ImplicitDescriptionUsageLookup::findImplicitUsages() - Code Metrics - Inspection of "Merge "Fix keys in ImplicitDescriptionUsageLookup:..." - wikimedia/mediawiki-extensions-Wikibase - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( 1326c7...574a68 )

unknown

created 2020-10-29 10:16 UTC

findImplicitUsages() B

↳ Parent: ImplicitDescriptionUsageLookup

Complexity

Conditions	7
Paths	27

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
dl	0
loc	35
rs	8.4266
c	0
b	0
f	0
cc	7
nc	27
nop	1

<?php

declare( strict_types = 1 );

namespace Wikibase\Client\Usage;

use MediaWiki\Cache\LinkBatchFactory;
use TitleFactory;
use Traversable;
use Wikibase\DataModel\Entity\EntityId;
use Wikibase\DataModel\Entity\ItemId;
use Wikibase\Lib\Store\SiteLinkLookup;

/**
 * A {@link UsageLookup} which decorates an inner lookup
 * and adds an implicit usage on a linked item’s description.
 *
 * An implicit usage is different from an ordinary, explicit usage
 * in that it is never recorded by a {@link UsageTracker}:
 * it is not tracked when the page is parsed and actually uses a part of an entity,
 * but rather synthesized by this class based on hard-coded knowledge
 * about where else an entity’s data is used in relation to a page.
 * However, implicit usages otherwise look exactly like explicit usages:
 * for a user of the {@link UsageLookup} interface, it is not (yet?) possible
 * to determine whether a usage from the lookup is explicit or implicit.
 *
 * This class implements one kind of implicit usage:
 * if a client page is linked to an item, it has an implicit usage
 * on that item’s description in the client wiki’s content language.
 * This is because the description is used, for example,
 * as part of the search result for the page (typically on mobile),
 * even if it is never used in the page itself.
 *
 * @see @ref md_docs_topics_usagetracking for virtual usage,
 * a similar but separate concept.
 *
 * @license GPL-2.0-or-later
 */
class ImplicitDescriptionUsageLookup implements UsageLookup {

	/** @var UsageLookup */
	private $usageLookup;

	/** @var TitleFactory */
	private $titleFactory;

	/** @var LinkBatchFactory */
	private $linkBatchFactory;

	/** @var string */
	private $globalSiteId;

	/** @var SiteLinkLookup */
	private $siteLinkLookup;

	/**
	 * @param UsageLookup $usageLookup The underlying/inner lookup.
	 * @param TitleFactory $titleFactory
	 * @param LinkBatchFactory $linkBatchFactory
	 * @param string $globalSiteId The global site ID of the client wiki.
	 * @param SiteLinkLookup $siteLinkLookup
	 */
	public function __construct(
		UsageLookup $usageLookup,
		TitleFactory $titleFactory,
		LinkBatchFactory $linkBatchFactory,
		string $globalSiteId,
		SiteLinkLookup $siteLinkLookup
	) {
		$this->usageLookup = $usageLookup;
		$this->titleFactory = $titleFactory;
		$this->linkBatchFactory = $linkBatchFactory;
		$this->globalSiteId = $globalSiteId;
		$this->siteLinkLookup = $siteLinkLookup;
	}

	public function getUsagesForPage( $pageId ): array {
		$usages = $this->usageLookup->getUsagesForPage( $pageId );
		$title = $this->titleFactory->newFromID( $pageId );
		if ( !$title ) {
			return $usages;
		}
		$entityId = $this->siteLinkLookup->getItemIdForLink(
			$this->globalSiteId,
			$title->getPrefixedText()
		);
		if ( !$entityId ) {
			return $usages;
		}
		$contentLanguage = $title->getPageLanguage()->getCode();

		$usage = new EntityUsage(
			$entityId,
			EntityUsage::DESCRIPTION_USAGE,
			$contentLanguage
		);
		// this might replace an existing usage but that’s okay
		$usages[$usage->getIdentityString()] = $usage;

		return $usages;
	}

	public function getPagesUsing( array $entityIds, array $aspects = [] ): Traversable {
		if ( !$this->aspectsMatchImplicitUsage( $aspects ) ) {
			// Caller is not interested in implicit usage,
			// no need to add anything
			return yield from $this->usageLookup->getPagesUsing( $entityIds, $aspects );
		}

		// Find the implicit usages that we’ll add – one per page / item ID / content language
		[ $itemIdsByPageId, $contentLanguagesByPageId ] = $this->findImplicitUsages( $entityIds );

		// Filter them according to the aspects
		[ $itemIdsByPageId, $contentLanguagesByPageId ] = $this->filterImplicitUsages(
			$itemIdsByPageId, $contentLanguagesByPageId, $aspects );

		// Now decorate the inner lookup’s usages with them
		foreach ( $this->usageLookup->getPagesUsing( $entityIds, $aspects ) as $pageEntityUsages ) {
			/** @var PageEntityUsages $pageEntityUsages */
			'@phan-var PageEntityUsages $pageEntityUsages';
			$pageId = $pageEntityUsages->getPageId();
			if ( isset( $itemIdsByPageId[$pageId] ) ) {
				// if equivalent usages already exist then addUsages() is a no-op
				$pageEntityUsages->addUsages( [
					new EntityUsage(
						$itemIdsByPageId[$pageId],
						EntityUsage::DESCRIPTION_USAGE,
						$contentLanguagesByPageId[$pageId]
					),
				] );
				unset( $itemIdsByPageId[$pageId] );
			}
			yield $pageEntityUsages;
		}

		// And yield any remaining pages that the inner lookup didn’t return at all
		foreach ( $itemIdsByPageId as $pageId => $itemId ) {
			yield new PageEntityUsages( $pageId, [
				new EntityUsage(
					$itemId,
					EntityUsage::DESCRIPTION_USAGE,
					$contentLanguagesByPageId[$pageId]
				),
			] );
		}
	}

	/**
	 * Whether the given aspects potentially match an implicit usage.
	 *
	 * @param string[] $aspects
	 * @return bool
	 */
	private function aspectsMatchImplicitUsage( array $aspects ): bool {
		if ( $aspects === [] ) {
			return true;
		}

		foreach ( $aspects as $aspectKey ) {
			if ( EntityUsage::stripModifier( $aspectKey ) === EntityUsage::DESCRIPTION_USAGE ) {
				// The implicit usage is on the description in the *page* content language,
				// not the wiki content language, so any description aspect matches.
				// (We’ll later filter based on the modifier and content language,
				// see filterImplicitUsages().)
				return true;
			}
		}
		return false;
	}

	/**
	 * Find the implicit usages on the given entity IDs.
	 *
	 * Returns two arrays, both keyed by page ID:
	 * the item ID linked to that page and the content language of the page.
	 *
	 * @param EntityId[] $entityIds
	 * @return array [ ItemId[] $itemIdsByPageId, string[] $contentLanguagesByPageId ]
	 */
	private function findImplicitUsages( array $entityIds ): array {
		$numericItemIds = [];
		foreach ( $entityIds as $entityId ) {
			if ( $entityId instanceof ItemId ) {
				$numericItemIds[] = $entityId->getNumericId();
			}
		}

		// each link is an array [ string $siteId, string $pageName, int $itemId ]
		$links = $this->siteLinkLookup->getLinks( $numericItemIds, [ $this->globalSiteId ] );
		// preload the titles in bulk (page ID and language)
		$titles = array_map( [ $this->titleFactory, 'newFromDBkey' ], array_column( $links, 1 ) );
		$this->linkBatchFactory->newLinkBatch( $titles )->execute();

		$itemIdsByPageId = [];
		foreach ( $links as [ $siteId, $pageName, $itemId ] ) {
			// note: this creates a new Title and looks up its page ID in the link cache;
			// this is simpler than finding the right existing Title in the $titles we have
			// (the $pageName is probably not exactly in DB key form)
			$pageId = $this->titleFactory->newFromDBkey( $pageName )->getArticleID();

			if ( $pageId ) {
				$itemIdsByPageId[$pageId] = ItemId::newFromNumber( $itemId );

			}
		}

		$contentLanguagesByPageId = [];
		foreach ( $titles as $title ) {
			$pageId = $title->getArticleID();
			if ( $pageId ) {
				$contentLanguagesByPageId[$pageId] = $title->getPageLanguage()->getCode();
			}
		}

		return [ $itemIdsByPageId, $contentLanguagesByPageId ];
	}

	/**
	 * Filter the implicit usages by the given aspects.
	 *
	 * Takes two arrays as returned by {@link findImplicitUsages}
	 * and returns similar arrays, but filtered if necessary.
	 *
	 * @param ItemId[] $itemIdsByPageId
	 * @param string[] $contentLanguagesByPageId
	 * @param string[] $aspects
	 * @return array [ ItemId[] $itemIdsByPageId, string[] $contentLanguagesByPageId ]
	 */
	private function filterImplicitUsages(
		array $itemIdsByPageId,
		array $contentLanguagesByPageId,
		array $aspects
	): array {
		if ( $aspects === [] ) {
			// caller is interested in all usages, don’t filter
			return [ $itemIdsByPageId, $contentLanguagesByPageId ];
		}

		$relevantLanguages = [];
		foreach ( $aspects as $aspectKey ) {
			[ $aspect, $modifier ] = EntityUsage::splitAspectKey( $aspectKey );

			if ( $aspect !== EntityUsage::DESCRIPTION_USAGE ) {

				continue;
			}
			if ( $modifier === null ) {
				// caller is interested in all description usages, don’t filter
				return [ $itemIdsByPageId, $contentLanguagesByPageId ];
			}
			$relevantLanguages[] = $modifier;
		}

		// caller is only interested in some description usages, filter
		foreach ( $contentLanguagesByPageId as $pageId => $contentLanguage ) {
			if ( !in_array( $contentLanguage, $relevantLanguages, /* strict */ true ) ) {
				unset( $itemIdsByPageId[$pageId] );
				unset( $contentLanguagesByPageId[$pageId] );
			}
		}
		return [ $itemIdsByPageId, $contentLanguagesByPageId ];
	}

	public function getUnusedEntities( array $entityIds ): array {
		// If a page is linked to an item, it has at least a sitelink usage on it;
		// therefore, the implicit usage can never make a difference for
		// whether an entity is used or unused.
		return $this->usageLookup->getUnusedEntities( $entityIds );
	}

}


1			<?php
2
3			declare( strict_types = 1 );
4
5			namespace Wikibase\Client\Usage;
6
7			use MediaWiki\Cache\LinkBatchFactory;
8			use TitleFactory;
9			use Traversable;
10			use Wikibase\DataModel\Entity\EntityId;
11			use Wikibase\DataModel\Entity\ItemId;
12			use Wikibase\Lib\Store\SiteLinkLookup;
13
14			/**
15			* A {@link UsageLookup} which decorates an inner lookup
16			* and adds an implicit usage on a linked item’s description.
17			*
18			* An implicit usage is different from an ordinary, explicit usage
19			* in that it is never recorded by a {@link UsageTracker}:
20			* it is not tracked when the page is parsed and actually uses a part of an entity,
21			* but rather synthesized by this class based on hard-coded knowledge
22			* about where else an entity’s data is used in relation to a page.
23			* However, implicit usages otherwise look exactly like explicit usages:
24			* for a user of the {@link UsageLookup} interface, it is not (yet?) possible
25			* to determine whether a usage from the lookup is explicit or implicit.
26			*
27			* This class implements one kind of implicit usage:
28			* if a client page is linked to an item, it has an implicit usage
29			* on that item’s description in the client wiki’s content language.
30			* This is because the description is used, for example,
31			* as part of the search result for the page (typically on mobile),
32			* even if it is never used in the page itself.
33			*
34			* @see @ref md_docs_topics_usagetracking for virtual usage,
35			* a similar but separate concept.
36			*
37			* @license GPL-2.0-or-later
38			*/
39			class ImplicitDescriptionUsageLookup implements UsageLookup {
40
41			/** @var UsageLookup */
42			private $usageLookup;
43
44			/** @var TitleFactory */
45			private $titleFactory;
46
47			/** @var LinkBatchFactory */
48			private $linkBatchFactory;
49
50			/** @var string */
51			private $globalSiteId;
52
53			/** @var SiteLinkLookup */
54			private $siteLinkLookup;
55
56			/**
57			* @param UsageLookup $usageLookup The underlying/inner lookup.
58			* @param TitleFactory $titleFactory
59			* @param LinkBatchFactory $linkBatchFactory
60			* @param string $globalSiteId The global site ID of the client wiki.
61			* @param SiteLinkLookup $siteLinkLookup
62			*/
63			public function __construct(
64			UsageLookup $usageLookup,
65			TitleFactory $titleFactory,
66			LinkBatchFactory $linkBatchFactory,
67			string $globalSiteId,
68			SiteLinkLookup $siteLinkLookup
69			) {
70			$this->usageLookup = $usageLookup;
71			$this->titleFactory = $titleFactory;
72			$this->linkBatchFactory = $linkBatchFactory;
73			$this->globalSiteId = $globalSiteId;
74			$this->siteLinkLookup = $siteLinkLookup;
75			}
76
77			public function getUsagesForPage( $pageId ): array {
78			$usages = $this->usageLookup->getUsagesForPage( $pageId );
79			$title = $this->titleFactory->newFromID( $pageId );
80			if ( !$title ) {
81			return $usages;
82			}
83			$entityId = $this->siteLinkLookup->getItemIdForLink(
84			$this->globalSiteId,
85			$title->getPrefixedText()
86			);
87			if ( !$entityId ) {
88			return $usages;
89			}
90			$contentLanguage = $title->getPageLanguage()->getCode();
91
92			$usage = new EntityUsage(
93			$entityId,
94			EntityUsage::DESCRIPTION_USAGE,
95			$contentLanguage
96			);
97			// this might replace an existing usage but that’s okay
98			$usages[$usage->getIdentityString()] = $usage;
99
100			return $usages;
101			}
102
103			public function getPagesUsing( array $entityIds, array $aspects = [] ): Traversable {
104			if ( !$this->aspectsMatchImplicitUsage( $aspects ) ) {
105			// Caller is not interested in implicit usage,
106			// no need to add anything
107			return yield from $this->usageLookup->getPagesUsing( $entityIds, $aspects );
108			}
109
110			// Find the implicit usages that we’ll add – one per page / item ID / content language
111			[ $itemIdsByPageId, $contentLanguagesByPageId ] = $this->findImplicitUsages( $entityIds );
			0 ignored issues – show Bug introduced 2020-10-28 11:24 UTC by Report Bug Copy Issue Report The variable `$itemIdsByPageId` does not exist. Did you forget to declare it? This check marks access to variables or properties that have not been declared yet. While PHP has no explicit notion of declaring a variable, accessing it before a value is assigned to it is most likely a bug. Loading history... Bug introduced 2020-10-28 11:24 UTC by Report Bug Copy Issue Report The variable `$contentLanguagesByPageId` does not exist. Did you forget to declare it? This check marks access to variables or properties that have not been declared yet. While PHP has no explicit notion of declaring a variable, accessing it before a value is assigned to it is most likely a bug. Loading history...
112			// Filter them according to the aspects
113			[ $itemIdsByPageId, $contentLanguagesByPageId ] = $this->filterImplicitUsages(
114			$itemIdsByPageId, $contentLanguagesByPageId, $aspects );
115
116			// Now decorate the inner lookup’s usages with them
117			foreach ( $this->usageLookup->getPagesUsing( $entityIds, $aspects ) as $pageEntityUsages ) {
118			/** @var PageEntityUsages $pageEntityUsages */
119			'@phan-var PageEntityUsages $pageEntityUsages';
120			$pageId = $pageEntityUsages->getPageId();
121			if ( isset( $itemIdsByPageId[$pageId] ) ) {
122			// if equivalent usages already exist then addUsages() is a no-op
123			$pageEntityUsages->addUsages( [
124			new EntityUsage(
125			$itemIdsByPageId[$pageId],
126			EntityUsage::DESCRIPTION_USAGE,
127			$contentLanguagesByPageId[$pageId]
128			),
129			] );
130			unset( $itemIdsByPageId[$pageId] );
131			}
132			yield $pageEntityUsages;
133			}
134
135			// And yield any remaining pages that the inner lookup didn’t return at all
136			foreach ( $itemIdsByPageId as $pageId => $itemId ) {
137			yield new PageEntityUsages( $pageId, [
138			new EntityUsage(
139			$itemId,
140			EntityUsage::DESCRIPTION_USAGE,
141			$contentLanguagesByPageId[$pageId]
142			),
143			] );
144			}
145			}
146
147			/**
148			* Whether the given aspects potentially match an implicit usage.
149			*
150			* @param string[] $aspects
151			* @return bool
152			*/
153			private function aspectsMatchImplicitUsage( array $aspects ): bool {
154			if ( $aspects === [] ) {
155			return true;
156			}
157
158			foreach ( $aspects as $aspectKey ) {
159			if ( EntityUsage::stripModifier( $aspectKey ) === EntityUsage::DESCRIPTION_USAGE ) {
160			// The implicit usage is on the description in the page content language,
161			// not the wiki content language, so any description aspect matches.
162			// (We’ll later filter based on the modifier and content language,
163			// see filterImplicitUsages().)
164			return true;
165			}
166			}
167			return false;
168			}
169
170			/**
171			* Find the implicit usages on the given entity IDs.
172			*
173			* Returns two arrays, both keyed by page ID:
174			* the item ID linked to that page and the content language of the page.
175			*
176			* @param EntityId[] $entityIds
177			* @return array [ ItemId[] $itemIdsByPageId, string[] $contentLanguagesByPageId ]
178			*/
179			private function findImplicitUsages( array $entityIds ): array {
180			$numericItemIds = [];
181			foreach ( $entityIds as $entityId ) {
182			if ( $entityId instanceof ItemId ) {
183			$numericItemIds[] = $entityId->getNumericId();
184			}
185			}
186
187			// each link is an array [ string $siteId, string $pageName, int $itemId ]
188			$links = $this->siteLinkLookup->getLinks( $numericItemIds, [ $this->globalSiteId ] );
189			// preload the titles in bulk (page ID and language)
190			$titles = array_map( [ $this->titleFactory, 'newFromDBkey' ], array_column( $links, 1 ) );
191			$this->linkBatchFactory->newLinkBatch( $titles )->execute();
192
193			$itemIdsByPageId = [];
194			foreach ( $links as [ $siteId, $pageName, $itemId ] ) {
195			// note: this creates a new Title and looks up its page ID in the link cache;
196			// this is simpler than finding the right existing Title in the $titles we have
197			// (the $pageName is probably not exactly in DB key form)
198			$pageId = $this->titleFactory->newFromDBkey( $pageName )->getArticleID();
			0 ignored issues – show Bug introduced 2020-10-28 11:24 UTC by Report Bug Copy Issue Report The variable `$pageName` does not exist. Did you forget to declare it? This check marks access to variables or properties that have not been declared yet. While PHP has no explicit notion of declaring a variable, accessing it before a value is assigned to it is most likely a bug. Loading history...
199			if ( $pageId ) {
200			$itemIdsByPageId[$pageId] = ItemId::newFromNumber( $itemId );
			0 ignored issues – show Bug introduced 2020-10-28 11:24 UTC by Report Bug Copy Issue Report The variable `$itemId` does not exist. Did you mean `$numericItemIds`? This check looks for variables that are accessed but have not been defined. It raises an issue if it finds another variable that has a similar name. The variable may have been renamed without also renaming all references. Loading history...
201			}
202			}
203
204			$contentLanguagesByPageId = [];
205			foreach ( $titles as $title ) {
206			$pageId = $title->getArticleID();
207			if ( $pageId ) {
208			$contentLanguagesByPageId[$pageId] = $title->getPageLanguage()->getCode();
209			}
210			}
211
212			return [ $itemIdsByPageId, $contentLanguagesByPageId ];
213			}
214
215			/**
216			* Filter the implicit usages by the given aspects.
217			*
218			* Takes two arrays as returned by {@link findImplicitUsages}
219			* and returns similar arrays, but filtered if necessary.
220			*
221			* @param ItemId[] $itemIdsByPageId
222			* @param string[] $contentLanguagesByPageId
223			* @param string[] $aspects
224			* @return array [ ItemId[] $itemIdsByPageId, string[] $contentLanguagesByPageId ]
225			*/
226			private function filterImplicitUsages(
227			array $itemIdsByPageId,
228			array $contentLanguagesByPageId,
229			array $aspects
230			): array {
231			if ( $aspects === [] ) {
232			// caller is interested in all usages, don’t filter
233			return [ $itemIdsByPageId, $contentLanguagesByPageId ];
234			}
235
236			$relevantLanguages = [];
237			foreach ( $aspects as $aspectKey ) {
238			[ $aspect, $modifier ] = EntityUsage::splitAspectKey( $aspectKey );
			0 ignored issues – show Bug introduced 2020-10-28 11:24 UTC by Report Bug Copy Issue Report The variable `$aspect` does not exist. Did you mean `$aspects`? This check looks for variables that are accessed but have not been defined. It raises an issue if it finds another variable that has a similar name. The variable may have been renamed without also renaming all references. Loading history... Bug introduced 2020-10-28 11:24 UTC by Report Bug Copy Issue Report The variable `$modifier` does not exist. Did you forget to declare it? This check marks access to variables or properties that have not been declared yet. While PHP has no explicit notion of declaring a variable, accessing it before a value is assigned to it is most likely a bug. Loading history...
239			if ( $aspect !== EntityUsage::DESCRIPTION_USAGE ) {
			0 ignored issues – show Bug introduced 2020-10-28 11:24 UTC by Report Bug Copy Issue Report The variable `$aspect` does not exist. Did you mean `$aspects`? This check looks for variables that are accessed but have not been defined. It raises an issue if it finds another variable that has a similar name. The variable may have been renamed without also renaming all references. Loading history...
240			continue;
241			}
242			if ( $modifier === null ) {
243			// caller is interested in all description usages, don’t filter
244			return [ $itemIdsByPageId, $contentLanguagesByPageId ];
245			}
246			$relevantLanguages[] = $modifier;
247			}
248
249			// caller is only interested in some description usages, filter
250			foreach ( $contentLanguagesByPageId as $pageId => $contentLanguage ) {
251			if ( !in_array( $contentLanguage, $relevantLanguages, /* strict */ true ) ) {
252			unset( $itemIdsByPageId[$pageId] );
253			unset( $contentLanguagesByPageId[$pageId] );
254			}
255			}
256			return [ $itemIdsByPageId, $contentLanguagesByPageId ];
257			}
258
259			public function getUnusedEntities( array $entityIds ): array {
260			// If a page is linked to an item, it has at least a sitelink usage on it;
261			// therefore, the implicit usage can never make a difference for
262			// whether an entity is used or unused.
263			return $this->usageLookup->getUnusedEntities( $entityIds );
264			}
265
266			}
267

wikimedia / mediawiki-extensions-Wikibase

Push — master ( 1326c7...574a68 )

findImplicitUsages() B

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like