function process_json_file_timeline($filepath, $dbh)
{
global $tweets_processed, $tweets_failed, $tweets_success, $valid_timeline, $empty_timeline, $invalid_timeline, $populated_timeline, $total_timeline, $all_tweet_ids, $all_users, $bin_name;
$tweetQueue = new TweetQueue();
$total_timeline++;
ini_set('auto_detect_line_endings', true);
$handle = @fopen($filepath, "r");
if ($handle) {
while (($buffer = fgets($handle, 40960)) !== false) {
$tweet = json_decode($buffer, true);
//var_export($tweet); print "\n\n";
$buffer = "";
$t = new Tweet();
$t->fromJSON($tweet);
if (!$t->isInBin($bin_name)) {
$tweetQueue->push($t, $bin_name);
if ($tweetQueue->length() > 100) {
$tweetQueue->insertDB();
}
$all_users[] = $t->from_user_id;
$all_tweet_ids[] = $t->id;
$tweets_processed++;
}
print ".";
}
if (!feof($handle)) {
echo "Error: unexpected fgets() fail\n";
}
fclose($handle);
}
if ($tweetQueue->length() > 0) {
$tweetQueue->insertDB();
}
}
function search($idlist)
{
global $twitter_keys, $current_key, $all_users, $all_tweet_ids, $bin_name, $dbh, $tweetQueue;
$keyinfo = getRESTKey(0);
$current_key = $keyinfo['key'];
$ratefree = $keyinfo['remaining'];
print "current key {$current_key} ratefree {$ratefree}\n";
$tmhOAuth = new tmhOAuth(array('consumer_key' => $twitter_keys[$current_key]['twitter_consumer_key'], 'consumer_secret' => $twitter_keys[$current_key]['twitter_consumer_secret'], 'token' => $twitter_keys[$current_key]['twitter_user_token'], 'secret' => $twitter_keys[$current_key]['twitter_user_secret']));
// by hundred
for ($i = 0; $i < sizeof($idlist); $i += 100) {
if ($ratefree <= 0 || $ratefree % 10 == 0) {
$keyinfo = getRESTKey($current_key);
$current_key = $keyinfo['key'];
$ratefree = $keyinfo['remaining'];
$tmhOAuth = new tmhOAuth(array('consumer_key' => $twitter_keys[$current_key]['twitter_consumer_key'], 'consumer_secret' => $twitter_keys[$current_key]['twitter_consumer_secret'], 'token' => $twitter_keys[$current_key]['twitter_user_token'], 'secret' => $twitter_keys[$current_key]['twitter_user_secret']));
}
$q = $idlist[$i];
$n = $i + 1;
while ($n < $i + 100) {
if (!isset($idlist[$n])) {
break;
}
$q .= "," . $idlist[$n];
$n++;
}
$params = array('id' => $q);
$code = $tmhOAuth->user_request(array('method' => 'GET', 'url' => $tmhOAuth->url('1.1/statuses/lookup'), 'params' => $params));
$ratefree--;
if ($tmhOAuth->response['code'] == 200) {
$data = json_decode($tmhOAuth->response['response'], true);
if (is_array($data) && empty($data)) {
// all tweets in set are deleted
continue;
}
$tweets = $data;
$tweet_ids = array();
foreach ($tweets as $tweet) {
$t = new Tweet();
$t->fromJSON($tweet);
if (!$t->isInBin($bin_name)) {
$all_users[] = $t->from_user_id;
$all_tweet_ids[] = $t->id;
$tweet_ids[] = $t->id;
$tweetQueue->push($t, $bin_name);
}
print ".";
}
sleep(1);
} else {
echo "Failure with code " . $tmhOAuth->response['response']['code'] . "\n";
var_dump($tmhOAuth->response['response']['info']);
var_dump($tmhOAuth->response['response']['error']);
var_dump($tmhOAuth->response['response']['errno']);
die;
}
$tweetQueue->insertDB();
}
}
function search($keywords, $max_id = null)
{
global $twitter_keys, $current_key, $ratefree, $bin_name, $dbh, $tweetQueue;
$ratefree--;
if ($ratefree < 1 || $ratefree % 10 == 0) {
$keyinfo = getRESTKey($current_key, 'search', 'tweets');
$current_key = $keyinfo['key'];
$ratefree = $keyinfo['remaining'];
}
$tmhOAuth = new tmhOAuth(array('consumer_key' => $twitter_keys[$current_key]['twitter_consumer_key'], 'consumer_secret' => $twitter_keys[$current_key]['twitter_consumer_secret'], 'token' => $twitter_keys[$current_key]['twitter_user_token'], 'secret' => $twitter_keys[$current_key]['twitter_user_secret']));
$params = array('q' => $keywords, 'count' => 100);
if (isset($max_id)) {
$params['max_id'] = $max_id;
}
$code = $tmhOAuth->user_request(array('method' => 'GET', 'url' => $tmhOAuth->url('1.1/search/tweets'), 'params' => $params));
if ($tmhOAuth->response['code'] == 200) {
$data = json_decode($tmhOAuth->response['response'], true);
$tweets = $data['statuses'];
$tweet_ids = array();
foreach ($tweets as $tweet) {
$t = new Tweet();
$t->fromJSON($tweet);
$tweet_ids[] = $t->id;
if (!$t->isInBin($bin_name)) {
$tweetQueue->push($t, $bin_name);
if ($tweetQueue->length() > 100) {
$tweetQueue->insertDB();
}
print ".";
}
}
if (!empty($tweet_ids)) {
print "\n";
if (count($tweet_ids) <= 1) {
print "no more tweets found\n\n";
return false;
}
$max_id = min($tweet_ids);
print "max id: " . $max_id . "\n";
} else {
print "0 tweets found\n\n";
return false;
}
sleep(1);
search($keywords, $max_id);
} else {
echo $tmhOAuth->response['response'] . "\n";
if ($tmhOAuth->response['response']['errors']['code'] == 130) {
// over capacity
sleep(1);
search($keywords, $max_id);
}
}
}
function process_json_file_timeline($filepath, $dbh)
{
global $tweets_processed, $tweets_failed, $tweets_success, $valid_timeline, $empty_timeline, $invalid_timeline, $populated_timeline, $total_timeline, $all_tweet_ids, $all_users, $bin_name;
$tweetQueue = new TweetQueue();
$total_timeline++;
$filestr = file_get_contents($filepath);
// sylvester stores multiple json exports in the same file,
// in order to decode it we will need to split it into its respective individual exports
$jsons = explode("}][{", $filestr);
print count($jsons) . " jsons found\n";
foreach ($jsons as $json) {
if (substr($json, 0, 2) != "[{") {
$json = "[{" . $json;
}
if (substr($json, -2) != "}]") {
$json = $json . "}]";
}
$timeline = json_decode($json);
if (is_array($timeline)) {
$valid_timeline++;
if (!empty($timeline)) {
$populated_timeline++;
} else {
$empty_timeline++;
}
} else {
$invalid_timeline++;
}
foreach ($timeline as $tweet) {
$t = new Tweet();
$t->fromJSON($tweet);
if (!$t->isInBin($bin_name)) {
$tweetQueue->push($t, $bin_name);
if ($tweetQueue->length() > 100) {
$tweetQueue->insertDB();
}
$all_users[] = $t->user->id;
$all_tweet_ids[] = $t->id;
$tweets_processed++;
}
}
}
if ($tweetQueue->length() > 0) {
$tweetQueue->insertDB();
}
}
function get_timeline($user_id, $type, $max_id = null)
{
print "doing {$user_id}\n";
global $twitter_keys, $current_key, $ratefree, $looped, $bin_name, $dbh, $tweetQueue;
$ratefree--;
if ($ratefree < 1 || $ratefree % 10 == 0) {
$keyinfo = getRESTKey($current_key, 'statuses', 'user_timeline');
$current_key = $keyinfo['key'];
$ratefree = $keyinfo['remaining'];
}
$tmhOAuth = new tmhOAuth(array('consumer_key' => $twitter_keys[$current_key]['twitter_consumer_key'], 'consumer_secret' => $twitter_keys[$current_key]['twitter_consumer_secret'], 'token' => $twitter_keys[$current_key]['twitter_user_token'], 'secret' => $twitter_keys[$current_key]['twitter_user_secret']));
$params = array('count' => 200, 'trim_user' => false, 'exclude_replies' => false, 'contributor_details' => true, 'include_rts' => 1);
if ($type == "user_id") {
$params['user_id'] = $user_id;
} else {
$params['screen_name'] = $user_id;
}
if (isset($max_id)) {
$params['max_id'] = $max_id;
}
$tmhOAuth->user_request(array('method' => 'GET', 'url' => $tmhOAuth->url('1.1/statuses/user_timeline'), 'params' => $params));
//var_export($params); print "\n";
if ($tmhOAuth->response['code'] == 200) {
$tweets = json_decode($tmhOAuth->response['response'], true);
// store in db
$tweet_ids = array();
foreach ($tweets as $tweet) {
$t = new Tweet();
$t->fromJSON($tweet);
$tweet_ids[] = $t->id;
if (!$t->isInBin($bin_name)) {
$tweetQueue->push($t, $bin_name);
print ".";
if ($tweetQueue->length() > 100) {
$tweetQueue->insertDB();
}
}
}
if (!empty($tweet_ids)) {
print "\n";
if (count($tweet_ids) <= 1) {
print "no more tweets found\n\n";
return false;
}
$max_id = min($tweet_ids);
print "max id: " . $max_id . "\n";
} else {
print "0 tweets found\n\n";
return false;
}
sleep(1);
get_timeline($user_id, $type, $max_id);
} else {
$error_code = json_decode($tmhOAuth->response['response'])->errors[0]->code;
if ($error_code == 130) {
print "Twitter is over capacity, sleeping 5 seconds before retry\n";
sleep(5);
get_timeline($user_id, $type, $max_id);
} elseif ($error_code == 88) {
print "API key rate limit exceeded, sleeping 60 seconds before retry\n";
sleep(60);
get_timeline($user_id, $type, $max_id);
} else {
echo "\nAPI error: " . $tmhOAuth->response['response'] . "\n";
}
}
}
function search($idlist)
{
global $twitter_keys, $current_key, $all_users, $all_tweet_ids, $bin_name, $dbh, $tweetQueue;
$keyinfo = getRESTKey(0);
$current_key = $keyinfo['key'];
$ratefree = $keyinfo['remaining'];
print "\ncurrent key {$current_key} ratefree {$ratefree}\n";
$tmhOAuth = new tmhOAuth(array('consumer_key' => $twitter_keys[$current_key]['twitter_consumer_key'], 'consumer_secret' => $twitter_keys[$current_key]['twitter_consumer_secret'], 'token' => $twitter_keys[$current_key]['twitter_user_token'], 'secret' => $twitter_keys[$current_key]['twitter_user_secret']));
// by hundred
for ($i = 0; $i < sizeof($idlist); $i += 100) {
if ($ratefree <= 0 || $ratefree % 10 == 0) {
print "\n";
$keyinfo = getRESTKey($current_key);
$current_key = $keyinfo['key'];
$ratefree = $keyinfo['remaining'];
$tmhOAuth = new tmhOAuth(array('consumer_key' => $twitter_keys[$current_key]['twitter_consumer_key'], 'consumer_secret' => $twitter_keys[$current_key]['twitter_consumer_secret'], 'token' => $twitter_keys[$current_key]['twitter_user_token'], 'secret' => $twitter_keys[$current_key]['twitter_user_secret']));
}
$q = $idlist[$i];
$n = $i + 1;
while ($n < $i + 100) {
if (!isset($idlist[$n])) {
break;
}
$q .= "," . $idlist[$n];
$n++;
}
$params = array('id' => $q);
$code = $tmhOAuth->user_request(array('method' => 'GET', 'url' => $tmhOAuth->url('1.1/statuses/lookup'), 'params' => $params));
$ratefree--;
$reset_connection = false;
if ($tmhOAuth->response['code'] == 200) {
$data = json_decode($tmhOAuth->response['response'], true);
if (is_array($data) && empty($data)) {
// all tweets in set are deleted
continue;
}
$tweets = $data;
$tweet_ids = array();
foreach ($tweets as $tweet) {
$t = new Tweet();
$t->fromJSON($tweet);
if (!$t->isInBin($bin_name)) {
$all_users[] = $t->from_user_id;
$all_tweet_ids[] = $t->id;
$tweet_ids[] = $t->id;
$tweetQueue->push($t, $bin_name);
}
print ".";
}
sleep(1);
$retries = 0;
// reset retry counter on success
} else {
if ($retries < 4 && $tmhOAuth->response['code'] == 503) {
/* this indicates problems on the Twitter side, such as overcapacity. we slow down and retry the connection */
print "!";
sleep(7);
$i--;
// rewind
$retries++;
$reset_connection = true;
} else {
if ($retries < 4) {
print "\n";
print "Failure with code " . $tmhOAuth->response['response']['code'] . "\n";
var_dump($tmhOAuth->response['response']['info']);
var_dump($tmhOAuth->response['response']['error']);
var_dump($tmhOAuth->response['response']['errno']);
print "The above error may not be permanent. We will sleep and retry the request.\n";
sleep(7);
$i--;
// rewind
$retries++;
$reset_connection = true;
} else {
print "\n";
print "Permanent error when querying the Twitter API. Please investigate the error output. Now stopping.\n";
exit(1);
}
}
}
if ($reset_connection) {
$tmhOAuth = new tmhOAuth(array('consumer_key' => $twitter_keys[$current_key]['twitter_consumer_key'], 'consumer_secret' => $twitter_keys[$current_key]['twitter_consumer_secret'], 'token' => $twitter_keys[$current_key]['twitter_user_token'], 'secret' => $twitter_keys[$current_key]['twitter_user_secret']));
$reset_connection = false;
} else {
$tweetQueue->insertDB();
}
}
}