1f5eb7cf0SAndreas Gohr<?php 2f5eb7cf0SAndreas Gohr/** 3f5eb7cf0SAndreas Gohr * DokuWiki fulltextsearch functions using the index 4f5eb7cf0SAndreas Gohr * 5f5eb7cf0SAndreas Gohr * @license GPL 2 (http://www.gnu.org/licenses/gpl.html) 6f5eb7cf0SAndreas Gohr * @author Andreas Gohr <andi@splitbrain.org> 7f5eb7cf0SAndreas Gohr */ 8f5eb7cf0SAndreas Gohr 9fa8adffeSAndreas Gohrif(!defined('DOKU_INC')) die('meh.'); 10f5eb7cf0SAndreas Gohr 11bd0293e7SAndreas Gohr/** 12bd0293e7SAndreas Gohr * create snippets for the first few results only 13bd0293e7SAndreas Gohr */ 14bd0293e7SAndreas Gohrif(!defined('FT_SNIPPET_NUMBER')) define('FT_SNIPPET_NUMBER',15); 15f5eb7cf0SAndreas Gohr 16f5eb7cf0SAndreas Gohr/** 17f5eb7cf0SAndreas Gohr * The fulltext search 18f5eb7cf0SAndreas Gohr * 19f5eb7cf0SAndreas Gohr * Returns a list of matching documents for the given query 20506fa893SAndreas Gohr * 216840140fSChris Smith * refactored into ft_pageSearch(), _ft_pageSearch() and trigger_event() 226840140fSChris Smith * 23f5eb7cf0SAndreas Gohr */ 24546d3a99SAndreas Gohrfunction ft_pageSearch($query,&$highlight){ 256840140fSChris Smith 266840140fSChris Smith $data['query'] = $query; 276840140fSChris Smith $data['highlight'] =& $highlight; 286840140fSChris Smith 296840140fSChris Smith return trigger_event('SEARCH_QUERY_FULLPAGE', $data, '_ft_pageSearch'); 306840140fSChris Smith} 31865c2687SKazutaka Miyasaka 32865c2687SKazutaka Miyasaka/** 33865c2687SKazutaka Miyasaka * Returns a list of matching documents for the given query 34865c2687SKazutaka Miyasaka * 35865c2687SKazutaka Miyasaka * @author Andreas Gohr <andi@splitbrain.org> 36865c2687SKazutaka Miyasaka * @author Kazutaka Miyasaka <kazmiya@gmail.com> 37865c2687SKazutaka Miyasaka */ 386840140fSChris Smithfunction _ft_pageSearch(&$data) { 399b41be24STom N Harris $Indexer = idx_get_indexer(); 409b41be24STom N Harris 41865c2687SKazutaka Miyasaka // parse the given query 429b41be24STom N Harris $q = ft_queryParser($Indexer, $data['query']); 43865c2687SKazutaka Miyasaka $data['highlight'] = $q['highlight']; 446840140fSChris Smith 45865c2687SKazutaka Miyasaka if (empty($q['parsed_ary'])) return array(); 46506fa893SAndreas Gohr 47f5eb7cf0SAndreas Gohr // lookup all words found in the query 489b41be24STom N Harris $lookup = $Indexer->lookup($q['words']); 49f5eb7cf0SAndreas Gohr 50865c2687SKazutaka Miyasaka // get all pages in this dokuwiki site (!: includes nonexistent pages) 51865c2687SKazutaka Miyasaka $pages_all = array(); 529b41be24STom N Harris foreach ($Indexer->getPages() as $id) { 539b41be24STom N Harris $pages_all[$id] = 0; // base: 0 hit 54f5eb7cf0SAndreas Gohr } 55f5eb7cf0SAndreas Gohr 56865c2687SKazutaka Miyasaka // process the query 57865c2687SKazutaka Miyasaka $stack = array(); 58865c2687SKazutaka Miyasaka foreach ($q['parsed_ary'] as $token) { 59865c2687SKazutaka Miyasaka switch (substr($token, 0, 3)) { 60865c2687SKazutaka Miyasaka case 'W+:': 612f502d70SKazutaka Miyasaka case 'W-:': 622f502d70SKazutaka Miyasaka case 'W_:': // word 63865c2687SKazutaka Miyasaka $word = substr($token, 3); 64865c2687SKazutaka Miyasaka $stack[] = (array) $lookup[$word]; 65865c2687SKazutaka Miyasaka break; 662f502d70SKazutaka Miyasaka case 'P+:': 672f502d70SKazutaka Miyasaka case 'P-:': // phrase 68865c2687SKazutaka Miyasaka $phrase = substr($token, 3); 69865c2687SKazutaka Miyasaka // since phrases are always parsed as ((W1)(W2)...(P)), 70865c2687SKazutaka Miyasaka // the end($stack) always points the pages that contain 71865c2687SKazutaka Miyasaka // all words in this phrase 72865c2687SKazutaka Miyasaka $pages = end($stack); 73865c2687SKazutaka Miyasaka $pages_matched = array(); 74865c2687SKazutaka Miyasaka foreach(array_keys($pages) as $id){ 75f5eb7cf0SAndreas Gohr $text = utf8_strtolower(rawWiki($id)); 76865c2687SKazutaka Miyasaka if (strpos($text, $phrase) !== false) { 77865c2687SKazutaka Miyasaka $pages_matched[$id] = 0; // phrase: always 0 hit 78865c2687SKazutaka Miyasaka } 79865c2687SKazutaka Miyasaka } 80865c2687SKazutaka Miyasaka $stack[] = $pages_matched; 81865c2687SKazutaka Miyasaka break; 822f502d70SKazutaka Miyasaka case 'N+:': 832f502d70SKazutaka Miyasaka case 'N-:': // namespace 84865c2687SKazutaka Miyasaka $ns = substr($token, 3); 85865c2687SKazutaka Miyasaka $pages_matched = array(); 86865c2687SKazutaka Miyasaka foreach (array_keys($pages_all) as $id) { 87865c2687SKazutaka Miyasaka if (strpos($id, $ns) === 0) { 88865c2687SKazutaka Miyasaka $pages_matched[$id] = 0; // namespace: always 0 hit 89865c2687SKazutaka Miyasaka } 90865c2687SKazutaka Miyasaka } 91865c2687SKazutaka Miyasaka $stack[] = $pages_matched; 92865c2687SKazutaka Miyasaka break; 93865c2687SKazutaka Miyasaka case 'AND': // and operation 94865c2687SKazutaka Miyasaka list($pages1, $pages2) = array_splice($stack, -2); 95865c2687SKazutaka Miyasaka $stack[] = ft_resultCombine(array($pages1, $pages2)); 96865c2687SKazutaka Miyasaka break; 97865c2687SKazutaka Miyasaka case 'OR': // or operation 98865c2687SKazutaka Miyasaka list($pages1, $pages2) = array_splice($stack, -2); 99865c2687SKazutaka Miyasaka $stack[] = ft_resultUnite(array($pages1, $pages2)); 100865c2687SKazutaka Miyasaka break; 101865c2687SKazutaka Miyasaka case 'NOT': // not operation (unary) 102865c2687SKazutaka Miyasaka $pages = array_pop($stack); 103865c2687SKazutaka Miyasaka $stack[] = ft_resultComplement(array($pages_all, $pages)); 104a21136cdSAndreas Gohr break; 105a21136cdSAndreas Gohr } 106f5eb7cf0SAndreas Gohr } 107865c2687SKazutaka Miyasaka $docs = array_pop($stack); 108865c2687SKazutaka Miyasaka 109865c2687SKazutaka Miyasaka if (empty($docs)) return array(); 110865c2687SKazutaka Miyasaka 111865c2687SKazutaka Miyasaka // check: settings, acls, existence 112865c2687SKazutaka Miyasaka foreach (array_keys($docs) as $id) { 113865c2687SKazutaka Miyasaka if (isHiddenPage($id) || auth_quickaclcheck($id) < AUTH_READ || !page_exists($id, '', false)) { 114865c2687SKazutaka Miyasaka unset($docs[$id]); 115f5eb7cf0SAndreas Gohr } 116f5eb7cf0SAndreas Gohr } 117f5eb7cf0SAndreas Gohr 118865c2687SKazutaka Miyasaka // sort docs by count 119f5eb7cf0SAndreas Gohr arsort($docs); 120f5eb7cf0SAndreas Gohr 121f5eb7cf0SAndreas Gohr return $docs; 122f5eb7cf0SAndreas Gohr} 123f5eb7cf0SAndreas Gohr 124f5eb7cf0SAndreas Gohr/** 12554f4c056SAndreas Gohr * Returns the backlinks for a given page 12654f4c056SAndreas Gohr * 127320f489aSMichael Hamann * Uses the metadata index. 12854f4c056SAndreas Gohr */ 12954f4c056SAndreas Gohrfunction ft_backlinks($id){ 13054f4c056SAndreas Gohr $result = array(); 13154f4c056SAndreas Gohr 132320f489aSMichael Hamann $result = idx_get_indexer()->lookupKey('relation_references', $id); 13354f4c056SAndreas Gohr 13463773904SAndreas Gohr if(!count($result)) return $result; 13563773904SAndreas Gohr 13663773904SAndreas Gohr // check ACL permissions 13763773904SAndreas Gohr foreach(array_keys($result) as $idx){ 13863773904SAndreas Gohr if(auth_quickaclcheck($result[$idx]) < AUTH_READ){ 13963773904SAndreas Gohr unset($result[$idx]); 14063773904SAndreas Gohr } 14163773904SAndreas Gohr } 14263773904SAndreas Gohr 14354f4c056SAndreas Gohr sort($result); 14454f4c056SAndreas Gohr return $result; 14554f4c056SAndreas Gohr} 14654f4c056SAndreas Gohr 14754f4c056SAndreas Gohr/** 148a05e297aSAndreas Gohr * Returns the pages that use a given media file 149a05e297aSAndreas Gohr * 150a05e297aSAndreas Gohr * Does a quick lookup with the fulltext index, then 151a05e297aSAndreas Gohr * evaluates the instructions of the found pages 152a05e297aSAndreas Gohr * 153a05e297aSAndreas Gohr * Aborts after $max found results 154a05e297aSAndreas Gohr */ 155a05e297aSAndreas Gohrfunction ft_mediause($id,$max){ 156a05e297aSAndreas Gohr if(!$max) $max = 1; // need to find at least one 157a05e297aSAndreas Gohr 158a05e297aSAndreas Gohr $result = array(); 159a05e297aSAndreas Gohr 160a05e297aSAndreas Gohr // quick lookup of the mediafile 1619b41be24STom N Harris // FIXME use metadata key lookup 162a05e297aSAndreas Gohr $media = noNS($id); 1639b41be24STom N Harris $matches = idx_lookup(idx_tokenizer($media)); 164a05e297aSAndreas Gohr $docs = array_keys(ft_resultCombine(array_values($matches))); 165a05e297aSAndreas Gohr if(!count($docs)) return $result; 166a05e297aSAndreas Gohr 167a05e297aSAndreas Gohr // go through all found pages 168a05e297aSAndreas Gohr $found = 0; 169a05e297aSAndreas Gohr $pcre = preg_quote($media,'/'); 170a05e297aSAndreas Gohr foreach($docs as $doc){ 171a05e297aSAndreas Gohr $ns = getNS($doc); 172a05e297aSAndreas Gohr preg_match_all('/\{\{([^|}]*'.$pcre.'[^|}]*)(|[^}]+)?\}\}/i',rawWiki($doc),$matches); 173a05e297aSAndreas Gohr foreach($matches[1] as $img){ 174a05e297aSAndreas Gohr $img = trim($img); 175a05e297aSAndreas Gohr if(preg_match('/^https?:\/\//i',$img)) continue; // skip external images 176a05e297aSAndreas Gohr list($img) = explode('?',$img); // remove any parameters 177a05e297aSAndreas Gohr resolve_mediaid($ns,$img,$exists); // resolve the possibly relative img 178a05e297aSAndreas Gohr 179a05e297aSAndreas Gohr if($img == $id){ // we have a match 180a05e297aSAndreas Gohr $result[] = $doc; 181a05e297aSAndreas Gohr $found++; 182a05e297aSAndreas Gohr break; 183a05e297aSAndreas Gohr } 184a05e297aSAndreas Gohr } 185a05e297aSAndreas Gohr if($found >= $max) break; 186a05e297aSAndreas Gohr } 187a05e297aSAndreas Gohr 188a05e297aSAndreas Gohr sort($result); 189a05e297aSAndreas Gohr return $result; 190a05e297aSAndreas Gohr} 191a05e297aSAndreas Gohr 192a05e297aSAndreas Gohr 193a05e297aSAndreas Gohr 194a05e297aSAndreas Gohr/** 195506fa893SAndreas Gohr * Quicksearch for pagenames 196506fa893SAndreas Gohr * 197506fa893SAndreas Gohr * By default it only matches the pagename and ignores the 19880423ab6SAdrian Lang * namespace. This can be changed with the second parameter. 19980423ab6SAdrian Lang * The third parameter allows to search in titles as well. 200506fa893SAndreas Gohr * 2018d22f1e9SAndreas Gohr * The function always returns titles as well 2026840140fSChris Smith * 2038d22f1e9SAndreas Gohr * @triggers SEARCH_QUERY_PAGELOOKUP 204506fa893SAndreas Gohr * @author Andreas Gohr <andi@splitbrain.org> 2058d22f1e9SAndreas Gohr * @author Adrian Lang <lang@cosmocode.de> 206506fa893SAndreas Gohr */ 2078d22f1e9SAndreas Gohrfunction ft_pageLookup($id, $in_ns=false, $in_title=false){ 2088d22f1e9SAndreas Gohr $data = compact('id', 'in_ns', 'in_title'); 2098d22f1e9SAndreas Gohr $data['has_titles'] = true; // for plugin backward compatibility check 2106840140fSChris Smith return trigger_event('SEARCH_QUERY_PAGELOOKUP', $data, '_ft_pageLookup'); 2116840140fSChris Smith} 2126840140fSChris Smith 2136840140fSChris Smithfunction _ft_pageLookup(&$data){ 21480423ab6SAdrian Lang // split out original parameters 2156840140fSChris Smith $id = $data['id']; 216b0f6db0cSAdrian Lang if (preg_match('/(?:^| )@(\w+)/', $id, $matches)) { 217b0f6db0cSAdrian Lang $ns = cleanID($matches[1]) . ':'; 218b0f6db0cSAdrian Lang $id = str_replace($matches[0], '', $id); 219b0f6db0cSAdrian Lang } 220b0f6db0cSAdrian Lang 2218d22f1e9SAndreas Gohr $in_ns = $data['in_ns']; 2228d22f1e9SAndreas Gohr $in_title = $data['in_title']; 22380423ab6SAdrian Lang $cleaned = cleanID($id); 2249b41be24STom N Harris 2259b41be24STom N Harris $Indexer = idx_get_indexer(); 2269b41be24STom N Harris $page_idx = $Indexer->getPages(); 2279b41be24STom N Harris 2289b41be24STom N Harris $pages = array(); 2295479a8c3SAndreas Gohr if ($id !== '' && $cleaned !== '') { 2309b41be24STom N Harris foreach ($page_idx as $p_id) { 2319b41be24STom N Harris if ((strpos($in_ns ? $p_id : noNSorNS($p_id), $cleaned) !== false)) { 2329b41be24STom N Harris if (!isset($pages[$p_id])) 2339b41be24STom N Harris $pages[$p_id] = p_get_first_heading($p_id, false); 234506fa893SAndreas Gohr } 235506fa893SAndreas Gohr } 236f078bb00STom N Harris if ($in_title) { 2377233c152SMichael Hamann $wildcard_id = "*$id*"; 2387233c152SMichael Hamann foreach ($Indexer->lookupKey('title', $wildcard_id) as $p_id) { 239f078bb00STom N Harris if (!isset($pages[$p_id])) 240f078bb00STom N Harris $pages[$p_id] = p_get_first_heading($p_id, false); 241f078bb00STom N Harris } 242f078bb00STom N Harris } 243d0bdf765SAdrian Lang } 244*0c074a52SMichael Hamann 245d0bdf765SAdrian Lang if (isset($ns)) { 246*0c074a52SMichael Hamann foreach (array_keys($pages) as $p_id) { 247*0c074a52SMichael Hamann if (strpos($p_id, $ns) !== 0) { 248*0c074a52SMichael Hamann unset($pages[$p_id]); 249d0bdf765SAdrian Lang } 250d0bdf765SAdrian Lang } 251506fa893SAndreas Gohr } 25263773904SAndreas Gohr 25380423ab6SAdrian Lang // discard hidden pages 25480423ab6SAdrian Lang // discard nonexistent pages 25563773904SAndreas Gohr // check ACL permissions 25663773904SAndreas Gohr foreach(array_keys($pages) as $idx){ 25780423ab6SAdrian Lang if(!isVisiblePage($idx) || !page_exists($idx) || 25880423ab6SAdrian Lang auth_quickaclcheck($idx) < AUTH_READ) { 25963773904SAndreas Gohr unset($pages[$idx]); 26063773904SAndreas Gohr } 26163773904SAndreas Gohr } 26263773904SAndreas Gohr 2633d2017d9SAdrian Lang uksort($pages,'ft_pagesorter'); 2648d22f1e9SAndreas Gohr return $pages; 265506fa893SAndreas Gohr} 266506fa893SAndreas Gohr 267506fa893SAndreas Gohr/** 268f31eb72bSAndreas Gohr * Sort pages based on their namespace level first, then on their string 269f31eb72bSAndreas Gohr * values. This makes higher hierarchy pages rank higher than lower hierarchy 270f31eb72bSAndreas Gohr * pages. 271f31eb72bSAndreas Gohr */ 272f31eb72bSAndreas Gohrfunction ft_pagesorter($a, $b){ 273f31eb72bSAndreas Gohr $ac = count(explode(':',$a)); 274f31eb72bSAndreas Gohr $bc = count(explode(':',$b)); 275f31eb72bSAndreas Gohr if($ac < $bc){ 276f31eb72bSAndreas Gohr return -1; 277f31eb72bSAndreas Gohr }elseif($ac > $bc){ 278f31eb72bSAndreas Gohr return 1; 279f31eb72bSAndreas Gohr } 280f31eb72bSAndreas Gohr return strcmp ($a,$b); 281f31eb72bSAndreas Gohr} 282f31eb72bSAndreas Gohr 283f31eb72bSAndreas Gohr/** 284506fa893SAndreas Gohr * Creates a snippet extract 285506fa893SAndreas Gohr * 286506fa893SAndreas Gohr * @author Andreas Gohr <andi@splitbrain.org> 28760e91a17SAndreas Gohr * @triggers FULLTEXT_SNIPPET_CREATE 288506fa893SAndreas Gohr */ 289546d3a99SAndreas Gohrfunction ft_snippet($id,$highlight){ 290506fa893SAndreas Gohr $text = rawWiki($id); 2914f0030ddSAndreas Gohr $text = str_replace("\xC2\xAD",'',$text); // remove soft-hyphens 29260e91a17SAndreas Gohr $evdata = array( 29360e91a17SAndreas Gohr 'id' => $id, 29460e91a17SAndreas Gohr 'text' => &$text, 29560e91a17SAndreas Gohr 'highlight' => &$highlight, 29660e91a17SAndreas Gohr 'snippet' => '', 29760e91a17SAndreas Gohr ); 29860e91a17SAndreas Gohr 29960e91a17SAndreas Gohr $evt = new Doku_Event('FULLTEXT_SNIPPET_CREATE',$evdata); 30060e91a17SAndreas Gohr if ($evt->advise_before()) { 301ced0762eSchris $match = array(); 302ced0762eSchris $snippets = array(); 3039ee93076Schris $utf8_offset = $offset = $end = 0; 304ced0762eSchris $len = utf8_strlen($text); 3059ee93076Schris 306546d3a99SAndreas Gohr // build a regexp from the phrases to highlight 3072237b4faSAndreas Gohr $re1 = '('.join('|',array_map('ft_snippet_re_preprocess', array_map('preg_quote_cb',array_filter((array) $highlight)))).')'; 308b571ff2dSChuck Kollars $re2 = "$re1.{0,75}(?!\\1)$re1"; 309b571ff2dSChuck Kollars $re3 = "$re1.{0,45}(?!\\1)$re1.{0,45}(?!\\1)(?!\\2)$re1"; 310546d3a99SAndreas Gohr 311b571ff2dSChuck Kollars for ($cnt=4; $cnt--;) { 312b571ff2dSChuck Kollars if (0) { 313b571ff2dSChuck Kollars } else if (preg_match('/'.$re3.'/iu',$text,$match,PREG_OFFSET_CAPTURE,$offset)) { 314b571ff2dSChuck Kollars } else if (preg_match('/'.$re2.'/iu',$text,$match,PREG_OFFSET_CAPTURE,$offset)) { 315b571ff2dSChuck Kollars } else if (preg_match('/'.$re1.'/iu',$text,$match,PREG_OFFSET_CAPTURE,$offset)) { 316b571ff2dSChuck Kollars } else { 317b571ff2dSChuck Kollars break; 318b571ff2dSChuck Kollars } 319ced0762eSchris 320ced0762eSchris list($str,$idx) = $match[0]; 321ced0762eSchris 322ced0762eSchris // convert $idx (a byte offset) into a utf8 character offset 323ced0762eSchris $utf8_idx = utf8_strlen(substr($text,0,$idx)); 324ced0762eSchris $utf8_len = utf8_strlen($str); 325ced0762eSchris 326ced0762eSchris // establish context, 100 bytes surrounding the match string 327ced0762eSchris // first look to see if we can go 100 either side, 328ced0762eSchris // then drop to 50 adding any excess if the other side can't go to 50, 329ced0762eSchris $pre = min($utf8_idx-$utf8_offset,100); 330ced0762eSchris $post = min($len-$utf8_idx-$utf8_len,100); 331ced0762eSchris 332ced0762eSchris if ($pre>50 && $post>50) { 333ced0762eSchris $pre = $post = 50; 334ced0762eSchris } else if ($pre>50) { 335ced0762eSchris $pre = min($pre,100-$post); 336ced0762eSchris } else if ($post>50) { 337ced0762eSchris $post = min($post, 100-$pre); 338ced0762eSchris } else { 339ced0762eSchris // both are less than 50, means the context is the whole string 34010ffc9ddSAndreas Gohr // make it so and break out of this loop - there is no need for the 34110ffc9ddSAndreas Gohr // complex snippet calculations 342ced0762eSchris $snippets = array($text); 343ced0762eSchris break; 344ced0762eSchris } 345ced0762eSchris 34610ffc9ddSAndreas Gohr // establish context start and end points, try to append to previous 34710ffc9ddSAndreas Gohr // context if possible 3489ee93076Schris $start = $utf8_idx - $pre; 349ced0762eSchris $append = ($start < $end) ? $end : false; // still the end of the previous context snippet 3509ee93076Schris $end = $utf8_idx + $utf8_len + $post; // now set it to the end of this context 351ced0762eSchris 352ced0762eSchris if ($append) { 353ced0762eSchris $snippets[count($snippets)-1] .= utf8_substr($text,$append,$end-$append); 354ced0762eSchris } else { 355ced0762eSchris $snippets[] = utf8_substr($text,$start,$end-$start); 356ced0762eSchris } 357ced0762eSchris 358ced0762eSchris // set $offset for next match attempt 35910ffc9ddSAndreas Gohr // substract strlen to avoid splitting a potential search success, 36010ffc9ddSAndreas Gohr // this is an approximation as the search pattern may match strings 36110ffc9ddSAndreas Gohr // of varying length and it will fail if the context snippet 362ced0762eSchris // boundary breaks a matching string longer than the current match 3639ee93076Schris $utf8_offset = $utf8_idx + $post; 3649ee93076Schris $offset = $idx + strlen(utf8_substr($text,$utf8_idx,$post)); 3659ee93076Schris $offset = utf8_correctIdx($text,$offset); 3669ee93076Schris } 3679ee93076Schris 368ced0762eSchris $m = "\1"; 369b571ff2dSChuck Kollars $snippets = preg_replace('/'.$re1.'/iu',$m.'$1'.$m,$snippets); 370b571ff2dSChuck Kollars $snippet = preg_replace('/'.$m.'([^'.$m.']*?)'.$m.'/iu','<strong class="search_hit">$1</strong>',hsc(join('... ',$snippets))); 371bd2cb6fcSchris 37260e91a17SAndreas Gohr $evdata['snippet'] = $snippet; 37360e91a17SAndreas Gohr } 37460e91a17SAndreas Gohr $evt->advise_after(); 37560e91a17SAndreas Gohr unset($evt); 37660e91a17SAndreas Gohr 37760e91a17SAndreas Gohr return $evdata['snippet']; 378506fa893SAndreas Gohr} 379506fa893SAndreas Gohr 380506fa893SAndreas Gohr/** 38126eb848cSGina Haeussge * Wraps a search term in regex boundary checks. 38226eb848cSGina Haeussge */ 3832237b4faSAndreas Gohrfunction ft_snippet_re_preprocess($term) { 38435594613SKazutaka Miyasaka // do not process asian terms where word boundaries are not explicit 38535594613SKazutaka Miyasaka if(preg_match('/'.IDX_ASIAN.'/u',$term)){ 38635594613SKazutaka Miyasaka return $term; 38735594613SKazutaka Miyasaka } 38835594613SKazutaka Miyasaka 3892237b4faSAndreas Gohr if(substr($term,0,2) == '\\*'){ 3902237b4faSAndreas Gohr $term = substr($term,2); 3912237b4faSAndreas Gohr }else{ 3922237b4faSAndreas Gohr $term = '\b'.$term; 3932237b4faSAndreas Gohr } 3942237b4faSAndreas Gohr 3952237b4faSAndreas Gohr if(substr($term,-2,2) == '\\*'){ 3962237b4faSAndreas Gohr $term = substr($term,0,-2); 3972237b4faSAndreas Gohr }else{ 3982237b4faSAndreas Gohr $term = $term.'\b'; 3992237b4faSAndreas Gohr } 4002237b4faSAndreas Gohr return $term; 40126eb848cSGina Haeussge} 40226eb848cSGina Haeussge 40326eb848cSGina Haeussge/** 404f5eb7cf0SAndreas Gohr * Combine found documents and sum up their scores 405f5eb7cf0SAndreas Gohr * 406f5eb7cf0SAndreas Gohr * This function is used to combine searched words with a logical 407f5eb7cf0SAndreas Gohr * AND. Only documents available in all arrays are returned. 408f5eb7cf0SAndreas Gohr * 409f5eb7cf0SAndreas Gohr * based upon PEAR's PHP_Compat function for array_intersect_key() 410f5eb7cf0SAndreas Gohr * 411f5eb7cf0SAndreas Gohr * @param array $args An array of page arrays 412f5eb7cf0SAndreas Gohr */ 413f5eb7cf0SAndreas Gohrfunction ft_resultCombine($args){ 414f5eb7cf0SAndreas Gohr $array_count = count($args); 415134f4ab2SAndreas Gohr if($array_count == 1){ 416134f4ab2SAndreas Gohr return $args[0]; 417134f4ab2SAndreas Gohr } 418134f4ab2SAndreas Gohr 419f5eb7cf0SAndreas Gohr $result = array(); 42009c27a6dSGuy Brand if ($array_count > 1) { 421a21136cdSAndreas Gohr foreach ($args[0] as $key => $value) { 422a21136cdSAndreas Gohr $result[$key] = $value; 423f5eb7cf0SAndreas Gohr for ($i = 1; $i !== $array_count; $i++) { 424a21136cdSAndreas Gohr if (!isset($args[$i][$key])) { 425a21136cdSAndreas Gohr unset($result[$key]); 426a21136cdSAndreas Gohr break; 427f5eb7cf0SAndreas Gohr } 428a21136cdSAndreas Gohr $result[$key] += $args[$i][$key]; 429f5eb7cf0SAndreas Gohr } 430f5eb7cf0SAndreas Gohr } 43109c27a6dSGuy Brand } 432f5eb7cf0SAndreas Gohr return $result; 433f5eb7cf0SAndreas Gohr} 434f5eb7cf0SAndreas Gohr 435f5eb7cf0SAndreas Gohr/** 436865c2687SKazutaka Miyasaka * Unites found documents and sum up their scores 437f5eb7cf0SAndreas Gohr * 438865c2687SKazutaka Miyasaka * based upon ft_resultCombine() function 439865c2687SKazutaka Miyasaka * 440865c2687SKazutaka Miyasaka * @param array $args An array of page arrays 441865c2687SKazutaka Miyasaka * @author Kazutaka Miyasaka <kazmiya@gmail.com> 442865c2687SKazutaka Miyasaka */ 443865c2687SKazutaka Miyasakafunction ft_resultUnite($args) { 444865c2687SKazutaka Miyasaka $array_count = count($args); 445865c2687SKazutaka Miyasaka if ($array_count === 1) { 446865c2687SKazutaka Miyasaka return $args[0]; 447865c2687SKazutaka Miyasaka } 448865c2687SKazutaka Miyasaka 449865c2687SKazutaka Miyasaka $result = $args[0]; 450865c2687SKazutaka Miyasaka for ($i = 1; $i !== $array_count; $i++) { 451865c2687SKazutaka Miyasaka foreach (array_keys($args[$i]) as $id) { 452865c2687SKazutaka Miyasaka $result[$id] += $args[$i][$id]; 453865c2687SKazutaka Miyasaka } 454865c2687SKazutaka Miyasaka } 455865c2687SKazutaka Miyasaka return $result; 456865c2687SKazutaka Miyasaka} 457865c2687SKazutaka Miyasaka 458865c2687SKazutaka Miyasaka/** 459865c2687SKazutaka Miyasaka * Computes the difference of documents using page id for comparison 460865c2687SKazutaka Miyasaka * 461865c2687SKazutaka Miyasaka * nearly identical to PHP5's array_diff_key() 462865c2687SKazutaka Miyasaka * 463865c2687SKazutaka Miyasaka * @param array $args An array of page arrays 464865c2687SKazutaka Miyasaka * @author Kazutaka Miyasaka <kazmiya@gmail.com> 465865c2687SKazutaka Miyasaka */ 466865c2687SKazutaka Miyasakafunction ft_resultComplement($args) { 467865c2687SKazutaka Miyasaka $array_count = count($args); 468865c2687SKazutaka Miyasaka if ($array_count === 1) { 469865c2687SKazutaka Miyasaka return $args[0]; 470865c2687SKazutaka Miyasaka } 471865c2687SKazutaka Miyasaka 472865c2687SKazutaka Miyasaka $result = $args[0]; 473865c2687SKazutaka Miyasaka foreach (array_keys($result) as $id) { 474865c2687SKazutaka Miyasaka for ($i = 1; $i !== $array_count; $i++) { 475865c2687SKazutaka Miyasaka if (isset($args[$i][$id])) unset($result[$id]); 476865c2687SKazutaka Miyasaka } 477865c2687SKazutaka Miyasaka } 478865c2687SKazutaka Miyasaka return $result; 479865c2687SKazutaka Miyasaka} 480865c2687SKazutaka Miyasaka 481865c2687SKazutaka Miyasaka/** 482865c2687SKazutaka Miyasaka * Parses a search query and builds an array of search formulas 483865c2687SKazutaka Miyasaka * 484865c2687SKazutaka Miyasaka * @author Andreas Gohr <andi@splitbrain.org> 485865c2687SKazutaka Miyasaka * @author Kazutaka Miyasaka <kazmiya@gmail.com> 486f5eb7cf0SAndreas Gohr */ 4879b41be24STom N Harrisfunction ft_queryParser($Indexer, $query){ 488865c2687SKazutaka Miyasaka /** 489865c2687SKazutaka Miyasaka * parse a search query and transform it into intermediate representation 490865c2687SKazutaka Miyasaka * 491865c2687SKazutaka Miyasaka * in a search query, you can use the following expressions: 492865c2687SKazutaka Miyasaka * 493865c2687SKazutaka Miyasaka * words: 494865c2687SKazutaka Miyasaka * include 495865c2687SKazutaka Miyasaka * -exclude 496865c2687SKazutaka Miyasaka * phrases: 497865c2687SKazutaka Miyasaka * "phrase to be included" 498865c2687SKazutaka Miyasaka * -"phrase you want to exclude" 499865c2687SKazutaka Miyasaka * namespaces: 500865c2687SKazutaka Miyasaka * @include:namespace (or ns:include:namespace) 501865c2687SKazutaka Miyasaka * ^exclude:namespace (or -ns:exclude:namespace) 502865c2687SKazutaka Miyasaka * groups: 503865c2687SKazutaka Miyasaka * () 504865c2687SKazutaka Miyasaka * -() 505865c2687SKazutaka Miyasaka * operators: 506865c2687SKazutaka Miyasaka * and ('and' is the default operator: you can always omit this) 5077871d415SKazutaka Miyasaka * or (or pipe symbol '|', lower precedence than 'and') 508865c2687SKazutaka Miyasaka * 509865c2687SKazutaka Miyasaka * e.g. a query [ aa "bb cc" @dd:ee ] means "search pages which contain 510865c2687SKazutaka Miyasaka * a word 'aa', a phrase 'bb cc' and are within a namespace 'dd:ee'". 511865c2687SKazutaka Miyasaka * this query is equivalent to [ -(-aa or -"bb cc" or -ns:dd:ee) ] 512865c2687SKazutaka Miyasaka * as long as you don't mind hit counts. 513865c2687SKazutaka Miyasaka * 514865c2687SKazutaka Miyasaka * intermediate representation consists of the following parts: 515865c2687SKazutaka Miyasaka * 516865c2687SKazutaka Miyasaka * ( ) - group 517865c2687SKazutaka Miyasaka * AND - logical and 518865c2687SKazutaka Miyasaka * OR - logical or 519865c2687SKazutaka Miyasaka * NOT - logical not 5202f502d70SKazutaka Miyasaka * W+:, W-:, W_: - word (underscore: no need to highlight) 5212f502d70SKazutaka Miyasaka * P+:, P-: - phrase (minus sign: logically in NOT group) 5222f502d70SKazutaka Miyasaka * N+:, N-: - namespace 523865c2687SKazutaka Miyasaka */ 524865c2687SKazutaka Miyasaka $parsed_query = ''; 525865c2687SKazutaka Miyasaka $parens_level = 0; 526865c2687SKazutaka Miyasaka $terms = preg_split('/(-?".*?")/u', utf8_strtolower($query), -1, PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY); 527865c2687SKazutaka Miyasaka 528865c2687SKazutaka Miyasaka foreach ($terms as $term) { 529865c2687SKazutaka Miyasaka $parsed = ''; 530865c2687SKazutaka Miyasaka if (preg_match('/^(-?)"(.+)"$/u', $term, $matches)) { 531865c2687SKazutaka Miyasaka // phrase-include and phrase-exclude 532865c2687SKazutaka Miyasaka $not = $matches[1] ? 'NOT' : ''; 5339b41be24STom N Harris $parsed = $not.ft_termParser($Indexer, $matches[2], false, true); 534f5eb7cf0SAndreas Gohr } else { 535865c2687SKazutaka Miyasaka // fix incomplete phrase 536865c2687SKazutaka Miyasaka $term = str_replace('"', ' ', $term); 537865c2687SKazutaka Miyasaka 538865c2687SKazutaka Miyasaka // fix parentheses 539865c2687SKazutaka Miyasaka $term = str_replace(')' , ' ) ', $term); 540865c2687SKazutaka Miyasaka $term = str_replace('(' , ' ( ', $term); 541865c2687SKazutaka Miyasaka $term = str_replace('- (', ' -(', $term); 542865c2687SKazutaka Miyasaka 5437871d415SKazutaka Miyasaka // treat pipe symbols as 'OR' operators 5447871d415SKazutaka Miyasaka $term = str_replace('|', ' or ', $term); 5457871d415SKazutaka Miyasaka 546865c2687SKazutaka Miyasaka // treat ideographic spaces (U+3000) as search term separators 547865c2687SKazutaka Miyasaka // FIXME: some more separators? 548865c2687SKazutaka Miyasaka $term = preg_replace('/[ \x{3000}]+/u', ' ', $term); 549865c2687SKazutaka Miyasaka $term = trim($term); 550865c2687SKazutaka Miyasaka if ($term === '') continue; 551865c2687SKazutaka Miyasaka 552865c2687SKazutaka Miyasaka $tokens = explode(' ', $term); 553865c2687SKazutaka Miyasaka foreach ($tokens as $token) { 554865c2687SKazutaka Miyasaka if ($token === '(') { 555865c2687SKazutaka Miyasaka // parenthesis-include-open 556865c2687SKazutaka Miyasaka $parsed .= '('; 557865c2687SKazutaka Miyasaka ++$parens_level; 558865c2687SKazutaka Miyasaka } elseif ($token === '-(') { 559865c2687SKazutaka Miyasaka // parenthesis-exclude-open 560865c2687SKazutaka Miyasaka $parsed .= 'NOT('; 561865c2687SKazutaka Miyasaka ++$parens_level; 562865c2687SKazutaka Miyasaka } elseif ($token === ')') { 563865c2687SKazutaka Miyasaka // parenthesis-any-close 564865c2687SKazutaka Miyasaka if ($parens_level === 0) continue; 565865c2687SKazutaka Miyasaka $parsed .= ')'; 566865c2687SKazutaka Miyasaka $parens_level--; 567865c2687SKazutaka Miyasaka } elseif ($token === 'and') { 568865c2687SKazutaka Miyasaka // logical-and (do nothing) 569865c2687SKazutaka Miyasaka } elseif ($token === 'or') { 570865c2687SKazutaka Miyasaka // logical-or 571865c2687SKazutaka Miyasaka $parsed .= 'OR'; 572865c2687SKazutaka Miyasaka } elseif (preg_match('/^(?:\^|-ns:)(.+)$/u', $token, $matches)) { 573865c2687SKazutaka Miyasaka // namespace-exclude 5742f502d70SKazutaka Miyasaka $parsed .= 'NOT(N+:'.$matches[1].')'; 575865c2687SKazutaka Miyasaka } elseif (preg_match('/^(?:@|ns:)(.+)$/u', $token, $matches)) { 576865c2687SKazutaka Miyasaka // namespace-include 5772f502d70SKazutaka Miyasaka $parsed .= '(N+:'.$matches[1].')'; 578865c2687SKazutaka Miyasaka } elseif (preg_match('/^-(.+)$/', $token, $matches)) { 579865c2687SKazutaka Miyasaka // word-exclude 5809b41be24STom N Harris $parsed .= 'NOT('.ft_termParser($Indexer, $matches[1]).')'; 581865c2687SKazutaka Miyasaka } else { 582865c2687SKazutaka Miyasaka // word-include 5839b41be24STom N Harris $parsed .= ft_termParser($Indexer, $token); 584865c2687SKazutaka Miyasaka } 585865c2687SKazutaka Miyasaka } 586865c2687SKazutaka Miyasaka } 587865c2687SKazutaka Miyasaka $parsed_query .= $parsed; 588f5eb7cf0SAndreas Gohr } 589f5eb7cf0SAndreas Gohr 590865c2687SKazutaka Miyasaka // cleanup (very sensitive) 591865c2687SKazutaka Miyasaka $parsed_query .= str_repeat(')', $parens_level); 592865c2687SKazutaka Miyasaka do { 593865c2687SKazutaka Miyasaka $parsed_query_old = $parsed_query; 594865c2687SKazutaka Miyasaka $parsed_query = preg_replace('/(NOT)?\(\)/u', '', $parsed_query); 595865c2687SKazutaka Miyasaka } while ($parsed_query !== $parsed_query_old); 596865c2687SKazutaka Miyasaka $parsed_query = preg_replace('/(NOT|OR)+\)/u', ')' , $parsed_query); 597865c2687SKazutaka Miyasaka $parsed_query = preg_replace('/(OR)+/u' , 'OR' , $parsed_query); 598865c2687SKazutaka Miyasaka $parsed_query = preg_replace('/\(OR/u' , '(' , $parsed_query); 599865c2687SKazutaka Miyasaka $parsed_query = preg_replace('/^OR|OR$/u' , '' , $parsed_query); 600865c2687SKazutaka Miyasaka $parsed_query = preg_replace('/\)(NOT)?\(/u' , ')AND$1(', $parsed_query); 601865c2687SKazutaka Miyasaka 6022f502d70SKazutaka Miyasaka // adjustment: make highlightings right 6032f502d70SKazutaka Miyasaka $parens_level = 0; 6042f502d70SKazutaka Miyasaka $notgrp_levels = array(); 6052f502d70SKazutaka Miyasaka $parsed_query_new = ''; 6062f502d70SKazutaka Miyasaka $tokens = preg_split('/(NOT\(|[()])/u', $parsed_query, -1, PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY); 6072f502d70SKazutaka Miyasaka foreach ($tokens as $token) { 6082f502d70SKazutaka Miyasaka if ($token === 'NOT(') { 6092f502d70SKazutaka Miyasaka $notgrp_levels[] = ++$parens_level; 6102f502d70SKazutaka Miyasaka } elseif ($token === '(') { 6112f502d70SKazutaka Miyasaka ++$parens_level; 6122f502d70SKazutaka Miyasaka } elseif ($token === ')') { 6132f502d70SKazutaka Miyasaka if ($parens_level-- === end($notgrp_levels)) array_pop($notgrp_levels); 6142f502d70SKazutaka Miyasaka } elseif (count($notgrp_levels) % 2 === 1) { 6152f502d70SKazutaka Miyasaka // turn highlight-flag off if terms are logically in "NOT" group 6162f502d70SKazutaka Miyasaka $token = preg_replace('/([WPN])\+\:/u', '$1-:', $token); 6172f502d70SKazutaka Miyasaka } 6182f502d70SKazutaka Miyasaka $parsed_query_new .= $token; 6192f502d70SKazutaka Miyasaka } 6202f502d70SKazutaka Miyasaka $parsed_query = $parsed_query_new; 6212f502d70SKazutaka Miyasaka 622865c2687SKazutaka Miyasaka /** 623865c2687SKazutaka Miyasaka * convert infix notation string into postfix (Reverse Polish notation) array 624865c2687SKazutaka Miyasaka * by Shunting-yard algorithm 625865c2687SKazutaka Miyasaka * 626865c2687SKazutaka Miyasaka * see: http://en.wikipedia.org/wiki/Reverse_Polish_notation 627865c2687SKazutaka Miyasaka * see: http://en.wikipedia.org/wiki/Shunting-yard_algorithm 628865c2687SKazutaka Miyasaka */ 629865c2687SKazutaka Miyasaka $parsed_ary = array(); 630865c2687SKazutaka Miyasaka $ope_stack = array(); 631865c2687SKazutaka Miyasaka $ope_precedence = array(')' => 1, 'OR' => 2, 'AND' => 3, 'NOT' => 4, '(' => 5); 632865c2687SKazutaka Miyasaka $ope_regex = '/([()]|OR|AND|NOT)/u'; 633865c2687SKazutaka Miyasaka 634865c2687SKazutaka Miyasaka $tokens = preg_split($ope_regex, $parsed_query, -1, PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY); 635865c2687SKazutaka Miyasaka foreach ($tokens as $token) { 636865c2687SKazutaka Miyasaka if (preg_match($ope_regex, $token)) { 637865c2687SKazutaka Miyasaka // operator 638865c2687SKazutaka Miyasaka $last_ope = end($ope_stack); 639865c2687SKazutaka Miyasaka while ($ope_precedence[$token] <= $ope_precedence[$last_ope] && $last_ope != '(') { 640865c2687SKazutaka Miyasaka $parsed_ary[] = array_pop($ope_stack); 641865c2687SKazutaka Miyasaka $last_ope = end($ope_stack); 642865c2687SKazutaka Miyasaka } 643865c2687SKazutaka Miyasaka if ($token == ')') { 644865c2687SKazutaka Miyasaka array_pop($ope_stack); // this array_pop always deletes '(' 645865c2687SKazutaka Miyasaka } else { 646865c2687SKazutaka Miyasaka $ope_stack[] = $token; 647865c2687SKazutaka Miyasaka } 648865c2687SKazutaka Miyasaka } else { 649865c2687SKazutaka Miyasaka // operand 650865c2687SKazutaka Miyasaka $token_decoded = str_replace(array('OP', 'CP'), array('(', ')'), $token); 651865c2687SKazutaka Miyasaka $parsed_ary[] = $token_decoded; 652865c2687SKazutaka Miyasaka } 653865c2687SKazutaka Miyasaka } 654865c2687SKazutaka Miyasaka $parsed_ary = array_values(array_merge($parsed_ary, array_reverse($ope_stack))); 655865c2687SKazutaka Miyasaka 656865c2687SKazutaka Miyasaka // cleanup: each double "NOT" in RPN array actually does nothing 657865c2687SKazutaka Miyasaka $parsed_ary_count = count($parsed_ary); 658865c2687SKazutaka Miyasaka for ($i = 1; $i < $parsed_ary_count; ++$i) { 659865c2687SKazutaka Miyasaka if ($parsed_ary[$i] === 'NOT' && $parsed_ary[$i - 1] === 'NOT') { 660865c2687SKazutaka Miyasaka unset($parsed_ary[$i], $parsed_ary[$i - 1]); 661865c2687SKazutaka Miyasaka } 662865c2687SKazutaka Miyasaka } 663865c2687SKazutaka Miyasaka $parsed_ary = array_values($parsed_ary); 664865c2687SKazutaka Miyasaka 665865c2687SKazutaka Miyasaka // build return value 666f5eb7cf0SAndreas Gohr $q = array(); 667f5eb7cf0SAndreas Gohr $q['query'] = $query; 668865c2687SKazutaka Miyasaka $q['parsed_str'] = $parsed_query; 669865c2687SKazutaka Miyasaka $q['parsed_ary'] = $parsed_ary; 670f5eb7cf0SAndreas Gohr 671865c2687SKazutaka Miyasaka foreach ($q['parsed_ary'] as $token) { 672865c2687SKazutaka Miyasaka if ($token[2] !== ':') continue; 673865c2687SKazutaka Miyasaka $body = substr($token, 3); 674865c2687SKazutaka Miyasaka 675865c2687SKazutaka Miyasaka switch (substr($token, 0, 3)) { 6762f502d70SKazutaka Miyasaka case 'N+:': 677865c2687SKazutaka Miyasaka $q['ns'][] = $body; // for backward compatibility 678865c2687SKazutaka Miyasaka break; 6792f502d70SKazutaka Miyasaka case 'N-:': 6802f502d70SKazutaka Miyasaka $q['notns'][] = $body; // for backward compatibility 6812f502d70SKazutaka Miyasaka break; 6822f502d70SKazutaka Miyasaka case 'W_:': 6832f502d70SKazutaka Miyasaka $q['words'][] = $body; 6842f502d70SKazutaka Miyasaka break; 685865c2687SKazutaka Miyasaka case 'W-:': 686865c2687SKazutaka Miyasaka $q['words'][] = $body; 6872f502d70SKazutaka Miyasaka $q['not'][] = $body; // for backward compatibility 688865c2687SKazutaka Miyasaka break; 689865c2687SKazutaka Miyasaka case 'W+:': 690865c2687SKazutaka Miyasaka $q['words'][] = $body; 6912237b4faSAndreas Gohr $q['highlight'][] = $body; 6922f502d70SKazutaka Miyasaka $q['and'][] = $body; // for backward compatibility 693865c2687SKazutaka Miyasaka break; 6942f502d70SKazutaka Miyasaka case 'P-:': 6952f502d70SKazutaka Miyasaka $q['phrases'][] = $body; 6962f502d70SKazutaka Miyasaka break; 6972f502d70SKazutaka Miyasaka case 'P+:': 698865c2687SKazutaka Miyasaka $q['phrases'][] = $body; 6992237b4faSAndreas Gohr $q['highlight'][] = $body; 700865c2687SKazutaka Miyasaka break; 701865c2687SKazutaka Miyasaka } 702865c2687SKazutaka Miyasaka } 7032f502d70SKazutaka Miyasaka foreach (array('words', 'phrases', 'highlight', 'ns', 'notns', 'and', 'not') as $key) { 704865c2687SKazutaka Miyasaka $q[$key] = empty($q[$key]) ? array() : array_values(array_unique($q[$key])); 705f5eb7cf0SAndreas Gohr } 706f5eb7cf0SAndreas Gohr 707f5eb7cf0SAndreas Gohr return $q; 708f5eb7cf0SAndreas Gohr} 709f5eb7cf0SAndreas Gohr 710865c2687SKazutaka Miyasaka/** 711865c2687SKazutaka Miyasaka * Transforms given search term into intermediate representation 712865c2687SKazutaka Miyasaka * 713865c2687SKazutaka Miyasaka * This function is used in ft_queryParser() and not for general purpose use. 714865c2687SKazutaka Miyasaka * 715865c2687SKazutaka Miyasaka * @author Kazutaka Miyasaka <kazmiya@gmail.com> 716865c2687SKazutaka Miyasaka */ 7179b41be24STom N Harrisfunction ft_termParser($Indexer, $term, $consider_asian = true, $phrase_mode = false) { 718865c2687SKazutaka Miyasaka $parsed = ''; 719865c2687SKazutaka Miyasaka if ($consider_asian) { 720865c2687SKazutaka Miyasaka // successive asian characters need to be searched as a phrase 721865c2687SKazutaka Miyasaka $words = preg_split('/('.IDX_ASIAN.'+)/u', $term, -1, PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY); 722865c2687SKazutaka Miyasaka foreach ($words as $word) { 7236ac2077aSKazutaka Miyasaka $phrase_mode = $phrase_mode ? true : preg_match('/'.IDX_ASIAN.'/u', $word); 7249b41be24STom N Harris $parsed .= ft_termParser($Indexer, $word, false, $phrase_mode); 725865c2687SKazutaka Miyasaka } 726865c2687SKazutaka Miyasaka } else { 727865c2687SKazutaka Miyasaka $term_noparen = str_replace(array('(', ')'), ' ', $term); 7289b41be24STom N Harris $words = $Indexer->tokenizer($term_noparen, true); 729865c2687SKazutaka Miyasaka 7302f502d70SKazutaka Miyasaka // W_: no need to highlight 731865c2687SKazutaka Miyasaka if (empty($words)) { 732865c2687SKazutaka Miyasaka $parsed = '()'; // important: do not remove 733865c2687SKazutaka Miyasaka } elseif ($words[0] === $term) { 734865c2687SKazutaka Miyasaka $parsed = '(W+:'.$words[0].')'; 735865c2687SKazutaka Miyasaka } elseif ($phrase_mode) { 736865c2687SKazutaka Miyasaka $term_encoded = str_replace(array('(', ')'), array('OP', 'CP'), $term); 7372f502d70SKazutaka Miyasaka $parsed = '((W_:'.implode(')(W_:', $words).')(P+:'.$term_encoded.'))'; 738865c2687SKazutaka Miyasaka } else { 739865c2687SKazutaka Miyasaka $parsed = '((W+:'.implode(')(W+:', $words).'))'; 740865c2687SKazutaka Miyasaka } 741865c2687SKazutaka Miyasaka } 742865c2687SKazutaka Miyasaka return $parsed; 743865c2687SKazutaka Miyasaka} 744865c2687SKazutaka Miyasaka 745e3776c06SMichael Hamann//Setup VIM: ex: et ts=4 : 746