1<?php 2/** 3 * DokuWiki fulltextsearch functions using the index 4 * 5 * @license GPL 2 (http://www.gnu.org/licenses/gpl.html) 6 * @author Andreas Gohr <andi@splitbrain.org> 7 */ 8 9if(!defined('DOKU_INC')) die('meh.'); 10 11/** 12 * create snippets for the first few results only 13 */ 14if(!defined('FT_SNIPPET_NUMBER')) define('FT_SNIPPET_NUMBER',15); 15 16/** 17 * The fulltext search 18 * 19 * Returns a list of matching documents for the given query 20 * 21 * refactored into ft_pageSearch(), _ft_pageSearch() and trigger_event() 22 * 23 * @param string $query 24 * @param array $highlight 25 * @return array 26 */ 27function ft_pageSearch($query,&$highlight){ 28 29 $data = array(); 30 $data['query'] = $query; 31 $data['highlight'] =& $highlight; 32 33 return trigger_event('SEARCH_QUERY_FULLPAGE', $data, '_ft_pageSearch'); 34} 35 36/** 37 * Returns a list of matching documents for the given query 38 * 39 * @author Andreas Gohr <andi@splitbrain.org> 40 * @author Kazutaka Miyasaka <kazmiya@gmail.com> 41 * 42 * @param array $data event data 43 * @return array matching documents 44 */ 45function _ft_pageSearch(&$data) { 46 $Indexer = idx_get_indexer(); 47 48 // parse the given query 49 $q = ft_queryParser($Indexer, $data['query']); 50 $data['highlight'] = $q['highlight']; 51 52 if (empty($q['parsed_ary'])) return array(); 53 54 // lookup all words found in the query 55 $lookup = $Indexer->lookup($q['words']); 56 57 // get all pages in this dokuwiki site (!: includes nonexistent pages) 58 $pages_all = array(); 59 foreach ($Indexer->getPages() as $id) { 60 $pages_all[$id] = 0; // base: 0 hit 61 } 62 63 // process the query 64 $stack = array(); 65 foreach ($q['parsed_ary'] as $token) { 66 switch (substr($token, 0, 3)) { 67 case 'W+:': 68 case 'W-:': 69 case 'W_:': // word 70 $word = substr($token, 3); 71 $stack[] = (array) $lookup[$word]; 72 break; 73 case 'P+:': 74 case 'P-:': // phrase 75 $phrase = substr($token, 3); 76 // since phrases are always parsed as ((W1)(W2)...(P)), 77 // the end($stack) always points the pages that contain 78 // all words in this phrase 79 $pages = end($stack); 80 $pages_matched = array(); 81 foreach(array_keys($pages) as $id){ 82 $evdata = array( 83 'id' => $id, 84 'phrase' => $phrase, 85 'text' => rawWiki($id) 86 ); 87 $evt = new Doku_Event('FULLTEXT_PHRASE_MATCH',$evdata); 88 if ($evt->advise_before() && $evt->result !== true) { 89 $text = utf8_strtolower($evdata['text']); 90 if (strpos($text, $phrase) !== false) { 91 $evt->result = true; 92 } 93 } 94 $evt->advise_after(); 95 if ($evt->result === true) { 96 $pages_matched[$id] = 0; // phrase: always 0 hit 97 } 98 } 99 $stack[] = $pages_matched; 100 break; 101 case 'N+:': 102 case 'N-:': // namespace 103 $ns = substr($token, 3); 104 $pages_matched = array(); 105 foreach (array_keys($pages_all) as $id) { 106 if (strpos($id, $ns) === 0) { 107 $pages_matched[$id] = 0; // namespace: always 0 hit 108 } 109 } 110 $stack[] = $pages_matched; 111 break; 112 case 'AND': // and operation 113 list($pages1, $pages2) = array_splice($stack, -2); 114 $stack[] = ft_resultCombine(array($pages1, $pages2)); 115 break; 116 case 'OR': // or operation 117 list($pages1, $pages2) = array_splice($stack, -2); 118 $stack[] = ft_resultUnite(array($pages1, $pages2)); 119 break; 120 case 'NOT': // not operation (unary) 121 $pages = array_pop($stack); 122 $stack[] = ft_resultComplement(array($pages_all, $pages)); 123 break; 124 } 125 } 126 $docs = array_pop($stack); 127 128 if (empty($docs)) return array(); 129 130 // check: settings, acls, existence 131 foreach (array_keys($docs) as $id) { 132 if (isHiddenPage($id) || auth_quickaclcheck($id) < AUTH_READ || !page_exists($id, '', false)) { 133 unset($docs[$id]); 134 } 135 } 136 137 // sort docs by count 138 arsort($docs); 139 140 return $docs; 141} 142 143/** 144 * Returns the backlinks for a given page 145 * 146 * Uses the metadata index. 147 * 148 * @param string $id The id for which links shall be returned 149 * @param bool $ignore_perms Ignore the fact that pages are hidden or read-protected 150 * @return array The pages that contain links to the given page 151 */ 152function ft_backlinks($id, $ignore_perms = false){ 153 $result = idx_get_indexer()->lookupKey('relation_references', $id); 154 155 if(!count($result)) return $result; 156 157 // check ACL permissions 158 foreach(array_keys($result) as $idx){ 159 if(($ignore_perms !== true && ( 160 isHiddenPage($result[$idx]) || auth_quickaclcheck($result[$idx]) < AUTH_READ 161 )) || !page_exists($result[$idx], '', false)){ 162 unset($result[$idx]); 163 } 164 } 165 166 sort($result); 167 return $result; 168} 169 170/** 171 * Returns the pages that use a given media file 172 * 173 * Uses the relation media metadata property and the metadata index. 174 * 175 * Note that before 2013-07-31 the second parameter was the maximum number of results and 176 * permissions were ignored. That's why the parameter is now checked to be explicitely set 177 * to true (with type bool) in order to be compatible with older uses of the function. 178 * 179 * @param string $id The media id to look for 180 * @param bool $ignore_perms Ignore hidden pages and acls (optional, default: false) 181 * @return array A list of pages that use the given media file 182 */ 183function ft_mediause($id, $ignore_perms = false){ 184 $result = idx_get_indexer()->lookupKey('relation_media', $id); 185 186 if(!count($result)) return $result; 187 188 // check ACL permissions 189 foreach(array_keys($result) as $idx){ 190 if(($ignore_perms !== true && ( 191 isHiddenPage($result[$idx]) || auth_quickaclcheck($result[$idx]) < AUTH_READ 192 )) || !page_exists($result[$idx], '', false)){ 193 unset($result[$idx]); 194 } 195 } 196 197 sort($result); 198 return $result; 199} 200 201 202 203/** 204 * Quicksearch for pagenames 205 * 206 * By default it only matches the pagename and ignores the 207 * namespace. This can be changed with the second parameter. 208 * The third parameter allows to search in titles as well. 209 * 210 * The function always returns titles as well 211 * 212 * @triggers SEARCH_QUERY_PAGELOOKUP 213 * @author Andreas Gohr <andi@splitbrain.org> 214 * @author Adrian Lang <lang@cosmocode.de> 215 * 216 * @param string $id page id 217 * @param bool $in_ns match against namespace as well? 218 * @param bool $in_title search in title? 219 * @return string[] 220 */ 221function ft_pageLookup($id, $in_ns=false, $in_title=false){ 222 $data = compact('id', 'in_ns', 'in_title'); 223 $data['has_titles'] = true; // for plugin backward compatibility check 224 return trigger_event('SEARCH_QUERY_PAGELOOKUP', $data, '_ft_pageLookup'); 225} 226 227/** 228 * Returns list of pages as array(pageid => First Heading) 229 * 230 * @param array &$data event data 231 * @return string[] 232 */ 233function _ft_pageLookup(&$data){ 234 // split out original parameters 235 $id = $data['id']; 236 if (preg_match('/(?:^| )(?:@|ns:)([\w:]+)/', $id, $matches)) { 237 $ns = cleanID($matches[1]) . ':'; 238 $id = str_replace($matches[0], '', $id); 239 } 240 241 $in_ns = $data['in_ns']; 242 $in_title = $data['in_title']; 243 $cleaned = cleanID($id); 244 245 $Indexer = idx_get_indexer(); 246 $page_idx = $Indexer->getPages(); 247 248 $pages = array(); 249 if ($id !== '' && $cleaned !== '') { 250 foreach ($page_idx as $p_id) { 251 if ((strpos($in_ns ? $p_id : noNSorNS($p_id), $cleaned) !== false)) { 252 if (!isset($pages[$p_id])) 253 $pages[$p_id] = p_get_first_heading($p_id, METADATA_DONT_RENDER); 254 } 255 } 256 if ($in_title) { 257 foreach ($Indexer->lookupKey('title', $id, '_ft_pageLookupTitleCompare') as $p_id) { 258 if (!isset($pages[$p_id])) 259 $pages[$p_id] = p_get_first_heading($p_id, METADATA_DONT_RENDER); 260 } 261 } 262 } 263 264 if (isset($ns)) { 265 foreach (array_keys($pages) as $p_id) { 266 if (strpos($p_id, $ns) !== 0) { 267 unset($pages[$p_id]); 268 } 269 } 270 } 271 272 // discard hidden pages 273 // discard nonexistent pages 274 // check ACL permissions 275 foreach(array_keys($pages) as $idx){ 276 if(!isVisiblePage($idx) || !page_exists($idx) || 277 auth_quickaclcheck($idx) < AUTH_READ) { 278 unset($pages[$idx]); 279 } 280 } 281 282 uksort($pages,'ft_pagesorter'); 283 return $pages; 284} 285 286/** 287 * Tiny helper function for comparing the searched title with the title 288 * from the search index. This function is a wrapper around stripos with 289 * adapted argument order and return value. 290 * 291 * @param string $search searched title 292 * @param string $title title from index 293 * @return bool 294 */ 295function _ft_pageLookupTitleCompare($search, $title) { 296 return stripos($title, $search) !== false; 297} 298 299/** 300 * Sort pages based on their namespace level first, then on their string 301 * values. This makes higher hierarchy pages rank higher than lower hierarchy 302 * pages. 303 * 304 * @param string $a 305 * @param string $b 306 * @return int Returns < 0 if $a is less than $b; > 0 if $a is greater than $b, and 0 if they are equal. 307 */ 308function ft_pagesorter($a, $b){ 309 $ac = count(explode(':',$a)); 310 $bc = count(explode(':',$b)); 311 if($ac < $bc){ 312 return -1; 313 }elseif($ac > $bc){ 314 return 1; 315 } 316 return strcmp ($a,$b); 317} 318 319/** 320 * Creates a snippet extract 321 * 322 * @author Andreas Gohr <andi@splitbrain.org> 323 * @triggers FULLTEXT_SNIPPET_CREATE 324 * 325 * @param string $id page id 326 * @param array $highlight 327 * @return mixed 328 */ 329function ft_snippet($id,$highlight){ 330 $text = rawWiki($id); 331 $text = str_replace("\xC2\xAD",'',$text); // remove soft-hyphens 332 $evdata = array( 333 'id' => $id, 334 'text' => &$text, 335 'highlight' => &$highlight, 336 'snippet' => '', 337 ); 338 339 $evt = new Doku_Event('FULLTEXT_SNIPPET_CREATE',$evdata); 340 if ($evt->advise_before()) { 341 $match = array(); 342 $snippets = array(); 343 $utf8_offset = $offset = $end = 0; 344 $len = utf8_strlen($text); 345 346 // build a regexp from the phrases to highlight 347 $re1 = '('.join('|',array_map('ft_snippet_re_preprocess', array_map('preg_quote_cb',array_filter((array) $highlight)))).')'; 348 $re2 = "$re1.{0,75}(?!\\1)$re1"; 349 $re3 = "$re1.{0,45}(?!\\1)$re1.{0,45}(?!\\1)(?!\\2)$re1"; 350 351 for ($cnt=4; $cnt--;) { 352 if (0) { 353 } else if (preg_match('/'.$re3.'/iu',$text,$match,PREG_OFFSET_CAPTURE,$offset)) { 354 } else if (preg_match('/'.$re2.'/iu',$text,$match,PREG_OFFSET_CAPTURE,$offset)) { 355 } else if (preg_match('/'.$re1.'/iu',$text,$match,PREG_OFFSET_CAPTURE,$offset)) { 356 } else { 357 break; 358 } 359 360 list($str,$idx) = $match[0]; 361 362 // convert $idx (a byte offset) into a utf8 character offset 363 $utf8_idx = utf8_strlen(substr($text,0,$idx)); 364 $utf8_len = utf8_strlen($str); 365 366 // establish context, 100 bytes surrounding the match string 367 // first look to see if we can go 100 either side, 368 // then drop to 50 adding any excess if the other side can't go to 50, 369 $pre = min($utf8_idx-$utf8_offset,100); 370 $post = min($len-$utf8_idx-$utf8_len,100); 371 372 if ($pre>50 && $post>50) { 373 $pre = $post = 50; 374 } else if ($pre>50) { 375 $pre = min($pre,100-$post); 376 } else if ($post>50) { 377 $post = min($post, 100-$pre); 378 } else if ($offset == 0) { 379 // both are less than 50, means the context is the whole string 380 // make it so and break out of this loop - there is no need for the 381 // complex snippet calculations 382 $snippets = array($text); 383 break; 384 } 385 386 // establish context start and end points, try to append to previous 387 // context if possible 388 $start = $utf8_idx - $pre; 389 $append = ($start < $end) ? $end : false; // still the end of the previous context snippet 390 $end = $utf8_idx + $utf8_len + $post; // now set it to the end of this context 391 392 if ($append) { 393 $snippets[count($snippets)-1] .= utf8_substr($text,$append,$end-$append); 394 } else { 395 $snippets[] = utf8_substr($text,$start,$end-$start); 396 } 397 398 // set $offset for next match attempt 399 // continue matching after the current match 400 // if the current match is not the longest possible match starting at the current offset 401 // this prevents further matching of this snippet but for possible matches of length 402 // smaller than match length + context (at least 50 characters) this match is part of the context 403 $utf8_offset = $utf8_idx + $utf8_len; 404 $offset = $idx + strlen(utf8_substr($text,$utf8_idx,$utf8_len)); 405 $offset = utf8_correctIdx($text,$offset); 406 } 407 408 $m = "\1"; 409 $snippets = preg_replace('/'.$re1.'/iu',$m.'$1'.$m,$snippets); 410 $snippet = preg_replace('/'.$m.'([^'.$m.']*?)'.$m.'/iu','<strong class="search_hit">$1</strong>',hsc(join('... ',$snippets))); 411 412 $evdata['snippet'] = $snippet; 413 } 414 $evt->advise_after(); 415 unset($evt); 416 417 return $evdata['snippet']; 418} 419 420/** 421 * Wraps a search term in regex boundary checks. 422 * 423 * @param string $term 424 * @return string 425 */ 426function ft_snippet_re_preprocess($term) { 427 // do not process asian terms where word boundaries are not explicit 428 if(preg_match('/'.IDX_ASIAN.'/u',$term)){ 429 return $term; 430 } 431 432 if (UTF8_PROPERTYSUPPORT) { 433 // unicode word boundaries 434 // see http://stackoverflow.com/a/2449017/172068 435 $BL = '(?<!\pL)'; 436 $BR = '(?!\pL)'; 437 } else { 438 // not as correct as above, but at least won't break 439 $BL = '\b'; 440 $BR = '\b'; 441 } 442 443 if(substr($term,0,2) == '\\*'){ 444 $term = substr($term,2); 445 }else{ 446 $term = $BL.$term; 447 } 448 449 if(substr($term,-2,2) == '\\*'){ 450 $term = substr($term,0,-2); 451 }else{ 452 $term = $term.$BR; 453 } 454 455 if($term == $BL || $term == $BR || $term == $BL.$BR) $term = ''; 456 return $term; 457} 458 459/** 460 * Combine found documents and sum up their scores 461 * 462 * This function is used to combine searched words with a logical 463 * AND. Only documents available in all arrays are returned. 464 * 465 * based upon PEAR's PHP_Compat function for array_intersect_key() 466 * 467 * @param array $args An array of page arrays 468 * @return array 469 */ 470function ft_resultCombine($args){ 471 $array_count = count($args); 472 if($array_count == 1){ 473 return $args[0]; 474 } 475 476 $result = array(); 477 if ($array_count > 1) { 478 foreach ($args[0] as $key => $value) { 479 $result[$key] = $value; 480 for ($i = 1; $i !== $array_count; $i++) { 481 if (!isset($args[$i][$key])) { 482 unset($result[$key]); 483 break; 484 } 485 $result[$key] += $args[$i][$key]; 486 } 487 } 488 } 489 return $result; 490} 491 492/** 493 * Unites found documents and sum up their scores 494 * 495 * based upon ft_resultCombine() function 496 * 497 * @param array $args An array of page arrays 498 * @return array 499 * 500 * @author Kazutaka Miyasaka <kazmiya@gmail.com> 501 */ 502function ft_resultUnite($args) { 503 $array_count = count($args); 504 if ($array_count === 1) { 505 return $args[0]; 506 } 507 508 $result = $args[0]; 509 for ($i = 1; $i !== $array_count; $i++) { 510 foreach (array_keys($args[$i]) as $id) { 511 $result[$id] += $args[$i][$id]; 512 } 513 } 514 return $result; 515} 516 517/** 518 * Computes the difference of documents using page id for comparison 519 * 520 * nearly identical to PHP5's array_diff_key() 521 * 522 * @param array $args An array of page arrays 523 * @return array 524 * 525 * @author Kazutaka Miyasaka <kazmiya@gmail.com> 526 */ 527function ft_resultComplement($args) { 528 $array_count = count($args); 529 if ($array_count === 1) { 530 return $args[0]; 531 } 532 533 $result = $args[0]; 534 foreach (array_keys($result) as $id) { 535 for ($i = 1; $i !== $array_count; $i++) { 536 if (isset($args[$i][$id])) unset($result[$id]); 537 } 538 } 539 return $result; 540} 541 542/** 543 * Parses a search query and builds an array of search formulas 544 * 545 * @author Andreas Gohr <andi@splitbrain.org> 546 * @author Kazutaka Miyasaka <kazmiya@gmail.com> 547 * 548 * @param Doku_Indexer $Indexer 549 * @param string $query search query 550 * @return array of search formulas 551 */ 552function ft_queryParser($Indexer, $query){ 553 /** 554 * parse a search query and transform it into intermediate representation 555 * 556 * in a search query, you can use the following expressions: 557 * 558 * words: 559 * include 560 * -exclude 561 * phrases: 562 * "phrase to be included" 563 * -"phrase you want to exclude" 564 * namespaces: 565 * @include:namespace (or ns:include:namespace) 566 * ^exclude:namespace (or -ns:exclude:namespace) 567 * groups: 568 * () 569 * -() 570 * operators: 571 * and ('and' is the default operator: you can always omit this) 572 * or (or pipe symbol '|', lower precedence than 'and') 573 * 574 * e.g. a query [ aa "bb cc" @dd:ee ] means "search pages which contain 575 * a word 'aa', a phrase 'bb cc' and are within a namespace 'dd:ee'". 576 * this query is equivalent to [ -(-aa or -"bb cc" or -ns:dd:ee) ] 577 * as long as you don't mind hit counts. 578 * 579 * intermediate representation consists of the following parts: 580 * 581 * ( ) - group 582 * AND - logical and 583 * OR - logical or 584 * NOT - logical not 585 * W+:, W-:, W_: - word (underscore: no need to highlight) 586 * P+:, P-: - phrase (minus sign: logically in NOT group) 587 * N+:, N-: - namespace 588 */ 589 $parsed_query = ''; 590 $parens_level = 0; 591 $terms = preg_split('/(-?".*?")/u', utf8_strtolower($query), -1, PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY); 592 593 foreach ($terms as $term) { 594 $parsed = ''; 595 if (preg_match('/^(-?)"(.+)"$/u', $term, $matches)) { 596 // phrase-include and phrase-exclude 597 $not = $matches[1] ? 'NOT' : ''; 598 $parsed = $not.ft_termParser($Indexer, $matches[2], false, true); 599 } else { 600 // fix incomplete phrase 601 $term = str_replace('"', ' ', $term); 602 603 // fix parentheses 604 $term = str_replace(')' , ' ) ', $term); 605 $term = str_replace('(' , ' ( ', $term); 606 $term = str_replace('- (', ' -(', $term); 607 608 // treat pipe symbols as 'OR' operators 609 $term = str_replace('|', ' or ', $term); 610 611 // treat ideographic spaces (U+3000) as search term separators 612 // FIXME: some more separators? 613 $term = preg_replace('/[ \x{3000}]+/u', ' ', $term); 614 $term = trim($term); 615 if ($term === '') continue; 616 617 $tokens = explode(' ', $term); 618 foreach ($tokens as $token) { 619 if ($token === '(') { 620 // parenthesis-include-open 621 $parsed .= '('; 622 ++$parens_level; 623 } elseif ($token === '-(') { 624 // parenthesis-exclude-open 625 $parsed .= 'NOT('; 626 ++$parens_level; 627 } elseif ($token === ')') { 628 // parenthesis-any-close 629 if ($parens_level === 0) continue; 630 $parsed .= ')'; 631 $parens_level--; 632 } elseif ($token === 'and') { 633 // logical-and (do nothing) 634 } elseif ($token === 'or') { 635 // logical-or 636 $parsed .= 'OR'; 637 } elseif (preg_match('/^(?:\^|-ns:)(.+)$/u', $token, $matches)) { 638 // namespace-exclude 639 $parsed .= 'NOT(N+:'.$matches[1].')'; 640 } elseif (preg_match('/^(?:@|ns:)(.+)$/u', $token, $matches)) { 641 // namespace-include 642 $parsed .= '(N+:'.$matches[1].')'; 643 } elseif (preg_match('/^-(.+)$/', $token, $matches)) { 644 // word-exclude 645 $parsed .= 'NOT('.ft_termParser($Indexer, $matches[1]).')'; 646 } else { 647 // word-include 648 $parsed .= ft_termParser($Indexer, $token); 649 } 650 } 651 } 652 $parsed_query .= $parsed; 653 } 654 655 // cleanup (very sensitive) 656 $parsed_query .= str_repeat(')', $parens_level); 657 do { 658 $parsed_query_old = $parsed_query; 659 $parsed_query = preg_replace('/(NOT)?\(\)/u', '', $parsed_query); 660 } while ($parsed_query !== $parsed_query_old); 661 $parsed_query = preg_replace('/(NOT|OR)+\)/u', ')' , $parsed_query); 662 $parsed_query = preg_replace('/(OR)+/u' , 'OR' , $parsed_query); 663 $parsed_query = preg_replace('/\(OR/u' , '(' , $parsed_query); 664 $parsed_query = preg_replace('/^OR|OR$/u' , '' , $parsed_query); 665 $parsed_query = preg_replace('/\)(NOT)?\(/u' , ')AND$1(', $parsed_query); 666 667 // adjustment: make highlightings right 668 $parens_level = 0; 669 $notgrp_levels = array(); 670 $parsed_query_new = ''; 671 $tokens = preg_split('/(NOT\(|[()])/u', $parsed_query, -1, PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY); 672 foreach ($tokens as $token) { 673 if ($token === 'NOT(') { 674 $notgrp_levels[] = ++$parens_level; 675 } elseif ($token === '(') { 676 ++$parens_level; 677 } elseif ($token === ')') { 678 if ($parens_level-- === end($notgrp_levels)) array_pop($notgrp_levels); 679 } elseif (count($notgrp_levels) % 2 === 1) { 680 // turn highlight-flag off if terms are logically in "NOT" group 681 $token = preg_replace('/([WPN])\+\:/u', '$1-:', $token); 682 } 683 $parsed_query_new .= $token; 684 } 685 $parsed_query = $parsed_query_new; 686 687 /** 688 * convert infix notation string into postfix (Reverse Polish notation) array 689 * by Shunting-yard algorithm 690 * 691 * see: http://en.wikipedia.org/wiki/Reverse_Polish_notation 692 * see: http://en.wikipedia.org/wiki/Shunting-yard_algorithm 693 */ 694 $parsed_ary = array(); 695 $ope_stack = array(); 696 $ope_precedence = array(')' => 1, 'OR' => 2, 'AND' => 3, 'NOT' => 4, '(' => 5); 697 $ope_regex = '/([()]|OR|AND|NOT)/u'; 698 699 $tokens = preg_split($ope_regex, $parsed_query, -1, PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY); 700 foreach ($tokens as $token) { 701 if (preg_match($ope_regex, $token)) { 702 // operator 703 $last_ope = end($ope_stack); 704 while ($last_ope !== false && $ope_precedence[$token] <= $ope_precedence[$last_ope] && $last_ope != '(') { 705 $parsed_ary[] = array_pop($ope_stack); 706 $last_ope = end($ope_stack); 707 } 708 if ($token == ')') { 709 array_pop($ope_stack); // this array_pop always deletes '(' 710 } else { 711 $ope_stack[] = $token; 712 } 713 } else { 714 // operand 715 $token_decoded = str_replace(array('OP', 'CP'), array('(', ')'), $token); 716 $parsed_ary[] = $token_decoded; 717 } 718 } 719 $parsed_ary = array_values(array_merge($parsed_ary, array_reverse($ope_stack))); 720 721 // cleanup: each double "NOT" in RPN array actually does nothing 722 $parsed_ary_count = count($parsed_ary); 723 for ($i = 1; $i < $parsed_ary_count; ++$i) { 724 if ($parsed_ary[$i] === 'NOT' && $parsed_ary[$i - 1] === 'NOT') { 725 unset($parsed_ary[$i], $parsed_ary[$i - 1]); 726 } 727 } 728 $parsed_ary = array_values($parsed_ary); 729 730 // build return value 731 $q = array(); 732 $q['query'] = $query; 733 $q['parsed_str'] = $parsed_query; 734 $q['parsed_ary'] = $parsed_ary; 735 736 foreach ($q['parsed_ary'] as $token) { 737 if ($token[2] !== ':') continue; 738 $body = substr($token, 3); 739 740 switch (substr($token, 0, 3)) { 741 case 'N+:': 742 $q['ns'][] = $body; // for backward compatibility 743 break; 744 case 'N-:': 745 $q['notns'][] = $body; // for backward compatibility 746 break; 747 case 'W_:': 748 $q['words'][] = $body; 749 break; 750 case 'W-:': 751 $q['words'][] = $body; 752 $q['not'][] = $body; // for backward compatibility 753 break; 754 case 'W+:': 755 $q['words'][] = $body; 756 $q['highlight'][] = $body; 757 $q['and'][] = $body; // for backward compatibility 758 break; 759 case 'P-:': 760 $q['phrases'][] = $body; 761 break; 762 case 'P+:': 763 $q['phrases'][] = $body; 764 $q['highlight'][] = $body; 765 break; 766 } 767 } 768 foreach (array('words', 'phrases', 'highlight', 'ns', 'notns', 'and', 'not') as $key) { 769 $q[$key] = empty($q[$key]) ? array() : array_values(array_unique($q[$key])); 770 } 771 772 return $q; 773} 774 775/** 776 * Transforms given search term into intermediate representation 777 * 778 * This function is used in ft_queryParser() and not for general purpose use. 779 * 780 * @author Kazutaka Miyasaka <kazmiya@gmail.com> 781 * 782 * @param Doku_Indexer $Indexer 783 * @param string $term 784 * @param bool $consider_asian 785 * @param bool $phrase_mode 786 * @return string 787 */ 788function ft_termParser($Indexer, $term, $consider_asian = true, $phrase_mode = false) { 789 $parsed = ''; 790 if ($consider_asian) { 791 // successive asian characters need to be searched as a phrase 792 $words = preg_split('/('.IDX_ASIAN.'+)/u', $term, -1, PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY); 793 foreach ($words as $word) { 794 $phrase_mode = $phrase_mode ? true : preg_match('/'.IDX_ASIAN.'/u', $word); 795 $parsed .= ft_termParser($Indexer, $word, false, $phrase_mode); 796 } 797 } else { 798 $term_noparen = str_replace(array('(', ')'), ' ', $term); 799 $words = $Indexer->tokenizer($term_noparen, true); 800 801 // W_: no need to highlight 802 if (empty($words)) { 803 $parsed = '()'; // important: do not remove 804 } elseif ($words[0] === $term) { 805 $parsed = '(W+:'.$words[0].')'; 806 } elseif ($phrase_mode) { 807 $term_encoded = str_replace(array('(', ')'), array('OP', 'CP'), $term); 808 $parsed = '((W_:'.implode(')(W_:', $words).')(P+:'.$term_encoded.'))'; 809 } else { 810 $parsed = '((W+:'.implode(')(W+:', $words).'))'; 811 } 812 } 813 return $parsed; 814} 815 816//Setup VIM: ex: et ts=4 : 817