/**
* Get or Create Solr Service
*
* @return void
*/
protected function _setSolrService()
{
if (null === $this->_solrService) {
$this->_solrService = new Apache_Solr_Service($this->_host, $this->_port, $this->_path);
if (!$this->_solrService->ping()) {
trigger_error("the Apache Solr Service is unavaliable");
exit;
}
}
}
/**
*
* @throws Opus_SolrSearch_Exception If connection to Solr server could not be established.
*/
public function __construct()
{
$this->log = Zend_Registry::get('Zend_Log');
$this->config = Zend_Registry::get('Zend_Config');
$this->solr_server = $this->getSolrServer();
if (false === $this->solr_server->ping()) {
$this->log->err('Connection to Solr server ' . $this->solr_server_url . ' could not be established.');
throw new Opus_SolrSearch_Exception('Solr server ' . $this->solr_server_url . ' is not responding.', Opus_SolrSearch_Exception::SERVER_UNREACHABLE);
}
$this->log->info('Connection to Solr server ' . $this->solr_server_url . ' was successfully established.');
}
public function getSolr()
{
require_once __DIR__ . '/../' . 'Lib/Apache/Solr/Apache_Solr_Service.php';
$em = $this->em;
$solrservers = $em->getRepository('EcoInformaticaAnalizadorBundle:SolrServer')->findAll();
if (count($solrservers) > 0) {
foreach ($solrservers as $solrserver) {
$USR = $solrserver->getUsername();
$PSWD = $solrserver->getPassword();
$HOST = $solrserver->getHost();
$PORT = $solrserver->getPort();
$SOLRPATH = $solrserver->getPath();
$solr = new \Apache_Solr_Service("{$USR}:{$PSWD}@{$HOST}", $PORT, $SOLRPATH);
if ($solr->ping()) {
break;
}
}
} else {
$solr = false;
}
return $solr;
}
/**
* This is a singleton class, thus the constructor should be private/protected
*
* @access protected
*
* @param string $core: The name of the core to use
*
* @return void
*/
protected function __construct($core)
{
// Load class.
if (!class_exists('Apache_Solr_Service')) {
require_once t3lib_div::getFileAbsFileName('EXT:' . self::$extKey . '/lib/SolrPhpClient/Apache/Solr/Service.php');
}
// Get Solr credentials.
$conf = unserialize($GLOBALS['TYPO3_CONF_VARS']['EXT']['extConf'][self::$extKey]);
$host = $conf['solrHost'] ? $conf['solrHost'] : 'localhost';
// Prepend username and password to hostname.
if ($conf['solrUser'] && $conf['solrPass']) {
$host = $conf['solrUser'] . ':' . $conf['solrPass'] . '@' . $host;
}
// Set port if not set.
$port = tx_dlf_helper::intInRange($conf['solrPort'], 1, 65535, 8180);
// Append core name to path.
$path = trim($conf['solrPath'], '/') . '/' . $core;
// Instantiate Apache_Solr_Service class.
$this->service = t3lib_div::makeInstance('Apache_Solr_Service', $host, $port, $path);
// Check if connection is established.
if ($this->service->ping() !== FALSE) {
// Do not collapse single value arrays.
$this->service->setCollapseSingleValueArrays = FALSE;
// Set core name.
$this->core = $core;
// Instantiation successful!
$this->ready = TRUE;
}
}
*/
function printMessage($message)
{
echo strftime('%d.%m.%Y %H:%M:%S', time()) . $message . "\n";
}
define('APPLICATION_ENV', 'production');
// basic bootstrapping
require_once dirname(__FILE__) . '/../common/bootstrap.php';
$config = Zend_Registry::get('Zend_Config');
$host = $config->searchengine->solr->host;
$port = $config->searchengine->solr->port;
$baseUri = $config->searchengine->solr->path;
$EOL = "\n";
$commitRange = 100;
$solr = new Apache_Solr_Service($host, $port, $baseUri);
if (false === $solr->ping()) {
echo 'Could not connect to solr service.' . $EOL;
return;
}
$startTime = time();
$docIds = Opus_Document::getAllIds();
$documents = array();
$conf = Zend_Registry::get('Zend_Config');
$baseFilePath = null;
if (true === isset($conf->file->destinationPath) and true === is_dir($conf->file->destinationPath)) {
$baseFilePath = $conf->file->destinationPath;
}
foreach ($docIds as $docId) {
printMessage(' Indexing document : ' . $docId);
$opusDoc = new Opus_Document($docId);
$solrDocument = Qucosa_Search_Solr_Document_OpusDocument::loadOpusDocument($opusDoc);
public function ping($timeout = self::DEFAULT_PING_TIMEOUT)
{
set_error_handler(array(get_class($this), 'ping_error'), E_ALL);
$ping = parent::ping($timeout);
restore_error_handler();
if ($this->_error || !$ping) {
$this->_error = false;
$this->_lastPing = NULL;
return false;
} else {
$this->_lastPing = $ping;
return true;
}
}
function db_store_article($article, $PageID = 0, $updating = false)
{
global $db;
global $config;
$update = false;
$id = 0;
// If we are editing an existing reference then we already know its id
if (isset($article->reference_id)) {
$id = $article->reference_id;
} else {
$id = db_find_article($article);
}
if ($id != 0) {
if ($updating) {
$update = true;
} else {
return $id;
}
}
// Try and trap empty references
if ($id == 0) {
$ok = false;
if (isset($article->title)) {
$ok = $article->title != '';
}
if (!$ok) {
return 0;
}
}
if (!isset($article->genre)) {
$article->genre = 'article';
}
$keys = array();
$values = array();
// Article metadata
foreach ($article as $k => $v) {
switch ($k) {
// Ignore as it's an array
case 'authors':
break;
case 'date':
$keys[] = 'date';
$values[] = $db->qstr($v);
if (!isset($article->year)) {
$keys[] = 'year';
$values[] = $db->qstr(year_from_date($v));
}
break;
// Don't store BHL URL here
// Don't store BHL URL here
case 'url':
if (preg_match('/^http:\\/\\/(www\\.)?biodiversitylibrary.org\\/page\\/(?<pageid>[0-9]+)/', $v)) {
} else {
// extract Handle if it exists
if (preg_match('/^http:\\/\\/hdl.handle.net\\/(?<hdl>.*)$/', $v, $m)) {
$keys[] = 'hdl';
$values[] = $db->qstr($m['hdl']);
} else {
$keys[] = $k;
$values[] = $db->qstr($v);
}
}
break;
// Things we store as is
// Things we store as is
case 'title':
case 'secondary_title':
case 'volume':
case 'series':
case 'issue':
case 'spage':
case 'epage':
case 'year':
case 'date':
case 'issn':
case 'genre':
case 'doi':
case 'hdl':
case 'lsid':
case 'oclc':
case 'pdf':
case 'abstract':
case 'pmid':
$keys[] = $k;
$values[] = $db->qstr($v);
break;
// Things we ignore
// Things we ignore
default:
break;
}
}
// Date
if (!isset($article->date) && isset($article->year)) {
$keys[] = 'date';
$values[] = $db->qstr($article->year . '-00-00');
}
// BHL PageID
if ($PageID != 0) {
$keys[] = 'PageID';
$values[] = $PageID;
}
// SICI
$s = new Sici();
$sici = $s->create($article);
if ($sici != '') {
$keys[] = 'sici';
$values[] = $db->qstr($sici);
}
if ($update) {
// Versioning?
// Delete links (author, pages, etc)
// Don't delete page range as we may loose plates, etc. outside range
/*
$sql = 'DELETE FROM rdmp_reference_page_joiner WHERE reference_id=' . $id;
$result = $db->Execute($sql);
if ($result == false) die("failed [" . __FILE__ . ":" . __LINE__ . "]: " . $sql);
*/
$sql = 'DELETE FROM rdmp_author_reference_joiner WHERE reference_id = ' . $id;
$result = $db->Execute($sql);
if ($result == false) {
die("failed [" . __FILE__ . ":" . __LINE__ . "]: " . $sql);
}
// update (updated timestamp will be automatically updated)
$sql = 'UPDATE rdmp_reference SET ';
$num_values = count($keys);
for ($i = 0; $i < $num_values; $i++) {
if ($i > 0) {
$sql .= ', ';
}
$sql .= $keys[$i] . '=' . $values[$i];
}
$sql .= ' WHERE reference_id=' . $id;
/* $cache_file = @fopen('/tmp/update.sql', "w+") or die("could't open file");
@fwrite($cache_file, $sql);
fclose($cache_file);
*/
$result = $db->Execute($sql);
if ($result == false) {
die("failed [" . __FILE__ . ":" . __LINE__ . "]: " . $sql);
}
} else {
// Adding article for first time so add 'created' and 'updated' timestamp
$keys[] = 'created';
$values[] = 'NOW()';
$keys[] = 'updated';
$values[] = 'NOW()';
$sql = 'INSERT INTO rdmp_reference (' . implode(",", $keys) . ') VALUES (' . implode(",", $values) . ')';
$result = $db->Execute($sql);
if ($result == false) {
die("failed [" . __FILE__ . ":" . __LINE__ . "]: " . $sql);
}
$id = $db->Insert_ID();
// Store reference_cluster_id which we can use to group duplicates, by default
// reference_cluster_id = reference_id
$sql = 'UPDATE rdmp_reference SET reference_cluster_id=' . $id . ' WHERE reference_id=' . $id;
$result = $db->Execute($sql);
if ($result == false) {
die("failed [" . __FILE__ . ":" . __LINE__ . "]: " . $sql);
}
}
// Indexing-------------------------------------------------------------------------------------
if (1) {
// solr
// this code is redundant with code in reference.php but I use different objects
// here and there (doh!). Also once we've added old stuff to solr this is the only place we
// should be calling solr
$solr = new Apache_Solr_Service('localhost', '8983', '/solr');
if (!$solr->ping()) {
echo 'Solr service not responding.';
exit;
}
$item = array();
$item['id'] = 'reference/' . $id;
$item['title'] = $article->title;
$item['publication_outlet'] = $article->secondary_title;
$item['year'] = $article->year;
$authors = array();
foreach ($article->authors as $a) {
$authors[] = $a->forename . ' ' . $a->surname;
}
$item['authors'] = $authors;
$citation = '';
$citation .= ' ' . $article->year;
$citation .= ' ' . $article->title;
$citation .= ' ' . $article->secondary_title;
$citation .= ' ' . $article->volume;
if (isset($article->issue)) {
$citation .= '(' . $article->issue . ')';
}
$citation .= ':';
$citation .= ' ';
$citation .= $article->spage;
if (isset($article->epage)) {
$citation .= '-' . $article->epage;
}
$item['citation'] = $citation;
$text = '';
$num_authors = count($article->authors);
$count = 0;
if ($num_authors > 0) {
foreach ($article->authors as $author) {
$text .= $author->forename . ' ' . $author->lastname;
if (isset($author->suffix)) {
$text .= ' ' . $author->suffix;
}
$count++;
if ($count == 2 && $num_authors > 3) {
$text .= ' et al.';
break;
}
if ($count < $num_authors - 1) {
$text .= ', ';
} else {
if ($count < $num_authors) {
$text .= ' and ';
}
}
}
}
$item['citation'] = $text . ' ' . $citation;
$parts = array();
$parts[] = $item;
//print_r($parts);
// add to solr
$documents = array();
foreach ($parts as $item => $fields) {
$part = new Apache_Solr_Document();
foreach ($fields as $key => $value) {
if (is_array($value)) {
foreach ($value as $datum) {
$part->setMultiValue($key, $datum);
}
} else {
$part->{$key} = $value;
}
}
$documents[] = $part;
}
//
//
// Load the documents into the index
//
try {
$solr->addDocuments($documents);
$solr->commit();
$solr->optimize();
} catch (Exception $e) {
echo $e->getMessage();
}
} else {
$sql = 'DELETE FROM rdmp_text_index WHERE (object_uri=' . $db->qstr($config['web_root'] . 'reference/' . $id) . ')';
$result = $db->Execute($sql);
if ($result == false) {
die("failed [" . __FILE__ . ":" . __LINE__ . "]: " . $sql);
}
// Only do this if we have a title, as sometimes we don't (e.g. CrossRef lacks metadata)
if (isset($article->title)) {
$sql = 'INSERT INTO rdmp_text_index(object_type, object_id, object_uri, object_text)
VALUES ("title"' . ', ' . $id . ', ' . $db->qstr($config['web_root'] . 'reference/' . $id) . ', ' . $db->qstr($article->title) . ')';
$result = $db->Execute($sql);
if ($result == false) {
die("failed [" . __FILE__ . ":" . __LINE__ . "]: " . $sql);
}
}
}
// Versioning-----------------------------------------------------------------------------------
// Store this object in version table so we can recover it if we overwrite item
$ip = getip();
$sql = 'INSERT INTO rdmp_reference_version(reference_id, ip, json) VALUES(' . $id . ', ' . 'INET_ATON(\'' . $ip . '\')' . ',' . $db->qstr(json_encode($article)) . ')';
$result = $db->Execute($sql);
if ($result == false) {
die("failed [" . __FILE__ . ":" . __LINE__ . "]: " . $sql);
}
// Author(s)------------------------------------------------------------------------------------
// Store author as and link to the article
if (isset($article->authors)) {
db_store_authors($id, $article->authors);
}
// Store page range (only if not updating, otherwise we may loose plates, etc.
// that aren't in page range)
if ($PageID != 0 && !$update) {
$page_range = array();
if (isset($article->spage) && isset($article->epage)) {
$page_range = bhl_page_range($PageID, $article->epage - $article->spage + 1);
} else {
// No epage, so just get spage (to do: how do we tell user we don't have page range?)
$page_range = bhl_page_range($PageID, 0);
}
//print_r($page_range);
$count = 0;
foreach ($page_range as $page) {
$sql = 'INSERT INTO rdmp_reference_page_joiner (reference_id, PageID, page_order)
VALUES (' . $id . ',' . $page . ',' . $count++ . ')';
$result = $db->Execute($sql);
if ($result == false) {
die("failed [" . __FILE__ . ":" . __LINE__ . "]: " . $sql);
}
}
}
// Tweet----------------------------------------------------------------------------------------
if (!$update) {
if ($config['twitter']) {
$url = $config['web_root'] . 'reference/' . $id . ' ' . '#bhlib';
// url + hashtag
$url_len = strlen($url);
$status = '';
if (isset($article->title)) {
$status = $article->title;
$status_len = strlen($status);
$extra = 140 - $status_len - $url_len - 1;
if ($extra < 0) {
$status_len += $extra;
$status_len -= 1;
$status = substr($status, 0, $status_len);
$status .= '…';
}
}
$status .= ' ' . $url;
tweet($status);
}
}
return $id;
}
public function isActive()
{
require_once __DIR__ . '/../' . 'Lib/Apache/Solr/Apache_Solr_Service.php';
$solr = new \Apache_Solr_Service("{$this->username}:{$this->password}@{$this->host}", $this->port, $this->path);
if ($solr->ping()) {
return true;
} else {
return false;
}
}
<?php
require_once dirname(dirname(__FILE__)) . '/Apache/Solr/Service.php';
require_once dirname(dirname(__FILE__)) . '/db.php';
require_once dirname(dirname(__FILE__)) . '/reference.php';
//
//
// Try to connect to the named server, port, and url
//
$solr = new Apache_Solr_Service('localhost', '8983', '/solr');
if (!$solr->ping()) {
echo 'Solr service not responding.';
exit;
}
$sql = "SELECT COUNT(reference_id) as c FROM rdmp_reference WHERE (PageID <> 0)";
$result = $db->Execute($sql);
if ($result == false) {
die("failed [" . __FILE__ . ":" . __LINE__ . "]: " . $sql);
}
$num = $result->fields['c'];
$page_size = 100;
$pages = $num / $page_size;
for ($page = 0; $page < $pages; $page++) {
$parts = array();
$sql = "SELECT reference_id FROM rdmp_reference WHERE (PageID <> 0) LIMIT " . $page * $page_size . "," . $page_size;
echo $sql . "\n";
$result = $db->Execute($sql);
if ($result == false) {
die("failed [" . __FILE__ . ":" . __LINE__ . "]: " . $sql);
}
$ids = array();
function deleteAction()
{
$this->_helper->viewRenderer->setNoRender(TRUE);
$title = "<h4>HUKUMONLINE INDONESIA: <small>search</small></h4><hr/>";
echo $title . '<br>';
$solr = new Apache_Solr_Service('nihki:sirkulasi@202.153.129.35', '8983', '/solr/core-catalog');
if (!$solr->ping()) {
echo 'Solr service not responding.';
exit;
}
$a = array('fl17956', 'fl1131');
foreach ($a as $c) {
$indexingEngine = Pandamp_Search::manager();
$indexingEngine->deleteCatalogFromIndex($c);
}
}
/**
* Iterate through available write services and select the first with a ping
* that satisfies configured timeout restrictions (or the default)
*
* @return Apache_Solr_Service
*
* @throws Exception If there are no write services that meet requirements
*/
private function _selectWriteService()
{
if (!$this->_currentWriteService || !isset($this->_writeableServices[$this->_currentWriteService])) {
foreach ($this->_writeableServices as $id => $service) {
if (is_array($service)) {
//convert the array definition to a client object
$service = new Apache_Solr_Service($service['host'], $service['port'], $service['path']);
$this->_writeableServices[$id] = $service;
}
//check the service
if ($service->ping($this->_writePingTimeout) !== false) {
$this->_currentWriteService = $id;
return $this->_writeableServices[$this->_currentWriteService];
}
}
throw new Exception('No write services were available');
}
return $this->_writeableServices[$this->_currentWriteService];
}
/**
* Test if Solr server is reachable within a given period.
* @param int $iTimeout Seconds to wait for reply from server.
* @return int Time it took to answer in microseconds or false if no answer.
*/
public function ping($iTimeout = 2)
{
wfProfileIn('BS::' . __METHOD__);
if ($this->bUseDifferentProtocolThanHttp) {
$ctx = stream_context_create(array('https' => array('timeout' => $iTimeout)));
$start = microtime(true);
$res = @file_get_contents($this->sUrl, 0, $ctx);
wfProfileOut('BS::' . __METHOD__);
return $res ? microtime(true) - $start : false;
}
wfProfileOut('BS::' . __METHOD__);
return parent::ping($iTimeout);
}
/**
* This is a singleton class, thus the constructor should be private/protected
*
* @access protected
*
* @param string $core: The name of the core to use
*
* @return void
*/
protected function __construct($core)
{
// Load class.
if (!class_exists('Apache_Solr_Service')) {
require_once \TYPO3\CMS\Core\Utility\GeneralUtility::getFileAbsFileName('EXT:' . self::$extKey . '/lib/SolrPhpClient/Apache/Solr/Service.php');
}
$solrInfo = self::getSolrConnectionInfo($core);
// Instantiate Apache_Solr_Service class.
$this->service = \TYPO3\CMS\Core\Utility\GeneralUtility::makeInstance('Apache_Solr_Service', $solrInfo['host'], $solrInfo['port'], $solrInfo['path']);
// Check if connection is established.
if ($this->service->ping() !== FALSE) {
// Do not collapse single value arrays.
$this->service->setCollapseSingleValueArrays = FALSE;
// Set core name.
$this->core = $core;
// Instantiation successful!
$this->ready = TRUE;
}
}
