t/get-bills.t
@@ -3,13 +3,17 @@ use FindBin;
use lib "$FindBin::Bin/../lib";
use Net::Parliament;
use Data::Dumper;
-use Net::Parliament::Member;
my $parl = Net::Parliament->new();
-my $members = $parl->Get_members();
-for my $member (@$members) {
- warn Dumper $member;
- last;
+if (0) {
+ my $members = $parl->Get_members();
+ for my $member (@$members) {
+ warn Dumper $member;
+ last;
+ }
}
+for my $bill (@{ $parl->Get_bills(parl => 40, session => 2) }) {
+ warn Dumper $bill;
+}
bin/scrape-parl
@@ -2,6 +2,7 @@ package Net::Parliament;
use Moose;
use Net::Parliament::UserAgent;
use HTML::TableExtract qw/tree/;
+use HTML::TreeBuilder;
=head1 NAME
@@ -30,8 +31,7 @@ and then parse it into hashrefs.
=cut
has '_members_base_url' => (
- is => 'ro',
- isa => 'Str',
+ is => 'ro', isa => 'Str',
default => 'http://webinfo.parl.gc.ca/MembersOfParliament/',
);
@@ -44,9 +44,19 @@ has 'members_html_url' => (
},
);
+has '_bills_base_domain' => (
+ is => 'ro', isa => 'Str',
+ default => 'http://www2.parl.gc.ca',
+);
+
+has '_bills_base_url' => (
+ is => 'ro', isa => 'Str',
+ default => 'http://www2.parl.gc.ca/HouseBills/billsgovernment.aspx?',
+);
+
has 'ua' => (
is => 'ro',
- isa => 'Net::Parliament::UserAgent',
+ isa => 'Object',
handles => ['get'],
default => sub { Net::Parliament::UserAgent->new },
);
@@ -105,6 +115,75 @@ sub Get_members {
return \@members;
}
+=head2 Get_bills()
+
+This method returns an arrayref containing a hashref for each
+Government Bill raised in parliament.
+
+=cut
+
+sub Get_bills {
+ my $self = shift;
+ my %opts = @_;
+
+ die "Must specify which Parliament" unless $opts{parl};
+ die "Must specify which Session" unless $opts{session};
+
+ my $url = $self->_bills_base_url . "Parl=$opts{parl}"
+ . "&Ses=$opts{session}";
+ my $html = $self->get($url);
+ my $block_oh_html = <<EOT;
+<div class="BillBlock BillBlockOdd" id="divBillBlockC2">
+ <span class="BillNumberCell">C-2</span>
+ <div class="BillSummary">
+ <span class="BillLongText">An Act to amend the Criminal Code and to make consequential amendments to other Acts</span>
+ <div class="BillSponsor"><a class="WebOption" onclick="GetWebOptions('PRISM','Affiliation',105824,'1');return false;" onmouseout="inDiv=0;setTimeout('TimeoutHide()',1000);return false;" href="/HousePublications/GetWebOptionsCallBack.aspx?SourceSystem=PRISM&ResourceType=Affiliation&ResourceID=105824&language=1&DisplayMode=2">The Minister of Justice</a></div>
+ <div>
+ <div><a class="BillVersionLink" href="/HouseBills/StaticLinkRedirector.aspx?Language=e&LinkTitle=%28C-2%29%20Legislative%20Summary&RedirectUrl=%2fSites%2fLOP%2fLEGISINFO%2findex.asp%3fList%3dls%26Language%3dE%26Query%3d5273%26Session%3d15&RefererUrl=X&StatsEnabled=true">Legislative Summary</a></div>
+ <div><a class="BillVersionLink" href="/HousePublications/Publication.aspx?DocId=3078412&Language=e&Mode=1">First Reading</a></div>
+ <div><a class="BillVersionLink" href="/HousePublications/Publication.aspx?DocId=3151626&Language=e&Mode=1">As passed by the House of Commons</a></div>
+ <div><a class="BillVersionLink" href="/HousePublications/Publication.aspx?DocId=3320180&Language=e&Mode=1">Royal Assent</a></div>
+ <div><a class="BillVersionLink" href="/housebills/BillVotes.aspx?Language=e&Mode=1&Parl=39&Ses=2&Bill=C2">Votes</a></div>
+ </div>
+ </div>
+</div>
+EOT
+
+ my $tree = HTML::TreeBuilder->new_from_content($html);
+ my @billblocks = $tree->look_down(class => qr/\bBillBlock\b/);
+ my @bills;
+ for my $b (@billblocks) {
+ my $bill = {};
+ $bill->{number} = $b->look_down(
+ class => 'BillNumberCell')->content->[0];
+ $bill->{summary} = $b->look_down(
+ class => 'BillLongText')->content->[0];
+ $bill->{sponsor} = $b->look_down(
+ class => 'BillSponsor')->content->[0];
+
+ if (ref($bill->{sponsor})) {
+ my $bs = $bill->{sponsor};
+ $bill->{sponsor} = $bs->content->[0];
+ my $url = $bs->look_down(
+ _tag => 'a')->attr('href');
+ if ($url =~ m/ResourceID=(\d+)/) {
+ $bill->{sponsor_id} = $1;
+ }
+ }
+
+ my @links = $b->look_down(class => 'BillVersionLink');
+ for my $link (@links) {
+ my $url = $self->_bills_base_domain . $link->attr('href');
+ $url =~ s/\s/%20/g;
+ push @{ $bill->{links} }, { $link->content->[0] => $url };
+ }
+
+ push @bills, $bill;
+ }
+ return \@bills;
+
+}
+
sub _load_member {
my $self = shift;
my $member = shift;
lib/Net/Parliament.pm
t/boilerplate.t
8c678be3525c03a3693fc81813afeca0e9b6c3b0
Luke Closs
gravatar@5thplane.com
http://github.com/lukec/cpan-net-parliament/commit/f03d2168a9d07dcfc40a401787b94eccc487dbbb
f03d2168a9d07dcfc40a401787b94eccc487dbbb
2009-05-20T23:07:31-07:00
2009-05-20T23:07:31-07:00
Add Get_bills() code and tests
cb23e41f9ce92a91c7abc103a5a38ce66a8c7c2a
Luke Closs
gravatar@5thplane.com