| Index: trunk/extensions/SemanticMediaWiki/maintenance/SMW_dumpRDF.php |
| — | — | @@ -17,7 +17,7 @@ |
| 18 | 18 | * --individuals do only pages that are no categories, properties, or types |
| 19 | 19 | * -d <delay> slows down the export in order to stress the server less, |
| 20 | 20 | * sleeping for <delay> milliseconds every now and then |
| 21 | | - * -e <each> after how many exported entities should the server take a nap? |
| | 21 | + * -e <each> after how many exported entities should the process take a nap? |
| 22 | 22 | * --server=<server> The protocol and server name to as base URLs, e.g. |
| 23 | 23 | * http://en.wikipedia.org. This is sometimes necessary because |
| 24 | 24 | * server name detection may fail in command line scripts. |
| — | — | @@ -33,7 +33,7 @@ |
| 34 | 34 | ? getenv( 'MW_INSTALL_PATH' ) . "/maintenance/commandLine.inc" |
| 35 | 35 | : dirname( __FILE__ ) . '/../../../maintenance/commandLine.inc' ); |
| 36 | 36 | global $smwgIP, $wgServer; |
| 37 | | -require_once( "$smwgIP/specials/Export/SMW_SpecialOWLExport.php" ); |
| | 37 | +//require_once( "$smwgIP/specials/Export/SMW_SpecialOWLExport.php" ); |
| 38 | 38 | |
| 39 | 39 | if ( !empty( $options['o'] ) ) { |
| 40 | 40 | $outfile = $options['o']; |
| — | — | @@ -76,5 +76,5 @@ |
| 77 | 77 | print "\nWriting OWL/RDF dump to file \"$outfile\" ...\n"; |
| 78 | 78 | } |
| 79 | 79 | |
| 80 | | -$exRDF = new SMWOWLExport(); |
| | 80 | +$exRDF = new SMWExportController( new SMWSerializer() ); |
| 81 | 81 | $exRDF->printAll( $outfile, $export_ns, $delay, $delayeach ); |
| Index: trunk/extensions/SemanticMediaWiki/specials/Export/SMW_SpecialOWLExport.php |
| — | — | @@ -27,6 +27,9 @@ |
| 28 | 28 | * @author Markus Krötzsch |
| 29 | 29 | */ |
| 30 | 30 | class SMWSpecialOWLExport extends SpecialPage { |
| | 31 | + |
| | 32 | + /// Export controller object to be used for serializing data |
| | 33 | + protected $export_controller; |
| 31 | 34 | |
| 32 | 35 | public function __construct() { |
| 33 | 36 | parent::__construct( 'ExportRDF' ); |
| — | — | @@ -34,26 +37,13 @@ |
| 35 | 38 | } |
| 36 | 39 | |
| 37 | 40 | public function execute( $page ) { |
| 38 | | - global $wgOut, $wgRequest, $wgUser, $smwgAllowRecursiveExport, $smwgExportBacklinks, $smwgExportAll; |
| 39 | | - |
| | 41 | + global $wgOut, $wgRequest; |
| 40 | 42 | $wgOut->setPageTitle( wfMsg( 'exportrdf' ) ); |
| 41 | | - |
| 42 | | - $recursive = 0; // default, no recursion |
| 43 | | - $backlinks = $smwgExportBacklinks; // default |
| 44 | | - |
| 45 | | - // Check whether we already know what to export. |
| 46 | | - if ( $page == '' ) { |
| 47 | | - // Try to get GET parameter; simple way of calling the export. |
| 48 | | - $page = $wgRequest->getVal( 'page' ); |
| 49 | | - } else { |
| 50 | | - // This is needed since MediaWiki 1.8, but it is wrong for 1.7. |
| 51 | | - $page = rawurldecode( $page ); |
| 52 | | - } |
| 53 | | - |
| 54 | | - if ( $page == '' ) { |
| 55 | | - // Try to get POST list; some settings are only available via POST. |
| | 43 | + |
| | 44 | + // see if we can find something to export: |
| | 45 | + $page = ( $page == '' ) ? $wgRequest->getVal( 'page' ) : $page = rawurldecode( $page ); |
| | 46 | + if ( $page == '' ) { // Try to get POST list; some settings are only available via POST. |
| 56 | 47 | $pageblob = $wgRequest->getText( 'pages' ); |
| 57 | | - |
| 58 | 48 | if ( $pageblob != '' ) { |
| 59 | 49 | $pages = explode( "\n", $pageblob ); |
| 60 | 50 | } |
| — | — | @@ -61,100 +51,108 @@ |
| 62 | 52 | $pages = array( $page ); |
| 63 | 53 | } |
| 64 | 54 | |
| 65 | | - if ( isset( $pages ) ) { // export to RDF |
| 66 | | - $wgOut->disable(); |
| 67 | | - ob_start(); |
| 68 | | - |
| 69 | | - // Only use rdf+xml mimetype if explicitly requested |
| 70 | | - // TODO: should the see also links in the exported RDF then have this parameter as well? |
| 71 | | - if ( $wgRequest->getVal( 'xmlmime' ) == 'rdf' ) { |
| 72 | | - header( "Content-type: application/rdf+xml; charset=UTF-8" ); |
| 73 | | - } else { |
| 74 | | - header( "Content-type: application/xml; charset=UTF-8" ); |
| 75 | | - } |
| 76 | | - |
| 77 | | - // Effect: assume "no" from missing parameters generated by checkboxes. |
| 78 | | - $postform = $wgRequest->getText( 'postform' ) == 1; |
| 79 | | - |
| 80 | | - $rec = $wgRequest->getText( 'recursive' ); |
| 81 | | - |
| 82 | | - if ( $rec == '' ) $rec = $wgRequest->getVal( 'recursive' ); |
| 83 | | - |
| 84 | | - if ( ( $rec == '1' ) && ( $smwgAllowRecursiveExport || $wgUser->isAllowed( 'delete' ) ) ) { |
| 85 | | - $recursive = 1; // users may be allowed to switch it on |
| 86 | | - } |
| 87 | | - |
| 88 | | - $bl = $wgRequest->getText( 'backlinks' ); |
| 89 | | - |
| 90 | | - if ( $bl == '' ) $bl = $wgRequest->getVal( 'backlinks' ); |
| 91 | | - |
| 92 | | - if ( ( $bl == '1' ) && ( $wgUser->isAllowed( 'delete' ) ) ) { |
| 93 | | - $backlinks = true; // admins can always switch on backlinks |
| 94 | | - } elseif ( ( $bl == '0' ) || ( '' == $bl && $postform ) ) { |
| 95 | | - $backlinks = false; // everybody can explicitly switch off backlinks |
| 96 | | - } |
| 97 | | - |
| 98 | | - $date = $wgRequest->getText( 'date' ); |
| 99 | | - |
| 100 | | - if ( $date == '' ) $date = $wgRequest->getVal( 'date' ); |
| 101 | | - |
| 102 | | - $exp = new SMWOWLExport(); |
| 103 | | - |
| 104 | | - if ( $date != '' ) $exp->setDate( $date ); |
| 105 | | - |
| 106 | | - $exp->printPages( $pages, $recursive, $backlinks ); |
| 107 | | - |
| | 55 | + if ( isset( $pages ) ) { |
| | 56 | + $this->exportPages( $pages ); |
| 108 | 57 | return; |
| 109 | 58 | } else { |
| 110 | 59 | $offset = $wgRequest->getVal( 'offset' ); |
| 111 | | - |
| 112 | 60 | if ( isset( $offset ) ) { |
| 113 | | - $wgOut->disable(); |
| 114 | | - |
| 115 | | - ob_start(); |
| 116 | | - header( "Content-type: application/xml; charset=UTF-8" ); |
| 117 | | - |
| 118 | | - $exp = new SMWOWLExport(); |
| 119 | | - $exp->printPageList( $offset ); |
| 120 | | - |
| | 61 | + $this->startRDFExport(); |
| | 62 | + $this->export_controller->printPageList( $offset ); |
| 121 | 63 | return; |
| 122 | 64 | } else { |
| 123 | 65 | $stats = $wgRequest->getVal( 'stats' ); |
| 124 | | - |
| 125 | 66 | if ( isset( $stats ) ) { |
| 126 | | - $wgOut->disable(); |
| 127 | | - |
| 128 | | - ob_start(); |
| 129 | | - header( "Content-type: application/xml; charset=UTF-8" ); |
| 130 | | - |
| 131 | | - $exp = new SMWOWLExport(); |
| 132 | | - $exp->printWikiInfo(); |
| | 67 | + $this->startRDFExport(); |
| | 68 | + $this->export_controller->printWikiInfo(); |
| | 69 | + return; |
| 133 | 70 | } |
| 134 | 71 | } |
| 135 | 72 | } |
| 136 | | - |
| 137 | 73 | // nothing exported yet; show user interface: |
| | 74 | + $this->showForm(); |
| | 75 | + } |
| | 76 | + |
| | 77 | + /** |
| | 78 | + * Create the HTML user interface for this special page. |
| | 79 | + */ |
| | 80 | + protected function showForm() { |
| | 81 | + global $wgOut, $wgUser, $smwgAllowRecursiveExport, $smwgExportBacklinks, $smwgExportAll; |
| | 82 | + |
| 138 | 83 | $html = '<form name="tripleSearch" action="" method="POST">' . "\n" . |
| 139 | | - '<p>' . wfMsg( 'smw_exportrdf_docu' ) . "</p>\n" . |
| 140 | | - '<input type="hidden" name="postform" value="1"/>' . "\n" . |
| 141 | | - '<textarea name="pages" cols="40" rows="10"></textarea><br />' . "\n"; |
| | 84 | + '<p>' . wfMsg( 'smw_exportrdf_docu' ) . "</p>\n" . |
| | 85 | + '<input type="hidden" name="postform" value="1"/>' . "\n" . |
| | 86 | + '<textarea name="pages" cols="40" rows="10"></textarea><br />' . "\n"; |
| 142 | 87 | |
| 143 | 88 | if ( $wgUser->isAllowed( 'delete' ) || $smwgAllowRecursiveExport ) { |
| 144 | 89 | $html .= '<input type="checkbox" name="recursive" value="1" id="rec"> <label for="rec">' . wfMsg( 'smw_exportrdf_recursive' ) . '</label></input><br />' . "\n"; |
| 145 | 90 | } |
| 146 | | - |
| 147 | 91 | if ( $wgUser->isAllowed( 'delete' ) || $smwgExportBacklinks ) { |
| 148 | 92 | $html .= '<input type="checkbox" name="backlinks" value="1" default="true" id="bl"> <label for="bl">' . wfMsg( 'smw_exportrdf_backlinks' ) . '</label></input><br />' . "\n"; |
| 149 | 93 | } |
| 150 | | - |
| 151 | 94 | if ( $wgUser->isAllowed( 'delete' ) || $smwgExportAll ) { |
| 152 | 95 | $html .= '<br />'; |
| 153 | 96 | $html .= '<input type="text" name="date" value="' . date( DATE_W3C, mktime( 0, 0, 0, 1, 1, 2000 ) ) . '" id="date"> <label for="ea">' . wfMsg( 'smw_exportrdf_lastdate' ) . '</label></input><br />' . "\n"; |
| 154 | 97 | } |
| | 98 | + $html .= '<br /><input type="submit" value="' . wfMsg( 'smw_exportrdf_submit' ) . "\"/>\n</form>"; |
| 155 | 99 | |
| 156 | | - $html .= "<br /><input type=\"submit\" value=\"" . wfMsg( 'smw_exportrdf_submit' ) . "\"/>\n</form>"; |
| 157 | | - |
| 158 | 100 | $wgOut->addHTML( $html ); |
| 159 | 101 | } |
| | 102 | + |
| | 103 | + /** |
| | 104 | + * Prepare $wgOut for printing non-HTML data. |
| | 105 | + */ |
| | 106 | + protected function startRDFExport() { |
| | 107 | + global $wgOut, $wgRequest; |
| | 108 | + $wgOut->disable(); |
| | 109 | + ob_start(); |
| | 110 | + // Only use rdf+xml mimetype if explicitly requested |
| | 111 | + // TODO: should the see also links in the exported RDF then have this parameter as well? |
| | 112 | + if ( $wgRequest->getVal( 'xmlmime' ) == 'rdf' ) { |
| | 113 | + header( "Content-type: application/rdf+xml; charset=UTF-8" ); |
| | 114 | + } else { |
| | 115 | + header( "Content-type: application/xml; charset=UTF-8" ); |
| | 116 | + } |
| | 117 | + $this->export_controller = new SMWExportController( new SMWSerializer() ); |
| | 118 | + } |
| | 119 | + |
| | 120 | + /** |
| | 121 | + * Export the given pages to RDF. |
| | 122 | + * @param array $pages containing the string names of pages to be exported |
| | 123 | + */ |
| | 124 | + protected function exportPages( $pages ) { |
| | 125 | + global $wgRequest, $smwgExportBacklinks; |
| 160 | 126 | |
| | 127 | + // Effect: assume "no" from missing parameters generated by checkboxes. |
| | 128 | + $postform = $wgRequest->getText( 'postform' ) == 1; |
| | 129 | + |
| | 130 | + $recursive = 0; // default, no recursion |
| | 131 | + $rec = $wgRequest->getText( 'recursive' ); |
| | 132 | + if ( $rec == '' ) $rec = $wgRequest->getVal( 'recursive' ); |
| | 133 | + if ( ( $rec == '1' ) && ( $smwgAllowRecursiveExport || $wgUser->isAllowed( 'delete' ) ) ) { |
| | 134 | + $recursive = 1; // users may be allowed to switch it on |
| | 135 | + } |
| | 136 | + |
| | 137 | + $backlinks = $smwgExportBacklinks; // default |
| | 138 | + $bl = $wgRequest->getText( 'backlinks' ); |
| | 139 | + if ( $bl == '' ) $bl = $wgRequest->getVal( 'backlinks' ); |
| | 140 | + if ( ( $bl == '1' ) && ( $wgUser->isAllowed( 'delete' ) ) ) { |
| | 141 | + $backlinks = true; // admins can always switch on backlinks |
| | 142 | + } elseif ( ( $bl == '0' ) || ( '' == $bl && $postform ) ) { |
| | 143 | + $backlinks = false; // everybody can explicitly switch off backlinks |
| | 144 | + } |
| | 145 | + |
| | 146 | + $date = $wgRequest->getText( 'date' ); |
| | 147 | + if ( $date == '' ) $date = $wgRequest->getVal( 'date' ); |
| | 148 | + if ( $date != '' ) { |
| | 149 | + $timeint = strtotime( $date ); |
| | 150 | + $stamp = date( "YmdHis", $timeint ); |
| | 151 | + $date = $stamp; |
| | 152 | + } |
| | 153 | + |
| | 154 | + $this->startRDFExport(); |
| | 155 | + $this->export_controller->enableBacklinks( $backlinks ); |
| | 156 | + $this->export_controller->printPages( $pages, $recursive, $date ); |
| | 157 | + } |
| | 158 | + |
| 161 | 159 | } |
| Index: trunk/extensions/SemanticMediaWiki/includes/export/SMW_OWLExport.php |
| — | — | @@ -1,836 +0,0 @@ |
| 2 | | -<?php |
| 3 | | - |
| 4 | | -/** |
| 5 | | - * File holding the SMWOWLExport class for OWL export, used by SMWSpecialOWLExport (Special:ExportRDF page). |
| 6 | | - * |
| 7 | | - * @file SMW_OWLExport.php |
| 8 | | - * @ingroup SMW |
| 9 | | - * |
| 10 | | - * @author Markus Krötzsch |
| 11 | | - */ |
| 12 | | - |
| 13 | | -/** |
| 14 | | - * Small data object holding the bare essentials of one title. |
| 15 | | - * Used to store processed and open pages for export. |
| 16 | | - * |
| 17 | | - * @ingroup SMW |
| 18 | | - */ |
| 19 | | -class SMWSmallTitle { |
| 20 | | - public $dbkey; |
| 21 | | - public $namespace; // MW namespace constant |
| 22 | | - public $modifier = ''; // e.g. a unit string |
| 23 | | - |
| 24 | | - public function getHash() { |
| 25 | | - return $this->dbkey . ' ' . $this->namespace . ' ' . $this->modifier; |
| 26 | | - } |
| 27 | | -} |
| 28 | | - |
| 29 | | - |
| 30 | | -/** |
| 31 | | - * Class for encapsulating the methods for RDF export. |
| 32 | | - * |
| 33 | | - * @ingroup SMW |
| 34 | | - */ |
| 35 | | -class SMWOWLExport { |
| 36 | | - /**#@+ |
| 37 | | - * @access private |
| 38 | | - */ |
| 39 | | - |
| 40 | | - const MAX_CACHE_SIZE = 5000; // do not let cache arrays get larger than this |
| 41 | | - const CACHE_BACKJUMP = 500; // kill this many cached entries if limit is reached, |
| 42 | | - // avoids too much array copying; <= MAX_CACHE_SIZE! |
| 43 | | - |
| 44 | | - /** |
| 45 | | - * An array that keeps track of the elements for which we still need to |
| 46 | | - * write auxilliary definitions. |
| 47 | | - */ |
| 48 | | - private $element_queue; |
| 49 | | - |
| 50 | | - /** |
| 51 | | - * An array that keeps track of the elements which have been exported already |
| 52 | | - */ |
| 53 | | - private $element_done; |
| 54 | | - |
| 55 | | - /** |
| 56 | | - * Date used to filter the export. If a page has not been changed since that |
| 57 | | - * date it will not be exported |
| 58 | | - */ |
| 59 | | - private $date; |
| 60 | | - |
| 61 | | - /** |
| 62 | | - * Array of additional namespaces (abbreviation => URI), flushed on |
| 63 | | - * closing the current namespace tag. Since we export RDF in a streamed |
| 64 | | - * way, it is not always possible to embed additional namespaces into |
| 65 | | - * the RDF-tag which might have been sent to the client already. But we |
| 66 | | - * wait with printing the current Description so that extra namespaces |
| 67 | | - * from this array can still be printed (note that you never know which |
| 68 | | - * extra namespaces you encounter during export). |
| 69 | | - */ |
| 70 | | - private $extra_namespaces; |
| 71 | | - |
| 72 | | - /** |
| 73 | | - * Array of namespaces that have been declared globally already. Contains |
| 74 | | - * entries of format 'namespace abbreviation' => true, assuming that the |
| 75 | | - * same abbreviation always refers to the same URI (i.e. you cannot import |
| 76 | | - * something as rdf:bla if you do not want rdf to be the standard |
| 77 | | - * namespace that is already given in every RDF export). |
| 78 | | - */ |
| 79 | | - private $global_namespaces; |
| 80 | | - |
| 81 | | - /** |
| 82 | | - * Unprinted XML is composed from the strings $pre_ns_buffer and $post_ns_buffer. |
| 83 | | - * The split between the two is such that one can append additional namespace |
| 84 | | - * declarations to $pre_ns_buffer so that they affect all current elements. The |
| 85 | | - * buffers are flushed during output in order to achieve "streaming" RDF export |
| 86 | | - * for larger files. |
| 87 | | - */ |
| 88 | | - private $pre_ns_buffer; |
| 89 | | - |
| 90 | | - /** |
| 91 | | - * See documentation for SMWOWLExport::pre_ns_buffer. |
| 92 | | - */ |
| 93 | | - private $post_ns_buffer; |
| 94 | | - |
| 95 | | - /** |
| 96 | | - * Boolean that is true as long as nothing was flushed yet. Indicates that |
| 97 | | - * extra namespaces can still become global. |
| 98 | | - */ |
| 99 | | - private $first_flush; |
| 100 | | - |
| 101 | | - /** |
| 102 | | - * Integer that counts down the number of objects we still process before |
| 103 | | - * doing the first flush. Aggregating some output before flushing is useful |
| 104 | | - * to get more namespaces global. Flushing will only happen if $delay_flush |
| 105 | | - * is 0. |
| 106 | | - */ |
| 107 | | - private $delay_flush; |
| 108 | | - |
| 109 | | - /** |
| 110 | | - * Constructor. |
| 111 | | - */ |
| 112 | | - public function __construct() { |
| 113 | | - $this->element_queue = array(); |
| 114 | | - $this->element_done = array(); |
| 115 | | - $this->date = ''; |
| 116 | | - } |
| 117 | | - |
| 118 | | - /** |
| 119 | | - * Sets a date as a filter. Any page that has not been changed since that date |
| 120 | | - * will not be exported. The date has to be a string in XML Schema format. |
| 121 | | - */ |
| 122 | | - public function setDate( $date ) { |
| 123 | | - $timeint = strtotime( $date ); |
| 124 | | - $stamp = date( "YmdHis", $timeint ); |
| 125 | | - $this->date = $stamp; |
| 126 | | - } |
| 127 | | - |
| 128 | | - /** |
| 129 | | - * This function prints all selected pages. The parameter $recursion determines |
| 130 | | - * how referenced ressources are treated: |
| 131 | | - * '0' : add brief declarations for each |
| 132 | | - * '1' : add full descriptions for each, thus beginning real recursion (and |
| 133 | | - * probably retrieving the whole wiki ...) |
| 134 | | - * else: ignore them, though -1 might become a synonym for "export *all*" in the future |
| 135 | | - * The parameter $backlinks determines whether or not subjects of incoming |
| 136 | | - * properties are exported as well. Enables "browsable RDF." |
| 137 | | - */ |
| 138 | | - public function printPages( $pages, $recursion = 1, $backlinks = true ) { |
| 139 | | - wfProfileIn( "RDF::PrintPages" ); |
| 140 | | - |
| 141 | | - $linkCache =& LinkCache::singleton(); |
| 142 | | - $this->pre_ns_buffer = ''; |
| 143 | | - $this->post_ns_buffer = ''; |
| 144 | | - $this->first_flush = true; |
| 145 | | - $this->delay_flush = 10; // flush only after (fully) printing 11 objects |
| 146 | | - $this->extra_namespaces = array(); |
| 147 | | - |
| 148 | | - if ( count( $pages ) == 1 ) { // ensure that ontologies that are retrieved as linked data are not confused with their subject! |
| 149 | | - $ontologyuri = SMWExporter::expandURI( '&export;' ) . '/' . urlencode( end( $pages ) ); |
| 150 | | - } else { // use empty URI, i.e. "location" as URI otherwise |
| 151 | | - $ontologyuri = ''; |
| 152 | | - } |
| 153 | | - |
| 154 | | - $this->printHeader( $ontologyuri ); // also inits global namespaces |
| 155 | | - |
| 156 | | - wfProfileIn( "RDF::PrintPages::PrepareQueue" ); |
| 157 | | - |
| 158 | | - // transform pages into queued export titles |
| 159 | | - $cur_queue = array(); |
| 160 | | - |
| 161 | | - foreach ( $pages as $page ) { |
| 162 | | - $title = Title::newFromText( $page ); |
| 163 | | - |
| 164 | | - if ( null === $title ) continue; // invalid title name given |
| 165 | | - |
| 166 | | - $st = new SMWSmallTitle(); |
| 167 | | - $st->dbkey = $title->getDBkey(); |
| 168 | | - $st->namespace = $title->getNamespace(); |
| 169 | | - |
| 170 | | - $cur_queue[] = $st; |
| 171 | | - } |
| 172 | | - |
| 173 | | - wfProfileOut( "RDF::PrintPages::PrepareQueue" ); |
| 174 | | - |
| 175 | | - while ( count( $cur_queue ) > 0 ) { |
| 176 | | - // first, print all selected pages |
| 177 | | - foreach ( $cur_queue as $st ) { |
| 178 | | - wfProfileIn( "RDF::PrintPages::PrintOne" ); |
| 179 | | - |
| 180 | | - $this->printObject( $st, true, $backlinks ); |
| 181 | | - |
| 182 | | - wfProfileOut( "RDF::PrintPages::PrintOne" ); |
| 183 | | - |
| 184 | | - if ( $this->delay_flush > 0 ) $this->delay_flush--; |
| 185 | | - } |
| 186 | | - |
| 187 | | - // prepare array for next iteration |
| 188 | | - $cur_queue = array(); |
| 189 | | - |
| 190 | | - if ( 1 == $recursion ) { |
| 191 | | - $cur_queue = $this->element_queue + $cur_queue; // make sure the array is *duplicated* instead of copying its ref |
| 192 | | - $this->element_queue = array(); |
| 193 | | - } |
| 194 | | - |
| 195 | | - $linkCache->clear(); |
| 196 | | - } |
| 197 | | - |
| 198 | | - // for pages not processed recursively, print at least basic declarations |
| 199 | | - wfProfileIn( "RDF::PrintPages::Auxiliary" ); |
| 200 | | - $this->date = ''; // no date restriction for the rest! |
| 201 | | - |
| 202 | | - if ( !empty( $this->element_queue ) ) { |
| 203 | | - if ( $this->pre_ns_buffer != '' ) { |
| 204 | | - $this->post_ns_buffer .= "\t<!-- auxiliary definitions -->\n"; |
| 205 | | - } else { |
| 206 | | - print "\t<!-- auxiliary definitions -->\n"; // just print this comment, so that later outputs still find the empty pre_ns_buffer! |
| 207 | | - } |
| 208 | | - |
| 209 | | - while ( !empty( $this->element_queue ) ) { |
| 210 | | - $st = array_pop( $this->element_queue ); |
| 211 | | - $this->printObject( $st, false, false ); |
| 212 | | - } |
| 213 | | - } |
| 214 | | - |
| 215 | | - wfProfileOut( "RDF::PrintPages::Auxiliary" ); |
| 216 | | - |
| 217 | | - $this->printFooter(); |
| 218 | | - $this->flushBuffers( true ); |
| 219 | | - |
| 220 | | - wfProfileOut( "RDF::PrintPages" ); |
| 221 | | - } |
| 222 | | - |
| 223 | | - /** |
| 224 | | - * This function prints RDF for *all* pages within the wiki, and for all |
| 225 | | - * elements that are referred to in the exported RDF. |
| 226 | | - */ |
| 227 | | - public function printAll( $outfile, $ns_restriction = false, $delay, $delayeach ) { |
| 228 | | - global $smwgNamespacesWithSemanticLinks; |
| 229 | | - |
| 230 | | - $linkCache =& LinkCache::singleton(); |
| 231 | | - |
| 232 | | - $db = & wfGetDB( DB_MASTER ); |
| 233 | | - $this->pre_ns_buffer = ''; |
| 234 | | - $this->post_ns_buffer = ''; |
| 235 | | - $this->first_flush = true; |
| 236 | | - |
| 237 | | - if ( $outfile === false ) { |
| 238 | | - // $this->delay_flush = 10000; //flush only after (fully) printing 10001 objects, |
| 239 | | - $this->delay_flush = - 1; // do not flush buffer at all |
| 240 | | - } else { |
| 241 | | - $file = fopen( $outfile, 'w' ); |
| 242 | | - |
| 243 | | - if ( !$file ) { |
| 244 | | - print "\nCannot open \"$outfile\" for writing.\n"; |
| 245 | | - return false; |
| 246 | | - } |
| 247 | | - |
| 248 | | - $this->delay_flush = - 1; // never flush, we flush in another way |
| 249 | | - } |
| 250 | | - |
| 251 | | - $this->extra_namespaces = array(); |
| 252 | | - $this->printHeader(); // also inits global namespaces |
| 253 | | - |
| 254 | | - $start = 1; |
| 255 | | - $end = $db->selectField( 'page', 'max(page_id)', false, $outfile ); |
| 256 | | - |
| 257 | | - $a_count = 0; $d_count = 0; // DEBUG |
| 258 | | - |
| 259 | | - $delaycount = $delayeach; |
| 260 | | - |
| 261 | | - for ( $id = $start; $id <= $end; $id++ ) { |
| 262 | | - $title = Title::newFromID( $id ); |
| 263 | | - |
| 264 | | - if ( ( $title === null ) || !smwfIsSemanticsProcessed( $title->getNamespace() ) ) continue; |
| 265 | | - if ( !SMWOWLExport::fitsNsRestriction( $ns_restriction, $title->getNamespace() ) ) continue; |
| 266 | | - |
| 267 | | - $st = new SMWSmallTitle(); |
| 268 | | - $st->dbkey = $title->getDBkey(); |
| 269 | | - $st->namespace = $title->getNamespace(); |
| 270 | | - |
| 271 | | - $cur_queue = array( $st ); |
| 272 | | - $a_count++; // DEBUG |
| 273 | | - $full_export = true; |
| 274 | | - |
| 275 | | - while ( count( $cur_queue ) > 0 ) { |
| 276 | | - foreach ( $cur_queue as $st ) { |
| 277 | | - wfProfileIn( "RDF::PrintAll::PrintOne" ); |
| 278 | | - $this->printObject( $st, $full_export, false ); |
| 279 | | - wfProfileOut( "RDF::PrintAll::PrintOne" ); |
| 280 | | - } |
| 281 | | - |
| 282 | | - $full_export = false; // make sure added dependencies do not pull more than needed |
| 283 | | - // resolve dependencies that will otherwise not be printed |
| 284 | | - $cur_queue = array(); |
| 285 | | - |
| 286 | | - foreach ( $this->element_queue as $key => $staux ) { |
| 287 | | - $taux = Title::makeTitle( $staux->namespace, $staux->dbkey ); |
| 288 | | - |
| 289 | | - if ( !smwfIsSemanticsProcessed( $staux->namespace ) || ( $staux->modifier !== '' ) || |
| 290 | | - !SMWOWLExport::fitsNsRestriction( $ns_restriction, $staux->namespace ) || |
| 291 | | - ( !$taux->exists() ) ) { |
| 292 | | - // Note: we do not need to check the cache to guess if an element was already |
| 293 | | - // printed. If so, it would not be included in the queue in the first place. |
| 294 | | - $cur_queue[] = $staux; |
| 295 | | - // $this->post_ns_buffer .= "<!-- Adding dependency '" . $staux->getHash() . "' -->"; //DEBUG |
| 296 | | - $d_count++; // DEBUG |
| 297 | | - } else { |
| 298 | | - unset( $this->element_queue[$key] ); // carrying around the values we do not |
| 299 | | - // want to export now is a potential memory leak |
| 300 | | - } |
| 301 | | - } |
| 302 | | - |
| 303 | | - // sleep each $delaycount for $delay ms to be nice to the server |
| 304 | | - if ( ( $delaycount-- < 0 ) && ( $delayeach != 0 ) ) { |
| 305 | | - usleep( $delay ); |
| 306 | | - $delaycount = $delayeach; |
| 307 | | - } |
| 308 | | - } |
| 309 | | - |
| 310 | | - if ( $outfile !== false ) { // flush buffer |
| 311 | | - fwrite( $file, $this->post_ns_buffer ); |
| 312 | | - $this->post_ns_buffer = ''; |
| 313 | | - } |
| 314 | | - |
| 315 | | - $linkCache->clear(); |
| 316 | | - } |
| 317 | | - |
| 318 | | - // DEBUG: |
| 319 | | - $this->post_ns_buffer .= "<!-- Processed $a_count regular articles. -->\n"; |
| 320 | | - $this->post_ns_buffer .= "<!-- Processed $d_count added dependencies. -->\n"; |
| 321 | | - $this->post_ns_buffer .= "<!-- Final cache size was " . sizeof( $this->element_done ) . ". -->\n"; |
| 322 | | - |
| 323 | | - $this->printFooter(); |
| 324 | | - |
| 325 | | - if ( $outfile === false ) { |
| 326 | | - $this->flushBuffers( true ); |
| 327 | | - } else { // prepend headers to file, there is no really efficient solution (`cat(1)`) for this it seems |
| 328 | | - // print head: |
| 329 | | - fclose( $file ); |
| 330 | | - |
| 331 | | - foreach ( $this->extra_namespaces as $nsshort => $nsuri ) { |
| 332 | | - $this->pre_ns_buffer .= "\n\txmlns:$nsshort=\"$nsuri\""; |
| 333 | | - } |
| 334 | | - |
| 335 | | - $full_export = file_get_contents( $outfile ); |
| 336 | | - $full_export = $this->pre_ns_buffer . $full_export . $this->post_ns_buffer; |
| 337 | | - |
| 338 | | - $file = fopen( $outfile, 'w' ); |
| 339 | | - fwrite( $file, $full_export ); |
| 340 | | - fclose( $file ); |
| 341 | | - } |
| 342 | | - } |
| 343 | | - |
| 344 | | - /** |
| 345 | | - * Print basic definitions a list of pages ordered by their page id. |
| 346 | | - * Offset and limit refer to the count of existing pages, not to the |
| 347 | | - * page id. |
| 348 | | - */ |
| 349 | | - public function printPageList( $offset = 0, $limit = 30 ) { |
| 350 | | - wfProfileIn( "RDF::PrintPageList" ); |
| 351 | | - |
| 352 | | - $db = & wfGetDB( DB_MASTER ); |
| 353 | | - $this->pre_ns_buffer = ''; |
| 354 | | - $this->post_ns_buffer = ''; |
| 355 | | - $this->first_flush = true; |
| 356 | | - $this->delay_flush = 10; // flush only after (fully) printing 11 objects |
| 357 | | - $this->extra_namespaces = array(); |
| 358 | | - $this->printHeader(); // also inits global namespaces |
| 359 | | - $linkCache =& LinkCache::singleton(); |
| 360 | | - |
| 361 | | - global $smwgNamespacesWithSemanticLinks; |
| 362 | | - $query = ''; |
| 363 | | - |
| 364 | | - foreach ( $smwgNamespacesWithSemanticLinks as $ns => $enabled ) { |
| 365 | | - if ( $enabled ) { |
| 366 | | - if ( $query != '' ) $query .= ' OR '; |
| 367 | | - $query .= 'page_namespace = ' . $db->addQuotes( $ns ); |
| 368 | | - } |
| 369 | | - } |
| 370 | | - |
| 371 | | - $res = $db->select( $db->tableName( 'page' ), |
| 372 | | - 'page_id,page_title,page_namespace', $query |
| 373 | | - , 'SMW::RDF::PrintPageList', array( 'ORDER BY' => 'page_id ASC', 'OFFSET' => $offset, 'LIMIT' => $limit ) ); |
| 374 | | - $foundpages = false; |
| 375 | | - |
| 376 | | - while ( $row = $db->fetchObject( $res ) ) { |
| 377 | | - $foundpages = true; |
| 378 | | - // $t = Title::makeTitle($row->page_namespace, $row->page_title); |
| 379 | | - // if ($t === null) continue; |
| 380 | | - // $et = new SMWExportTitle($t, $this); |
| 381 | | - $st = new SMWSmallTitle(); |
| 382 | | - |
| 383 | | - $st->dbkey = $row->page_title; |
| 384 | | - $st->namespace = $row->page_namespace; |
| 385 | | - |
| 386 | | - $this->printObject( $st, false, false ); |
| 387 | | - |
| 388 | | - if ( $this->delay_flush > 0 ) $this->delay_flush--; |
| 389 | | - |
| 390 | | - $linkCache->clear(); |
| 391 | | - } |
| 392 | | - if ( $foundpages ) { // add link to next result page |
| 393 | | - if ( strpos( SMWExporter::expandURI( '&wikiurl;' ), '?' ) === false ) { // check whether we have title as a first parameter or in URL |
| 394 | | - $nexturl = SMWExporter::expandURI( '&export;?offset=' ) . ( $offset + $limit ); |
| 395 | | - } else { |
| 396 | | - $nexturl = SMWExporter::expandURI( '&export;&offset=' ) . ( $offset + $limit ); |
| 397 | | - } |
| 398 | | - |
| 399 | | - $this->post_ns_buffer .= |
| 400 | | - "\t<!-- Link to next set of results -->\n" . |
| 401 | | - "\t<owl:Thing rdf:about=\"$nexturl\">\n" . |
| 402 | | - "\t\t<rdfs:isDefinedBy rdf:resource=\"$nexturl\"/>\n" . |
| 403 | | - "\t</owl:Thing>\n"; |
| 404 | | - } |
| 405 | | - |
| 406 | | - $this->printFooter(); |
| 407 | | - $this->flushBuffers( true ); |
| 408 | | - |
| 409 | | - wfProfileOut( "RDF::PrintPageList" ); |
| 410 | | - } |
| 411 | | - |
| 412 | | - |
| 413 | | - /** |
| 414 | | - * Print basic information about this site. |
| 415 | | - */ |
| 416 | | - public function printWikiInfo() { |
| 417 | | - wfProfileIn( "RDF::PrintWikiInfo" ); |
| 418 | | - |
| 419 | | - global $wgSitename, $wgLanguageCode; |
| 420 | | - |
| 421 | | - $db = & wfGetDB( DB_MASTER ); |
| 422 | | - $this->pre_ns_buffer = ''; |
| 423 | | - $this->post_ns_buffer = ''; |
| 424 | | - $this->extra_namespaces = array(); |
| 425 | | - $data = new SMWExpData( new SMWExpResource( '&wiki;#wiki' ) ); |
| 426 | | - |
| 427 | | - $ed = new SMWExpData( SMWExporter::getSpecialElement( 'swivt', 'Wikisite' ) ); |
| 428 | | - $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'rdf', 'type' ), $ed ); |
| 429 | | - |
| 430 | | - $ed = new SMWExpData( new SMWExpLiteral( $wgSitename ) ); |
| 431 | | - $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'rdfs', 'label' ), $ed ); |
| 432 | | - |
| 433 | | - $ed = new SMWExpData( new SMWExpLiteral( $wgSitename, null, 'http://www.w3.org/2001/XMLSchema#string' ) ); |
| 434 | | - $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'siteName' ), $ed ); |
| 435 | | - |
| 436 | | - $ed = new SMWExpData( new SMWExpLiteral( SMWExporter::expandURI( '&wikiurl;' ), null, 'http://www.w3.org/2001/XMLSchema#string' ) ); |
| 437 | | - $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'pagePrefix' ), $ed ); |
| 438 | | - |
| 439 | | - $ed = new SMWExpData( new SMWExpLiteral( SMW_VERSION, null, 'http://www.w3.org/2001/XMLSchema#string' ) ); |
| 440 | | - $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'smwVersion' ), $ed ); |
| 441 | | - |
| 442 | | - $ed = new SMWExpData( new SMWExpLiteral( $wgLanguageCode, null, 'http://www.w3.org/2001/XMLSchema#string' ) ); |
| 443 | | - $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'langCode' ), $ed ); |
| 444 | | - |
| 445 | | - // stats |
| 446 | | - $ed = new SMWExpData( new SMWExpLiteral( SiteStats::pages(), null, 'http://www.w3.org/2001/XMLSchema#int' ) ); |
| 447 | | - $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'pageCount' ), $ed ); |
| 448 | | - |
| 449 | | - $ed = new SMWExpData( new SMWExpLiteral( SiteStats::articles(), null, 'http://www.w3.org/2001/XMLSchema#int' ) ); |
| 450 | | - $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'contentPageCount' ), $ed ); |
| 451 | | - |
| 452 | | - $ed = new SMWExpData( new SMWExpLiteral( SiteStats::images(), null, 'http://www.w3.org/2001/XMLSchema#int' ) ); |
| 453 | | - $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'mediaCount' ), $ed ); |
| 454 | | - |
| 455 | | - $ed = new SMWExpData( new SMWExpLiteral( SiteStats::edits(), null, 'http://www.w3.org/2001/XMLSchema#int' ) ); |
| 456 | | - $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'editCount' ), $ed ); |
| 457 | | - |
| 458 | | - $ed = new SMWExpData( new SMWExpLiteral( SiteStats::views(), null, 'http://www.w3.org/2001/XMLSchema#int' ) ); |
| 459 | | - $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'viewCount' ), $ed ); |
| 460 | | - |
| 461 | | - $ed = new SMWExpData( new SMWExpLiteral( SiteStats::users(), null, 'http://www.w3.org/2001/XMLSchema#int' ) ); |
| 462 | | - $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'userCount' ), $ed ); |
| 463 | | - |
| 464 | | - $ed = new SMWExpData( new SMWExpLiteral( SiteStats::admins(), null, 'http://www.w3.org/2001/XMLSchema#int' ) ); |
| 465 | | - $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'adminCount' ), $ed ); |
| 466 | | - |
| 467 | | - $mainpage = Title::newMainPage(); |
| 468 | | - |
| 469 | | - if ( $mainpage !== null ) { |
| 470 | | - $ed = new SMWExpData( new SMWExpResource( $mainpage->getFullURL() ) ); |
| 471 | | - $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'mainPage' ), $ed ); |
| 472 | | - } |
| 473 | | - |
| 474 | | - $this->printHeader(); // also inits global namespaces |
| 475 | | - $this->printExpData( $data ); |
| 476 | | - |
| 477 | | - if ( strpos( SMWExporter::expandURI( '&wikiurl;' ), '?' ) === false ) { // check whether we have title as a first parameter or in URL |
| 478 | | - $nexturl = SMWExporter::expandURI( '&export;?offset=0' ); |
| 479 | | - } else { |
| 480 | | - $nexturl = SMWExporter::expandURI( '&export;&offset=0' ); |
| 481 | | - } |
| 482 | | - |
| 483 | | - $this->post_ns_buffer .= |
| 484 | | - "\t<!-- Link to semantic page list -->\n" . |
| 485 | | - "\t<owl:Thing rdf:about=\"$nexturl\">\n" . |
| 486 | | - "\t\t<rdfs:isDefinedBy rdf:resource=\"$nexturl\"/>\n" . |
| 487 | | - "\t</owl:Thing>\n"; |
| 488 | | - |
| 489 | | - $this->printFooter(); |
| 490 | | - $this->flushBuffers( true ); |
| 491 | | - |
| 492 | | - wfProfileOut( "RDF::PrintWikiInfo" ); |
| 493 | | - } |
| 494 | | - |
| 495 | | - /* Functions for exporting RDF */ |
| 496 | | - |
| 497 | | - protected function makeValueEntityString( $string ) { |
| 498 | | - return "'" . str_replace( '%','%',$string ) . "'"; |
| 499 | | - } |
| 500 | | - |
| 501 | | - protected function printHeader( $ontologyuri = '' ) { |
| 502 | | - global $wgContLang; |
| 503 | | - |
| 504 | | - $this->pre_ns_buffer .= |
| 505 | | - "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n" . |
| 506 | | - "<!DOCTYPE rdf:RDF[\n" . |
| 507 | | - "\t<!ENTITY rdf " . $this->makeValueEntityString( SMWExporter::expandURI( '&rdf;' ) ) . ">\n" . |
| 508 | | - "\t<!ENTITY rdfs " . $this->makeValueEntityString( SMWExporter::expandURI( '&rdfs;' ) ) . ">\n" . |
| 509 | | - "\t<!ENTITY owl " . $this->makeValueEntityString( SMWExporter::expandURI( '&owl;' ) ) . ">\n" . |
| 510 | | - "\t<!ENTITY swivt " . $this->makeValueEntityString( SMWExporter::expandURI( '&swivt;' ) ) . ">\n" . |
| 511 | | - // A note on "wiki": this namespace is crucial as a fallback when it would be illegal to start e.g. with a number. In this case, one can always use wiki:... followed by "_" and possibly some namespace, since _ is legal as a first character. |
| 512 | | - "\t<!ENTITY wiki " . $this->makeValueEntityString( SMWExporter::expandURI( '&wiki;' ) ) . ">\n" . |
| 513 | | - "\t<!ENTITY property " . $this->makeValueEntityString( SMWExporter::expandURI( '&property;' ) ) . ">\n" . |
| 514 | | - "\t<!ENTITY wikiurl " . $this->makeValueEntityString( SMWExporter::expandURI( '&wikiurl;' ) ) . ">\n" . |
| 515 | | - "]>\n\n" . |
| 516 | | - "<rdf:RDF\n" . |
| 517 | | - "\txmlns:rdf=\"&rdf;\"\n" . |
| 518 | | - "\txmlns:rdfs=\"&rdfs;\"\n" . |
| 519 | | - "\txmlns:owl =\"&owl;\"\n" . |
| 520 | | - "\txmlns:swivt=\"&swivt;\"\n" . |
| 521 | | - "\txmlns:wiki=\"&wiki;\"\n" . |
| 522 | | - "\txmlns:property=\"&property;\""; |
| 523 | | - $this->global_namespaces = array( 'rdf' => true, 'rdfs' => true, 'owl' => true, 'swivt' => true, 'wiki' => true, 'property' => true ); |
| 524 | | - |
| 525 | | - $this->post_ns_buffer .= |
| 526 | | - ">\n\t<!-- Ontology header -->\n" . |
| 527 | | - "\t<owl:Ontology rdf:about=\"$ontologyuri\">\n" . |
| 528 | | - "\t\t<swivt:creationDate rdf:datatype=\"http://www.w3.org/2001/XMLSchema#dateTime\">" . date( DATE_W3C ) . "</swivt:creationDate>\n" . |
| 529 | | - "\t\t<owl:imports rdf:resource=\"http://semantic-mediawiki.org/swivt/1.0\" />\n" . |
| 530 | | - "\t</owl:Ontology>\n" . |
| 531 | | - "\t<!-- exported page data -->\n"; |
| 532 | | - } |
| 533 | | - |
| 534 | | - /** |
| 535 | | - * Prints the footer. |
| 536 | | - */ |
| 537 | | - protected function printFooter() { |
| 538 | | - $this->post_ns_buffer .= "\t<!-- Created by Semantic MediaWiki, http://semantic-mediawiki.org/ -->\n"; |
| 539 | | - $this->post_ns_buffer .= '</rdf:RDF>'; |
| 540 | | - } |
| 541 | | - |
| 542 | | - /** |
| 543 | | - * Serialise the given semantic data. |
| 544 | | - */ |
| 545 | | - protected function printExpData( /*SMWExpData*/ $data, $indent = '' ) { |
| 546 | | - $type = $data->extractMainType()->getQName(); |
| 547 | | - |
| 548 | | - if ( '' == $this->pre_ns_buffer ) { // start new ns block |
| 549 | | - $this->pre_ns_buffer .= "\t$indent<$type"; |
| 550 | | - } else { |
| 551 | | - $this->post_ns_buffer .= "\t$indent<$type"; |
| 552 | | - } |
| 553 | | - |
| 554 | | - if ( ( $data->getSubject() instanceof SMWExpLiteral ) || ( $data->getSubject() instanceof SMWExpResource ) ) { |
| 555 | | - $this->post_ns_buffer .= ' rdf:about="' . $data->getSubject()->getName() . '"'; |
| 556 | | - } // else: blank node |
| 557 | | - |
| 558 | | - if ( count( $data->getProperties() ) == 0 ) { |
| 559 | | - $this->post_ns_buffer .= " />\n"; |
| 560 | | - } else { |
| 561 | | - $this->post_ns_buffer .= ">\n"; |
| 562 | | - |
| 563 | | - foreach ( $data->getProperties() as $property ) { |
| 564 | | - $this->queueElement( $property ); |
| 565 | | - |
| 566 | | - foreach ( $data->getValues( $property ) as $value ) { |
| 567 | | - $this->post_ns_buffer .= "\t\t$indent<" . $property->getQName(); |
| 568 | | - $this->addExtraNamespace( $property->getNamespaceID(), $property->getNamespace() ); |
| 569 | | - $object = $value->getSubject(); |
| 570 | | - |
| 571 | | - if ( $object instanceof SMWExpLiteral ) { |
| 572 | | - if ( $object->getDatatype() != '' ) { |
| 573 | | - $this->post_ns_buffer .= ' rdf:datatype="' . $object->getDatatype() . '"'; |
| 574 | | - } |
| 575 | | - |
| 576 | | - $this->post_ns_buffer .= '>' . |
| 577 | | - str_replace( array( '&', '>', '<' ), array( '&', '>', '<' ), $object->getName() ) . |
| 578 | | - '</' . $property->getQName() . ">\n"; |
| 579 | | - } else { // bnode or resource, may have subdescriptions |
| 580 | | - $collection = $value->getCollection(); |
| 581 | | - |
| 582 | | - if ( $collection != false ) { |
| 583 | | - $this->post_ns_buffer .= " rdf:parseType=\"Collection\">\n"; |
| 584 | | - |
| 585 | | - foreach ( $collection as $subvalue ) { |
| 586 | | - $this->printExpData( $subvalue, $indent . "\t\t" ); |
| 587 | | - } |
| 588 | | - |
| 589 | | - $this->post_ns_buffer .= "\t\t$indent</" . $property->getQName() . ">\n"; |
| 590 | | - } elseif ( count( $value->getProperties() ) > 0 ) { |
| 591 | | - $this->post_ns_buffer .= ">\n"; |
| 592 | | - $this->printExpData( $value, $indent . "\t\t" ); |
| 593 | | - $this->post_ns_buffer .= "\t\t$indent</" . $property->getQName() . ">\n"; |
| 594 | | - } else { |
| 595 | | - if ( $object instanceof SMWExpResource ) { |
| 596 | | - $this->post_ns_buffer .= ' rdf:resource="' . $object->getName() . '"'; |
| 597 | | - $this->queueElement( $object ); // queue only non-explicated resources |
| 598 | | - } |
| 599 | | - |
| 600 | | - $this->post_ns_buffer .= "/>\n"; |
| 601 | | - } |
| 602 | | - } |
| 603 | | - } |
| 604 | | - } |
| 605 | | - |
| 606 | | - $this->post_ns_buffer .= "\t$indent</" . $type . ">\n"; |
| 607 | | - } |
| 608 | | - |
| 609 | | - $this->flushBuffers(); |
| 610 | | - } |
| 611 | | - |
| 612 | | - /** |
| 613 | | - * Print the triples associated to a specific page, and references those needed. |
| 614 | | - * They get printed in the printFooter-function. |
| 615 | | - * |
| 616 | | - * @param $st SMWSmallTitle wrapping the page to be exported |
| 617 | | - * @param $fullexport Boolean to define whether all the data for the page should |
| 618 | | - * be exported, or whether just a definition of the given title. |
| 619 | | - * @param $backlinks Boolean specifying if properties linking to the exported title |
| 620 | | - * should be included. |
| 621 | | - */ |
| 622 | | - protected function printObject( /*SMWSmallTitle*/ $st, $fullexport = true, $backlinks = false ) { |
| 623 | | - global $smwgMW_1_14; |
| 624 | | - |
| 625 | | - if ( array_key_exists( $st->getHash(), $this->element_done ) ) return; // do not export twice |
| 626 | | - |
| 627 | | - $value = SMWWikiPageValue::makePage( $st->dbkey, $st->namespace ); |
| 628 | | - |
| 629 | | - if ( $this->date !== '' ) { // check date restriction if given |
| 630 | | - $rev = $smwgMW_1_14 ? Revision::getTimeStampFromID( $value->getTitle(), $value->getTitle()->getLatestRevID() ):Revision::getTimeStampFromID( $value->getTitle()->getLatestRevID() ); |
| 631 | | - if ( $rev < $this->date ) return; |
| 632 | | - } |
| 633 | | - |
| 634 | | - $data = SMWExporter::makeExportData( $this->getSemanticData( $value, !$fullexport ), $st->modifier ); |
| 635 | | - $this->printExpData( $data ); // serialise |
| 636 | | - |
| 637 | | - // let other extensions add additional RDF data for this page |
| 638 | | - $additionalDataArray = array(); |
| 639 | | - wfRunHooks( 'smwAddToRDFExport', array( $value->getTitle(), &$additionalDataArray, $fullexport, $backlinks ) ); |
| 640 | | - foreach ( $additionalDataArray as $additionalData ) { |
| 641 | | - $this->printExpData( $additionalData ); // serialise |
| 642 | | - } |
| 643 | | - |
| 644 | | - $this->markAsDone( $st ); |
| 645 | | - |
| 646 | | - // possibly add backlinks |
| 647 | | - if ( ( $fullexport ) && ( $backlinks ) ) { |
| 648 | | - wfProfileIn( "RDF::PrintPages::GetBacklinks" ); |
| 649 | | - $inRels = smwfGetStore()->getInProperties( $value ); |
| 650 | | - |
| 651 | | - foreach ( $inRels as $inRel ) { |
| 652 | | - $inSubs = smwfGetStore()->getPropertySubjects( $inRel, $value ); |
| 653 | | - |
| 654 | | - foreach ( $inSubs as $inSub ) { |
| 655 | | - $stb = new SMWSmallTitle(); |
| 656 | | - $stb->dbkey = $inSub->getDBkey(); |
| 657 | | - $stb->namespace = $inSub->getNamespace(); |
| 658 | | - |
| 659 | | - if ( !array_key_exists( $stb->getHash(), $this->element_done ) ) { |
| 660 | | - $semdata = $this->getSemanticData( $inSub, true ); |
| 661 | | - $semdata->addPropertyObjectValue( $inRel, $value ); |
| 662 | | - $data = SMWExporter::makeExportData( $semdata ); |
| 663 | | - $this->printExpData( $data ); |
| 664 | | - } |
| 665 | | - } |
| 666 | | - } |
| 667 | | - |
| 668 | | - if ( NS_CATEGORY === $value->getNamespace() ) { // also print elements of categories |
| 669 | | - $options = new SMWRequestOptions(); |
| 670 | | - $options->limit = 100; // Categories can be large, always use limit |
| 671 | | - $instances = smwfGetStore()->getPropertySubjects( SMWPropertyValue::makeProperty( '_INST' ), $value, $options ); |
| 672 | | - $pinst = SMWPropertyValue::makeProperty( '_INST' ); |
| 673 | | - |
| 674 | | - foreach ( $instances as $instance ) { |
| 675 | | - $stb = new SMWSmallTitle(); |
| 676 | | - $stb->dbkey = $instance->getDBkey(); |
| 677 | | - $stb->namespace = $instance->getNamespace(); |
| 678 | | - |
| 679 | | - if ( !array_key_exists( $stb->getHash(), $this->element_done ) ) { |
| 680 | | - $semdata = $this->getSemanticData( $instance, true ); |
| 681 | | - $semdata->addPropertyObjectValue( $pinst, $value ); |
| 682 | | - $data = SMWExporter::makeExportData( $semdata ); |
| 683 | | - $this->printExpData( $data ); |
| 684 | | - } |
| 685 | | - } |
| 686 | | - } elseif ( SMW_NS_CONCEPT === $value->getNamespace() ) { // print concept members (slightly different code) |
| 687 | | - $desc = new SMWConceptDescription( $value->getTitle() ); |
| 688 | | - $desc->addPrintRequest( new SMWPrintRequest( SMWPrintRequest::PRINT_THIS, '' ) ); |
| 689 | | - $query = new SMWQuery( $desc ); |
| 690 | | - $query->setLimit( 100 ); |
| 691 | | - |
| 692 | | - $res = smwfGetStore()->getQueryResult( $query ); |
| 693 | | - $resarray = $res->getNext(); |
| 694 | | - $pinst = SMWPropertyValue::makeProperty( '_INST' ); |
| 695 | | - |
| 696 | | - while ( $resarray !== false ) { |
| 697 | | - $instance = end( $resarray )->getNextObject(); |
| 698 | | - |
| 699 | | - $stb = new SMWSmallTitle(); |
| 700 | | - $stb->dbkey = $instance->getDBkey(); |
| 701 | | - $stb->namespace = $instance->getNamespace(); |
| 702 | | - |
| 703 | | - if ( !array_key_exists( $stb->getHash(), $this->element_done ) ) { |
| 704 | | - $semdata = $this->getSemanticData( $instance, true ); |
| 705 | | - $semdata->addPropertyObjectValue( $pinst, $value ); |
| 706 | | - $data = SMWExporter::makeExportData( $semdata ); |
| 707 | | - $this->printExpData( $data ); |
| 708 | | - } |
| 709 | | - |
| 710 | | - $resarray = $res->getNext(); |
| 711 | | - } |
| 712 | | - } |
| 713 | | - |
| 714 | | - wfProfileOut( "RDF::PrintPages::GetBacklinks" ); |
| 715 | | - } |
| 716 | | - } |
| 717 | | - |
| 718 | | - /** |
| 719 | | - * Flush all buffers and extra namespaces by printing them to stdout and flushing |
| 720 | | - * the output buffers afterwards. |
| 721 | | - * |
| 722 | | - * @param force if true, the flush cannot be delayed any longer |
| 723 | | - */ |
| 724 | | - protected function flushBuffers( $force = false ) { |
| 725 | | - if ( $this->post_ns_buffer == '' ) return; // nothing to flush (every non-empty pre_ns_buffer also requires a non-empty post_ns_buffer) |
| 726 | | - if ( ( 0 != $this->delay_flush ) && !$force ) return; // wait a little longer |
| 727 | | - |
| 728 | | - print $this->pre_ns_buffer; |
| 729 | | - $this->pre_ns_buffer = ''; |
| 730 | | - |
| 731 | | - foreach ( $this->extra_namespaces as $nsshort => $nsuri ) { |
| 732 | | - if ( $this->first_flush ) { |
| 733 | | - $this->global_namespaces[$nsshort] = true; |
| 734 | | - print "\n\t"; |
| 735 | | - } else print ' '; |
| 736 | | - |
| 737 | | - print "xmlns:$nsshort=\"$nsuri\""; |
| 738 | | - } |
| 739 | | - |
| 740 | | - $this->extra_namespaces = array(); |
| 741 | | - print $this->post_ns_buffer; |
| 742 | | - $this->post_ns_buffer = ''; |
| 743 | | - |
| 744 | | - // Ship data in small chunks (even though browsers often do not display anything |
| 745 | | - // before the file is complete -- this might be due to syntax highlighting features |
| 746 | | - // for app/xml). You may want to sleep(1) here for debugging this. |
| 747 | | - ob_flush(); |
| 748 | | - flush(); |
| 749 | | - |
| 750 | | - $this->first_flush = false; |
| 751 | | - } |
| 752 | | - |
| 753 | | - /** |
| 754 | | - * Add an extra namespace that was encountered during output. The method |
| 755 | | - * checks whether the required namespace is available globally and adds |
| 756 | | - * it to the list of extra_namespace otherwise. |
| 757 | | - */ |
| 758 | | - public function addExtraNamespace( $nsshort, $nsuri ) { |
| 759 | | - if ( !array_key_exists( $nsshort, $this->global_namespaces ) ) { |
| 760 | | - $this->extra_namespaces[$nsshort] = $nsuri; |
| 761 | | - } |
| 762 | | - } |
| 763 | | - |
| 764 | | - /** |
| 765 | | - * Add a given SMWExpResource to the export queue if needed. |
| 766 | | - */ |
| 767 | | - public function queueElement( $element ) { |
| 768 | | - if ( !( $element instanceof SMWExpResource ) ) return; // only Resources are queued |
| 769 | | - $title = $element->getDataValue(); |
| 770 | | - |
| 771 | | - if ( $title instanceof SMWWikiPageValue ) { |
| 772 | | - $spt = new SMWSmallTitle(); |
| 773 | | - $title = $title->getTitle(); |
| 774 | | - $spt->dbkey = $title->getDBkey(); |
| 775 | | - $spt->namespace = $title->getNamespace(); |
| 776 | | - $spt->modifier = $element->getModifier(); |
| 777 | | - |
| 778 | | - if ( !array_key_exists( $spt->getHash(), $this->element_done ) ) { |
| 779 | | - $this->element_queue[$spt->getHash()] = $spt; |
| 780 | | - } |
| 781 | | - } |
| 782 | | - } |
| 783 | | - |
| 784 | | - /** |
| 785 | | - * Mark an article as done while making sure that the cache used for this |
| 786 | | - * stays reasonably small. Input is given as an SMWExportArticle object. |
| 787 | | - */ |
| 788 | | - protected function markAsDone( $st ) { |
| 789 | | - if ( count( $this->element_done ) >= self::MAX_CACHE_SIZE ) { |
| 790 | | - $this->element_done = array_slice( $this->element_done, |
| 791 | | - self::CACHE_BACKJUMP, |
| 792 | | - self::MAX_CACHE_SIZE - self::CACHE_BACKJUMP, |
| 793 | | - true ); |
| 794 | | - } |
| 795 | | - $this->element_done[$st->getHash()] = $st; // mark title as done |
| 796 | | - unset( $this->element_queue[$st->getHash()] ); // make sure it is not in the queue |
| 797 | | - } |
| 798 | | - |
| 799 | | - /** |
| 800 | | - * Retrieve a copy of the semantic data for a wiki page, possibly filtering |
| 801 | | - * it so that only essential properties are included (in some cases, we only |
| 802 | | - * want to export stub information about a page). |
| 803 | | - * We make a copy of the object since we may want to add more data later on |
| 804 | | - * and we do not want to modify the store's result which may be used for |
| 805 | | - * caching purposes elsewhere. |
| 806 | | - */ |
| 807 | | - protected function getSemanticData( $pagevalue, $core_props_only ) { |
| 808 | | - $semdata = smwfGetStore()->getSemanticData( $pagevalue, $core_props_only ? array( '__spu', '__typ', '__imp' ) : false ); // advise store to retrieve only core things |
| 809 | | - if ( $core_props_only ) { // be sure to filter all non-relevant things that may still be present in the retrieved |
| 810 | | - $result = new SMWSemanticData( $pagevalue ); |
| 811 | | - foreach ( array( '__spu', '__typ', '__imp' ) as $propid ) { |
| 812 | | - $prop = SMWPropertyValue::makeProperty( $propid ); |
| 813 | | - $values = $semdata->getPropertyValues( $prop ); |
| 814 | | - foreach ( $values as $dv ) { |
| 815 | | - $result->addPropertyObjectValue( $prop, $dv ); |
| 816 | | - } |
| 817 | | - } |
| 818 | | - } else { |
| 819 | | - $result = clone $semdata; |
| 820 | | - } |
| 821 | | - return $result; |
| 822 | | - } |
| 823 | | - |
| 824 | | - /** |
| 825 | | - * This function checks whether some article fits into a given namespace restriction. |
| 826 | | - * FALSE means "no restriction," non-negative restictions require to check whether |
| 827 | | - * the given number equals the given namespace. A restriction of -1 requires the |
| 828 | | - * namespace to be different from Category:, Relation:, Attribute:, and Type:. |
| 829 | | - */ |
| 830 | | - static public function fitsNsRestriction( $res, $ns ) { |
| 831 | | - if ( $res === false ) return true; |
| 832 | | - if ( is_array( $res ) ) return in_array( $ns, $res ); |
| 833 | | - if ( $res >= 0 ) return ( $res == $ns ); |
| 834 | | - return ( ( $res != NS_CATEGORY ) && ( $res != SMW_NS_PROPERTY ) && ( $res != SMW_NS_TYPE ) ); |
| 835 | | - } |
| 836 | | - |
| 837 | | -} |
| Index: trunk/extensions/SemanticMediaWiki/includes/export/SMW_Exp_Data.php |
| — | — | @@ -40,7 +40,7 @@ |
| 41 | 41 | $rdftype = SMWExporter::getSpecialElement( 'rdf', 'type' ); |
| 42 | 42 | $rdffirst = SMWExporter::getSpecialElement( 'rdf', 'first' ); |
| 43 | 43 | $rdfrest = SMWExporter::getSpecialElement( 'rdf', 'rest' ); |
| 44 | | - $result = new SMWExpData( new SMWExpElement( '' ) ); // bnode |
| | 44 | + $result = new SMWExpData( new SMWExpResource( '' ) ); // bnode |
| 45 | 45 | $result->addPropertyObjectValue( $rdftype, new SMWExpData( SMWExporter::getSpecialElement( 'rdf', 'List' ) ) ); |
| 46 | 46 | $result->addPropertyObjectValue( $rdffirst, array_shift( $elements ) ); |
| 47 | 47 | $result->addPropertyObjectValue( $rdfrest, SMWExpData::makeCollection( $elements ) ); |
| — | — | @@ -58,7 +58,7 @@ |
| 59 | 59 | /** |
| 60 | 60 | * Set the subject element. |
| 61 | 61 | */ |
| 62 | | - public function setSubject( SMWExpElement $subject ) { |
| | 62 | + public function setSubject( SMWExpResource $subject ) { |
| 63 | 63 | $this->m_subject = $subject; |
| 64 | 64 | } |
| 65 | 65 | |
| — | — | @@ -66,7 +66,7 @@ |
| 67 | 67 | * Store a value for a property identified by its title object. No duplicate elimination as this |
| 68 | 68 | * is usually done in SMWSemanticData already (which is typically used to generate this object) |
| 69 | 69 | */ |
| 70 | | - public function addPropertyObjectValue( SMWExpElement $property, SMWExpData $child ) { |
| | 70 | + public function addPropertyObjectValue( SMWExpResource $property, SMWExpData $child ) { |
| 71 | 71 | if ( !array_key_exists( $property->getName(), $this->m_edges ) ) { |
| 72 | 72 | $this->m_children[$property->getName()] = array(); |
| 73 | 73 | $this->m_edges[$property->getName()] = $property; |
| — | — | @@ -84,7 +84,7 @@ |
| 85 | 85 | /** |
| 86 | 86 | * Return the list of SMWExpData values associated to some property (element) |
| 87 | 87 | */ |
| 88 | | - public function getValues( /*SMWExpElement*/ $property ) { |
| | 88 | + public function getValues( SMWExpResource $property ) { |
| 89 | 89 | if ( array_key_exists( $property->getName(), $this->m_children ) ) { |
| 90 | 90 | return $this->m_children[$property->getName()]; |
| 91 | 91 | } else { |
| — | — | @@ -185,7 +185,7 @@ |
| 186 | 186 | if ( ( $name == '' ) || ( $name[0] == '_' ) ) { // bnode, rename ID to avoid unifying bnodes of different contexts |
| 187 | 187 | // TODO: should we really rename bnodes of the form "_id" here? |
| 188 | 188 | $child = clone $child; |
| 189 | | - $subject = new SMWExpElement( '_' . $smwgBnodeCount++, $child->getSubject()->getDataValue() ); |
| | 189 | + $subject = new SMWExpResource( '_' . $smwgBnodeCount++, $child->getSubject()->getDataValue() ); |
| 190 | 190 | $child->setSubject( $subject ); |
| 191 | 191 | } |
| 192 | 192 | $result[] = array( $this->m_subject, $edge, $child->getSubject() ); |
| Index: trunk/extensions/SemanticMediaWiki/includes/export/SMW_Exp_Element.php |
| — | — | @@ -1,7 +1,7 @@ |
| 2 | 2 | <?php |
| 3 | 3 | /** |
| 4 | | - * SMWExpElement is a class for representing single elements that appear in exported |
| 5 | | - * data, such as individual resources, data literals, or blank nodes. |
| | 4 | + * SMWExpElement is a class for representing single elements that appear in |
| | 5 | + * exported data, such as individual resources, data literals, or blank nodes. |
| 6 | 6 | * |
| 7 | 7 | * @author Markus Krötzsch |
| 8 | 8 | * @file |
| — | — | @@ -9,18 +9,14 @@ |
| 10 | 10 | */ |
| 11 | 11 | |
| 12 | 12 | /** |
| 13 | | - * A single element for export, e.g. a data literal, instance name, or blank node. |
| 14 | | - * Supports various serialisation aids for creating URIs or other strings for export. |
| 15 | | - * This abstract base class declares the basic common functionality of export elements. |
| | 13 | + * A single element for export, e.g. a data literal, instance name, or blank |
| | 14 | + * node. This abstract base class declares the basic common functionality of |
| | 15 | + * export elements (which is not much, really). |
| | 16 | + * @note This class should not be instantiated directly. |
| 16 | 17 | * |
| 17 | | - * This class can also be used to represent blank nodes: It is assumed that all objects |
| 18 | | - * of class SMWExpElement or any of its subclasses do represent blank node if their name |
| 19 | | - * is empty or of the form "_id" where "id" is any identifier string. IDs are local to the |
| 20 | | - * current context, such as a list of triples or an SMWExpData container. |
| 21 | 18 | * @ingroup SMW |
| 22 | 19 | */ |
| 23 | 20 | class SMWExpElement { |
| 24 | | - |
| 25 | 21 | protected $m_dv; |
| 26 | 22 | protected $m_name; |
| 27 | 23 | |
| — | — | @@ -51,8 +47,17 @@ |
| 52 | 48 | } |
| 53 | 49 | |
| 54 | 50 | /** |
| 55 | | - * A single resource (individual) for export. Defined by a URI, and possibly also providing |
| 56 | | - * abbreviated forms (QNames). |
| | 51 | + * A single resource (individual) for export. Defined by a URI, and possibly |
| | 52 | + * also providing abbreviated forms (QNames). |
| | 53 | + * This class can also be used to represent blank nodes: It is assumed that all |
| | 54 | + * objects of class SMWExpElement or any of its subclasses do represent blank |
| | 55 | + * node if their name is empty or of the form "_id" where "id" is any |
| | 56 | + * identifier string. IDs are local to the current context, such as a list of |
| | 57 | + * triples or an SMWExpData container. |
| | 58 | + * |
| | 59 | + * @todo This class should be split into two: one general resource class, and |
| | 60 | + * one that only supports resources with namespace/qname form, because the |
| | 61 | + * latter is strictly necessary in some places where resources are used. |
| 57 | 62 | * @ingroup SMW |
| 58 | 63 | */ |
| 59 | 64 | class SMWExpResource extends SMWExpElement { |
| — | — | @@ -79,6 +84,13 @@ |
| 80 | 85 | SMWExpElement::__construct( $name, $dv ); |
| 81 | 86 | } |
| 82 | 87 | } |
| | 88 | + |
| | 89 | + /** |
| | 90 | + * Return true of this resource represents a blank node. |
| | 91 | + */ |
| | 92 | + public function isBlankNode() { |
| | 93 | + return ( $this->m_name == '' ) || ( $this->m_name{0} == '_' ); |
| | 94 | + } |
| 83 | 95 | |
| 84 | 96 | /** |
| 85 | 97 | * SMW uses URI-Refs (#) to make "variants" of some base URI, e.g. to create multiple |
| Index: trunk/extensions/SemanticMediaWiki/includes/export/SMW_ExportController.php |
| — | — | @@ -0,0 +1,652 @@ |
| | 2 | +<?php |
| | 3 | + |
| | 4 | +/** |
| | 5 | + * File holding the SMWExportController class that provides basic functions for |
| | 6 | + * exporting pages to RDF and OWL. |
| | 7 | + * |
| | 8 | + * @file SMW_ExportController.php |
| | 9 | + * @ingroup SMW |
| | 10 | + * |
| | 11 | + * @author Markus Krötzsch |
| | 12 | + */ |
| | 13 | + |
| | 14 | +/** |
| | 15 | + * Small data object that specifies one wiki page to be serialised. |
| | 16 | + * SMWSmallTitle objects are used to queue pages for serialisation, hence it |
| | 17 | + * should be small to save memory. |
| | 18 | + * |
| | 19 | + * @ingroup SMW |
| | 20 | + */ |
| | 21 | +class SMWSmallTitle { |
| | 22 | + /// DB key version of the title. |
| | 23 | + public $dbkey; |
| | 24 | + /// MediaWiki namespace constant. |
| | 25 | + public $namespace; |
| | 26 | + /** |
| | 27 | + * Recursion depth for serialising this object. Depth of 1 or above means |
| | 28 | + * the object is serialised with all property values, and referenced |
| | 29 | + * objects are serialised with depth reduced by 1. Depth 0 means that only |
| | 30 | + * minimal declarations are serialised, so no dependencies are added. A |
| | 31 | + * depth of -1 encodes "infinite" depth, i.e. a complete recursive |
| | 32 | + * serialisation without limit. |
| | 33 | + * @var integer |
| | 34 | + */ |
| | 35 | + public $recdepth = 1; |
| | 36 | + |
| | 37 | + public function getHash() { |
| | 38 | + return $this->dbkey . ' ' . $this->namespace; |
| | 39 | + } |
| | 40 | +} |
| | 41 | + |
| | 42 | +/** |
| | 43 | + * Class for controlling the export of SMW page data, supporting high-level |
| | 44 | + * features such as recursive export and backlink inclusion. The class controls |
| | 45 | + * export independent of the serialisation syntax that is used. |
| | 46 | + * |
| | 47 | + * @ingroup SMW |
| | 48 | + */ |
| | 49 | +class SMWExportController { |
| | 50 | + const MAX_CACHE_SIZE = 5000; // do not let cache arrays get larger than this |
| | 51 | + const CACHE_BACKJUMP = 500; // kill this many cached entries if limit is reached, |
| | 52 | + // avoids too much array copying; <= MAX_CACHE_SIZE! |
| | 53 | + /** |
| | 54 | + * The object used for serialisation. |
| | 55 | + * @var SMWSerializer |
| | 56 | + */ |
| | 57 | + protected $serializer; |
| | 58 | + /** |
| | 59 | + * An array that keeps track of the elements for which we still need to |
| | 60 | + * write auxiliary definitions/declarations. |
| | 61 | + */ |
| | 62 | + protected $element_queue; |
| | 63 | + /** |
| | 64 | + * An array that keeps track of the recursion depth with which each object |
| | 65 | + * has been serialised. |
| | 66 | + */ |
| | 67 | + protected $element_done; |
| | 68 | + /** |
| | 69 | + * Boolean to indicate whether all objects that are exported in full (with |
| | 70 | + * all data) should also lead to the inclusion of all "inlinks" that they |
| | 71 | + * receive from other objects. If yes, these other objects are also |
| | 72 | + * serialised with at least the relevant inlinking properties included. |
| | 73 | + * Adding such dependencies counts as "recursive serialisation" and whether |
| | 74 | + * or not inlinking objects are included in full depends on the setting for |
| | 75 | + * recursion depth. Setting this to true enables "browsable RDF". |
| | 76 | + */ |
| | 77 | + protected $add_backlinks; |
| | 78 | + /** |
| | 79 | + * Controls how long to wait until flushing content to output. Flushing |
| | 80 | + * early may reduce the memory footprint of serialization functions. |
| | 81 | + * Flushing later has some advantages for export formats like RDF/XML where |
| | 82 | + * global namespace declarations are only possible by modifying the header, |
| | 83 | + * so that only local declarations are possible after the first flush. |
| | 84 | + */ |
| | 85 | + protected $delay_flush; |
| | 86 | + /** |
| | 87 | + * File handle for a potential output file to write to, or null if printing |
| | 88 | + * to standard output. |
| | 89 | + */ |
| | 90 | + protected $outputfile; |
| | 91 | + |
| | 92 | + /** |
| | 93 | + * Constructor. |
| | 94 | + * @param SMWSerializer $serializer defining the object used for syntactic |
| | 95 | + * serialization. |
| | 96 | + * @param boolean $enable_backlinks defining if backlinks are included, |
| | 97 | + * see $add_backlinks for details. |
| | 98 | + */ |
| | 99 | + public function __construct( SMWSerializer $serializer, $enable_backlinks = false ) { |
| | 100 | + $this->serializer = $serializer; |
| | 101 | + $this->outputfile = null; |
| | 102 | + $this->add_backlinks = $enable_backlinks; |
| | 103 | + } |
| | 104 | + |
| | 105 | + /** |
| | 106 | + * Enable or disable inclusion of backlinks into the output. |
| | 107 | + * @param boolean $enable |
| | 108 | + */ |
| | 109 | + public function enableBacklinks( $enable ) { |
| | 110 | + $this->add_backlinks = $enable; |
| | 111 | + } |
| | 112 | + |
| | 113 | + /** |
| | 114 | + * Initialize all internal structures to begin with some serialization. |
| | 115 | + * Returns true if initialization was successful (this means that the |
| | 116 | + * optional output file is writable). |
| | 117 | + * @param string $outfilename URL of the file that output should be written |
| | 118 | + * to, or empty string for writting to the standard output. |
| | 119 | + */ |
| | 120 | + protected function prepareSerialization( $outfilename = '' ) { |
| | 121 | + $this->serializer->clear(); |
| | 122 | + $this->element_queue = array(); |
| | 123 | + $this->element_done = array(); |
| | 124 | + if ( $outfilename != '' ) { |
| | 125 | + $this->outputfile = fopen( $outfilename, 'w' ); |
| | 126 | + if ( !$this->outputfile ) { // TODO Rather throw an exception here. |
| | 127 | + print "\nCannot open \"$outfile\" for writing.\n"; |
| | 128 | + return false; |
| | 129 | + } |
| | 130 | + } |
| | 131 | + return true; |
| | 132 | + } |
| | 133 | + |
| | 134 | + /** |
| | 135 | + * Serialize data associated to a specific page. This method works on the |
| | 136 | + * level of pages, i.e. it serialises parts of SMW content and implements |
| | 137 | + * features like recursive export or backlinks that are available for this |
| | 138 | + * type of data. |
| | 139 | + * |
| | 140 | + * @param SMWWikiPageValue $value specifying the page to be exported |
| | 141 | + * @param integer $recursiondepth specifying the depth of recursion, see |
| | 142 | + * SMWSmallTitle::$recdepth |
| | 143 | + */ |
| | 144 | + protected function serializePage( SMWWikiPageValue $value, $recursiondepth = 1 ) { |
| | 145 | + $st = new SMWSmallTitle(); |
| | 146 | + $st->dbkey = $value->getDBKey(); |
| | 147 | + $st->namespace = $value->getNamespace(); |
| | 148 | + $st->recdepth = $recursiondepth; |
| | 149 | + if ( $this->isDone( $st ) ) return; // do not export twice |
| | 150 | + $this->markAsDone( $st ); |
| | 151 | + $data = SMWExporter::makeExportData( $this->getSemanticData( $value, ( $recursiondepth == 0 ) ) ); |
| | 152 | + $this->serializer->serializeExpData( $data, $recursiondepth ); |
| | 153 | + |
| | 154 | + // let other extensions add additional RDF data for this page |
| | 155 | + $additionalDataArray = array(); |
| | 156 | + wfRunHooks( 'smwAddToRDFExport', array( $value->getTitle(), &$additionalDataArray, ( $recursiondepth != 0 ), $this->add_backlinks ) ); |
| | 157 | + foreach ( $additionalDataArray as $additionalData ) { |
| | 158 | + $this->serializer->serializeExpData( $additionalData ); // serialise |
| | 159 | + } |
| | 160 | + |
| | 161 | + if ( $recursiondepth != 0 ) { |
| | 162 | + $subrecdepth = ($recursiondepth>0) ? ($recursiondepth-1) : ($recursiondepth==0 ? 0 : -1); |
| | 163 | + |
| | 164 | + foreach ( $data->getProperties() as $property ) { |
| | 165 | + if ( $property->getDataValue() instanceof SMWWikiPageValue ) { |
| | 166 | + // TODO This currently drops modifiers (units of measurement) |
| | 167 | + $this->queuePage( $property->getDataValue(), 0 ); // no real recursion along properties |
| | 168 | + } |
| | 169 | + $wikipagevalues = false; |
| | 170 | + foreach ( $data->getValues( $property ) as $expdata ) { |
| | 171 | + $subject = $expdata->getSubject(); |
| | 172 | + if ( !$wikipagevalues && ( $subject->getDataValue() instanceof SMWWikiPageValue ) ) { |
| | 173 | + $wikipagevalues = true; |
| | 174 | + } elseif ( !$wikipagevalues ) { |
| | 175 | + break; |
| | 176 | + } |
| | 177 | + $this->queuePage( $subject->getDatavalue(), $subrecdepth ); |
| | 178 | + } |
| | 179 | + } |
| | 180 | + |
| | 181 | + // Add backlinks: |
| | 182 | + // Note: Backlinks are different from recursive serialisations, since |
| | 183 | + // stub declarations (recdepth==0) still need to have the property that |
| | 184 | + // links back to the object. So objects that would be exported with |
| | 185 | + // recdepth 0 cannot be put into the main queue but must be done right |
| | 186 | + // away. They also might be required many times, if they link back to |
| | 187 | + // many different objects in many ways (we cannot consider them "Done" |
| | 188 | + // if they were serialised at recdepth 0 only). |
| | 189 | + if ( $this->add_backlinks ) { |
| | 190 | + wfProfileIn( "RDF::PrintPages::GetBacklinks" ); |
| | 191 | + $inprops = smwfGetStore()->getInProperties( $value ); |
| | 192 | + foreach ( $inprops as $inprop ) { |
| | 193 | + if ( $inprop->getWikiPageValue() instanceof SMWWikiPageValue ) { |
| | 194 | + $this->queuePage( $inprop->getWikiPageValue(), 0 ); // no real recursion along properties |
| | 195 | + } |
| | 196 | + $inSubs = smwfGetStore()->getPropertySubjects( $inprop, $value ); |
| | 197 | + foreach ( $inSubs as $inSub ) { |
| | 198 | + $stb = new SMWSmallTitle(); |
| | 199 | + $stb->dbkey = $inSub->getDBkey(); |
| | 200 | + $stb->namespace = $inSub->getNamespace(); |
| | 201 | + $stb->recdepth = $subrecdepth; |
| | 202 | + if ( !$this->isDone($stb) ) { |
| | 203 | + $semdata = $this->getSemanticData( $inSub, true ); |
| | 204 | + $semdata->addPropertyObjectValue( $inprop, $value ); |
| | 205 | + $data = SMWExporter::makeExportData( $semdata ); |
| | 206 | + $this->serializer->serializeExpData( $data, $subrecdepth ); |
| | 207 | + } |
| | 208 | + } |
| | 209 | + } |
| | 210 | + |
| | 211 | + if ( NS_CATEGORY === $value->getNamespace() ) { // also print elements of categories |
| | 212 | + $options = new SMWRequestOptions(); |
| | 213 | + $options->limit = 100; // Categories can be large, always use limit |
| | 214 | + $instances = smwfGetStore()->getPropertySubjects( SMWPropertyValue::makeProperty( '_INST' ), $value, $options ); |
| | 215 | + $pinst = SMWPropertyValue::makeProperty( '_INST' ); |
| | 216 | + |
| | 217 | + foreach ( $instances as $instance ) { |
| | 218 | + $stb = new SMWSmallTitle(); |
| | 219 | + $stb->dbkey = $instance->getDBkey(); |
| | 220 | + $stb->namespace = $instance->getNamespace(); |
| | 221 | + |
| | 222 | + if ( !array_key_exists( $stb->getHash(), $this->element_done ) ) { |
| | 223 | + $semdata = $this->getSemanticData( $instance, true ); |
| | 224 | + $semdata->addPropertyObjectValue( $pinst, $value ); |
| | 225 | + $data = SMWExporter::makeExportData( $semdata ); |
| | 226 | + $this->serializer->serializeExpData( $data, $subrecdepth ); |
| | 227 | + } |
| | 228 | + } |
| | 229 | + } elseif ( SMW_NS_CONCEPT === $value->getNamespace() ) { // print concept members (slightly different code) |
| | 230 | + $desc = new SMWConceptDescription( $value->getTitle() ); |
| | 231 | + $desc->addPrintRequest( new SMWPrintRequest( SMWPrintRequest::PRINT_THIS, '' ) ); |
| | 232 | + $query = new SMWQuery( $desc ); |
| | 233 | + $query->setLimit( 100 ); |
| | 234 | + |
| | 235 | + $res = smwfGetStore()->getQueryResult( $query ); |
| | 236 | + $resarray = $res->getNext(); |
| | 237 | + $pinst = SMWPropertyValue::makeProperty( '_INST' ); |
| | 238 | + |
| | 239 | + while ( $resarray !== false ) { |
| | 240 | + $instance = end( $resarray )->getNextObject(); |
| | 241 | + |
| | 242 | + $stb = new SMWSmallTitle(); |
| | 243 | + $stb->dbkey = $instance->getDBkey(); |
| | 244 | + $stb->namespace = $instance->getNamespace(); |
| | 245 | + |
| | 246 | + if ( !array_key_exists( $stb->getHash(), $this->element_done ) ) { |
| | 247 | + $semdata = $this->getSemanticData( $instance, true ); |
| | 248 | + $semdata->addPropertyObjectValue( $pinst, $value ); |
| | 249 | + $data = SMWExporter::makeExportData( $semdata ); |
| | 250 | + $this->serializer->serializeExpData( $data ); |
| | 251 | + } |
| | 252 | + |
| | 253 | + $resarray = $res->getNext(); |
| | 254 | + } |
| | 255 | + } |
| | 256 | + wfProfileOut( "RDF::PrintPages::GetBacklinks" ); |
| | 257 | + } |
| | 258 | + } |
| | 259 | + } |
| | 260 | + |
| | 261 | + /** |
| | 262 | + * Serialize data associated to a specific page. |
| | 263 | + * |
| | 264 | + * @param SMWSmallTitle $st specifying the page to be exported |
| | 265 | + */ |
| | 266 | + protected function serializeSmallTitle( SMWSmallTitle $st ) { |
| | 267 | + if ( $this->isDone( $st ) ) return; // do not export twice |
| | 268 | + $value = SMWWikiPageValue::makePage( $st->dbkey, $st->namespace ); |
| | 269 | + $this->serializePage( $value, $st->recdepth ); |
| | 270 | + } |
| | 271 | + |
| | 272 | + /** |
| | 273 | + * Add a given SMWWikiPageValue to the export queue if needed. |
| | 274 | + */ |
| | 275 | + protected function queuePage( SMWWikiPageValue $pagevalue, $recursiondepth ) { |
| | 276 | + $spt = new SMWSmallTitle(); |
| | 277 | + $spt->dbkey = $pagevalue->getDBkey(); |
| | 278 | + $spt->namespace = $pagevalue->getNamespace(); |
| | 279 | + $spt->recdepth = $recursiondepth; |
| | 280 | + if ( !$this->isDone( $spt ) ) { |
| | 281 | + $this->element_queue[$spt->getHash()] = $spt; |
| | 282 | + } |
| | 283 | + } |
| | 284 | + |
| | 285 | + /** |
| | 286 | + * Mark an article as done while making sure that the cache used for this |
| | 287 | + * stays reasonably small. Input is given as an SMWSmallTitle object. |
| | 288 | + */ |
| | 289 | + protected function markAsDone( $st ) { |
| | 290 | + if ( count( $this->element_done ) >= self::MAX_CACHE_SIZE ) { |
| | 291 | + $this->element_done = array_slice( $this->element_done, |
| | 292 | + self::CACHE_BACKJUMP, |
| | 293 | + self::MAX_CACHE_SIZE - self::CACHE_BACKJUMP, |
| | 294 | + true ); |
| | 295 | + } |
| | 296 | + $hash = $st->getHash(); |
| | 297 | + if ( !$this->isDone( $st ) ) { |
| | 298 | + $this->element_done[$hash] = $st->recdepth; // mark title as done, with given recursion |
| | 299 | + } |
| | 300 | + unset( $this->element_queue[$hash] ); // make sure it is not in the queue |
| | 301 | + } |
| | 302 | + |
| | 303 | + /** |
| | 304 | + * Check if the given object has already been serialised at sufficient |
| | 305 | + * recursion depth. |
| | 306 | + * @param SMWSmallTitle $st specifying the object to check |
| | 307 | + */ |
| | 308 | + protected function isDone( SMWSmallTitle $st ) { |
| | 309 | + $hash = $st->getHash(); |
| | 310 | + return ( ( array_key_exists( $hash, $this->element_done ) ) && |
| | 311 | + ( ( $this->element_done[$hash] == -1 ) || |
| | 312 | + ( ( $st->recdepth != -1 ) && ( $this->element_done[$hash] >= $st->recdepth ) ) ) ); |
| | 313 | + } |
| | 314 | + |
| | 315 | + /** |
| | 316 | + * Retrieve a copy of the semantic data for a wiki page, possibly filtering |
| | 317 | + * it so that only essential properties are included (in some cases, we only |
| | 318 | + * want to export stub information about a page). |
| | 319 | + * We make a copy of the object since we may want to add more data later on |
| | 320 | + * and we do not want to modify the store's result which may be used for |
| | 321 | + * caching purposes elsewhere. |
| | 322 | + */ |
| | 323 | + protected function getSemanticData( $pagevalue, $core_props_only ) { |
| | 324 | + $semdata = smwfGetStore()->getSemanticData( $pagevalue, $core_props_only ? array( '__spu', '__typ', '__imp' ) : false ); // advise store to retrieve only core things |
| | 325 | + if ( $core_props_only ) { // be sure to filter all non-relevant things that may still be present in the retrieved |
| | 326 | + $result = new SMWSemanticData( $pagevalue ); |
| | 327 | + foreach ( array( '_URI', '_TYPE', '_IMPO' ) as $propid ) { |
| | 328 | + $prop = SMWPropertyValue::makeProperty( $propid ); |
| | 329 | + $values = $semdata->getPropertyValues( $prop ); |
| | 330 | + foreach ( $values as $dv ) { |
| | 331 | + $result->addPropertyObjectValue( $prop, $dv ); |
| | 332 | + } |
| | 333 | + } |
| | 334 | + } else { |
| | 335 | + $result = clone $semdata; |
| | 336 | + } |
| | 337 | + return $result; |
| | 338 | + } |
| | 339 | + |
| | 340 | + /** |
| | 341 | + * Send to the output what has been serialized so far. The flush might |
| | 342 | + * be deferred until later unless $force is true. |
| | 343 | + */ |
| | 344 | + protected function flush( $force = false ) { |
| | 345 | + if ( !$force && ( $this->delay_flush > 0 ) ) { |
| | 346 | + $this->delay_flush -= 1; |
| | 347 | + } elseif ( $this->outputfile !== null ) { |
| | 348 | + fwrite( $this->outputfile, $this->serializer->flushContent() ); |
| | 349 | + } else { |
| | 350 | + print $this->serializer->flushContent(); |
| | 351 | + // Ship data in small chunks (even though browsers often do not display anything |
| | 352 | + // before the file is complete -- this might be due to syntax highlighting features |
| | 353 | + // for app/xml). You may want to sleep(1) here for debugging this. |
| | 354 | + ob_flush(); |
| | 355 | + flush(); |
| | 356 | + } |
| | 357 | + } |
| | 358 | + |
| | 359 | + /** |
| | 360 | + * Create an SMWExpData container that encodes the ontology header for an |
| | 361 | + * SMW exported OWL file. |
| | 362 | + * |
| | 363 | + * @param string $ontologyuri specifying the URI of the ontology, possibly |
| | 364 | + * empty |
| | 365 | + */ |
| | 366 | + protected function getOntologyExpData( $ontologyuri ) { |
| | 367 | + $data = new SMWExpData( new SMWExpResource( $ontologyuri ) ); |
| | 368 | + $ed = new SMWExpData( SMWExporter::getSpecialElement( 'owl', 'Ontology' ) ); |
| | 369 | + $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'rdf', 'type' ), $ed ); |
| | 370 | + $ed = new SMWExpData( new SMWExpLiteral( date( DATE_W3C ), null, 'http://www.w3.org/2001/XMLSchema#dateTime' ) ); |
| | 371 | + $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'creationDate' ), $ed ); |
| | 372 | + $ed = new SMWExpData( new SMWExpResource( 'http://semantic-mediawiki.org/swivt/1.0' ) ); |
| | 373 | + $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'owl', 'imports' ), $ed ); |
| | 374 | + return $data; |
| | 375 | + } |
| | 376 | + |
| | 377 | + /** |
| | 378 | + * This function prints all selected pages, specified as an array of page |
| | 379 | + * names (strings with namespace identifiers). |
| | 380 | + * |
| | 381 | + * @param array $pages list of page names to export |
| | 382 | + * @param integer $recursion determines how pages are exported recursively: |
| | 383 | + * "0" means that referenced resources are only declared briefly, "1" means |
| | 384 | + * that all referenced resources are also exported recursively (propbably |
| | 385 | + * retrieving the whole wiki). |
| | 386 | + * @param string $revisiondate filter page list by including only pages |
| | 387 | + * that have been changed since this date; format "YmdHis" |
| | 388 | + * |
| | 389 | + * @todo Consider dropping the $revisiondate filtering and all associated |
| | 390 | + * functionality. Is anybody using this? |
| | 391 | + */ |
| | 392 | + public function printPages( $pages, $recursion = 1, $revisiondate = false ) { |
| | 393 | + global $smwgMW_1_14; |
| | 394 | + wfProfileIn( "RDF::PrintPages" ); |
| | 395 | + |
| | 396 | + $linkCache =& LinkCache::singleton(); |
| | 397 | + $this->prepareSerialization(); |
| | 398 | + $this->delay_flush = 10; // flush only after (fully) printing 11 objects |
| | 399 | + |
| | 400 | + // transform pages into queued short titles |
| | 401 | + foreach ( $pages as $page ) { |
| | 402 | + $title = Title::newFromText( $page ); |
| | 403 | + if ( null === $title ) continue; // invalid title name given |
| | 404 | + if ( $revisiondate !== '' ) { // filter page list by revision date |
| | 405 | + $rev = $smwgMW_1_14 ? Revision::getTimeStampFromID( $title, $title->getLatestRevID() ) : Revision::getTimeStampFromID( $title->getLatestRevID() ); |
| | 406 | + if ( $rev < $revisiondate ) continue; |
| | 407 | + } |
| | 408 | + $st = new SMWSmallTitle(); |
| | 409 | + $st->dbkey = $title->getDBkey(); |
| | 410 | + $st->namespace = $title->getNamespace(); |
| | 411 | + $st->recdepth = $recursion==1 ? -1 : 1; |
| | 412 | + $this->element_queue[$st->getHash()] = $st; |
| | 413 | + } |
| | 414 | + |
| | 415 | + $this->serializer->serializeHeader(); |
| | 416 | + |
| | 417 | + if ( count( $pages ) == 1 ) { // ensure that ontologies that are retrieved as linked data are not confused with their subject! |
| | 418 | + $ontologyuri = SMWExporter::expandURI( '&export;' ) . '/' . urlencode( end( $pages ) ); |
| | 419 | + } else { // use empty URI, i.e. "location" as URI otherwise |
| | 420 | + $ontologyuri = ''; |
| | 421 | + } |
| | 422 | + $this->serializer->serializeExpData( $this->getOntologyExpData( $ontologyuri ) ); |
| | 423 | + |
| | 424 | + while ( count( $this->element_queue ) > 0 ) { |
| | 425 | + $this->serializeSmallTitle( reset( $this->element_queue ) ); |
| | 426 | + $this->flush(); |
| | 427 | + $linkCache->clear(); // avoid potential memory leak |
| | 428 | + } |
| | 429 | + $this->serializer->serializeFooter(); |
| | 430 | + $this->flush( true ); |
| | 431 | + |
| | 432 | + wfProfileOut( "RDF::PrintPages" ); |
| | 433 | + } |
| | 434 | + |
| | 435 | + |
| | 436 | + /** |
| | 437 | + * This function exports the semantic data for all pages within the wiki, |
| | 438 | + * and for all elements that are referred to in the exported data. |
| | 439 | + * |
| | 440 | + * @param string $outfile the output file URI, or false if printing to stdout |
| | 441 | + * @param mixed $ns_restriction namespace restriction, see fitsNsRestriction() |
| | 442 | + * @param integer $delay number of microseconds for which to sleep during |
| | 443 | + * export to reduce server load in long-running operations |
| | 444 | + * @param integer $delayeach number of pages to process between two sleeps |
| | 445 | + */ |
| | 446 | + public function printAll( $outfile, $ns_restriction = false, $delay, $delayeach ) { |
| | 447 | + global $smwgNamespacesWithSemanticLinks; |
| | 448 | + $linkCache =& LinkCache::singleton(); |
| | 449 | + $db = wfGetDB( DB_SLAVE ); |
| | 450 | + |
| | 451 | + $this->delay_flush = 10; |
| | 452 | + if ( !$this->prepareSerialization( $outfile ) ) return; |
| | 453 | + |
| | 454 | + $this->serializer->serializeHeader(); |
| | 455 | + $this->serializer->serializeExpData( $this->getOntologyExpData( '' ) ); |
| | 456 | + |
| | 457 | + $end = $db->selectField( 'page', 'max(page_id)', false, $outfile ); |
| | 458 | + $a_count = 0; $d_count = 0; // DEBUG |
| | 459 | + $delaycount = $delayeach; |
| | 460 | + |
| | 461 | + for ( $id = 1; $id <= $end; $id += 1 ) { |
| | 462 | + $title = Title::newFromID( $id ); |
| | 463 | + if ( ( $title === null ) || !smwfIsSemanticsProcessed( $title->getNamespace() ) ) continue; |
| | 464 | + if ( !SMWExportController::fitsNsRestriction( $ns_restriction, $title->getNamespace() ) ) continue; |
| | 465 | + $a_count += 1; // DEBUG |
| | 466 | + |
| | 467 | + $st = new SMWSmallTitle(); |
| | 468 | + $st->dbkey = $title->getDBkey(); |
| | 469 | + $st->namespace = $title->getNamespace(); |
| | 470 | + $st->recdepth = 1; |
| | 471 | + $this->element_queue[$st->getHash()] = $st; |
| | 472 | + |
| | 473 | + while ( count( $this->element_queue ) > 0 ) { |
| | 474 | + $this->serializeSmallTitle( reset( $this->element_queue ) ); |
| | 475 | + // resolve dependencies that will otherwise not be printed |
| | 476 | + foreach ( $this->element_queue as $key => $staux ) { |
| | 477 | + if ( !smwfIsSemanticsProcessed( $staux->namespace ) || //( $staux->modifier !== '' ) || |
| | 478 | + !SMWExportController::fitsNsRestriction( $ns_restriction, $staux->namespace ) ) { |
| | 479 | + // Note: we do not need to check the cache to guess if an element was already |
| | 480 | + // printed. If so, it would not be included in the queue in the first place. |
| | 481 | + $d_count += 1; // DEBUG |
| | 482 | + } else { // don't carry values that you do not want to export (yet) |
| | 483 | + unset( $this->element_queue[$key] ); |
| | 484 | + } |
| | 485 | + } |
| | 486 | + // sleep each $delaycount for $delay µs to be nice to the server |
| | 487 | + if ( ( $delaycount-- < 0 ) && ( $delayeach != 0 ) ) { |
| | 488 | + usleep( $delay ); |
| | 489 | + $delaycount = $delayeach; |
| | 490 | + } |
| | 491 | + } |
| | 492 | + |
| | 493 | + $this->flush(); |
| | 494 | + $linkCache->clear(); |
| | 495 | + } |
| | 496 | + |
| | 497 | + $this->serializer->serializeFooter(); |
| | 498 | + $this->flush( true ); |
| | 499 | + } |
| | 500 | + |
| | 501 | + /** |
| | 502 | + * Print basic definitions a list of pages ordered by their page id. |
| | 503 | + * Offset and limit refer to the count of existing pages, not to the |
| | 504 | + * page id. |
| | 505 | + * @param integer $offset the number of the first (existing) page to |
| | 506 | + * serialize a declaration for |
| | 507 | + * @param integer $limit the number of pages to serialize |
| | 508 | + */ |
| | 509 | + public function printPageList( $offset = 0, $limit = 30 ) { |
| | 510 | + global $smwgNamespacesWithSemanticLinks; |
| | 511 | + wfProfileIn( "RDF::PrintPageList" ); |
| | 512 | + |
| | 513 | + $db = wfGetDB( DB_SLAVE ); |
| | 514 | + $this->prepareSerialization(); |
| | 515 | + $this->delay_flush = 35; // don't do intermediate flushes with default parameters |
| | 516 | + $linkCache = LinkCache::singleton(); |
| | 517 | + |
| | 518 | + $this->serializer->serializeHeader(); |
| | 519 | + $this->serializer->serializeExpData( $this->getOntologyExpData( '' ) ); |
| | 520 | + |
| | 521 | + $query = ''; |
| | 522 | + foreach ( $smwgNamespacesWithSemanticLinks as $ns => $enabled ) { |
| | 523 | + if ( $enabled ) { |
| | 524 | + if ( $query != '' ) $query .= ' OR '; |
| | 525 | + $query .= 'page_namespace = ' . $db->addQuotes( $ns ); |
| | 526 | + } |
| | 527 | + } |
| | 528 | + $res = $db->select( $db->tableName( 'page' ), |
| | 529 | + 'page_id,page_title,page_namespace', $query |
| | 530 | + , 'SMW::RDF::PrintPageList', array( 'ORDER BY' => 'page_id ASC', 'OFFSET' => $offset, 'LIMIT' => $limit ) ); |
| | 531 | + $foundpages = false; |
| | 532 | + |
| | 533 | + while ( $row = $db->fetchObject( $res ) ) { |
| | 534 | + $foundpages = true; |
| | 535 | + $st = new SMWSmallTitle(); |
| | 536 | + $st->dbkey = $row->page_title; |
| | 537 | + $st->namespace = $row->page_namespace; |
| | 538 | + $st->recdepth = 0; |
| | 539 | + $this->serializeSmallTitle( $st ); |
| | 540 | + $this->flush(); |
| | 541 | + $linkCache->clear(); |
| | 542 | + } |
| | 543 | + if ( $foundpages ) { // add link to next result page |
| | 544 | + if ( strpos( SMWExporter::expandURI( '&wikiurl;' ), '?' ) === false ) { // check whether we have title as a first parameter or in URL |
| | 545 | + $nexturl = SMWExporter::expandURI( '&export;?offset=' ) . ( $offset + $limit ); |
| | 546 | + } else { |
| | 547 | + $nexturl = SMWExporter::expandURI( '&export;&offset=' ) . ( $offset + $limit ); |
| | 548 | + } |
| | 549 | + |
| | 550 | + $data = new SMWExpData( new SMWExpResource( $nexturl ) ); |
| | 551 | + $ed = new SMWExpData( SMWExporter::getSpecialElement( 'owl', 'Thing' ) ); |
| | 552 | + $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'rdf', 'type' ), $ed ); |
| | 553 | + $ed = new SMWExpData( new SMWExpResource( $nexturl ) ); |
| | 554 | + $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'rdfs', 'isDefinedBy' ), $ed ); |
| | 555 | + $this->serializer->serializeExpData( $data ); |
| | 556 | + } |
| | 557 | + |
| | 558 | + $this->serializer->serializeFooter(); |
| | 559 | + $this->flush( true ); |
| | 560 | + |
| | 561 | + wfProfileOut( "RDF::PrintPageList" ); |
| | 562 | + } |
| | 563 | + |
| | 564 | + |
| | 565 | + /** |
| | 566 | + * Print basic information about this site. |
| | 567 | + */ |
| | 568 | + public function printWikiInfo() { |
| | 569 | + wfProfileIn( "RDF::PrintWikiInfo" ); |
| | 570 | + |
| | 571 | + global $wgSitename, $wgLanguageCode; |
| | 572 | + |
| | 573 | + $db = & wfGetDB( DB_SLAVE ); |
| | 574 | + $this->prepareSerialization(); |
| | 575 | + $this->delay_flush = 35; // don't do intermediate flushes with default parameters |
| | 576 | + $linkCache = LinkCache::singleton(); |
| | 577 | + |
| | 578 | + // assemble export data: |
| | 579 | + $data = new SMWExpData( new SMWExpResource( '&wiki;#wiki' ) ); |
| | 580 | + $ed = new SMWExpData( SMWExporter::getSpecialElement( 'swivt', 'Wikisite' ) ); |
| | 581 | + $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'rdf', 'type' ), $ed ); |
| | 582 | + // basic wiki information |
| | 583 | + $ed = new SMWExpData( new SMWExpLiteral( $wgSitename ) ); |
| | 584 | + $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'rdfs', 'label' ), $ed ); |
| | 585 | + $ed = new SMWExpData( new SMWExpLiteral( $wgSitename, null, 'http://www.w3.org/2001/XMLSchema#string' ) ); |
| | 586 | + $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'siteName' ), $ed ); |
| | 587 | + $ed = new SMWExpData( new SMWExpLiteral( SMWExporter::expandURI( '&wikiurl;' ), null, 'http://www.w3.org/2001/XMLSchema#string' ) ); |
| | 588 | + $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'pagePrefix' ), $ed ); |
| | 589 | + $ed = new SMWExpData( new SMWExpLiteral( SMW_VERSION, null, 'http://www.w3.org/2001/XMLSchema#string' ) ); |
| | 590 | + $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'smwVersion' ), $ed ); |
| | 591 | + $ed = new SMWExpData( new SMWExpLiteral( $wgLanguageCode, null, 'http://www.w3.org/2001/XMLSchema#string' ) ); |
| | 592 | + $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'langCode' ), $ed ); |
| | 593 | + $mainpage = Title::newMainPage(); |
| | 594 | + if ( $mainpage !== null ) { |
| | 595 | + $ed = new SMWExpData( new SMWExpResource( $mainpage->getFullURL() ) ); |
| | 596 | + $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'mainPage' ), $ed ); |
| | 597 | + } |
| | 598 | + // statistical information |
| | 599 | + $ed = new SMWExpData( new SMWExpLiteral( SiteStats::pages(), null, 'http://www.w3.org/2001/XMLSchema#int' ) ); |
| | 600 | + $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'pageCount' ), $ed ); |
| | 601 | + $ed = new SMWExpData( new SMWExpLiteral( SiteStats::articles(), null, 'http://www.w3.org/2001/XMLSchema#int' ) ); |
| | 602 | + $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'contentPageCount' ), $ed ); |
| | 603 | + $ed = new SMWExpData( new SMWExpLiteral( SiteStats::images(), null, 'http://www.w3.org/2001/XMLSchema#int' ) ); |
| | 604 | + $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'mediaCount' ), $ed ); |
| | 605 | + $ed = new SMWExpData( new SMWExpLiteral( SiteStats::edits(), null, 'http://www.w3.org/2001/XMLSchema#int' ) ); |
| | 606 | + $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'editCount' ), $ed ); |
| | 607 | + $ed = new SMWExpData( new SMWExpLiteral( SiteStats::views(), null, 'http://www.w3.org/2001/XMLSchema#int' ) ); |
| | 608 | + $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'viewCount' ), $ed ); |
| | 609 | + $ed = new SMWExpData( new SMWExpLiteral( SiteStats::users(), null, 'http://www.w3.org/2001/XMLSchema#int' ) ); |
| | 610 | + $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'userCount' ), $ed ); |
| | 611 | + $ed = new SMWExpData( new SMWExpLiteral( SiteStats::admins(), null, 'http://www.w3.org/2001/XMLSchema#int' ) ); |
| | 612 | + $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'swivt', 'adminCount' ), $ed ); |
| | 613 | + |
| | 614 | + $this->serializer->serializeHeader(); |
| | 615 | + $this->serializer->serializeExpData( $this->getOntologyExpData( '' ) ); |
| | 616 | + $this->serializer->serializeExpData( $data ); |
| | 617 | + |
| | 618 | + // link to list of existing pages: |
| | 619 | + if ( strpos( SMWExporter::expandURI( '&wikiurl;' ), '?' ) === false ) { // check whether we have title as a first parameter or in URL |
| | 620 | + $nexturl = SMWExporter::expandURI( '&export;?offset=0' ); |
| | 621 | + } else { |
| | 622 | + $nexturl = SMWExporter::expandURI( '&export;&offset=0' ); |
| | 623 | + } |
| | 624 | + $data = new SMWExpData( new SMWExpResource( $nexturl ) ); |
| | 625 | + $ed = new SMWExpData( SMWExporter::getSpecialElement( 'owl', 'Thing' ) ); |
| | 626 | + $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'rdf', 'type' ), $ed ); |
| | 627 | + $ed = new SMWExpData( new SMWExpResource( $nexturl ) ); |
| | 628 | + $data->addPropertyObjectValue( SMWExporter::getSpecialElement( 'rdfs', 'isDefinedBy' ), $ed ); |
| | 629 | + $this->serializer->serializeExpData( $data ); |
| | 630 | + |
| | 631 | + $this->serializer->serializeFooter(); |
| | 632 | + $this->flush( true ); |
| | 633 | + |
| | 634 | + wfProfileOut( "RDF::PrintWikiInfo" ); |
| | 635 | + } |
| | 636 | + |
| | 637 | + /** |
| | 638 | + * This function checks whether some article fits into a given namespace |
| | 639 | + * restriction. Restrictions are encoded as follows: a non-negative number |
| | 640 | + * requires the namespace to be identical to the given number; "-1" |
| | 641 | + * requires the namespace to be different from Category, Property, and |
| | 642 | + * Type; "false" means "no restriction". |
| | 643 | + * @param $res mixed encoding the restriction as described above |
| | 644 | + * @param $ns integer the namespace constant to be checked |
| | 645 | + */ |
| | 646 | + static public function fitsNsRestriction( $res, $ns ) { |
| | 647 | + if ( $res === false ) return true; |
| | 648 | + if ( is_array( $res ) ) return in_array( $ns, $res ); |
| | 649 | + if ( $res >= 0 ) return ( $res == $ns ); |
| | 650 | + return ( ( $res != NS_CATEGORY ) && ( $res != SMW_NS_PROPERTY ) && ( $res != SMW_NS_TYPE ) ); |
| | 651 | + } |
| | 652 | + |
| | 653 | +} |
| Property changes on: trunk/extensions/SemanticMediaWiki/includes/export/SMW_ExportController.php |
| ___________________________________________________________________ |
| Added: svn:eol-style |
| 1 | 654 | + native |
| Index: trunk/extensions/SemanticMediaWiki/includes/export/SMW_Exporter.php |
| — | — | @@ -149,7 +149,7 @@ |
| 150 | 150 | if ( $pe !== null ) { |
| 151 | 151 | foreach ( $semdata->getPropertyValues( $property ) as $dv ) { |
| 152 | 152 | if ( $cat_only ) { |
| 153 | | - if ( !( $dv instanceof SMWWikiPageValue ) || ( $dv->etNamespace != NS_CATEGORY ) ) { |
| | 153 | + if ( !( $dv instanceof SMWWikiPageValue ) || ( $dv->getNamespace() != NS_CATEGORY ) ) { |
| 154 | 154 | continue; |
| 155 | 155 | } |
| 156 | 156 | } |
| Index: trunk/extensions/SemanticMediaWiki/includes/export/SMW_Serializer.php |
| — | — | @@ -0,0 +1,419 @@ |
| | 2 | +<?php |
| | 3 | + |
| | 4 | +/** |
| | 5 | + * File holding the SMWSerializer class that provides basic functions for |
| | 6 | + * serialising data in OWL and RDF syntaxes. |
| | 7 | + * |
| | 8 | + * @file SMW_Serializer.php |
| | 9 | + * @ingroup SMW |
| | 10 | + * |
| | 11 | + * @author Markus Krötzsch |
| | 12 | + */ |
| | 13 | + |
| | 14 | +define( 'SMW_SERIALIZER_DECL_CLASS', 1 ); |
| | 15 | +define( 'SMW_SERIALIZER_DECL_OPROP', 2 ); |
| | 16 | +define( 'SMW_SERIALIZER_DECL_APROP', 4 ); |
| | 17 | + |
| | 18 | +/** |
| | 19 | + * Class for serializing exported data (encoded as SMWExpData object) in a |
| | 20 | + * concrete syntactic format such as Turtle or RDF/XML. The serializer |
| | 21 | + * adds object serialisations to an internal string that can be retrieved for |
| | 22 | + * pushing it to an output. RDF and OWL have two types of dependencies that are |
| | 23 | + * managed: namespaces (and similar abbreviations) and element declarations. |
| | 24 | + * The former need to be defined before being used, while the latter can occur |
| | 25 | + * at some later point in the serialization. Declarations are relevant to the |
| | 26 | + * OWL data model, being one of Class, DatatypeProperty, and ObjectProperty |
| | 27 | + * (only the latter two are mutually exclusive). This class determines the |
| | 28 | + * required declaration from the context in which an element is used. |
| | 29 | + * |
| | 30 | + * @ingroup SMW |
| | 31 | + */ |
| | 32 | +class SMWSerializer { |
| | 33 | + /** |
| | 34 | + * Array for recording required declarations; format: |
| | 35 | + * resourcename => decl-flag, where decl-flag is a sum of flags |
| | 36 | + * SMW_SERIALIZER_DECL_CLASS, SMW_SERIALIZER_DECL_OPROP, |
| | 37 | + * SMW_SERIALIZER_DECL_APROP. |
| | 38 | + */ |
| | 39 | + protected $decl_todo; |
| | 40 | + /** |
| | 41 | + * Array for recording previous declarations; format: |
| | 42 | + * resourcename => decl-flag, where decl-flag is a sum of flags |
| | 43 | + * SMW_SERIALIZER_DECL_CLASS, SMW_SERIALIZER_DECL_OPROP, |
| | 44 | + * SMW_SERIALIZER_DECL_APROP. |
| | 45 | + */ |
| | 46 | + protected $decl_done; |
| | 47 | + /** |
| | 48 | + * Array of additional namespaces (abbreviation => URI), flushed on |
| | 49 | + * closing the current namespace tag. Since we export in a streamed |
| | 50 | + * way, it is not always possible to embed additional namespaces into |
| | 51 | + * a syntactic block (e.g. an RDF/XML tag) which might have been sent to |
| | 52 | + * the client already. But we wait with printing the current block so that |
| | 53 | + * extra namespaces from this array can still be printed (note that one |
| | 54 | + * never know which extra namespaces you encounter during export). |
| | 55 | + */ |
| | 56 | + protected $extra_namespaces; |
| | 57 | + /** |
| | 58 | + * Array of namespaces that have been declared globally already. Contains |
| | 59 | + * entries of format 'namespace abbreviation' => true, assuming that the |
| | 60 | + * same abbreviation always refers to the same URI. |
| | 61 | + */ |
| | 62 | + protected $global_namespaces; |
| | 63 | + /** |
| | 64 | + * The current working string is obtained by concatenating the strings |
| | 65 | + * $pre_ns_buffer and $post_ns_buffer. The split between the two is such |
| | 66 | + * that one can append additional namespace declarations to $pre_ns_buffer |
| | 67 | + * so that they affect all current elements. The buffers are flushed during |
| | 68 | + * output in order to achieve "streaming" RDF export for larger files. |
| | 69 | + */ |
| | 70 | + protected $pre_ns_buffer; |
| | 71 | + /** |
| | 72 | + * See documentation for $pre_ns_buffer. |
| | 73 | + */ |
| | 74 | + protected $post_ns_buffer; |
| | 75 | + /** |
| | 76 | + * True if the $pre_ns_buffer contains the beginning of a namespace |
| | 77 | + * declaration block to which further declarations for the current |
| | 78 | + * context can be appended. |
| | 79 | + */ |
| | 80 | + protected $namespace_block_started; |
| | 81 | + /** |
| | 82 | + * True if the namespaces that are added at the current serialization stage |
| | 83 | + * become global, i.e. remain available for all later contexts. This is the |
| | 84 | + * case in RDF/XML only as long as the header has not been streamed to the |
| | 85 | + * client (reflected herein by calling flushContent()). Later, namespaces |
| | 86 | + * can only be added locally to individual elements, thus requiring them to |
| | 87 | + * be re-added multiple times if used in many elements. |
| | 88 | + */ |
| | 89 | + protected $namespaces_are_global; |
| | 90 | + |
| | 91 | + /** |
| | 92 | + * Constructor. |
| | 93 | + */ |
| | 94 | + public function __construct() { |
| | 95 | + $this->clear(); |
| | 96 | + } |
| | 97 | + |
| | 98 | + /** |
| | 99 | + * Clear internal states to start a new serialization. |
| | 100 | + */ |
| | 101 | + public function clear() { |
| | 102 | + $this->element_queue = array(); |
| | 103 | + $this->element_done = array(); |
| | 104 | + $this->decl_todo = array(); |
| | 105 | + $this->decl_done = array(); |
| | 106 | + $this->pre_ns_buffer = ''; |
| | 107 | + $this->post_ns_buffer = ''; |
| | 108 | + $this->namespaces_are_global = false; |
| | 109 | + $this->extra_namespaces = array(); |
| | 110 | + } |
| | 111 | + |
| | 112 | + /* Functions for exporting RDF */ |
| | 113 | + |
| | 114 | + public function serializeHeader() { |
| | 115 | + $this->clear(); |
| | 116 | + $this->namespaces_are_global = true; |
| | 117 | + $this->namespace_block_started = true; |
| | 118 | + $this->pre_ns_buffer = |
| | 119 | + "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n" . |
| | 120 | + "<!DOCTYPE rdf:RDF[\n" . |
| | 121 | + "\t<!ENTITY rdf " . $this->makeValueEntityString( SMWExporter::expandURI( '&rdf;' ) ) . ">\n" . |
| | 122 | + "\t<!ENTITY rdfs " . $this->makeValueEntityString( SMWExporter::expandURI( '&rdfs;' ) ) . ">\n" . |
| | 123 | + "\t<!ENTITY owl " . $this->makeValueEntityString( SMWExporter::expandURI( '&owl;' ) ) . ">\n" . |
| | 124 | + "\t<!ENTITY swivt " . $this->makeValueEntityString( SMWExporter::expandURI( '&swivt;' ) ) . ">\n" . |
| | 125 | + // A note on "wiki": this namespace is crucial as a fallback when it would be illegal to start e.g. with a number. |
| | 126 | + // In this case, one can always use wiki:... followed by "_" and possibly some namespace, since _ is legal as a first character. |
| | 127 | + "\t<!ENTITY wiki " . $this->makeValueEntityString( SMWExporter::expandURI( '&wiki;' ) ) . ">\n" . |
| | 128 | + "\t<!ENTITY property " . $this->makeValueEntityString( SMWExporter::expandURI( '&property;' ) ) . ">\n" . |
| | 129 | + "\t<!ENTITY wikiurl " . $this->makeValueEntityString( SMWExporter::expandURI( '&wikiurl;' ) ) . ">\n" . |
| | 130 | + "]>\n\n" . |
| | 131 | + "<rdf:RDF\n" . |
| | 132 | + "\txmlns:rdf=\"&rdf;\"\n" . |
| | 133 | + "\txmlns:rdfs=\"&rdfs;\"\n" . |
| | 134 | + "\txmlns:owl =\"&owl;\"\n" . |
| | 135 | + "\txmlns:swivt=\"&swivt;\"\n" . |
| | 136 | + "\txmlns:wiki=\"&wiki;\"\n" . |
| | 137 | + "\txmlns:property=\"&property;\""; |
| | 138 | + $this->global_namespaces = array( 'rdf' => true, 'rdfs' => true, 'owl' => true, 'swivt' => true, 'wiki' => true, 'property' => true ); |
| | 139 | + $this->post_ns_buffer .= ">\n\n"; |
| | 140 | + } |
| | 141 | + |
| | 142 | + /** |
| | 143 | + * Serialise the footer. |
| | 144 | + */ |
| | 145 | + public function serializeFooter() { |
| | 146 | + $this->serializeDeclarations(); |
| | 147 | + $this->post_ns_buffer .= "\t<!-- Created by Semantic MediaWiki, http://semantic-mediawiki.org/ -->\n"; |
| | 148 | + $this->post_ns_buffer .= '</rdf:RDF>'; |
| | 149 | + } |
| | 150 | + |
| | 151 | + /** |
| | 152 | + * Serialize any declarations that have been found to be missing while |
| | 153 | + * serializing other elements. |
| | 154 | + */ |
| | 155 | + public function serializeDeclarations() { |
| | 156 | + foreach ( $this->decl_todo as $name => $flag ) { |
| | 157 | + $types = array(); |
| | 158 | + if ( $flag & SMW_SERIALIZER_DECL_CLASS ) $types[] = 'owl:Class'; |
| | 159 | + if ( $flag & SMW_SERIALIZER_DECL_OPROP ) $types[] = 'owl:ObjectProperty'; |
| | 160 | + if ( $flag & SMW_SERIALIZER_DECL_APROP ) $types[] = 'owl:DatatypeProperty'; |
| | 161 | + foreach ( $types as $type ) { |
| | 162 | + $this->post_ns_buffer .= "\t<$type rdf:about=\"$name\" />\n"; |
| | 163 | + } |
| | 164 | + $curdone = array_key_exists( $name, $this->decl_done ) ? $this->decl_done[$name] : 0; |
| | 165 | + $this->decl_done[$name] = $curdone | $flag; |
| | 166 | + } |
| | 167 | + $this->decl_todo = array(); // reset all |
| | 168 | + } |
| | 169 | + |
| | 170 | + /** |
| | 171 | + * Serialise the given SMWExpData object. The method does not assume that |
| | 172 | + * the exported data refers to wiki pages or other SMW data, and it makes |
| | 173 | + * sure that all required auxiliary declarations for obtaining proper OWL |
| | 174 | + * are included anyway. |
| | 175 | + * |
| | 176 | + * @param $data SMWExpData containing the data to be serialised. |
| | 177 | + */ |
| | 178 | + public function serializeExpData( SMWExpData $data ) { |
| | 179 | + $this->serializeNestedExpData( $data, '' ); |
| | 180 | + $this->serializeNamespaces(); |
| | 181 | + if ( !$this->namespaces_are_global ) { |
| | 182 | + $this->pre_ns_buffer .= $this->post_ns_buffer; |
| | 183 | + $this->post_ns_buffer = ''; |
| | 184 | + $this->namespace_block_started = false; |
| | 185 | + } |
| | 186 | + } |
| | 187 | + |
| | 188 | + /** |
| | 189 | + * Serialise the given SMWExpData object, possibly recursively with |
| | 190 | + * increased indentation. |
| | 191 | + * |
| | 192 | + * @param $data SMWExpData containing the data to be serialised. |
| | 193 | + * @param $indent string specifying a prefix for indentation (usually a sequence of tabs) |
| | 194 | + */ |
| | 195 | + protected function serializeNestedExpData( SMWExpData $data, $indent ) { |
| | 196 | + $this->recordDeclarationTypes( $data ); |
| | 197 | + |
| | 198 | + $type = $data->extractMainType()->getQName(); |
| | 199 | + if ( !$this->namespace_block_started ) { // start new ns block |
| | 200 | + $this->pre_ns_buffer .= "\t$indent<$type"; |
| | 201 | + $this->namespace_block_started = true; |
| | 202 | + } else { // continue running block |
| | 203 | + $this->post_ns_buffer .= "\t$indent<$type"; |
| | 204 | + } |
| | 205 | + |
| | 206 | + if ( ( $data->getSubject() instanceof SMWExpLiteral ) || |
| | 207 | + ( $data->getSubject() instanceof SMWExpResource ) ) { |
| | 208 | + $this->post_ns_buffer .= ' rdf:about="' . $data->getSubject()->getName() . '"'; |
| | 209 | + } // else: blank node, no "rdf:about" |
| | 210 | + |
| | 211 | + if ( count( $data->getProperties() ) == 0 ) { // nothing else to export |
| | 212 | + $this->post_ns_buffer .= " />\n"; |
| | 213 | + } else { // process data |
| | 214 | + $this->post_ns_buffer .= ">\n"; |
| | 215 | + |
| | 216 | + foreach ( $data->getProperties() as $property ) { |
| | 217 | + $prop_decl_queued = false; |
| | 218 | + $prop_decl_type = 0; |
| | 219 | + $class_type_prop = $this->isOWLClassTypeProperty( $property ); |
| | 220 | + |
| | 221 | + foreach ( $data->getValues( $property ) as $value ) { |
| | 222 | + $this->post_ns_buffer .= "\t\t$indent<" . $property->getQName(); |
| | 223 | + $this->requireNamespace( $property->getNamespaceID(), $property->getNamespace() ); |
| | 224 | + $object = $value->getSubject(); |
| | 225 | + |
| | 226 | + if ( $object instanceof SMWExpLiteral ) { |
| | 227 | + $prop_decl_type = SMW_SERIALIZER_DECL_APROP; |
| | 228 | + if ( $object->getDatatype() != '' ) { |
| | 229 | + $this->post_ns_buffer .= ' rdf:datatype="' . $object->getDatatype() . '"'; |
| | 230 | + } |
| | 231 | + $this->post_ns_buffer .= '>' . |
| | 232 | + str_replace( array( '&', '>', '<' ), array( '&', '>', '<' ), $object->getName() ) . |
| | 233 | + '</' . $property->getQName() . ">\n"; |
| | 234 | + } else { // resource (maybe blank node), could have subdescriptions |
| | 235 | + $prop_decl_type = SMW_SERIALIZER_DECL_OPROP; |
| | 236 | + $collection = $value->getCollection(); |
| | 237 | + if ( $collection !== false ) { // RDF-style collection (list) |
| | 238 | + $this->post_ns_buffer .= " rdf:parseType=\"Collection\">\n"; |
| | 239 | + foreach ( $collection as $subvalue ) { |
| | 240 | + $this->serializeNestedExpData( $subvalue, $indent . "\t\t" ); |
| | 241 | + if ( $class_type_prop ) { |
| | 242 | + $this->requireDeclaration( $subvalue, SMW_SERIALIZER_DECL_CLASS ); |
| | 243 | + } |
| | 244 | + } |
| | 245 | + $this->post_ns_buffer .= "\t\t$indent</" . $property->getQName() . ">\n"; |
| | 246 | + } else { |
| | 247 | + if ( $class_type_prop ) { |
| | 248 | + $this->requireDeclaration( $object, SMW_SERIALIZER_DECL_CLASS ); |
| | 249 | + } |
| | 250 | + if ( count( $value->getProperties() ) > 0 ) { // resource with data: serialise |
| | 251 | + $this->post_ns_buffer .= ">\n"; |
| | 252 | + $this->serializeNestedExpData( $value, $indent . "\t\t" ); |
| | 253 | + $this->post_ns_buffer .= "\t\t$indent</" . $property->getQName() . ">\n"; |
| | 254 | + } else { // resource without data: may need to be queued |
| | 255 | + if ( !$object->isBlankNode() ) { |
| | 256 | + $this->post_ns_buffer .= ' rdf:resource="' . $object->getName() . '"'; |
| | 257 | + } |
| | 258 | + $this->post_ns_buffer .= "/>\n"; |
| | 259 | + } |
| | 260 | + } |
| | 261 | + } |
| | 262 | + if ( !$prop_decl_queued ) { |
| | 263 | + $this->requireDeclaration( $property, $prop_decl_type ); |
| | 264 | + $prop_decl_queued = true; |
| | 265 | + } |
| | 266 | + } |
| | 267 | + } |
| | 268 | + $this->post_ns_buffer .= "\t$indent</" . $type . ">\n"; |
| | 269 | + } |
| | 270 | + } |
| | 271 | + |
| | 272 | + /** |
| | 273 | + * Get the string that has been serialized so far. This function also |
| | 274 | + * resets the internal buffers for serilized strings and namespaces |
| | 275 | + * (what is flushed is gone). |
| | 276 | + */ |
| | 277 | + public function flushContent() { |
| | 278 | + if ( ( $this->pre_ns_buffer == '' ) && ( $this->post_ns_buffer == '' ) ) return ''; |
| | 279 | + $this->serializeNamespaces(); |
| | 280 | + $this->namespaces_are_global = false; |
| | 281 | + $result = $this->pre_ns_buffer . $this->post_ns_buffer; |
| | 282 | + $this->pre_ns_buffer = ''; |
| | 283 | + $this->post_ns_buffer = ''; |
| | 284 | + $this->namespace_block_started = false; |
| | 285 | + return $result; |
| | 286 | + } |
| | 287 | + |
| | 288 | + /** |
| | 289 | + * Require an additional namespace to be declared in the serialization. |
| | 290 | + * The function checks whether the required namespace is available globally |
| | 291 | + * and add it to the list of required namespaces otherwise. |
| | 292 | + */ |
| | 293 | + protected function requireNamespace( $nsshort, $nsuri ) { |
| | 294 | + if ( !array_key_exists( $nsshort, $this->global_namespaces ) ) { |
| | 295 | + $this->extra_namespaces[$nsshort] = $nsuri; |
| | 296 | + } |
| | 297 | + } |
| | 298 | + |
| | 299 | + /** |
| | 300 | + * Include collected namespace information into the serialization. |
| | 301 | + */ |
| | 302 | + protected function serializeNamespaces() { |
| | 303 | + foreach ( $this->extra_namespaces as $nsshort => $nsuri ) { |
| | 304 | + if ( $this->namespaces_are_global ) { |
| | 305 | + $this->global_namespaces[$nsshort] = true; |
| | 306 | + $this->pre_ns_buffer .= "\n\t"; |
| | 307 | + } else { |
| | 308 | + $this->pre_ns_buffer .= ' '; |
| | 309 | + } |
| | 310 | + $this->pre_ns_buffer .= "xmlns:$nsshort=\"$nsuri\""; |
| | 311 | + } |
| | 312 | + $this->extra_namespaces = array(); |
| | 313 | + } |
| | 314 | + |
| | 315 | + /** |
| | 316 | + * State that a certain declaration is needed. The method checks if the |
| | 317 | + * declaration is already available, and records a todo otherwise. |
| | 318 | + */ |
| | 319 | + protected function requireDeclaration( SMWExpResource $resource, $decltype ) { |
| | 320 | + $namespaceid = $resource->getNamespaceID(); |
| | 321 | + // Do not declare predefined OWL language constructs: |
| | 322 | + if ( ( $namespaceid == 'owl' ) || ( $namespaceid == 'rdf' ) || ( $namespaceid == 'rdfs' ) ) return; |
| | 323 | + // Do not declare blank nodes: |
| | 324 | + if ( $resource->isBlankNode() ) return; |
| | 325 | + |
| | 326 | + $name = $resource->getName(); |
| | 327 | + if ( array_key_exists( $name, $this->decl_done ) && ( $this->decl_done[$name] & $decltype ) ) { |
| | 328 | + return; |
| | 329 | + } |
| | 330 | + if ( !array_key_exists( $name, $this->decl_todo ) ) { |
| | 331 | + $this->decl_todo[$name] = $decltype; |
| | 332 | + } else { |
| | 333 | + $this->decl_todo[$name] = $this->decl_todo[$name] | $decltype; |
| | 334 | + } |
| | 335 | + } |
| | 336 | + |
| | 337 | + /** |
| | 338 | + * Update the declaration "todo" and "done" lists for the case that the |
| | 339 | + * given data has been serialized with the type information it provides. |
| | 340 | + * |
| | 341 | + * @param $data specifying the type data upon which declarations are based |
| | 342 | + */ |
| | 343 | + protected function recordDeclarationTypes( SMWExpData $data ) { |
| | 344 | + foreach ( $data->getSpecialValues( 'rdf', 'type') as $typedata ) { |
| | 345 | + $typeresource = $typedata->getSubject(); |
| | 346 | + if ( $typeresource instanceof SMWExpResource ) { |
| | 347 | + switch ( $typeresource->getQName() ) { |
| | 348 | + case 'owl:Class': $typeflag = SMW_SERIALIZER_DECL_CLASS; break; |
| | 349 | + case 'owl:ObjectProperty': $typeflag = SMW_SERIALIZER_DECL_OPROP; break; |
| | 350 | + case 'owl:DatatypeProperty': $typeflag = SMW_SERIALIZER_DECL_APROP; break; |
| | 351 | + default: $typeflag = 0; |
| | 352 | + } |
| | 353 | + if ( $typeflag != 0 ) { |
| | 354 | + $this->declarationDone( $data->getSubject(), $typeflag ); |
| | 355 | + } |
| | 356 | + } |
| | 357 | + } |
| | 358 | + } |
| | 359 | + |
| | 360 | + /** |
| | 361 | + * Update the declaration "todo" and "done" lists to reflect the fact that |
| | 362 | + * the given element has been declared to has the given type. |
| | 363 | + * |
| | 364 | + * @param $element SMWExpResource specifying the element to update |
| | 365 | + * @param $typeflag integer specifying the type (e.g. SMW_SERIALIZER_DECL_CLASS) |
| | 366 | + */ |
| | 367 | + protected function declarationDone( SMWExpResource $element, $typeflag ) { |
| | 368 | + $name = $element->getName(); |
| | 369 | + $curdone = array_key_exists( $name, $this->decl_done ) ? $this->decl_done[$name] : 0; |
| | 370 | + $this->decl_done[$name] = $curdone | $typeflag; |
| | 371 | + if ( array_key_exists( $name, $this->decl_todo ) ) { |
| | 372 | + $this->decl_todo[$name] = $this->decl_todo[$name] & ( ~$typeflag ); |
| | 373 | + if ( $this->decl_todo[$name] == 0 ) { |
| | 374 | + unset( $this->decl_todo[$name] ); |
| | 375 | + } |
| | 376 | + } |
| | 377 | + } |
| | 378 | + |
| | 379 | + /** |
| | 380 | + * Check if the given property is one of the special properties of the OWL |
| | 381 | + * language that require their values to be classes or RDF lists of |
| | 382 | + * classes. In these cases, it is necessary to declare this in the exported |
| | 383 | + * data. |
| | 384 | + * |
| | 385 | + * @note The list of properties checked here is not complete for the OWL |
| | 386 | + * language but covers what is used in SMW. |
| | 387 | + * @note OWL 2 allows URIs to refer to both classes and individual elements |
| | 388 | + * in different contexts. We only need declarations for classes that are |
| | 389 | + * used as such, hence it is enough to check the property. Moreover, we do |
| | 390 | + * not use OWL Datatypes in SMW, so rdf:type, rdfs:domain, etc. always |
| | 391 | + * refer to classes. |
| | 392 | + * @param SMWExpResource $property |
| | 393 | + */ |
| | 394 | + protected function isOWLClassTypeProperty( SMWExpResource $property ) { |
| | 395 | + $locname = $property->getLocalName(); |
| | 396 | + if ( $property->getNamespaceID() == 'rdf' ) { |
| | 397 | + return ( $locname == 'type' ); |
| | 398 | + } elseif ( $property->getNamespaceID() == 'owl' ) { |
| | 399 | + return ( $locname == 'intersectionOf' ) || ( $locname == 'unionOf' ) || |
| | 400 | + ( $locname == 'equivalentClass' ) || |
| | 401 | + ( $locname == 'complementOf' ) || ( $locname == 'someValuesFrom' ) || |
| | 402 | + ( $locname == 'allValuesFrom' ) || ( $locname == 'onClass' ); |
| | 403 | + } elseif ( $property->getNamespaceID() == 'rdfs' ) { |
| | 404 | + return ( $locname == 'subClassOf' ) || ( $locname == 'range' ) || ( $locname == 'domain' ); |
| | 405 | + } else { |
| | 406 | + return false; |
| | 407 | + } |
| | 408 | + } |
| | 409 | + |
| | 410 | + /** |
| | 411 | + * Escape a string in the special form that is required for values in |
| | 412 | + * DTD entity declarations in XML. Namely, this require the percent sign |
| | 413 | + * to be replaced. |
| | 414 | + * @param string $string to be escaped |
| | 415 | + */ |
| | 416 | + protected function makeValueEntityString( $string ) { |
| | 417 | + return "'" . str_replace( '%','%',$string ) . "'"; |
| | 418 | + } |
| | 419 | + |
| | 420 | +} |
| Property changes on: trunk/extensions/SemanticMediaWiki/includes/export/SMW_Serializer.php |
| ___________________________________________________________________ |
| Added: svn:eol-style |
| 1 | 421 | + native |
| Index: trunk/extensions/SemanticMediaWiki/includes/datavalues/SMW_DV_Record.php |
| — | — | @@ -151,7 +151,7 @@ |
| 152 | 152 | public function getExportData() { |
| 153 | 153 | if ( !$this->isValid() ) return null; |
| 154 | 154 | |
| 155 | | - $result = new SMWExpData( new SMWExpElement( '', $this ) ); // bnode |
| | 155 | + $result = new SMWExpData( new SMWExpResource( '', $this ) ); // bnode |
| 156 | 156 | $ed = new SMWExpData( SMWExporter::getSpecialElement( 'swivt', 'Container' ) ); |
| 157 | 157 | $result->addPropertyObjectValue( SMWExporter::getSpecialElement( 'rdf', 'type' ), $ed ); |
| 158 | 158 | $count = 0; |
| Index: trunk/extensions/SemanticMediaWiki/includes/datavalues/SMW_DV_Concept.php |
| — | — | @@ -102,7 +102,7 @@ |
| 103 | 103 | $element = new SMWExpData( SMWExporter::getSpecialElement( 'owl', 'Thing' ) ); |
| 104 | 104 | } |
| 105 | 105 | if ( !$exact ) { |
| 106 | | - $result = new SMWExpData( new SMWExpElement( '' ) ); |
| | 106 | + $result = new SMWExpData( new SMWExpResource( '' ) ); |
| 107 | 107 | $result->addPropertyObjectValue( SMWExporter::getSpecialElement( 'rdf', 'type' ), |
| 108 | 108 | new SMWExpData( SMWExporter::getSpecialElement( 'owl', 'Class' ) ) ); |
| 109 | 109 | $result->addPropertyObjectValue( SMWExporter::getSpecialElement( 'rdfs', 'subClassOf' ), $owldesc ); |
| — | — | @@ -117,7 +117,7 @@ |
| 118 | 118 | |
| 119 | 119 | public function descriptionToExpData( $desc, &$exact ) { |
| 120 | 120 | if ( ( $desc instanceof SMWConjunction ) || ( $desc instanceof SMWDisjunction ) ) { |
| 121 | | - $result = new SMWExpData( new SMWExpElement( '' ) ); |
| | 121 | + $result = new SMWExpData( new SMWExpResource( '' ) ); |
| 122 | 122 | $result->addPropertyObjectValue( SMWExporter::getSpecialElement( 'rdf', 'type' ), |
| 123 | 123 | new SMWExpData( SMWExporter::getSpecialElement( 'owl', 'Class' ) ) ); |
| 124 | 124 | $elements = array(); |
| — | — | @@ -135,7 +135,7 @@ |
| 136 | 136 | if ( count( $desc->getCategories() ) == 1 ) { // single category |
| 137 | 137 | $result = new SMWExpData( SMWExporter::getResourceElement( end( $desc->getCategories() ) ) ); |
| 138 | 138 | } else { // disjunction of categories |
| 139 | | - $result = new SMWExpData( new SMWExpElement( '' ) ); |
| | 139 | + $result = new SMWExpData( new SMWExpResource( '' ) ); |
| 140 | 140 | $elements = array(); |
| 141 | 141 | foreach ( $desc->getCategories() as $cat ) { |
| 142 | 142 | $elements[] = new SMWExpData( SMWExporter::getResourceElement( $cat ) ); ; |
| — | — | @@ -148,7 +148,7 @@ |
| 149 | 149 | } elseif ( $desc instanceof SMWConceptDescription ) { |
| 150 | 150 | $result = new SMWExpData( SMWExporter::getResourceElement( $desc->getConcept() ) ); |
| 151 | 151 | } elseif ( $desc instanceof SMWSomeProperty ) { |
| 152 | | - $result = new SMWExpData( new SMWExpElement( '' ) ); |
| | 152 | + $result = new SMWExpData( new SMWExpResource( '' ) ); |
| 153 | 153 | $result->addPropertyObjectValue( SMWExporter::getSpecialElement( 'rdf', 'type' ), |
| 154 | 154 | new SMWExpData( SMWExporter::getSpecialElement( 'owl', 'Restriction' ) ) ); |
| 155 | 155 | $result->addPropertyObjectValue( SMWExporter::getSpecialElement( 'owl', 'onProperty' ), |
| Index: trunk/extensions/SemanticMediaWiki/includes/SMW_SetupLight.php |
| — | — | @@ -130,11 +130,14 @@ |
| 131 | 131 | |
| 132 | 132 | |
| 133 | 133 | // Export |
| 134 | | -// $wgAutoloadClasses['SMWExporter'] = $smwgIP . 'includes/export/SMW_Exporter.php'; |
| 135 | | -// $wgAutoloadClasses['SMWExpData'] = $smwgIP . 'includes/export/SMW_Exp_Data.php'; |
| 136 | | -// $wgAutoloadClasses['SMWExpElement'] = $smwgIP . 'includes/export/SMW_Exp_Element.php'; |
| 137 | | -// $wgAutoloadClasses['SMWExpLiteral'] = $smwgIP . 'includes/export/SMW_Exp_Element.php'; |
| 138 | | -// $wgAutoloadClasses['SMWExpResource'] = $smwgIP . 'includes/export/SMW_Exp_Element.php'; |
| | 134 | +// $expDir = $smwgIP . 'includes/export/'; |
| | 135 | +// $wgAutoloadClasses['SMWExporter'] = $expDir . 'SMW_Exporter.php'; |
| | 136 | +// $wgAutoloadClasses['SMWExpData'] = $expDir . 'SMW_Exp_Data.php'; |
| | 137 | +// $wgAutoloadClasses['SMWExpElement'] = $expDir . 'SMW_Exp_Element.php'; |
| | 138 | +// $wgAutoloadClasses['SMWExpLiteral'] = $expDir . 'SMW_Exp_Element.php'; |
| | 139 | +// $wgAutoloadClasses['SMWExpResource'] = $expDir . 'SMW_Exp_Element.php'; |
| | 140 | +// $wgAutoloadClasses['SMWExportController'] = $expDir . 'SMW_ExportController.php'; |
| | 141 | +// $wgAutoloadClasses['SMWSerializer'] = $expDir . 'SMW_Serializer.php'; |
| 139 | 142 | |
| 140 | 143 | // Parser hooks |
| 141 | 144 | $phDir = $smwgIP . 'includes/parserhooks/'; |
| Index: trunk/extensions/SemanticMediaWiki/includes/SMW_Setup.php |
| — | — | @@ -156,6 +156,8 @@ |
| 157 | 157 | $wgAutoloadClasses['SMWExpElement'] = $expDir . 'SMW_Exp_Element.php'; |
| 158 | 158 | $wgAutoloadClasses['SMWExpLiteral'] = $expDir . 'SMW_Exp_Element.php'; |
| 159 | 159 | $wgAutoloadClasses['SMWExpResource'] = $expDir . 'SMW_Exp_Element.php'; |
| | 160 | + $wgAutoloadClasses['SMWExportController'] = $expDir . 'SMW_ExportController.php'; |
| | 161 | + $wgAutoloadClasses['SMWSerializer'] = $expDir . 'SMW_Serializer.php'; |
| 160 | 162 | |
| 161 | 163 | // Parser hooks |
| 162 | 164 | $phDir = $smwgIP . 'includes/parserhooks/'; |
| — | — | @@ -231,7 +233,6 @@ |
| 232 | 234 | $wgSpecialPages['SemanticStatistics'] = array( 'SMWSpecialSemanticStatistics' ); |
| 233 | 235 | $wgSpecialPageGroups['SemanticStatistics'] = 'wiki'; // Similar to Special:Statistics |
| 234 | 236 | |
| 235 | | - $wgAutoloadClasses['SMWOWLExport'] = $smwgIP . 'includes/export/SMW_OWLExport.php'; |
| 236 | 237 | $wgAutoloadClasses['SMWSpecialOWLExport'] = $smwgIP . 'specials/Export/SMW_SpecialOWLExport.php'; |
| 237 | 238 | $wgSpecialPages['ExportRDF'] = array( 'SMWSpecialOWLExport' ); |
| 238 | 239 | $wgSpecialPageGroups['ExportRDF'] = 'smw_group'; |