r13009 MediaWiki - Code Review archive

Repository:	MediaWiki
Revision:	< r13008‎ \| r13009 \| r13010 >
Date:	23:39, 13 February 2006
Author:	evanprodromou
Status:	old
Tags:
Comment:	Now uses memcached to cache models if available. Still has some problems with Category namespace, but at least doesn't barf.
Modified paths:	/trunk/extensions/RDF/README.RDF.txt (modified) (history) /trunk/extensions/RDF/extensions/MwRdf.php (modified) (history)

Diff [purge]

Index: trunk/extensions/RDF/extensions/MwRdf.php
—	—	@@ -22,14 +22,16 @@
23	23	* @subpackage Extensions
24	24	*/
25	25	if (defined('MEDIAWIKI')) {
	26	+ require_once('GlobalFunctions.php');
26	27	if (!defined('RDFAPI_INCLUDE_DIR')) {
27	28	wfDebugDieBacktrace("MwRdf: you must install RAP (RDF API for PHP) " .
28	29	"and define 'RDFAPI_INCLUDE_DIR' in LocalSettings.php");
29	30	}
30	31	require_once(RDFAPI_INCLUDE_DIR . "RdfAPI.php");
31	32	require_once(RDFAPI_INCLUDE_DIR . PACKAGE_VOCABULARY);
32		-
33		~~- define('MWRDF_VERSION', '0.3');~~
	33	+ require_once('SpecialPage.php');
	34	+
	35	+ define('MWRDF_VERSION', '0.4');
34	36	define('MWRDF_XML_TYPE_PREFS',
35	37	"application/rdf+xml,text/xml;q=0.7," .
36	38	"application/xml;q=0.5,text/rdf;q=0.1");
—	—	@@ -72,13 +74,14 @@
73	75	'ntriples' => MwRdfOutputNtriples);
74	76
75	77	$wgRdfNamespaces = array('cc' => CC_NS);
76		-
	78	+ $wgRdfCacheExpiry = 86400;
	79	+
77	80	/* Config end */
78	81
79	82	$wgExtensionFunctions[] = 'setupMwRdf';
80	83
81	84	function setupMwRdf() {
82		~~- global $wgParser, $wgMessageCache, $wgRequest, $wgOut;~~
	85	+ global $wgParser, $wgMessageCache, $wgRequest, $wgOut, $wgHooks;
83	86
84	87	$wgMessageCache->addMessages(array('rdf' => 'Rdf',
85	88	'rdf-inpage' => "Embedded In-page Turtle",
—	—	@@ -128,7 +131,8 @@
129	132	if (isset($title) && mb_strlen($title) > 0) {
130	133	$nt =& Title::newFromText($title);
131	134
132		~~- if ($action == 'view' &&~~
	135	+ if (isset($nt) &&
	136	+ $action == 'view' &&
133	137	$nt->getNamespace() != NS_SPECIAL)
134	138	{
135	139	$rdft =& Title::makeTitle(NS_SPECIAL, "Rdf");
—	—	@@ -139,6 +143,13 @@
140	144	$wgOut->addMetadataLink($linkdata);
141	145	}
142	146	}
	147	+
	148	+ # We set some hooks for invalidating the cache
	149	+
	150	+ $wgHooks['ArticleSave'][] = 'MwRdfOnArticleSave';
	151	+ $wgHooks['ArticleSaveComplete'][] = 'MwRdfOnArticleSaveComplete';
	152	+ $wgHooks['TitleMoveComplete'][] = 'MwRdfOnTitleMoveComplete';
	153	+ $wgHooks['ArticleDeleteComplete'][] = 'MwRdfOnArticleDeleteComplete';
143	154	}
144	155
145	156	function wfSpecialRdf($par) {
—	—	@@ -283,8 +294,9 @@
284	295	}
285	296
286	297	$fullModel = ModelFactory::getDefaultModel();
	298	+ $title = $article->mTitle;
287	299
288		~~- $uri = $article->mTitle->getFullURL();~~
	300	+ $uri = $title->getFullURL();
289	301	$fullModel->setBaseURI($uri);
290	302
291	303	foreach ($modelnames as $modelname) {
—	—	@@ -296,11 +308,19 @@
297	309	" for model '$modelname'.");
298	310	}
299	311
300		~~- $model = $modelfunc($article);~~
301		~~- if ($model != null) {~~
	312	+ # Check the cache...
	313	+
	314	+ $model = MwRdfGetCache($title, $modelname);
	315	+
	316	+ # If it's not there, regenerate.
	317	+
	318	+ if (!isset($model) \|\| !$model) {
	319	+ $model = $modelfunc($article);
	320	+ MwRdfSetCache($title, $modelname, $model);
	321	+ }
	322	+
	323	+ if (isset($model)) {
302	324	$fullModel->addModel($model);
303		~~- # free the memory~~
304		~~- $model->close();~~
305	325	}
306	326	}
307	327
—	—	@@ -366,7 +386,7 @@
367	387	$parser = new N3Parser();
368	388	$parser->baseURI = $article->mTitle->getFullURL();
369	389
370		~~- $prefixes = array_merge($default_prefixes, MwRdfNamespacePrefixes(), $wgRdfNamespaces);~~
	390	+ $prefixes = array_merge($default_prefixes, $wgRdfNamespaces, MwRdfNamespacePrefixes());
371	391
372	392	$prelude = "";
373	393
—	—	@@ -383,7 +403,7 @@
384	404	$RDFS_comment,
385	405	MwRdfLiteral("Error parsing in-page RDF: "
386	406	. $parser->errors[0] .
387		~~- "\n code here: \n '" . $rdf . "'", null,~~
	407	+ "\n code here: \n '" . $prelude . $rdf . "'" , null,
388	408	"en")));
389	409	}
390	410	}
—	—	@@ -664,20 +684,26 @@
665	685	# Find prefixed links
666	686	preg_match_all("/\[\[([^\|\]]+:[^\|\]]+)(\\|.*)?\]\]/", $text, $matches);
667	687
668		~~- foreach ($matches[1] as $linktext) {~~
669		~~- $iwlink = Title::newFromText($linktext);~~
670		~~- $pfx = $iwlink->getInterwiki();~~
671		~~- if (mb_strlen($pfx) > 0) {~~
672		~~- $iwr = MwRdfTitleResource($iwlink);~~
673		~~- # XXX: Wikitravel uses some 4+ prefixes for sister site links~~
674		~~- if ($wgContLang->getLanguageName($pfx) && mb_strlen($pfx) < 4) {~~
675		~~- $model->add(new Statement($tr, $DCTERM['hasVersion'], $iwr));~~
676		~~- $model->add(new Statement($iwr, $DCMES['language'],~~
677		~~- MwRdfLanguage($pfx)));~~
678		~~- } else {~~
679		~~- # XXX: Express the "sister site" relationship better~~
680		~~- $model->add(new Statement($tr, $RDFS_seeAlso,~~
681		~~- $iwr));~~
	688	+ # XXX: this fails for Category: namespace; why?
	689	+
	690	+ if (isset($matches)) {
	691	+ foreach ($matches[1] as $linktext) {
	692	+ $iwlink = Title::newFromText($linktext);
	693	+ if (isset($iwlink)) {
	694	+ $pfx = $iwlink->getInterwiki();
	695	+ if (mb_strlen($pfx) > 0) {
	696	+ $iwr = MwRdfTitleResource($iwlink);
	697	+ # XXX: Wikitravel uses some 4+ prefixes for sister site links
	698	+ if ($wgContLang->getLanguageName($pfx) && mb_strlen($pfx) < 4) {
	699	+ $model->add(new Statement($tr, $DCTERM['hasVersion'], $iwr));
	700	+ $model->add(new Statement($iwr, $DCMES['language'],
	701	+ MwRdfLanguage($pfx)));
	702	+ } else {
	703	+ # XXX: Express the "sister site" relationship better
	704	+ $model->add(new Statement($tr, $RDFS_seeAlso,
	705	+ $iwr));
	706	+ }
	707	+ }
682	708	}
683	709	}
684	710	}
—	—	@@ -844,16 +870,132 @@
845	871	$prefixes = array();
846	872	$spaces = $wgLang->getNamespaces();
847	873	foreach ($spaces as $code => $text) {
848		~~- $prefix = str_replace(' ', '_', $text);~~
849		~~- # XXX: figure out a less sneaky way to do this~~
850		~~- # XXX: won't work if article title isn't at the end of the URL~~
851		~~- $title = Title::makeTitle($code, '');~~
852		~~- $uri = $title->getFullURL();~~
853		~~- $prefixes[$prefix] = $uri;~~
	874	+ $prefix = urlencode(str_replace(' ', '_', $text));
	875	+ # FIXME: this is a hack
	876	+ if (strpos($prefix, '%') === false) {
	877	+ # XXX: figure out a less sneaky way to do this
	878	+ # XXX: won't work if article title isn't at the end of the URL
	879	+ $title = Title::makeTitle($code, '');
	880	+ $uri = $title->getFullURL();
	881	+ $prefixes[$prefix] = $uri;
	882	+ }
854	883	}
855	884	}
856	885	return $prefixes;
857	886	}
	887	+
	888	+ function MwRdfGetCache($title, $modelname) {
	889	+ global $wgMemc;
	890	+ if (!isset($wgMemc)) {
	891	+ return false;
	892	+ } else {
	893	+ $ntrip = $wgMemc->get(MwRdfCacheKey($title, $modelname));
	894	+ if (isset($ntrip) && $ntrip) {
	895	+ return MwRdfNTriplesToModel($ntrip);
	896	+ } else {
	897	+ return null;
	898	+ }
	899	+ }
	900	+ }
	901	+
	902	+ function MwRdfClearCache($title, $modelname) {
	903	+ global $wgMemc;
	904	+ if (!isset($wgMemc)) {
	905	+ return false;
	906	+ } else {
	907	+ return $wgMemc->delete(MwRdfCacheKey($title, $modelname));
	908	+ }
	909	+ }
	910	+
	911	+ function MwRdfClearCacheAll($title) {
	912	+ global $wgRdfModelFunctions;
	913	+ $nt = $title;
	914	+ foreach (array_keys($wgRdfModelFunctions) as $modelname) {
	915	+ MwRdfClearCache($nt, $modelname);
	916	+ }
	917	+ }
	918	+
	919	+ function MwRdfSetCache($title, $modelname, $model) {
	920	+ global $wgMemc, $wgRdfCacheExpiry;
	921	+ if (!isset($wgMemc)) {
	922	+ return false;
	923	+ } else {
	924	+ return $wgMemc->set(MwRdfCacheKey($title, $modelname),
	925	+ MwRdfModelToNTriples($model),
	926	+ $wgRdfCacheExpiry);
	927	+ }
	928	+ }
	929	+
	930	+ function MwRdfCacheKey($title, $modelname) {
	931	+ if (!isset($title)) {
	932	+ return null;
	933	+ } else {
	934	+ global $wgDBname;
	935	+ $dbkey = $title->getDBkey();
	936	+ $ns = $title->getNamespace();
	937	+ return "$wgDBname:rdf:$ns:$dbkey:$modelname";
	938	+ }
	939	+ }
	940	+
	941	+ # Before saving, we clear the cache for articles this article links to
	942	+
	943	+ function MwRdfOnArticleSave($article, $dc1, $dc2, $dc3, $dc4, $dc5, $dc6) {
	944	+ $id = $article->mTitle->getArticleID();
	945	+ if ($id != 0) {
	946	+ $dbr =& wfGetDB(DB_SLAVE);
	947	+ $res = $dbr->select(array('cur', 'links'),
	948	+ array('cur_namespace', 'cur_title'),
	949	+ array('cur_id = l_to',
	950	+ 'l_from = ' . $id),
	951	+ 'MwRdfOnArticleSave',
	952	+ array('ORDER BY' => 'cur_namespace, cur_title'));
	953	+ while ($res && $row = $dbr->fetchObject($res)) {
	954	+ $lt = Title::makeTitle($row->cur_namespace, $row->cur_title);
	955	+ MwRdfClearCache($lt, 'linksto');
	956	+ }
	957	+ }
	958	+ return true;
	959	+ }
	960	+
	961	+ # Clear the cache when the article is saved
	962	+
	963	+ function MwRdfOnArticleSaveComplete($article, $dc1, $dc2, $dc3, $dc4, $dc5, $dc6) {
	964	+ MwRdfClearCacheAll($article->mTitle);
	965	+ return true;
	966	+ }
	967	+
	968	+ # Clear the cache when an article is moved
	969	+
	970	+ function MwRdfOnTitleMoveComplete($oldt, $newt, $user, $oldid, $newid) {
	971	+ MwRdfClearCacheAll($newt);
	972	+ MwRdfClearCacheAll($oldt);
	973	+ return true;
	974	+ }
	975	+
	976	+ # Clear the cache when an article is deleted
	977	+
	978	+ function MwRdfOnArticleDeleteComplete($article, $user, $reason) {
	979	+ MwRdfClearCacheAll($article->mTitle);
	980	+ return true;
	981	+ }
	982	+
	983	+ function MwRdfNTriplesToModel($ntrip) {
	984	+ require_once(RDFAPI_INCLUDE_DIR.PACKAGE_SYNTAX_N3);
	985	+ $parser = new N3Parser();
	986	+ $model = $parser->parse2model($ntrip);
	987	+ return $model;
	988	+ }
	989	+
	990	+ function MwRdfModelToNTriples($model) {
	991	+ # Make sure serializer is loaded
	992	+ if (!isset($model) \|\| $model->size() == 0) {
	993	+ return '';
	994	+ } else {
	995	+ require_once(RDFAPI_INCLUDE_DIR . PACKAGE_SYNTAX_N3);
	996	+ $ser = new NTripleSerializer();
	997	+ return $ser->serialize($model);
	998	+ }
	999	+ }
858	1000	}
859	1001
860		~~-?>~~
\ No newline at end of file
	1002	+?>
Index: trunk/extensions/RDF/README.RDF.txt
—	—	@@ -1,7 +1,7 @@
2	2	MediaWiki RDF extension
3	3
4		~~-version 0.3~~
5		~~-16 November 2005~~
	4	+version 0.4
	5	+24 January 2006
6	6
7	7	This is the README file for the RDF extension for MediaWiki
8	8	software. The extension is only useful if you've got a MediaWiki
—	—	@@ -19,7 +19,7 @@
20	20
21	21	== License ==
22	22
23		~~-Copyright 2005 Evan Prodromou <evan@wikitravel.org>~~
	23	+Copyright 2005, 2006 Evan Prodromou <evan@wikitravel.org>.
24	24
25	25	This program is free software; you can redistribute it and/or modify
26	26	it under the terms of the GNU General Public License as published by
—	—	@@ -255,7 +255,8 @@
256	256	$wgRdfOutputFunctions -- A map of output format to functions that
257	257	generate that output. You can add new output
258	258	formats by adding to this array.
259		-
	259	+$wgRdfCacheExpiry -- time in seconds to expire cached items
	260	+
260	261	== Extending ==
261	262
262	263	You can add new RDF models to the framework by creating a model
—	—	@@ -300,7 +301,7 @@
301	302
302	303	* Store statements in DB: statements could be stored in the database
303	304	when the page is saved and retrieved when needed. This would make it
304		~~- to do extended queries based on information about all pages.~~
	305	+ possible to do extended queries based on information about all pages.
305	306	* Performance: there wasn't much performance tuning and there are
306	307	probably way too many DB hits and reads and such.
307	308	* Semantic tuning: I'd like to make sure that the statements in the

Status & tagging log

01:58, 13 October 2010 😂 (talk | contribs) changed the status of r13009 [removed: new added: old]