/**
* Sphinx server connector initialization
*
* @return SphinxClient
*/
function init_sphinx()
{
$this->sphinx = new SphinxClient();
$this->sphinx->SetServer($this->admin_options['sphinx_host'], intval($this->admin_options['sphinx_port']));
$this->sphinx->SetMatchMode(SPH_MATCH_EXTENDED2);
return $this->sphinx;
}
/**
* Initialize the Sphinx search engine
* @return void
*/
public function initialize()
{
$this->_connectionOptions = array('host_name' => $this->modx->getOption('discuss.sphinx.host_name', null, 'localhost'), 'port' => $this->modx->getOption('discuss.sphinx.port', null, 9312), 'connection_timeout' => $this->modx->getOption('discuss.sphinx.connection_timeout', null, 30), 'searchd_retries' => $this->modx->getOption('discuss.sphinx.searchd_retries', null, 3), 'searchd_retry_delay' => $this->modx->getOption('discuss.sphinx.searchd_retry_delay', null, 10000));
$this->_indices = $this->modx->getOption('discuss.sphinx.indexes', null, 'discuss_posts');
$this->client = new SphinxClient();
$this->client->SetServer($this->_connectionOptions['host_name'], $this->_connectionOptions['port']);
$this->client->SetConnectTimeout($this->_connectionOptions['connection_timeout']);
$this->client->SetMatchMode(SPH_MATCH_EXTENDED);
return true;
}
/**
* 初始化搜索引擎
*/
private function _initSphinx()
{
$this->_loadCore('Help_SphinxClient');
$this->_sphinx = new SphinxClient();
$this->_sphinx->SetServer(SPHINX_HOST, SPHINX_PORT);
$this->_sphinx->SetConnectTimeout(5);
//连接时间
$this->_sphinx->SetArrayResult(true);
$this->_sphinx->SetMaxQueryTime(10);
//设置最大超时时间
$this->_sphinx->SetMatchMode(SPH_MATCH_ANY);
//匹配模式
}
/**
* 初始化搜索引擎
*/
private function _initSphinx()
{
import('@.Util.SphinxClient');
$this->_sphinx = new SphinxClient();
$this->_sphinx->SetServer(C('SPHINX_HOST'), C('SPHINX_PORT'));
$this->_sphinx->SetConnectTimeout(5);
//连接时间
$this->_sphinx->SetArrayResult(true);
//设置数组返回
$this->_sphinx->SetMaxQueryTime(10);
//设置最大超时时间
$this->_sphinx->SetMatchMode(SPH_MATCH_ANY);
//匹配模式
}
private function get_sugg_trigrams($word, SearchEngineOptions $options)
{
$trigrams = $this->BuildTrigrams($word);
$query = "\"{$trigrams}\"/1";
$len = strlen($word);
$this->resetSphinx();
$this->suggestionClient->SetMatchMode(SPH_MATCH_EXTENDED2);
$this->suggestionClient->SetRankingMode(SPH_RANK_WORDCOUNT);
$this->suggestionClient->SetFilterRange("len", $len - 2, $len + 4);
$this->suggestionClient->SetSortMode(SPH_SORT_EXTENDED, "@weight DESC");
$this->suggestionClient->SetLimits(0, 10);
$indexes = [];
foreach ($options->getDataboxes() as $databox) {
$indexes[] = 'suggest' . $this->CRCdatabox($databox);
}
$index = implode(',', $indexes);
$res = $this->suggestionClient->Query($query, $index);
if ($this->suggestionClient->Status() === false) {
return [];
}
if (!$res || !isset($res["matches"])) {
return [];
}
$words = [];
foreach ($res["matches"] as $match) {
$words[] = $match['attrs']['keyword'];
}
return $words;
}
/**
*
* @param string $query
* @return array of integers - taskIds
*/
public static function searchTasks($query)
{
$fieldWeights = array('description' => 10, 'note' => 6);
$indexName = 'plancake_tasks';
$client = new SphinxClient();
// $client->SetServer (sfConfig::get('app_sphinx_host'), sfConfig::get('app_sphinx_port'));
$client->SetFilter("author_id", array(PcUserPeer::getLoggedInUser()->getId()));
$client->SetConnectTimeout(1);
$client->SetMatchMode(SPH_MATCH_ANY);
$client->SetSortMode(SPH_SORT_RELEVANCE);
$client->SetRankingMode(SPH_RANK_PROXIMITY_BM25);
$client->SetArrayResult(true);
$client->SetFieldWeights($fieldWeights);
$client->setLimits(0, 100);
$results = $client->query($client->EscapeString($query), $indexName);
if ($results === false) {
$error = "Sphinx Error - " . $client->GetLastError();
sfErrorNotifier::alert($error);
}
$ids = array();
if (isset($results['matches']) && count($results['matches'])) {
foreach ($results['matches'] as $match) {
$ids[] = $match['id'];
}
}
return PcTaskPeer::retrieveByPKs($ids);
}
function hook_search($search)
{
$offset = 0;
$limit = 500;
$sphinxClient = new SphinxClient();
$sphinxpair = explode(":", SPHINX_SERVER, 2);
$sphinxClient->SetServer($sphinxpair[0], (int) $sphinxpair[1]);
$sphinxClient->SetConnectTimeout(1);
$sphinxClient->SetFieldWeights(array('title' => 70, 'content' => 30, 'feed_title' => 20));
$sphinxClient->SetMatchMode(SPH_MATCH_EXTENDED2);
$sphinxClient->SetRankingMode(SPH_RANK_PROXIMITY_BM25);
$sphinxClient->SetLimits($offset, $limit, 1000);
$sphinxClient->SetArrayResult(false);
$sphinxClient->SetFilter('owner_uid', array($_SESSION['uid']));
$result = $sphinxClient->Query($search, SPHINX_INDEX);
$ids = array();
if (is_array($result['matches'])) {
foreach (array_keys($result['matches']) as $int_id) {
$ref_id = $result['matches'][$int_id]['attrs']['ref_id'];
array_push($ids, $ref_id);
}
}
$ids = join(",", $ids);
if ($ids) {
return array("ref_id IN ({$ids})", array());
} else {
return array("ref_id = -1", array());
}
}
/**
* Performs a search on an author's posts without caring about message contents. Depends on display specific params
*
* @param string $type contains either posts or topics depending on what should be searched for
* @param boolean $firstpost_only if true, only topic starting posts will be considered
* @param array $sort_by_sql contains SQL code for the ORDER BY part of a query
* @param string $sort_key is the key of $sort_by_sql for the selected sorting
* @param string $sort_dir is either a or d representing ASC and DESC
* @param string $sort_days specifies the maximum amount of days a post may be old
* @param array $ex_fid_ary specifies an array of forum ids which should not be searched
* @param string $post_visibility specifies which types of posts the user can view in which forums
* @param int $topic_id is set to 0 or a topic id, if it is not 0 then only posts in this topic should be searched
* @param array $author_ary an array of author ids
* @param string $author_name specifies the author match, when ANONYMOUS is also a search-match
* @param array &$id_ary passed by reference, to be filled with ids for the page specified by $start and $per_page, should be ordered
* @param int $start indicates the first index of the page
* @param int $per_page number of ids each page is supposed to contain
* @return boolean|int total number of results
*/
public function author_search($type, $firstpost_only, $sort_by_sql, $sort_key, $sort_dir, $sort_days, $ex_fid_ary, $post_visibility, $topic_id, $author_ary, $author_name, &$id_ary, $start, $per_page)
{
$this->search_query = '';
$this->sphinx->SetMatchMode(SPH_MATCH_FULLSCAN);
$fields = $firstpost_only ? 'firstpost' : 'all';
$terms = 'all';
return $this->keyword_search($type, $fields, $terms, $sort_by_sql, $sort_key, $sort_dir, $sort_days, $ex_fid_ary, $post_visibility, $topic_id, $author_ary, $author_name, $id_ary, $start, $per_page);
}
function sphinx_add_result_forum($items) {
$inCore = cmsCore::getInstance();
global $_LANG;
cmsCore::loadLanguage('components/forum');
$config = $inCore->loadComponentConfig('forum');
$search_model = cms_model_search::initModel();
foreach ($items as $id => $item) {
if (!cmsCore::checkContentAccess($item['attrs']['access_list'])) { continue; }
$pages = ceil($item['attrs']['post_count'] / $config['pp_thread']);
$result_array = array(
'link' => '/forum/thread'. $id .'-'. $pages .'.html',
'place' => $item['attrs']['forum'],
'placelink' => '/forum/'. $item['attrs']['forum_id'],
'description' => $search_model->getProposalWithSearchWord($item['attrs']['description']),
'title' => $item['attrs']['title'],
'pubdate' => date('Y-m-d H:i:s', $item['attrs']['pubdate'])
);
$search_model->addResult($result_array);
}
// Ищем в тексте постов
$cl = new SphinxClient();
$cl->SetServer('127.0.0.1', 9312);
$cl->SetMatchMode(SPH_MATCH_EXTENDED2);
$cl->SetLimits(0, 100);
$result = $cl->Query($search_model->against, $search_model->config['Sphinx_Search']['prefix'] .'_forum_posts');
if ($result !== false) {
foreach ($result['matches'] as $id => $item) {
$pages = ceil($item['attrs']['post_count'] / $config['pp_thread']);
$post_page = ($pages > 1) ? postPage::getPage($item['attrs']['thread_id'], $id, $config['pp_thread']) : 1;
$result_array = array(
'link' => '/forum/thread'. $item['attrs']['thread_id'] .'-'. $post_page .'.html#'. $id,
'place' => $_LANG['FORUM_POST'],
'placelink' => '/forum/thread'. $item['attrs']['thread_id'] .'-'. $post_page .'.html#'. $id,
'description' => $search_model->getProposalWithSearchWord($item['attrs']['content_html']),
'title' => $item['attrs']['thread'],
'imageurl' => $item['attrs']['fileurl'],
'pubdate' => date('Y-m-d H:i:s', $item['attrs']['pubdate'])
);
$search_model->addResult($result_array);
}
}
return;
}
private function _getSphinxClient()
{
require_once SCRIPT_BASE . 'lib/sphinx-2.1.9/sphinxapi.php';
$sphinxClient = new SphinxClient();
$sphinxClient->SetServer('127.0.0.1', 9312);
$sphinxClient->SetConnectTimeout(20);
$sphinxClient->SetArrayResult(true);
$sphinxClient->SetWeights(array(1000, 1));
$sphinxClient->SetMatchMode(SPH_MATCH_EXTENDED);
return $sphinxClient;
}
function query($query, $index, $offset = 0)
{
require_once DIR . "lib/sphinx/sphinxapi.php";
$sphinx = new SphinxClient();
$sphinx->setServer(SPHINX_HOST, SPHINX_PORT);
$sphinx->SetLimits($offset, 100, 10000000);
$sphinx->SetMatchMode(SPH_MATCH_EXTENDED);
$sphinx->SetSortMode(SPH_SORT_ATTR_DESC, 'date_posted');
$res = $sphinx->Query($query, $index);
return $res;
}
/**
* Непосредственно сам поиск
*
* @param string $sTerms Поисковый запрос
* @param string $sObjType Тип поиска
* @param int $iOffset Сдвиг элементов
* @param int $iLimit Количество элементов
* @param array $aExtraFilters Список фильтров
* @return array
*/
public function FindContent($sTerms, $sObjType, $iOffset, $iLimit, $aExtraFilters)
{
/**
* используем кеширование при поиске
*/
$sExtraFilters = serialize($aExtraFilters);
$cacheKey = Config::Get('module.search.entity_prefix') . "searchResult_{$sObjType}_{$sTerms}_{$iOffset}_{$iLimit}_{$sExtraFilters}";
if (false === ($data = $this->Cache_Get($cacheKey))) {
/**
* Параметры поиска
*/
$this->oSphinx->SetMatchMode(SPH_MATCH_ALL);
$this->oSphinx->SetLimits($iOffset, $iLimit);
/**
* Устанавливаем атрибуты поиска
*/
$this->oSphinx->ResetFilters();
if (!is_null($aExtraFilters)) {
foreach ($aExtraFilters as $sAttribName => $sAttribValue) {
$this->oSphinx->SetFilter($sAttribName, is_array($sAttribValue) ? $sAttribValue : array($sAttribValue));
}
}
/**
* Ищем
*/
if (!is_array($data = $this->oSphinx->Query($sTerms, Config::Get('module.search.entity_prefix') . $sObjType . 'Index'))) {
return FALSE;
// Скорее всего недоступен демон searchd
}
/**
* Если результатов нет, то и в кеш писать не стоит...
* хотя тут момент спорный
*/
if ($data['total'] > 0) {
$this->Cache_Set($data, $cacheKey, array(), 60 * 15);
}
}
return $data;
}
/**
* __construct()
* 构造函数
*/
public function __construct()
{
parent::__construct();
parent::SetServer(SPH_HOST, 9312);
//parent::SetServer("localhost",9312);
//parent::SetConnectTimeout(10);
parent::SetMatchMode(SPH_MATCH_EXTENDED2);
//parent::SetFieldWeights($this->weights);
parent::SetRankingMode(SPH_RANK_WORDCOUNT);
//
parent::SetSortMode(SPH_SORT_EXTENDED, '@weight desc');
parent::SetArrayResult(TRUE);
$this->SetLimits($this->offset, $this->limit);
}
protected function resetClient()
{
$this->sphinxClient->ResetFilters();
$this->sphinxClient->ResetGroupBy();
$this->sphinxClient->ResetOverrides();
$this->sphinxClient->SetLimits(0, 20);
$this->sphinxClient->SetArrayResult(true);
$this->sphinxClient->SetFieldWeights(array());
$this->sphinxClient->SetIDRange(0, 0);
$this->sphinxClient->SetIndexWeights(array());
$this->sphinxClient->SetMatchMode(SPH_MATCH_EXTENDED2);
$this->sphinxClient->SetRankingMode(SPH_RANK_NONE);
$this->sphinxClient->SetSortMode(SPH_SORT_RELEVANCE, "");
$this->sphinxClient->SetSelect("*");
}
public function __construct($query)
{
$this->query = $query;
$max_results = intval(SphinxSearch_Config_Plugin::getValue("results", "maxresults"));
$this->limit = $max_results;
$SphinxClient = new SphinxClient();
$this->SphinxClient = $SphinxClient;
$SphinxClient->SetMatchMode(SPH_MATCH_EXTENDED2);
$SphinxClient->SetSortMode(SPH_SORT_EXTENDED, "@weight DESC");
$SphinxClient->setServer("localhost", SphinxSearch_Config_Plugin::getValue("searchd", "port"));
// Sphinx Client is to always return everything - it's just IDs
// Paginator is then to cast the necessary Items, this can be done
// with offset/limit
$SphinxClient->setLimits(0, $max_results, $max_results);
}
function sphinx_add_result_clubs($items) {
global $_LANG;
cmsCore::m('clubs');
$search_model = cms_model_search::initModel();
foreach ($items as $id => $item) {
$result_array = array(
'link' => cmsCore::m('clubs')->getPostURL($item['attrs']['user_id'], $item['attrs']['seolink']),
'place' => ' «'. $item['attrs']['cat_title'] .'»',
'placelink' => cmsCore::m('clubs')->getBlogURL($item['attrs']['user_id']),
'description' => $search_model->getProposalWithSearchWord($item['attrs']['content_html']),
'title' => $item['attrs']['title'],
'imageurl' => $item['fileurl'],
'pubdate' => date('Y-m-d H:i:s', $item['attrs']['pubdate'])
);
$search_model->addResult($result_array);
}
/////// поиск по клубным фоткам //////////
$cl = new SphinxClient();
$cl->SetServer('127.0.0.1', 9312);
$cl->SetMatchMode(SPH_MATCH_EXTENDED2);
$cl->SetLimits(0, 100);
$result = $cl->Query($search_model->against, $search_model->config['Sphinx_Search']['prefix'] .'_clubs_photos');
if ($result !== false) {
foreach ($result['matches'] as $id => $item) {
$result_array = array(
'link' => '/clubs/photo'. $id .'.html',
'place' => $_LANG['CLUBS_PHOTOALBUM'] .' «'. $item['attrs']['cat_title'] .'»',
'placelink' => '/clubs/photoalbum'. $item['attrs']['cat_id'],
'description' => $search_model->getProposalWithSearchWord($item['attrs']['description']),
'title' => $item['attrs']['title'],
'imageurl' => (file_exists(PATH .'/images/photos/medium/'. $item['attrs']['file']) ? '/images/photos/medium/'. $item['attrs']['file'] : ''),
'pubdate' => date('Y-m-d H:i:s', $item['attrs']['pubdate'])
);
$search_model->addResult($result_array);
}
}
return;
}
/**
* Непосредственно сам поиск
*
* @param string $sQuery Поисковый запрос
* @param string $sObjType Тип поиска
* @param int $iOffset Сдвиг элементов
* @param int $iLimit Количество элементов
* @param array $aExtraFilters Список фильтров
*
* @return array
*/
public function FindContent($sQuery, $sObjType, $iOffset, $iLimit, $aExtraFilters)
{
// * используем кеширование при поиске
$sExtraFilters = serialize($aExtraFilters);
$cacheKey = Config::Get('plugin.sphinx.prefix') . "searchResult_{$sObjType}_{$sQuery}_{$iOffset}_{$iLimit}_{$sExtraFilters}";
if (false === ($data = E::ModuleCache()->Get($cacheKey))) {
// * Параметры поиска
$this->oSphinx->SetMatchMode(SPH_MATCH_ALL);
$this->oSphinx->SetLimits($iOffset, $iLimit, 1000);
// * Устанавливаем атрибуты поиска
$this->oSphinx->ResetFilters();
if (!is_null($aExtraFilters)) {
foreach ($aExtraFilters as $sAttribName => $sAttribValue) {
$this->oSphinx->SetFilter($sAttribName, is_array($sAttribValue) ? $sAttribValue : array($sAttribValue));
}
}
// * Ищем
$sIndex = Config::Get('plugin.sphinx.prefix') . $sObjType . 'Index';
$data = $this->oSphinx->Query($sQuery, $sIndex);
if (!is_array($data)) {
// Если false, то, скорее всего, ошибка и ее пишем в лог
$sError = $this->GetLastError();
if ($sError) {
$sError .= "\nquery:{$sQuery}\nindex:{$sIndex}";
if ($aExtraFilters) {
$sError .= "\nfilters:";
foreach ($aExtraFilters as $sAttribName => $sAttribValue) {
$sError .= $sAttribName . '=(' . (is_array($sAttribValue) ? join(',', $sAttribValue) : $sAttribValue) . ')';
}
}
$this->LogError($sError);
}
return false;
}
/**
* Если результатов нет, то и в кеш писать не стоит...
* хотя тут момент спорный
*/
if ($data['total'] > 0) {
E::ModuleCache()->Set($data, $cacheKey, array(), 60 * 15);
}
}
return $data;
}
function MakeSuggestion($keyword)
{
$trigrams = BuildTrigrams($keyword);
$query = "\"{$trigrams}\"/1";
$len = strlen($keyword);
$delta = LENGTH_THRESHOLD;
$cl = new SphinxClient();
$cl->SetMatchMode(SPH_MATCH_EXTENDED2);
$cl->SetRankingMode(SPH_RANK_WORDCOUNT);
$cl->SetFilterRange("len", $len - $delta, $len + $delta);
$cl->SetSelect("*, @weight+{$delta}-abs(len-{$len}) AS myrank");
$cl->SetSortMode(SPH_SORT_EXTENDED, "myrank DESC, freq DESC");
$cl->SetArrayResult(true);
// pull top-N best trigram matches and run them through Levenshtein
$cl->SetLimits(0, TOP_COUNT);
$res = $cl->Query($query, "suggest");
if (!$res || !$res["matches"]) {
return false;
}
if (SUGGEST_DEBUG) {
print "--- DEBUG START ---\n";
foreach ($res["matches"] as $match) {
$w = $match["attrs"]["keyword"];
$myrank = @$match["attrs"]["myrank"];
if ($myrank) {
$myrank = ", myrank={$myrank}";
}
// FIXME? add costs?
// FIXME! does not work with UTF-8.. THIS! IS!! PHP!!!
$levdist = levenshtein($keyword, $w);
print "id={$match['id']}, weight={$match['weight']}, freq={$match[attrs][freq]}{$myrank}, word={$w}, levdist={$levdist}\n";
}
print "--- DEBUG END ---\n";
}
// further restrict trigram matches with a sane Levenshtein distance limit
foreach ($res["matches"] as $match) {
$suggested = $match["attrs"]["keyword"];
if (levenshtein($keyword, $suggested) <= LEVENSHTEIN_THRESHOLD) {
return $suggested;
}
}
return $keyword;
}
public function getSphinxAdapter()
{
require_once Mage::getBaseDir('lib') . DIRECTORY_SEPARATOR . 'sphinxapi.php';
// Connect to our Sphinx Search Engine and run our queries
$sphinx = new SphinxClient();
$host = Mage::getStoreConfig('sphinxsearch/server/host');
$port = Mage::getStoreConfig('sphinxsearch/server/port');
if (empty($host)) {
return $sphinx;
}
if (empty($port)) {
$port = 9312;
}
$sphinx->SetServer($host, $port);
$sphinx->SetMatchMode(SPH_MATCH_EXTENDED2);
$sphinx->setFieldWeights(array('name' => 7, 'category' => 1, 'name_attributes' => 1, 'data_index' => 3));
$sphinx->setLimits(0, 200, 1000, 5000);
// SPH_RANK_PROXIMITY_BM25 ist default
$sphinx->SetRankingMode(SPH_RANK_SPH04, "");
// 2nd parameter is rank expr?
return $sphinx;
}
public function getResultByTag($keyword = "", $offset = 0, $limit = 0, $searchParams = array())
{
$sphinx = $this->config->item('sphinx');
$query = array();
$cl = new SphinxClient();
$cl->SetServer($sphinx['ip'], $sphinx['port']);
// 注意这里的主机
$cl->SetConnectTimeout($sphinx['timeout']);
$cl->SetArrayResult(true);
// $cl->SetIDRange(89,90);//过滤ID
if (isset($searchParams['provice_sid']) && $searchParams['provice_sid']) {
$cl->setFilter('provice_sid', array($searchParams['provice_sid']));
}
if (isset($searchParams['city_sid']) && $searchParams['city_sid']) {
$cl->setFilter('city_sid', array($searchParams['city_sid']));
}
if (isset($searchParams['piccode']) && $searchParams['piccode']) {
$cl->setFilter('piccode', array($searchParams['piccode']));
}
if (isset($searchParams['recent']) && $searchParams['recent']) {
$cl->SetFilterRange('createtime', time() - 86400 * 30, time());
//近期1个月
}
if (isset($searchParams['searchtype']) && $searchParams['searchtype']) {
//精确:模糊
$searchtype = SPH_MATCH_ALL;
} else {
$searchtype = SPH_MATCH_ANY;
}
$cl->SetLimits($offset, $limit);
$cl->SetMatchMode($searchtype);
// 使用多字段模式
$cl->SetSortMode(SPH_SORT_EXTENDED, "@weight desc,@id desc");
$index = "*";
$query = $cl->Query($keyword, $index);
$cl->close();
return $query;
}
/**
* sphinx search
*/
public function sphinxSearch(Request $request, \SphinxClient $sphinx)
{
$search = $request->get('search');
//sphinx的主机名和端口 //mysql -h 127.0.0.1 -P 9306
$sphinx->SetServer('127.0.0.1', 9312);
//设定搜索模式 SPH_MATCH_ALL(匹配所有的查询词)
$sphinx->SetMatchMode(SPH_MATCH_ALL);
//设置返回结果集为数组格式
$sphinx->SetArrayResult(true);
//匹配结果的偏移量，参数的意义依次为：起始位置，返回结果条数，最大匹配条数
$sphinx->SetLimits(0, 20, 1000);
//最大搜索时间
$sphinx->SetMaxQueryTime(10);
//索引源是配置文件中的 index 类，如果有多个索引源可使用,号隔开：'email,diary' 或者使用'*'号代表全部索引源
$result = $sphinx->query($search, '*');
//返回值说明 total 本次查询返回条数 total_found 一共检索到多少条 docs 在多少文档中出现 hits——共出现了多少次
//关闭查询连接
$sphinx->close();
//打印结果
/*echo "<pre>";
print_r($result);
echo "</pre>";exit();*/
$ids = [0];
if (!empty($result)) {
foreach ($result['matches'] as $key => $val) {
$ids[] = $val['id'];
}
}
$ids = implode(',', array_unique($ids));
$list = DB::select("SELECT * from documents WHERE id IN ({$ids})");
if (!empty($list)) {
foreach ($list as $key => $val) {
$val->content = str_replace($search, '<span style="color: red;">' . $search . '</span>', $val->content);
$val->title = str_replace($search, '<span style="color: red;">' . $search . '</span>', $val->title);
}
}
return view('/sphinx.search')->with('data', array('total' => $result['total'] ? $result['total'] : 0, 'time' => $result['time'] ? $result['time'] : 0, 'list' => $list));
}
private function newSphinxClient()
{
$s = new SphinxClient();
$s->setServer(Yii::app()->params['sphinx_servername'], Yii::app()->params['sphinx_port']);
$s->setMaxQueryTime(5000);
$s->SetLimits(0, 100);
$s->SetMatchMode(SPH_MATCH_EXTENDED);
return $s;
}
}
$SphinxAPI = realpath($CATSHome . '/' . SPHINX_API);
if (!file_exists($SphinxAPI)) {
fwrite($stderr, "Config Error: SPHINX_API could not be found.\n");
exit(1);
}
include $SphinxAPI;
/* Sphinx API likes to throw PHP errors *AND* use it's own error
* handling.
*/
assert_options(ASSERT_WARNING, 0);
/* Execute the Sphinx query. */
$sphinx = new SphinxClient();
$sphinx->SetServer(SPHINX_HOST, SPHINX_PORT);
$sphinx->SetWeights(array(0, 100, 0, 0, 50));
$sphinx->SetMatchMode(SPH_MATCH_BOOLEAN);
$sphinx->SetLimits(0, 10);
$sphinx->SetSortMode(SPH_SORT_TIME_SEGMENTS, 'date_added');
$sphinx->SetFilter('site_id', TEST_SITE_ID);
/* Execute the Sphinx query. Sphinx can ask us to retry if its
* maxed out. Retry up to 5 times.
*/
$tries = 0;
do {
/* Wait for one second if this isn't out first attempt. */
if (++$tries > 1) {
sleep(1);
}
$results = $sphinx->Query(TEST_QUERY, SPHINX_INDEX);
$errorMessage = $sphinx->GetLastError();
} while ($results === false && strpos($errorMessage, 'server maxed out, retry') !== false && $tries <= 5);
public function index()
{
C('TOKEN_ON', false);
$seo = seo();
$this->assign("seo", $seo);
if (isset($_GET['q'])) {
G('search');
//关键字
$q = Input::forSearch(safe_replace($this->_get("q")));
$q = htmlspecialchars(strip_tags($q));
//时间范围
$time = $this->_get("time");
//模型
$mid = (int) $this->_get("modelid");
//栏目
$catid = (int) $this->_get("catid");
//排序
$order = array("adddate" => "DESC", "searchid" => "DESC");
//搜索历史记录
$shistory = cookie("shistory");
if (!$shistory) {
$shistory = array();
}
$model = F("Model");
$category = F("Category");
if (trim($_GET['q']) == '') {
header('Location: ' . U("Search/Index/index"));
exit;
}
array_unshift($shistory, $q);
$shistory = array_slice(array_unique($shistory), 0, 10);
//加入搜索历史
cookie("shistory", $shistory);
$where = array();
//每页显示条数
$pagesize = $this->config['pagesize'] ? $this->config['pagesize'] : 10;
//缓存时间
$cachetime = (int) $this->config['cachetime'];
//按时间搜索
if ($time == 'day') {
//一天
$search_time = time() - 86400;
$where['adddate'] = array("GT", $search_time);
} elseif ($time == 'week') {
//一周
$search_time = time() - 604800;
$where['adddate'] = array("GT", $search_time);
} elseif ($time == 'month') {
//一月
$search_time = time() - 2592000;
$where['adddate'] = array("GT", $search_time);
} elseif ($time == 'year') {
//一年
$search_time = time() - 31536000;
$where['adddate'] = array("GT", $search_time);
} else {
$search_time = 0;
}
//可用数据源
$this->config['modelid'] = $this->config['modelid'] ? $this->config['modelid'] : array();
//按模型搜索
if ($mid && in_array($mid, $this->config['modelid'])) {
$where['modelid'] = array("EQ", (int) $mid);
}
//按栏目搜索
if ($catid) {
//不支持多栏目搜索，和父栏目搜索。
$where['catid'] = array("EQ", (int) $catid);
}
//分页模板
$TP = '共有{recordcount}条信息 {pageindex}/{pagecount} {first}{prev}{liststart}{list}{listend}{next}{last}';
//如果开启sphinx
if ($this->config['sphinxenable']) {
import("Sphinxapi", APP_PATH . C("APP_GROUP_PATH") . '/Search/Class/');
$sphinxhost = $this->config['sphinxhost'];
$sphinxport = $this->config['sphinxport'];
$cl = new SphinxClient();
//设置searchd的主机名和TCP端口
$cl->SetServer($sphinxhost, $sphinxport);
//设置连接超时
$cl->SetConnectTimeout(1);
//控制搜索结果集的返回格式
$cl->SetArrayResult(true);
//设置全文查询的匹配模式 api http://docs.php.net/manual/zh/sphinxclient.setmatchmode.php
$cl->SetMatchMode(SPH_MATCH_EXTENDED2);
//设置排名模式 api http://docs.php.net/manual/zh/sphinxclient.setrankingmode.php
$cl->SetRankingMode(SPH_RANK_PROXIMITY_BM25);
//按一种类似SQL的方式将列组合起来，升序或降序排列。用weight是权重排序
$cl->SetSortMode(SPH_SORT_EXTENDED, "@weight desc");
//设置返回结果集偏移量和数目
$page = (int) $this->_get(C("VAR_PAGE"));
$page = $page < 1 ? 1 : $page;
$offset = $pagesize * ($page - 1);
$cl->SetLimits($offset, $pagesize, $pagesize > 1000 ? $pagesize : 1000);
if (in_array($time, array("day", "week", "month", "year"))) {
//过滤时间
$cl->SetFilterRange('adddate', $search_time, time(), false);
}
if ($mid && in_array($mid, $this->config['modelid'])) {
//过滤模型
$cl->SetFilter('modelid', (int) $mid);
}
if ($catid) {
//过滤栏目
$cl->SetFilter('catid', (int) $catid);
}
//执行搜索 api http://docs.php.net/manual/zh/sphinxclient.query.php
$res = $cl->Query($q, "*");
//信息总数
$count = $res['total'];
//如果结果不为空
if (!empty($res['matches'])) {
$result_sphinx = $res['matches'];
}
$result = array();
//数组重新组合
foreach ($result_sphinx as $k => $v) {
$result[$k] = array("searchid" => $v['id'], "adddate" => $v['attrs']['adddate'], "catid" => $v['attrs']['catid'], "id" => $v['attrs']['id'], "modelid" => $v['attrs']['modelid']);
}
$words = array();
//搜索关键字
foreach ($res['words'] as $k => $v) {
$words[] = $k;
}
$page = page($count, $pagesize);
$page->SetPager('default', $TP, array("listlong" => "6", "first" => "首页", "last" => "尾页", "prev" => "上一页", "next" => "下一页", "list" => "*", "disabledclass" => ""));
$this->assign("Page", $page->show('default'));
} else {
import("Segment", APP_PATH . C("APP_GROUP_PATH") . '/Search/Class/');
$Segment = new Segment();
//分词结果
$segment_q = $Segment->get_keyword($Segment->split_result($q));
$words = explode(" ", $segment_q);
if (!empty($segment_q)) {
$where['_string'] = " MATCH (`data`) AGAINST ('{$segment_q}' IN BOOLEAN MODE) ";
} else {
//这种搜索最不行
$where['data'] = array('like', "%{$q}%");
}
//查询结果缓存
if ($cachetime) {
//统计
$count = M("Search")->cache(true, $cachetime)->where($where)->count();
$page = page($count, $pagesize);
$result = M("Search")->cache(true, $cachetime)->where($where)->limit($page->firstRow . ',' . $page->listRows)->order($order)->select();
} else {
$count = M("Search")->where($where)->count();
$page = $this->page($count, $pagesize);
$result = M("Search")->where($where)->limit($page->firstRow . ',' . $page->listRows)->order($order)->select();
}
$page->SetPager('default', $TP, array("listlong" => "6", "first" => "首页", "last" => "尾页", "prev" => "上一页", "next" => "下一页", "list" => "*", "disabledclass" => ""));
$this->assign("Page", $page->show('default'));
}
//搜索结果处理
if ($result && is_array($result)) {
foreach ($result as $k => $r) {
$modelid = $r['modelid'];
$id = $r['id'];
$tablename = ucwords($model[$modelid]['tablename']);
if ($tablename) {
$result[$k] = M($tablename)->where(array("id" => $id))->find();
}
}
}
//搜索记录
if (strlen($q) < 17 && strlen($q) > 1 && $result) {
$res = M("SearchKeyword")->where(array('keyword' => $q))->find();
if ($res) {
//关键词搜索数+1
M("SearchKeyword")->where(array('keyword' => $q))->setInc("searchnums");
} else {
//关键词转换为拼音
load("@.iconvfunc");
$pinyin = gbk_to_pinyin(iconv('utf-8', 'gbk//IGNORE', $q));
if (is_array($pinyin)) {
$pinyin = implode('', $pinyin);
}
M("SearchKeyword")->add(array('keyword' => $q, 'searchnums' => 1, 'data' => $segment_q, 'pinyin' => $pinyin));
}
}
//相关搜索功能
if ($this->config['relationenble']) {
$map = array();
//相关搜索
if (!empty($segment_q)) {
$relation_q = str_replace(' ', '%', $segment_q);
} else {
$relation_q = $q;
}
$map['_string'] = " MATCH (`data`) AGAINST ('%{$relation_q}%' IN BOOLEAN MODE) ";
$relation = M("SearchKeyword")->where($map)->select();
$this->assign("relation", $relation);
}
foreach ($this->config['modelid'] as $modelid) {
$source[$modelid] = array("name" => $model[$modelid]['name'], "modelid" => $modelid);
}
//搜索结果
$this->assign("result", $result);
//运行时间
$search_time = G('search', 'end', 6);
$this->assign("count", $count ? $count : 0);
$this->assign("search_time", $search_time);
$this->assign("keyword", $q);
$this->assign("category", $category);
$this->assign("source", $source);
$this->assign("time", $time);
$this->assign("modelid", $mid);
$this->assign("shistory", $shistory);
//分词后的搜索关键字
$this->assign("words", $words);
$this->display("search");
} else {
$this->display();
}
}
/**
* {@inheritdoc }.
*
* SPH_MATCH_ALL, matches all query words (<b>default mode</b>);
* SPH_MATCH_ANY, matches any of the query words;
* SPH_MATCH_PHRASE, matches query as a phrase, requiring perfect match;
* SPH_MATCH_BOOLEAN, matches query as a boolean expression;
* SPH_MATCH_EXTENDED, matches query as an expression in Sphinx internal query language;
* SPH_MATCH_FULLSCAN, matches query, forcibly using the "full scan" mode as below.
*
* @see http://sphinxsearch.com/docs/current.html#matching-modes Matching modes
*/
public function SetMatchMode($mode)
{
return parent::SetMatchMode($mode);
}
require_once '../tools/db.php';
require_once '../tools/main.php';
include_once "../tools/sphinxapi.php";
if (isset($_REQUEST["number"]) && intval($_REQUEST["number"]) > 9) {
$number = $_REQUEST["number"];
} else {
$number = 9;
}
$pageNum = isset($_REQUEST["page"]) ? intval($_REQUEST["page"]) : 0;
$mdb = new MeekroDB(DB_HOST, DB_USERNAME, DB_PASSWORD, DB_TABLE_NAME, DB_PORT, DB_CHARSET);
$sphinx = new SphinxClient();
$sphinx->SetServer(SPHINX_HOST, SPHINX_PORT);
$sphinx->SetArrayResult(true);
$sphinx->SetLimits($pageNum * $number, $number, 5000);
$sphinx->SetMaxQueryTime(10);
$sphinx->SetMatchMode(SPH_MATCH_EXTENDED2);
$sphinx->SetFilter('enabled', array(1));
if (isset($_REQUEST["type"])) {
$r1 = $mdb->queryFirstRow("SELECT id, name FROM media_type WHERE name=%s", $_REQUEST["type"]);
$type = $r1["id"];
//set type
if ($type == "") {
echo json_encode(array("result" => null, "status" => false, "error" => "请求的类型不存在"));
exit;
}
$sphinx->SetFilter('type', array($type));
}
if (isset($_REQUEST['language'])) {
$r2 = $mdb->queryFirstRow("SELECT id, name FROM media_language WHERE name=%s", $_REQUEST["language"]);
$language = $r2["id"];
//set language
function do_query($search_str)
{
//$tmp_var = array(array('itemName' => "test1"), array('itemName' => "test2"), array('itemName' => "test3"));
//echo implode(",",tmp_var);
//echo json_encode($tmp_var);
//return tmp_var;
$q = "";
$sql = "";
$mode = SPH_MATCH_ALL;
$host = "localhost";
$port = 9312;
$index = "*";
$groupby = "";
$groupsort = "@group desc";
$filter = "group_id";
$filtervals = array();
$distinct = "";
$sortby = "";
$sortexpr = "";
$limit = 20;
$ranker = SPH_RANK_PROXIMITY_BM25;
$select = "*";
$cl = new SphinxClient();
$cl->SetServer($host, $port);
$cl->SetConnectTimeout(1);
$cl->SetArrayResult(true);
$cl->SetWeights(array(100, 1));
$cl->SetMatchMode($mode);
if (count($filtervals)) {
$cl->SetFilter($filter, $filtervals);
}
if ($groupby) {
$cl->SetGroupBy($groupby, SPH_GROUPBY_ATTR, $groupsort);
}
if ($sortby) {
$cl->SetSortMode(SPH_SORT_EXTENDED, $sortby);
}
if ($sortexpr) {
$cl->SetSortMode(SPH_SORT_EXPR, $sortexpr);
}
if ($distinct) {
$cl->SetGroupDistinct($distinct);
}
if ($select) {
$cl->SetSelect($select);
}
if ($limit) {
$cl->SetLimits(0, $limit, $limit > 1000 ? $limit : 1000);
}
$cl->SetRankingMode($ranker);
$res = $cl->Query($search_str, $index);
//return $res;
if (is_array($res["matches"])) {
$results = array();
$n = 1;
//print "Matches:\n";
foreach ($res["matches"] as $docinfo) {
//print "$n. doc_id=$docinfo[id], weight=$docinfo[weight]";
$attr_array = array();
$results[$docinfo[id]];
foreach ($res["attrs"] as $attrname => $attrtype) {
$value = $docinfo["attrs"][$attrname];
if ($attrtype == SPH_ATTR_MULTI || $attrtype == SPH_ATTR_MULTI64) {
$value = "(" . join(",", $value) . ")";
} else {
if ($attrtype == SPH_ATTR_TIMESTAMP) {
$value = date("Y-m-d H:i:s", $value);
}
}
$attr_array[$attrname] = $value;
//print $value;
}
$results[$docinfo[id]] = $attr_array;
$n++;
//print implode("",$results)."\n";
}
return $results;
}
}
function getCategory($keywords)
{
$q = $keywords;
$mode = SPH_MATCH_EXTENDED;
$item = array();
$comma_separated = "";
//Init config
$host = SPHINX_SERVER;
$port = SPHINX_PORT;
$index = '*';
$ranker = SPH_RANK_PROXIMITY_BM25;
$cl = new SphinxClient();
$cl->SetServer($host, $port);
$cl->SetConnectTimeout(1);
$cl->SetWeights(array(100, 1));
$cl->SetMatchMode($mode);
$cl->SetRankingMode($ranker);
$cl->SetArrayResult(true);
$cl->SetFilter('status', array('1'));
$cl->SetGroupBy("level_1_category_id", SPH_GROUPBY_ATTR, "@count DESC");
$res = $cl->Query($q, $index);
$arr = array();
if ($res && isset($res["matches"])) {
if (is_array($res["matches"])) {
foreach ($res["matches"] as $results) {
$arr[] = $results["attrs"];
}
}
}
return $arr;
}
public function __construct($rowsPerPage, $currentPage, $siteID, $wildCardString, $sortBy, $sortDirection)
{
$this->_db = DatabaseConnection::getInstance();
$this->_siteID = $siteID;
$this->_sortByFields = array('firstName', 'lastName', 'city', 'state', 'dateModifiedSort', 'dateCreatedSort', 'ownerSort');
if (ENABLE_SPHINX) {
/* Sphinx API likes to throw PHP errors *AND* use it's own error
* handling.
*/
assert_options(ASSERT_WARNING, 0);
$sphinx = new SphinxClient();
$sphinx->SetServer(SPHINX_HOST, SPHINX_PORT);
$sphinx->SetWeights(array(0, 100, 0, 0, 50));
$sphinx->SetMatchMode(SPH_MATCH_EXTENDED);
$sphinx->SetLimits(0, 1000);
$sphinx->SetSortMode(SPH_SORT_TIME_SEGMENTS, 'date_added');
// FIXME: This can be sped up a bit by actually grouping ranges of
// site IDs into their own index's. Maybe every 500 or so at
// least on the Hosted system.
$sphinx->SetFilter('site_id', array($this->_siteID));
/* Create the Sphinx query string. */
$wildCardString = DatabaseSearch::humanToSphinxBoolean($wildCardString);
/* Execute the Sphinx query. Sphinx can ask us to retry if its
* maxed out. Retry up to 5 times.
*/
$tries = 0;
do {
/* Wait for one second if this isn't out first attempt. */
if (++$tries > 1) {
sleep(1);
}
$results = $sphinx->Query($wildCardString, SPHINX_INDEX);
$errorMessage = $sphinx->GetLastError();
} while ($results === false && strpos($errorMessage, 'server maxed out, retry') !== false && $tries <= 5);
/* Throw a fatal error if Sphinx errors occurred. */
if ($results === false) {
$this->fatal('Sphinx Error: ' . ucfirst($errorMessage) . '.');
}
/* Throw a fatal error (for now) if Sphinx warnings occurred. */
$lastWarning = $sphinx->GetLastWarning();
if (!empty($lastWarning)) {
// FIXME: Just display a warning, and notify dev team.
$this->fatal('Sphinx Warning: ' . ucfirst($lastWarning) . '.');
}
/* Show warnings for assert()s again. */
assert_options(ASSERT_WARNING, 1);
if (empty($results['matches'])) {
$this->_WHERE = '0';
} else {
$attachmentIDs = implode(',', array_keys($results['matches']));
$this->_WHERE = 'attachment.attachment_id IN(' . $attachmentIDs . ')';
}
} else {
$this->_WHERE = DatabaseSearch::makeBooleanSQLWhere(DatabaseSearch::fulltextEncode($wildCardString), $this->_db, 'attachment.text');
}
/* How many companies do we have? */
$sql = sprintf("SELECT\n COUNT(*) AS count\n FROM\n attachment\n LEFT JOIN candidate\n ON attachment.data_item_id = candidate.candidate_id\n AND attachment.data_item_type = %s\n AND attachment.site_id = candidate.site_id\n LEFT JOIN user AS owner_user\n ON candidate.owner = owner_user.user_id\n WHERE\n resume = 1\n AND\n %s\n AND\n (ISNULL(candidate.is_admin_hidden) OR (candidate.is_admin_hidden = 0))\n AND\n (ISNULL(candidate.is_active) OR (candidate.is_active = 1))\n AND\n attachment.site_id = %s", DATA_ITEM_CANDIDATE, $this->_WHERE, $this->_siteID);
$rs = $this->_db->getAssoc($sql);
/* Pass "Search By Resume"-specific parameters to Pager constructor. */
parent::__construct($rs['count'], $rowsPerPage, $currentPage);
}
/**
* ¬озвращает список публичных типовых услуг по заданным услови¤м и пагинацией
*
* @return array
*/
public function getList($excluded_ids = array())
{
$criteria = array($this->category_id, $this->city_id, $this->country_id, $this->keywords, $this->limit, $this->offset, $this->price_ranges, $this->price_max, $this->order, $excluded_ids, $this->user_id);
$membuf = new memBuff();
$memkey = __METHOD__ . '#' . md5(serialize($criteria));
if (false !== ($result = $membuf->get($memkey)) && is_release()) {
return $result;
}
$sort = $this->getSort();
# @see http://sphinxsearch.com/forum/view.html?id=11538 about city = x or country = y
$sphinxClient = new SphinxClient();
$sphinxClient->SetServer(SEARCHHOST, SEARCHPORT);
$sphinxClient->SetLimits($this->offset, $this->limit, 20000);
$sphinxClient->SetSortMode(SPH_SORT_EXTENDED, $sort);
$sphinxClient->SetFieldWeights(array('title' => 2, 'extra_title' => 1));
//$sphinxClient->SetRankingMode(SPH_RANK_PROXIMITY_BM25);
$selectExpression = '*';
// все колонки
if ($this->user_id) {
$selectExpression .= ", IF(user_id = {$this->user_id}, 1, 0) as match_user";
$sphinxClient->setFilter('match_user', array(1));
}
if ($this->category_id) {
$selectExpression .= ", IF(category_id = {$this->category_id} or category_parent_id = {$this->category_id}, 1, 0) as match_category";
$sphinxClient->setFilter('match_category', array(1));
}
if ($this->country_id) {
$selectExpression .= ", IF(user_country_id = {$this->country_id} or country_id = {$this->country_id}, 1, 0) as match_country";
$sphinxClient->setFilter('match_country', array(1));
}
if ($this->city_id) {
$selectExpression .= ", IF(user_city_id = {$this->city_id} or city_id = {$this->city_id}, 1, 0) as match_city";
$sphinxClient->setFilter('match_city', array(1));
}
if (count($this->price_ranges)) {
$match_price_exprs = array();
foreach ($this->getPriceRanges() as $i => $price_range) {
if (!isset($this->price_ranges[$i])) {
continue;
}
$match_price_exprs[] = "price_{$i} = 1";
}
$match_price_exprs = implode(' or ', $match_price_exprs);
$selectExpression .= ", IF({$match_price_exprs}, 1, 0) as match_price";
$sphinxClient->setFilter('match_price', array(1));
}
if ($this->price_max > 0) {
$selectExpression .= ", IF(price <= {$this->price_max}, 1, 0) as match_price_max";
$sphinxClient->setFilter('match_price_max', array(1));
}
$searchString = '';
if (!empty($this->keywords)) {
$keywords = implode(' ', array_filter(preg_split('/\\s*,\\s*/', $this->keywords)));
$searchString = trim($keywords);
//$searchString = $this->GetSphinxKeyword($searchString);
$sphinxClient->SetMatchMode(SPH_MATCH_ANY);
//SPH_MATCH_EXTENDED2);
}
if (count($excluded_ids)) {
$sphinxClient->setFilter('tservice_id', $excluded_ids, true);
}
$sphinxClient->SetSelect($selectExpression);
$queryResult = $sphinxClient->query($searchString, "tservices;delta_tservices");
//echo '<pre>error: ', $sphinxClient->GetLastError(), '</pre>';
//echo '<pre>warn : ', $sphinxClient->GetLastWarning(), '</pre>';
$list = array();
$total = 0;
if (isset($queryResult['matches'])) {
foreach ($queryResult['matches'] as $id => $row) {
$row['attrs']['id'] = $id;
$list[] = $row['attrs'];
}
$total = $queryResult['total_found'] < $queryResult['total'] ? $queryResult['total_found'] : $queryResult['total'];
}
$result = array('list' => $list, 'total' => $total);
if ($this->_ttl) {
$membuf->set($memkey, $result, $this->_ttl);
}
return $result;
}
