xref: /dokuwiki/inc/Search/MetadataSearch.php (revision 9329b002986cc3f43c18c207dd9d0fdfd0f8a5e8)
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{
17*9329b002SSatoshi Sahara    /** @var MetadataSearch $instance */
18*9329b002SSatoshi Sahara    protected static $instance = null;
19*9329b002SSatoshi Sahara
20*9329b002SSatoshi Sahara    /**
21*9329b002SSatoshi Sahara     * Get new or existing singleton instance of the MetadataSearch
22*9329b002SSatoshi Sahara     *
23*9329b002SSatoshi Sahara     * @return MetadataSearch
24*9329b002SSatoshi Sahara     */
25*9329b002SSatoshi Sahara    public static function getInstance()
26*9329b002SSatoshi Sahara    {
27*9329b002SSatoshi Sahara        if (is_null(static::$instance)) {
28*9329b002SSatoshi Sahara            static::$instance = new static();
29*9329b002SSatoshi Sahara        }
30*9329b002SSatoshi Sahara        return static::$instance;
31*9329b002SSatoshi Sahara    }
32*9329b002SSatoshi Sahara
33fe2d1da1SSatoshi Sahara    /**
34fe2d1da1SSatoshi Sahara     *  Metadata Search constructor. prevent direct object creation
35fe2d1da1SSatoshi Sahara     */
36fe2d1da1SSatoshi Sahara    protected function __construct() {}
37fe2d1da1SSatoshi Sahara
38fe2d1da1SSatoshi Sahara    /**
39fe2d1da1SSatoshi Sahara     * Quicksearch for pagenames
40fe2d1da1SSatoshi Sahara     *
41fe2d1da1SSatoshi Sahara     * By default it only matches the pagename and ignores the namespace.
42fe2d1da1SSatoshi Sahara     * This can be changed with the second parameter.
43fe2d1da1SSatoshi Sahara     * The third parameter allows to search in titles as well.
44fe2d1da1SSatoshi Sahara     *
45fe2d1da1SSatoshi Sahara     * The function always returns titles as well
46fe2d1da1SSatoshi Sahara     *
47fe2d1da1SSatoshi Sahara     * @triggers SEARCH_QUERY_PAGELOOKUP
48fe2d1da1SSatoshi Sahara     * @author   Andreas Gohr <andi@splitbrain.org>
49fe2d1da1SSatoshi Sahara     * @author   Adrian Lang <lang@cosmocode.de>
50fe2d1da1SSatoshi Sahara     *
51fe2d1da1SSatoshi Sahara     * @param string     $id       page id
52fe2d1da1SSatoshi Sahara     * @param bool       $in_ns    match against namespace as well?
53fe2d1da1SSatoshi Sahara     * @param bool       $in_title search in title?
54fe2d1da1SSatoshi Sahara     * @param int|string $after    only show results with mtime after this date,
55fe2d1da1SSatoshi Sahara     *                             accepts timestap or strtotime arguments
56fe2d1da1SSatoshi Sahara     * @param int|string $before   only show results with mtime before this date,
57fe2d1da1SSatoshi Sahara     *                             accepts timestap or strtotime arguments
58fe2d1da1SSatoshi Sahara     *
59fe2d1da1SSatoshi Sahara     * @return string[]
60fe2d1da1SSatoshi Sahara     */
61*9329b002SSatoshi Sahara    public function pageLookup($id, $in_ns = false, $in_title = false, $after = null, $before = null)
62fe2d1da1SSatoshi Sahara    {
63fe2d1da1SSatoshi Sahara        $data = [
64fe2d1da1SSatoshi Sahara            'id' => $id,
65fe2d1da1SSatoshi Sahara            'in_ns' => $in_ns,
66fe2d1da1SSatoshi Sahara            'in_title' => $in_title,
67fe2d1da1SSatoshi Sahara            'after' => $after,
68fe2d1da1SSatoshi Sahara            'before' => $before
69fe2d1da1SSatoshi Sahara        ];
70fe2d1da1SSatoshi Sahara        $data['has_titles'] = true; // for plugin backward compatibility check
71*9329b002SSatoshi Sahara        $action = [$this, 'pageLookupCallBack'];
72fe2d1da1SSatoshi Sahara        return Event::createAndTrigger('SEARCH_QUERY_PAGELOOKUP', $data, $action);
73fe2d1da1SSatoshi Sahara    }
74fe2d1da1SSatoshi Sahara
75fe2d1da1SSatoshi Sahara    /**
76fe2d1da1SSatoshi Sahara     * Returns list of pages as array(pageid => First Heading)
77fe2d1da1SSatoshi Sahara     *
78fe2d1da1SSatoshi Sahara     * @param array $data  event data
79fe2d1da1SSatoshi Sahara     * @return string[]
80fe2d1da1SSatoshi Sahara     */
81*9329b002SSatoshi Sahara    public function pageLookupCallBack(&$data)
82fe2d1da1SSatoshi Sahara    {
83fe2d1da1SSatoshi Sahara        // split out original parameters
84fe2d1da1SSatoshi Sahara        $id = $data['id'];
85*9329b002SSatoshi Sahara        $parsedQuery = (new QueryParser)->convert($id);
86fe2d1da1SSatoshi Sahara
87fe2d1da1SSatoshi Sahara        if (count($parsedQuery['ns']) > 0) {
88fe2d1da1SSatoshi Sahara            $ns = cleanID($parsedQuery['ns'][0]) . ':';
89fe2d1da1SSatoshi Sahara            $id = implode(' ', $parsedQuery['highlight']);
90fe2d1da1SSatoshi Sahara        }
91fe2d1da1SSatoshi Sahara
92fe2d1da1SSatoshi Sahara        $in_ns    = $data['in_ns'];
93fe2d1da1SSatoshi Sahara        $in_title = $data['in_title'];
94fe2d1da1SSatoshi Sahara        $cleaned = cleanID($id);
95fe2d1da1SSatoshi Sahara
96fe2d1da1SSatoshi Sahara        $pages = array();
97fe2d1da1SSatoshi Sahara        if ($id !== '' && $cleaned !== '') {
9802361d2aSSatoshi Sahara            $MetadataIndex = MetadataIndex::getInstance();
9902361d2aSSatoshi Sahara            $page_idx = $MetadataIndex->getPages();
100fe2d1da1SSatoshi Sahara            foreach ($page_idx as $p_id) {
101fe2d1da1SSatoshi Sahara                if ((strpos($in_ns ? $p_id : noNSorNS($p_id), $cleaned) !== false)) {
102fe2d1da1SSatoshi Sahara                    if (!isset($pages[$p_id])) {
103fe2d1da1SSatoshi Sahara                        $pages[$p_id] = p_get_first_heading($p_id, METADATA_DONT_RENDER);
104fe2d1da1SSatoshi Sahara                    }
105fe2d1da1SSatoshi Sahara                }
106fe2d1da1SSatoshi Sahara            }
107fe2d1da1SSatoshi Sahara            if ($in_title) {
108*9329b002SSatoshi Sahara                $func = [$this, 'pageLookupTitleCompare'];
109be5c1ea2SSatoshi Sahara                foreach ($MetadataIndex->lookupKey('title', $id, $func) as $p_id) {
110fe2d1da1SSatoshi Sahara                    if (!isset($pages[$p_id])) {
111fe2d1da1SSatoshi Sahara                        $pages[$p_id] = p_get_first_heading($p_id, METADATA_DONT_RENDER);
112fe2d1da1SSatoshi Sahara                    }
113fe2d1da1SSatoshi Sahara                }
114fe2d1da1SSatoshi Sahara            }
115fe2d1da1SSatoshi Sahara        }
116fe2d1da1SSatoshi Sahara
117fe2d1da1SSatoshi Sahara        if (isset($ns)) {
118fe2d1da1SSatoshi Sahara            foreach (array_keys($pages) as $p_id) {
119fe2d1da1SSatoshi Sahara                if (strpos($p_id, $ns) !== 0) {
120fe2d1da1SSatoshi Sahara                    unset($pages[$p_id]);
121fe2d1da1SSatoshi Sahara                }
122fe2d1da1SSatoshi Sahara            }
123fe2d1da1SSatoshi Sahara        }
124fe2d1da1SSatoshi Sahara
125fe2d1da1SSatoshi Sahara        // discard hidden pages
126fe2d1da1SSatoshi Sahara        // discard nonexistent pages
127fe2d1da1SSatoshi Sahara        // check ACL permissions
128fe2d1da1SSatoshi Sahara        foreach (array_keys($pages) as $idx) {
129fe2d1da1SSatoshi Sahara            if (!isVisiblePage($idx) || !page_exists($idx) || auth_quickaclcheck($idx) < AUTH_READ) {
130fe2d1da1SSatoshi Sahara                unset($pages[$idx]);
131fe2d1da1SSatoshi Sahara            }
132fe2d1da1SSatoshi Sahara        }
133fe2d1da1SSatoshi Sahara
134*9329b002SSatoshi Sahara        $pages = $this->filterResultsByTime($pages, $data['after'], $data['before']);
135fe2d1da1SSatoshi Sahara
136*9329b002SSatoshi Sahara        uksort($pages, [$this, 'pagesorter']);
137fe2d1da1SSatoshi Sahara        return $pages;
138fe2d1da1SSatoshi Sahara    }
139fe2d1da1SSatoshi Sahara
140fe2d1da1SSatoshi Sahara    /**
141fe2d1da1SSatoshi Sahara     * Tiny helper function for comparing the searched title with the title
142fe2d1da1SSatoshi Sahara     * from the search index. This function is a wrapper around stripos with
143fe2d1da1SSatoshi Sahara     * adapted argument order and return value.
144fe2d1da1SSatoshi Sahara     *
145fe2d1da1SSatoshi Sahara     * @param string $search searched title
146fe2d1da1SSatoshi Sahara     * @param string $title  title from index
147fe2d1da1SSatoshi Sahara     * @return bool
148fe2d1da1SSatoshi Sahara     */
149*9329b002SSatoshi Sahara    protected function pageLookupTitleCompare($search, $title)
150fe2d1da1SSatoshi Sahara    {
151fe2d1da1SSatoshi Sahara        return stripos($title, $search) !== false;
152fe2d1da1SSatoshi Sahara    }
153fe2d1da1SSatoshi Sahara
154fe2d1da1SSatoshi Sahara    /**
155fe2d1da1SSatoshi Sahara     * Sort pages based on their namespace level first, then on their string
156fe2d1da1SSatoshi Sahara     * values. This makes higher hierarchy pages rank higher than lower hierarchy
157fe2d1da1SSatoshi Sahara     * pages.
158fe2d1da1SSatoshi Sahara     *
159fe2d1da1SSatoshi Sahara     * @param string $a
160fe2d1da1SSatoshi Sahara     * @param string $b
161fe2d1da1SSatoshi Sahara     * @return int Returns < 0 if $a is less than $b; > 0 if $a is greater than $b,
162fe2d1da1SSatoshi Sahara     *             and 0 if they are equal.
163fe2d1da1SSatoshi Sahara     */
164*9329b002SSatoshi Sahara    protected function pagesorter($a, $b)
165fe2d1da1SSatoshi Sahara    {
166fe2d1da1SSatoshi Sahara        $ac = count(explode(':',$a));
167fe2d1da1SSatoshi Sahara        $bc = count(explode(':',$b));
168fe2d1da1SSatoshi Sahara        if ($ac < $bc) {
169fe2d1da1SSatoshi Sahara            return -1;
170fe2d1da1SSatoshi Sahara        } elseif ($ac > $bc) {
171fe2d1da1SSatoshi Sahara            return 1;
172fe2d1da1SSatoshi Sahara        }
173fe2d1da1SSatoshi Sahara        return strcmp ($a,$b);
174fe2d1da1SSatoshi Sahara    }
175fe2d1da1SSatoshi Sahara
176fe2d1da1SSatoshi Sahara    /**
177fe2d1da1SSatoshi Sahara     * @param array      $results search results in the form pageid => value
178fe2d1da1SSatoshi Sahara     * @param int|string $after   only returns results with mtime after this date,
179fe2d1da1SSatoshi Sahara     *                            accepts timestap or strtotime arguments
180fe2d1da1SSatoshi Sahara     * @param int|string $before  only returns results with mtime after this date,
181fe2d1da1SSatoshi Sahara     *                            accepts timestap or strtotime arguments
182fe2d1da1SSatoshi Sahara     *
183fe2d1da1SSatoshi Sahara     * @return array
184fe2d1da1SSatoshi Sahara     */
185*9329b002SSatoshi Sahara    protected function filterResultsByTime(array $results, $after, $before)
186fe2d1da1SSatoshi Sahara    {
187fe2d1da1SSatoshi Sahara        if ($after || $before) {
188fe2d1da1SSatoshi Sahara            $after = is_int($after) ? $after : strtotime($after);
189fe2d1da1SSatoshi Sahara            $before = is_int($before) ? $before : strtotime($before);
190fe2d1da1SSatoshi Sahara
191fe2d1da1SSatoshi Sahara            foreach ($results as $id => $value) {
192fe2d1da1SSatoshi Sahara                $mTime = filemtime(wikiFN($id));
193fe2d1da1SSatoshi Sahara                if ($after && $after > $mTime) {
194fe2d1da1SSatoshi Sahara                    unset($results[$id]);
195fe2d1da1SSatoshi Sahara                    continue;
196fe2d1da1SSatoshi Sahara                }
197fe2d1da1SSatoshi Sahara                if ($before && $before < $mTime) {
198fe2d1da1SSatoshi Sahara                    unset($results[$id]);
199fe2d1da1SSatoshi Sahara                }
200fe2d1da1SSatoshi Sahara            }
201fe2d1da1SSatoshi Sahara        }
202fe2d1da1SSatoshi Sahara        return $results;
203fe2d1da1SSatoshi Sahara    }
204fe2d1da1SSatoshi Sahara}
205