function __construct($pageID, $language)
 {
     $this->uri = RDFTriple::page($pageID);
     $this->language = $language;
     if (Options::isOptionSet('predicateFilter')) {
         $p = Options::getOption('predicateFilter');
         foreach ($p as $one) {
             //echo $one."\n";
             $this->predicateFilterList[] = trim($one);
         }
     }
     if (Options::isOptionSet('objectFilter')) {
         $o = Options::getOption('objectFilter');
         foreach ($o as $one) {
             //echo $one."\n";
             $this->objectFilterList[] = trim($one);
         }
     }
     if (Options::isOptionSet('predicateObjectFilter')) {
         $po = Options::getOption('predicateObjectFilter');
         foreach ($po as $one) {
             //echo $one."\n";
             $pos = strpos($one, " ");
             $currentp = trim(substr($one, 0, $pos));
             $currento = trim(substr($one, $pos + 1));
             //echo $currentp."\n";
             //echo $currento."\n";
             $this->predicateFilterList[] = $one;
         }
     }
 }
 function __construct($metainfo)
 {
     $this->uri = RDFtriple::page($metainfo['pageTitle']);
     $this->language = $metainfo['language'];
     $this->oaiId = $metainfo['oaiId'];
     $this->graphURI = Options::getOption('graphURI');
     $this->annotationGraphURI = Options::getOption('annotationGraphURI');
     $this->generateOWLAxiomAnnotations = Options::getOption('generateOWLAxiomAnnotations');
     $this->languageProperties = Options::getOption('stringPredicateWithForeignlanguages');
     $this->debug_turn_off_insert = Options::getOption('debug_turn_off_insert');
     $this->debug_run_tests = Options::getOption('debug_run_tests');
     $this->hash = new Hash($this->oaiId, $this->uri->getURI());
     //$this->metainfo = $metainfo;
     $this->subjectSPARULpattern = $this->uri->toSPARULPattern();
     if (Options::isOptionSet('predicateFilter')) {
         $p = Options::getOption('predicateFilter');
         $this->log(WARN, 'currently not working');
     }
     if (Options::isOptionSet('objectFilter')) {
         $o = Options::getOption('objectFilter');
         $this->log(WARN, 'currently not working');
     }
     if (Options::isOptionSet('predicateObjectFilter')) {
         $po = Options::getOption('predicateObjectFilter');
         $this->log(WARN, 'currently not working');
     }
 }
Exemple #3
0
function printAll($lastarticles, $language)
{
    $statisticdir = Options::getOption('statisticdir');
    Timer::start('main::glob');
    if (Options::getOption('noglob')) {
        Statistics::setArticleQueue('deactivated for speed');
    } else {
        Statistics::setArticleQueue(count(glob(Options::getOption('oaiRecords') . "/*.*")));
    }
    Timer::stop('main::glob');
    /*
     *      //too slow
            Timer::start('main::lswc');
            Statistics::setArticleQueue(exec ('ls -1 '.Options::getOption('oaiRecords').' | wc -l'));
            Timer::stop('main::lswc');
    */
    $data = array();
    $data['lastarticles'] = $lastarticles;
    Timer::stop("main::Runtime");
    $timeString = Timer::getTimeAsString();
    $data['timeString'] = $timeString;
    $data['time'] = Timer::$time;
    Timer::start("main::Runtime");
    $overall = array();
    $overall['startingtime'] = Timer::$startingTime;
    $overall['lasttime'] = microtime(true);
    $data['timeOverall'] = $overall;
    Timer::stop("main::processing");
    $data['processingTime'] = Timer::$time["main::processing"]['total'];
    Timer::start("main::processing");
    //toFile($time, 'timeString.txt', $statisticdir);
    //do statistics
    //Timer::timeToFile($statisticdir);
    //      Statistics::statisticsToFile($statisticdir);
    $data['triples'] = Statistics::$countArr;
    $data['extractorMeta'] = Statistics::$extractorMeta;
    $s = Timer::getElapsedSeconds();
    $t = Statistics::getTotalTriples();
    $memory = "memory_get_usage  (true ): " . memory_get_usage(true) . "\n";
    $memory .= "memory_get_usage  (false ): " . memory_get_usage(false) . "\n";
    $memory .= "memory_get_peak_usage  (true ): " . memory_get_peak_usage(true) . "\n";
    $memory .= "memory_get_peak_usage  (false ): " . memory_get_peak_usage(false) . "\n";
    $data['memory'] = $memory;
    //toFile($memory, 'memory.txt', $statisticdir);
    //echo $s."\n";
    $general = "Seconds per article: " . $s / Statistics::getTotalArticles() . "\n";
    $general .= "Articles per second: " . Statistics::getTotalArticles() / $s . "\n";
    $general .= "Articles per hour: " . Statistics::getTotalArticles() / $s * 3600 . "\n";
    $general .= "Category per second: " . Statistics::getTotalCategories() / $s . "\n";
    $general .= "Category per hour: " . Statistics::getTotalCategories() / $s * 3600 . "\n";
    $general .= "Redirect per second: " . Statistics::getTotalRedirects() / $s . "\n";
    $general .= "Redirect per hour: " . Statistics::getTotalRedirects() / $s * 3600 . "\n";
    $general .= "Triples per second: " . $t / $s . "\n";
    $general .= "Triples per minute: " . $t / $s * 60 . "\n";
    $general .= "Triples per hour: " . $t / $s * 3600 . "\n";
    $general .= "Triples per day: " . $t / $s * 3600 * 24 . "\n";
    $data['general'] = $general;
    $append = '';
    if (Options::isOptionSet('processID')) {
        $append = Options::$config['processID'];
    }
    $indexfile = 'index' . $append . '.html';
    toFile(Statistics::generateStatisticHTML(Options::getOption('linkeddataresourceprefix'), $language, $data), $indexfile, $statisticdir);
    Logger::info('wrote html file to ' . $statisticdir . '/' . $indexfile);
}