8 /* private */ var $mId = 0, $mNamespace, $mTitle, $mText;
9 /* private */ var $mTitleWords;
11 function SearchUpdate( $id, $title, $text = false ) {
12 $nt = Title
::newFromText( $title );
17 $this->mNamespace
= $nt->getNamespace();
18 $this->mTitle
= $nt->getText(); # Discard namespace
20 $this->mTitleWords
= $this->mTextWords
= array();
22 wfDebug( "SearchUpdate object created with invalid title '$title'\n" );
27 global $wgContLang, $wgDisableSearchUpdate;
29 if( $wgDisableSearchUpdate ||
!$this->mId
) {
32 $fname = 'SearchUpdate::doUpdate';
33 wfProfileIn( $fname );
35 $search = SearchEngine
::create();
36 $lc = SearchEngine
::legalSearchChars() . '&#;';
38 if( $this->mText
=== false ) {
39 $search->updateTitle($this->mId
,
40 Title
::indexTitle( $this->mNamespace
, $this->mTitle
));
41 wfProfileOut( $fname );
45 # Language-specific strip/conversion
46 $text = $wgContLang->stripForSearch( $this->mText
);
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
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 );
60 $p1 = "/([^\\[])\\[({$protos}):[{$uc}]+]/";
61 $p2 = "/([^\\[])\\[({$protos}):[{$uc}]+\\s+([^\\]]+)]/";
62 $text = preg_replace( $p1, "\\1 ", $text );
63 $text = preg_replace( $p2, "\\1 \\3 ", $text );
65 # Internal image links
66 $pat2 = "/\\[\\[image:([{$uc}]+)\\.(gif|png|jpg|jpeg)([^{$uc}])/i";
67 $text = preg_replace( $pat2, " \\1 \\3", $text );
69 $text = preg_replace( "/([^{$lc}])([{$lc}]+)]]([a-z]+)/",
70 "\\1\\2 \\2\\3", $text ); # Handle [[game]]s
72 # Strip all remaining non-search characters
73 $text = preg_replace( "/[^{$lc}]+/", " ", $text );
77 # $text = preg_replace( "/([{$lc}]+)'s /", "\\1 \\1's ", $text );
78 # $text = preg_replace( "/([{$lc}]+)s' /", "\\1s ", $text );
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.
87 # Following are reversed versions which are consistently fast
88 # (about 3 milliseconds on 1GHz-level processor).
90 $text = strrev( preg_replace( "/ s'([{$lc}]+)/", " s'\\1 \\1", strrev( $text ) ) );
91 $text = strrev( preg_replace( "/ 's([{$lc}]+)/", " s\\1", strrev( $text ) ) );
93 # Strip wiki '' and '''
94 $text = preg_replace( "/''[']*/", " ", $text );
95 wfProfileOut( "$fname-regexps" );
97 wfRunHooks( 'SearchUpdate', array( $this->mId
, $this->mNamespace
, $this->mTitle
, &$text ) );
99 # Perform the actual update
100 $search->update($this->mId
, Title
::indexTitle( $this->mNamespace
, $this->mTitle
),
103 wfProfileOut( $fname );
111 class SearchUpdateMyISAM
extends SearchUpdate
{
112 # Inherits everything