[Toolserver-l] Five questions

Thu Sep 21 13:10:41 UTC 2006

On Thu, 21 Sep 2006 09:54:31 +0200, <toolserver-l-request at Wikipedia.org>  
wrote:

> Message: 13
> Date: Thu, 21 Sep 2006 09:54:29 +0200
> From: Stefan K?hn <kuehn-s at gmx.net>
> Subject: [Toolserver-l] Five questions
> To: Mailingliste Toolserver <toolserver-l at Wikipedia.org>
> Message-ID: <45124535.8050706 at gmx.net>
> Content-Type: text/plain; charset=ISO-8859-15; format=flowed

> Can I use Perl to get some data from the database? (I work since three
> weeks with Perl and CGI, so I need a small example.)
Yes. I have no experience with it (I use python, which -imo- is easier  
than perl), but google showed
http://www.codeproject.com/perl/perldbi.asp
which could be useful. Perl/SQL is - for example- used at slashdot.org

> Question 2:
> Can I use a SQL-Question with perl? For example: I want the category's
> of page "xy" in DE.
See the previous link. Make a SQL query for it; something like
"select * from dewiki_p.templatelinks where tl_namespace=<template  
namespace> and tl_title=<template title>".

> Question 3.
> When I scan all articles from the Dump (XML-File) for coordinates I need
> with Perl in EN 45 minutes and in DE 15 minutes. If I use in the future
> hopefully the MySQL-database I think this process will be to long for
> the database. So that all other services have a problem. Is this right?
> Or is the power of the database strong enough for this full-text search?
There is no need for a full-text search. If you get all articles that have  
the template on it, you will only need the text from those pages. Getting  
the text then regexping it on hemlock (with a low priority) probably is  
the best way to do it. You will need the text tables for it though, and  
they are not yet available.

> Question 4.
> After the full-text search I will put the results in the database.
> Therefore I need also a Perl-example. Please help me.
See the same link.

> Question 5.
> If I can not do the full-text search at the MySQL-database I think it
> would be very helpful to make one directory for all users with all dumps
> (XML-files). So that every user can use this dumps. At the moment I have
> the current dump (DE+EN) in my home-directory, but I am sure that other
> users also have dumps. What did you think about this?
Wait until the text tables are there ;) - a FULLTEXT index of the text  
table would be useful, but probably very space-consuming. However, if  
zedler is overloaded by all the database requests, it could be an idea to  
do xml-fulltext-searches on hemlock (on low priority). Just my $0.02

-valhallasw