<?php
require_once "config.php";
$xml = safe_scrape_page('http://www.wansbeck.gov.uk/planning.cfm?day=17&month=7&year=2007');
$parsed_applications = simplexml_load_string($xml);
//Loop through the applications, add tinyurl / google maps etc and add to array
if (sizeof($parsed_applications) > 0) {
foreach ($parsed_applications->applications->application as $parsed_application) {
}
}
print "ddd";
function parse_applications($feed_url, $authority_id)
{
$return_applications = array();
//reset warnings
//Grab the XML
$xml = "";
try {
$xml = safe_scrape_page($feed_url);
} catch (exception $e) {
array_push($this->log, "ERROR: problem occured when grabbing feed: " . $feed_url . " ---->>>" . $e);
}
if ($xml == false) {
$this->store_log("ERROR: empty feed feed: " . $feed_url);
}
//Turn the xml into an object
$parsed_applications = simplexml_load_string($xml);
//Loop through the applications, add tinyurl / google maps etc and add to array
if (sizeof($parsed_applications) > 0) {
foreach ($parsed_applications->applications->application as $parsed_application) {
$application = new application();
//Grab basic data from the xml
$application->authority_id = $authority_id;
$application->council_reference = $parsed_application->council_reference;
$date_received_dmy = split("/", $parsed_application->date_received);
if (count($date_received_dmy) == 3) {
$application->date_received = "{$date_received_dmy['2']}-{$date_received_dmy['1']}-{$date_received_dmy['0']}";
} else {
// Make a best effort attempt to parse the date
$ts = strtotime($parsed_application->date_received);
if ($ts != FALSE && $ts != -1) {
$application->date_received = date("Y-m-d", $ts);
}
}
$application->address = $parsed_application->address;
$application->description = $parsed_application->description;
$application->info_url = $parsed_application->info_url;
$application->comment_url = $parsed_application->comment_url;
$application->date_scraped = mysql_date(time());
//Make the urls
$info_tiny_url = tiny_url($application->info_url);
if ($info_tiny_url == "") {
$this->store_log("ERROR: Created blank info tiny url");
}
$comment_tiny_url = tiny_url($application->comment_url);
if ($comment_tiny_url == "") {
$this->store_log("ERROR: Created blank comment tiny url");
}
if (isset($parsed_application->postcode)) {
//Workout the XY location from postcode
$xy = postcode_to_location($parsed_application->postcode);
$application->postcode = $parsed_application->postcode;
$application->x = $xy[0];
$application->y = $xy[1];
} else {
if (isset($parsed_application->easting) && isset($parsed_application->northing)) {
$postcode = location_to_postcode($parsed_application->easting, $parsed_application->northing);
$application->postcode = $postcode;
$application->x = $parsed_application->easting;
$application->y = $parsed_application->northing;
}
}
$application->info_tinyurl = $info_tiny_url;
$application->comment_tinyurl = $comment_tiny_url;
$application->map_url = googlemap_url_from_postcode($application->postcode);
//Add to array
array_push($return_applications, $application);
}
}
return $return_applications;
}
function scrape_applications_islington($search_url, $info_url_base, $comment_url_base)
{
$applications = array();
$application_pattern = '/<TR>([^<]*)<TD class="lg" valign="top" >([^<]*)<a href([^<]*)<a href=wphappcriteria.display>Search Criteria(.*)([^<]*)<(.*)>([^<]*)<TD class="lg" >([^<]*)<\\/TD>([^<]*)<TD class="lg" >([^<]*)<INPUT TYPE=HIDDEN NAME([^>]*)([^<]*)/';
//grab the page
$html = safe_scrape_page($search_url);
preg_match_all($application_pattern, $html, $application_matches, PREG_PATTERN_ORDER);
foreach ($application_matches[0] as $application_match) {
$application_string = str_replace("\n", "", $application_match);
$reference_pattern = '/Search Results<\\/a>">([^<]*)/';
preg_match_all($reference_pattern, $application_string, $reference_matches, PREG_PATTERN_ORDER);
$application = new Application();
//match the applicaiton number
$application->council_reference = str_replace('Search Results</a>">', "", $reference_matches[0][0]);
//Comment and info urls
$application->info_url = $info_url_base . $application->council_reference;
$application->comment_url = $comment_url_base . $application->council_reference;
//get full details
$details_html = "";
$details_html = safe_scrape_page($info_url_base . $application->council_reference);
$details_html = str_replace("\r\n", "", $details_html);
//Details
$full_detail_pattern = '/Proposal:<\\/label><\\/td>([^<]*)<td colspan="3">([^<]*)/';
preg_match($full_detail_pattern, $details_html, $full_detail_matches);
if (isset($full_detail_matches[2])) {
$application->description = $full_detail_matches[2];
}
//Address
$address_pattern = '/Main location:<\\/label><\\/td>([^<]*)<td colspan="3">([^<]*)/';
$address = "";
preg_match($address_pattern, $details_html, $address_matches);
if (isset($address_matches[2])) {
$application->address = $address_matches[2];
}
//postcode
$postcode_pattern = "/[A-Z][A-Z]?[0-9][A-Z0-9]? ?[0-9][ABDEFGHJLNPQRSTUWXYZ]{2}/";
preg_match($postcode_pattern, $application->address, $postcode_matches);
if (isset($postcode_matches[0])) {
$application->postcode = $postcode_matches[0];
}
//only add it if we have a postcode (bit useless otherwise)
if (is_postcode($application->postcode)) {
array_push($applications, $application);
}
}
//return
return $applications;
}
