includes/parser/Parser_LinkHooks.php

   1 <?php
   2 /**
   3  * Parser with LinkHooks experiment
   4  * @ingroup Parser
   5  */
   6 class Parser_LinkHooks extends Parser
   7 {
   8         /**
   9          * Update this version number when the ParserOutput format
  10          * changes in an incompatible way, so the parser cache
  11          * can automatically discard old data.
  12          */
  13         const VERSION = '1.6.4';
  14
  15         # Flags for Parser::setLinkHook
  16         # Also available as global constants from Defines.php
  17         const SLH_PATTERN = 1;
  18
  19         # Constants needed for external link processing
  20         # Everything except bracket, space, or control characters
  21         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  22         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  23                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  24
  25         /**#@+
  26          * @private
  27          */
  28         # Persistent:
  29         var $mLinkHooks;
  30
  31         /**#@-*/
  32
  33         /**
  34          * Constructor
  35          *
  36          * @public
  37          */
  38         function __construct( $conf = array() ) {
  39                 parent::__construct( $conf );
  40                 $this->mLinkHooks = array();
  41         }
  42
  43         /**
  44          * Do various kinds of initialisation on the first call of the parser
  45          */
  46         function firstCallInit() {
  47                 parent::__construct();
  48                 if ( !$this->mFirstCall ) {
  49                         return;
  50                 }
  51                 $this->mFirstCall = false;
  52
  53                 wfProfileIn( __METHOD__ );
  54
  55                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
  56                 CoreParserFunctions::register( $this );
  57                 CoreLinkFunctions::register( $this );
  58                 $this->initialiseVariables();
  59
  60                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
  61                 wfProfileOut( __METHOD__ );
  62         }
  63
  64         /**
  65          * Create a link hook, e.g. [[Namepsace:...|display}}
  66          * The callback function should have the form:
  67          *    function myLinkCallback( $parser, $holders, $markers,
  68          *      Title $title, $titleText, &$sortText = null, &$leadingColon = false ) { ... }
  69          *
  70          * Or with SLH_PATTERN:
  71          *    function myLinkCallback( $parser, $holders, $markers, )
  72          *      &$titleText, &$sortText = null, &$leadingColon = false ) { ... }
  73          *
  74          * The callback may either return a number of different possible values:
  75          * String) Text result of the link
  76          * True) (Treat as link) Parse the link according to normal link rules
  77          * False) (Bad link) Just output the raw wikitext (You may modify the text first)
  78          *
  79          * @public
  80          *
  81          * @param integer|string $ns The Namespace ID or regex pattern if SLH_PATTERN is set
  82          * @param mixed $callback The callback function (and object) to use
  83          * @param integer $flags a combination of the following flags:
  84          *     SLH_PATTERN   Use a regex link pattern rather than a namespace
  85          *
  86          * @return The old callback function for this name, if any
  87          */
  88         function setLinkHook( $ns, $callback, $flags = 0 ) {
  89                 if( $flags & SLH_PATTERN && !is_string($ns) )
  90                         throw new MWException( __METHOD__.'() expecting a regex string pattern.' );
  91                 elseif( $flags | ~SLH_PATTERN && !is_int($ns) )
  92                         throw new MWException( __METHOD__.'() expecting a namespace index.' );
  93                 $oldVal = isset( $this->mLinkHooks[$ns] ) ? $this->mLinkHooks[$ns][0] : null;
  94                 $this->mLinkHooks[$ns] = array( $callback, $flags );
  95                 return $oldVal;
  96         }
  97
  98         /**
  99          * Get all registered link hook identifiers
 100          *
 101          * @return array
 102          */
 103         function getLinkHooks() {
 104                 return array_keys( $this->mLinkHooks );
 105         }
 106
 107         /**
 108          * Process [[ ]] wikilinks
 109          * @return LinkHolderArray
 110          *
 111          * @private
 112          */
 113         function replaceInternalLinks2( &$s ) {
 114                 global $wgContLang;
 115
 116                 wfProfileIn( __METHOD__ );
 117
 118                 wfProfileIn( __METHOD__.'-setup' );
 119                 static $tc = FALSE, $titleRegex;//$e1, $e1_img;
 120                 if( !$tc ) {
 121                         # the % is needed to support urlencoded titles as well
 122                         $tc = Title::legalChars() . '#%';
 123                         # Match a link having the form [[namespace:link|alternate]]trail
 124                         //$e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
 125                         # Match cases where there is no "]]", which might still be images
 126                         //$e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
 127                         # Match a valid plain title
 128                         $titleRegex = "/^([{$tc}]+)$/sD";
 129                 }
 130
 131                 $sk = $this->mOptions->getSkin();
 132                 $holders = new LinkHolderArray( $this );
 133
 134                 if( is_null( $this->mTitle ) ) {
 135                         wfProfileOut( __METHOD__ );
 136                         wfProfileOut( __METHOD__.'-setup' );
 137                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
 138                 }
 139                 $nottalk = !$this->mTitle->isTalkPage();
 140
 141                 if($wgContLang->hasVariants()) {
 142                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
 143                 } else {
 144                         $selflink = array($this->mTitle->getPrefixedText());
 145                 }
 146                 wfProfileOut( __METHOD__.'-setup' );
 147
 148                 $offset = 0;
 149                 $offsetStack = array();
 150                 $markers = new LinkMarkerReplacer( $this, $holders, array( &$this, 'replaceInternalLinksCallback' ) );
 151                 while( true ) {
 152                         $startBracketOffset = strpos( $s, '[[', $offset );
 153                         $endBracketOffset   = strpos( $s, ']]', $offset );
 154                         # Finish when there are no more brackets
 155                         if( $startBracketOffset === false && $endBracketOffset === false ) break;
 156                         # Determine if the bracket is a starting or ending bracket
 157                         # When we find both, use the first one
 158                         elseif( $startBracketOffset !== false && $endBracketOffset !== false )
 159                              $isStart = $startBracketOffset <= $endBracketOffset;
 160                         # When we only found one, check which it is
 161                         else $isStart = $startBracketOffset !== false;
 162                         $bracketOffset = $isStart ? $startBracketOffset : $endBracketOffset;
 163                         if( $isStart ) {
 164                                 /** Opening bracket **/
 165                                 # Just push our current offset in the string onto the stack
 166                                 $offsetStack[] = $startBracketOffset;
 167                         } else {
 168                                 /** Closing bracket **/
 169                                 # Pop the start pos for our current link zone off the stack
 170                                 $startBracketOffset = array_pop($offsetStack);
 171                                 # Just to clean up the code, lets place offsets on the outer ends
 172                                 $endBracketOffset += 2;
 173
 174                                 # Only do logic if we actually have a opening bracket for this
 175                                 if( isset($startBracketOffset) ) {
 176                                         # Extract text inside the link
 177                                         @list( $titleText, $paramText ) = explode('|',
 178                                                 substr($s, $startBracketOffset+2, $endBracketOffset-$startBracketOffset-4), 2);
 179                                         # Create markers only for valid links
 180                                         if( preg_match( $titleRegex, $titleText ) ) {
 181                                                 # Store the text for the marker
 182                                                 $marker = $markers->addMarker($titleText, $paramText);
 183                                                 # Replace the current link with the marker
 184                                                 $s = substr($s,0,$startBracketOffset).
 185                                                         $marker.
 186                                                         substr($s, $endBracketOffset);
 187                                                 # We have modified $s, because of this we need to set the
 188                                                 # offset manually since the end position is different now
 189                                                 $offset = $startBracketOffset+strlen($marker);
 190                                                 continue;
 191                                         }
 192                                         # ToDo: Some LinkHooks may allow recursive links inside of
 193                                         # the link text, create a regex that also matches our
 194                                         # <!-- LINKMARKER ### --> sequence in titles
 195                                         # ToDO: Some LinkHooks use patterns rather than namespaces
 196                                         # these need to be tested at this point here
 197                                 }
 198
 199                         }
 200                         # Bump our offset to after our current bracket
 201                         $offset = $bracketOffset+2;
 202                 }
 203
 204
 205                 # Now expand our tree
 206                 wfProfileIn( __METHOD__.'-expand' );
 207                 $s = $markers->expand( $s );
 208                 wfProfileOut( __METHOD__.'-expand' );
 209
 210                 wfProfileOut( __METHOD__ );
 211                 return $holders;
 212         }
 213
 214         function replaceInternalLinksCallback( $parser, $holders, $markers, $titleText, $paramText ) {
 215                 wfProfileIn( __METHOD__ );
 216                 $wt = isset($paramText) ? "[[$titleText|$paramText]]" : "[[$titleText]]";
 217                 wfProfileIn( __METHOD__."-misc" );
 218                 # Don't allow internal links to pages containing
 219                 # PROTO: where PROTO is a valid URL protocol; these
 220                 # should be external links.
 221                 if( preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $titleText) ) {
 222                         wfProfileOut( __METHOD__ );
 223                         return $wt;
 224                 }
 225
 226                 # Make subpage if necessary
 227                 if( $this->areSubpagesAllowed() ) {
 228                         $titleText = $this->maybeDoSubpageLink( $titleText, $paramText );
 229                 }
 230
 231                 # Check for a leading colon and strip it if it is there
 232                 $leadingColon = $titleText[0] == ':';
 233                 if( $leadingColon ) $titleText = substr( $titleText, 1 );
 234
 235                 wfProfileOut( __METHOD__."-misc" );
 236                 # Make title object
 237                 wfProfileIn( __METHOD__."-title" );
 238                 $title = Title::newFromText( $this->mStripState->unstripNoWiki($titleText) );
 239                 if( !$title ) {
 240                         wfProfileOut( __METHOD__."-title" );
 241                         wfProfileOut( __METHOD__ );
 242                         return $wt;
 243                 }
 244                 $ns = $title->getNamespace();
 245                 wfProfileOut( __METHOD__."-title" );
 246
 247                 # Default for Namespaces is a default link
 248                 # ToDo: Default for patterns is plain wikitext
 249                 $return = true;
 250                 if( isset($this->mLinkHooks[$ns]) ) {
 251                         list( $callback, $flags ) = $this->mLinkHooks[$ns];
 252                         if( $flags & SLH_PATTERN ) {
 253                                 $args = array( $parser, $holders, $markers, $titleText, &$paramText, &$leadingColon );
 254                         } else {
 255                                 $args = array( $parser, $holders, $markers, $title, $titleText, &$paramText, &$leadingColon );
 256                         }
 257                         # Workaround for PHP bug 35229 and similar
 258                         if ( !is_callable( $callback ) ) {
 259                                 throw new MWException( "Tag hook for $name is not callable\n" );
 260                         }
 261                         $return = call_user_func_array( $callback, $args );
 262                 }
 263                 if( $return === true ) {
 264                         # True (treat as plain link) was returned, call the defaultLinkHook
 265                         $args = array( $parser, $holders, $markers, $title, $titleText, &$paramText, &$leadingColon );
 266                         $return = call_user_func_array( array( 'CoreLinkFunctions', 'defaultLinkHook' ), $args );
 267                 }
 268                 if( $return === false ) {
 269                         # False (no link) was returned, output plain wikitext
 270                         # Build it again as the hook is allowed to modify $paramText
 271                         return isset($paramText) ? "[[$titleText|$paramText]]" : "[[$titleText]]";
 272                 }
 273                 # Content was returned, return it
 274                 return $return;
 275         }
 276
 277 }
 278
 279 class LinkMarkerReplacer {
 280
 281         protected $markers, $nextId, $parser, $holders, $callback;
 282
 283         function __construct( $parser, $holders, $callback ) {
 284                 $this->nextId   = 0;
 285                 $this->markers  = array();
 286                 $this->parser   = $parser;
 287                 $this->holders  = $holders;
 288                 $this->callback = $callback;
 289         }
 290
 291         function addMarker($titleText, $paramText) {
 292                 $id = $this->nextId++;
 293                 $this->markers[$id] = array( $titleText, $paramText );
 294                 return "<!-- LINKMARKER $id -->";
 295         }
 296
 297         function findMarker( $string ) {
 298                 return (bool) preg_match('/<!-- LINKMARKER [0-9]+ -->/', $string );
 299         }
 300
 301         function expand( $string ) {
 302                 return StringUtils::delimiterReplaceCallback( "<!-- LINKMARKER ", " -->", array( &$this, 'callback' ), $string );
 303         }
 304
 305         function callback( $m ) {
 306                 $id = intval($m[1]);
 307                 if( !array_key_exists($id, $this->markers) ) return $m[0];
 308                 $args = $this->markers[$id];
 309                 array_unshift( $args, $this );
 310                 array_unshift( $args, $this->holders );
 311                 array_unshift( $args, $this->parser );
 312                 return call_user_func_array( $this->callback, $args );
 313         }
 314
 315 }