Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
6.67% |
5 / 75 |
|
10.00% |
2 / 20 |
CRAP | |
0.00% |
0 / 1 |
| MathIdGenerator | |
6.67% |
5 / 75 |
|
10.00% |
2 / 20 |
1275.63 | |
0.00% |
0 / 1 |
| newFromRevisionRecord | |
0.00% |
0 / 12 |
|
0.00% |
0 / 1 |
12 | |||
| getKeys | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| __construct | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
2 | |||
| newFromRevisionId | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
2 | |||
| newFromTitle | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getIdList | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| formatIds | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| parserKey2fId | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
20 | |||
| getInputHash | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getIdsFromContent | |
75.00% |
3 / 4 |
|
0.00% |
0 / 1 |
2.06 | |||
| getContentIdMap | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
12 | |||
| guessIdFromContent | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
20 | |||
| getMathTags | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getWikiText | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getRevisionId | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setUseCustomIds | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getTagFromId | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
30 | |||
| getUniqueFromId | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
30 | |||
| formatKey | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| getUserInputTex | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| 1 | <?php |
| 2 | |
| 3 | use MediaWiki\Extension\Math\MathSource; |
| 4 | use MediaWiki\MediaWikiServices; |
| 5 | use MediaWiki\Parser\Parser; |
| 6 | use MediaWiki\Parser\Sanitizer; |
| 7 | use MediaWiki\Revision\RevisionRecord; |
| 8 | use MediaWiki\Revision\SlotRecord; |
| 9 | use MediaWiki\Title\Title; |
| 10 | |
| 11 | class MathIdGenerator { |
| 12 | |
| 13 | public const CONTENT_POS = 1; |
| 14 | public const ATTRIB_POS = 2; |
| 15 | |
| 16 | private string $wikiText; |
| 17 | /** |
| 18 | * @var array<string,array{0:string,1:string|null,2:array,3:string}> Filtered result from |
| 19 | * {@see Parser::extractTagsAndParams} with only <math> tags |
| 20 | */ |
| 21 | private array $mathTags; |
| 22 | private int $revisionId; |
| 23 | /** @var int[] */ |
| 24 | private $contentAccessStats = []; |
| 25 | private string $format = "math.%d.%d"; |
| 26 | private bool $useCustomIds = false; |
| 27 | /** @var int[]|null */ |
| 28 | private $keys; |
| 29 | /** @var array<string,?string[]>|null */ |
| 30 | private $contentIdMap; |
| 31 | |
| 32 | public static function newFromRevisionRecord( RevisionRecord $revisionRecord ): MathIdGenerator { |
| 33 | $contentModel = $revisionRecord |
| 34 | ->getSlot( SlotRecord::MAIN, RevisionRecord::RAW ) |
| 35 | ->getModel(); |
| 36 | if ( $contentModel !== CONTENT_MODEL_WIKITEXT ) { |
| 37 | throw new RuntimeException( "MathIdGenerator supports only CONTENT_MODEL_WIKITEXT" ); |
| 38 | } |
| 39 | $content = $revisionRecord->getContent( SlotRecord::MAIN ); |
| 40 | if ( !$content instanceof TextContent ) { |
| 41 | throw new RuntimeException( "MathIdGenerator supports only TextContent" ); |
| 42 | } |
| 43 | return new self( |
| 44 | $content->getText(), |
| 45 | $revisionRecord->getId() |
| 46 | ); |
| 47 | } |
| 48 | |
| 49 | /** |
| 50 | * @return array<string,int> Array mapping key names to their position |
| 51 | */ |
| 52 | public function getKeys() { |
| 53 | $this->keys ??= array_flip( array_keys( $this->mathTags ) ); |
| 54 | return $this->keys; |
| 55 | } |
| 56 | |
| 57 | public function __construct( string $wikiText, int $revisionId = 0 ) { |
| 58 | $wikiText = Sanitizer::removeHTMLcomments( $wikiText ); |
| 59 | $this->wikiText = |
| 60 | Parser::extractTagsAndParams( [ 'nowki', 'syntaxhighlight', 'math' ], $wikiText, |
| 61 | $tags ); |
| 62 | $this->mathTags = array_filter( $tags, static function ( $v ) { |
| 63 | return $v[0] === 'math'; |
| 64 | } ); |
| 65 | $this->revisionId = $revisionId; |
| 66 | } |
| 67 | |
| 68 | public static function newFromRevisionId( int $revId ): MathIdGenerator { |
| 69 | $revisionRecord = MediaWikiServices::getInstance() |
| 70 | ->getRevisionLookup() |
| 71 | ->getRevisionById( $revId ); |
| 72 | |
| 73 | return self::newFromRevisionRecord( $revisionRecord ); |
| 74 | } |
| 75 | |
| 76 | public static function newFromTitle( Title $title ): MathIdGenerator { |
| 77 | return self::newFromRevisionId( $title->getLatestRevID() ); |
| 78 | } |
| 79 | |
| 80 | public function getIdList() { |
| 81 | return $this->formatIds( $this->mathTags ); |
| 82 | } |
| 83 | |
| 84 | /** |
| 85 | * @param array<string,mixed> $mathTags |
| 86 | * |
| 87 | * @return string[] |
| 88 | */ |
| 89 | public function formatIds( $mathTags ) { |
| 90 | return array_map( [ $this, 'parserKey2fId' ], array_keys( $mathTags ) ); |
| 91 | } |
| 92 | |
| 93 | /** |
| 94 | * @param string $key |
| 95 | * |
| 96 | * @return string|null |
| 97 | */ |
| 98 | public function parserKey2fId( $key ) { |
| 99 | if ( $this->useCustomIds ) { |
| 100 | if ( isset( $this->mathTags[$key][self::ATTRIB_POS]['id'] ) ) { |
| 101 | return $this->mathTags[$key][self::ATTRIB_POS]['id']; |
| 102 | } |
| 103 | } |
| 104 | if ( isset( $this->mathTags[$key] ) ) { |
| 105 | return $this->formatKey( $key ); |
| 106 | } |
| 107 | } |
| 108 | |
| 109 | public function getInputHash( $inputTex ) { |
| 110 | return pack( "H32", md5( $inputTex ) ); |
| 111 | } |
| 112 | |
| 113 | /** |
| 114 | * @param string $content |
| 115 | * |
| 116 | * @return ?string[] |
| 117 | */ |
| 118 | public function getIdsFromContent( $content ) { |
| 119 | $contentIdMap = $this->getContentIdMap(); |
| 120 | if ( array_key_exists( $content, $contentIdMap ) ) { |
| 121 | return $contentIdMap[$content]; |
| 122 | } |
| 123 | return []; |
| 124 | } |
| 125 | |
| 126 | /** |
| 127 | * @return array<string,?string[]> |
| 128 | */ |
| 129 | public function getContentIdMap() { |
| 130 | if ( !$this->contentIdMap ) { |
| 131 | $this->contentIdMap = []; |
| 132 | foreach ( $this->mathTags as $key => $tag ) { |
| 133 | $userInputTex = $this->getUserInputTex( $tag ); |
| 134 | $this->contentIdMap[$userInputTex][] = $this->parserKey2fId( $key ); |
| 135 | } |
| 136 | } |
| 137 | return $this->contentIdMap; |
| 138 | } |
| 139 | |
| 140 | public function guessIdFromContent( string $content ): ?string { |
| 141 | $allIds = $this->getIdsFromContent( $content ); |
| 142 | $size = count( $allIds ); |
| 143 | if ( $size == 0 ) { |
| 144 | return null; |
| 145 | } |
| 146 | if ( $size == 1 ) { |
| 147 | return $allIds[0]; |
| 148 | } |
| 149 | if ( array_key_exists( $content, $this->contentAccessStats ) ) { |
| 150 | $this->contentAccessStats[$content]++; |
| 151 | } else { |
| 152 | $this->contentAccessStats[$content] = 0; |
| 153 | } |
| 154 | $currentIndex = $this->contentAccessStats[$content] % $size; |
| 155 | return $allIds[$currentIndex]; |
| 156 | } |
| 157 | |
| 158 | /** |
| 159 | * @return array |
| 160 | */ |
| 161 | public function getMathTags(): array { |
| 162 | return $this->mathTags; |
| 163 | } |
| 164 | |
| 165 | public function getWikiText(): string { |
| 166 | return $this->wikiText; |
| 167 | } |
| 168 | |
| 169 | public function getRevisionId(): int { |
| 170 | return $this->revisionId; |
| 171 | } |
| 172 | |
| 173 | public function setUseCustomIds( bool $useCustomIds ): void { |
| 174 | $this->useCustomIds = $useCustomIds; |
| 175 | } |
| 176 | |
| 177 | /** |
| 178 | * @param string $eid |
| 179 | * @return array|null |
| 180 | */ |
| 181 | public function getTagFromId( string $eid ): ?array { |
| 182 | foreach ( $this->mathTags as $key => $mathTag ) { |
| 183 | if ( $eid == $this->formatKey( $key ) ) { |
| 184 | return $mathTag; |
| 185 | } |
| 186 | if ( isset( $mathTag[self::ATTRIB_POS]['id'] ) && $eid == $mathTag[self::ATTRIB_POS]['id'] ) { |
| 187 | return $mathTag; |
| 188 | } |
| 189 | } |
| 190 | return null; |
| 191 | } |
| 192 | |
| 193 | public function getUniqueFromId( string $eid ): ?string { |
| 194 | foreach ( $this->mathTags as $key => $mathTag ) { |
| 195 | if ( $eid == $this->formatKey( $key ) ) { |
| 196 | return $key; |
| 197 | } |
| 198 | if ( isset( $mathTag[self::ATTRIB_POS]['id'] ) && $eid == $mathTag[self::ATTRIB_POS]['id'] ) { |
| 199 | return $key; |
| 200 | } |
| 201 | } |
| 202 | return null; |
| 203 | } |
| 204 | |
| 205 | private function formatKey( string $key ): string { |
| 206 | $keys = $this->getKeys(); |
| 207 | return sprintf( $this->format, $this->revisionId, $keys[$key] ); |
| 208 | } |
| 209 | |
| 210 | /** |
| 211 | * @param array{1:string,2:array} $tag |
| 212 | * @return string |
| 213 | */ |
| 214 | public function getUserInputTex( array $tag ): string { |
| 215 | return ( new MathSource( $tag[self::CONTENT_POS], $tag[self::ATTRIB_POS] ) )->getUserInputTex(); |
| 216 | } |
| 217 | } |