includes/site/MediaWikiSite.php

   1 <?php
   2
   3 /**
   4  * Class representing a MediaWiki site.
   5  *
   6  * @since 1.21
   7  *
   8  * @file
   9  * @ingroup Site
  10  *
  11  * @license GNU GPL v2+
  12  * @author John Erling Blad < jeblad@gmail.com >
  13  * @author Daniel Kinzler
  14  * @author Jeroen De Dauw < jeroendedauw@gmail.com >
  15  */
  16 class MediaWikiSite extends SiteObject {
  17
  18         const PATH_FILE = 'file_path';
  19         const PATH_PAGE = 'page_path';
  20
  21         /**
  22          * @since 1.21
  23          *
  24          * @param integer $globalId
  25          *
  26          * @return MediaWikiSite
  27          */
  28         public static function newFromGlobalId( $globalId ) {
  29                 return SitesTable::singleton()->newRow( array(
  30                         'type' => Site::TYPE_MEDIAWIKI,
  31                         'global_key' => $globalId,
  32                 ), true );
  33         }
  34
  35         /**
  36          * Returns the database form of the given title.
  37          *
  38          * @since 1.21
  39          *
  40          * @param String $title the target page's title, in normalized form.
  41          *
  42          * @return String
  43          */
  44         public function toDBKey( $title ) {
  45                 return str_replace( ' ', '_', $title );
  46         }
  47
  48         /**
  49          * Returns the normalized form of the given page title, using the normalization rules of the given site.
  50          * If the given title is a redirect, the redirect weill be resolved and the redirect target is returned.
  51          *
  52          * @note  : This actually makes an API request to the remote site, so beware that this function is slow and depends
  53          *          on an external service.
  54          *
  55          * @note  : If MW_PHPUNIT_TEST is defined, the call to the external site is skipped, and the title
  56          *          is normalized using the local normalization rules as implemented by the Title class.
  57          *
  58          * @see Site::normalizePageName
  59          *
  60          * @since 1.21
  61          *
  62          * @param string $pageName
  63          *
  64          * @return string
  65          * @throws MWException
  66          */
  67         public function normalizePageName( $pageName ) {
  68
  69                 // Check if we have strings as arguments.
  70                 if ( !is_string( $pageName ) ) {
  71                         throw new MWException( '$pageName must be a string' );
  72                 }
  73
  74                 // Go on call the external site
  75                 if ( defined( 'MW_PHPUNIT_TEST' ) ) {
  76                         // If the code is under test, don't call out to other sites, just normalize locally.
  77                         // Note: this may cause results to be inconsistent with the actual normalization used by the respective remote site!
  78
  79                         $t = Title::newFromText( $pageName );
  80                         return $t->getPrefixedText();
  81                 } else {
  82
  83                         // Make sure the string is normalized into NFC (due to the bug 40017)
  84                         // but do nothing to the whitespaces, that should work appropriately.
  85                         // @see https://bugzilla.wikimedia.org/show_bug.cgi?id=40017
  86                         $pageName = UtfNormal::cleanUp( $pageName );
  87
  88                         // Build the args for the specific call
  89                         $args = array(
  90                                 'action' => 'query',
  91                                 'prop' => 'info',
  92                                 'redirects' => true,
  93                                 'converttitles' => true,
  94                                 'format' => 'json',
  95                                 'titles' => $pageName,
  96                                 //@todo: options for maxlag and maxage
  97                                 // Note that maxlag will lead to a long delay before a reply is made,
  98                                 // but that maxage can avoid the extreme delay. On the other hand
  99                                 // maxage could be nice to use anyhow as it stops unnecessary requests.
 100                                 // Also consider smaxage if maxage is used.
 101                         );
 102
 103                         $url = $this->getFileUrl( 'api.php' ) . '?' . wfArrayToCgi( $args );
 104
 105                         // Go on call the external site
 106                         //@todo: we need a good way to specify a timeout here.
 107                         $ret = Http::get( $url );
 108                 }
 109
 110                 if ( $ret === false ) {
 111                         wfDebugLog( "MediaWikiSite", "call to external site failed: $url" );
 112                         return false;
 113                 }
 114
 115                 $data = FormatJson::decode( $ret, true );
 116
 117                 if ( !is_array( $data ) ) {
 118                         wfDebugLog( "MediaWikiSite", "call to <$url> returned bad json: " . $ret );
 119                         return false;
 120                 }
 121
 122                 $page = static::extractPageRecord( $data, $pageName );
 123
 124                 if ( isset( $page['missing'] ) ) {
 125                         wfDebugLog( "MediaWikiSite", "call to <$url> returned a marker for a missing page title! " . $ret );
 126                         return false;
 127                 }
 128
 129                 if ( isset( $page['invalid'] ) ) {
 130                         wfDebugLog( "MediaWikiSite", "call to <$url> returned a marker for an invalid page title! " . $ret );
 131                         return false;
 132                 }
 133
 134                 if ( !isset( $page['title'] ) ) {
 135                         wfDebugLog( "MediaWikiSite", "call to <$url> did not return a page title! " . $ret );
 136                         return false;
 137                 }
 138
 139                 return $page['title'];
 140         }
 141
 142
 143         /**
 144          * Get normalization record for a given page title from an API response.
 145          *
 146          * @since 1.21
 147          *
 148          * @param array $externalData A reply from the API on a external server.
 149          * @param string $pageTitle Identifies the page at the external site, needing normalization.
 150          *
 151          * @return array|false a 'page' structure representing the page identified by $pageTitle.
 152          */
 153         private static function extractPageRecord( $externalData, $pageTitle ) {
 154                 // If there is a special case with only one returned page
 155                 // we can cheat, and only return
 156                 // the single page in the "pages" substructure.
 157                 if ( isset( $externalData['query']['pages'] ) ) {
 158                         $pages = array_values( $externalData['query']['pages'] );
 159                         if ( count( $pages) === 1 ) {
 160                                 return $pages[0];
 161                         }
 162                 }
 163                 // This is only used during internal testing, as it is assumed
 164                 // a more optimal (and lossfree) storage.
 165                 // Make initial checks and return if prerequisites are not meet.
 166                 if ( !is_array( $externalData ) || !isset( $externalData['query'] ) ) {
 167                         return false;
 168                 }
 169                 // Loop over the tree different named structures, that otherwise are similar
 170                 $structs = array(
 171                         'normalized' => 'from',
 172                         'converted' => 'from',
 173                         'redirects' => 'from',
 174                         'pages' => 'title'
 175                 );
 176                 foreach ( $structs as $listId => $fieldId ) {
 177                         // Check if the substructure exist at all.
 178                         if ( !isset( $externalData['query'][$listId] ) ) {
 179                                 continue;
 180                         }
 181                         // Filter the substructure down to what we actually are using.
 182                         $collectedHits = array_filter(
 183                                 array_values( $externalData['query'][$listId] ),
 184                                 function( $a ) use ( $fieldId, $pageTitle ) {
 185                                         return $a[$fieldId] === $pageTitle;
 186                                 }
 187                         );
 188                         // If still looping over normalization, conversion or redirects,
 189                         // then we need to keep the new page title for later rounds.
 190                         if ( $fieldId === 'from' && is_array( $collectedHits ) ) {
 191                                 switch ( count( $collectedHits ) ) {
 192                                         case 0:
 193                                                 break;
 194                                         case 1:
 195                                                 $pageTitle = $collectedHits[0]['to'];
 196                                                 break;
 197                                         default:
 198                                                 return false;
 199                                 }
 200                         }
 201                         // If on the pages structure we should prepare for returning.
 202                         elseif ( $fieldId === 'title' && is_array( $collectedHits ) ) {
 203                                 switch ( count( $collectedHits ) ) {
 204                                         case 0:
 205                                                 return false;
 206                                         case 1:
 207                                                 return array_shift( $collectedHits );
 208                                         default:
 209                                                 return false;
 210                                 }
 211                         }
 212                 }
 213                 // should never be here
 214                 return false;
 215         }
 216
 217         /**
 218          * @see Site::getLinkPathType
 219          * Returns Site::PATH_PAGE
 220          *
 221          * @since 1.21
 222          *
 223          * @return string
 224          */
 225         public function getLinkPathType() {
 226                 return self::PATH_PAGE;
 227         }
 228
 229         /**
 230          * Returns the relative page path.
 231          *
 232          * @since 1.21
 233          *
 234          * @return string
 235          */
 236         public function getRelativePagePath() {
 237                 return parse_url( $this->getPath( self::PATH_PAGE ), PHP_URL_PATH );
 238         }
 239
 240         /**
 241          * Returns the relative file path.
 242          *
 243          * @since 1.21
 244          *
 245          * @return string
 246          */
 247         public function getRelativeFilePath() {
 248                 return parse_url( $this->getPath( self::PATH_FILE ), PHP_URL_PATH );
 249         }
 250
 251         /**
 252          * Sets the relative page path.
 253          *
 254          * @since 1.21
 255          *
 256          * @param string $path
 257          */
 258         public function setPagePath( $path ) {
 259                 $this->setPath( self::PATH_PAGE, $path );
 260         }
 261
 262         /**
 263          * Sets the relative file path.
 264          *
 265          * @since 1.21
 266          *
 267          * @param string $path
 268          */
 269         public function setFilePath( $path ) {
 270                 $this->setPath( self::PATH_FILE, $path );
 271         }
 272
 273         /**
 274          * @see Site::getPagePath
 275          *
 276          * This implementation returns a URL constructed using the path returned by getLinkPath().
 277          * In addition to the default behaviour implemented by SiteObject::getPageUrl(), this
 278          * method converts the $pageName to DBKey-format by replacing spaces with underscores
 279          * before using it in the URL.
 280          *
 281          * @since 1.21
 282          *
 283          * @param $pagename string: Page name (default: false)
 284          *
 285          * @return string
 286          */
 287         public function getPageUrl( $pageName = false ) {
 288                 $url = $this->getLinkPath();
 289
 290                 if ( $url === false ) {
 291                         return false;
 292                 }
 293
 294                 if ( $pageName !== false ) {
 295                         $pageName = $this->toDBKey( trim( $pageName ) );
 296                         $url = str_replace( '$1', wfUrlencode( $pageName ), $url ) ;
 297                 }
 298
 299                 return $url;
 300         }
 301
 302         /**
 303          * Returns the full file path (ie site url + relative file path).
 304          * The path should go at the $1 marker. If the $path
 305          * argument is provided, the marker will be replaced by it's value.
 306          *
 307          * @since 1.21
 308          *
 309          * @param string|false $path
 310          *
 311          * @return string
 312          */
 313         public function getFileUrl( $path = false ) {
 314                 $filePath = $this->getPath( self::PATH_FILE );
 315
 316                 if ( $filePath !== false ) {
 317                         $filePath = str_replace( '$1', $path, $filePath );
 318                 }
 319
 320                 return $filePath;
 321         }
 322
 323 }