<?php
include_once __DIR__ . '/../config.php';
include_once __DIR__ . '/../common/constants.php';
if (defined("ADMIN_USER") && ADMIN_USER != "" && (!isset($_SERVER['PHP_AUTH_USER']) || $_SERVER['PHP_AUTH_USER'] != ADMIN_USER)) {
die("Go away, you evil hacker!");
}
include_once __DIR__ . '/query_manager.php';
include_once __DIR__ . '/../common/functions.php';
include_once __DIR__ . '/../common/upgrade.php';
include_once __DIR__ . '/../capture/common/functions.php';
create_admin();
create_error_logs();
$captureroles = unserialize(CAPTUREROLES);
$querybins = getBins();
$activePhrases = getNrOfActivePhrases();
$activeGeoboxes = getNrOfActiveGeoboxes();
$activeUsers = getNrOfActiveUsers();
$lastRateLimitHit = getLastRateLimitHit();
?>
<html>
<head>
<title>DMI-TCAT query manager</title>
<meta charset='<?php
echo mb_internal_encoding();
?>
'>
<style type="text/css">
body,html { font-family:Arial, Helvetica, sans-serif; font-size:12px; }
function tracker_run()
{
global $dbuser, $dbpass, $database, $hostname, $tweetQueue;
// We need the tcat_status table
create_error_logs();
// We need the tcat_captured_phrases table
create_admin();
$tweetQueue = new TweetQueue();
$tweetQueue->setoption('replace', false);
if (defined('USE_INSERT_DELAYED') && USE_INSERT_DELAYED) {
$tweetQueue->setoption('delayed', true);
}
if (defined('DISABLE_INSERT_IGNORE') && DISABLE_INSERT_IGNORE) {
$tweetQueue->setoption('ignore', false);
} else {
$tweetQueue->setoption('ignore', true);
}
if (!defined("CAPTURE")) {
/* logged to no file in particular, because we don't know which one. this should not happen. */
error_log("tracker_run() called without defining CAPTURE. have you set up config.php ?");
die;
}
$roles = unserialize(CAPTUREROLES);
if (!in_array(CAPTURE, $roles)) {
/* incorrect script execution, report back error to user */
error_log("tracker_run() role " . CAPTURE . " is not configured to run");
die;
}
// log execution environment
$phpstring = phpversion() . " in mode " . php_sapi_name() . " with extensions ";
$extensions = get_loaded_extensions();
$first = true;
foreach ($extensions as $ext) {
if ($first) {
$first = false;
} else {
$phpstring .= ',';
}
$phpstring .= "{$ext}";
}
$phpstring .= " (ini file: " . php_ini_loaded_file() . ")";
logit(CAPTURE . ".error.log", "running php version {$phpstring}");
// install the signal handler
if (function_exists('pcntl_signal')) {
// tick use required as of PHP 4.3.0
declare (ticks=1);
// See signal method discussion:
// http://darrendev.blogspot.nl/2010/11/php-53-ticks-pcntlsignal.html
logit(CAPTURE . ".error.log", "installing term signal handler for this script");
// setup signal handlers
pcntl_signal(SIGTERM, "capture_signal_handler_term");
} else {
logit(CAPTURE . ".error.log", "your php installation does not support signal handlers. graceful reload will not work");
}
// sanity check for geo bins functions
if (geophp_sane()) {
logit(CAPTURE . ".error.log", "geoPHP library is fully functional");
} elseif (geobinsActive()) {
logit(CAPTURE . ".error.log", "refusing to track until geobins are stopped or geo is functional");
exit(1);
} else {
logit(CAPTURE . ".error.log", "geoPHP functions are not yet available, see documentation for instructions");
}
global $rl_current_record, $rl_registering_minute;
global $last_insert_id;
global $tracker_started_at;
$rl_current_record = 0;
// how many tweets have been ratelimited this MINUTE?
$rl_registering_minute = get_current_minute();
// what is the minute we are registering (as soon as the current minute differs from this, we insert our record in the database)
$last_insert_id = -1;
// needed to make INSERT DELAYED work, see the function database_activity()
$tracker_started_at = time();
// the walltime when this script was started
global $twitter_consumer_key, $twitter_consumer_secret, $twitter_user_token, $twitter_user_secret, $lastinsert;
$pid = getmypid();
logit(CAPTURE . ".error.log", "started script " . CAPTURE . " with pid {$pid}");
$lastinsert = time();
$procfilename = __DIR__ . "/../../proc/" . CAPTURE . ".procinfo";
if (file_put_contents($procfilename, $pid . "|" . time()) === FALSE) {
logit(CAPTURE . ".error.log", "cannot register capture script start time (file \"{$procfilename}\" is not WRITABLE. make sure the proc/ directory exists in your webroot and is writable by the cron user)");
die;
}
$networkpath = isset($GLOBALS["HOSTROLE"][CAPTURE]) ? $GLOBALS["HOSTROLE"][CAPTURE] : 'https://stream.twitter.com/';
// prepare queries
if (CAPTURE == "track") {
// check for geolocation bins
$locations = geobinsActive() ? getActiveLocationsImploded() : false;
// assemble query
$querylist = getActivePhrases();
if (empty($querylist) && !geobinsActive()) {
logit(CAPTURE . ".error.log", "empty query list, aborting!");
return;
}
$method = $networkpath . '1.1/statuses/filter.json';
$track = implode(",", $querylist);
$params = array();
if (geobinsActive()) {
$params['locations'] = $locations;
}
if (!empty($querylist)) {
$params['track'] = $track;
}
} elseif (CAPTURE == "follow") {
$querylist = getActiveUsers();
if (empty($querylist)) {
logit(CAPTURE . ".error.log", "empty query list, aborting!");
return;
}
$method = $networkpath . '1.1/statuses/filter.json';
$params = array("follow" => implode(",", $querylist));
} elseif (CAPTURE == "onepercent") {
$method = $networkpath . '1.1/statuses/sample.json';
$params = array('stall_warnings' => 'true');
}
logit(CAPTURE . ".error.log", "connecting to API socket");
$tmhOAuth = new tmhOAuth(array('consumer_key' => $twitter_consumer_key, 'consumer_secret' => $twitter_consumer_secret, 'token' => $twitter_user_token, 'secret' => $twitter_user_secret, 'host' => 'stream.twitter.com'));
$tmhOAuth->request_settings['headers']['Host'] = 'stream.twitter.com';
if (CAPTURE == "track" || CAPTURE == "follow") {
logit(CAPTURE . ".error.log", "connecting - query " . var_export($params, 1));
} elseif (CAPTURE == "onepercent") {
logit(CAPTURE . ".error.log", "connecting to sample stream");
}
$capturebucket = array();
$tmhOAuth->streaming_request('POST', $method, $params, 'tracker_streamCallback', array('Host' => 'stream.twitter.com'));
// output any response we get back AFTER the Stream has stopped -- or it errors
logit(CAPTURE . ".error.log", "stream stopped - error " . var_export($tmhOAuth, 1));
logit(CAPTURE . ".error.log", "processing buffer before exit");
processtweets($capturebucket);
}
