Warning: chmod(): Operation not permitted in /var/www/hopeinstoughton_www/wp-content/plugins/simple-universal-google-analytics/main.php on line 8
[0] in function chmod in /var/www/hopeinstoughton_www/wp-content/plugins/simple-universal-google-analytics/main.php on line 8
[1] in function include_once in /var/www/hopeinstoughton_www/wp-settings.php on line 560
[2] in function require_once in /var/www/hopeinstoughton_www/wp-config.php on line 85
[3] in function require_once in /var/www/hopeinstoughton_www/wp-load.php on line 50
[4] in function require_once in /var/www/hopeinstoughton_www/wp-blog-header.php on line 13
[5] in function require in /var/www/hopeinstoughton_www/index.php on line 17

<?php

require_once __DIR__ . '/xapian.php';
//require_once __DIR__ . '/elasticsearch.php';
require_once realpath(__DIR__ . '/../../media-manager/cdn/include/vtt_class.php');

// uses remove_accents($string) from wordpress

class class_xap {
    
    private $doctype;
    private $docid;
    private $content;
    private $dbfilename;
    private $db;
    private $termgenerator = NULL;
    
    public function DocumentType() {
        if (func_num_args()) {
            $this->doctype = func_get_arg(0);
            return $this;
        }
        return $this->doctype;
    }
    public function DocumentId() {
        if (func_num_args()) {
            $this->docid = func_get_arg(0);
            return $this;
        }
        return $this->docid;
    }
    public function FilenameDocument() {
        if (func_num_args()) {
            $filename = func_get_arg(0);
            $this->content = file_get_contents($filename);
            return $this;
        }
        return NULL;
    }
    public function Content() {
        if (func_num_args()) {
            $this->content = func_get_arg(0);
            return $this;
        }
        return $this->content;
    }
    
    public function FilenameDatabase() {
        if (func_num_args()) {
            $this->dbfilename = func_get_arg(0);
            return $this;
        }
        return $this->dbfilename;
    }

    // only one process can open it read/write
    public function Open($readonly = TRUE) {
		if (SEARCH == 'elastic') {
			$this->db = new \MyElastic;
        } elseif ($readonly) {
            $this->db = new XapianDatabase($this->dbfilename);
        } else {
            $this->db = new XapianWritableDatabase($this->dbfilename, Xapian::DB_CREATE_OR_OPEN);
        }
    }

    public function Close() {
        if (isset($this->db)) {
            $this->termgenerator = NULL;
        }
        $this->db = NULL;
        return $this;
    }

	private function _SearchXapian($queryString) {
        $enquire = new XapianEnquire($this->db);
        $qp = new XapianQueryParser();
        $qp->set_stemmer(new XapianStem('en'));
        $qp->set_database($this->db);
        $qp->set_stemming_strategy(XapianQueryParser::STEM_SOME);
        $query = $qp->parse_query($queryString, XapianQueryParser::FLAG_DEFAULT | XapianQueryParser::FLAG_PARTIAL);
        $enquire->set_query($query);
		$matches = $enquire->get_mset(0, 300);
        $res = [];
        $c = $matches->size();
        for($i = 0; $i < $c; $i++) {
            $doc = $matches->get_document($i);
//            $docid = $matches->get_docid($i);
            $pct = $matches->get_document_percentage($i);
            $data = json_decode($doc->get_data(), TRUE);
            $snipped = $matches->snippet(join(' ', $data['body']), 220); // , NULL, XapianMSet::SNIPPET_BACKGROUND_MODEL or XapianMSet::SNIPPET_EXHAUSTIVE

            $node = [
                'doctype'  => $data['doctype'],
                'docid'    => $data['docid'],
//              'id'       => $docid,       // don't need to know the xapian document number
                'snip'     => $snipped,
                'pct'      => $pct,
				'offset'   => $this->_determineSnippetOffset($data, $snipped),
            ];
            $res[] = $node;
        }
        return $res;
	}
    
    public function Search($queryString) {
		if (SEARCH == 'elastic') {
			return $this->db->DocumentSearch($queryString);
		} else {
			$queryString = remove_accents($queryString);
			return $this->_SearchXapian($queryString);
		}
    }
    
    private function _determineSnippetOffset($data, $snipped) {
        $snipped = str_replace(['...', '<b>', '</b>'], '', $snipped);
        $source = join(' ', $data['body']);
        if ($snipped == '') return '0:00';
        $ps = strpos($source, $snipped);
        if ($ps === FALSE) return '0:00';
        
        $offset = 0;
        foreach($data['body'] as $index => $line) {
            $l = strlen($line) + 1;
            if ($ps >= $offset && $ps < $offset + $l) {
                return $data['timing'][$index];
            }
            $offset += $l;
        }
        return '0:00';
    }

    // load vtt, strip out times and sequence numbers
    public function Reindex() {
		
		$vtt = new vtt_class();
		$content = $this->Content();
		$vtt->Content($content);
		$vttdata = $vtt->getEntries();
		if (!count($vttdata)) return;	       		// no transcript available

		[$output, $times] = $vtt->AsReindex();

		if (SEARCH == 'elastic') {

			$this->db->DocumentStore($this->DocumentId(), explode("\t", $output), $times); 

		} else {
			if (is_null($this->termgenerator)) {
				$this->termgenerator = new XapianTermGenerator();
				$this->termgenerator->set_stemmer(new XapianStem('en'));
			}

			// Add to the database
			$doc = new XapianDocument();
			$this->termgenerator->set_document($doc);
			
			$safe_text = remove_accents($output);
			$this->termgenerator->index_text($safe_text);
			$doc->add_boolean_term('T' . $this->DocumentType());
			$doc->add_boolean_term('I' . $this->DocumentId());

			// store all the fields for display purposes
			$payload = [
				'doctype'  => $this->DocumentType(),
				'docid'    => $this->DocumentId(),
				'modified' => 0,
				'timing'   => $times,
				'body'     => explode("\t", $output),
			];
			$doc->set_data(json_encode($payload));

			// we use the identifier to ensure each object ends up
			// in the database only once no matter how many times
			// we run the indexer
			$idterm = 'Q' . $this->DocumentId();
			$doc->add_term($idterm);

			$this->db->replace_document($idterm, $doc);
		}
        return $this;
    }
    
    public function Remove() {
        $idterm = 'Q' . $this->DocumentId();
        $this->db->delete_document($idterm);
        return $this;
    }
    
}