+ $last = $index;
+ }
+ if( $extract )
+ $extract .= '<b> ... </b>';
+
+ $processed = array();
+ foreach($terms as $term){
+ if( ! isset($processed[$term]) ){
+ $pat3 = "/$patPre(".$term.")$patPost/ui"; // highlight word
+ $extract = preg_replace( $pat3,
+ "\\1<span class='searchmatch'>\\2</span>\\3", $extract );
+ $processed[$term] = true;
+ }
+ }
+
+ wfProfileOut( "$fname-extract" );
+
+ return $extract;
+ }
+
+ /**
+ * Split text into lines and add it to extracts array
+ *
+ * @param array $extracts index -> $line
+ * @param int $count
+ * @param string $text
+ */
+ function splitAndAdd(&$extracts, &$count, $text){
+ $split = explode( "\n", $this->mCleanWikitext? $this->removeWiki($text) : $text );
+ foreach($split as $line){
+ $tt = trim($line);
+ if( $tt )
+ $extracts[$count++] = $tt;
+ }
+ }
+
+ /**
+ * Do manual case conversion for non-ascii chars
+ *
+ * @param unknown_type $matches
+ */
+ function caseCallback($matches){
+ global $wgContLang;
+ if( strlen($matches[0]) > 1 ){
+ return '['.$wgContLang->lc($matches[0]).$wgContLang->uc($matches[0]).']';
+ } else
+ return $matches[0];
+ }
+
+ /**
+ * Extract part of the text from start to end, but by
+ * not chopping up words
+ * @param string $text
+ * @param int $start
+ * @param int $end
+ * @param int $posStart (out) actual start position
+ * @param int $posEnd (out) actual end position
+ * @return string
+ */
+ function extract($text, $start, $end, &$posStart = null, &$posEnd = null ){
+ global $wgContLang;
+
+ if( $start != 0)
+ $start = $this->position( $text, $start, 1 );
+ if( $end >= strlen($text) )
+ $end = strlen($text);
+ else
+ $end = $this->position( $text, $end );
+
+ if(!is_null($posStart))
+ $posStart = $start;
+ if(!is_null($posEnd))
+ $posEnd = $end;
+
+ if($end > $start)
+ return substr($text, $start, $end-$start);
+ else
+ return '';
+ }
+
+ /**
+ * Find a nonletter near a point (index) in the text
+ *
+ * @param string $text
+ * @param int $point
+ * @param int $offset to found index
+ * @return int nearest nonletter index, or beginning of utf8 char if none
+ */
+ function position($text, $point, $offset=0 ){
+ $tolerance = 10;
+ $s = max( 0, $point - $tolerance );
+ $l = min( strlen($text), $point + $tolerance ) - $s;
+ $m = array();
+ if( preg_match('/[ ,.!?~!@#$%^&*\(\)+=\-\\\|\[\]"\'<>]/', substr($text,$s,$l), $m, PREG_OFFSET_CAPTURE ) ){
+ return $m[0][1] + $s + $offset;
+ } else{
+ // check if point is on a valid first UTF8 char
+ $char = ord( $text[$point] );
+ while( $char >= 0x80 && $char < 0xc0 ) {
+ // skip trailing bytes
+ $point++;
+ if($point >= strlen($text))
+ return strlen($text);
+ $char = ord( $text[$point] );
+ }
+ return $point;
+
+ }
+ }
+
+ /**
+ * Search extracts for a pattern, and return snippets
+ *
+ * @param string $pattern regexp for matching lines
+ * @param array $extracts extracts to search
+ * @param int $linesleft number of extracts to make
+ * @param int $contextchars length of snippet
+ * @param array $out map for highlighted snippets
+ * @param array $offsets map of starting points of snippets
+ * @protected
+ */
+ function process( $pattern, $extracts, &$linesleft, &$contextchars, &$out, &$offsets ){
+ if($linesleft == 0)
+ return; // nothing to do
+ foreach($extracts as $index => $line){
+ if( array_key_exists($index,$out) )
+ continue; // this line already highlighted
+
+ $m = array();
+ if ( !preg_match( $pattern, $line, $m, PREG_OFFSET_CAPTURE ) )
+ continue;
+
+ $offset = $m[0][1];
+ $len = strlen($m[0][0]);
+ if($offset + $len < $contextchars)
+ $begin = 0;
+ elseif( $len > $contextchars)
+ $begin = $offset;
+ else
+ $begin = $offset + intval( ($len - $contextchars) / 2 );
+
+ $end = $begin + $contextchars;
+
+ $posBegin = $begin;
+ // basic snippet from this line
+ $out[$index] = $this->extract($line,$begin,$end,$posBegin);
+ $offsets[$index] = $posBegin;
+ $linesleft--;
+ if($linesleft == 0)
+ return;
+ }
+ }
+
+ /**
+ * Basic wikitext removal
+ * @protected
+ */
+ function removeWiki($text) {
+ $fname = __METHOD__;
+ wfProfileIn( $fname );
+
+ //$text = preg_replace("/'{2,5}/", "", $text);
+ //$text = preg_replace("/\[[a-z]+:\/\/[^ ]+ ([^]]+)\]/", "\\2", $text);
+ //$text = preg_replace("/\[\[([^]|]+)\]\]/", "\\1", $text);
+ //$text = preg_replace("/\[\[([^]]+\|)?([^|]]+)\]\]/", "\\2", $text);
+ //$text = preg_replace("/\\{\\|(.*?)\\|\\}/", "", $text);
+ //$text = preg_replace("/\\[\\[[A-Za-z_-]+:([^|]+?)\\]\\]/", "", $text);
+ $text = preg_replace("/\\{\\{([^|]+?)\\}\\}/", "", $text);
+ $text = preg_replace("/\\{\\{([^|]+\\|)(.*?)\\}\\}/", "\\2", $text);
+ $text = preg_replace("/\\[\\[([^|]+?)\\]\\]/", "\\1", $text);
+ $text = preg_replace_callback("/\\[\\[([^|]+\\|)(.*?)\\]\\]/", array($this,'linkReplace'), $text);
+ //$text = preg_replace("/\\[\\[([^|]+\\|)(.*?)\\]\\]/", "\\2", $text);
+ $text = preg_replace("/<\/?[^>]+>/", "", $text);
+ $text = preg_replace("/'''''/", "", $text);
+ $text = preg_replace("/('''|<\/?[iIuUbB]>)/", "", $text);
+ $text = preg_replace("/''/", "", $text);
+
+ wfProfileOut( $fname );
+ return $text;
+ }
+
+ /**
+ * callback to replace [[target|caption]] kind of links, if
+ * the target is category or image, leave it
+ *
+ * @param array $matches
+ */
+ function linkReplace($matches){
+ $colon = strpos( $matches[1], ':' );
+ if( $colon === false )
+ return $matches[2]; // replace with caption
+ global $wgContLang;
+ $ns = substr( $matches[1], 0, $colon );
+ $index = $wgContLang->getNsIndex($ns);
+ if( $index !== false && ($index == NS_FILE || $index == NS_CATEGORY) )
+ return $matches[0]; // return the whole thing
+ else
+ return $matches[2];
+
+ }
+
+ /**
+ * Simple & fast snippet extraction, but gives completely unrelevant
+ * snippets
+ *
+ * @param string $text
+ * @param array $terms
+ * @param int $contextlines
+ * @param int $contextchars
+ * @return string
+ */
+ public function highlightSimple( $text, $terms, $contextlines, $contextchars ) {
+ global $wgLang, $wgContLang;
+ $fname = __METHOD__;
+
+ $lines = explode( "\n", $text );
+
+ $terms = implode( '|', $terms );
+ $max = intval( $contextchars ) + 1;
+ $pat1 = "/(.*)($terms)(.{0,$max})/i";
+
+ $lineno = 0;
+
+ $extract = "";
+ wfProfileIn( "$fname-extract" );
+ foreach ( $lines as $line ) {
+ if ( 0 == $contextlines ) {
+ break;
+ }
+ ++$lineno;
+ $m = array();
+ if ( ! preg_match( $pat1, $line, $m ) ) {
+ continue;
+ }
+ --$contextlines;
+ $pre = $wgContLang->truncate( $m[1], -$contextchars );
+
+ if ( count( $m ) < 3 ) {
+ $post = '';
+ } else {
+ $post = $wgContLang->truncate( $m[3], $contextchars );
+ }
+
+ $found = $m[2];
+
+ $line = htmlspecialchars( $pre . $found . $post );
+ $pat2 = '/(' . $terms . ")/i";
+ $line = preg_replace( $pat2,
+ "<span class='searchmatch'>\\1</span>", $line );
+
+ $extract .= "${line}\n";
+ }
+ wfProfileOut( "$fname-extract" );
+
+ return $extract;
+ }
+
+}
+
+/**
+ * Dummy class to be used when non-supported Database engine is present.
+ * @fixme Dummy class should probably try something at least mildly useful,
+ * such as a LIKE search through titles.
+ * @ingroup Search
+ */
+class SearchEngineDummy extends SearchEngine {
+ // no-op
+}