Commit 04f76c53 authored by Birte Kristina Friesel's avatar Birte Kristina Friesel
Browse files

Use XML::LibXML directly

parent dde575a1

bin/efa

+25 −16
Original line number Diff line number Diff line
@@ -7,7 +7,7 @@ use warnings; 
use 5.010;



use Getopt::Long qw/:config no_ignore_case/;

use HTML::TreeBuilder::LibXML;

use XML::LibXML;

use WWW::Mechanize;



my $firsturl = 'http://efa.vrr.de/vrr/XSLT_TRIP_REQUEST2?language=de&itdLPxx_transpCompany=vrr';
@@ -31,15 +31,18 @@ sub check_ambiguous { 
	my ($full_tree) = @_;

	my $ambiguous = 0;



	foreach my $select (@{$full_tree->findnodes('//select')}) {

	my $xp_select = XML::LibXML::XPathExpression->new('//select');

	my $xp_option = XML::LibXML::XPathExpression->new('./option');



	foreach my $select (@{$full_tree->findnodes($xp_select)}) {

		$ambiguous = 1;

		printf(

			"Ambiguous input for %s\n",

			$select->attr('name'),

			$select->getAttribute('name'),

		);

		foreach my $val ($select->findnodes('./option')) {

		foreach my $val ($select->findnodes($xp_option)) {

			print "\t";

			say $val->as_trimmed_text();

			say $val->textContent();

		}

	}

	if ($ambiguous) {
@@ -52,6 +55,13 @@ sub display_connection { 


	for my $con (@{$con_parts}) {



		# Note: Idets @{$con} elements

		foreach my $str (@{$con}) {

			$str =~ s/[\s\n\t]+/ /gs;

			$str =~ s/^ //;

			$str =~ s/ $//;

		}



		if (@{$con} < 5) {

			foreach my $str (@{$con}) {

				say "# $str";
@@ -198,19 +208,20 @@ sub parse_tree { 
	my $con_no = 0;

	my $cons;



	foreach my $td (@{$full_tree->findnodes('//table//table/tr/td')}) {

	my $xp_td = XML::LibXML::XPathExpression->new('//table//table/tr/td');

	my $xp_img = XML::LibXML::XPathExpression->new('./img');



	foreach my $td (@{$full_tree->findnodes($xp_td)}) {



		my $colspan = $td->attr('colspan') // 0;

		my $class   = $td->attr('class')   // q{};

		my $colspan = $td->getAttribute('colspan') // 0;

		my $class   = $td->getAttribute('class')   // q{};



		# Putting these into the XPath expression would lead to noticable (1

		# to 2 seconds) performance penalties

		if ( $colspan != 8 and $class !~ /^bgColor2?$/ ) {

			next;

		}



		if ($colspan == 8) {

			if ($td->as_trimmed_text() =~ / (?<no> \d+ ) \. .+ Fahrt /x) {

			if ($td->textContent() =~ / (?<no> \d+ ) \. .+ Fahrt /x) {

				$con_no = $+{'no'} - 1;

				$con_part = 0;

				next;
@@ -226,8 +237,8 @@ sub parse_tree { 
			}

		}



		if (not @{$td->findnodes('./img')} and $td->as_text() !~ /^\s*$/) {

			push(@{$cons->[$con_no]->[$con_part]}, $td->as_trimmed_text());

		if (not @{$td->findnodes($xp_img)} and $td->textContent() !~ /^\s*$/) {

			push(@{$cons->[$con_no]->[$con_part]}, $td->textContent());

		}

	}

	return $cons;
@@ -326,9 +337,7 @@ if ($test_dump) { 
	exit 0

}



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

$tree->parse($content);

$tree->eof();

my $tree = XML::LibXML->load_html(string => $content);



check_ambiguous($tree);

test/parse_ambiguous

+7 −7

File changed.

Contains only whitespace changes.