Set difference with XPath 1.0 - how do I get .//table without .//table//table?

Question

I'm trying to find all the tables below my current node without also including the nested tables. In other words, if I have this, i want to find "yes" and not "no":

<table> <!-- outer table - no -->
  <tr><td>
    <div> <!-- *** context node *** -->
      <table> <!-- yes -->
        <tr><td>
          <table> ... </table> <!-- no -->
        </td></tr>
      </table>
      <table> <!-- yes -->
        <tr><td>
          <table> ... </table> <!-- no -->
        </td></tr>
      </table>
    </div>
  </td></tr>
</table>

Is there any easy way to do this in XPath 1.0? (In 2.0, it'd be .//table except .//table//table, but I don't have a 2.0 as an option.)

EDIT: please, the answers so far are not respecting the idea of current context node. I don't know how far down the first layer of table might be (and it might differ), and I also don't know if I might be inside another table (or two or three).

Literally, I want what .//table except .//table//table in XPath 2.0 would be, but I have only XPath 1.

Answer 1

After investigating it here and elsewhere, the answer seems to be "you can't, and that's why we have XPath 2.0". Oh well.

Answer 2

I think you want child::table aka table

#!/usr/bin/perl --
use strict;
use warnings;

use HTML::TreeBuilder;
{
  my $tree = HTML::TreeBuilder->new();

  $tree->parse(<<'__HTML__');
<table> <!-- outer table - no -->
  <tr><td>
    <div> <!-- *** context node *** -->
      <table> <!-- yes -->
        <tr><td>
          <table> ... </table> <!-- no -->
        </td></tr>
      </table>
      <table> <!-- yes -->
        <tr><td>
          <table> ... </table> <!-- no -->
        </td></tr>
      </table>
    </div>
  </td></tr>
</table>
__HTML__

  sub HTML::Element::addressx {
    return join(
      '/',
      '/', # // ROOT
      reverse(    # so it starts at the top
        map {
          my $n = $_->pindex() || '0';
          my $t = $_->tag;
          $t . '['. $n .']'
          }         # so that root's undef -> '0'
          $_[0],    # self and...
        $_[0]->lineage
      )
    );
  } ## end sub HTML::Element::addressx

  for my $td ( $tree->look_down( _tag => qr/div|table/i ) ) {
    print $td->addressx, "\n";
  }
  $tree->delete;
  undef $tree;
}
__END__
//html[0]/body[1]/table[0]
//html[0]/body[1]/table[0]/tr[0]/td[0]/div[0]
//html[0]/body[1]/table[0]/tr[0]/td[0]/div[0]/table[0]
//html[0]/body[1]/table[0]/tr[0]/td[0]/div[0]/table[0]/tr[0]/td[0]/table[0]
//html[0]/body[1]/table[0]/tr[0]/td[0]/div[0]/table[1]
//html[0]/body[1]/table[0]/tr[0]/td[0]/div[0]/table[1]/tr[0]/td[0]/table[0]

and second part

#!/usr/bin/perl --

use strict;
use warnings;

use HTML::TreeBuilder::XPath;

my $tree = HTML::TreeBuilder::XPath->new;
$tree->parse_content(<<'__HTML__');
<table> <!-- outer table - no -->
  <tr><td>
    <div> <!-- *** context node *** -->
      <table> <!-- yes -->
        <tr><td>
          <table> ... </table> <!-- no -->
        </td></tr>
      </table>
      <table> <!-- yes -->
        <tr><td>
          <table> ... </table> <!-- no -->
        </td></tr>
      </table>
    </div>
  </td></tr>
</table>
__HTML__



#~ for my $result ($tree->findnodes(q{//html[0]/body[1]/table[0]/tr[0]/td[0]/div[0]})) {
for my $result ($tree->findnodes(q{/html/body/table/tr/td/div})) {
    print $result->as_HTML,"\n\n";
    for my $table( $result->findnodes(q{table}) ){ ## child::table
        print "$table\n";
        print $table->as_HTML,"\n\n\n";
    }

}

__END__
<div><table><tr><td><table><tr><td> ... </td></tr></table></td></tr></table><table><tr><td><table><tr><td> ... </td></tr></table></td></tr></table></div>


HTML::Element=HASH(0xc6c964)
<table><tr><td><table><tr><td> ... </td></tr></table></td></tr></table>



HTML::Element=HASH(0xc6cbf4)
<table><tr><td><table><tr><td> ... </td></tr></table></td></tr></table>

Answer 3

Well, if I understand it, the content_list can solve:

my $table_one = $tree->findnodes('/html//table')->[1];

for ( $table_one->content_list ) {
    last if $_->exists('table');
    print $_->as_text;
}   

:)

Answer 4

What about .//table[not(.//table)]? Sorry for brevity, I'm on my phone.

Answer 5

I don't know how to get the context node to be evaluated in the nested predicates, but what you need is something like this:

descendant::table[not(ancestor::table[ancestor::div])]

only with the ability to reference the context node, instead of div

EDIT: If you set a variable for the context node,

<xsl:variable name="contextNode" select="." />

then you can reference it in the XPATH predicate:

descendant::table[not(ancestor::table[ancestor::*[generate-id(.)=generate-id($contextNode)]])]