check that $wgArticle is an instance of the Article class in Skin::pageStats() per...
[lhc/web/wiklou.git] / includes / SearchUpdate.php
1 <?php
2 /**
3 * See deferred.txt
4 * @ingroup Search
5 */
6 class SearchUpdate {
7
8 /* private */ var $mId = 0, $mNamespace, $mTitle, $mText;
9 /* private */ var $mTitleWords;
10
11 function SearchUpdate( $id, $title, $text = false ) {
12 $nt = Title::newFromText( $title );
13 if( $nt ) {
14 $this->mId = $id;
15 $this->mText = $text;
16
17 $this->mNamespace = $nt->getNamespace();
18 $this->mTitle = $nt->getText(); # Discard namespace
19
20 $this->mTitleWords = $this->mTextWords = array();
21 } else {
22 wfDebug( "SearchUpdate object created with invalid title '$title'\n" );
23 }
24 }
25
26 function doUpdate() {
27 global $wgContLang, $wgDisableSearchUpdate;
28
29 if( $wgDisableSearchUpdate || !$this->mId ) {
30 return false;
31 }
32 $fname = 'SearchUpdate::doUpdate';
33 wfProfileIn( $fname );
34
35 $search = SearchEngine::create();
36 $lc = SearchEngine::legalSearchChars() . '&#;';
37
38 if( $this->mText === false ) {
39 $search->updateTitle($this->mId,
40 Title::indexTitle( $this->mNamespace, $this->mTitle ));
41 wfProfileOut( $fname );
42 return;
43 }
44
45 # Language-specific strip/conversion
46 $text = $wgContLang->stripForSearch( $this->mText );
47
48 wfProfileIn( $fname.'-regexps' );
49 $text = preg_replace( "/<\\/?\\s*[A-Za-z][A-Za-z0-9]*\\s*([^>]*?)>/",
50 ' ', strtolower( " " . $text /*$this->mText*/ . " " ) ); # Strip HTML markup
51 $text = preg_replace( "/(^|\\n)==\\s*([^\\n]+)\\s*==(\\s)/sD",
52 "\\1\\2 \\2 \\2\\3", $text ); # Emphasize headings
53
54 # Strip external URLs
55 $uc = "A-Za-z0-9_\\/:.,~%\\-+&;#?!=()@\\xA0-\\xFF";
56 $protos = "http|https|ftp|mailto|news|gopher";
57 $pat = "/(^|[^\\[])({$protos}):[{$uc}]+([^{$uc}]|$)/";
58 $text = preg_replace( $pat, "\\1 \\3", $text );
59
60 $p1 = "/([^\\[])\\[({$protos}):[{$uc}]+]/";
61 $p2 = "/([^\\[])\\[({$protos}):[{$uc}]+\\s+([^\\]]+)]/";
62 $text = preg_replace( $p1, "\\1 ", $text );
63 $text = preg_replace( $p2, "\\1 \\3 ", $text );
64
65 # Internal image links
66 $pat2 = "/\\[\\[image:([{$uc}]+)\\.(gif|png|jpg|jpeg)([^{$uc}])/i";
67 $text = preg_replace( $pat2, " \\1 \\3", $text );
68
69 $text = preg_replace( "/([^{$lc}])([{$lc}]+)]]([a-z]+)/",
70 "\\1\\2 \\2\\3", $text ); # Handle [[game]]s
71
72 # Strip all remaining non-search characters
73 $text = preg_replace( "/[^{$lc}]+/", " ", $text );
74
75 # Handle 's, s'
76 #
77 # $text = preg_replace( "/([{$lc}]+)'s /", "\\1 \\1's ", $text );
78 # $text = preg_replace( "/([{$lc}]+)s' /", "\\1s ", $text );
79 #
80 # These tail-anchored regexps are insanely slow. The worst case comes
81 # when Japanese or Chinese text (ie, no word spacing) is written on
82 # a wiki configured for Western UTF-8 mode. The Unicode characters are
83 # expanded to hex codes and the "words" are very long paragraph-length
84 # monstrosities. On a large page the above regexps may take over 20
85 # seconds *each* on a 1GHz-level processor.
86 #
87 # Following are reversed versions which are consistently fast
88 # (about 3 milliseconds on 1GHz-level processor).
89 #
90 $text = strrev( preg_replace( "/ s'([{$lc}]+)/", " s'\\1 \\1", strrev( $text ) ) );
91 $text = strrev( preg_replace( "/ 's([{$lc}]+)/", " s\\1", strrev( $text ) ) );
92
93 # Strip wiki '' and '''
94 $text = preg_replace( "/''[']*/", " ", $text );
95 wfProfileOut( "$fname-regexps" );
96
97 wfRunHooks( 'SearchUpdate', array( $this->mId, $this->mNamespace, $this->mTitle, &$text ) );
98
99 # Perform the actual update
100 $search->update($this->mId, Title::indexTitle( $this->mNamespace, $this->mTitle ),
101 $text);
102
103 wfProfileOut( $fname );
104 }
105 }
106
107 /**
108 * Placeholder class
109 * @ingroup Search
110 */
111 class SearchUpdateMyISAM extends SearchUpdate {
112 # Inherits everything
113 }