various things
[mediawiki.git] / includes / OutputPage.php
blobadc63df86e4242d6e3eddbc9527177bb19b9de10
1 <?
2 # See design.doc
4 if($wgUseTeX) include_once( "Math.php" );
6 class OutputPage {
7 var $mHeaders, $mCookies, $mMetatags, $mKeywords;
8 var $mLinktags, $mPagetitle, $mBodytext, $mDebugtext;
9 var $mHTMLtitle, $mRobotpolicy, $mIsarticle, $mPrintable;
10 var $mSubtitle, $mRedirect, $mAutonumber, $mHeadtext;
11 var $mLastModified, $mCategoryLinks;
13 var $mDTopen, $mLastSection; # Used for processing DL, PRE
14 var $mLanguageLinks, $mSupressQuickbar;
16 function OutputPage()
18 $this->mHeaders = $this->mCookies = $this->mMetatags =
19 $this->mKeywords = $this->mLinktags = array();
20 $this->mHTMLtitle = $this->mPagetitle = $this->mBodytext =
21 $this->mLastSection = $this->mRedirect = $this->mLastModified =
22 $this->mSubtitle = $this->mDebugtext = $this->mRobotpolicy = "";
23 $this->mIsarticle = $this->mPrintable = true;
24 $this->mSupressQuickbar = $this->mDTopen = $this->mPrintable = false;
25 $this->mLanguageLinks = array();
26 $this->mCategoryLinks = array() ;
27 $this->mAutonumber = 0;
30 function addHeader( $name, $val ) { array_push( $this->mHeaders, "$name: $val" ) ; }
31 function addCookie( $name, $val ) { array_push( $this->mCookies, array( $name, $val ) ); }
32 function redirect( $url ) { $this->mRedirect = $url; }
34 # To add an http-equiv meta tag, precede the name with "http:"
35 function addMeta( $name, $val ) { array_push( $this->mMetatags, array( $name, $val ) ); }
36 function addKeyword( $text ) { array_push( $this->mKeywords, $text ); }
37 function addLink( $rel, $rev, $target ) { array_push( $this->mLinktags, array( $rel, $rev, $target ) ); }
39 function checkLastModified ( $timestamp )
41 global $wgLang, $wgCachePages, $wgUser;
42 if( !$wgCachePages ) {
43 wfDebug( "CACHE DISABLED\n", false );
44 return;
46 if( preg_match( '/MSIE ([1-4]|5\.0)/', $_SERVER["HTTP_USER_AGENT"] ) ) {
47 # IE 5.0 has probs with our caching
48 wfDebug( "-- bad client, not caching\n", false );
49 return;
51 if( $wgUser->getOption( "nocache" ) ) {
52 wfDebug( "USER DISABLED CACHE\n", false );
53 return;
56 $lastmod = gmdate( "D, j M Y H:i:s", wfTimestamp2Unix(
57 max( $timestamp, $wgUser->mTouched ) ) ) . " GMT";
59 if( $_SERVER["HTTP_IF_MODIFIED_SINCE"] != "" ) {
60 # IE sends sizes after the date for compressed pages:
61 # Wed, 20 Aug 2003 06:51:19 GMT; length=5202
62 # this breaks strtotime().
63 $modsince = preg_replace( '/;.*$/', '', $_SERVER["HTTP_IF_MODIFIED_SINCE"] );
64 $ismodsince = wfUnix2Timestamp( strtotime( $modsince ) );
65 wfDebug( "-- client send If-Modified-Since: " . $modsince . "\n", false );
66 wfDebug( "-- we might send Last-Modified : $lastmod\n", false );
68 if( ($ismodsince >= $timestamp ) and $wgUser->validateCache( $ismodsince ) ) {
69 # Make sure you're in a place you can leave when you call us!
70 header( "HTTP/1.0 304 Not Modified" );
71 header( "Expires: Mon, 15 Jan 2001 00:00:00 GMT" ); # Cachers always validate the page!
72 header( "Cache-Control: private, must-revalidate, max-age=0" );
73 header( "Last-Modified: {$lastmod}" );
74 wfDebug( "CACHED client: $ismodsince ; user: $wgUser->mTouched ; page: $timestamp\n", false );
75 exit;
76 } else {
77 wfDebug( "READY client: $ismodsince ; user: $wgUser->mTouched ; page: $timestamp\n", false );
78 $this->mLastModified = $lastmod;
80 } else {
81 wfDebug( "We're confused.\n", false );
82 $this->mLastModified = $lastmod;
86 function setRobotpolicy( $str ) { $this->mRobotpolicy = $str; }
87 function setHTMLtitle( $name ) { $this->mHTMLtitle = $name; }
88 function setPageTitle( $name ) { $this->mPagetitle = $name; }
89 function getPageTitle() { return $this->mPagetitle; }
90 function setSubtitle( $str ) { $this->mSubtitle = $str; }
91 function getSubtitle() { return $this->mSubtitle; }
92 function setArticleFlag( $v ) { $this->mIsarticle = $v; }
93 function isArticle() { return $this->mIsarticle; }
94 function setPrintable() { $this->mPrintable = true; }
95 function isPrintable() { return $this->mPrintable; }
97 function getLanguageLinks() {
98 global $wgUseNewInterlanguage, $wgTitle, $wgLanguageCode;
99 global $wgDBconnection, $wgDBname, $wgDBintlname;
101 if ( ! $wgUseNewInterlanguage )
102 return $this->mLanguageLinks;
104 mysql_select_db( $wgDBintlname, $wgDBconnection ) or die(
105 htmlspecialchars(mysql_error()) );
107 $list = array();
108 $sql = "SELECT * FROM ilinks WHERE lang_from=\"" .
109 "{$wgLanguageCode}\" AND title_from=\"" . $wgTitle->getDBkey() . "\"";
110 $res = mysql_query( $sql, $wgDBconnection );
112 while ( $q = mysql_fetch_object ( $res ) ) {
113 $list[] = $q->lang_to . ":" . $q->title_to;
115 mysql_free_result( $res );
116 mysql_select_db( $wgDBname, $wgDBconnection ) or die(
117 htmlspecialchars(mysql_error()) );
119 return $list;
122 function supressQuickbar() { $this->mSupressQuickbar = true; }
123 function isQuickbarSupressed() { return $this->mSupressQuickbar; }
125 function addHTML( $text ) { $this->mBodytext .= $text; }
126 function addHeadtext( $text ) { $this->mHeadtext .= $text; }
127 function debug( $text ) { $this->mDebugtext .= $text; }
129 # First pass--just handle <nowiki> sections, pass the rest off
130 # to doWikiPass2() which does all the real work.
133 function addWikiText( $text, $linestart = true )
135 global $wgUseTeX;
136 $fname = "OutputPage::addWikiText";
137 wfProfileIn( $fname );
138 $unique = "3iyZiyA7iMwg5rhxP0Dcc9oTnj8qD1jm1Sfv4";
139 $unique2 = "4LIQ9nXtiYFPCSfitVwDw7EYwQlL4GeeQ7qSO";
140 $unique3 = "fPaA8gDfdLBqzj68Yjg9Hil3qEF8JGO0uszIp";
141 $nwlist = array();
142 $nwsecs = 0;
143 $mathlist = array();
144 $mathsecs = 0;
145 $prelist = array ();
146 $presecs = 0;
147 $stripped = "";
148 $stripped2 = "";
149 $stripped3 = "";
151 while ( "" != $text ) {
152 $p = preg_split( "/<\\s*nowiki\\s*>/i", $text, 2 );
153 $stripped .= $p[0];
154 if ( ( count( $p ) < 2 ) || ( "" == $p[1] ) ) { $text = ""; }
155 else {
156 $q = preg_split( "/<\\/\\s*nowiki\\s*>/i", $p[1], 2 );
157 ++$nwsecs;
158 $nwlist[$nwsecs] = wfEscapeHTMLTagsOnly($q[0]);
159 $stripped .= $unique;
160 $text = $q[1];
164 if( $wgUseTeX ) {
165 while ( "" != $stripped ) {
166 $p = preg_split( "/<\\s*math\\s*>/i", $stripped, 2 );
167 $stripped2 .= $p[0];
168 if ( ( count( $p ) < 2 ) || ( "" == $p[1] ) ) { $stripped = ""; }
169 else {
170 $q = preg_split( "/<\\/\\s*math\\s*>/i", $p[1], 2 );
171 ++$mathsecs;
172 $mathlist[$mathsecs] = renderMath($q[0]);
173 $stripped2 .= $unique2;
174 $stripped = $q[1];
177 } else {
178 $stripped2 = $stripped;
181 while ( "" != $stripped2 ) {
182 $p = preg_split( "/<\\s*pre\\s*>/i", $stripped2, 2 );
183 $stripped3 .= $p[0];
184 if ( ( count( $p ) < 2 ) || ( "" == $p[1] ) ) { $stripped2 = ""; }
185 else {
186 $q = preg_split( "/<\\/\\s*pre\\s*>/i", $p[1], 2 );
187 ++$presecs;
188 $prelist[$presecs] = "<pre>". wfEscapeHTMLTagsOnly($q[0]). "</pre>";
189 $stripped3 .= $unique3;
190 $stripped2 = $q[1];
194 $text = $this->doWikiPass2( $stripped3, $linestart );
196 $specialChars = array("\\", "$");
197 $escapedChars = array("\\\\", "\\$");
198 for ( $i = 1; $i <= $presecs; ++$i ) {
199 $text = preg_replace( "/{$unique3}/", str_replace( $specialChars,
200 $escapedChars, $prelist[$i] ), $text, 1 );
203 for ( $i = 1; $i <= $mathsecs; ++$i ) {
204 $text = preg_replace( "/{$unique2}/", str_replace( $specialChars,
205 $escapedChars, $mathlist[$i] ), $text, 1 );
208 for ( $i = 1; $i <= $nwsecs; ++$i ) {
209 $text = preg_replace( "/{$unique}/", str_replace( $specialChars,
210 $escapedChars, $nwlist[$i] ), $text, 1 );
212 $this->addHTML( $text );
213 wfProfileOut( $fname );
216 function sendCacheControl() {
217 global $wgUseGzip;
218 if( $this->mLastModified != "" ) {
219 wfDebug( "** private caching; {$this->mLastModified} **\n", false );
220 header( "Cache-Control: private, must-revalidate, max-age=0" );
221 header( "Last-modified: {$this->mLastModified}" );
222 if( $wgUseGzip ) {
223 # We should put in Accept-Encoding, but IE chokes on anything but
224 # User-Agent in a Vary: header (at least through 6.0)
225 header( "Vary: User-Agent" );
227 } else {
228 wfDebug( "** no caching **\n", false );
229 header( "Cache-Control: no-cache" ); # Experimental - see below
230 header( "Pragma: no-cache" );
231 header( "Last-modified: " . gmdate( "D, j M Y H:i:s" ) . " GMT" );
233 header( "Expires: Mon, 15 Jan 2001 00:00:00 GMT" ); # Cachers always validate the page!
236 # Finally, all the text has been munged and accumulated into
237 # the object, let's actually output it:
239 function output()
241 global $wgUser, $wgLang, $wgDebugComments, $wgCookieExpiration;
242 global $wgInputEncoding, $wgOutputEncoding, $wgLanguageCode;
244 $fname = "OutputPage::output";
245 wfProfileIn( $fname );
247 $sk = $wgUser->getSkin();
249 $this->sendCacheControl();
251 header( "Content-type: text/html; charset={$wgOutputEncoding}" );
252 header( "Content-language: {$wgLanguageCode}" );
254 if ( "" != $this->mRedirect ) {
255 header( "Location: {$this->mRedirect}" );
256 return;
259 $exp = time() + $wgCookieExpiration;
260 foreach( $this->mCookies as $name => $val ) {
261 setcookie( $name, $val, $exp, "/" );
264 $sk->outputPage( $this );
265 flush();
268 function out( $ins )
270 global $wgInputEncoding, $wgOutputEncoding, $wgLang;
271 if ( 0 == strcmp( $wgInputEncoding, $wgOutputEncoding ) ) {
272 $outs = $ins;
273 } else {
274 $outs = $wgLang->iconv( $wgInputEncoding, $wgOutputEncoding, $ins );
275 if ( false === $outs ) { $outs = $ins; }
277 print $outs;
280 function setEncodings()
282 global $HTTP_SERVER_VARS, $wgInputEncoding, $wgOutputEncoding;
283 global $wgUser, $wgLang;
285 $wgInputEncoding = strtolower( $wgInputEncoding );
286 $s = $HTTP_SERVER_VARS['HTTP_ACCEPT_CHARSET'];
288 if( $wgUser->getOption( 'altencoding' ) ) {
289 $wgLang->setAltEncoding();
290 return;
293 if ( "" == $s ) {
294 $wgOutputEncoding = strtolower( $wgOutputEncoding );
295 return;
297 $a = explode( ",", $s );
298 $best = 0.0;
299 $bestset = "*";
301 foreach ( $a as $s ) {
302 if ( preg_match( "/(.*);q=(.*)/", $s, $m ) ) {
303 $set = $m[1];
304 $q = (float)($m[2]);
305 } else {
306 $set = $s;
307 $q = 1.0;
309 if ( $q > $best ) {
310 $bestset = $set;
311 $best = $q;
314 #if ( "*" == $bestset ) { $bestset = "iso-8859-1"; }
315 if ( "*" == $bestset ) { $bestset = $wgOutputEncoding; }
316 $wgOutputEncoding = strtolower( $bestset );
318 # Disable for now
320 $wgOutputEncoding = $wgInputEncoding;
323 function reportTime()
325 global $wgRequestTime, $wgDebugLogFile, $HTTP_SERVER_VARS;
326 global $wgProfiling, $wgProfileStack, $wgUser;
328 list( $usec, $sec ) = explode( " ", microtime() );
329 $now = (float)$sec + (float)$usec;
331 list( $usec, $sec ) = explode( " ", $wgRequestTime );
332 $start = (float)$sec + (float)$usec;
333 $elapsed = $now - $start;
335 if ( "" != $wgDebugLogFile ) {
336 $prof = wfGetProfilingOutput( $start, $elapsed );
338 if( $forward = $HTTP_SERVER_VARS['HTTP_X_FORWARDED_FOR'] )
339 $forward = " forwarded for $forward";
340 if( $client = $HTTP_SERVER_VARS['HTTP_CLIENT_IP'] )
341 $forward .= " client IP $client";
342 if( $from = $HTTP_SERVER_VARS['HTTP_FROM'] )
343 $forward .= " from $from";
344 if( $forward )
345 $forward = "\t(proxied via {$HTTP_SERVER_VARS['REMOTE_ADDR']}{$forward})";
346 if($wgUser->getId() == 0)
347 $forward .= " anon";
348 $log = sprintf( "%s\t%04.3f\t%s\n",
349 gmdate( "YmdHis" ), $elapsed,
350 urldecode( $HTTP_SERVER_VARS['REQUEST_URI'] . $forward ) );
351 error_log( $log . $prof, 3, $wgDebugLogFile );
353 $com = sprintf( "<!-- Time since request: %01.2f secs. -->",
354 $elapsed );
355 return $com;
358 # Note: these arguments are keys into wfMsg(), not text!
360 function errorpage( $title, $msg )
362 global $wgTitle;
364 $this->mDebugtext .= "Original title: " .
365 $wgTitle->getPrefixedText() . "\n";
366 $this->setHTMLTitle( wfMsg( "errorpagetitle" ) );
367 $this->setPageTitle( wfMsg( $title ) );
368 $this->setRobotpolicy( "noindex,nofollow" );
369 $this->setArticleFlag( false );
371 $this->mBodytext = "";
372 $this->addHTML( "<p>" . wfMsg( $msg ) . "\n" );
373 $this->returnToMain( false );
375 $this->output();
376 exit;
379 function sysopRequired()
381 global $wgUser;
383 $this->setHTMLTitle( wfMsg( "errorpagetitle" ) );
384 $this->setPageTitle( wfMsg( "sysoptitle" ) );
385 $this->setRobotpolicy( "noindex,nofollow" );
386 $this->setArticleFlag( false );
387 $this->mBodytext = "";
389 $sk = $wgUser->getSkin();
390 $ap = $sk->makeKnownLink( wfMsg( "administrators" ), "" );
391 $text = str_replace( "$1", $ap, wfMsg( "sysoptext" ) );
392 $this->addHTML( $text );
393 $this->returnToMain();
396 function developerRequired()
398 global $wgUser;
400 $this->setHTMLTitle( wfMsg( "errorpagetitle" ) );
401 $this->setPageTitle( wfMsg( "developertitle" ) );
402 $this->setRobotpolicy( "noindex,nofollow" );
403 $this->setArticleFlag( false );
404 $this->mBodytext = "";
406 $sk = $wgUser->getSkin();
407 $ap = $sk->makeKnownLink( wfMsg( "administrators" ), "" );
408 $text = str_replace( "$1", $ap, wfMsg( "developertext" ) );
409 $this->addHTML( $text );
410 $this->returnToMain();
413 function databaseError( $fname )
415 global $wgUser, $wgCommandLineMode;
417 $this->setPageTitle( wfMsgNoDB( "databaseerror" ) );
418 $this->setRobotpolicy( "noindex,nofollow" );
419 $this->setArticleFlag( false );
421 if ( $wgCommandLineMode ) {
422 $msg = wfMsgNoDB( "dberrortextcl" );
423 } else {
424 $msg = wfMsgNoDB( "dberrortextcl" );
427 $msg = str_replace( "$1", htmlspecialchars( wfLastDBquery() ), $msg );
428 $msg = str_replace( "$2", htmlspecialchars( $fname ), $msg );
429 $msg = str_replace( "$3", wfLastErrno(), $msg );
430 $msg = str_replace( "$4", htmlspecialchars( wfLastError() ), $msg );
432 if ( $wgCommandLineMode ) {
433 print "$msg\n";
434 exit();
436 $sk = $wgUser->getSkin();
437 $shlink = $sk->makeKnownLink( wfMsgNoDB( "searchhelppage" ),
438 wfMsgNoDB( "searchingwikipedia" ) );
439 $msg = str_replace( "$5", $shlink, $msg );
441 $this->mBodytext = $msg;
442 $this->output();
443 exit();
446 function readOnlyPage( $source = "" )
448 global $wgUser, $wgReadOnlyFile;
450 $this->setPageTitle( wfMsg( "readonly" ) );
451 $this->setRobotpolicy( "noindex,nofollow" );
452 $this->setArticleFlag( false );
454 $reason = implode( "", file( $wgReadOnlyFile ) );
455 $text = str_replace( "$1", $reason, wfMsg( "readonlytext" ) );
457 if($source) {
458 $rows = $wgUser->getOption( "rows" );
459 $cols = $wgUser->getOption( "cols" );
460 $text .= "</p>\n<textarea cols='$cols' rows='$rows' readonly>" .
461 htmlspecialchars( $source ) . "\n</textarea>";
464 $this->addHTML( $text );
465 $this->returnToMain( false );
468 function fatalError( $message )
470 $this->setPageTitle( wfMsg( "internalerror" ) );
471 $this->setRobotpolicy( "noindex,nofollow" );
472 $this->setArticleFlag( false );
474 $this->mBodytext = $message;
475 $this->output();
476 exit;
479 function unexpectedValueError( $name, $val )
481 $msg = str_replace( "$1", $name, wfMsg( "unexpected" ) );
482 $msg = str_replace( "$2", $val, $msg );
483 $this->fatalError( $msg );
486 function fileCopyError( $old, $new )
488 $msg = str_replace( "$1", $old, wfMsg( "filecopyerror" ) );
489 $msg = str_replace( "$2", $new, $msg );
490 $this->fatalError( $msg );
493 function fileRenameError( $old, $new )
495 $msg = str_replace( "$1", $old, wfMsg( "filerenameerror" ) );
496 $msg = str_replace( "$2", $new, $msg );
497 $this->fatalError( $msg );
500 function fileDeleteError( $name )
502 $msg = str_replace( "$1", $name, wfMsg( "filedeleteerror" ) );
503 $this->fatalError( $msg );
506 function fileNotFoundError( $name )
508 $msg = str_replace( "$1", $name, wfMsg( "filenotfound" ) );
509 $this->fatalError( $msg );
512 function returnToMain( $auto = true )
514 global $wgUser, $wgOut, $returnto;
516 $sk = $wgUser->getSkin();
517 if ( "" == $returnto ) {
518 $returnto = wfMsg( "mainpage" );
520 $link = $sk->makeKnownLink( $returnto, "" );
522 $r = str_replace( "$1", $link, wfMsg( "returnto" ) );
523 if ( $auto ) {
524 $wgOut->addMeta( "http:Refresh", "10;url=" .
525 wfLocalUrlE( wfUrlencode( $returnto ) ) );
527 $wgOut->addHTML( "\n<p>$r\n" );
531 function categoryMagic ()
533 global $wgTitle , $wgUseCategoryMagic ;
534 if ( !isset ( $wgUseCategoryMagic ) || !$wgUseCategoryMagic ) return ;
535 $id = $wgTitle->getArticleID() ;
536 $cat = ucfirst ( wfMsg ( "category" ) ) ;
537 $ti = $wgTitle->getText() ;
538 $ti = explode ( ":" , $ti , 2 ) ;
539 if ( $cat != $ti[0] ) return "" ;
540 $r = "<br break=all>\n" ;
542 $articles = array() ;
543 $parents = array () ;
544 $children = array() ;
547 global $wgUser ;
548 $sk = $wgUser->getSkin() ;
549 $sql = "SELECT l_from FROM links WHERE l_to={$id}" ;
550 $res = wfQuery ( $sql, DB_READ ) ;
551 while ( $x = wfFetchObject ( $res ) )
553 # $t = new Title ;
554 # $t->newFromDBkey ( $x->l_from ) ;
555 # $t = $t->getText() ;
556 $t = $x->l_from ;
557 $y = explode ( ":" , $t , 2 ) ;
558 if ( count ( $y ) == 2 && $y[0] == $cat ) {
559 array_push ( $children , $sk->makeLink ( $t , $y[1] ) ) ;
560 } else {
561 array_push ( $articles , $sk->makeLink ( $t ) ) ;
564 wfFreeResult ( $res ) ;
566 # Children
567 if ( count ( $children ) > 0 )
569 asort ( $children ) ;
570 $r .= "<h2>".wfMsg("subcategories")."</h2>\n" ;
571 $r .= implode ( ", " , $children ) ;
574 # Articles
575 if ( count ( $articles ) > 0 )
577 asort ( $articles ) ;
578 $h = str_replace ( "$1" , $ti[1] , wfMsg("category_header") ) ;
579 $r .= "<h2>{$h}</h2>\n" ;
580 $r .= implode ( ", " , $articles ) ;
584 return $r ;
587 function getHTMLattrs ()
589 $htmlattrs = array( # Allowed attributes--no scripting, etc.
590 "title", "align", "lang", "dir", "width", "height",
591 "bgcolor", "clear", /* BR */ "noshade", /* HR */
592 "cite", /* BLOCKQUOTE, Q */ "size", "face", "color",
593 /* FONT */ "type", "start", "value", "compact",
594 /* For various lists, mostly deprecated but safe */
595 "summary", "width", "border", "frame", "rules",
596 "cellspacing", "cellpadding", "valign", "char",
597 "charoff", "colgroup", "col", "span", "abbr", "axis",
598 "headers", "scope", "rowspan", "colspan", /* Tables */
599 "id", "class", "name", "style" /* For CSS */
601 return $htmlattrs ;
604 function fixTableTags ( $t )
606 if ( trim ( $t ) == "" ) return "" ; # Saves runtime ;-)
607 $htmlattrs = $this->getHTMLattrs() ;
609 # Strip non-approved attributes from the tag
610 $t = preg_replace(
611 "/(\\w+)(\\s*=\\s*([^\\s\">]+|\"[^\">]*\"))?/e",
612 "(in_array(strtolower(\"\$1\"),\$htmlattrs)?(\"\$1\".((\"x\$3\" != \"x\")?\"=\$3\":'')):'')",
613 $t);
615 return trim ( $t ) ;
618 function doTableStuff ( $t )
620 $t = explode ( "\n" , $t ) ;
621 $td = array () ; # Is currently a td tag open?
622 $ltd = array () ; # Was it TD or TH?
623 $tr = array () ; # Is currently a tr tag open?
624 $ltr = array () ; # tr attributes
625 foreach ( $t AS $k => $x )
627 $x = rtrim ( $x ) ;
628 $fc = substr ( $x , 0 , 1 ) ;
629 if ( "{|" == substr ( $x , 0 , 2 ) )
631 $t[$k] = "<table " . $this->fixTableTags ( substr ( $x , 3 ) ) . ">" ;
632 array_push ( $td , false ) ;
633 array_push ( $ltd , "" ) ;
634 array_push ( $tr , false ) ;
635 array_push ( $ltr , "" ) ;
637 else if ( count ( $td ) == 0 ) { } # Don't do any of the following
638 else if ( "|}" == substr ( $x , 0 , 2 ) )
640 $z = "</table>\n" ;
641 $l = array_pop ( $ltd ) ;
642 if ( array_pop ( $tr ) ) $z = "</tr>" . $z ;
643 if ( array_pop ( $td ) ) $z = "</{$l}>" . $z ;
644 array_pop ( $ltr ) ;
645 $t[$k] = $z ;
647 /* else if ( "|_" == substr ( $x , 0 , 2 ) ) # Caption
649 $z = trim ( substr ( $x , 2 ) ) ;
650 $t[$k] = "<caption>{$z}</caption>\n" ;
652 else if ( "|-" == substr ( $x , 0 , 2 ) ) # Allows for |---------------
654 $x = substr ( $x , 1 ) ;
655 while ( $x != "" && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
656 $z = "" ;
657 $l = array_pop ( $ltd ) ;
658 if ( array_pop ( $tr ) ) $z = "</tr>" . $z ;
659 if ( array_pop ( $td ) ) $z = "</{$l}>" . $z ;
660 array_pop ( $ltr ) ;
661 $t[$k] = $z ;
662 array_push ( $tr , false ) ;
663 array_push ( $td , false ) ;
664 array_push ( $ltd , "" ) ;
665 array_push ( $ltr , $this->fixTableTags ( $x ) ) ;
667 else if ( "|" == $fc || "!" == $fc || "|+" == substr ( $x , 0 , 2 ) ) # Caption
669 if ( "|+" == substr ( $x , 0 , 2 ) )
671 $fc = "+" ;
672 $x = substr ( $x , 1 ) ;
674 $after = substr ( $x , 1 ) ;
675 if ( $fc == "!" ) $after = str_replace ( "!!" , "||" , $after ) ;
676 $after = explode ( "||" , $after ) ;
677 $t[$k] = "" ;
678 foreach ( $after AS $theline )
680 $z = "" ;
681 $tra = array_pop ( $ltr ) ;
682 if ( !array_pop ( $tr ) ) $z = "<tr {$tra}>\n" ;
683 array_push ( $tr , true ) ;
684 array_push ( $ltr , "" ) ;
686 $l = array_pop ( $ltd ) ;
687 if ( array_pop ( $td ) ) $z = "</{$l}>" . $z ;
688 if ( $fc == "|" ) $l = "TD" ;
689 else if ( $fc == "!" ) $l = "TH" ;
690 else if ( $fc == "+" ) $l = "CAPTION" ;
691 else $l = "" ;
692 array_push ( $ltd , $l ) ;
693 $y = explode ( "|" , $theline , 2 ) ;
694 if ( count ( $y ) == 1 ) $y = "{$z}<{$l}>{$y[0]}" ;
695 else $y = $y = "{$z}<{$l} ".$this->fixTableTags($y[0]).">{$y[1]}" ;
696 $t[$k] .= $y ;
697 array_push ( $td , true ) ;
702 # Closing open td, tr && table
703 while ( count ( $td ) > 0 )
705 if ( array_pop ( $td ) ) $t[] = "</td>" ;
706 if ( array_pop ( $tr ) ) $t[] = "</tr>" ;
707 $t[] = "</table>" ;
710 $t = implode ( "\n" , $t ) ;
711 # $t = $this->removeHTMLtags( $t );
712 return $t ;
715 # Well, OK, it's actually about 14 passes. But since all the
716 # hard lifting is done inside PHP's regex code, it probably
717 # wouldn't speed things up much to add a real parser.
719 function doWikiPass2( $text, $linestart )
721 global $wgUser, $wgLang, $wgUseDynamicDates;
722 $fname = "OutputPage::doWikiPass2";
723 wfProfileIn( $fname );
725 $text = $this->removeHTMLtags( $text );
726 $text = $this->replaceVariables( $text );
728 $text = preg_replace( "/(^|\n)-----*/", "\\1<hr>", $text );
729 $text = str_replace ( "<HR>", "<hr>", $text );
731 $text = $this->doAllQuotes( $text );
732 $text = $this->doHeadings( $text );
733 $text = $this->doBlockLevels( $text, $linestart );
735 if($wgUseDynamicDates) {
736 $text = $wgLang->replaceDates( $text );
739 $text = $this->replaceExternalLinks( $text );
740 $text = $this->replaceInternalLinks ( $text );
741 $text = $this->doTableStuff ( $text ) ;
743 $text = $this->magicISBN( $text );
744 $text = $this->magicRFC( $text );
745 $text = $this->formatHeadings( $text );
747 $sk = $wgUser->getSkin();
748 $text = $sk->transformContent( $text );
749 $text .= $this->categoryMagic () ;
751 wfProfileOut( $fname );
752 return $text;
755 /* private */ function doAllQuotes( $text )
757 $outtext = "";
758 $lines = explode( "\r\n", $text );
759 foreach ( $lines as $line ) {
760 $outtext .= $this->doQuotes ( "", $line, "" ) . "\r\n";
762 return $outtext;
765 /* private */ function doQuotes( $pre, $text, $mode )
767 if ( preg_match( "/^(.*)''(.*)$/sU", $text, $m ) ) {
768 $m1_strong = ($m[1] == "") ? "" : "<strong>{$m[1]}</strong>";
769 $m1_em = ($m[1] == "") ? "" : "<em>{$m[1]}</em>";
770 if ( substr ($m[2], 0, 1) == "'" ) {
771 $m[2] = substr ($m[2], 1);
772 if ($mode == "em") {
773 return $this->doQuotes ( $m[1], $m[2], ($m[1] == "") ? "both" : "emstrong" );
774 } else if ($mode == "strong") {
775 return $m1_strong . $this->doQuotes ( "", $m[2], "" );
776 } else if (($mode == "emstrong") || ($mode == "both")) {
777 return $this->doQuotes ( "", $pre.$m1_strong.$m[2], "em" );
778 } else if ($mode == "strongem") {
779 return "<strong>{$pre}{$m1_em}</strong>" . $this->doQuotes ( "", $m[2], "em" );
780 } else {
781 return $m[1] . $this->doQuotes ( "", $m[2], "strong" );
783 } else {
784 if ($mode == "strong") {
785 return $this->doQuotes ( $m[1], $m[2], ($m[1] == "") ? "both" : "strongem" );
786 } else if ($mode == "em") {
787 return $m1_em . $this->doQuotes ( "", $m[2], "" );
788 } else if ($mode == "emstrong") {
789 return "<em>{$pre}{$m1_strong}</em>" . $this->doQuotes ( "", $m[2], "strong" );
790 } else if (($mode == "strongem") || ($mode == "both")) {
791 return $this->doQuotes ( "", $pre.$m1_em.$m[2], "strong" );
792 } else {
793 return $m[1] . $this->doQuotes ( "", $m[2], "em" );
796 } else {
797 $text_strong = ($text == "") ? "" : "<strong>{$text}</strong>";
798 $text_em = ($text == "") ? "" : "<em>{$text}</em>";
799 if ($mode == "") {
800 return $pre . $text;
801 } else if ($mode == "em") {
802 return $pre . $text_em;
803 } else if ($mode == "strong") {
804 return $pre . $text_strong;
805 } else if ($mode == "strongem") {
806 return (($pre == "") && ($text == "")) ? "" : "<strong>{$pre}{$text_em}</strong>";
807 } else {
808 return (($pre == "") && ($text == "")) ? "" : "<em>{$pre}{$text_strong}</em>";
813 /* private */ function doHeadings( $text )
815 for ( $i = 6; $i >= 1; --$i ) {
816 $h = substr( "======", 0, $i );
817 $text = preg_replace( "/^{$h}([^=]+){$h}(\\s|$)/m",
818 "<h{$i}>\\1</h{$i}>\\2", $text );
820 return $text;
823 # Note: we have to do external links before the internal ones,
824 # and otherwise take great care in the order of things here, so
825 # that we don't end up interpreting some URLs twice.
827 /* private */ function replaceExternalLinks( $text )
829 $fname = "OutputPage::replaceExternalLinks";
830 wfProfileIn( $fname );
831 $text = $this->subReplaceExternalLinks( $text, "http", true );
832 $text = $this->subReplaceExternalLinks( $text, "https", true );
833 $text = $this->subReplaceExternalLinks( $text, "ftp", false );
834 $text = $this->subReplaceExternalLinks( $text, "gopher", false );
835 $text = $this->subReplaceExternalLinks( $text, "news", false );
836 $text = $this->subReplaceExternalLinks( $text, "mailto", false );
837 wfProfileOut( $fname );
838 return $text;
841 /* private */ function subReplaceExternalLinks( $s, $protocol, $autonumber )
843 global $wgUser, $printable;
844 global $wgAllowExternalImages;
847 $unique = "4jzAfzB8hNvf4sqyO9Edd8pSmk9rE2in0Tgw3";
848 $uc = "A-Za-z0-9_\\/~%\\-+&*#?!=()@\\x80-\\xFF";
850 # this is the list of separators that should be ignored if they
851 # are the last character of an URL but that should be included
852 # if they occur within the URL, e.g. "go to www.foo.com, where .."
853 # in this case, the last comma should not become part of the URL,
854 # but in "www.foo.com/123,2342,32.htm" it should.
855 $sep = ",;\.:";
856 $fnc = "A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF";
857 $images = "gif|png|jpg|jpeg";
859 # PLEASE NOTE: The curly braces { } are not part of the regex,
860 # they are interpreted as part of the string (used to tell PHP
861 # that the content of the string should be inserted there).
862 $e1 = "/(^|[^\\[])({$protocol}:)([{$uc}{$sep}]+)\\/([{$fnc}]+)\\." .
863 "((?i){$images})([^{$uc}]|$)/";
865 $e2 = "/(^|[^\\[])({$protocol}:)(([".$uc."]|[".$sep."][".$uc."])+)([^". $uc . $sep. "]|[".$sep."]|$)/";
866 $sk = $wgUser->getSkin();
868 if ( $autonumber and $wgAllowExternalImages) { # Use img tags only for HTTP urls
869 $s = preg_replace( $e1, "\\1" . $sk->makeImage( "{$unique}:\\3" .
870 "/\\4.\\5", "\\4.\\5" ) . "\\6", $s );
872 $s = preg_replace( $e2, "\\1" . "<a href=\"{$unique}:\\3\"" .
873 $sk->getExternalLinkAttributes( "{$unique}:\\3", wfEscapeHTML(
874 "{$unique}:\\3" ) ) . ">" . wfEscapeHTML( "{$unique}:\\3" ) .
875 "</a>\\5", $s );
876 $s = str_replace( $unique, $protocol, $s );
878 $a = explode( "[{$protocol}:", " " . $s );
879 $s = array_shift( $a );
880 $s = substr( $s, 1 );
882 $e1 = "/^([{$uc}"."{$sep}]+)](.*)\$/sD";
883 $e2 = "/^([{$uc}"."{$sep}]+)\\s+([^\\]]+)](.*)\$/sD";
885 foreach ( $a as $line ) {
886 if ( preg_match( $e1, $line, $m ) ) {
887 $link = "{$protocol}:{$m[1]}";
888 $trail = $m[2];
889 if ( $autonumber ) { $text = "[" . ++$this->mAutonumber . "]"; }
890 else { $text = wfEscapeHTML( $link ); }
891 } else if ( preg_match( $e2, $line, $m ) ) {
892 $link = "{$protocol}:{$m[1]}";
893 $text = $m[2];
894 $trail = $m[3];
895 } else {
896 $s .= "[{$protocol}:" . $line;
897 continue;
899 if ( $printable == "yes") $paren = " (<i>" . htmlspecialchars ( $link ) . "</i>)";
900 else $paren = "";
901 $la = $sk->getExternalLinkAttributes( $link, $text );
902 $s .= "<a href='{$link}'{$la}>{$text}</a>{$paren}{$trail}";
905 return $s;
908 /* private */ function replaceInternalLinks( $s )
910 global $wgTitle, $wgUser, $wgLang;
911 global $wgLinkCache, $wgInterwikiMagic, $wgUseCategoryMagic;
912 global $wgNamespacesWithSubpages, $wgLanguageCode;
913 wfProfileIn( $fname = "OutputPage::replaceInternalLinks" );
915 wfProfileIn( "$fname-setup" );
916 $tc = Title::legalChars() . "#";
917 $sk = $wgUser->getSkin();
919 $a = explode( "[[", " " . $s );
920 $s = array_shift( $a );
921 $s = substr( $s, 1 );
923 $e1 = "/^([{$tc}]+)\\|([^]]+)]](.*)\$/sD";
924 $e2 = "/^([{$tc}]+)]](.*)\$/sD";
925 wfProfileOut( "$fname-setup" );
927 foreach ( $a as $line ) {
928 if ( preg_match( $e1, $line, $m ) ) { # page with alternate text
930 $text = $m[2];
931 $trail = $m[3];
933 } else if ( preg_match( $e2, $line, $m ) ) { # page with normal text
935 $text = "";
936 $trail = $m[2];
939 else { # Invalid form; output directly
940 $s .= "[[" . $line ;
941 wfProfileOut( "$fname-loop1" );
942 continue;
944 if(substr($m[1],0,1)=="/") { # subpage
945 if(substr($m[1],-1,1)=="/") { # / at end means we don't want the slash to be shown
946 $m[1]=substr($m[1],1,strlen($m[1])-2);
947 $noslash=$m[1];
949 } else {
950 $noslash=substr($m[1],1);
952 if($wgNamespacesWithSubpages[$wgTitle->getNamespace()]) { # subpages allowed here
953 $link = $wgTitle->getPrefixedText(). "/" . trim($noslash);
954 if(!$text) {
955 $text= $m[1];
956 } # this might be changed for ugliness reasons
957 } else {
958 $link = $noslash; # no subpage allowed, use standard link
960 } else { # no subpage
961 $link = $m[1];
964 if ( preg_match( "/^((?:i|x|[a-z]{2,3})(?:-[a-z0-9]+)?|[A-Za-z\\x80-\\xff]+):(.*)\$/", $link, $m ) ) {
965 $pre = strtolower( $m[1] );
966 $suf = trim($m[2]);
967 if( empty( $suf ) ) {
968 $s .= $trail;
969 } else if ( $wgLang->getNsIndex( $pre ) ==
970 Namespace::getImage() ) {
971 $nt = Title::newFromText( $suf );
972 $name = $nt->getDBkey();
973 if ( "" == $text ) { $text = $nt->GetText(); }
975 $wgLinkCache->addImageLink( $name );
976 $s .= $sk->makeImageLink( $name,
977 wfImageUrl( $name ), $text );
978 $s .= $trail;
979 } else if ( "media" == $pre ) {
980 $nt = Title::newFromText( $suf );
981 $name = $nt->getDBkey();
982 if ( "" == $text ) { $text = $nt->GetText(); }
984 $wgLinkCache->addImageLink( $name );
985 $s .= $sk->makeMediaLink( $name,
986 wfImageUrl( $name ), $text );
987 $s .= $trail;
988 } else if ( isset($wgUseCategoryMagic) && $wgUseCategoryMagic && $pre == wfMsg ( "category" ) ) {
989 $l = $sk->makeLink ( $pre.":".ucfirst( $m[2] ), ucfirst ( $m[2] ) ) ;
990 array_push ( $this->mCategoryLinks , $l ) ;
991 $s .= $trail ;
992 } else {
993 $l = $wgLang->getLanguageName( $pre );
994 if ( "" == $l or !$wgInterwikiMagic or Namespace::isTalk( $wgTitle->getNamespace() ) ) {
995 if ( "" == $text ) {
996 $text = $link;
998 $s .= $sk->makeLink( $link, $text, "", $trail );
999 } else if ( $pre != $wgLanguageCode ) {
1000 array_push( $this->mLanguageLinks, "$pre:$suf" );
1001 $s .= $trail;
1004 # } else if ( 0 == strcmp( "##", substr( $link, 0, 2 ) ) ) {
1005 # $link = substr( $link, 2 );
1006 # $s .= "<a name=\"{$link}\">{$text}</a>{$trail}";
1007 } else {
1008 if ( "" == $text ) { $text = $link; }
1009 # Hotspot:
1010 $s .= $sk->makeLink( $link, $text, "", $trail );
1013 wfProfileOut( $fname );
1014 return $s;
1017 # Some functions here used by doBlockLevels()
1019 /* private */ function closeParagraph()
1021 $result = "";
1022 if ( 0 != strcmp( "p", $this->mLastSection ) &&
1023 0 != strcmp( "", $this->mLastSection ) ) {
1024 $result = "</" . $this->mLastSection . ">";
1026 $this->mLastSection = "";
1027 return $result;
1029 # getCommon() returns the length of the longest common substring
1030 # of both arguments, starting at the beginning of both.
1032 /* private */ function getCommon( $st1, $st2 )
1034 $fl = strlen( $st1 );
1035 $shorter = strlen( $st2 );
1036 if ( $fl < $shorter ) { $shorter = $fl; }
1038 for ( $i = 0; $i < $shorter; ++$i ) {
1039 if ( $st1{$i} != $st2{$i} ) { break; }
1041 return $i;
1043 # These next three functions open, continue, and close the list
1044 # element appropriate to the prefix character passed into them.
1046 /* private */ function openList( $char )
1048 $result = $this->closeParagraph();
1050 if ( "*" == $char ) { $result .= "<ul><li>"; }
1051 else if ( "#" == $char ) { $result .= "<ol><li>"; }
1052 else if ( ":" == $char ) { $result .= "<dl><dd>"; }
1053 else if ( ";" == $char ) {
1054 $result .= "<dl><dt>";
1055 $this->mDTopen = true;
1057 else { $result = "<!-- ERR 1 -->"; }
1059 return $result;
1062 /* private */ function nextItem( $char )
1064 if ( "*" == $char || "#" == $char ) { return "</li><li>"; }
1065 else if ( ":" == $char || ";" == $char ) {
1066 $close = "</dd>";
1067 if ( $this->mDTopen ) { $close = "</dt>"; }
1068 if ( ";" == $char ) {
1069 $this->mDTopen = true;
1070 return $close . "<dt>";
1071 } else {
1072 $this->mDTopen = false;
1073 return $close . "<dd>";
1076 return "<!-- ERR 2 -->";
1079 /* private */function closeList( $char )
1081 if ( "*" == $char ) { return "</li></ul>"; }
1082 else if ( "#" == $char ) { return "</li></ol>"; }
1083 else if ( ":" == $char ) {
1084 if ( $this->mDTopen ) {
1085 $this->mDTopen = false;
1086 return "</dt></dl>";
1087 } else {
1088 return "</dd></dl>";
1091 return "<!-- ERR 3 -->";
1094 /* private */ function doBlockLevels( $text, $linestart )
1096 $fname = "OutputPage::doBlockLevels";
1097 wfProfileIn( $fname );
1098 # Parsing through the text line by line. The main thing
1099 # happening here is handling of block-level elements p, pre,
1100 # and making lists from lines starting with * # : etc.
1102 $a = explode( "\n", $text );
1103 $text = $lastPref = "";
1104 $this->mDTopen = $inBlockElem = false;
1106 if ( ! $linestart ) { $text .= array_shift( $a ); }
1107 foreach ( $a as $t ) {
1108 if ( "" != $text ) { $text .= "\n"; }
1110 $oLine = $t;
1111 $opl = strlen( $lastPref );
1112 $npl = strspn( $t, "*#:;" );
1113 $pref = substr( $t, 0, $npl );
1114 $pref2 = str_replace( ";", ":", $pref );
1115 $t = substr( $t, $npl );
1117 if ( 0 != $npl && 0 == strcmp( $lastPref, $pref2 ) ) {
1118 $text .= $this->nextItem( substr( $pref, -1 ) );
1120 if ( ";" == substr( $pref, -1 ) ) {
1121 $cpos = strpos( $t, ":" );
1122 if ( ! ( false === $cpos ) ) {
1123 $term = substr( $t, 0, $cpos );
1124 $text .= $term . $this->nextItem( ":" );
1125 $t = substr( $t, $cpos + 1 );
1128 } else if (0 != $npl || 0 != $opl) {
1129 $cpl = $this->getCommon( $pref, $lastPref );
1131 while ( $cpl < $opl ) {
1132 $text .= $this->closeList( $lastPref{$opl-1} );
1133 --$opl;
1135 if ( $npl <= $cpl && $cpl > 0 ) {
1136 $text .= $this->nextItem( $pref{$cpl-1} );
1138 while ( $npl > $cpl ) {
1139 $char = substr( $pref, $cpl, 1 );
1140 $text .= $this->openList( $char );
1142 if ( ";" == $char ) {
1143 $cpos = strpos( $t, ":" );
1144 if ( ! ( false === $cpos ) ) {
1145 $term = substr( $t, 0, $cpos );
1146 $text .= $term . $this->nextItem( ":" );
1147 $t = substr( $t, $cpos + 1 );
1150 ++$cpl;
1152 $lastPref = $pref2;
1154 if ( 0 == $npl ) { # No prefix--go to paragraph mode
1155 if ( preg_match(
1156 "/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6)/i", $t ) ) {
1157 $text .= $this->closeParagraph();
1158 $inBlockElem = true;
1160 if ( ! $inBlockElem ) {
1161 if ( " " == $t{0} ) {
1162 $newSection = "pre";
1163 # $t = wfEscapeHTML( $t );
1165 else { $newSection = "p"; }
1167 if ( 0 == strcmp( "", trim( $oLine ) ) ) {
1168 $text .= $this->closeParagraph();
1169 $text .= "<" . $newSection . ">";
1170 } else if ( 0 != strcmp( $this->mLastSection,
1171 $newSection ) ) {
1172 $text .= $this->closeParagraph();
1173 if ( 0 != strcmp( "p", $newSection ) ) {
1174 $text .= "<" . $newSection . ">";
1177 $this->mLastSection = $newSection;
1179 if ( $inBlockElem &&
1180 preg_match( "/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6)/i", $t ) ) {
1181 $inBlockElem = false;
1184 $text .= $t;
1186 while ( $npl ) {
1187 $text .= $this->closeList( $pref2{$npl-1} );
1188 --$npl;
1190 if ( "" != $this->mLastSection ) {
1191 if ( "p" != $this->mLastSection ) {
1192 $text .= "</" . $this->mLastSection . ">";
1194 $this->mLastSection = "";
1196 wfProfileOut( $fname );
1197 return $text;
1200 /* private */ function replaceVariables( $text )
1202 global $wgLang;
1203 $fname = "OutputPage::replaceVariables";
1204 wfProfileIn( $fname );
1206 /* As with sigs, use server's local time --
1207 ensure this is appropriate for your audience! */
1208 $v = date( "m" );
1209 $mw =& MagicWord::get( MAG_CURRENTMONTH );
1210 $text = $mw->replace( $v, $text );
1212 $v = $wgLang->getMonthName( date( "n" ) );
1213 $mw =& MagicWord::get( MAG_CURRENTMONTHNAME );
1214 $text = $mw->replace( $v, $text );
1216 $v = $wgLang->getMonthNameGen( date( "n" ) );
1217 $mw =& MagicWord::get( MAG_CURRENTMONTHNAMEGEN );
1218 $text = $mw->replace( $v, $text );
1220 $v = date( "j" );
1221 $mw = MagicWord::get( MAG_CURRENTDAY );
1222 $text = $mw->replace( $v, $text );
1224 $v = $wgLang->getWeekdayName( date( "w" )+1 );
1225 $mw =& MagicWord::get( MAG_CURRENTDAYNAME );
1226 $text = $mw->replace( $v, $text );
1228 $v = date( "Y" );
1229 $mw =& MagicWord::get( MAG_CURRENTYEAR );
1230 $text = $mw->replace( $v, $text );
1232 $v = $wgLang->time( wfTimestampNow(), false );
1233 $mw =& MagicWord::get( MAG_CURRENTTIME );
1234 $text = $mw->replace( $v, $text );
1236 $mw =& MagicWord::get( MAG_NUMBEROFARTICLES );
1237 if ( $mw->match( $text ) ) {
1238 $v = wfNumberOfArticles();
1239 $text = $mw->replace( $v, $text );
1242 # The callbacks are in GlobalFunctions.php
1243 $mw =& MagicWord::get( MAG_MSG );
1244 $text = $mw->substituteCallback( $text, "replaceMsgVar" );
1246 $mw =& MagicWord::get( MAG_MSGNW );
1247 $text = $mw->substituteCallback( $text, "replaceMsgVarNw" );
1249 wfProfileOut( $fname );
1250 return $text;
1253 /* private */ function removeHTMLtags( $text )
1255 $fname = "OutputPage::removeHTMLtags";
1256 wfProfileIn( $fname );
1257 $htmlpairs = array( # Tags that must be closed
1258 "b", "i", "u", "font", "big", "small", "sub", "sup", "h1",
1259 "h2", "h3", "h4", "h5", "h6", "cite", "code", "em", "s",
1260 "strike", "strong", "tt", "var", "div", "center",
1261 "blockquote", "ol", "ul", "dl", "table", "caption", "pre",
1262 "ruby", "rt" , "rb" , "rp"
1264 $htmlsingle = array(
1265 "br", "p", "hr", "li", "dt", "dd"
1267 $htmlnest = array( # Tags that can be nested--??
1268 "table", "tr", "td", "th", "div", "blockquote", "ol", "ul",
1269 "dl", "font", "big", "small", "sub", "sup"
1271 $tabletags = array( # Can only appear inside table
1272 "td", "th", "tr"
1275 $htmlsingle = array_merge( $tabletags, $htmlsingle );
1276 $htmlelements = array_merge( $htmlsingle, $htmlpairs );
1278 $htmlattrs = $this->getHTMLattrs () ;
1280 # Remove HTML comments
1281 $text = preg_replace( "/<!--.*-->/sU", "", $text );
1283 $bits = explode( "<", $text );
1284 $text = array_shift( $bits );
1285 $tagstack = array(); $tablestack = array();
1287 foreach ( $bits as $x ) {
1288 $prev = error_reporting( E_ALL & ~( E_NOTICE | E_WARNING ) );
1289 preg_match( "/^(\\/?)(\\w+)([^>]*)(\\/{0,1}>)([^<]*)$/",
1290 $x, $regs );
1291 list( $qbar, $slash, $t, $params, $brace, $rest ) = $regs;
1292 error_reporting( $prev );
1294 $badtag = 0 ;
1295 if ( in_array( $t = strtolower( $t ), $htmlelements ) ) {
1296 # Check our stack
1297 if ( $slash ) {
1298 # Closing a tag...
1299 if ( ! in_array( $t, $htmlsingle ) &&
1300 ( $ot = array_pop( $tagstack ) ) != $t ) {
1301 array_push( $tagstack, $ot );
1302 $badtag = 1;
1303 } else {
1304 if ( $t == "table" ) {
1305 $tagstack = array_pop( $tablestack );
1307 $newparams = "";
1309 } else {
1310 # Keep track for later
1311 if ( in_array( $t, $tabletags ) &&
1312 ! in_array( "table", $tagstack ) ) {
1313 $badtag = 1;
1314 } else if ( in_array( $t, $tagstack ) &&
1315 ! in_array ( $t , $htmlnest ) ) {
1316 $badtag = 1 ;
1317 } else if ( ! in_array( $t, $htmlsingle ) ) {
1318 if ( $t == "table" ) {
1319 array_push( $tablestack, $tagstack );
1320 $tagstack = array();
1322 array_push( $tagstack, $t );
1324 # Strip non-approved attributes from the tag
1325 $newparams = preg_replace(
1326 "/(\\w+)(\\s*=\\s*([^\\s\">]+|\"[^\">]*\"))?/e",
1327 "(in_array(strtolower(\"\$1\"),\$htmlattrs)?(\"\$1\".((\"x\$3\" != \"x\")?\"=\$3\":'')):'')",
1328 $params);
1330 if ( ! $badtag ) {
1331 $rest = str_replace( ">", "&gt;", $rest );
1332 $text .= "<$slash$t$newparams$brace$rest";
1333 continue;
1336 $text .= "&lt;" . str_replace( ">", "&gt;", $x);
1338 # Close off any remaining tags
1339 while ( $t = array_pop( $tagstack ) ) {
1340 $text .= "</$t>\n";
1341 if ( $t == "table" ) { $tagstack = array_pop( $tablestack ); }
1343 wfProfileOut( $fname );
1344 return $text;
1350 * This function accomplishes several tasks:
1351 * 1) Auto-number headings if that option is enabled
1352 * 2) Add an [edit] link to sections for logged in users who have enabled the option
1353 * 3) Add a Table of contents on the top for users who have enabled the option
1354 * 4) Auto-anchor headings
1356 * It loops through all headlines, collects the necessary data, then splits up the
1357 * string and re-inserts the newly formatted headlines.
1359 * */
1360 /* private */ function formatHeadings( $text )
1362 global $wgUser,$wgArticle,$wgTitle,$wpPreview;
1363 $nh=$wgUser->getOption( "numberheadings" );
1364 $st=$wgUser->getOption( "showtoc" );
1365 if(!$wgTitle->userCanEdit()) {
1366 $es=0;
1367 $esr=0;
1368 } else {
1369 $es=$wgUser->getID() && $wgUser->getOption( "editsection" );
1370 $esr=$wgUser->getID() && $wgUser->getOption( "editsectiononrightclick" );
1373 # Inhibit editsection links if requested in the page
1374 if ($es) {
1375 $esw=& MagicWord::get(MAG_NOEDITSECTION);
1376 if ($esw->matchAndRemove( $text )) {
1377 $es=0;
1380 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
1381 # do not add TOC
1382 $mw =& MagicWord::get( MAG_NOTOC );
1383 if ($mw->matchAndRemove( $text ))
1385 $st = 0;
1388 # never add the TOC to the Main Page. This is an entry page that should not
1389 # be more than 1-2 screens large anyway
1390 if($wgTitle->getPrefixedText()==wfMsg("mainpage")) {$st=0;}
1392 # We need this to perform operations on the HTML
1393 $sk=$wgUser->getSkin();
1395 # Get all headlines for numbering them and adding funky stuff like [edit]
1396 # links
1397 preg_match_all("/<H([1-6])(.*?>)(.*?)<\/H[1-6]>/i",$text,$matches);
1399 # headline counter
1400 $c=0;
1402 # Ugh .. the TOC should have neat indentation levels which can be
1403 # passed to the skin functions. These are determined here
1404 foreach($matches[3] as $headline) {
1405 if($level) { $prevlevel=$level;}
1406 $level=$matches[1][$c];
1407 if(($nh||$st) && $prevlevel && $level>$prevlevel) {
1409 $h[$level]=0; // reset when we enter a new level
1410 $toc.=$sk->tocIndent($level-$prevlevel);
1411 $toclevel+=$level-$prevlevel;
1414 if(($nh||$st) && $level<$prevlevel) {
1415 $h[$level+1]=0; // reset when we step back a level
1416 $toc.=$sk->tocUnindent($prevlevel-$level);
1417 $toclevel-=$prevlevel-$level;
1420 $h[$level]++; // count number of headlines for each level
1422 if($nh||$st) {
1423 for($i=1;$i<=$level;$i++) {
1424 if($h[$i]) {
1425 if($dot) {$numbering.=".";}
1426 $numbering.=$h[$i];
1427 $dot=1;
1433 $canonized_headline=preg_replace("/<.*?>/","",$headline); // strip out HTML
1434 $tocline=$canonized_headline;
1435 $canonized_headline=str_replace('"',"",$canonized_headline);
1436 $canonized_headline=str_replace(" ","_",trim($canonized_headline));
1437 $refer[$c]=$canonized_headline;
1438 $refers[$canonized_headline]++; // count how many in assoc. array so we can track dupes in anchors
1439 $refcount[$c]=$refers[$canonized_headline];
1440 if($nh||$st) {
1441 $tocline=$numbering ." ". $tocline;
1442 if($nh) {
1443 $headline=$numbering . " " . $headline; // the two are different if the line contains a link
1446 $anchor=$canonized_headline;
1447 if($refcount[$c]>1) {$anchor.="_".$refcount[$c];}
1448 if($st) {
1449 $toc.=$sk->tocLine($anchor,$tocline,$toclevel);
1451 if($es && !isset($wpPreview)) {
1452 $head[$c].=$sk->editSectionLink($c+1);
1454 $head[$c].="<H".$level.$matches[2][$c]
1455 ."<a name=\"".$anchor."\">"
1456 .$headline
1457 ."</a>"
1458 ."</H".$level.">";
1459 if($esr && !isset($wpPreview)) {
1460 $head[$c]=$sk->editSectionScript($c+1,$head[$c]);
1462 $numbering="";
1463 $c++;
1464 $dot=0;
1467 if($st) {
1468 $toclines=$c;
1469 $toc.=$sk->tocUnindent($toclevel);
1470 $toc=$sk->tocTable($toc);
1473 // split up and insert constructed headlines
1475 $blocks=preg_split("/<H[1-6].*?>.*?<\/H[1-6]>/i",$text);
1476 $i=0;
1479 foreach($blocks as $block) {
1480 if(($es) && !isset($wpPreview) && $c>0 && $i==0) {
1481 # This is the [edit] link that appears for the top block of text when
1482 # section editing is enabled
1483 $full.=$sk->editSectionLink(0);
1485 $full.=$block;
1486 if($st && $toclines>3 && !$i) {
1487 # Let's add a top anchor just in case we want to link to the top of the page
1488 $full="<a name=\"top\"></a>".$full.$toc;
1491 $full.=$head[$i];
1492 $i++;
1494 return $full;
1497 /* private */ function magicISBN( $text )
1499 global $wgLang;
1501 $a = split( "ISBN ", " $text" );
1502 if ( count ( $a ) < 2 ) return $text;
1503 $text = substr( array_shift( $a ), 1);
1504 $valid = "0123456789-ABCDEFGHIJKLMNOPQRSTUVWXYZ";
1506 foreach ( $a as $x ) {
1507 $isbn = $blank = "" ;
1508 while ( " " == $x{0} ) {
1509 $blank .= " ";
1510 $x = substr( $x, 1 );
1512 while ( strstr( $valid, $x{0} ) != false ) {
1513 $isbn .= $x{0};
1514 $x = substr( $x, 1 );
1516 $num = str_replace( "-", "", $isbn );
1517 $num = str_replace( " ", "", $num );
1519 if ( "" == $num ) {
1520 $text .= "ISBN $blank$x";
1521 } else {
1522 $text .= "<a href=\"" . wfLocalUrlE( $wgLang->specialPage(
1523 "Booksources"), "isbn={$num}" ) . "\" CLASS=\"internal\">ISBN $isbn</a>";
1524 $text .= $x;
1527 return $text;
1530 /* private */ function magicRFC( $text )
1532 return $text;
1535 /* private */ function headElement()
1537 global $wgDocType, $wgDTD, $wgUser, $wgLanguageCode, $wgOutputEncoding, $wgLang;
1539 $ret = "<!DOCTYPE HTML PUBLIC \"$wgDocType\"\n \"$wgDTD\">\n";
1541 if ( "" == $this->mHTMLtitle ) {
1542 $this->mHTMLtitle = $this->mPagetitle;
1544 $rtl = $wgLang->isRTL() ? " dir='RTL'" : "";
1545 $ret .= "<html lang=\"$wgLanguageCode\"$rtl><head><title>{$this->mHTMLtitle}</title>\n";
1546 array_push( $this->mMetatags, array( "http:Content-type", "text/html; charset={$wgOutputEncoding}" ) );
1547 foreach ( $this->mMetatags as $tag ) {
1548 if ( 0 == strcasecmp( "http:", substr( $tag[0], 0, 5 ) ) ) {
1549 $a = "http-equiv";
1550 $tag[0] = substr( $tag[0], 5 );
1551 } else {
1552 $a = "name";
1554 $ret .= "<meta $a=\"{$tag[0]}\" content=\"{$tag[1]}\">\n";
1556 $p = $this->mRobotpolicy;
1557 if ( "" == $p ) { $p = "index,follow"; }
1558 $ret .= "<meta name=\"robots\" content=\"$p\">\n";
1560 if ( count( $this->mKeywords ) > 0 ) {
1561 $ret .= "<meta name=\"keywords\" content=\"" .
1562 implode( ",", $this->mKeywords ) . "\">\n";
1564 foreach ( $this->mLinktags as $tag ) {
1565 $ret .= "<link ";
1566 if ( "" != $tag[0] ) { $ret .= "rel=\"{$tag[0]}\" "; }
1567 if ( "" != $tag[1] ) { $ret .= "rev=\"{$tag[1]}\" "; }
1568 $ret .= "href=\"{$tag[2]}\">\n";
1570 $sk = $wgUser->getSkin();
1571 $ret .= $sk->getHeadScripts();
1572 $ret .= $sk->getUserStyles();
1574 $ret .= "</head>\n";
1575 return $ret;