--- /usr/lib/perl5/site_perl/5.8.0/WWW/Search/Pagesjaunes.pm 2003-12-05 20:00:45.000000000 +010~0
+++ lib/WWW/Search/Pagesjaunes.pm 2004-01-05 16:30:31.000000000 +0100
@@ -1,25 +1,26 @@
package WWW::Search::Pagesjaunes;
use strict;
-#use Carp qw(carp);
-#use locale;
+use Carp qw(carp croak);
use HTML::Form;
use HTML::TokeParser;
use LWP::UserAgent;
-$WWW::Search::Pagesjaunes::VERSION = '0.07';
+$WWW::Search::Pagesjaunes::VERSION = '0.08_01';
sub ROOT_URL() { '
http://www.pagesjaunes.fr' }
sub new {
my $class = shift;
my $self = {};
- my $ua = shift () || LWP::UserAgent->new(
+ my $ua = shift() || LWP::UserAgent->new(
env_proxy => 1,
keep_alive => 1,
timeout => 30,
);
- $ua->agent( "WWW::Search::Pagesjaunes/$WWW::Search::Pagesjaunes::VERSION "
+
+ $ua->agent( "WXWW::Search::Pagesjaunes/$WWW::Search::Pagesjaunes::VERSION "
. $ua->agent );
+
$self->{ua} = $ua;
$self->{limit} = 50;
@@ -36,15 +37,20 @@
# Make the first request to pagesjaunes.fr
$self->{URL} = ROOT_URL . ( $opt{activite} ? '/pj.cgi' : '/pb.cgi' );
- my @forms = HTML::Form->parse(
- $self->{ua}->request( HTTP::Request->new( 'GET', $self->{URL} ) )
- ->content,
- $self->{URL}
- );
+ my $req = $self->{ua}->request( HTTP::Request->new( 'GET', $self->{URL} ) );
+
+ $DB::single = 1;
+ if ( !$req->content || !$req->is_success ) {
+ croak('Error while retrieving the HTML page');
+ }
- my $form = $opt{activite} ? $forms[1] : $forms[0];
+ my @forms = HTML::Form->parse( $req->content, $self->{URL} );
+
+ #my $form = $opt{activite} ? $forms[1] : $forms[0];
+ my $form = $forms[0];
{
+
# HTML::Form complains when you change hidden fields values.
local $^W;
$form->value( 'lang', $self->{lang} );
@@ -71,7 +77,9 @@
my $parser = HTML::TokeParser->new( \$result_page );
# All the <br> tags are transformed to whitespace
- $parser->{textify} = { 'br' => sub() { " " } };
+ $parser->{textify} = {
+ 'br' => sub() { " " }
+ };
my @results;
@@ -116,7 +124,7 @@
my $phone = _trim( $parser->get_trimmed_text('/td') );
$phone =~ s/^\W(.*)$/$1/g;
- push (
+ push(
@results,
WWW::Search::Pagesjaunes::Entry->new(
$name, $address, $phone, 0
@@ -137,14 +145,17 @@
}
# If there was no result, we look for an error message in the HTML page
- if ( !@results && $self->{error} ){
+ if ( !@results && $self->{error} ) {
$parser = HTML::TokeParser->new( \$result_page );
- while ( my $token = $parser->get_tag("font") ){
- next unless $token->[1]
+ while ( my $token = $parser->get_tag("font") ) {
+ next
+ unless $token->[1]
&& $token->[1]{color}
&& $token->[1]{color} eq '#ff0000';
- $parser->{textify} = { 'br' => sub() { " " } };
- print STDERR _trim($parser->get_trimmed_text('/font')) . "\n";
+ $parser->{textify} = {
+ 'br' => sub() { " " }
+ };
+ print STDERR _trim( $parser->get_trimmed_text('/font') ) . "\n";
}
}
