web::scraper using an xpath

[ag4ve]

i'm trying to get data out of a web page. i figured i'd use web::scraper (it looked easy enough) so, i wrote this:

my $pagedata = scraper {
  process '//*/table[@class="someclass"]', 'table[]' => scraper {
     process '//tr/td[1]', 'name' => 'TEXT';
     process '//tr/td[2]', 'attr' => 'TEXT';
  };
};
.......
  my $res = $pagedata->scrape( $content )
     or die "Can't define content to parser $!";

print Dumper( $res );

and, data dumper says i'm only getting the first line of the table. where the table would look something like this and i want all of it:

<table class="someclass" style="width:508px;" id="Any_20">
 <tbody>
 <tr>
  <td>name</td>
  <td>attribute</td>

  <td>name2</td>
  <td>attribute2</td>

  <td>possible name3</td>
  <td>possible attribute3</td>

  <td>
....
  </tr><tr>
   <td>... etc

so, i'm guessing that the question is how to set the //td/tr xpath strings up so that i get multiple rows? or if not, how do i setup the scraper function so that it loops to the next row? or what other module might i try to accomplish this?

UPDATE

ok, so thanks to the anonymous poster, here's what i've come up with that works damn nice:

#!/usr/bin/perl

use strict;
use warnings;


use LWP::UserAgent;
use LWP::Simple;
use Web::Scraper;
use Data::Dumper::Simple;

my( $infile ) = $ARGV[ 0 ] =~ m/^([\ A-Z0-9_.-]+)$/ig;

my $pagedata = scraper {
  process '//*/table[@class="someclass"]//tr', 'table[]' => scraper {
      my $count = 1;
      process '//tr/td[' . $count++ . ']', 'name' => 'TEXT';
      process '//tr/td[' . $count++ . ']', 'attr' => 'TEXT';
  };
};


open( FILE, "< $infile" );

my $content = do { local $/; <FILE> };

  my $res = $pagedata->scrape( $content )
     or die "Can't define content to parser $!";

print Dumper( $res );

thanks all, and sorry about not posting full code and just the snip - tried not to make the post too long.

[jethro]

I know nothing about XPATH or Web::Scraper, but you could try to add tbody to your first process call, i.e. process '//*/table[@class="someclass"]/tbody' !??

[Anonymous Monk]

Changing it to '//*/table[@class="someclass"]//tr' should work as intended

[jethro]

Are you sure? In that case I would guess that he must also remove the 'tr' from his further process statements.

[Corion]

You can't go "pairwise" over a list with XPath.

You will need to capture all TD contents and then do the reconstruction of name+attribute in Perl if the table is really structured (and classed) the way you show.

[Anonymous Monk]

I'd use a search like site:search.cpan.org xpath html which turns up HTML::TreeBuilder::XPath. Then you'd need to fetch the HTML so probably use LWP::Simple.

But then, that search didn't turn up Web::Scraper (that I noticed), so you might want to try variations on it.

[Anonymous Monk]

Um, Web::Scraper already does all that ...