xref: /dokuwiki/inc/Search/MetadataSearch.php (revision fab81cc8d9f8ae6dbaa16254159007a2d6d9a8d0)
1fe2d1da1SSatoshi Sahara<?php
246b83514SSatoshi Sahara
3fe2d1da1SSatoshi Saharanamespace dokuwiki\Search;
4fe2d1da1SSatoshi Sahara
5fe2d1da1SSatoshi Saharause dokuwiki\Extension\Event;
686fc7283SSatoshi Saharause dokuwiki\Search\MetadataIndex;
7fe2d1da1SSatoshi Saharause dokuwiki\Search\QueryParser;
8fe2d1da1SSatoshi Sahara
9fe2d1da1SSatoshi Sahara/**
10fe2d1da1SSatoshi Sahara * Class DokuWiki Metadata Search
11fe2d1da1SSatoshi Sahara *
12fe2d1da1SSatoshi Sahara * @license    GPL 2 (http://www.gnu.org/licenses/gpl.html)
13fe2d1da1SSatoshi Sahara * @author     Andreas Gohr <andi@splitbrain.org>
14fe2d1da1SSatoshi Sahara */
15fe2d1da1SSatoshi Saharaclass MetadataSearch
16fe2d1da1SSatoshi Sahara{
17fe2d1da1SSatoshi Sahara    /**
18fe2d1da1SSatoshi Sahara     * Quicksearch for pagenames
19fe2d1da1SSatoshi Sahara     *
20fe2d1da1SSatoshi Sahara     * By default it only matches the pagename and ignores the namespace.
21fe2d1da1SSatoshi Sahara     * This can be changed with the second parameter.
22fe2d1da1SSatoshi Sahara     * The third parameter allows to search in titles as well.
23fe2d1da1SSatoshi Sahara     *
24fe2d1da1SSatoshi Sahara     * The function always returns titles as well
25fe2d1da1SSatoshi Sahara     *
26fe2d1da1SSatoshi Sahara     * @triggers SEARCH_QUERY_PAGELOOKUP
27fe2d1da1SSatoshi Sahara     * @author   Andreas Gohr <andi@splitbrain.org>
28fe2d1da1SSatoshi Sahara     * @author   Adrian Lang <lang@cosmocode.de>
29fe2d1da1SSatoshi Sahara     *
30fe2d1da1SSatoshi Sahara     * @param string     $id       page id
31fe2d1da1SSatoshi Sahara     * @param bool       $in_ns    match against namespace as well?
32fe2d1da1SSatoshi Sahara     * @param bool       $in_title search in title?
33fe2d1da1SSatoshi Sahara     * @param int|string $after    only show results with mtime after this date,
34fe2d1da1SSatoshi Sahara     *                             accepts timestap or strtotime arguments
35fe2d1da1SSatoshi Sahara     * @param int|string $before   only show results with mtime before this date,
36fe2d1da1SSatoshi Sahara     *                             accepts timestap or strtotime arguments
37fe2d1da1SSatoshi Sahara     *
38fe2d1da1SSatoshi Sahara     * @return string[]
39fe2d1da1SSatoshi Sahara     */
409329b002SSatoshi Sahara    public function pageLookup($id, $in_ns = false, $in_title = false, $after = null, $before = null)
41fe2d1da1SSatoshi Sahara    {
42fe2d1da1SSatoshi Sahara        $data = [
43fe2d1da1SSatoshi Sahara            'id' => $id,
44fe2d1da1SSatoshi Sahara            'in_ns' => $in_ns,
45fe2d1da1SSatoshi Sahara            'in_title' => $in_title,
46fe2d1da1SSatoshi Sahara            'after' => $after,
47fe2d1da1SSatoshi Sahara            'before' => $before
48fe2d1da1SSatoshi Sahara        ];
49fe2d1da1SSatoshi Sahara        $data['has_titles'] = true; // for plugin backward compatibility check
509329b002SSatoshi Sahara        $action = [$this, 'pageLookupCallBack'];
51fe2d1da1SSatoshi Sahara        return Event::createAndTrigger('SEARCH_QUERY_PAGELOOKUP', $data, $action);
52fe2d1da1SSatoshi Sahara    }
53fe2d1da1SSatoshi Sahara
54fe2d1da1SSatoshi Sahara    /**
55fe2d1da1SSatoshi Sahara     * Returns list of pages as array(pageid => First Heading)
56fe2d1da1SSatoshi Sahara     *
57fe2d1da1SSatoshi Sahara     * @param array $data  event data
58fe2d1da1SSatoshi Sahara     * @return string[]
59fe2d1da1SSatoshi Sahara     */
609329b002SSatoshi Sahara    public function pageLookupCallBack(&$data)
61fe2d1da1SSatoshi Sahara    {
62fe2d1da1SSatoshi Sahara        // split out original parameters
63fe2d1da1SSatoshi Sahara        $id = $data['id'];
649329b002SSatoshi Sahara        $parsedQuery = (new QueryParser)->convert($id);
65fe2d1da1SSatoshi Sahara
66fe2d1da1SSatoshi Sahara        if (count($parsedQuery['ns']) > 0) {
67fe2d1da1SSatoshi Sahara            $ns = cleanID($parsedQuery['ns'][0]) . ':';
68fe2d1da1SSatoshi Sahara            $id = implode(' ', $parsedQuery['highlight']);
69fe2d1da1SSatoshi Sahara        }
70*fab81cc8SSatoshi Sahara        if (count($parsedQuery['notns']) > 0) {
71*fab81cc8SSatoshi Sahara            $notns = cleanID($parsedQuery['notns'][0]) . ':';
72*fab81cc8SSatoshi Sahara            $id = implode(' ', $parsedQuery['highlight']);
73*fab81cc8SSatoshi Sahara        }
74fe2d1da1SSatoshi Sahara
75fe2d1da1SSatoshi Sahara        $in_ns    = $data['in_ns'];
76fe2d1da1SSatoshi Sahara        $in_title = $data['in_title'];
77fe2d1da1SSatoshi Sahara        $cleaned = cleanID($id);
78fe2d1da1SSatoshi Sahara
79a32da6ddSSatoshi Sahara        $MetadataIndex = new MetadataIndex();
8002361d2aSSatoshi Sahara        $page_idx = $MetadataIndex->getPages();
81*fab81cc8SSatoshi Sahara
82*fab81cc8SSatoshi Sahara        $pages = array();
83*fab81cc8SSatoshi Sahara        if ($id !== '' && $cleaned !== '') {
84fe2d1da1SSatoshi Sahara            foreach ($page_idx as $p_id) {
85fe2d1da1SSatoshi Sahara                if ((strpos($in_ns ? $p_id : noNSorNS($p_id), $cleaned) !== false)) {
86fe2d1da1SSatoshi Sahara                    if (!isset($pages[$p_id])) {
87fe2d1da1SSatoshi Sahara                        $pages[$p_id] = p_get_first_heading($p_id, METADATA_DONT_RENDER);
88fe2d1da1SSatoshi Sahara                    }
89fe2d1da1SSatoshi Sahara                }
90fe2d1da1SSatoshi Sahara            }
91fe2d1da1SSatoshi Sahara            if ($in_title) {
929329b002SSatoshi Sahara                $func = [$this, 'pageLookupTitleCompare'];
93be5c1ea2SSatoshi Sahara                foreach ($MetadataIndex->lookupKey('title', $id, $func) as $p_id) {
94fe2d1da1SSatoshi Sahara                    if (!isset($pages[$p_id])) {
95fe2d1da1SSatoshi Sahara                        $pages[$p_id] = p_get_first_heading($p_id, METADATA_DONT_RENDER);
96fe2d1da1SSatoshi Sahara                    }
97fe2d1da1SSatoshi Sahara                }
98fe2d1da1SSatoshi Sahara            }
99fe2d1da1SSatoshi Sahara        }
100fe2d1da1SSatoshi Sahara
101fe2d1da1SSatoshi Sahara        if (isset($ns)) {
102fe2d1da1SSatoshi Sahara            foreach (array_keys($pages) as $p_id) {
103fe2d1da1SSatoshi Sahara                if (strpos($p_id, $ns) !== 0) {
104fe2d1da1SSatoshi Sahara                    unset($pages[$p_id]);
105fe2d1da1SSatoshi Sahara                }
106fe2d1da1SSatoshi Sahara            }
107fe2d1da1SSatoshi Sahara        }
108*fab81cc8SSatoshi Sahara        if (isset($notns)) {
109*fab81cc8SSatoshi Sahara            foreach (array_keys($pages) as $p_id) {
110*fab81cc8SSatoshi Sahara                if (strpos($p_id, $notns) === 0) {
111*fab81cc8SSatoshi Sahara                    unset($pages[$p_id]);
112*fab81cc8SSatoshi Sahara                }
113*fab81cc8SSatoshi Sahara            }
114*fab81cc8SSatoshi Sahara        }
115fe2d1da1SSatoshi Sahara
116fe2d1da1SSatoshi Sahara        // discard hidden pages
117fe2d1da1SSatoshi Sahara        // discard nonexistent pages
118fe2d1da1SSatoshi Sahara        // check ACL permissions
119fe2d1da1SSatoshi Sahara        foreach (array_keys($pages) as $idx) {
120fe2d1da1SSatoshi Sahara            if (!isVisiblePage($idx) || !page_exists($idx) || auth_quickaclcheck($idx) < AUTH_READ) {
121fe2d1da1SSatoshi Sahara                unset($pages[$idx]);
122fe2d1da1SSatoshi Sahara            }
123fe2d1da1SSatoshi Sahara        }
124fe2d1da1SSatoshi Sahara
1259329b002SSatoshi Sahara        $pages = $this->filterResultsByTime($pages, $data['after'], $data['before']);
126fe2d1da1SSatoshi Sahara
1279329b002SSatoshi Sahara        uksort($pages, [$this, 'pagesorter']);
128fe2d1da1SSatoshi Sahara        return $pages;
129fe2d1da1SSatoshi Sahara    }
130fe2d1da1SSatoshi Sahara
131fe2d1da1SSatoshi Sahara    /**
132fe2d1da1SSatoshi Sahara     * Tiny helper function for comparing the searched title with the title
133fe2d1da1SSatoshi Sahara     * from the search index. This function is a wrapper around stripos with
134fe2d1da1SSatoshi Sahara     * adapted argument order and return value.
135fe2d1da1SSatoshi Sahara     *
136fe2d1da1SSatoshi Sahara     * @param string $search searched title
137fe2d1da1SSatoshi Sahara     * @param string $title  title from index
138fe2d1da1SSatoshi Sahara     * @return bool
139fe2d1da1SSatoshi Sahara     */
1409329b002SSatoshi Sahara    protected function pageLookupTitleCompare($search, $title)
141fe2d1da1SSatoshi Sahara    {
142fe2d1da1SSatoshi Sahara        return stripos($title, $search) !== false;
143fe2d1da1SSatoshi Sahara    }
144fe2d1da1SSatoshi Sahara
145fe2d1da1SSatoshi Sahara    /**
146fe2d1da1SSatoshi Sahara     * Sort pages based on their namespace level first, then on their string
147fe2d1da1SSatoshi Sahara     * values. This makes higher hierarchy pages rank higher than lower hierarchy
148fe2d1da1SSatoshi Sahara     * pages.
149fe2d1da1SSatoshi Sahara     *
150fe2d1da1SSatoshi Sahara     * @param string $a
151fe2d1da1SSatoshi Sahara     * @param string $b
152fe2d1da1SSatoshi Sahara     * @return int Returns < 0 if $a is less than $b; > 0 if $a is greater than $b,
153fe2d1da1SSatoshi Sahara     *             and 0 if they are equal.
154fe2d1da1SSatoshi Sahara     */
1559329b002SSatoshi Sahara    protected function pagesorter($a, $b)
156fe2d1da1SSatoshi Sahara    {
157fe2d1da1SSatoshi Sahara        $ac = count(explode(':',$a));
158fe2d1da1SSatoshi Sahara        $bc = count(explode(':',$b));
159fe2d1da1SSatoshi Sahara        if ($ac < $bc) {
160fe2d1da1SSatoshi Sahara            return -1;
161fe2d1da1SSatoshi Sahara        } elseif ($ac > $bc) {
162fe2d1da1SSatoshi Sahara            return 1;
163fe2d1da1SSatoshi Sahara        }
164fe2d1da1SSatoshi Sahara        return strcmp ($a,$b);
165fe2d1da1SSatoshi Sahara    }
166fe2d1da1SSatoshi Sahara
167fe2d1da1SSatoshi Sahara    /**
168fe2d1da1SSatoshi Sahara     * @param array      $results search results in the form pageid => value
169fe2d1da1SSatoshi Sahara     * @param int|string $after   only returns results with mtime after this date,
170fe2d1da1SSatoshi Sahara     *                            accepts timestap or strtotime arguments
171fe2d1da1SSatoshi Sahara     * @param int|string $before  only returns results with mtime after this date,
172fe2d1da1SSatoshi Sahara     *                            accepts timestap or strtotime arguments
173fe2d1da1SSatoshi Sahara     *
174fe2d1da1SSatoshi Sahara     * @return array
175fe2d1da1SSatoshi Sahara     */
1769329b002SSatoshi Sahara    protected function filterResultsByTime(array $results, $after, $before)
177fe2d1da1SSatoshi Sahara    {
178fe2d1da1SSatoshi Sahara        if ($after || $before) {
179fe2d1da1SSatoshi Sahara            $after = is_int($after) ? $after : strtotime($after);
180fe2d1da1SSatoshi Sahara            $before = is_int($before) ? $before : strtotime($before);
181fe2d1da1SSatoshi Sahara
182fe2d1da1SSatoshi Sahara            foreach ($results as $id => $value) {
183fe2d1da1SSatoshi Sahara                $mTime = filemtime(wikiFN($id));
184fe2d1da1SSatoshi Sahara                if ($after && $after > $mTime) {
185fe2d1da1SSatoshi Sahara                    unset($results[$id]);
186fe2d1da1SSatoshi Sahara                    continue;
187fe2d1da1SSatoshi Sahara                }
188fe2d1da1SSatoshi Sahara                if ($before && $before < $mTime) {
189fe2d1da1SSatoshi Sahara                    unset($results[$id]);
190fe2d1da1SSatoshi Sahara                }
191fe2d1da1SSatoshi Sahara            }
192fe2d1da1SSatoshi Sahara        }
193fe2d1da1SSatoshi Sahara        return $results;
194fe2d1da1SSatoshi Sahara    }
195fe2d1da1SSatoshi Sahara}
196