From: Phil Rhoades <phil@...>
Date: 2008-04-27T05:26:38+09:00
Subject: Re: Howto get array.agrep (NOT array.grep)

jens,


On Sun, 2008-04-27 at 02:50 +0900, Jens Wille wrote:
> Phil Rhoades [2008-04-26 19:13]:
> > This might work but it would be more difficult without regexs -
> > the current application does a system call to agrep but of course
> > it is very slow for large numbers of calls.  A typical call is
> > something like:
> > 
> >   agrep -2 "Smith\|J.*12345" list1.txt list2.txt list3.txt
> > 
> > This allows two differences on a minimum amount of information 
> > consisting of last name, first initial and zip code.  If I use
> > the Enumerable version, I would have to use the whole, delimited,
> > name & address string and increase the differences/distance
> > number?
>
> i think something like that could work in your case (requires the
> Text gem):
> 
>   File.open('list1.txt').select { |line|
>     # extract name and zip code from line
>     line =~ /\A(.*?\|.).*\b(\d{5})\b/  # adjust appropriately!
> 
>     # name may have two errors, zip only one -- or whatever...
>     Text::Levenshtein.distance($1, 'Smith|J') <= 2 &&
> Text::Levenshtein.distance($2, '12345') <= 1
>   }


I see what you are doing but this would have to be repeated for the
three different lists (list1.txt, list2.txt, list3.txt) - I guess that
should still be faster than a single system call . .


> > Did you just do that hack now?
> that's right. but i just read a bit on agrep's algorithm and it
> might be fun to implement it in ruby (though a bit slow, probably).


I don't know if it helps but there is this:

http://www.koders.com/ruby/fidCEAEDCAA28D4A59A76ADF20A0DA2A3858438834D.aspx


> as an alternative, it might be even worth writing ruby bindings to
> agrep. who knows, if time permits... ;-)


I was wondering about something like that but I have never created a
Ruby binding before . .


> > - how do I get/install it? (Fedora 8).
> well, i don't think that particular implementation suits your needs
> and is obviously easily adapted (after all, it's just a select with
> an appropriate block utilizing Text::Levenshtein.distance). but you
> can get ruby-nuggets from rubyforge (gem install ruby-nuggets), or,
> if the new version hasn't found its way onto the mirrors yet, from
> our own gem server at http://prometheus.khi.uni-koeln.de/rubygems/.


Thanks!

Phil.
-- 
Philip Rhoades

Pricom Pty Limited  (ACN 003 252 275  ABN 91 003 252 275)
GPO Box 3411
Sydney NSW      2001
Australia
Fax:     +61:(0)2-8221-9599
E-mail:  phil@pricom.com.au