10Syaroslav@ivinco.com<?php
2*124Syaroslav@ivinco.com/*
30Syaroslav@ivinco.com * To change this template, choose Tools | Templates
40Syaroslav@ivinco.com * and open the template in the editor.
50Syaroslav@ivinco.com */
60Syaroslav@ivinco.com
70Syaroslav@ivinco.comclass SphinxSearch
80Syaroslav@ivinco.com{
90Syaroslav@ivinco.com    private $_sphinx = null;
100Syaroslav@ivinco.com    private $_result = array();
110Syaroslav@ivinco.com    private $_index = null;
1215Syaroslav@ivinco.com    private $_query = '';
1319Syaroslav@ivinco.com
1416Syaroslav@ivinco.com    private $_snippetSize = 256;
1516Syaroslav@ivinco.com    private $_aroundKeyword = 5;
1669Syaroslav@ivinco.com    private $_resultsPerPage = 10;
1719Syaroslav@ivinco.com
1880Syaroslav@ivinco.com    private $_titlePriority = 1;
1980Syaroslav@ivinco.com    private $_bodyPriority = 1;
2080Syaroslav@ivinco.com    private $_namespacePriority = 1;
2180Syaroslav@ivinco.com    private $_pagenamePriority = 1;
22*124Syaroslav@ivinco.com
230Syaroslav@ivinco.com    public function  __construct($host, $port, $index)
240Syaroslav@ivinco.com    {
250Syaroslav@ivinco.com        $this->_sphinx = new SphinxClient();
260Syaroslav@ivinco.com        $this->_sphinx->SetServer($host, $port);
27*124Syaroslav@ivinco.com        $this->_sphinx->SetMatchMode(SPH_MATCH_EXTENDED2);
280Syaroslav@ivinco.com
290Syaroslav@ivinco.com        $this->_index = $index;
300Syaroslav@ivinco.com    }
310Syaroslav@ivinco.com
3269Syaroslav@ivinco.com    public function setSearchAllQuery($keywords, $categories)
3369Syaroslav@ivinco.com    {
3497Syaroslav@ivinco.com        $keywords = $this->_sphinx->EscapeString($keywords);
35*124Syaroslav@ivinco.com        $keywords = $this->_enableQuotesAndDefis($keywords);
3670Syaroslav@ivinco.com        $starKeyword = $this->starQuery($keywords);
3780Syaroslav@ivinco.com        $this->_query = "(@(namespace,pagename) $starKeyword) | (@(body,title) {$keywords})";
3869Syaroslav@ivinco.com    }
3969Syaroslav@ivinco.com
4069Syaroslav@ivinco.com    public function setSearchAllQueryWithCategoryFilter($keywords, $categories)
4119Syaroslav@ivinco.com    {
4297Syaroslav@ivinco.com        $keywords = $this->_sphinx->EscapeString($keywords);
43*124Syaroslav@ivinco.com        $keywords = $this->_enableQuotesAndDefis($keywords);
4469Syaroslav@ivinco.com        $starKeyword = $this->starQuery($keywords);
4569Syaroslav@ivinco.com        if(strpos($categories, "-") === 0){
4669Syaroslav@ivinco.com            $categories = '-"'.substr($categories, 1).'"';
4769Syaroslav@ivinco.com        }
4880Syaroslav@ivinco.com        $this->_query = "(@(namespace,pagename) {$categories}) & ((@(body,title) {$keywords}) | (@(namespace,pagename) {$starKeyword}))";
4969Syaroslav@ivinco.com    }
5069Syaroslav@ivinco.com
5170Syaroslav@ivinco.com    public function setSearchCategoryQuery($keywords, $categories)
5269Syaroslav@ivinco.com    {
5397Syaroslav@ivinco.com        $keywords = $this->_sphinx->EscapeString($keywords);
54*124Syaroslav@ivinco.com        $keywords = $this->_enableQuotesAndDefis($keywords);
55*124Syaroslav@ivinco.com
5670Syaroslav@ivinco.com        $starKeyword = $this->starQuery($keywords);
5772Syaroslav@ivinco.com        if (!empty($categories)){
5880Syaroslav@ivinco.com            $this->_query = "(@(namespace,pagename) $categories $starKeyword)";
5972Syaroslav@ivinco.com        } else {
6080Syaroslav@ivinco.com            $this->_query = "(@(namespace,pagename) $starKeyword)";
6172Syaroslav@ivinco.com        }
62*124Syaroslav@ivinco.com
6369Syaroslav@ivinco.com    }
64*124Syaroslav@ivinco.com
65*124Syaroslav@ivinco.com    public function setSearchOnlyPagename()
66*124Syaroslav@ivinco.com    {
67*124Syaroslav@ivinco.com    	$this->_query = "(@(pagename) {$this->_query})";
68*124Syaroslav@ivinco.com    }
69*124Syaroslav@ivinco.com
7069Syaroslav@ivinco.com    public function search($start, $resultsPerPage = 10)
7169Syaroslav@ivinco.com    {
7269Syaroslav@ivinco.com        $this->_resultsPerPage = $resultsPerPage;
7369Syaroslav@ivinco.com
7480Syaroslav@ivinco.com        $this->_sphinx->SetFieldWeights(array(
7580Syaroslav@ivinco.com            'namespace' => $this->_namespacePriority,
7680Syaroslav@ivinco.com            'pagename' => $this->_pagenamePriority,
7780Syaroslav@ivinco.com            'title' => $this->_titlePriority,
7880Syaroslav@ivinco.com            'body' => $this->_bodyPriority)
7980Syaroslav@ivinco.com        );
8019Syaroslav@ivinco.com
8127Syaroslav@ivinco.com        $this->_sphinx->SetLimits($start, $resultsPerPage+100);
8269Syaroslav@ivinco.com
8369Syaroslav@ivinco.com        $res = $this->_sphinx->Query($this->_query, $this->_index);
84119Syaroslav
850Syaroslav@ivinco.com        $this->_result = $res;
8628Syaroslav@ivinco.com
870Syaroslav@ivinco.com        if (empty($res['matches'])) {
880Syaroslav@ivinco.com            return false;
890Syaroslav@ivinco.com	}
9069Syaroslav@ivinco.com        return true;
9169Syaroslav@ivinco.com    }
920Syaroslav@ivinco.com
9377Syaroslav@ivinco.com    public function getPages($keywords)
9469Syaroslav@ivinco.com    {
9592Syaroslav@ivinco.com        if (empty($this->_result['matches'])) {
9692Syaroslav@ivinco.com            return false;
9792Syaroslav@ivinco.com	}
98119Syaroslav
9969Syaroslav@ivinco.com        $pagesIdsAll = $this->getPagesIds();
10039Syaroslav@ivinco.com        $this->_offset = 0;
10139Syaroslav@ivinco.com        $counter = 0;
10243Syaroslav@ivinco.com        $tmpRes = array();
10343Syaroslav@ivinco.com        $pagesIds = array();
10439Syaroslav@ivinco.com        foreach($pagesIdsAll as $id => $pageData){
10539Syaroslav@ivinco.com            $this->_offset++;
10639Syaroslav@ivinco.com            if(auth_quickaclcheck($pageData['page']) >= AUTH_READ){
10743Syaroslav@ivinco.com                if(!isset($tmpRes[$pageData['page']])){
10843Syaroslav@ivinco.com                    $tmpRes[$pageData['page']] = 1;
10943Syaroslav@ivinco.com                    $counter++;
110*124Syaroslav@ivinco.com                }
11139Syaroslav@ivinco.com                $pagesIds[$id] = $pageData;
11269Syaroslav@ivinco.com                if ($counter == $this->_resultsPerPage){
11339Syaroslav@ivinco.com                    break;
11439Syaroslav@ivinco.com                }
115*124Syaroslav@ivinco.com            }
116*124Syaroslav@ivinco.com        }
117*124Syaroslav@ivinco.com        if (empty($pagesIds)){
11827Syaroslav@ivinco.com            return false;
11927Syaroslav@ivinco.com        }
1200Syaroslav@ivinco.com
1210Syaroslav@ivinco.com        $pagesList = array();
12210Syaroslav@ivinco.com        $body = array();
12319Syaroslav@ivinco.com        $titleText = array();
12410Syaroslav@ivinco.com        $category = array();
12527Syaroslav@ivinco.com        foreach ($pagesIds as $crc => $data){
126*124Syaroslav@ivinco.com            if (empty($data['page'])){
127*124Syaroslav@ivinco.com                continue;
128*124Syaroslav@ivinco.com            }
12927Syaroslav@ivinco.com            if (!empty($data['hid'])){
13027Syaroslav@ivinco.com                $bodyHtml = p_render('xhtml',p_get_instructions(getSectionByTitleLevel($data['page'], $data['title'], true)),$info);
1316Syaroslav@ivinco.com            } else {
13227Syaroslav@ivinco.com                $bodyHtml = p_wiki_xhtml($data['page']);
1338Syaroslav@ivinco.com            }
13460Syaroslav@ivinco.com            $bodyHtml = preg_replace("#[\s]+?</li>#", "</li>;", $bodyHtml);
13557Syaroslav@ivinco.com            $bodyHtml = htmlspecialchars_decode($bodyHtml);
13610Syaroslav@ivinco.com            $body[$crc] = strip_tags($bodyHtml);
13779Syaroslav@ivinco.com            if(!empty($data['title_text'])){
13879Syaroslav@ivinco.com                $titleText[$crc] = strip_tags($data['title_text']);
13979Syaroslav@ivinco.com            } else {
14079Syaroslav@ivinco.com                $titleText[$crc] = $data['page'];
14179Syaroslav@ivinco.com            }
14227Syaroslav@ivinco.com            $category[$crc] = $data['page'];
14357Syaroslav@ivinco.com        }
144*124Syaroslav@ivinco.com
14580Syaroslav@ivinco.com        //$starQuery = $this->starQuery($keywords);
14680Syaroslav@ivinco.com        $bodyExcerpt = $this->getExcerpt($body, $keywords);
14780Syaroslav@ivinco.com        $titleTextExcerpt = $this->getExcerpt($titleText, $keywords);
1480Syaroslav@ivinco.com        $i = 0;
1494Syaroslav@ivinco.com        $results = array();
15010Syaroslav@ivinco.com        foreach($body as $crc => $notused){
15110Syaroslav@ivinco.com            $results[$crc] = array(
15210Syaroslav@ivinco.com                'page' => $pagesIds[$crc]['page'],
15310Syaroslav@ivinco.com                'bodyExcerpt' => $bodyExcerpt[$i],
15419Syaroslav@ivinco.com                'titleTextExcerpt' => $titleTextExcerpt[$i],
15510Syaroslav@ivinco.com                'hid' => $pagesIds[$crc]['hid'],
15619Syaroslav@ivinco.com                'title' => $pagesIds[$crc]['title'],
15719Syaroslav@ivinco.com                'title_text' => $pagesIds[$crc]['title_text']
15810Syaroslav@ivinco.com            );
15910Syaroslav@ivinco.com            $i++;
1600Syaroslav@ivinco.com        }
1614Syaroslav@ivinco.com        return $results;
1620Syaroslav@ivinco.com    }
1630Syaroslav@ivinco.com
16469Syaroslav@ivinco.com    public function getPagesIds()
16569Syaroslav@ivinco.com    {
16669Syaroslav@ivinco.com        $pageMapper = new PageMapper();
16769Syaroslav@ivinco.com
16869Syaroslav@ivinco.com        return $pageMapper->getByCrc(array_keys($this->_result['matches']));
16969Syaroslav@ivinco.com    }
17069Syaroslav@ivinco.com
17139Syaroslav@ivinco.com    public function getOffset()
17239Syaroslav@ivinco.com    {
17339Syaroslav@ivinco.com        return $this->_offset;
17439Syaroslav@ivinco.com    }
17539Syaroslav@ivinco.com
17628Syaroslav@ivinco.com    public function getError()
17728Syaroslav@ivinco.com    {
17828Syaroslav@ivinco.com        return $this->_sphinx->GetLastError();
17928Syaroslav@ivinco.com    }
18028Syaroslav@ivinco.com
1810Syaroslav@ivinco.com    public function getTotalFound()
1820Syaroslav@ivinco.com    {
1830Syaroslav@ivinco.com        return !empty($this->_result['total_found'])?$this->_result['total_found'] : 0;
1840Syaroslav@ivinco.com    }
18510Syaroslav@ivinco.com
18610Syaroslav@ivinco.com    public function getExcerpt($data, $query)
18710Syaroslav@ivinco.com    {
18816Syaroslav@ivinco.com        return $this->_sphinx->BuildExcerpts($data, $this->_index, $query,
18916Syaroslav@ivinco.com                    array(
19016Syaroslav@ivinco.com                        'limit' => $this->_snippetSize,
19139Syaroslav@ivinco.com                        'around' => $this->_aroundKeyword,
19239Syaroslav@ivinco.com                        'weight_order' => 1,
19339Syaroslav@ivinco.com                        'sp' => 1
19416Syaroslav@ivinco.com                    )
19516Syaroslav@ivinco.com                );
19610Syaroslav@ivinco.com    }
19710Syaroslav@ivinco.com
19810Syaroslav@ivinco.com    public function starQuery($query)
19910Syaroslav@ivinco.com    {
20061Syaroslav@ivinco.com        $query = $this->removeStars($query);
20110Syaroslav@ivinco.com        $words = explode(" ", $query);
20261Syaroslav@ivinco.com        foreach($words as $id => $word){
20361Syaroslav@ivinco.com            $words[$id] = "*".$word."*";
20410Syaroslav@ivinco.com        }
20561Syaroslav@ivinco.com        return implode(" ", $words);
20661Syaroslav@ivinco.com    }
20761Syaroslav@ivinco.com
20861Syaroslav@ivinco.com    public function removeStars($query)
20961Syaroslav@ivinco.com    {
21061Syaroslav@ivinco.com        $words = explode(" ", $query);
21161Syaroslav@ivinco.com        foreach($words as $id => $word){
21261Syaroslav@ivinco.com            $words[$id] = trim($word, "*");
21361Syaroslav@ivinco.com        }
21461Syaroslav@ivinco.com        return implode(" ", $words);
21510Syaroslav@ivinco.com    }
21615Syaroslav@ivinco.com
21715Syaroslav@ivinco.com    public function getQuery()
21815Syaroslav@ivinco.com    {
21915Syaroslav@ivinco.com        return $this->_query;
22015Syaroslav@ivinco.com    }
22116Syaroslav@ivinco.com
22216Syaroslav@ivinco.com    public function setSnippetSize($symbols = 256)
22316Syaroslav@ivinco.com    {
22416Syaroslav@ivinco.com        $this->_snippetSize = $symbols;
22516Syaroslav@ivinco.com    }
22616Syaroslav@ivinco.com
22716Syaroslav@ivinco.com    public function setArroundWordsCount($words = 5)
22816Syaroslav@ivinco.com    {
22916Syaroslav@ivinco.com        $this->_aroundKeyword = $words;
23016Syaroslav@ivinco.com    }
23119Syaroslav@ivinco.com
23219Syaroslav@ivinco.com    public function setTitlePriority($priority)
23319Syaroslav@ivinco.com    {
23419Syaroslav@ivinco.com        $this->_titlePriority = $priority;
23519Syaroslav@ivinco.com    }
23619Syaroslav@ivinco.com
23719Syaroslav@ivinco.com    public function setBodyPriority($priority)
23819Syaroslav@ivinco.com    {
23919Syaroslav@ivinco.com        $this->_bodyPriority = $priority;
24019Syaroslav@ivinco.com    }
24119Syaroslav@ivinco.com
24280Syaroslav@ivinco.com    public function setNamespacePriority($priority)
24319Syaroslav@ivinco.com    {
24480Syaroslav@ivinco.com        $this->_namespacePriority = $priority;
24580Syaroslav@ivinco.com    }
24680Syaroslav@ivinco.com
24780Syaroslav@ivinco.com    public function setPagenamePriority($priority)
24880Syaroslav@ivinco.com    {
24980Syaroslav@ivinco.com        $this->_pagenamePriority = $priority;
25019Syaroslav@ivinco.com    }
251*124Syaroslav@ivinco.com
252*124Syaroslav@ivinco.com    private function _enableQuotesAndDefis($query)
253*124Syaroslav@ivinco.com    {
254*124Syaroslav@ivinco.com        $query = ' '. $query;
255*124Syaroslav@ivinco.com        $quotesCount = count(explode('"', $query))-1;
256*124Syaroslav@ivinco.com        if ($quotesCount && $quotesCount%2 == 0){
257*124Syaroslav@ivinco.com            $query = str_replace('\"', '"', $query);
258*124Syaroslav@ivinco.com        }
259*124Syaroslav@ivinco.com        $query = preg_replace("#\s\\\-(\w)#ui", " -$1", $query);
260*124Syaroslav@ivinco.com
261*124Syaroslav@ivinco.com        $query = substr($query, 1);
262*124Syaroslav@ivinco.com
263*124Syaroslav@ivinco.com        return $query;
264*124Syaroslav@ivinco.com    }
2650Syaroslav@ivinco.com}
266