function generate($what, $filename) { global $tsv, $network, $esc, $titles, $database, $interval, $outputformat; require_once __DIR__ . '/CSV.class.php'; // initialize variables $tweets = $times = $from_user_names = $results = $urls = $urls_expanded = $hosts = $hashtags = array(); $csv = new CSV($filename, $outputformat); $collation = current_collation(); // determine interval $sql = "SELECT MIN(t.created_at) AS min, MAX(t.created_at) AS max FROM " . $esc['mysql']['dataset'] . "_tweets t "; $sql .= sqlSubset(); //print $sql . "<bR>"; $rec = mysql_query($sql); $res = mysql_fetch_assoc($rec); // get frequencies if ($what == "hashtag") { $results = frequencyTable("hashtags", "text"); } elseif ($what == "urls") { $results = frequencyTable("urls", "url_followed"); } elseif ($what == "hosts") { $results = frequencyTable("urls", "domain"); } elseif ($what == "mention") { $results = frequencyTable("mentions", "to_user"); // get other things } else { // @todo, this could also use database grouping $sql = "SELECT id,text COLLATE {$collation} as text,created_at,from_user_name COLLATE {$collation} as from_user_name FROM " . $esc['mysql']['dataset'] . "_tweets t "; $sql .= sqlSubset(); // get slice and its min and max time $rec = mysql_query($sql); if ($rec && mysql_num_rows($rec) > 0) { while ($res = mysql_fetch_assoc($rec)) { $tweets[] = $res['text']; $ids[] = $res['id']; $times[] = $res['created_at']; $from_user_names[] = strtolower($res['from_user_name']); } } // extract desired things ($what) and group per interval foreach ($tweets as $key => $tweet) { $time = $times[$key]; switch ($interval) { case "hourly": $group = strftime("%Y-%m-%d %Hh", strtotime($time)); break; case "weekly": $group = strftime("%Y %u", strtotime($time)); break; case "monthly": $group = strftime("%Y-%m", strtotime($time)); break; case "yearly": $group = strftime("%Y-%m", strtotime($time)); break; case "overall": $group = "overall"; break; case "custom": $group = groupByInterval(strftime("%Y-%m-%d", strtotime($time))); break; default: $group = strftime("%Y-%m-%d", strtotime($time)); // default daily } switch ($what) { //case "hashtag": // foreach ($hashtags as $hashtag) // $results[$group][] = $hashtag; // break; //case "mention": // @todo, mentions might be taken from own table // $stuff = get_replies($tweet); // foreach ($stuff as $thing) // $results[$group][] = $thing; // break; case "user": $results[$group][] = $from_user_names[$key]; break; case "user-mention": $stuff = get_replies($tweet); foreach ($stuff as $thing) { $results[$group]['mentions'][] = $thing; } $results[$group]['users'][] = $from_user_names[$key]; //var_dump($results); break; case "retweet": $results[$group][] = $tweet; // TODO, write stemming function break; //case "urls": // if (isset($urls_expanded[$ids[$key]])) // $results[$group][] = $urls_expanded[$ids[$key]]; // break; //case "hosts": // if (isset($urls_expanded[$ids[$key]])) // $results[$group][] = $hosts[$ids[$key]]; // break; //case "urls": // if (isset($urls_expanded[$ids[$key]])) // $results[$group][] = $urls_expanded[$ids[$key]]; // break; //case "hosts": // if (isset($urls_expanded[$ids[$key]])) // $results[$group][] = $hosts[$ids[$key]]; // break; default: break; } } // count frequency of occurence of thing, per interval if ($what != "user-mention") { foreach ($results as $group => $things) { $counted_things = array_count_values($things); arsort($counted_things); $results[$group] = $counted_things; } } } // network output for users if ($what == "user-mention") { foreach ($results as $group => $things) { $tmp_mentions = array_count_values($things['mentions']); $tmp_users = array_count_values($things['users']); $counted_things = array(); // add all from_user_names foreach ($tmp_users as $user => $count) { if (isset($tmp_mentions["@" . $user])) { $counted_things[$user] = $tmp_mentions["@" . $user] . "," . $count; } else { $counted_things[$user] = "0," . $count; } } // add all users which were replied but not in the set foreach ($tmp_mentions as $user => $count) { $user = str_replace("@", "", $user); if (!isset($counted_things[$user])) { $counted_things[$user] = $count . ",0"; } } ksort($counted_things); $results[$group] = $counted_things; } if (isset($titles[$what])) { if (!empty($esc['shell']['query'])) { $q = " with search " . $esc['shell']['query']; } else { $q = ""; } $csv->writeheader(array($titles[$what] . $q . " from " . $esc['date']["startdate"] . " to " . $esc['date']["enddate"])); } $csv->writeheader(array("date", "user", "mentions", "tweets")); foreach ($results as $group => $things) { foreach ($things as $thing => $count) { $csv->newrow(); $csv->addfield($group); $csv->addfield($thing); $exp = explode(",", $count); // unpack what we packed $csv->addfield($exp[0]); $csv->addfield($exp[1]); $csv->writerow(); } } // write tsv output } elseif (in_array($what, $tsv) !== false) { ksort($results); // construct file if (isset($titles[$what])) { if (!empty($esc['shell']['query'])) { $q = " with search " . $esc['shell']['query']; } else { $q = ""; } $csv->writeheader(array($titles[$what] . " for " . $esc['shell']['datasetname'] . $q . " from " . $esc['date']["startdate"] . " to " . $esc['date']["enddate"])); } if ($what == "urls") { $csv->writeheader(array("date", "frequency", "tweetedurl")); } elseif ($what == "hosts") { $csv->writeheader(array("date", "frequency", "domain", "name")); } else { $csv->writeheader(array("date", "frequency", $what)); } foreach ($results as $group => $things) { arsort($things); foreach ($things as $thing => $count) { if (empty($thing)) { continue; } if ($count < $esc['shell']['minf']) { continue; } $csv->newrow(); $csv->addfield($group); $csv->addfield($count); $csv->addfield($thing); $csv->writerow(); } } } else { die('no valid output format found'); } $csv->close(); }
// @toponder what to compare it with full selection (shows overal likelihood) or compared to previous intervals (shows timely progress but has initialization problem) } // put data in right format for visualization $vis_data = array(); // intialize vis_data with all intervals $sql = "SELECT "; $sql .= sqlInterval(); $sql .= "FROM " . $esc['mysql']['dataset'] . "_tweets t "; $sql .= sqlSubset(); $sql .= "GROUP BY datepart"; //print $sql."<bR>"; $sqlresults = mysql_query($sql); while ($res = mysql_fetch_assoc($sqlresults)) { $date = $res['datepart']; if (!empty($intervalDates)) { $date = groupByInterval($res['datepart']); } $vis_data[$date] = array(); } $excludeFromGraph = array(); if (isset($_REQUEST['excludeFromGraph'])) { $excludeFromGraph = explode(",", $_REQUEST['excludeFromGraph']); foreach ($excludeFromGraph as $k => $v) { $excludeFromGraph[$k] = trim($v); } } foreach ($ap[$keywordToTrack] as $date => $cowords) { foreach ($cowords as $word => $frequency_coword) { if ($frequency_coword_total[$word] >= $minimumCowordFrequencyOverall && $frequency_coword >= $minimumCowordFrequencyInterval) { if (array_search($word, $excludeFromGraph) !== false) { continue;