Pl
Workhorse Computing                                   St Louis, MO 63110
lembark at wrkhors.com                                      +1 888 359 3508