Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
71.57% |
73 / 102 |
|
88.89% |
8 / 9 |
CRAP | |
0.00% |
0 / 1 |
| Cleaner | |
71.57% |
73 / 102 |
|
88.89% |
8 / 9 |
60.57 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| cleanHtml | |
100.00% |
11 / 11 |
|
100.00% |
1 / 1 |
4 | |||
| createDomDocument | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
1 | |||
| addContent | |
100.00% |
27 / 27 |
|
100.00% |
1 / 1 |
11 | |||
| matchesRemove | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
7 | |||
| nodeHasClass | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
2 | |||
| addPartOfContent | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
2 | |||
| lastElement | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| cleanHtmlDom | |
0.00% |
0 / 29 |
|
0.00% |
0 / 1 |
20 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace MediaWiki\Wikispeech\Segment; |
| 4 | |
| 5 | /** |
| 6 | * @file |
| 7 | * @ingroup Extensions |
| 8 | * @license GPL-2.0-or-later |
| 9 | */ |
| 10 | |
| 11 | use DOMComment; |
| 12 | use DOMDocument; |
| 13 | use DOMElement; |
| 14 | use DOMNode; |
| 15 | use DOMXPath; |
| 16 | use LogicException; |
| 17 | use MediaWiki\Wikispeech\Segment\PartOfContent\Link; |
| 18 | |
| 19 | /** |
| 20 | * Used for cleaning text with HTML markup. The cleaned text is used |
| 21 | * as input for `Segmenter`. |
| 22 | * |
| 23 | * @since 0.1.13 Add `$partOfContent` |
| 24 | * @since 0.0.1 |
| 25 | */ |
| 26 | class Cleaner { |
| 27 | /** |
| 28 | * An array of tags that should be removed completely during cleaning. |
| 29 | * |
| 30 | * @var array |
| 31 | */ |
| 32 | private $removeTags; |
| 33 | |
| 34 | /** |
| 35 | * An array of tags that should add a segment break during cleaning. |
| 36 | * |
| 37 | * @var array |
| 38 | */ |
| 39 | private $segmentBreakingTags; |
| 40 | |
| 41 | /** |
| 42 | * If true certain tags add extra content that is read before any text. |
| 43 | * |
| 44 | * @var bool |
| 45 | */ |
| 46 | private $partOfContent; |
| 47 | |
| 48 | /** |
| 49 | * An array of `CleanedText`s and `SegmentBreak`s. |
| 50 | * |
| 51 | * @var SegmentContent[] |
| 52 | */ |
| 53 | private $cleanedContent; |
| 54 | |
| 55 | /** |
| 56 | * @since 0.1.13 Add `$partOfContent` |
| 57 | * @param array $removeTags An array of tags that should be |
| 58 | * removed completely during cleaning. |
| 59 | * @param array $segmentBreakingTags An array of `CleanedText`s |
| 60 | * and `SegmentBreak`s. |
| 61 | * @param bool $partOfContent If true certain tags add extra content that is read before any text. |
| 62 | */ |
| 63 | public function __construct( |
| 64 | $removeTags, |
| 65 | $segmentBreakingTags, |
| 66 | bool $partOfContent |
| 67 | ) { |
| 68 | $this->removeTags = $removeTags; |
| 69 | $this->segmentBreakingTags = $segmentBreakingTags; |
| 70 | $this->partOfContent = $partOfContent; |
| 71 | } |
| 72 | |
| 73 | /** |
| 74 | * Clean HTML tags from a string. |
| 75 | * |
| 76 | * Separates any HTML tags from the text. |
| 77 | * |
| 78 | * @since 0.0.1 |
| 79 | * @param string $markedUpText Input text that may contain HTML |
| 80 | * tags. |
| 81 | * @return SegmentContent[] Represents the DOM nodes. |
| 82 | */ |
| 83 | public function cleanHtml( $markedUpText ): array { |
| 84 | $dom = self::createDomDocument( $markedUpText ); |
| 85 | $xpath = new DOMXPath( $dom ); |
| 86 | // Only add elements below the dummy element. These are the |
| 87 | // elements from the original HTML. |
| 88 | $top = $xpath->evaluate( '/meta/dummy' )->item( 0 ); |
| 89 | $this->cleanedContent = []; |
| 90 | $this->addContent( $top ); |
| 91 | // Remove any segment break at the start or end of the array, |
| 92 | // since they won't do anything. |
| 93 | if ( |
| 94 | $this->cleanedContent && |
| 95 | $this->cleanedContent[0] instanceof SegmentBreak |
| 96 | ) { |
| 97 | array_shift( $this->cleanedContent ); |
| 98 | } |
| 99 | if ( self::lastElement( $this->cleanedContent ) instanceof SegmentBreak ) { |
| 100 | array_pop( $this->cleanedContent ); |
| 101 | } |
| 102 | return $this->cleanedContent; |
| 103 | } |
| 104 | |
| 105 | /** |
| 106 | * Create a DOMDocument from an HTML string. |
| 107 | * |
| 108 | * A dummy element is added as top node. |
| 109 | * |
| 110 | * @since 0.0.1 |
| 111 | * @param string $markedUpText The string to create the |
| 112 | * DOMDocument. |
| 113 | * @return DOMDocument The created DOMDocument. |
| 114 | */ |
| 115 | private static function createDomDocument( $markedUpText ): DOMDocument { |
| 116 | $dom = new DOMDocument(); |
| 117 | // Add encoding information and wrap the input text in a dummy |
| 118 | // tag to prevent p tags from being added for text nodes. |
| 119 | $wrappedText = '<head><meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>' . |
| 120 | '<dummy>' . $markedUpText . '</dummy></head>'; |
| 121 | libxml_use_internal_errors( true ); |
| 122 | $dom->loadHTML( |
| 123 | $wrappedText, |
| 124 | LIBXML_HTML_NODEFDTD | LIBXML_HTML_NOIMPLIED |
| 125 | ); |
| 126 | return $dom; |
| 127 | } |
| 128 | |
| 129 | /** |
| 130 | * Recursively add items to the cleaned content. |
| 131 | * |
| 132 | * Goes through all the child nodes of $node and adds their |
| 133 | * content text. Adds segment breaks for appropriate tags. |
| 134 | * |
| 135 | * @since 0.0.1 |
| 136 | * @param DOMNode $node The top node to add from. |
| 137 | */ |
| 138 | private function addContent( $node ): void { |
| 139 | if ( $node instanceof DOMComment ) { |
| 140 | return; |
| 141 | } |
| 142 | |
| 143 | if ( $this->matchesRemove( $node ) ) { |
| 144 | return; |
| 145 | } |
| 146 | |
| 147 | foreach ( $node->childNodes as $child ) { |
| 148 | if ( |
| 149 | !self::lastElement( $this->cleanedContent ) |
| 150 | instanceof SegmentBreak && |
| 151 | in_array( |
| 152 | $child->nodeName, |
| 153 | $this->segmentBreakingTags |
| 154 | ) |
| 155 | ) { |
| 156 | // Add segment breaks for start tags specified in |
| 157 | // the config, unless the previous item is a break |
| 158 | // or this is the first item. |
| 159 | $this->cleanedContent[] = new SegmentBreak(); |
| 160 | } |
| 161 | |
| 162 | if ( $child->nodeType == XML_TEXT_NODE ) { |
| 163 | // Remove the path to the dummy node and instead |
| 164 | // add "." to match when used with context. |
| 165 | $path = preg_replace( |
| 166 | '!^/meta/dummy' . '!', |
| 167 | '.', |
| 168 | $child->getNodePath() |
| 169 | ); |
| 170 | $this->cleanedContent[] = new CleanedText( $child->textContent, $path ); |
| 171 | } else { |
| 172 | if ( $this->partOfContent && $child instanceof DOMElement ) { |
| 173 | $this->addPartOfContent( $child ); |
| 174 | } |
| 175 | $this->addContent( $child ); |
| 176 | } |
| 177 | if ( |
| 178 | !self::lastElement( $this->cleanedContent ) instanceof SegmentBreak && |
| 179 | in_array( |
| 180 | $child->nodeName, |
| 181 | $this->segmentBreakingTags |
| 182 | ) |
| 183 | ) { |
| 184 | // Add segment breaks for end tags specified in |
| 185 | // the config. |
| 186 | $this->cleanedContent[] = new SegmentBreak(); |
| 187 | } |
| 188 | } |
| 189 | } |
| 190 | |
| 191 | /** |
| 192 | * Check if a node matches criteria for removal. |
| 193 | * |
| 194 | * The node is compared to the removal criteria from the |
| 195 | * configuration, to determine if it should be removed completely. |
| 196 | * |
| 197 | * @since 0.0.1 |
| 198 | * @param DOMNode $node The node to check. |
| 199 | * @return bool true if the node match removal criteria, otherwise |
| 200 | * false. |
| 201 | */ |
| 202 | private function matchesRemove( $node ): bool { |
| 203 | if ( !array_key_exists( $node->nodeName, $this->removeTags ) ) { |
| 204 | // The node name isn't found in the removal list. |
| 205 | return false; |
| 206 | } |
| 207 | $removeCriteria = $this->removeTags[$node->nodeName]; |
| 208 | if ( $removeCriteria === true ) { |
| 209 | // Node name is found and there are no extra criteria. |
| 210 | return true; |
| 211 | } elseif ( is_array( $removeCriteria ) ) { |
| 212 | // If there are multiple classes for a tag, check if any |
| 213 | // of them match. |
| 214 | foreach ( $removeCriteria as $class ) { |
| 215 | if ( self::nodeHasClass( $node, $class ) ) { |
| 216 | return true; |
| 217 | } |
| 218 | } |
| 219 | } elseif ( self::nodeHasClass( $node, $removeCriteria ) ) { |
| 220 | // Node name and class name match. |
| 221 | return true; |
| 222 | } |
| 223 | return false; |
| 224 | } |
| 225 | |
| 226 | /** |
| 227 | * Check if a node has a class attribute, containing a string. |
| 228 | * |
| 229 | * Since this is for checking HTML tag classes, the class |
| 230 | * attribute, if present, is assumed to be a string of substrings, |
| 231 | * separated by spaces. |
| 232 | * |
| 233 | * @since 0.0.1 |
| 234 | * @param DOMNode $node The node to check. |
| 235 | * @param string $className The name of the class to check for. |
| 236 | * @return bool true if the node's class attribute contain |
| 237 | * $className, otherwise false. |
| 238 | */ |
| 239 | private static function nodeHasClass( $node, $className ): bool { |
| 240 | $classNode = $node->attributes->getNamedItem( 'class' ); |
| 241 | if ( $classNode == null ) { |
| 242 | return false; |
| 243 | } |
| 244 | $classString = $classNode->nodeValue; |
| 245 | $nodeClasses = explode( ' ', $classString ); |
| 246 | return in_array( $className, $nodeClasses ); |
| 247 | } |
| 248 | |
| 249 | /** |
| 250 | * Add an object to `$cleanedContent` if `$element` is of a certain type. |
| 251 | * |
| 252 | * @since 0.1.13 |
| 253 | * @param DOMElement $element |
| 254 | */ |
| 255 | private function addPartOfContent( $element ) { |
| 256 | // TODO: It may make more sense to have the logic live in the respective classes. |
| 257 | if ( $element->nodeName === 'a' ) { |
| 258 | $this->cleanedContent[] = new Link(); |
| 259 | } |
| 260 | } |
| 261 | |
| 262 | /** |
| 263 | * Get the last element in an array. |
| 264 | * |
| 265 | * @since 0.0.1 |
| 266 | * @param array $array The array to get the last element from. |
| 267 | * @return mixed|null The last element in the array, null if array is empty. |
| 268 | */ |
| 269 | private static function lastElement( $array ) { |
| 270 | if ( !count( $array ) ) { |
| 271 | return null; |
| 272 | } else { |
| 273 | return $array[count( $array ) - 1]; |
| 274 | } |
| 275 | } |
| 276 | |
| 277 | /** |
| 278 | * Cleans title and content. |
| 279 | * |
| 280 | * @since 0.1.10 |
| 281 | * @param string $displayTitle |
| 282 | * @param string $pageContent |
| 283 | * @return SegmentContent[] Title and content represented as `CleanedText`s and `SegmentBreak`s |
| 284 | * @throws LogicException If segmented title text is not an instance of CleanedText |
| 285 | */ |
| 286 | public function cleanHtmlDom( |
| 287 | string $displayTitle, |
| 288 | string $pageContent |
| 289 | ): array { |
| 290 | // Clean HTML. |
| 291 | $cleanedText = null; |
| 292 | // Parse latest revision, using parser cache. |
| 293 | $cleanedText = $this->cleanHtml( $pageContent ); |
| 294 | // Create a DOM for the title to get the Xpath, in case there |
| 295 | // are elements within the title. This happens e.g. when the |
| 296 | // title is italicized. |
| 297 | $dom = new DOMDocument(); |
| 298 | $dom->loadHTML( |
| 299 | '<h1>' . $displayTitle . '</h1>', |
| 300 | LIBXML_HTML_NODEFDTD | LIBXML_HTML_NOIMPLIED |
| 301 | ); |
| 302 | $xpath = new DOMXPath( $dom ); |
| 303 | $titleSegments = []; |
| 304 | $i = 0; |
| 305 | foreach ( $this->cleanHtml( $displayTitle ) as $titlePart ) { |
| 306 | if ( !$titlePart instanceof CleanedText ) { |
| 307 | throw new LogicException( |
| 308 | 'Segmented title is not an instance of CleanedText!' |
| 309 | ); |
| 310 | } |
| 311 | |
| 312 | $node = $xpath->evaluate( '//text()' )->item( $i ); |
| 313 | $titlePart->setPath( '/' . $node->getNodePath() ); |
| 314 | $titleSegments[] = $titlePart; |
| 315 | $titleSegments[] = new SegmentBreak(); |
| 316 | $i++; |
| 317 | } |
| 318 | array_pop( $titleSegments ); |
| 319 | if ( $cleanedText ) { |
| 320 | $cleanedText = array_merge( |
| 321 | $titleSegments, |
| 322 | [ new SegmentBreak() ], |
| 323 | $cleanedText |
| 324 | ); |
| 325 | } else { |
| 326 | $cleanedText = $titleSegments; |
| 327 | } |
| 328 | return $cleanedText; |
| 329 | } |
| 330 | |
| 331 | } |