function tdr_promotions_social_media_scraper($social_network)
{
// Get list of campaigns
$promotion = new tdr_promotions();
$campaign_list = $promotion->get_the_decoded_promotions();
// Run Scrape for $campaign_slug with contest page url $contest_permalink
foreach ($campaign_list as $campaign_slug => $campaign_data) {
if ($campaign_data['campaign_ready'] === true) {
// Only scrape campaigns in production status
// Find contest page permalink based on campaign slug
// Get list of pages with the contest template
$contest_page_list = get_pages(array('meta_key' => '_wp_page_template', 'meta_value' => 'template-contest-' . $campaign_slug . '.php', 'hierarchical' => false));
$contest_page_found = false;
foreach ($contest_page_list as $contest_page) {
if ($campaign_data['url_slug'] === $contest_page->post_name) {
$contest_page_id = $contest_page->ID;
$contest_page_found = true;
break;
}
}
if ($contest_page_found) {
$contest_permalink = get_permalink($contest_page_id);
// Contest permalink for scraping
}
$db_slug = str_replace('-', '_', $campaign_slug);
switch ($social_network) {
case 'facebook':
$api_base_url = 'https://graph.facebook.com/search';
$transient_name = 'tdr_promo_fb_scr_' . $db_slug;
/*
$site_title_format = get_option( 'tdr_promotions_site_title_function' ); // Get title function for site -- extract from within
// Site title function is defined and is callable
if ( ( false !== $site_title_format ) && ( function_exists( $site_title_format ) ) ) {
$contest_page_title = $site_title_format( $contest_page->ID );
$contest_page_title = str_replace('|', '\\|', $contest_page_title );
}
// Fallback to page title
else {
$contest_page_title = get_the_title( $contest_page->ID );
}
*/
// DEBUG echo( 'using ' . $contest_page_title . ' for facebook scrapes' );
$request_array = array('q' => $campaign_data['facebook_share_title'], 'type' => 'post', 'access_token' => '', 'limit' => '100', 'date_format' => 'U');
break;
default:
return;
}
// If transient not set, do new query
$refresh_api_query = get_transient($transient_name);
if (true) {
// DEBUG echo('no transient found');
switch ($social_network) {
case 'facebook':
$api_request = $api_base_url . '?' . http_build_query($request_array);
break;
}
// END STANDARD QUERY
} else {
// DEBUG echo('transient was found');
switch ($social_network) {
case 'facebook':
$api_request = $refresh_api_query;
// Transient holds full request url
break;
}
}
// Perform API GET request
// DEBUG echo('making request to '. $api_request );
$wp_remote_get_args = array('method' => 'GET', 'timeout' => 25, 'redirection' => 5, 'user-agent' => 'tdr-promos/1.0', 'blocking' => true, 'compress' => true, 'decompress' => true, 'sslverify' => true, 'httpversion' => '1.1', 'headers' => array(), 'body' => null, 'cookies' => array());
$request_results = wp_remote_get($api_request, $wp_remote_get_args);
// On errors, retry with exponential backing off
$wait_time = 1;
while (is_wp_error($request_results)) {
// Api request failure -- retry with exponential backing off
sleep($wait_time);
$request_results = wp_remote_get($api_request, $wp_remote_get_args);
if ($wait_time > 600) {
return;
// End function on repeated errors beyond 10 minutes retry wait ( aggregate of 21 minutes after timeouts )
}
$wait_time = $wait_time * 2;
/*DEBUG var_dump( $request_results );
echo( 'first request died' );
echo( '<br>' . $api_request ); */
}
$parsed_results = json_decode($request_results['body'], true);
// Parse response JSON
// Update transient with refresh URL
switch ($social_network) {
case 'facebook':
if (!empty($parsed_results['data'])) {
// Get unix timestamp of most recent result
$most_recent_result = $parsed_results['data'][0]['created_time'];
// Most recent result is returned first by fb graph api
$request_array['since'] = $most_recent_result;
// Add since parameter to request array
$refresh_query = $api_base_url . '?' . http_build_query($request_array);
// Build url for api refresh request
} else {
$refresh_query = $api_request;
// Next time repeat a normal query if there were no results this time
}
break;
}
set_transient($transient_name, $refresh_query, 60 * 60 * 24 * 7);
// Give transient an expiration of a week
$referral_id_list = array();
// Create empty array to hold referall ids from search results
// Look through the search results
switch ($social_network) {
case 'facebook':
$i = 0;
// Set a loop counter so pagination links can be followed after the first iteration
$pagination_api_request = '';
// Set a variable for holding pagination urls in scope for all iterations of the loop
$time_array = array();
while (true) {
// Loop until the results are empty
if ($i > 4) {
break;
}
// Follow pagination links for new iterations of the loop
if ($i > 0) {
/* PROFILING CODE
// Start time logging*/
//$start_time = microtime( true );
$request_results = wp_remote_get($pagination_api_request, $wp_remote_get_args);
//$end_time = microtime( true );
//$execution_time = round( $end_time - $start_time, 4 );
//$time_array [] = $execution_time;
// END PROFILING CODE
$timeout = false;
$wait_time = 1;
while (is_wp_error($request_results)) {
// Api request failure -- retry with exponential backing off
sleep($wait_time);
$request_results = wp_remote_get($pagination_api_request, $wp_remote_get_args);
if ($wait_time > 150) {
$timeout = true;
break;
// End function on repeated errors beyond 2.5 minutes retry wait ( aggregate of ~7.5 minutes after timeouts )
}
$wait_time = $wait_time * 2;
/* DEBUG echo( 'died at iteration ' . $i );
var_dump( $pagination_api_request );
var_dump( $request_results ); */
}
if ($timeout) {
break;
// Stop scraping and process any collected entries
}
$parsed_results = json_decode($request_results['body'], true);
// Parse response JSON
}
// Get pagination link, if any
if (array_key_exists('paging', $parsed_results)) {
$pagination_link = $parsed_results['paging']['next'];
$pagination_api_request = $pagination_link;
} else {
break;
// End of results-- end loop
}
// Filter the pagination so only non-scraped results are shown
$url_params = parse_url($api_request, PHP_URL_QUERY);
// Get query string from link
parse_str($url_params, $url_param_array);
// Parse query string into array
if (array_key_exists('since', $url_param_array)) {
// Check if there was a since parameter in the original request
$pagination_api_request .= '&since=' . $url_param_array['since'];
// Append it to facebook's suggested pagination links
}
// Loop through the results for the current page
foreach ($parsed_results['data'] as $result) {
// Look for relevant entity
$url_params = parse_url($result['link'], PHP_URL_QUERY);
// Get query string from link
parse_str($url_params, $url_param_array);
// Parse query string into array
if (array_key_exists('ref', $url_param_array)) {
// 'ref' param was found
$located_referral_id = $url_param_array['ref'];
// Add to list
$located_referral_id = strtoupper($located_referral_id);
// Capitalize referral ids just in case
// User previously found in results
if (array_key_exists($located_referral_id, $referral_id_list)) {
$referral_id_list[$located_referral_id]++;
// Increment user's count
} else {
$referral_id_list[$located_referral_id] = 1;
// Add referral id to list
}
} else {
// Not found
// DEBUG echo('no match');
}
}
$i++;
// Increment loop counter
sleep(1);
// Be nice to facebook and impose a rate limit of 1 second between loops
}
// Write to /tmp
break;
}
/*
* PROFILING CODE
$fp = fopen('/tmp/twitterscrape.txt', 'w');
fwrite($fp, $i);
fwrite($fp, "\n min " . min( $time_array ) . " / max " . max( $time_array ) . " / average " . array_sum($time_array) / count(array_filter($time_array)) . " / total " . array_sum( $time_array ) );
fclose($fp);
*/
// DEBUG var_dump( $referral_id_list );
// User query for referral ID IN list
// Credit with tweet/facebook post
$query_referral_id_list = array_keys($referral_id_list);
// Define the wordpress user query parameters
$contest_user_query_args = array('role' => 'promo-' . $campaign_slug, 'meta_key' => 'referral_id', 'meta_value' => $query_referral_id_list, 'meta_compare' => 'IN');
// Perform the user query
$contest_user_query = new WP_User_Query($contest_user_query_args);
// If results found, fetch them
if ($contest_user_query->total_users != "0") {
$contest_users_from_scrape = $contest_user_query->get_results();
foreach ($contest_users_from_scrape as $contest_user) {
$current_contest_entries = get_user_meta($contest_user->ID, 'contest_entries', true);
// Get the current meta value
$referral_id = get_user_meta($contest_user->ID, 'referral_id', true);
// Get the user's referral id
$current_contest_entries['sharing'][$social_network] += $referral_id_list[$referral_id];
// Credit the user for the number of new shares found
// Perform the meta update
update_usermeta($contest_user->ID, 'contest_entries', $current_contest_entries);
tdr_check_entry_cap($contest_user->ID, $campaign_slug);
// Check to see if entry cap is reached; Move to capped group if so
// DEBUG var_dump( $contest_user->ID );
}
}
// DEBUG var_dump( $contest_user_query );
// DEBUG var_dump( $contest_users_from_scrape );
//$fp = fopen('/tmp/twitterscrape.txt', 'w');
//fwrite($fp, $request_results);
//fwrite($fp, var_dump( $user_list ) );
//fclose($fp);
// Write to /tmp
}
}
}
