123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959 |
- <?php
- /**
- * DokuWiki fulltextsearch functions using the index
- *
- * @license GPL 2 (http://www.gnu.org/licenses/gpl.html)
- * @author Andreas Gohr <andi@splitbrain.org>
- */
- use dokuwiki\Extension\Event;
- use dokuwiki\Utf8\Clean;
- use dokuwiki\Utf8\PhpString;
- use dokuwiki\Utf8\Sort;
- /**
- * create snippets for the first few results only
- */
- if(!defined('FT_SNIPPET_NUMBER')) define('FT_SNIPPET_NUMBER',15);
- /**
- * The fulltext search
- *
- * Returns a list of matching documents for the given query
- *
- * refactored into ft_pageSearch(), _ft_pageSearch() and trigger_event()
- *
- * @param string $query
- * @param array $highlight
- * @param string $sort
- * @param int|string $after only show results with mtime after this date, accepts timestap or strtotime arguments
- * @param int|string $before only show results with mtime before this date, accepts timestap or strtotime arguments
- *
- * @return array
- */
- function ft_pageSearch($query,&$highlight, $sort = null, $after = null, $before = null){
- if ($sort === null) {
- $sort = 'hits';
- }
- $data = [
- 'query' => $query,
- 'sort' => $sort,
- 'after' => $after,
- 'before' => $before
- ];
- $data['highlight'] =& $highlight;
- return Event::createAndTrigger('SEARCH_QUERY_FULLPAGE', $data, '_ft_pageSearch');
- }
- /**
- * Returns a list of matching documents for the given query
- *
- * @author Andreas Gohr <andi@splitbrain.org>
- * @author Kazutaka Miyasaka <kazmiya@gmail.com>
- *
- * @param array $data event data
- * @return array matching documents
- */
- function _ft_pageSearch(&$data) {
- $Indexer = idx_get_indexer();
- // parse the given query
- $q = ft_queryParser($Indexer, $data['query']);
- $data['highlight'] = $q['highlight'];
- if (empty($q['parsed_ary'])) return array();
- // lookup all words found in the query
- $lookup = $Indexer->lookup($q['words']);
- // get all pages in this dokuwiki site (!: includes nonexistent pages)
- $pages_all = array();
- foreach ($Indexer->getPages() as $id) {
- $pages_all[$id] = 0; // base: 0 hit
- }
- // process the query
- $stack = array();
- foreach ($q['parsed_ary'] as $token) {
- switch (substr($token, 0, 3)) {
- case 'W+:':
- case 'W-:':
- case 'W_:': // word
- $word = substr($token, 3);
- if(isset($lookup[$word])) {
- $stack[] = (array)$lookup[$word];
- }
- break;
- case 'P+:':
- case 'P-:': // phrase
- $phrase = substr($token, 3);
- // since phrases are always parsed as ((W1)(W2)...(P)),
- // the end($stack) always points the pages that contain
- // all words in this phrase
- $pages = end($stack);
- $pages_matched = array();
- foreach(array_keys($pages) as $id){
- $evdata = array(
- 'id' => $id,
- 'phrase' => $phrase,
- 'text' => rawWiki($id)
- );
- $evt = new Event('FULLTEXT_PHRASE_MATCH',$evdata);
- if ($evt->advise_before() && $evt->result !== true) {
- $text = PhpString::strtolower($evdata['text']);
- if (strpos($text, $phrase) !== false) {
- $evt->result = true;
- }
- }
- $evt->advise_after();
- if ($evt->result === true) {
- $pages_matched[$id] = 0; // phrase: always 0 hit
- }
- }
- $stack[] = $pages_matched;
- break;
- case 'N+:':
- case 'N-:': // namespace
- $ns = cleanID(substr($token, 3)) . ':';
- $pages_matched = array();
- foreach (array_keys($pages_all) as $id) {
- if (strpos($id, $ns) === 0) {
- $pages_matched[$id] = 0; // namespace: always 0 hit
- }
- }
- $stack[] = $pages_matched;
- break;
- case 'AND': // and operation
- list($pages1, $pages2) = array_splice($stack, -2);
- $stack[] = ft_resultCombine(array($pages1, $pages2));
- break;
- case 'OR': // or operation
- list($pages1, $pages2) = array_splice($stack, -2);
- $stack[] = ft_resultUnite(array($pages1, $pages2));
- break;
- case 'NOT': // not operation (unary)
- $pages = array_pop($stack);
- $stack[] = ft_resultComplement(array($pages_all, $pages));
- break;
- }
- }
- $docs = array_pop($stack);
- if (empty($docs)) return array();
- // check: settings, acls, existence
- foreach (array_keys($docs) as $id) {
- if (isHiddenPage($id) || auth_quickaclcheck($id) < AUTH_READ || !page_exists($id, '', false)) {
- unset($docs[$id]);
- }
- }
- $docs = _ft_filterResultsByTime($docs, $data['after'], $data['before']);
- if ($data['sort'] === 'mtime') {
- uksort($docs, 'ft_pagemtimesorter');
- } else {
- // sort docs by count
- uksort($docs, 'ft_pagesorter');
- arsort($docs);
- }
- return $docs;
- }
- /**
- * Returns the backlinks for a given page
- *
- * Uses the metadata index.
- *
- * @param string $id The id for which links shall be returned
- * @param bool $ignore_perms Ignore the fact that pages are hidden or read-protected
- * @return array The pages that contain links to the given page
- */
- function ft_backlinks($id, $ignore_perms = false){
- $result = idx_get_indexer()->lookupKey('relation_references', $id);
- if(!count($result)) return $result;
- // check ACL permissions
- foreach(array_keys($result) as $idx){
- if(($ignore_perms !== true && (
- isHiddenPage($result[$idx]) || auth_quickaclcheck($result[$idx]) < AUTH_READ
- )) || !page_exists($result[$idx], '', false)){
- unset($result[$idx]);
- }
- }
- Sort::sort($result);
- return $result;
- }
- /**
- * Returns the pages that use a given media file
- *
- * Uses the relation media metadata property and the metadata index.
- *
- * Note that before 2013-07-31 the second parameter was the maximum number of results and
- * permissions were ignored. That's why the parameter is now checked to be explicitely set
- * to true (with type bool) in order to be compatible with older uses of the function.
- *
- * @param string $id The media id to look for
- * @param bool $ignore_perms Ignore hidden pages and acls (optional, default: false)
- * @return array A list of pages that use the given media file
- */
- function ft_mediause($id, $ignore_perms = false){
- $result = idx_get_indexer()->lookupKey('relation_media', $id);
- if(!count($result)) return $result;
- // check ACL permissions
- foreach(array_keys($result) as $idx){
- if(($ignore_perms !== true && (
- isHiddenPage($result[$idx]) || auth_quickaclcheck($result[$idx]) < AUTH_READ
- )) || !page_exists($result[$idx], '', false)){
- unset($result[$idx]);
- }
- }
- Sort::sort($result);
- return $result;
- }
- /**
- * Quicksearch for pagenames
- *
- * By default it only matches the pagename and ignores the
- * namespace. This can be changed with the second parameter.
- * The third parameter allows to search in titles as well.
- *
- * The function always returns titles as well
- *
- * @triggers SEARCH_QUERY_PAGELOOKUP
- * @author Andreas Gohr <andi@splitbrain.org>
- * @author Adrian Lang <lang@cosmocode.de>
- *
- * @param string $id page id
- * @param bool $in_ns match against namespace as well?
- * @param bool $in_title search in title?
- * @param int|string $after only show results with mtime after this date, accepts timestap or strtotime arguments
- * @param int|string $before only show results with mtime before this date, accepts timestap or strtotime arguments
- *
- * @return string[]
- */
- function ft_pageLookup($id, $in_ns=false, $in_title=false, $after = null, $before = null){
- $data = [
- 'id' => $id,
- 'in_ns' => $in_ns,
- 'in_title' => $in_title,
- 'after' => $after,
- 'before' => $before
- ];
- $data['has_titles'] = true; // for plugin backward compatibility check
- return Event::createAndTrigger('SEARCH_QUERY_PAGELOOKUP', $data, '_ft_pageLookup');
- }
- /**
- * Returns list of pages as array(pageid => First Heading)
- *
- * @param array &$data event data
- * @return string[]
- */
- function _ft_pageLookup(&$data){
- // split out original parameters
- $id = $data['id'];
- $Indexer = idx_get_indexer();
- $parsedQuery = ft_queryParser($Indexer, $id);
- if (count($parsedQuery['ns']) > 0) {
- $ns = cleanID($parsedQuery['ns'][0]) . ':';
- $id = implode(' ', $parsedQuery['highlight']);
- }
- if (count($parsedQuery['notns']) > 0) {
- $notns = cleanID($parsedQuery['notns'][0]) . ':';
- $id = implode(' ', $parsedQuery['highlight']);
- }
- $in_ns = $data['in_ns'];
- $in_title = $data['in_title'];
- $cleaned = cleanID($id);
- $Indexer = idx_get_indexer();
- $page_idx = $Indexer->getPages();
- $pages = array();
- if ($id !== '' && $cleaned !== '') {
- foreach ($page_idx as $p_id) {
- if ((strpos($in_ns ? $p_id : noNSorNS($p_id), $cleaned) !== false)) {
- if (!isset($pages[$p_id]))
- $pages[$p_id] = p_get_first_heading($p_id, METADATA_DONT_RENDER);
- }
- }
- if ($in_title) {
- foreach ($Indexer->lookupKey('title', $id, '_ft_pageLookupTitleCompare') as $p_id) {
- if (!isset($pages[$p_id]))
- $pages[$p_id] = p_get_first_heading($p_id, METADATA_DONT_RENDER);
- }
- }
- }
- if (isset($ns)) {
- foreach (array_keys($pages) as $p_id) {
- if (strpos($p_id, $ns) !== 0) {
- unset($pages[$p_id]);
- }
- }
- }
- if (isset($notns)) {
- foreach (array_keys($pages) as $p_id) {
- if (strpos($p_id, $notns) === 0) {
- unset($pages[$p_id]);
- }
- }
- }
- // discard hidden pages
- // discard nonexistent pages
- // check ACL permissions
- foreach(array_keys($pages) as $idx){
- if(!isVisiblePage($idx) || !page_exists($idx) ||
- auth_quickaclcheck($idx) < AUTH_READ) {
- unset($pages[$idx]);
- }
- }
- $pages = _ft_filterResultsByTime($pages, $data['after'], $data['before']);
- uksort($pages,'ft_pagesorter');
- return $pages;
- }
- /**
- * @param array $results search results in the form pageid => value
- * @param int|string $after only returns results with mtime after this date, accepts timestap or strtotime arguments
- * @param int|string $before only returns results with mtime after this date, accepts timestap or strtotime arguments
- *
- * @return array
- */
- function _ft_filterResultsByTime(array $results, $after, $before) {
- if ($after || $before) {
- $after = is_int($after) ? $after : strtotime($after);
- $before = is_int($before) ? $before : strtotime($before);
- foreach ($results as $id => $value) {
- $mTime = filemtime(wikiFN($id));
- if ($after && $after > $mTime) {
- unset($results[$id]);
- continue;
- }
- if ($before && $before < $mTime) {
- unset($results[$id]);
- }
- }
- }
- return $results;
- }
- /**
- * Tiny helper function for comparing the searched title with the title
- * from the search index. This function is a wrapper around stripos with
- * adapted argument order and return value.
- *
- * @param string $search searched title
- * @param string $title title from index
- * @return bool
- */
- function _ft_pageLookupTitleCompare($search, $title) {
- if (Clean::isASCII($search)) {
- $pos = stripos($title, $search);
- } else {
- $pos = PhpString::strpos(
- PhpString::strtolower($title),
- PhpString::strtolower($search)
- );
- }
- return $pos !== false;
- }
- /**
- * Sort pages based on their namespace level first, then on their string
- * values. This makes higher hierarchy pages rank higher than lower hierarchy
- * pages.
- *
- * @param string $a
- * @param string $b
- * @return int Returns < 0 if $a is less than $b; > 0 if $a is greater than $b, and 0 if they are equal.
- */
- function ft_pagesorter($a, $b){
- $ac = count(explode(':',$a));
- $bc = count(explode(':',$b));
- if($ac < $bc){
- return -1;
- }elseif($ac > $bc){
- return 1;
- }
- return Sort::strcmp($a,$b);
- }
- /**
- * Sort pages by their mtime, from newest to oldest
- *
- * @param string $a
- * @param string $b
- *
- * @return int Returns < 0 if $a is newer than $b, > 0 if $b is newer than $a and 0 if they are of the same age
- */
- function ft_pagemtimesorter($a, $b) {
- $mtimeA = filemtime(wikiFN($a));
- $mtimeB = filemtime(wikiFN($b));
- return $mtimeB - $mtimeA;
- }
- /**
- * Creates a snippet extract
- *
- * @author Andreas Gohr <andi@splitbrain.org>
- * @triggers FULLTEXT_SNIPPET_CREATE
- *
- * @param string $id page id
- * @param array $highlight
- * @return mixed
- */
- function ft_snippet($id,$highlight){
- $text = rawWiki($id);
- $text = str_replace("\xC2\xAD",'',$text); // remove soft-hyphens
- $evdata = array(
- 'id' => $id,
- 'text' => &$text,
- 'highlight' => &$highlight,
- 'snippet' => '',
- );
- $evt = new Event('FULLTEXT_SNIPPET_CREATE',$evdata);
- if ($evt->advise_before()) {
- $match = array();
- $snippets = array();
- $utf8_offset = $offset = $end = 0;
- $len = PhpString::strlen($text);
- // build a regexp from the phrases to highlight
- $re1 = '(' .
- join(
- '|',
- array_map(
- 'ft_snippet_re_preprocess',
- array_map(
- 'preg_quote_cb',
- array_filter((array) $highlight)
- )
- )
- ) .
- ')';
- $re2 = "$re1.{0,75}(?!\\1)$re1";
- $re3 = "$re1.{0,45}(?!\\1)$re1.{0,45}(?!\\1)(?!\\2)$re1";
- for ($cnt=4; $cnt--;) {
- if (0) {
- } else if (preg_match('/'.$re3.'/iu',$text,$match,PREG_OFFSET_CAPTURE,$offset)) {
- } else if (preg_match('/'.$re2.'/iu',$text,$match,PREG_OFFSET_CAPTURE,$offset)) {
- } else if (preg_match('/'.$re1.'/iu',$text,$match,PREG_OFFSET_CAPTURE,$offset)) {
- } else {
- break;
- }
- list($str,$idx) = $match[0];
- // convert $idx (a byte offset) into a utf8 character offset
- $utf8_idx = PhpString::strlen(substr($text,0,$idx));
- $utf8_len = PhpString::strlen($str);
- // establish context, 100 bytes surrounding the match string
- // first look to see if we can go 100 either side,
- // then drop to 50 adding any excess if the other side can't go to 50,
- $pre = min($utf8_idx-$utf8_offset,100);
- $post = min($len-$utf8_idx-$utf8_len,100);
- if ($pre>50 && $post>50) {
- $pre = $post = 50;
- } else if ($pre>50) {
- $pre = min($pre,100-$post);
- } else if ($post>50) {
- $post = min($post, 100-$pre);
- } else if ($offset == 0) {
- // both are less than 50, means the context is the whole string
- // make it so and break out of this loop - there is no need for the
- // complex snippet calculations
- $snippets = array($text);
- break;
- }
- // establish context start and end points, try to append to previous
- // context if possible
- $start = $utf8_idx - $pre;
- $append = ($start < $end) ? $end : false; // still the end of the previous context snippet
- $end = $utf8_idx + $utf8_len + $post; // now set it to the end of this context
- if ($append) {
- $snippets[count($snippets)-1] .= PhpString::substr($text,$append,$end-$append);
- } else {
- $snippets[] = PhpString::substr($text,$start,$end-$start);
- }
- // set $offset for next match attempt
- // continue matching after the current match
- // if the current match is not the longest possible match starting at the current offset
- // this prevents further matching of this snippet but for possible matches of length
- // smaller than match length + context (at least 50 characters) this match is part of the context
- $utf8_offset = $utf8_idx + $utf8_len;
- $offset = $idx + strlen(PhpString::substr($text,$utf8_idx,$utf8_len));
- $offset = Clean::correctIdx($text,$offset);
- }
- $m = "\1";
- $snippets = preg_replace('/'.$re1.'/iu',$m.'$1'.$m,$snippets);
- $snippet = preg_replace(
- '/' . $m . '([^' . $m . ']*?)' . $m . '/iu',
- '<strong class="search_hit">$1</strong>',
- hsc(join('... ', $snippets))
- );
- $evdata['snippet'] = $snippet;
- }
- $evt->advise_after();
- unset($evt);
- return $evdata['snippet'];
- }
- /**
- * Wraps a search term in regex boundary checks.
- *
- * @param string $term
- * @return string
- */
- function ft_snippet_re_preprocess($term) {
- // do not process asian terms where word boundaries are not explicit
- if(\dokuwiki\Utf8\Asian::isAsianWords($term)) return $term;
- if (UTF8_PROPERTYSUPPORT) {
- // unicode word boundaries
- // see http://stackoverflow.com/a/2449017/172068
- $BL = '(?<!\pL)';
- $BR = '(?!\pL)';
- } else {
- // not as correct as above, but at least won't break
- $BL = '\b';
- $BR = '\b';
- }
- if(substr($term,0,2) == '\\*'){
- $term = substr($term,2);
- }else{
- $term = $BL.$term;
- }
- if(substr($term,-2,2) == '\\*'){
- $term = substr($term,0,-2);
- }else{
- $term = $term.$BR;
- }
- if($term == $BL || $term == $BR || $term == $BL.$BR) $term = '';
- return $term;
- }
- /**
- * Combine found documents and sum up their scores
- *
- * This function is used to combine searched words with a logical
- * AND. Only documents available in all arrays are returned.
- *
- * based upon PEAR's PHP_Compat function for array_intersect_key()
- *
- * @param array $args An array of page arrays
- * @return array
- */
- function ft_resultCombine($args){
- $array_count = count($args);
- if($array_count == 1){
- return $args[0];
- }
- $result = array();
- if ($array_count > 1) {
- foreach ($args[0] as $key => $value) {
- $result[$key] = $value;
- for ($i = 1; $i !== $array_count; $i++) {
- if (!isset($args[$i][$key])) {
- unset($result[$key]);
- break;
- }
- $result[$key] += $args[$i][$key];
- }
- }
- }
- return $result;
- }
- /**
- * Unites found documents and sum up their scores
- *
- * based upon ft_resultCombine() function
- *
- * @param array $args An array of page arrays
- * @return array
- *
- * @author Kazutaka Miyasaka <kazmiya@gmail.com>
- */
- function ft_resultUnite($args) {
- $array_count = count($args);
- if ($array_count === 1) {
- return $args[0];
- }
- $result = $args[0];
- for ($i = 1; $i !== $array_count; $i++) {
- foreach (array_keys($args[$i]) as $id) {
- $result[$id] += $args[$i][$id];
- }
- }
- return $result;
- }
- /**
- * Computes the difference of documents using page id for comparison
- *
- * nearly identical to PHP5's array_diff_key()
- *
- * @param array $args An array of page arrays
- * @return array
- *
- * @author Kazutaka Miyasaka <kazmiya@gmail.com>
- */
- function ft_resultComplement($args) {
- $array_count = count($args);
- if ($array_count === 1) {
- return $args[0];
- }
- $result = $args[0];
- foreach (array_keys($result) as $id) {
- for ($i = 1; $i !== $array_count; $i++) {
- if (isset($args[$i][$id])) unset($result[$id]);
- }
- }
- return $result;
- }
- /**
- * Parses a search query and builds an array of search formulas
- *
- * @author Andreas Gohr <andi@splitbrain.org>
- * @author Kazutaka Miyasaka <kazmiya@gmail.com>
- *
- * @param dokuwiki\Search\Indexer $Indexer
- * @param string $query search query
- * @return array of search formulas
- */
- function ft_queryParser($Indexer, $query){
- /**
- * parse a search query and transform it into intermediate representation
- *
- * in a search query, you can use the following expressions:
- *
- * words:
- * include
- * -exclude
- * phrases:
- * "phrase to be included"
- * -"phrase you want to exclude"
- * namespaces:
- * @include:namespace (or ns:include:namespace)
- * ^exclude:namespace (or -ns:exclude:namespace)
- * groups:
- * ()
- * -()
- * operators:
- * and ('and' is the default operator: you can always omit this)
- * or (or pipe symbol '|', lower precedence than 'and')
- *
- * e.g. a query [ aa "bb cc" @dd:ee ] means "search pages which contain
- * a word 'aa', a phrase 'bb cc' and are within a namespace 'dd:ee'".
- * this query is equivalent to [ -(-aa or -"bb cc" or -ns:dd:ee) ]
- * as long as you don't mind hit counts.
- *
- * intermediate representation consists of the following parts:
- *
- * ( ) - group
- * AND - logical and
- * OR - logical or
- * NOT - logical not
- * W+:, W-:, W_: - word (underscore: no need to highlight)
- * P+:, P-: - phrase (minus sign: logically in NOT group)
- * N+:, N-: - namespace
- */
- $parsed_query = '';
- $parens_level = 0;
- $terms = preg_split('/(-?".*?")/u', PhpString::strtolower($query),
- -1, PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY);
- foreach ($terms as $term) {
- $parsed = '';
- if (preg_match('/^(-?)"(.+)"$/u', $term, $matches)) {
- // phrase-include and phrase-exclude
- $not = $matches[1] ? 'NOT' : '';
- $parsed = $not.ft_termParser($Indexer, $matches[2], false, true);
- } else {
- // fix incomplete phrase
- $term = str_replace('"', ' ', $term);
- // fix parentheses
- $term = str_replace(')' , ' ) ', $term);
- $term = str_replace('(' , ' ( ', $term);
- $term = str_replace('- (', ' -(', $term);
- // treat pipe symbols as 'OR' operators
- $term = str_replace('|', ' or ', $term);
- // treat ideographic spaces (U+3000) as search term separators
- // FIXME: some more separators?
- $term = preg_replace('/[ \x{3000}]+/u', ' ', $term);
- $term = trim($term);
- if ($term === '') continue;
- $tokens = explode(' ', $term);
- foreach ($tokens as $token) {
- if ($token === '(') {
- // parenthesis-include-open
- $parsed .= '(';
- ++$parens_level;
- } elseif ($token === '-(') {
- // parenthesis-exclude-open
- $parsed .= 'NOT(';
- ++$parens_level;
- } elseif ($token === ')') {
- // parenthesis-any-close
- if ($parens_level === 0) continue;
- $parsed .= ')';
- $parens_level--;
- } elseif ($token === 'and') {
- // logical-and (do nothing)
- } elseif ($token === 'or') {
- // logical-or
- $parsed .= 'OR';
- } elseif (preg_match('/^(?:\^|-ns:)(.+)$/u', $token, $matches)) {
- // namespace-exclude
- $parsed .= 'NOT(N+:'.$matches[1].')';
- } elseif (preg_match('/^(?:@|ns:)(.+)$/u', $token, $matches)) {
- // namespace-include
- $parsed .= '(N+:'.$matches[1].')';
- } elseif (preg_match('/^-(.+)$/', $token, $matches)) {
- // word-exclude
- $parsed .= 'NOT('.ft_termParser($Indexer, $matches[1]).')';
- } else {
- // word-include
- $parsed .= ft_termParser($Indexer, $token);
- }
- }
- }
- $parsed_query .= $parsed;
- }
- // cleanup (very sensitive)
- $parsed_query .= str_repeat(')', $parens_level);
- do {
- $parsed_query_old = $parsed_query;
- $parsed_query = preg_replace('/(NOT)?\(\)/u', '', $parsed_query);
- } while ($parsed_query !== $parsed_query_old);
- $parsed_query = preg_replace('/(NOT|OR)+\)/u', ')' , $parsed_query);
- $parsed_query = preg_replace('/(OR)+/u' , 'OR' , $parsed_query);
- $parsed_query = preg_replace('/\(OR/u' , '(' , $parsed_query);
- $parsed_query = preg_replace('/^OR|OR$/u' , '' , $parsed_query);
- $parsed_query = preg_replace('/\)(NOT)?\(/u' , ')AND$1(', $parsed_query);
- // adjustment: make highlightings right
- $parens_level = 0;
- $notgrp_levels = array();
- $parsed_query_new = '';
- $tokens = preg_split('/(NOT\(|[()])/u', $parsed_query, -1, PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY);
- foreach ($tokens as $token) {
- if ($token === 'NOT(') {
- $notgrp_levels[] = ++$parens_level;
- } elseif ($token === '(') {
- ++$parens_level;
- } elseif ($token === ')') {
- if ($parens_level-- === end($notgrp_levels)) array_pop($notgrp_levels);
- } elseif (count($notgrp_levels) % 2 === 1) {
- // turn highlight-flag off if terms are logically in "NOT" group
- $token = preg_replace('/([WPN])\+\:/u', '$1-:', $token);
- }
- $parsed_query_new .= $token;
- }
- $parsed_query = $parsed_query_new;
- /**
- * convert infix notation string into postfix (Reverse Polish notation) array
- * by Shunting-yard algorithm
- *
- * see: http://en.wikipedia.org/wiki/Reverse_Polish_notation
- * see: http://en.wikipedia.org/wiki/Shunting-yard_algorithm
- */
- $parsed_ary = array();
- $ope_stack = array();
- $ope_precedence = array(')' => 1, 'OR' => 2, 'AND' => 3, 'NOT' => 4, '(' => 5);
- $ope_regex = '/([()]|OR|AND|NOT)/u';
- $tokens = preg_split($ope_regex, $parsed_query, -1, PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY);
- foreach ($tokens as $token) {
- if (preg_match($ope_regex, $token)) {
- // operator
- $last_ope = end($ope_stack);
- while ($last_ope !== false && $ope_precedence[$token] <= $ope_precedence[$last_ope] && $last_ope != '(') {
- $parsed_ary[] = array_pop($ope_stack);
- $last_ope = end($ope_stack);
- }
- if ($token == ')') {
- array_pop($ope_stack); // this array_pop always deletes '('
- } else {
- $ope_stack[] = $token;
- }
- } else {
- // operand
- $token_decoded = str_replace(array('OP', 'CP'), array('(', ')'), $token);
- $parsed_ary[] = $token_decoded;
- }
- }
- $parsed_ary = array_values(array_merge($parsed_ary, array_reverse($ope_stack)));
- // cleanup: each double "NOT" in RPN array actually does nothing
- $parsed_ary_count = count($parsed_ary);
- for ($i = 1; $i < $parsed_ary_count; ++$i) {
- if ($parsed_ary[$i] === 'NOT' && $parsed_ary[$i - 1] === 'NOT') {
- unset($parsed_ary[$i], $parsed_ary[$i - 1]);
- }
- }
- $parsed_ary = array_values($parsed_ary);
- // build return value
- $q = array();
- $q['query'] = $query;
- $q['parsed_str'] = $parsed_query;
- $q['parsed_ary'] = $parsed_ary;
- foreach ($q['parsed_ary'] as $token) {
- if (strlen($token) < 3 || $token[2] !== ':') continue;
- $body = substr($token, 3);
- switch (substr($token, 0, 3)) {
- case 'N+:':
- $q['ns'][] = $body; // for backward compatibility
- break;
- case 'N-:':
- $q['notns'][] = $body; // for backward compatibility
- break;
- case 'W_:':
- $q['words'][] = $body;
- break;
- case 'W-:':
- $q['words'][] = $body;
- $q['not'][] = $body; // for backward compatibility
- break;
- case 'W+:':
- $q['words'][] = $body;
- $q['highlight'][] = $body;
- $q['and'][] = $body; // for backward compatibility
- break;
- case 'P-:':
- $q['phrases'][] = $body;
- break;
- case 'P+:':
- $q['phrases'][] = $body;
- $q['highlight'][] = $body;
- break;
- }
- }
- foreach (array('words', 'phrases', 'highlight', 'ns', 'notns', 'and', 'not') as $key) {
- $q[$key] = empty($q[$key]) ? array() : array_values(array_unique($q[$key]));
- }
- return $q;
- }
- /**
- * Transforms given search term into intermediate representation
- *
- * This function is used in ft_queryParser() and not for general purpose use.
- *
- * @author Kazutaka Miyasaka <kazmiya@gmail.com>
- *
- * @param dokuwiki\Search\Indexer $Indexer
- * @param string $term
- * @param bool $consider_asian
- * @param bool $phrase_mode
- * @return string
- */
- function ft_termParser($Indexer, $term, $consider_asian = true, $phrase_mode = false) {
- $parsed = '';
- if ($consider_asian) {
- // successive asian characters need to be searched as a phrase
- $words = \dokuwiki\Utf8\Asian::splitAsianWords($term);
- foreach ($words as $word) {
- $phrase_mode = $phrase_mode ? true : \dokuwiki\Utf8\Asian::isAsianWords($word);
- $parsed .= ft_termParser($Indexer, $word, false, $phrase_mode);
- }
- } else {
- $term_noparen = str_replace(array('(', ')'), ' ', $term);
- $words = $Indexer->tokenizer($term_noparen, true);
- // W_: no need to highlight
- if (empty($words)) {
- $parsed = '()'; // important: do not remove
- } elseif ($words[0] === $term) {
- $parsed = '(W+:'.$words[0].')';
- } elseif ($phrase_mode) {
- $term_encoded = str_replace(array('(', ')'), array('OP', 'CP'), $term);
- $parsed = '((W_:'.implode(')(W_:', $words).')(P+:'.$term_encoded.'))';
- } else {
- $parsed = '((W+:'.implode(')(W+:', $words).'))';
- }
- }
- return $parsed;
- }
- /**
- * Recreate a search query string based on parsed parts, doesn't support negated phrases and `OR` searches
- *
- * @param array $and
- * @param array $not
- * @param array $phrases
- * @param array $ns
- * @param array $notns
- *
- * @return string
- */
- function ft_queryUnparser_simple(array $and, array $not, array $phrases, array $ns, array $notns) {
- $query = implode(' ', $and);
- if (!empty($not)) {
- $query .= ' -' . implode(' -', $not);
- }
- if (!empty($phrases)) {
- $query .= ' "' . implode('" "', $phrases) . '"';
- }
- if (!empty($ns)) {
- $query .= ' @' . implode(' @', $ns);
- }
- if (!empty($notns)) {
- $query .= ' ^' . implode(' ^', $notns);
- }
- return $query;
- }
- //Setup VIM: ex: et ts=4 :
|