//save the record scraperwiki::save(array('id'), $record); } } require 'scraperwiki/simple_html_dom.php'; ini_set('max_execution_time', 600); setlocale(LC_ALL, 'fr_FR.UTF8'); //the firt time, we set manually the first page $value = "/recherche/resultats/map/Centre/tr/by"; //if $value is a valid page while ($value != "") { $htmlCentre = scraperWiki::scrape("http://www.athome.lu" . $value); $domCentre = new simple_html_dom(); $domCentre->load($htmlCentre); //find the announces in the current page and store the records findAnnounces($domCentre); //look for the next page link $value = ""; foreach ($domCentre->find('.next') as $data) { $value = $data->href; } } /***************** Functions *********************/ function findAnnounces($strDataDom) { //for each page, there are up to 10 records, each of them marked with DOM class 'plus' foreach ($strDataDom->find('.plus') as $data) { $value = $data->href; //is it a valid URL for an announce? if (strrpos($value, "www.athome.lu/")) { //go into the announce
} } } require 'scraperwiki/simple_html_dom.php'; ini_set('max_execution_time', 1200); setlocale(LC_ALL, 'fr_FR.UTF8'); $counter = 0; //the first time, set manually the first page $value = "/recherche/resultats/tr/rt/w/d,Luxembourg%7CBelair,Limpertsberg,Rollingergrund,Merl%3Ba,Bertrange%3Ba,Strassen/ig/h,f"; //while $value is not blank while ($value != "") { $htmlCentre = scraperWiki::scrape("http://www.athome.lu" . $value); $domCentre = new simple_html_dom(); $domCentre->load($htmlCentre); //find the announces in the current page and store the records findAnnounces($domCentre, $counter); //look for the next page link $value = ""; foreach ($domCentre->find('.next') as $data) { $value = $data->href; } } //if there are no records, insert a blank record /* if ($counter == 0){ $record=array(); $record["id"] = 0; $record["url"] = "No records for today :("; $record["submitter"] = 0; //save the record scraperwiki::save(array('id'), $record);