10Syaroslav@ivinco.com<?php
2124Syaroslav@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 $_pagenamePriority = 1;
22124Syaroslav@ivinco.com
230Syaroslav@ivinco.com    public function  __construct($host, $port, $index)
240Syaroslav@ivinco.com    {
250Syaroslav@ivinco.com        $this->_sphinx = new SphinxClient();
260Syaroslav@ivinco.com        $this->_sphinx->SetServer($host, $port);
27124Syaroslav@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);
35124Syaroslav@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);
43124Syaroslav@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);
54124Syaroslav@ivinco.com        $keywords = $this->_enableQuotesAndDefis($keywords);
55124Syaroslav@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        }
62124Syaroslav@ivinco.com
6369Syaroslav@ivinco.com    }
64124Syaroslav@ivinco.com
65124Syaroslav@ivinco.com    public function setSearchOnlyPagename()
66124Syaroslav@ivinco.com    {
67124Syaroslav@ivinco.com    	$this->_query = "(@(pagename) {$this->_query})";
68124Syaroslav@ivinco.com    }
69124Syaroslav@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
81133Sandrey        $this->_sphinx->SetLimits($start, $resultsPerPage+100, 1000);
8269Syaroslav@ivinco.com
83133Sandrey        $this->_result = $this->_sphinx->Query($this->_query, $this->_index);
84119Syaroslav
85133Sandrey        if (empty($this->_result['matches'])) {
860Syaroslav@ivinco.com            return false;
87133Sandrey	    }
8869Syaroslav@ivinco.com        return true;
8969Syaroslav@ivinco.com    }
900Syaroslav@ivinco.com
9177Syaroslav@ivinco.com    public function getPages($keywords)
9269Syaroslav@ivinco.com    {
9392Syaroslav@ivinco.com        if (empty($this->_result['matches'])) {
9492Syaroslav@ivinco.com            return false;
95133Sandrey	    }
96119Syaroslav
9769Syaroslav@ivinco.com        $pagesIdsAll = $this->getPagesIds();
9839Syaroslav@ivinco.com        $this->_offset = 0;
9939Syaroslav@ivinco.com        $counter = 0;
10043Syaroslav@ivinco.com        $tmpRes = array();
10143Syaroslav@ivinco.com        $pagesIds = array();
10239Syaroslav@ivinco.com        foreach($pagesIdsAll as $id => $pageData){
10339Syaroslav@ivinco.com            $this->_offset++;
10439Syaroslav@ivinco.com            if(auth_quickaclcheck($pageData['page']) >= AUTH_READ){
10543Syaroslav@ivinco.com                if(!isset($tmpRes[$pageData['page']])){
10643Syaroslav@ivinco.com                    $tmpRes[$pageData['page']] = 1;
10743Syaroslav@ivinco.com                    $counter++;
108124Syaroslav@ivinco.com                }
10939Syaroslav@ivinco.com                $pagesIds[$id] = $pageData;
11069Syaroslav@ivinco.com                if ($counter == $this->_resultsPerPage){
11139Syaroslav@ivinco.com                    break;
11239Syaroslav@ivinco.com                }
113124Syaroslav@ivinco.com            }
114124Syaroslav@ivinco.com        }
115124Syaroslav@ivinco.com        if (empty($pagesIds)){
11627Syaroslav@ivinco.com            return false;
11727Syaroslav@ivinco.com        }
1180Syaroslav@ivinco.com
1190Syaroslav@ivinco.com        $pagesList = array();
12010Syaroslav@ivinco.com        $body = array();
12119Syaroslav@ivinco.com        $titleText = array();
12210Syaroslav@ivinco.com        $category = array();
12327Syaroslav@ivinco.com        foreach ($pagesIds as $crc => $data){
124124Syaroslav@ivinco.com            if (empty($data['page'])){
125124Syaroslav@ivinco.com                continue;
126124Syaroslav@ivinco.com            }
12727Syaroslav@ivinco.com            if (!empty($data['hid'])){
12827Syaroslav@ivinco.com                $bodyHtml = p_render('xhtml',p_get_instructions(getSectionByTitleLevel($data['page'], $data['title'], true)),$info);
1296Syaroslav@ivinco.com            } else {
13027Syaroslav@ivinco.com                $bodyHtml = p_wiki_xhtml($data['page']);
1318Syaroslav@ivinco.com            }
13260Syaroslav@ivinco.com            $bodyHtml = preg_replace("#[\s]+?</li>#", "</li>;", $bodyHtml);
13357Syaroslav@ivinco.com            $bodyHtml = htmlspecialchars_decode($bodyHtml);
13410Syaroslav@ivinco.com            $body[$crc] = strip_tags($bodyHtml);
13579Syaroslav@ivinco.com            if(!empty($data['title_text'])){
13679Syaroslav@ivinco.com                $titleText[$crc] = strip_tags($data['title_text']);
13779Syaroslav@ivinco.com            } else {
13879Syaroslav@ivinco.com                $titleText[$crc] = $data['page'];
13979Syaroslav@ivinco.com            }
14027Syaroslav@ivinco.com            $category[$crc] = $data['page'];
14157Syaroslav@ivinco.com        }
142*134Sandrey
14380Syaroslav@ivinco.com        //$starQuery = $this->starQuery($keywords);
14480Syaroslav@ivinco.com        $bodyExcerpt = $this->getExcerpt($body, $keywords);
14580Syaroslav@ivinco.com        $titleTextExcerpt = $this->getExcerpt($titleText, $keywords);
1460Syaroslav@ivinco.com        $i = 0;
1474Syaroslav@ivinco.com        $results = array();
14810Syaroslav@ivinco.com        foreach($body as $crc => $notused){
14910Syaroslav@ivinco.com            $results[$crc] = array(
15010Syaroslav@ivinco.com                'page' => $pagesIds[$crc]['page'],
15110Syaroslav@ivinco.com                'bodyExcerpt' => $bodyExcerpt[$i],
15219Syaroslav@ivinco.com                'titleTextExcerpt' => $titleTextExcerpt[$i],
15310Syaroslav@ivinco.com                'hid' => $pagesIds[$crc]['hid'],
15419Syaroslav@ivinco.com                'title' => $pagesIds[$crc]['title'],
15519Syaroslav@ivinco.com                'title_text' => $pagesIds[$crc]['title_text']
15610Syaroslav@ivinco.com            );
15710Syaroslav@ivinco.com            $i++;
1580Syaroslav@ivinco.com        }
1594Syaroslav@ivinco.com        return $results;
1600Syaroslav@ivinco.com    }
1610Syaroslav@ivinco.com
16269Syaroslav@ivinco.com    public function getPagesIds()
16369Syaroslav@ivinco.com    {
16469Syaroslav@ivinco.com        $pageMapper = new PageMapper();
16569Syaroslav@ivinco.com
16669Syaroslav@ivinco.com        return $pageMapper->getByCrc(array_keys($this->_result['matches']));
16769Syaroslav@ivinco.com    }
16869Syaroslav@ivinco.com
16939Syaroslav@ivinco.com    public function getOffset()
17039Syaroslav@ivinco.com    {
17139Syaroslav@ivinco.com        return $this->_offset;
17239Syaroslav@ivinco.com    }
17339Syaroslav@ivinco.com
174*134Sandrey    public function getPageNames()
175*134Sandrey    {
176*134Sandrey        $pageIds = $this->getPagesIds();
177*134Sandrey
178*134Sandrey        $matchPages = array();
179*134Sandrey        foreach ($pageIds as $page) {
180*134Sandrey            if(auth_quickaclcheck($page['page']) < AUTH_READ){
181*134Sandrey                continue;
182*134Sandrey            }
183*134Sandrey            /*if(auth_quickaclcheck($pageData['page']) >= AUTH_READ){
184*134Sandrey                if(!isset($tmpRes[$pageData['page']])){
185*134Sandrey                    $tmpRes[$pageData['page']] = 1;
186*134Sandrey                    $counter++;
187*134Sandrey                }
188*134Sandrey                $pagesIds[$id] = $pageData;
189*134Sandrey                if ($counter == $this->_resultsPerPage){
190*134Sandrey                    break;
191*134Sandrey                }
192*134Sandrey            }*/
193*134Sandrey
194*134Sandrey            $matchPages[$page['page']] = $page['hid'];
195*134Sandrey        }
196*134Sandrey
197*134Sandrey        return $matchPages;
198*134Sandrey    }
199*134Sandrey
20028Syaroslav@ivinco.com    public function getError()
20128Syaroslav@ivinco.com    {
20228Syaroslav@ivinco.com        return $this->_sphinx->GetLastError();
20328Syaroslav@ivinco.com    }
20428Syaroslav@ivinco.com
2050Syaroslav@ivinco.com    public function getTotalFound()
2060Syaroslav@ivinco.com    {
2070Syaroslav@ivinco.com        return !empty($this->_result['total_found'])?$this->_result['total_found'] : 0;
2080Syaroslav@ivinco.com    }
20910Syaroslav@ivinco.com
21010Syaroslav@ivinco.com    public function getExcerpt($data, $query)
21110Syaroslav@ivinco.com    {
21216Syaroslav@ivinco.com        return $this->_sphinx->BuildExcerpts($data, $this->_index, $query,
21316Syaroslav@ivinco.com                    array(
21416Syaroslav@ivinco.com                        'limit' => $this->_snippetSize,
21539Syaroslav@ivinco.com                        'around' => $this->_aroundKeyword,
21639Syaroslav@ivinco.com                        'weight_order' => 1,
21739Syaroslav@ivinco.com                        'sp' => 1
21816Syaroslav@ivinco.com                    )
21916Syaroslav@ivinco.com                );
22010Syaroslav@ivinco.com    }
22110Syaroslav@ivinco.com
22210Syaroslav@ivinco.com    public function starQuery($query)
22310Syaroslav@ivinco.com    {
22461Syaroslav@ivinco.com        $query = $this->removeStars($query);
22510Syaroslav@ivinco.com        $words = explode(" ", $query);
22661Syaroslav@ivinco.com        foreach($words as $id => $word){
22761Syaroslav@ivinco.com            $words[$id] = "*".$word."*";
22810Syaroslav@ivinco.com        }
22961Syaroslav@ivinco.com        return implode(" ", $words);
23061Syaroslav@ivinco.com    }
23161Syaroslav@ivinco.com
23261Syaroslav@ivinco.com    public function removeStars($query)
23361Syaroslav@ivinco.com    {
23461Syaroslav@ivinco.com        $words = explode(" ", $query);
23561Syaroslav@ivinco.com        foreach($words as $id => $word){
23661Syaroslav@ivinco.com            $words[$id] = trim($word, "*");
23761Syaroslav@ivinco.com        }
23861Syaroslav@ivinco.com        return implode(" ", $words);
23910Syaroslav@ivinco.com    }
24015Syaroslav@ivinco.com
24115Syaroslav@ivinco.com    public function getQuery()
24215Syaroslav@ivinco.com    {
24315Syaroslav@ivinco.com        return $this->_query;
24415Syaroslav@ivinco.com    }
24516Syaroslav@ivinco.com
24616Syaroslav@ivinco.com    public function setSnippetSize($symbols = 256)
24716Syaroslav@ivinco.com    {
24816Syaroslav@ivinco.com        $this->_snippetSize = $symbols;
24916Syaroslav@ivinco.com    }
25016Syaroslav@ivinco.com
25116Syaroslav@ivinco.com    public function setArroundWordsCount($words = 5)
25216Syaroslav@ivinco.com    {
25316Syaroslav@ivinco.com        $this->_aroundKeyword = $words;
25416Syaroslav@ivinco.com    }
25519Syaroslav@ivinco.com
25619Syaroslav@ivinco.com    public function setTitlePriority($priority)
25719Syaroslav@ivinco.com    {
25819Syaroslav@ivinco.com        $this->_titlePriority = $priority;
25919Syaroslav@ivinco.com    }
26019Syaroslav@ivinco.com
26119Syaroslav@ivinco.com    public function setBodyPriority($priority)
26219Syaroslav@ivinco.com    {
26319Syaroslav@ivinco.com        $this->_bodyPriority = $priority;
26419Syaroslav@ivinco.com    }
26519Syaroslav@ivinco.com
26680Syaroslav@ivinco.com    public function setNamespacePriority($priority)
26719Syaroslav@ivinco.com    {
26880Syaroslav@ivinco.com        $this->_namespacePriority = $priority;
26980Syaroslav@ivinco.com    }
27080Syaroslav@ivinco.com
27180Syaroslav@ivinco.com    public function setPagenamePriority($priority)
27280Syaroslav@ivinco.com    {
27380Syaroslav@ivinco.com        $this->_pagenamePriority = $priority;
27419Syaroslav@ivinco.com    }
275124Syaroslav@ivinco.com
276124Syaroslav@ivinco.com    private function _enableQuotesAndDefis($query)
277124Syaroslav@ivinco.com    {
278124Syaroslav@ivinco.com        $query = ' '. $query;
279124Syaroslav@ivinco.com        $quotesCount = count(explode('"', $query))-1;
280124Syaroslav@ivinco.com        if ($quotesCount && $quotesCount%2 == 0){
281124Syaroslav@ivinco.com            $query = str_replace('\"', '"', $query);
282124Syaroslav@ivinco.com        }
283124Syaroslav@ivinco.com        $query = preg_replace("#\s\\\-(\w)#ui", " -$1", $query);
284124Syaroslav@ivinco.com
285124Syaroslav@ivinco.com        $query = substr($query, 1);
286124Syaroslav@ivinco.com
287124Syaroslav@ivinco.com        return $query;
288124Syaroslav@ivinco.com    }
2890Syaroslav@ivinco.com}
290