view release on metacpan or  search on metacpan

lib/NLP/GATE/Document.pm  view on Meta::CPAN

  my $offset = 0;
  my $text = "";
  for my $el ($doc->findnodes("/GateDocument/TextWithNodes")) {
    foreach my $c ($el->childNodes()) {
      if($c->nodeType() == 1) {  # element node
        ## get the attribute id
        my $nodeid = _getAttr($c,"id");
        $nodemap{$nodeid} = $offset;
      } elsif($c->nodeType() == 3 || $c->nodeType() == 4) {
        ## 3: text
        ## 4: cdata
        my $t = $c->textContent();
        $offset += length($t);
        $text .= $t;
      } else {
        croak "Invalid node type encountered: ",$c->nodeType(),"\n";
      }
    }
  }
  $self->{text} = $text;
  ## process the annotation features, replacing node ids with offset information