domains gotomyjob.com, first local database, phpmyadmin=> INSERT INTO `prio` ( `id` , `source1` , `priority` ) VALUES (NULL , 'XXXXXXputHere$source', '0');
$_GET['bx']=1;//1 means no DB insert
$testmode=1;//stop looping at 5th result
$local_debug=0;//0 means open website location, 1 means read data file from local server
//###debug-end###
include "func.php";
$counter=1;
$source="jobbuilding.be";
$company="Job Building";
$today=getdate();
$month=$today['mon'];
$mday=$today['mday'];
$year=$today['year'];
$hours= "00";
$minutes="00";
$seconds="00";
$today=$year.'-'.$month.'-'.$mday.' '.$hours.':'.$minutes.':'.$seconds;
$date=$today;
$link_base="https://www.jobbuilding.be/";
echo "".$link_base."
"."\n";
$industry_arr['1.'] = "Ontwikkeling & Expansie";
$industry_arr['2.'] = "Beheer & Management";
$industry_arr['3.'] = "Deskundigen";
$industry_arr['4.'] = "Bouw & Techniek";
$industry_arr['5.'] = "Verhuur & Verkoop";
$industry_arr['6.'] = "Financieel & Verzekeringen";
$industry_arr['7.'] = "Ontwerp & Creatief";
$industry_arr['8.'] = "Juridisch";
$industry_arr['9.'] = "Nevenfuncties";
$industry_arr['10.'] = "Administratief";
$industry_arr['11.'] = "Consultancy";
$industry_arr['12.'] = "Franchising & Filiaalhouders";
$industry_arr['13.'] = "Directie & Topmanagement";
$industry_arr['14.'] = "Overheidsdiensten";
$industry_arr['374.'] = "Opleiding & Onderzoek";
$industry_arr['381.'] = "Andere functies";
$snoopy=new Snoopy;
$snoopy->agent="Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1)";
$snoopy->referer="http://www.google.com/";
$snoopy->rawheaders["Pragma"]="no-cache";
$snoopy->maxredirs=2;
$snoopy->offsiteok=false;
$snoopy->expandlinks=false;
$snoopy->passcookies=true;
$snoopy->set_submit_normal();
$snoopy->setcookies();
if(!sizeof($industry_arr)) $industry_arr=array('');
foreach ($industry_arr as $cat_link=>$industry) {
$already_submitted=0;
$next_page_exists=true;
$page_nr=1;
while ($next_page_exists) {
$link_base_while=$link_base."Search.aspx?&p=3.&q=&page=".($page_nr-1);
echo "".$link_base_while."".($industry ? "\n"."
".$industry." Jobs" : "")."
"."\n";
if(!$local_debug) {
$snoopy->fetch($link_base_while);
$htmlsource=$snoopy->results;
} else {
//$htmlsource = getSource ('http://computer4.server.intranet/clients/other/gotomyjob.com/cron/_tmp/www.nationalevacaturebank.nl_vacatures.html');
}
if (!eregi('type="button" name="Laatste"',$htmlsource)) {
$next_page_exists=false;
}
echo $htmlsource;exit;
ExtractAndDelete($htmlsource,'Volgende pagina','<');
do {
echo "
";
/* $date_extracted = ExtractAndDelete($htmlsource,'
','<');
if($link!="-1" && $date_extracted!="-1") {
$date_exploded=explode("/",trim($date_extracted));
$date=$date_exploded[2]."-".(strlen($date_exploded[1])==1?"0":"").$date_exploded[1]."-".(strlen($date_exploded[0])==1?"0":"").$date_exploded[0];
if (strtotime($date) <= strtotime(date('Y-m-d', mktime(date('H'),0,0,date('m'),date('d')-$delete_days,date('Y'))))) {
echo "
Expired Jobs
";
continue;
}
} // */
$link=ExtractAndDelete($htmlsource,'="ab" href=\'','\'');
$title=ExtractAndDelete($htmlsource,'>','<');
$title=preg_replace("/^\((.*)\)/","",$title);
$city=ucwords(strtolower(ExtractAndDelete($htmlsource,'','<')));
$link=$link_base.$link;
echo "Details Page: ".$link."
"."\n";
//$industry
$occupation=$title;
if(substr($link,-2)=="-1")
$link="-1";
if($link!="-1") { //*
$snoopy->fetch($link);
$htmlsource1=$snoopy->results;
ExtractAndDelete($htmlsource1,'Functieomschrijving','');
$description=RemoveAllTrash(ExtractAndDelete($htmlsource1,' | ','
'));
// */
}
$thetime="fulltime";
$hhh=strtolower($htmlsource1);
if(strpos($hhh,"tijdelijk")>0)$thetime="parttime";
else
if(strpos($hhh,"parttime")>0)$thetime="parttime";
else
if(strpos($hhh,"part time")>0)$thetime="parttime";
if(strcmp($link,"-1")!=0){
//$link=$link_base.$link;
echo "\n"."
--".$counter++."--";
$already_submitted+=AddToDatabase($title,$link,$description,$city,$date,$source,$occupation,$industry,$company,$thetime,$thelogo);
}
if($already_submitted>=3) {
echo "\n"."
Found 3 entries already submitted in the database. Exiting ...
"."\n";
break 2;
}
//###debug-start### after the last AddToDatabase, but ouside the condition
if(strcmp($link,"-1")!=0) {
if($title=="-1" | $title=="")
$bad_title[$link_base.$industry_link]++;
if($description=="-1" | $description=="")
$bad_description[$link_base.$industry_link]++;
if($city=="-1" | $city=="")
$bad_city[$link_base.$industry_link]++;
if($occupation=="-1" | $occupation=="")
$bad_occupation[$link_base.$industry_link]++;
if($industry=="-1" | $industry=="")
$bad_industry[$link_base.$industry_link]++;
if($company=="-1" | $company=="")
$bad_company[$link_base.$industry_link]++;
if ($testmode && (($counter>5 && $testmode==1) | ($counter>$testmode && $testmode!=1))) {
echo "\n"."
Stopped at ".($counter)." job. Exiting...";
cron_debug_check_errors();
exit;
}
} else {
$bad_link[$link_base.$industry_link]++;
}
//###debug-end###
ob_flush();
flush();
if(substr($link,-2)=="-1")
echo "\n"."
--- BAD LINK ---------------------------------------------------------------------------
"."\n";
} while (strcmp($link,"-1"));
echo "############################################################################
"."\n"."End of Page: ".($page_nr)."\n"."
############################################################################
"."\n";
$page_nr+=1;
}
}
echo "\n"."
------------------------------------------"."\n"."
Finishing the process!!!
"."\n";
//###debug-start### bottom of page
function cron_debug_check_errors() {
global $bad_link, $bad_title, $bad_description, $bad_city, $bad_occupation, $bad_industry, $bad_company;
echo "\n"."
Debug messages:
"."\n";
if(sizeof($bad_link)) {
foreach ($bad_link as $key=>$value) {
echo "\n"."
Bad links: Error in URL: ".$key." - nr. of bad links: ".$value."";
}
}
if(sizeof($bad_title)) {
foreach ($bad_title as $key=>$value) {
echo "\n"."
Bad titles: Error in URL: ".$key." - nr. of bad titles: ".$value."";
}
}
if(sizeof($bad_description)) {
foreach ($bad_description as $key=>$value) {
echo "\n"."
Bad descriptions: Error in URL: ".$key." - nr. of bad descriptions: ".$value."";
}
}
if(sizeof($bad_city)) {
foreach ($bad_city as $key=>$value) {
echo "\n"."
Bad cities:Error in URL: ".$key." - nr. of bad cities: ".$value."";
}
}
if(sizeof($bad_occupation)) {
foreach ($bad_occupation as $key=>$value) {
echo "\n"."
Bad occupations: Error in URL: ".$key." - nr. of bad occupations: ".$value."";
}
}
if(sizeof($bad_industry)) {
foreach ($bad_industry as $key=>$value) {
echo "\n"."
Bad industries: Error in URL: ".$key." - nr. of bad industries: ".$value."";
}
}
if(sizeof($bad_company)) {
foreach ($bad_company as $key=>$value) {
echo "\n"."
Bad companies: Error in URL: ".$key." - nr. of bad companies: ".$value."";
}
}
}
echo "\n\n"."
";
cron_debug_check_errors();
/* // generate the necessary code for this cron file: select the code below then press CTRL+[-] then use it
print("\n------------------------------------------------------------------------------------------------------------------------")
local job_site=''
for m in editor:match("[$]source[ ]*=[ ]*.*", SCFIND_REGEXP) do
job_site=string.gsub(m.text,"(.*)=\"(.*)\"(.*)","%2")
end
print("INSERT INTO `prio` ( `id` , `source1` , `priority` ) VALUES (NULL , '"..job_site.."', '0');")
print("-> upload cron_bx_xxx.php file");
print("add to CRON_STAT.PHP : $job_sites['"..job_site.."']=\""..job_site.."\";");
print("11 23 * * * /usr/bin/php /var/www/vhosts/gotomyjob.com/httpdocs/"..filename().." >/dev/null");
print("------------------------------------------------------------------------------------------------------------------------")
*/
//###debug-end###
?>