Commit dde575a1 authored by Birte Kristina Friesel's avatar Birte Kristina Friesel
Browse files

Use HTML::TreeBuilder::LibXML. Major performance enhancement.

parent 8413c8a4

bin/efa

+11 −8
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::XPath;

use HTML::TreeBuilder::LibXML;

use WWW::Mechanize;



my $firsturl = 'http://efa.vrr.de/vrr/XSLT_TRIP_REQUEST2?language=de&itdLPxx_transpCompany=vrr';
@@ -37,8 +37,9 @@ sub check_ambiguous { 
			"Ambiguous input for %s\n",

			$select->attr('name'),

		);

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

			say "\t$val";

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

			print "\t";

			say $val->as_trimmed_text();

		}

	}

	if ($ambiguous) {
@@ -209,7 +210,7 @@ sub parse_tree { 
		}



		if ($colspan == 8) {

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

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

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

				$con_part = 0;

				next;
@@ -225,8 +226,8 @@ sub parse_tree { 
			}

		}



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

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

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

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

		}

	}

	return $cons;
@@ -325,7 +326,9 @@ if ($test_dump) { 
	exit 0

}



my $tree = HTML::TreeBuilder::XPath->new_from_content($content);

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

$tree->parse($content);

$tree->eof();



check_ambiguous($tree);

test/parse_ambiguous

+7 −7

File changed.

Contains only whitespace changes.