function fetch_page_email($pageNo)
{
$DOT = ".";
$g_ofile = "mc.page" . $pageNo . $DOT . "email";
$fhandle = fopen($g_ofile, "w");
//get page
$pageUrl = "http://www.mediaclubofindia.com/profiles/friend/list?page=" . $pageNo;
$pageHtml = fetch_page($pageUrl);
printf("page = %s \n", $pageUrl);
$profiles = get_profile_url($pageHtml);
//print_r($profiles); exit ;
foreach ($profiles as $profile) {
$buffer = get_profile_email($profile);
fwrite($fhandle, $buffer);
}
}
function main_loop()
{
$root = "http://toostep.com/knowledge/a/all/popular/";
$ofile = "toostep.email";
$fhandle = fopen($ofile, "a");
//rajeev - till 1000
for ($page = 101; $page <= 200; $page++) {
$pageUrl = $root . $page;
$profiles = get_profile_url($pageUrl);
foreach ($profiles as $profile) {
$data = get_profile_email($profile);
$buffer = NULL;
if (empty($data) || is_null($data)) {
$buffer = sprintf("__NO_DATA__ %s \n", $profile);
} else {
$buffer = sprintf("__DATA__ %s|%s|%s|%s \n", $data["email"], $data["name"], $data["title"], $data["company"]);
}
fwrite($fhandle, $buffer);
}
}
//close resources
fclose($fhandle);
}
function main_loop()
{
global $g_root;
global $g_ofile;
$fhandle = fopen($g_ofile, "w");
//rajeev - till 1000
for ($page = 201; $page <= 300; $page++) {
$pageUrl = $g_root . $page;
$profiles = get_profile_url($pageUrl);
foreach ($profiles as $profile) {
$data = get_profile_email($profile);
$buffer = NULL;
if (empty($data) || is_null($data)) {
$buffer = sprintf("__NO_DATA__ %s \n", $profile);
} else {
$buffer = sprintf("__DATA__ %s|%s|%s|%s \n", $data["email"], $data["name"], $data["title"], $data["company"]);
}
fwrite($fhandle, $buffer);
}
}
//close resources
fclose($fhandle);
}
function run_on_name($ch, $name, $pageNo)
{
global $g_debug;
$DOT = ".";
$g_ofile = urlencode($name) . $DOT . time() . $DOT . "email";
$fhandle = fopen($g_ofile, "w");
while (1) {
$dtime = time() . "999";
$params = array("pageNo" => $pageNo, "type" => "page", "archive" => "0", "d" => $dtime);
$qstring = http_build_query($params);
$pageUrl = "http://toostep.com/searchUserAjax.html?" . $qstring;
if ($g_debug) {
print_r($params);
printf("page = %s \n", $pageUrl);
}
$pageHtml = do_post($ch, $pageUrl, $name);
if ($g_debug) {
echo $pageHtml;
}
$profiles = get_profile_url($pageHtml);
if (sizeof($profiles) == 0) {
fclose($fhandle);
//name processing finished
return;
}
//print_r($profiles); exit ;
//use profiles to get emails
foreach ($profiles as $profile) {
$data = get_profile_email($profile);
$buffer = NULL;
if (empty($data) || is_null($data)) {
$buffer = sprintf("__NO_DATA__ %s \n", $profile);
} else {
$buffer = sprintf("__DATA__ %s|%s|%s|%s \n", $data["email"], $data["name"], $data["title"], $data["company"]);
}
fwrite($fhandle, $buffer);
}
$pageNo++;
}
}
function run_on_name($name, $pageNo, $lookup)
{
global $g_debug;
$ch = curl_init();
curl_setopt($ch, CURLOPT_USERAGENT, "Mozilla/5.0 (X11; Linux x86_64; rv:10.0.7) Gecko/20100101 Firefox/10.0.7 Iceweasel/10.0.7");
curl_setopt($ch, CURLOPT_TIMEOUT, 60);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
curl_setopt($ch, CURLOPT_FOLLOWLOCATION, 1);
$prefix = date("d-M-y");
$g_ofile = $prefix . "/" . $name . ".email." . time();
if (!file_exists(dirname($g_ofile))) {
mkdir(dirname($g_ofile), 0755, true);
}
$fhandle = fopen($g_ofile, "w");
while (1) {
$dtime = time() . "999";
$params = array("pageNo" => $pageNo, "type" => "page", "archive" => "0", "d" => $dtime);
$qstring = http_build_query($params);
$pageUrl = "http://toostep.com/searchUserAjax.html?" . $qstring;
$pageHtml = do_post($ch, $pageUrl, $name);
if ($g_debug) {
printf(" \n\n ------ dumping search page having profiles ----- \n\n");
echo $pageHtml;
}
$profiles = get_profile_url($pageHtml);
if (sizeof($profiles) == 0) {
//free resources
curl_close($ch);
fclose($fhandle);
//name processing finished
return;
}
if ($g_debug) {
printf(" \n\n ------- dumping extracted profiles from search page ----- \n\n ");
print_r($profiles);
}
//use profiles to get emails
foreach ($profiles as $profile) {
$data = get_profile_email($profile);
$buffer = NULL;
if (empty($data) || is_null($data)) {
$buffer = sprintf("__NO_DATA__ %s \n", $profile);
} else {
$buffer = sprintf("__DATA__ %s|%s|%s|%s \n", $data["email"], $data["name"], $data["title"], $data["company"]);
}
fwrite($fhandle, $buffer);
if ($g_debug) {
printf(" \n ----- wrote [%s] ----- \n ", $buffer);
}
}
printf("processed %s - page %s \n", $name, $pageNo);
$pageNo++;
$lookup[$name] = $pageNo;
//save name -page_no to disk
save_lookup($lookup);
}
}
