Example #1
0
 }
 if ($_G['adminid'] != '1' && $_G['setting']['search']['forum']['maxspm']) {
     if (DB::result_first("SELECT COUNT(*) FROM " . DB::table('common_searchindex') . " WHERE srchmod='{$srchmod}' AND dateline>'{$_G['timestamp']}'-60") >= $_G['setting']['search']['forum']['maxspm']) {
         showmessage('search_toomany', 'search.php?mod=forum', array('maxspm' => $_G['setting']['search']['forum']['maxspm']));
     }
 }
 if ($srchtype == 'fulltext' && $_G['setting']['sphinxon']) {
     require_once libfile('class/sphinx');
     $s = new SphinxClient();
     $s->setServer($_G['setting']['sphinxhost'], intval($_G['setting']['sphinxport']));
     $s->setMaxQueryTime(intval($_G['setting']['sphinxmaxquerytime']));
     $s->SetRankingMode($_G['setting']['sphinxrank']);
     $s->setLimits(0, intval($_G['setting']['sphinxlimit']), intval($_G['setting']['sphinxlimit']));
     $s->setGroupBy('tid', SPH_GROUPBY_ATTR);
     if ($srchfilter == 'digest') {
         $s->setFilterRange('digest', 1, 3, false);
     }
     if ($srchfilter == 'top') {
         $s->setFilterRange('displayorder', 1, 2, false);
     } else {
         $s->setFilterRange('displayorder', 0, 2, false);
     }
     if (!empty($srchfrom) && empty($srchtxt) && empty($srchuid) && empty($srchuname)) {
         $expiration = TIMESTAMP + $cachelife_time;
         $keywords = '';
         if ($before) {
             $spx_timemix = 0;
             $spx_timemax = TIMESTAMP - $srchfrom;
         } else {
             $spx_timemix = TIMESTAMP - $srchfrom;
             $spx_timemax = TIMESTAMP;
Example #2
0
 /**
  * Does an index search via Sphinx and returns the results
  *
  * @param string $type Search type.  Valid types are: author, title, series, subject, keyword (default)
  * @param string $term Search term/phrase
  * @param int $limit Number of results to return
  * @param int $offset Where to begin result set -- for pagination purposes
  * @param array $sort_array Numerically keyed array of sort parameters.  Valid options are: newest, oldest
  * @param array $location_array Numerically keyed array of location params.  NOT IMPLEMENTED YET
  * @param array $facet_args String-keyed array of facet parameters. See code below for array structure
  * @return array String-keyed result set
  */
 public function search($type, $term, $limit, $offset, $sort_opt = NULL, $format_array = array(), $location_array = array(), $facet_args = array(), $override_search_filter = FALSE, $limit_available = FALSE, $show_inactive = FALSE)
 {
     if (is_callable(array(__CLASS__ . '_hook', __FUNCTION__))) {
         eval('$hook = new ' . __CLASS__ . '_hook;');
         return $hook->{__FUNCTION__}($type, $term, $limit, $offset, $sort_opt, $format_array, $location_array, $facet_args, $override_search_filter, $limit_available);
     }
     require_once $this->locum_config['sphinx_config']['api_path'] . '/sphinxapi.php';
     $db =& MDB2::connect($this->dsn);
     $term_arr = explode('?', trim(preg_replace('/\\//', ' ', $term)));
     $term = trim($term_arr[0]);
     if ($term == '*' || $term == '**') {
         $term = '';
     } else {
         $term_prestrip = $term;
         //$term = preg_replace('/[^A-Za-z0-9*\- ]/iD', '', $term);
         $term = preg_replace('/\\*\\*/', '*', $term);
         //fix for how iii used to do wildcards
     }
     $final_result_set['term'] = $term;
     $final_result_set['type'] = trim($type);
     $cl = new SphinxClient();
     $cl->SetServer($this->locum_config['sphinx_config']['server_addr'], (int) $this->locum_config['sphinx_config']['server_port']);
     // Defaults to 'keyword', non-boolean
     $bool = FALSE;
     $cl->SetMatchMode(SPH_MATCH_ALL);
     if (!$term) {
         // Searches for everything (usually for browsing purposes--Hot/New Items, etc..)
         $cl->SetMatchMode(SPH_MATCH_EXTENDED2);
     } else {
         $picturebook = array('picturebook', 'picture book');
         $picbk_search = '(@callnum ^E)';
         $term = str_ireplace($picturebook, $picbk_search, $term);
         if ($type == 'keyword') {
             // Custom fiction and non-fiction search
             $nonfic_search = ' (@callnum "0*" | @callnum "1*" | @callnum "2*" | @callnum "3*" | @callnum "4*" | @callnum "5*" | @callnum "6*" | @callnum "7*" | @callnum "8*" | @callnum "9*")';
             $fiction_search = ' (@title fiction | @subjects fiction | @callnum mystery | @callnum fantasy | @callnum fiction | @callnum western | @callnum romance)';
             if (stripos($term, 'nonfiction') !== FALSE) {
                 $term = '@@relaxed ' . str_ireplace('nonfiction', '', $term) . $nonfic_search;
             } else {
                 if (strpos($term, 'non-fiction') !== FALSE) {
                     $term = '@@relaxed ' . str_ireplace('non-fiction', '', $term) . $nonfic_search;
                 } else {
                     if (strpos($term, 'fiction') !== FALSE) {
                         $term = '@@relaxed ' . str_ireplace('fiction', '', $term) . $fiction_search;
                     }
                 }
             }
         }
         // Is it a boolean search?
         if (preg_match("/ \\| /i", $term) || preg_match("/ \\-/i", $term) || preg_match("/ \\!/i", $term)) {
             $cl->SetMatchMode(SPH_MATCH_BOOLEAN);
             $bool = TRUE;
         }
         if (preg_match("/ OR /i", $term)) {
             $cl->SetMatchMode(SPH_MATCH_BOOLEAN);
             $term = preg_replace('/ OR /i', ' | ', $term);
             $bool = TRUE;
         }
         // Is it a phrase search?
         if (preg_match("/\"/i", $term) || preg_match("/\\@/i", $term)) {
             $cl->SetMatchMode(SPH_MATCH_EXTENDED2);
             $bool = TRUE;
         }
     }
     // Set up for the various search types
     switch ($type) {
         case 'author':
             $cl->SetFieldWeights(array('author' => 50, 'addl_author' => 30));
             $idx = 'bib_items_author';
             break;
         case 'title':
             $cl->SetFieldWeights(array('title' => 50, 'title_medium' => 50, 'series' => 30));
             $idx = 'bib_items_title';
             break;
         case 'series':
             $cl->SetFieldWeights(array('title' => 5, 'series' => 80));
             $idx = 'bib_items_title';
             break;
         case 'subject':
             $idx = 'bib_items_subject';
             break;
         case 'callnum':
             $cl->SetFieldWeights(array('callnum' => 100));
             $idx = 'bib_items_callnum';
             //$cl->SetMatchMode(SPH_MATCH_ANY);
             break;
         case 'tags':
             $cl->SetFieldWeights(array('tag_idx' => 100));
             $idx = 'bib_items_tags';
             //$cl->SetMatchMode(SPH_MATCH_PHRASE);
             break;
         case 'reviews':
             $cl->SetFieldWeights(array('review_idx' => 100));
             $idx = 'bib_items_reviews';
             break;
         case 'keyword':
         default:
             $cl->SetFieldWeights(array('title' => 400, 'title_medium' => 30, 'author' => 70, 'addl_author' => 40, 'tag_idx' => 25, 'series' => 25, 'review_idx' => 10, 'notes' => 10, 'subjects' => 5));
             $idx = 'bib_items_keyword';
             break;
     }
     // Filter out the records we don't want shown, per locum.ini
     if (!$override_search_filter) {
         if (trim($this->locum_config['location_limits']['no_search'])) {
             $cfg_filter_arr = $this->csv_parser($this->locum_config['location_limits']['no_search']);
             foreach ($cfg_filter_arr as $cfg_filter) {
                 $cfg_filter_vals[] = $this->string_poly($cfg_filter);
             }
             $cl->SetFilter('loc_code', $cfg_filter_vals, TRUE);
         }
     }
     // Valid sort types are 'newest' and 'oldest'.  Default is relevance.
     switch ($sort_opt) {
         case 'newest':
             $cl->SetSortMode(SPH_SORT_EXTENDED, 'pub_year DESC, @relevance DESC');
             break;
         case 'oldest':
             $cl->SetSortMode(SPH_SORT_EXTENDED, 'pub_year ASC, @relevance DESC');
             break;
         case 'catalog_newest':
             $cl->SetSortMode(SPH_SORT_EXTENDED, 'bib_created DESC, @relevance DESC');
             break;
         case 'catalog_oldest':
             $cl->SetSortMode(SPH_SORT_EXTENDED, 'bib_created ASC, @relevance DESC');
             break;
         case 'title':
             $cl->SetSortMode(SPH_SORT_ATTR_ASC, 'title_ord');
             break;
         case 'author':
             $cl->SetSortMode(SPH_SORT_EXTENDED, 'author_null ASC, author_ord ASC');
             break;
         case 'top_rated':
             $cl->SetSortMode(SPH_SORT_ATTR_DESC, 'rating_idx');
             break;
         case 'popular_week':
             $cl->SetSortMode(SPH_SORT_ATTR_DESC, 'hold_count_week');
             break;
         case 'popular_month':
             $cl->SetSortMode(SPH_SORT_ATTR_DESC, 'hold_count_month');
             break;
         case 'popular_year':
             $cl->SetSortMode(SPH_SORT_ATTR_DESC, 'hold_count_year');
             break;
         case 'popular_total':
             $cl->SetSortMode(SPH_SORT_ATTR_DESC, 'hold_count_total');
             break;
         case 'atoz':
             $cl->SetSortMode(SPH_SORT_ATTR_ASC, 'title_ord');
             break;
         case 'ztoa':
             $cl->SetSortMode(SPH_SORT_ATTR_DESC, 'title_ord');
             break;
         default:
             $cl->SetSortMode(SPH_SORT_EXPR, "@weight + (hold_count_total)*0.02");
             break;
     }
     // Filter by material types
     if (is_array($format_array)) {
         foreach ($format_array as $format) {
             if (strtolower($format) != 'all') {
                 $filter_arr_mat[] = $this->string_poly(trim($format));
             }
         }
         if (count($filter_arr_mat)) {
             $cl->SetFilter('mat_code', $filter_arr_mat);
         }
     }
     // Filter by location
     if (count($location_array)) {
         foreach ($location_array as $location) {
             if (strtolower($location) != 'all') {
                 $filter_arr_loc[] = $this->string_poly(trim($location));
             }
         }
         if (count($filter_arr_loc)) {
             $cl->SetFilter('loc_code', $filter_arr_loc);
         }
     }
     // Filter by pub_year
     if ($facet_args['facet_year']) {
         if (strpos($facet_args['facet_year'][0], '-') !== FALSE) {
             $min_year = 1;
             $max_year = 9999;
             $args = explode('-', $facet_args['facet_year'][0]);
             $min_arg = (int) $args[0];
             $max_arg = (int) $args[1];
             if ($min_arg && $min_arg > $min_year) {
                 $min_year = $min_arg;
             }
             if ($max_arg && $max_arg < $max_year) {
                 $max_year = $max_arg;
             }
             $cl->setFilterRange('pub_year', $min_year, $max_year);
         } else {
             $cl->SetFilter('pub_year', $facet_args['facet_year']);
         }
     }
     // Filter by pub_decade
     if ($facet_args['facet_decade']) {
         $cl->SetFilter('pub_decade', $facet_args['facet_decade']);
     }
     // Filter by lexile
     if ($facet_args['facet_lexile']) {
         $cl->SetFilter('lexile', $facet_args['facet_lexile']);
     }
     // Filter by Series
     if (count($facet_args['facet_series'])) {
         foreach ($facet_args['facet_series'] as &$facet_series) {
             $facet_series = $this->string_poly($facet_series);
         }
         $cl->SetFilter('series_attr', $facet_args['facet_series']);
     }
     // Filter by Language
     if (count($facet_args['facet_lang'])) {
         foreach ($facet_args['facet_lang'] as &$facet_lang) {
             $facet_lang = $this->string_poly($facet_lang);
         }
         $cl->SetFilter('lang', $facet_args['facet_lang']);
     }
     // Filter inactive records
     if (!$show_inactive) {
         $cl->SetFilter('active', array('0'), TRUE);
     }
     // Filter by age
     if (count($facet_args['age'])) {
         foreach ($facet_args['age'] as $age_facet) {
             $cl->SetFilter('ages', array($this->string_poly($age_facet)));
         }
     }
     // Filter by availability
     if ($limit_available) {
         $cl->SetFilter('branches', array($this->string_poly($limit_available)));
     }
     $cl->SetRankingMode(SPH_RANK_SPH04);
     $proximity_check = $cl->Query($term, $idx);
     // Quick check on number of results
     // If original match didn't return any results, try a proximity search
     if (empty($proximity_check['matches']) && $bool == FALSE && $term != "*" && $type != "tags") {
         $term = '"' . $term . '"/1';
         $cl->SetMatchMode(SPH_MATCH_EXTENDED);
         $forcedchange = 'yes';
     }
     // Paging/browsing through the result set.
     $sort_limit = 2000;
     if ($offset + $limit > $sort_limit) {
         $sort_limit = $offset + $limit;
     }
     $cl->SetLimits((int) $offset, (int) $limit, (int) $sort_limit);
     // And finally.... we search.
     $cl->AddQuery($term, $idx);
     // CREATE FACETS
     $cl->SetLimits(0, 1000);
     // Up to 1000 facets
     $cl->SetArrayResult(TRUE);
     // Allow duplicate documents in result, for facet grouping
     $cl->SetGroupBy('pub_year', SPH_GROUPBY_ATTR);
     $cl->AddQuery($term, $idx);
     $cl->ResetGroupBy();
     $cl->SetGroupBy('pub_decade', SPH_GROUPBY_ATTR);
     $cl->AddQuery($term, $idx);
     $cl->ResetGroupBy();
     $cl->SetGroupBy('mat_code', SPH_GROUPBY_ATTR, '@count desc');
     $cl->AddQuery($term, $idx);
     $cl->ResetGroupBy();
     $cl->SetGroupBy('branches', SPH_GROUPBY_ATTR, '@count desc');
     $cl->AddQuery($term, $idx);
     $cl->ResetGroupBy();
     $cl->SetGroupBy('ages', SPH_GROUPBY_ATTR, '@count desc');
     $cl->AddQuery($term, $idx);
     $cl->ResetGroupBy();
     $cl->SetGroupBy('lang', SPH_GROUPBY_ATTR, '@count desc');
     $cl->AddQuery($term, $idx);
     $cl->ResetGroupBy();
     $cl->SetGroupBy('series_attr', SPH_GROUPBY_ATTR, '@count desc');
     $cl->AddQuery($term, $idx);
     $cl->ResetGroupBy();
     $cl->SetGroupBy('lexile', SPH_GROUPBY_ATTR);
     $cl->AddQuery($term, $idx);
     $cl->ResetGroupBy();
     $results = $cl->RunQueries();
     // Include descriptors
     $final_result_set['num_hits'] = $results[0]['total_found'];
     if ($results[0]['total'] <= $this->locum_config['api_config']['suggestion_threshold'] || $forcedchange == 'yes') {
         if ($this->locum_config['api_config']['use_yahoo_suggest'] == TRUE) {
             $final_result_set['suggestion'] = $this->yahoo_suggest($term_prestrip);
         }
     }
     // Pull full records out of Couch
     if ($final_result_set['num_hits']) {
         $skip_avail = $this->csv_parser($this->locum_config['format_special']['skip_avail']);
         $bib_hits = array();
         foreach ($results[0]['matches'] as $match) {
             $bib_hits[] = (string) $match['id'];
         }
         $final_result_set['results'] = $this->get_bib_items_arr($bib_hits);
         foreach ($final_result_set['results'] as &$result) {
             $result = $result['value'];
             if ($result['bnum']) {
                 // Get availability (Only cached)
                 $result['status'] = $this->get_item_status($result['bnum'], FALSE, TRUE);
             }
         }
     }
     $final_result_set['facets'] = $this->sphinx_facetizer($results);
     if ($forcedchange == 'yes') {
         $final_result_set['changed'] = 'yes';
     }
     return $final_result_set;
 }
Example #3
0
function sphinx_search($keyword)
{
    $fid = 0;
    $daterange = 0;
    $orderby = 'match';
    $page = 1;
    $pagesize = 60;
    global $conf, $time;
    $cl = new SphinxClient();
    $cl->SetServer($conf['sphinx_host'], $conf['sphinx_port']);
    $cl->SetConnectTimeout(3);
    $cl->SetArrayResult(TRUE);
    $cl->SetWeights(array(100, 1, 5));
    // 标题权重100,内容权重1,作者权重10
    $fid && $cl->SetFilter('fid', array($fid));
    $daterange && $cl->setFilterRange('dateline', $time - $daterange * 86400, $time);
    $cl->SetMatchMode(SPH_MATCH_ALL);
    if ($orderby == 'match') {
        $cl->SetSortMode(SPH_SORT_RELEVANCE);
        // 如果不设置,默认按照权重排序!但是TMD是正序!
    } elseif ($orderby == 'timeasc') {
        $cl->SetSortMode(SPH_SORT_ATTR_ASC, 'tid');
    } elseif ($orderby == 'timedesc') {
        $cl->SetSortMode(SPH_SORT_ATTR_DESC, 'tid');
    }
    //$cl->SetSortMode (SPH_SORT_ATTR_DESC, 'tid');	// 如果不设置,默认按照权重排序!但是TMD是正序!
    /*
    $cl->SetMatchMode ( SPH_MATCH_EXTENDED );	//设置模式
    $cl->SetRankingMode ( SPH_RANK_PROXIMITY );	//设置评分模式
    $cl->SetFieldWeights (array('subject'=>100,'message'=>10,'username'=>1));//设置字段的权重,如果area命中,那么权重算2
    $cl->SetSortMode ('SPH_SORT_EXPR','@weight');	//按照权重排序
    */
    // --------------> 优先搜索增量索引
    $newlist = array();
    $forums = array();
    if ($page == 1) {
        $cl->SetLimits(0, $pagesize, 1000);
        // 最大结果集
        $res = $cl->Query($keyword, $conf['sphinx_deltasrc']);
        // * 为所有的索引
        if (!empty($cl->_error)) {
            return xn_error(-1, 'Sphinx 错误:' . $cl->_error);
        }
        if (!empty($res) && !empty($res['total'])) {
            $deltamatch = $res['matches'];
        }
        $res['matches'] && arrlist_change_key($res['matches'], 'id');
        $newlist = array();
        $forums = array();
        foreach ((array) $res['matches'] as $v) {
            if (empty($v['attrs'])) {
                continue;
            }
            if (empty($v['attrs']['fid'])) {
                continue;
            }
            $fid = $v['attrs']['fid'];
            $thread = thread_read($v['attrs']['tid']);
            if (empty($thread)) {
                continue;
            }
            if (stripos($thread['subject'], $keyword) === FALSE) {
                continue;
            }
            $thread['subject'] = str_replace($keyword, '<span class="red">' . $keyword . '</span>', $thread['subject']);
            $newlist[] = $thread;
        }
    }
    // --------------> 再搜索主索引
    $start = ($page - 1) * $pagesize;
    $cl->SetLimits($start, $pagesize, 1000);
    // 最大结果集
    $res = $cl->Query($keyword, $conf['sphinx_datasrc']);
    if (!empty($cl->_error)) {
        return xn_error(-1, 'Sphinx 错误:' . $cl->_error);
    }
    if (empty($res) || empty($res['total'])) {
        $res['matches'] = $deltamatch;
    } else {
        arrlist_change_key($res['matches'], 'id');
    }
    $threadlist = array();
    foreach ((array) $res['matches'] as $v) {
        if (empty($v['attrs'])) {
            continue;
        }
        if (empty($v['attrs']['fid'])) {
            continue;
        }
        $fid = $v['attrs']['fid'];
        $thread = thread_read($v['attrs']['tid']);
        if (empty($thread)) {
            continue;
        }
        $thread['subject'] = str_replace($keyword, '<span class="red">' . $keyword . '</span>', $thread['subject']);
        $threadlist[] = $thread;
    }
    $arrlist = $newlist + $threadlist;
    return $arrlist;
}
Example #4
0
 public function fetch()
 {
     if (!class_exists('SphinxClient')) {
         return false;
     }
     $s = new SphinxClient();
     $s->setServer($this->_sphinxHost, $this->_sphinxPort);
     if (count($this->_arrSearchOutRangeColumnMinMax) > 0) {
         foreach ($this->_arrSearchOutRangeColumnMinMax as $value) {
             $d = explode(',', $value);
             $s->setFilterRange($d[0], $d[1], $d[2], true);
         }
     }
     if (count($this->_arrSearchInRangeColumnMinMax) > 0) {
         foreach ($this->_arrSearchInRangeColumnMinMax as $value) {
             $d = explode(',', $value);
             $s->setFilterRange($d[0], $d[1], $d[2], false);
         }
     }
     $s->setConnectTimeout($this->_connectTimeout);
     $s->setMaxQueryTime($this->{$_maxquerytime});
     //            $s->setRetries ( int $this->retriesCount , int $this->retriesDelay  );
     //
     $s->setMatchMode($this->searchMode);
     $s->setFieldWeights($this->_fieldweights);
     //            $s->setFilter ( string $attribute , array $values [, bool $exclude = false ] );
     //            $s->setFilterFloatRange ( string $attribute , float $min , float $max [, bool $exclude = false ] );
     //            $s->setFilterRange ( string $attribute , int $min , int $max [, bool $exclude = false ] );
     //            $s->setGeoAnchor ( string $attrlat , string $attrlong , float $latitude , float $longitude );
     //            $s->setGroupBy ( string $attribute , int $func [, string $groupsort = "@group desc" ] );
     //            $s->setGroupDistinct ( string $attribute );
     //            $s->setIDRange ( int $min , int $max );
     $s->setIndexWeights($this->_arrIndexweights);
     //            $s->setLimits ( int $offset , int $limit [, int $max_matches = 0 [, int $cutoff = 0 ]] );
     $s->setMatchMode($this->searchMode);
     //            $s->setOverride ( string $attribute , int $type , array $values );
     $s->setRankingMode($this->rankMode);
     //            $s->setSelect ( string $clause );
     //            $s->setSortMode ( int $mode [, string $sortby ] );
     return $s->query($this->_query);
 }
 /**
  * Binds the filter to a SphinxClient instance
  *
  * @param \SphinxClient $sphinx
  *
  * @return FilterRange
  */
 public function bindToSphinx(\SphinxClient $sphinx)
 {
     $sphinx->setFilterRange($this->name, $this->min, $this->max, $this->exclude);
     return $this;
 }