From 54a24331cc6f6d7b77fe39ca04a35a26f47bd8e9 Mon Sep 17 00:00:00 2001 From: memespring Date: Sun, 29 Apr 2007 12:09:12 +0000 Subject: [PATCH] more of toms scrapers --- trunk/cgi-bin/Dacorum.cgi | 108 +++++++++++++++++++++++++++++++ trunk/cgi-bin/EastHerts.cgi | 122 ++++++++++++++++++++++++++++++++++++ trunk/cgi-bin/Enfield.cgi | 120 +++++++++++++++++++++++++++++++++++ 3 files changed, 350 insertions(+) create mode 100644 trunk/cgi-bin/Dacorum.cgi create mode 100644 trunk/cgi-bin/EastHerts.cgi create mode 100644 trunk/cgi-bin/Enfield.cgi diff --git a/trunk/cgi-bin/Dacorum.cgi b/trunk/cgi-bin/Dacorum.cgi new file mode 100644 index 0000000..1a27715 --- /dev/null +++ b/trunk/cgi-bin/Dacorum.cgi @@ -0,0 +1,108 @@ +#!/usr/bin/perl + +use strict; +use warnings; + +use CGI qw(:cgi); +use HTML::TreeBuilder; +use LWP::UserAgent; +use XML::Writer; + +# The master URLs for the Dacorum planning search +our $SearchURL = "http://www.dacorum.gov.uk/default.aspx?page=1495"; +our $InfoURL = "http://www.dacorum.gov.uk/Default.aspx?page=1497&ID="; +our $CommentURL = "http://www.dacorum.gov.uk/Default.aspx?page=2847&ID="; + +# We're a CGI script... +my $query = CGI->new(); + +# Construct an LWP user agent +our $UA = LWP::UserAgent->new(env_proxy => 1, + cookie_jar => {}, + requests_redirectable => [ 'GET', 'HEAD', 'POST' ]); + +# Post the URL to get an initial blank form +my $state = get_state(do_post()); + +# Do the search +my $page = do_post({"__VIEWSTATE" => $state, + "Template:_ctl10:_ctl0:btnSearch" => "Search", + "Template:_ctl10:_ctl0:tbRegistrationFromDay" => $query->param("day"), + "Template:_ctl10:_ctl0:tbRegistrationFromMon" => $query->param("month"), + "Template:_ctl10:_ctl0:tbRegistrationFromYear" => $query->param("year"), + "Template:_ctl10:_ctl0:tbRegistrationToDay" => $query->param("day"), + "Template:_ctl10:_ctl0:tbRegistrationToMon" => $query->param("month"), + "Template:_ctl10:_ctl0:tbRegistrationToYear" => $query->param("year")}); + +# Output an HTTP response header +print $query->header(-type => "text/xml"); + +# Create an XML output stream +my $Writer = XML::Writer->new(DATA_MODE => 1); + +# Output the XML header data +$Writer->xmlDecl("UTF-8"); +$Writer->startTag("planning"); +$Writer->dataElement("authority_name", "Dacorum Borough Council"); +$Writer->dataElement("authority_short_name", "Dacorum"); +$Writer->startTag("applications"); + +# Find the result table +my $table = $page->look_down("_tag" => "table", "class" => "FormDataGrid"); + +# Process each row of the results +foreach my $row ($table->look_down("_tag" => "tr")) +{ + my @cells = $row->look_down("_tag" => "td"); + + if ($cells[0]->attr("class") eq "FormGridDataItem" || + $cells[0]->attr("class") eq "FormGridAlternatingDataItem") + { + my $reference = $cells[0]->as_trimmed_text; + my $address = $cells[1]->as_trimmed_text; + my $description = $cells[2]->as_trimmed_text; + my $date = $cells[3]->as_trimmed_text; + my $postcode; + + if ($address =~ /\s+([A-Z]+\d+\s+\d+[A-Z]+)$/) + { + $postcode = $1; + } + + $Writer->startTag("application"); + $Writer->dataElement("council_reference", $reference); + $Writer->dataElement("address", $address); + $Writer->dataElement("postcode", $postcode); + $Writer->dataElement("description", $description); + $Writer->dataElement("info_url", $InfoURL . $reference); + $Writer->dataElement("comment_url", $CommentURL . $reference); + $Writer->dataElement("date_received", $date); + $Writer->endTag("application"); + } +} + +# Finish off XML output +$Writer->endTag("applications"); +$Writer->endTag("planning"); +$Writer->end(); + +exit 0; + +# Extract the state from a page so we can repost it +sub get_state +{ + my $page = shift; + my $viewstate = $page->look_down("_tag" => "input", "name" => "__VIEWSTATE"); + + return $viewstate->attr("value"); +} + +# Post to the planning search page +sub do_post +{ + my $response = $UA->post($SearchURL, @_); + + die $response->status_line unless $response->is_success; + + return HTML::TreeBuilder->new_from_content($response->content); +} diff --git a/trunk/cgi-bin/EastHerts.cgi b/trunk/cgi-bin/EastHerts.cgi new file mode 100644 index 0000000..34dec27 --- /dev/null +++ b/trunk/cgi-bin/EastHerts.cgi @@ -0,0 +1,122 @@ +#!/usr/bin/perl + +use strict; +use warnings; + +use CGI qw(:cgi); +use HTML::TreeBuilder; +use LWP::UserAgent; +use XML::Writer; + +# The master URLs for the East Herts planning search +our $SearchURL = "http://e-services.eastherts.gov.uk/swiftlg/apas/run/WPHAPPCRITERIA"; +our $InfoURL = "http://e-services.eastherts.gov.uk/swiftlg/apas/run/WPHAPPDETAIL.DisplayUrl?theApnID="; +our $CommentURL = "http://e-services.eastherts.gov.uk/swiftlg/apas/run/wphmakerep.displayURL?ApnID="; + +# We're a CGI script... +my $query = CGI->new(); + +# Get the date to fetch +my $date = $query->param("day") . "/" . $query->param("month") . "/" . $query->param("year"); + +# Construct an LWP user agent +our $UA = LWP::UserAgent->new(env_proxy => 1); + +# Do the search +my $page = do_post($SearchURL, + {"REGFROMDATE.MAINBODY.WPACIS.1." => $date, + "REGTODATE.MAINBODY.WPACIS.1." => $date, + "SEARCHBUTTON.MAINBODY.WPACIS.1." => "Search"}); + +# Output an HTTP response header +print $query->header(-type => "text/xml"); + +# Create an XML output stream +my $Writer = XML::Writer->new(DATA_MODE => 1); + +# Output the XML header data +$Writer->xmlDecl("UTF-8"); +$Writer->startTag("planning"); +$Writer->dataElement("authority_name", "East Herts Council"); +$Writer->dataElement("authority_short_name", "East Herts"); +$Writer->startTag("applications"); + +# Output any applications on the first page +output_applications($page); + +# Loop over any additional results pages +foreach my $link ($page->look_down("_tag" => "a", "href" => qr/^WPHAPPSEARCHRES\.displayResultsURL/)) +{ + # Fetch this page... + $page = do_get(URI->new_abs($link->attr("href"), $SearchURL)); + + # ...and output the applications from it + output_applications($page); +} + +# Finish off XML output +$Writer->endTag("applications"); +$Writer->endTag("planning"); +$Writer->end(); + +exit 0; + +# Make a GET request +sub do_get +{ + my $response = $UA->get(@_); + + die $response->status_line unless $response->is_success; + + return HTML::TreeBuilder->new_from_content($response->content); +} + +# Make a POST request +sub do_post +{ + my $response = $UA->post(@_); + + die $response->status_line unless $response->is_success; + + return HTML::TreeBuilder->new_from_content($response->content); +} + +# Output applications from a results page +sub output_applications +{ + my $page = shift; + + # Find the result table + my $table = $page->look_down("_tag" => "table", "cellspacing" => "2", "cellpadding" => "2"); + + # Process each row of the results + foreach my $row ($table->look_down("_tag" => "tr")) + { + my @cells = $row->look_down("_tag" => "td"); + + if (@cells >= 3) + { + my $reference = $cells[0]->as_trimmed_text; + my $description = $cells[1]->as_trimmed_text; + my $address = $cells[2]->as_trimmed_text; + my $postcode; + + if ($address =~ /\s+([A-Z]+\d+\s+\d+[A-Z]+)$/) + { + $postcode = $1; + } + + $Writer->startTag("application"); + $Writer->dataElement("council_reference", $reference); + $Writer->dataElement("address", $address); + $Writer->dataElement("postcode", $postcode); + $Writer->dataElement("description", $description); + $Writer->dataElement("info_url", $InfoURL . $reference); + $Writer->dataElement("comment_url", $CommentURL . $reference); + $Writer->dataElement("date_received", $date); + $Writer->endTag("application"); + } + } + + return; +} diff --git a/trunk/cgi-bin/Enfield.cgi b/trunk/cgi-bin/Enfield.cgi new file mode 100644 index 0000000..bfb0df7 --- /dev/null +++ b/trunk/cgi-bin/Enfield.cgi @@ -0,0 +1,120 @@ +#!/usr/bin/perl + +use strict; +use warnings; + +use CGI qw(:cgi); +use HTML::TreeBuilder; +use LWP::UserAgent; +use XML::Writer; + +# The master URLs for the Enfield planning search +our $SearchURL = "http://forms.enfield.gov.uk/swiftlg/apas/run/WPHAPPCRITERIA"; +our $InfoURL = "http://forms.enfield.gov.uk/swiftlg/apas/run/WPHAPPDETAIL.DisplayUrl?theApnID="; + +# We're a CGI script... +my $query = CGI->new(); + +# Get the date to fetch +my $date = $query->param("day") . "/" . $query->param("month") . "/" . $query->param("year"); + +# Construct an LWP user agent +our $UA = LWP::UserAgent->new(env_proxy => 1); + +# Do the search +my $page = do_post($SearchURL, + {"REGFROMDATE.MAINBODY.WPACIS.1." => $date, + "REGTODATE.MAINBODY.WPACIS.1." => $date, + "SEARCHBUTTON.MAINBODY.WPACIS.1." => "Search"}); + +# Output an HTTP response header +print $query->header(-type => "text/xml"); + +# Create an XML output stream +my $Writer = XML::Writer->new(DATA_MODE => 1); + +# Output the XML header data +$Writer->xmlDecl("UTF-8"); +$Writer->startTag("planning"); +$Writer->dataElement("authority_name", "Enfield Council"); +$Writer->dataElement("authority_short_name", "Enfield"); +$Writer->startTag("applications"); + +# Output any applications on the first page +output_applications($page); + +# Loop over any additional results pages +foreach my $link ($page->look_down("_tag" => "a", "href" => qr/^WPHAPPSEARCHRES\.displayResultsURL/)) +{ + # Fetch this page... + $page = do_get(URI->new_abs($link->attr("href"), $SearchURL)); + + # ...and output the applications from it + output_applications($page); +} + +# Finish off XML output +$Writer->endTag("applications"); +$Writer->endTag("planning"); +$Writer->end(); + +exit 0; + +# Make a GET request +sub do_get +{ + my $response = $UA->get(@_); + + die $response->status_line unless $response->is_success; + + return HTML::TreeBuilder->new_from_content($response->content); +} + +# Make a POST request +sub do_post +{ + my $response = $UA->post(@_); + + die $response->status_line unless $response->is_success; + + return HTML::TreeBuilder->new_from_content($response->content); +} + +# Output applications from a results page +sub output_applications +{ + my $page = shift; + + # Find the result table + my $table = $page->look_down("_tag" => "table", "class" => "apas_tbl"); + + # Process each row of the results + foreach my $row ($table->look_down("_tag" => "tr")) + { + my @cells = $row->look_down("_tag" => "td"); + + if (@cells >= 3) + { + my $reference = $cells[0]->as_trimmed_text; + my $description = $cells[1]->as_trimmed_text; + my $address = $cells[2]->as_trimmed_text; + my $postcode; + + if ($address =~ /\s+([A-Z]+\d+\s+\d+[A-Z]+)$/) + { + $postcode = $1; + } + + $Writer->startTag("application"); + $Writer->dataElement("council_reference", $reference); + $Writer->dataElement("address", $address); + $Writer->dataElement("postcode", $postcode); + $Writer->dataElement("description", $description); + $Writer->dataElement("info_url", $InfoURL . $reference); + $Writer->dataElement("date_received", $date); + $Writer->endTag("application"); + } + } + + return; +}